Embed Size (px)
Citation preview
Breastfeeding Practices of Inuit Canadians
by
Kathryn Elizabeth McIsaac
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Dalla Lana School of Public Health University of Toronto
© Copyright by Kathryn Elizabeth McIsaac 2013
ii
Breastfeeding Practices of Inuit Canadians
Kathryn Elizabeth McIsaac
Doctor of Philosophy
Dalla Lana School of Public Health
University of Toronto
2013
Abstract
Background: Inuit populations are less likely to breastfeed than other Canadians and are
disproportionately affected by many conditions which breastfeeding is protective against. No
population-based research has explored the determinants of breastfeeding initiation and exclusive
breastfeeding to six months, as recommended by Health Canada, in this population.
Objective: The aim of this research is to examine the prevalence of breastfeeding initiation and
exclusive breastfeeding to six months among Canadian Inuit and to identify factors that are
associated with both breastfeeding practices.
Methods: Data for this research come primarily from two cross-sectional population-based
surveys: the Nunavut Inuit Child Health Survey and the IPY Inuit Health Survey. Possible
determinants of breastfeeding practices include household, maternal, infant and community
characteristics as well as engagement in traditional lifestyle. Logistic regression procedures (i.e.
multiple logistic regression; multinomial logistic regression; hierarchal logistic regression) were
used to model the determinants of breastfeeding initiation and exclusive breastfeeding as
recommended.
Results: The prevalence of breastfeeding initiation was 67% (95% CI: 62.8-72.4) in the entire
sample and 76% (95% CI: 70.6-82.1) in children raised by their biological mother. Twenty-three
iii
percent of Inuit Canadians practiced exclusive breastfeeding as recommended (95% CI: 16.2-
29.3). Among children being raised by their biological mother, infants of low birth weight and
whose mothers smoked during pregnancy were less likely to begin breastfeeding (OR: 0.16,
95%CI: 0.10-0.27; OR: 0.34, 95%CI: 0.19-0.61, respectively), while infants in crowded
households were more likely to begin breastfeeding (OR=2.01, 95%CI: 1.39-2.94). Families
currently receiving income support were less likely to stop exclusive breastfeeding their infant
before 5.5 months (OR≤1month=0.39, 95%CI: 0.14-1.05; OR1-<5.5 months=0.34, 95%CI: 0.13-0.85).
There was no relationship between classes of traditional lifestyle and either breastfeeding
initiation or exclusive breastfeeding as recommended.
Conclusions: These data suggest substantial room for improving breastfeeding practices of Inuit
Canadians, which may provide downstream health benefits to Inuit children. Increasing
exclusive breastfeeding to six months will require targeted interventions.
iv
Acknowledgments
I would like to express my sincere appreciation to my supervisor, Dr. Kue Young and my
committee members, Dr. Daniel Sellen and Dr. Wendy Lou, for their mentorship, guidance and
incisive feedback during the development and writing of this thesis. I would also like to
acknowledge my internal reviewers, Dr. Susan Bondy and Dr. Cameron Mustard, and external
reviewer, Dr. Noreen Willows, for their helpful suggestions.
The Inuit Health Survey would not have been possible without the tireless efforts of the
Qanuippitali Steering committee, chaired by Grace Egeland. I am most grateful to Dr. Egeland
for extending me the privilege of working with such a rich data set. I would also like to
acknowledge members of her team at the Centre for Indigenous Peoples’ Nutrition and
Environment, McGill University, particularly Helga Saudny and Nelofar Sheikh.
I extend my sincere appreciation to the Inuit who participated in this survey as well as the Inuit
organizations with whom I consulted during the preparation and writing of this thesis,
particularly those at Inuit Tuttarvingat, National Aboriginal Health Organization. Their time and
efforts are most appreciated.
I consider myself extremely fortunate to have developed a network of peers from the PhD
program who were always available for consultation. In particular: Laura Anderson, Fiona
Kouyoumdjian, Mana Rezai, Brendan Smith, and Sarah Jane Steele. And of course, David
Stock. Their critical “internal reviews” helped shape and inform many components of this
dissertation.
Most importantly, a huge thank you to my parents, Howard and Julie McIsaac, and my sister
Caroline McIsaac Johal for their continual love and support: I am who I am because of you.
My doctoral studies were largely funded by the Canadian Institutes of Health Research (2009-
2011). Additional funding support was provided by Health Canada (2009, 2012), the Centre for
Indigenous Peoples’ Nutrition and Environment (2010, 2012) and the University of Toronto
(2012-2013).
v
Table of Contents
Acknowledgments .......................................................................................................................... iv
Table of Contents ............................................................................................................................ v
List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................ xi
List of Appendices ........................................................................................................................ xii
Chapter 1 :Introduction and Objectives .......................................................................................... 1
1.1 Thesis Overview ................................................................................................................. 1
1.2 Thesis Objectives ................................................................................................................ 1
1.3 Roles and Responsibilities .................................................................................................. 2
1.4 Structure of Thesis .............................................................................................................. 3
1.5 Section References .............................................................................................................. 4
Chapter 2 : Background and Literature Review ............................................................................. 6
2 Background Overview ............................................................................................................... 6
2.1 Breastfeeding Definitions and Recommendations .............................................................. 6
2.2 Breastfeeding in Canada ..................................................................................................... 7
2.2.1 Historical Breastfeeding Practices .......................................................................... 7
2.2.2 Current Breastfeeding Practices .............................................................................. 8
2.3 Benefits of Breastfeeding .................................................................................................. 10
2.3.1 Infant Benefits ....................................................................................................... 11
2.3.2 Maternal Benefits .................................................................................................. 16
2.4 Factors Associated with Breastfeeding Practices ............................................................. 18
2.4.1 Maternal Characteristics ....................................................................................... 19
2.4.2 Infant Characteristics ............................................................................................ 24
2.4.3 Supportive Environments ...................................................................................... 25
vi
2.4.4 Contextual Factors ................................................................................................ 27
2.4.5 Factors Unique to an Inuit Setting ........................................................................ 29
2.5 Motivation for Research Project ....................................................................................... 31
2.6 Section References ............................................................................................................ 35
Chapter 3 :Study Methods ............................................................................................................ 52
3 Methods Overview ................................................................................................................... 52
3.1 Data Sources ..................................................................................................................... 52
3.2 Setting and Design of the IPY Inuit Health Survey and the Nunavut Inuit Child
Health Survey .................................................................................................................... 53
3.2.1 Demographic Characteristics of Regions Included in the IPY Inuit Health
Survey and Nunavut Inuit Child Health Survey ................................................... 53
3.3 Methods for the Nunavut Inuit Child Health Survey ........................................................ 56
3.3.1 Participants and recruitment strategies ................................................................. 56
3.3.2 Data collection strategies ...................................................................................... 57
3.3.3 Study variables: Measurement and classification ................................................. 59
3.4 Methods for the IPY Inuit Health Survey ......................................................................... 67
3.4.1 Participants and recruitment strategies ................................................................. 67
3.4.2 Data collection strategies ...................................................................................... 67
3.4.3 Study variables: Measurement and classification ................................................. 70
3.5 Statistical Analyses ........................................................................................................... 71
3.5.1 Complex survey methodology .............................................................................. 71
3.5.2 Model building strategies ...................................................................................... 73
3.5.3 Multilevel modeling .............................................................................................. 74
3.5.4 Latent class analysis .............................................................................................. 76
3.6 Ethics ................................................................................................................................. 77
3.7 Section References ............................................................................................................ 78
Chapter 4 : Manuscript 1 ............................................................................................................... 84
vii
4 Study Overview ........................................................................................................................ 84
4.1 Prevalence and characteristics associated with breastfeeding initiation among
Canadian Inuit from the 2007-2008 Nunavut Inuit Child Health Survey ......................... 85
Chapter 5 : Manuscript 2 ............................................................................................................. 108
5 Study Overview ...................................................................................................................... 108
5.1 Exclusive breastfeeding among Canadian Inuit: results from the Nunavut Inuit Child
Health Survey .................................................................................................................. 109
Chapter 6 : Manuscript 3 ............................................................................................................. 136
6 Study Overview ...................................................................................................................... 136
6.1 Measuring level of engagement in traditional activities among Inuit Canadians by a
latent class analysis and the application of this analysis to describe breastfeeding
practices among Canadian Inuit: IPY (International Polar Year) Inuit Health Survey .. 137
Chapter 7 : Discussion and Conclusions ..................................................................................... 163
7 Discussion Overview ............................................................................................................. 163
7.1 Main findings .................................................................................................................. 163
7.2 Strengths ......................................................................................................................... 165
7.3 Limitations ...................................................................................................................... 167
7.3.1 Selection Bias ...................................................................................................... 167
7.3.2 Information Bias ................................................................................................. 170
7.3.3 Confounding ....................................................................................................... 172
7.3.4 Statistical Concerns ............................................................................................. 172
7.3.5 External Validity ................................................................................................. 175
7.4 Public Health Significance .............................................................................................. 175
7.5 Future Directions for Further Research .......................................................................... 176
7.6 Section References .......................................................................................................... 178
viii
List of Tables
Chapters 1-3, 7 and Appendices
Table 1: Prevalence of Specific Health Events for which breastfeeding is protective against in
Inuit Populations and Comparator Populations ............................................................................ 33
Table 2: Communities participating in the IPY Inuit Health Survey and Nunavut Inuit Child
Health Survey ................................................................................................................................ 54
Table 3: Information collected in select survey modules: Nunavut Inuit Child Health Survey ... 58
Table 4: Information collected in each survey module: IPY Inuit Health Survey ........................ 69
Table 5: Child Inuit Health Survey 2007-2008 Data Collection Schedule ................................. 182
Table 6: Adult Inuit Health Survey 2007-2008 Data Collection Schedule ................................ 183
Chapter 4: Manuscript 1
Table 1: Unweighted household, maternal, infant and community characteristics of study
respondents, by caregiver subgroup, Nunavut Inuit Child Health Survey (n=361) ……………102
Table 2: Weighted household, maternal, infant and community characteristics of study
respondents, by caregiver subgroup. Nunavut Inuit Child Health Survey (n=1088) ………….103
Table 3: Weighted, bivariable prevalence odds ratios (pOR) and 95% confidence intervals (95%
CI) for tests of the association between selected household, maternal, infant and community
characteristics with breastfeeding initiation amongst Canadian Inuit, by caregiver subgroup.
Nunavut Inuit Child Health Survey (n=1088) ………………………………………………... 104
Table 4: Weighted multivariable prevalence odds ratios (pOR) and 95% confidence intervals
(95%CI) of factors associated with breastfeeding initiation, by caregiver subgroup. Nunavut
Inuit Child Health Survey (n=1088) ………………………………………………………….. 106
ix
Table 5: Fixed (prevalence odds ratios (pOR), 95% confidence intervals (95%CI)) and random
effect estimates from hierarchical linear model of breastfeeding initiation, by caregiver subgroup.
Nunavut Inuit Child Health Survey (n=1088) …………………………………………………107
Chapter 5: Manuscript 2
Table 1: Selected unweighted (n=188) and weighted (n=752) household, maternal, infant
and community characteristics of respondents, Nunavut Inuit Child Health Survey
(2007-2008) ……………………………………………………………………………..……...129
Table 2: Distribution of exclusive breastfeeding practices (weighted), by select household,
maternal, infant and community characteristics of respondents, Nunavut Inuit Child Health,
(n=752) Survey (2007-2008) .. ………………………………………………………………..130
Table 3: Weighted bivariate prevalence odds ratios (pOR) and 95% confidence interval of
selected household, maternal, infant and community characteristics from multinomial
logistic regression models, by exclusive breastfeeding duration. Nunavut Inuit Child Health
Survey, 2007-2008 (nweighted = 752) ……………………………………………………………133
Table 4: Multivariate adjusted weighted prevalence odds ratios and 95% confidence limits, by
exclusive breastfeeding duration. Nunavut Inuit Child Health Survey (2007-2008) (nweighted =
752) …………………………………………………………………………………………….135
Chapter 6: Manuscript 3
Table 1: Manifest variables that were considered for latent class analysis of traditional
behaviours amongst Canadian Inuit, IPY Inuit Health Survey ……………………………… 157
Table 2: Demographic characteristics of study respondents. IPY Inuit Health Survey and
Nunavut Inuit Child Health Survey …………………………………………………………... 158
Table 3: Traditional behaviours in study respondents. IPY Inuit Health Survey and Nunavut
Inuit Child Health Survey …………………………………………………………………….. 159
Table 4: Fit Statistics for latent class analysis of traditional behaviours amongst Canadian Inuit.
IPY Inuit Health Survey (n=2595) ……………………………………………………………..159
x
Table 5: Class and item membership probabilities for traditional behaviours amongst Canadian
Inuit and odds of class membership by age and sex, IPY Inuit Health Survey ………………..160
Table 6: Interpretations of classes from latent class analysis of traditional behaviours amongst
Canadian Inuit. IPY Inuit Health Survey (n=2595) ……………………………….. ………...161
Table 7: Odds of breastfeeding initiation, by class membership of traditional behaviours.
IPY Inuit Health Survey and Nunavut Inuit Child Health Survey (n=135) …………………... 162
Table 8: Odds of exclusive breastfeeding as recommended by class membership of traditional
behaviours. IPY Inuit Health Survey and Nunavut Inuit Child Health Survey ……………… 162
xi
List of Figures
Chapter 4: Manuscript 1
Figure 1: Survey response patterns, Nunavut Inuit Child Heath Survey ................................... 101
Chapter 5: Manuscript 2
Figure 1: Flow diagram of participants from the Nunavut Inuit Child Health Survey (2007-
2008) included in the present analysis ………………………………………………………...127
Figure 2: Exclusive breastfeeding duration (n=752), Nunavut Inuit Child Health Survey
(2007-2008) …………………………………………………………………………………… 128
Chapter 6: Manuscript 3
Figure 1: Flow diagram of participants in the IPY Inuit Health Survey (2007-2008) and
Nunavut Inuit Child Health Survey (2007-2008) included in these analyses …………………156
xii
List of Appendices
Appendix A: Data Collection Schedule ...................................................................................... 182
Appendix B: Child Questionnaire .............................................................................................. 185
Appendix C: Ethics Approval ..................................................................................................... 194
Appendix D: Summary of Discussion Group at the National Aboriginal Health Organization . 195
Appendix E: Power Calculations ................................................................................................ 197
1
Chapter 1 :Introduction and Objectives
1.1 Thesis Overview
Exclusive breastfeeding to six months offers substantial short and long term health benefits to
both the infant and mother1-5
. In 2010, 26% of new mothers in Canada practiced exclusive
breastfeeding to six months6 as recommended by Health Canada
7 and the World Health
Organization8, a trend which has been increasing in recent years (i.e. from 17.3% in 2003)
6.
Although attention has been given to the determinants of breastfeeding initiation and exclusive
breastfeeding duration in all Canadians9, as well as specific Canadian subpopulations such as
First Nations 10-12
, research in Inuit Canadians is limited13
.
The decision to begin and continue breastfeeding is strongly linked to social norms: it varies
substantially between regions9,14
, populations15
and cultures16
. This alone provides justification
to explore more comprehensively factors related to both initiation and duration from an Inuit
Canadian lens, as information from other populations is likely not directly generalizable to Inuit
Canadians. Furthermore, Inuit Canadians also experience a disproportionate burden of
morbidities which breastfeeding is protective against (e.g. respiratory infections)17
, and these
conditions are likely to be more severe compared to other Canadians18,19
. Encouraging a health
promoting behaviour like exclusive breastfeeding to six months can have substantial public
health benefit.
To examine the breastfeeding practices of Inuit Canadian women, data from two population-
based surveys conducted in Canada’s North were used: the Nunavut Inuit Child Health Survey
and the IPY (International Polar Year) Inuit Health Survey. Community measures, collected
from various sources (e.g. Government of Nunavut), were also included.
1.2 Thesis Objectives
Objective 1: To describe the prevalence of breastfeeding initiation among Inuit Canadians
Objective 2: To identify factors associated with breastfeeding initiation in Inuit Canadians
2
Objective 3: To describe the distribution of exclusive breastfeeding duration among Inuit
Canadians
Objective 4: To identify factors associated with exclusive breastfeeding, as recommended (i.e.
to six months) among Inuit Canadians.
Objective 5: To develop a latent model that describes heterogeneity in traditional activities
amongst Canadian Inuit using the IPY Inuit Health Survey.
Objective 6: To evaluate the association between traditional activities and i) breastfeeding
initiation, and ii) recommended breastfeeding duration (i.e. 6 months) using data
from both the IPY Inuit Health Survey and the Nunavut Inuit Child Heath Survey.
1.3 Roles and Responsibilities
Data for the Nunavut Inuit Child Health Survey and IPY Inuit Health Survey were collected
during the International Polar Year (2007-2008) through efforts coordinated by Dr. Grace
Egeland (McGill University). The surveys were developed to better understand the factors
contributing to Inuit health and the Inuit spirit of thriving in the face of social and environmental
changes20
and included self-reported information on many different aspects of health. This
particular study is my contribution to a collection of several existing studies about the Health and
Wellness of Inuit Canadians.
I identified the research question for this thesis and gained approval for the project from the IPY
Inuit Health Survey Steering Committee and the University of Toronto’s Research Ethics Board.
I reviewed the literature, identified and defined the variables for research, identified the
appropriate analytic strategies and analyzed and interpreted the results for each study.
Manuscript writing was an iterative process: I completed the first draft and circulated to my
committee for feedback. Each manuscript was revised, as necessary, after consultation and
discussions with my committee members. I also conducted ad hoc consultations with Inuit
women to help better understand the research question.
3
1.4 Structure of Thesis
This thesis is organized into seven chapters. Chapter 1 provides a brief overview of the overall
thesis. Chapter 2 provides a review of the current literature about research on the determinants
of breastfeeding amongst Inuit as well as other Canadians and identifies gaps in knowledge.
Chapter 3 contains detailed study methods. Chapters 4, 5 and 6 contain three separate
manuscripts: in the first manuscript, factors related to breastfeeding initiation among Canadian
Inuit are identified (Objectives 1 and 2); in the second manuscript, factors related to exclusive
breastfeeding to six months among Canadian Inuit are identified (Objectives 3 and 4); and in the
third manuscript, the effects of engaging in traditional activities on breastfeeding practices are
evaluated (Objectives 5 and 6). The final chapter (Chapter 7) provides a general discussion of
the thesis results.
4
1.5 Section References
1. Kramer M, Kakuma R. Optimal Duration of exclusive breastfeeding. Cochrane Database
of Systematic Reviews. 2002;1.
2. Raisler J, Alexander C, O'Campo P. Breast-feeding and infant illness: a dose-response
relationship? Am J Public Health. 1999;89(1):25-30.
3. Oddy WH, Li J, Landsborough L, Kendall GE, Henderson S, Downie J. The association
of maternal overweight and obesity with breastfeeding duration. The Journal of
Pediatrics. 2006;149(2):185-191.
4. Davis M, Savitz D, Graubard B. Infant feeding and childhood cancer. Lancet. 1988;2:365
- 368.
5. Davis MK. Breastfeeding and Chronic Disease in Childhood and Adolescence. Pediatric
Clinics of North America. 2001;48(1):125-141.
6. Health Canada. Trends in breastfeeding practices in Canada (2001-2009-2010). 2012;
http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/prenatal/trends-tendances-
eng.php. Accessed Dec 8, 2012.
7. Health Canada. Exclusive Breastfeeding Duration: 2004 Health Canada
Recommendation. 2004; http://www.hc-sc.gc.ca/fn-an/alt_formats/hpfb-
dgpsa/pdf/nutrition/excl_bf_dur-dur_am_excl-eng.pdf. Accessed June 3, 2012.
8. World Health Organization. The World Health Organization's infant feeding
recommendation. 2001;
http://www.who.int/nutrition/topics/infantfeeding_recommendation/en/index.html.
Accessed June 20, 2009.
9. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
10. Clifford TJ, Campbell MK, Speechley KN, Gorodzinsky F. Factors Influencing Full
Breastfeeding in a Southwestern Ontario Community: Assessments at 1 Week and at 6
Months Postpartum. J Hum Lact. 2006;22(3):292-304.
11. Black R, Godwin M, Ponka D. Breastfeeding among the Ontario James Bay Cree: a
retrospective study. Can J Public Health. 2008;99(2):98-101.
12. Simard I, O'Brien H, Beaudoin A, et al. Factors influencing the initiation and duration of
breastfeeding among low-income women followed by the Canada prenatal nutrition
program in 4 regions of Quebec. J Hum Lact. 2005;21(3):327-337.
13. Dufour R. Breast feeding and adoption among the Inuit population of Northern Quebec.
Can J Publ Health. 1984;75(2):137-140.
5
14. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
15. Callen J, Pinelli J. Incidence and duration of breastfeeding for term infants in Canada,
United States, Europe and Australia: a literature review. Birth. 2004;31(4):285-292.
16. Stuart MacAdam P, Dettwyler KA. Breastfeeding: biocultural perspectives. New York:
Transaction Publishers; 1995.
17. Chantry C, Howard C, Auinger P. Full breastfeeding duration and associated decrease in
respiratory tract infection in US Children. Pediatrics. 2006;117(2):425-432.
18. Orr PH. Respiratory tract infections in Inuit children: "Set thine house in order". CMAJ.
2007;177(2):167-168.
19. Bulkow LR, Singleton RJ, Karron RA, Harrison LH. Risk Factors for Severe Respiratory
Syncytial Virus Infection Among Alaska Native Children. Pediatrics. 2002;109(2):210-
216.
20. Saudny H, Leggee D, Egeland G. Design and methods of the Adult Inuit Health Survey
2007–2008. International Journal of Circumpolar Health.
2012;71:10.3402/ijch.v3471i3400.19752.
6
Chapter 2 : Background and Literature Review
2 Background Overview
This chapter reviews the literature about breastfeeding practices to provide a critical context for
the thesis. Specifically, this chapter reviews current recommendations for breastfeeding from
prominent health organizations, provides an overview of historic breastfeeding practices in
Canada from the early 1900’s until the present day, reviews the international literature about
breastfeeding benefits to both the infant and mother, and identifies factors that have been
associated with breastfeeding practices. When available, information on determinants of
breastfeeding initiation and exclusive breastfeeding duration in Canadian Aboriginal groups are
provided.
2.1 Breastfeeding Definitions and Recommendations
Exclusive breastfeeding to six months and continued breastfeeding for up to two years and
beyond is recommended by many national and international health organizations, including the
World Health Organization1, Health Canada
2,3 and the American Academy of Pediatrics
4,5. An
infant is considered exclusively breastfed when they receive only breast milk (can be expressed
or provided by a wet nurse6). Under this definition, an infant cannot receive any other food or
drink, even water. Oral hydration solutions, drops and syrups containing vitamins, minerals and
medicines are permitted1.
The recommendation to exclusive breastfeed to six months is largely based on a systematic
review of studies around the world that compared exclusive breastfeeding for four (or less than
four) months with exclusive breastfeeding for between four and six months7, as well as results
from a randomized control trial in Belarus (the Promotion of Breastfeeding Intervention Trial or
PROBIT trial)8. Prior to 2001, the World Health Organization recommended exclusive
breastfeeding for between four and six months (thereby introducing complementary foods into
the diet between four and six months). One of the primary concerns with later introduction of
complementary foods (i.e. around six months) into the infant diet was that breast milk alone
would be nutritionally insufficient for the growth needs of infants in both developed and
7
developing countries (specifically in terms of iron and zinc)7. However, data from the systematic
review8, which was further supported by the randomized trial
9, found no adverse growth effects
associated with longer (i.e. six months duration) of exclusive breastfeeding.
Most babies can begin breastfeeding: there are very few contraindications5. Infants with a rare
genetic disorder, galactosemia are unable to metabolize breast milk and should receive other
forms of nutrition.10
Mothers with certain infectious diseases should also avoid breastfeeding
(i.e. Infectious Tuberculosis (TB); Human Immunodeficiency Virus (HIV); Human T-cell
Lymphotrophic Virus type I or II (HTLV I, II); herpes virus with lesions on the breast)11
.
Breastfeeding is also contraindicated with specific drugs, both prescription (i.e. group II anti-
parasitics; chemotherapy agents) and illicit street drugs (i.e. marijuana, PCP, cocaine)5. Finally,
mothers should minimize their alcohol consumption and wait for at least two hours after
consuming two drinks before breastfeeding their infant5.
2.2 Breastfeeding in Canada
Breastfeeding has been a cornerstone of public health policy in Canada since the 1920’s when
the government released The Canadian Mother’s Book, which contained the country’s first
official recommendation around breastfeeding practices: exclusive breastfeeding to nine months
and continued breastfeeding to two years12
. Although recommendations about infant feeding
practices have changed over time adapting to cultural trends, medical beliefs and scientific
evidence, “breast is best” has resonated as a central theme in all of the Government of Canada’s
recommendations. Yet, Canadians have not always embraced this philosophy.
Data from a variety of sources suggest breastfeeding has waxed and waned in popularity since
the 1900’s12
. Information on the breastfeeding practices of Canadians prior to 1960 comes either
from medical records or small community based samples. National surveys measuring
breastfeeding practices of Canadian women began in the 1960’s with the Nutrition Canada
Survey12
.
2.2.1 Historical Breastfeeding Practices
Information in the historical breastfeeding practices of Canadians was obtained from a recent
book The One Best Way: breastfeeding history, politics and policy in Canada published by two
researchers from the University of British Columbia (Nathoo) and the University of Victoria
8
(Ostry). The book neatly synthesizes breastfeeding practices in Canada over the past 100 years,
and provides historical, cultural, political and economic context for these practices.13
Breastfeeding was prevalent in Canada around the turn of the century until the First World War.
In 1917, 80% of mothers in the lower economic class initiated breastfeeding and 61% continued
breastfeeding for at least three months12
. During this period, infant mortality from diarrheal
disease was particularly high (i.e. total infant mortality in 1926 was 102.1 deaths per 1000 live
births). With the growing popularity of the germ theory of disease, came an awareness that cow’s
milk contained micro-organisms that could potentially cause illness and even death to an infant12
.
Breastfeeding was seen as a way to reduce this mortality.
By 1963, only 36% of mothers initiated breastfeeding and 34% were practicing any
breastfeeding at three months12
. Several factors precipitated this shift. Improved sanitation had
a direct impact on infant mortality and there was no longer an apparent benefit of breastfeeding
compared to other infant feeding alternatives. At the same time, infant feeding formulas were
being developed and marketed aggressively to new mothers and health care providers. Lastly,
birth shifted away from the community and into the hospital: 45% of infants were born in the
hospital in 1960 compared to 95% in 194012
. These births were attended largely by male
physicians with limited knowledge and training about overcoming breastfeeding barriers.
Offering formula food was a practical solution12
.
Spurred by feminism and grassroots movements (e.g. La Leche League) breastfeeding
experienced a revival from the late 1960’s and onwards. Initiation rates increased 135% between
1965-1971 and 1978 from 26% to 61%12
. Generally, rates have remained stable or increased
since that time. These increases have been supported by strong local, provincial, national and
even international campaigns (e.g. The Baby Friendly Hospital Initiative, see Section 2.4.4)12
.
2.2.2 Current Breastfeeding Practices
More recent data confirm that breastfeeding practices continue to improve in Canada. Between
2001 and 2010, the initiation rate increased from 81.5% to 87.3%14
. In 2003, 37.3% of new
mothers practiced exclusive breastfeeding to at least four months and 17.3% to six months,
while, in 2010, these numbers had increased to 44.2% and 25.9%14
, respectively. During this
time (2004), Health Canada revised their recommendations from exclusive breastfeeding
9
between four and six months12
to six months, in line with recently revised (2001)
recommendations from the World Health Organization1.
In spite of improvements in breastfeeding practices, notable variations exist across population
subgroups (further described in Section 2.4). Younger mothers are particularly vulnerable to
sub-optimal breastfeeding practices. The 2006 Maternity Experiences Survey found that 83.3%
of mothers aged 15 to 19 years initiate breastfeeding and 22.8% practice any breastfeeding to six
months15
. This is compared to 92.5% of mothers aged 40 and older who initiate breastfeeding
and 69.7% who practice any breastfeeding to six months15
. The same survey found that
breastfeeding practices varied by income and education: 82.8% of mothers with less than a high
school education initiate breastfeeding compared to 95.9% of mothers who are university
graduates, while 86.9% of mothers whose household income is below the low income cut-off
initiate breastfeeding (i.e. families likely to spend 20 percentage points more than the average
family on food, shelter and clothing, taking into account family size and costs of living in
different regions16
) compared to 91.1% of mothers whose household income is above the low
income cut-off15
.
Place of residence is also related to breastfeeding practices in Canada (further information in
Section 2.4). British Columbia and Yukon report the highest rates of both initiation and
exclusive breastfeeding to six months, while Newfoundland and Prince Edward Island report the
lowest rates (Yukon: 96.7% initiate; 34.2% practice exclusive breastfeeding to six months;
Prince Edward Island: 71.9% initiate; 10.0% practice exclusive breastfeeding to six months)15
.
In fully adjusted models1, mothers living in British Columbia were 1.94 times (95% CI: 1.42,
2.64) more likely and mothers in the Northern Territories (i.e. Northwest Territories, Nunavut,
Yukon) were 3.02 times more likely (95% CI: 2.21-4.12) to practice exclusive breastfeeding to
six months compared to mothers living in the Atlantic provinces (i.e. New Brunswick;
Newfoundland; Nova Scotia; Prince Edward Island)17
.
1 Models adjusted for: marital status; maternal self-rated health; maternal smoking during pregnancy; location of
infant’s birth (e.g. hospital, private home); delivery mode (i.e. vaginal, Caesarean); infant admission to neo-natal
intensive care unit; maternal employment <6 months after delivery; maternal education; maternal age at first
pregnancy; number of past pregnancies; pre-pregnancy body mass index.
10
Aboriginal Canadians (i.e. First Nations, Inuit, Metis) are less likely than other Canadians to both
initiate and exclusive breastfed as recommended according to population-based surveys14,18
. In
2010, fewer Aboriginal mothers living off reserve initiated and practiced exclusive breastfeeding
to six months compared to other Canadians (77.8% vs. 88.0%; 17.5% vs. 23.5% respectively)14
.
Among Inuit Canadians, the Aboriginal Children’s Survey (2006) reports 76% of mothers
initiate breastfeeding and 58% practice any breastfeeding at 6 months. The proportion of Inuit
who are practicing breastfeeding exclusively at six months is unclear18
. Trends over time in
breastfeeding practices among all Aboriginal Canadian groups have not been reported.
2.3 Benefits of Breastfeeding
Breastfeeding is viewed as an important public and population health concern because of its
noted health benefits to the infant and the mother5. Some benefits of breastfeeding for the infant
occur only while the child is receiving breastfeeding, while others extend into adulthood19,20
.
Research suggests infants who are breastfed have lower risk of infection in infancy (e.g.
gastrointestinal21
; respiratory22
), reduced risk of conditions arising in childhood (e.g. otitis
media19
, atopic dermatitis23
) and reduced risk of conditions developing in early and later
adulthood (e.g. type II diabetes24
and obesity25
). Breastfeeding may also be related to improved
cognitive development although research is inconsistent26-28
.
The benefits of breastfeeding extend beyond the infant to the mother. Some of the noted
maternal benefits of breastfeeding include: improved birth spacing29
; lower risk of postpartum
depression19
; lower risk of reproductive cancers (breast30
; ovarian19
); lower risk of developing
type II diabetes31
; and lower risk of cardiovascular disease32
. Although some research has
suggested a protective effect of breastfeeding against both osteoporosis and a high body mass
index, a recent systematic review did not find sufficient evidence to support these associations.19
It should be noted that there is a dose-response effect for some of the observed benefits of
breastfeeding, wherein longer durations of exclusive breastfeeding provide the most benefit (e.g.
gastrointestinal infections)33
. Moreover, infants who receive any breastfeeding may have some
of the same benefits, although to a smaller degree compared to infants who receive no
breastfeeding at all33,34
.
11
2.3.1 Infant Benefits
2.3.1.1 Infection Risk
There is a strong body of literature to support that infants who receive some breastfeeding are
less likely to be hospitalized from lower respiratory tract infection. A meta-analysis of seven
studies report those infants who were breastfed for at least four months (compared to none) had a
0.28 times lower risk (95% CI: 0.14-0.54) of hospitalization from a lower respiratory tract
infection during infancy.22
The relationship persisted even after stratifying for smoking status
and socioeconomic position, two strong confounding variables22
. A more recent systematic
review further substantiates a strong protective effect of breastfeeding against infant
development of respiratory tract infection35
.
The effect of breastfeeding on the prevention of respiratory tract infections has been studied in
Aboriginal populations in Australia36
, the United States37
and Greenland38
. Although
breastfeeding was defined in various ways in the separate studies (e.g. ever vs. never37
; exclusive
breastfeeding for six months vs. never breastfeeding38
), each found that some breastfeeding was
associated with a reduced risk of respiratory infection. One of these studies, a community based
cohort study of Greenlandic Inuit children, reported that Inuit infants who were never breastfed
had a 3.73 times greater risk (95% CI: 1.24-11.20) of a lower respiratory tract infection
compared to those exclusively breastfed for six months38
.
In terms of gastrointestinal infection, a meta-analysis of 16 observational studies reported that
among infants less than one year of age, those who received some breastfeeding were 0.36 times
as likely (95% CI: 0.32-0.41) to acquire a gastrointestinal infection compared to those who did
not, although significant heterogeneity was noted21
. Heterogeneity is likely related to
inconsistent definitions of breastfeeding across studies and differential definitions of infection.21
The relationship between gastrointestinal infection and breastfeeding is also supported in the
PROBIT randomized control trail of exclusive breastfeeding for three or six months, conducted
in Belarus (OR=0.60, 95%CI: 0.40=0.91).8
2.3.1.2 Otitis Media
In a 2009 meta-analysis of five studies, Ip et al19
found that the risk of otitis media was lower
amongst children who received any breastfeeding compared to children who were exclusively
12
formula fed (OR=0.77, 95% CI: 0.64-0.91). Restricting these analyses to studies that compared
exclusive breastfeeding for at least 3 months with exclusive formula feeding further strengthens
the protective effect of breastfeeding against otitis media development (OR=0.50, 95%CI: 0.36-
0.70).
A cohort of Australian Aboriginal children did not find a significant association between
breastfeeding for less than three months (compared to more) and developing recurrent ear
infection (OR=1.12, 95%CI: 0.85-1.47).36
; however this study included children up to the age of
18 years, and the immune protective benefits of breast milk likely do not extend much beyond
infancy39
.
Yet, a more recent Canadian study finds that beginning breastfeeding will only lower risk of
initial otitis media infection (OR=0.87, 95% CI: 0.81-0.92)40
and will not affect the risk of
recurrent otitis media infection, a concern in Canadian Inuit communities.41-43
2.3.1.3 Sudden Infant Death Syndrome
Breastfeeding appears to confer protection against infant mortality from sudden infant death
syndrome (SIDS). A meta-analysis of 23 studies found that infants who were exclusively
formula fed had a 2.11 times greater risk of SIDS compared to infants who began breastfeeding
(95%CI: 1.66-2.68)44
. An updated meta-analysis further supports this research and demonstrates
a dose-response protective effect of breastfeeding against SIDS: the combined effect of any
breastfeeding compared to no breastfeeding was 0.55 (95% CI: 0.44-0.69)45
, while the combined
effect of any duration of exclusive breastfeeding compared to exclusive formula feeding was
0.27 (95% CI: 0.24-0.31)45
, although significant heterogeneity was noted in the latter. This
estimate was unadjusted, which may account for at least part of the heterogeneity. The varied
definitions of exclusive breastfeeding across studies may also be contributing to heterogeneity.45
2.3.1.4 Asthma
Two meta-analyses of twelve46
and fifteen19
prospective studies report that, among children with
no family history of asthma, risk of developing asthma is lowest among children who receive any
breastfeeding compared to children who are exclusively formula fed (OR=0.70, 95% CI: 0.60-
0.8146
; OR=0.73, 95% CI: 0.59-0.92)19
. Although a more recent meta-analysis may seem to
contradict these findings (OR=0.88, 95% CI: 0.71-1.08)47
, these results did not stratify by family
13
history. Previous research suggests that the association between beginning breastfeeding and the
incidence of asthma is modified by familial history19
, wherein the protective effect of
breastfeeding is greater in children in families with as opposed to without a family history.
Combining these two strata together does not provide an accurate etiologic representation of the
protective effect of breastfeeding against the development of asthma19
.
2.3.1.5 Atopic Dermatitis
Breastfeeding appears to protect against development of atopic dermatitis in childhood. A meta-
analysis of 18 prospective studies, which followed children an average of 4.5 years after birth,
found that children who received any breastfeeding were less likely to develop atopic dermatitis
compared to children who did not begin breastfeeding (OR=0.78, 95% CI: 0.62-0.99)23
. A more
recent meta-analysis of 21 observational studies also suggest a modest protective effect of any
breastfeeding compared to no breastfeeding on development of childhood atopic dermatitis
(OR=0.70, 95% CI: 0.50-0.99).48
This finding is consistent with data from the PROBIT trial
comparing exclusive breastfeeding for three months to exclusive breastfeeding for six months8.
More recently, research among more than 15,000 Danish infants suggest that the protective
effect of breastfeeding on atopic dermatitis may only be important for children with a familial
history of the disease49
.
2.3.1.6 Childhood Cancer
The association between breastfeeding practices and development of childhood leukemia was
investigated in a meta-analysis of 22 case-control studies50
. There were 14 studies that explored
the protective benefit of continued breastfeeding to six months or more compared to no
breastfeeding on the development of acute lymphoblastic leukemia. There were 8 studies that
explored the protective effect of continued breastfeeding to six months or more compared to no
breastfeeding on the development of acute myeloblastic leukemia. Although the authors note a
modest, significant protective effect of breastfeeding practices on both forms of leukemia, the
effect was strongest for acute lymphoblastic leukemia50
. Moreover, the effect disappeared after
only including studies that controlled for socioeconomic status50
. An updated meta-analysis
including an additional three high to fair quality case-control studies suggests that the protective
effect of six months of breastfeeding is only observed among cases with acute lymphoblastic
leukemia (OR=0.81, 95% CI: 0.71-0.91)19
and not acute myeloblastic leukemia.
14
2.3.1.7 Cognitive Development
The effect of breastfeeding practices on cognitive development in children is unclear. Two
meta-analyses report that children who received any breastfeeding score approximately three
points higher on various measures of cognitive performance compared to children who did not
receive any breastfeeding (mean=2.89, 95% CI: 2.41-3.3726
; mean=3.37, 95% CI: 1.73-6.7427
).
However, failing to control for important covariates may be driving these associations51,52
. A
meta-analysis by Der et al report that for each one standard deviation increase in maternal
intelligence, the odds of maternal breastfeeding increase 2.3 times (95% CI: 2.16-2.44)27
. After
adjusting for maternal intelligence and other important covariates, the effect of breastfeeding on
cognitive development attenuated and was no longer statistically significant (mean=0.52, 95%
CI: -0.19-1.23). Moreover, the effect of breastfeeding on cognitive development may be
modified by birth weight, wherein low birth weight babies receive greater benefit26
, but many
studies provide only pooled estimates of development, potentially masking an important
component on the causal pathway. A randomized trial of over 10,000 infants in Belarus, does
support an association between exclusive breastfeeding to six months compared to three months
on several cognitive performance measures.28
2.3.1.8 Inflammatory Bowel Disease
A meta-analysis of 14 studies found a reduced risk of any breastfeeding compared to no
breastfeeding for both ulcerative colitis (UC) and Crohn’s disease (CD) (ORUC=0.86, 95% CI:
0.76-0.96; ORCD=0.83, 95% CI: 0.75-0.95), although significant heterogeneity was reported53
.
After including only those studies deemed to be of high quality, the effect was strengthened
(ORUC=0.46, 95%CI: 0.32-0.65; ORCD=0.57 95% CI: 0.41-0.79), although marginal
heterogeneity was noted (pUC=0.063; pCD=0.27). Some of this heterogeneity likely relates to the
various definitions of breastfeeding included in these studies (e.g. ever vs. never; <2 months vs.
never)53
, as well as the varied age ranges of participants between published studies (e.g. children,
adults, both).
2.3.1.9 Celiac Disease
Any breastfeeding compared to no breastfeeding at the time gluten was introduced in an infant’s
diet was associated with a reduced risk of developing celiac disease in a meta-analysis of four
case-control studies (OR=0.48, 95% CI: 0.40-0.59)54
. These studies included mostly younger
15
children (up to age 7) but also included one study where participants ranged in age from 0 to
14.9 years54
. Although the protective effect of breastfeeding against celiac disease was fairly
strong, maternal recall of the timing and types of complementary foods introduced in the diet is
fairly poor.55
Prospective research is needed to confirm this association.
2.3.1.10 Type II Diabetes
Children who receive any breastfeeding are less likely to develop Type II Diabetes later in life56
compared to children who do not begin breastfeeding (OR=0.61, 95% CI: 0.44-0.85)24
according
to a recent meta-analysis that compiled 23 studies. This protective effect was seen in studies in
younger children57
and adults followed for as many as 71 years 58
.
There have been two studies specific to Aboriginal populations which examined early onset of
Type II Diabetes – both were included in the meta-analysis57,59
. Among Pima Indians aged 10 to
39 years, some breastfeeding (OR=0.64, 95% CI: 0.43-0.99) and excusive breastfeeding to six
months (OR=0.41, 95% CI: 0.18-0.93)59
, in comparison to infants who were exclusively formula
fed, were both associated with a lower risk of developing Type II Diabetes. In Canadian First
Nations, a population where early onset Type II Diabetes is becoming increasingly prevalent60
,
continued breastfeeding beyond 12 months was associated with a lower risk of types II diabetes
(RR=0.24, 95% CI: 0.07-0.84)57
. Type II Diabetes risk is not homogeneous in Canadian
Aboriginal groups, however, and Inuit Canadians actually report a lower or similar risk of the
disease compared to other Canadians.60
2.3.1.11 Body Mass Index
The effect of breastfeeding on childhood and adult obesity is unclear. Three meta-analyses19,25,61
of published observational studies suggest modest benefits of any breastfeeding on developing
obesity in childhood or in later life (e.g. OR per additional year breastfed=0.96, 95%CI: 0.94-0.9825
;
OR=0.78 (95%CI: 0.72-0.84)61
). However, findings from PROBIT, a randomized trial in
Belarus comparing exclusive breastfeeding to 3 months to exclusive breastfeeding to 6 months
do not support an association 6.5 years after birth62
. Moreover, a Canadian study using data
from the Nunavut Inuit Child Health Survey did not find an association between breastfeeding
and childhood obesity in Inuit children (aged 3 to 5), though this finding may be related to study
power63
.
16
2.3.1.12 Other Research in Aboriginal Populations
Some additional benefits of breastfeeding have been explored in Canadian Aboriginal
populations. A recent analyses using data from the 2006 Aboriginal Children’s Survey found
that caregivers of First Nations and Metis children who were under the age of five and living off
reserve were 1.5 times (95% CI: 1.2-1.8)64
more likely to report their child was in excellent or
very good health if the child began breastfeeding as opposed to receiving exclusively formula.
Caregiver perception of child’s health is related to activity restrictions, chronic conditions, health
difficulties and health care usage and is generally considered a reasonable proxy for the child’s
overall health65
.
2.3.2 Maternal Benefits
2.3.2.1 Improved Birth Spacing
The contraceptive benefit of prolonged breastfeeding has been noted for several decades29
.
Breastfeeding an infant, in both developed66,67
and developing67
countries is associated with
longer durations of postpartum amenorrhea.
2.3.2.2 Post-partum depression
A systematic review of six studies suggests that women who breastfeed have a lower risk of
postpartum depression19
, although many of the studies included were noted to be of low quality.
Two more recent studies, one in Scotland and the other in Canada further corroborate these
findings68,69
. The Canadian study, which included 1423 women, found that the risk of major
post-partum depression was 2.12 times greater (95% CI: 1.21-3.70) among mothers who were
not breastfeeding at eight weeks compared to those who were.69
One cannot discount the
possibility of reverse causality, wherein mothers experiencing post-partum depression are more
likely to cease exclusive breastfeeding before the recommended six months. Prospective
research with frequent measures of maternal state during the post-partum period would be
helpful to further clarify this reported association.
2.3.2.3 Breast Cancer
Breastfeeding an infant confers protection against maternal development of breast cancer. One
meta-analysis of 23 studies found that women who initiated breastfeeding had a 0.82 times lower
17
risk of breast cancer (95% CI: 0.76-0.89) compared to women who did not initiate
breastfeeding30
. The effect was similar by menopausal status (Pre-menopausal: OR =0.81, 95%
CI: 0.72-0.91; Post-menopausal: OR: 0.84 95% CI: 0.69, 1.03), although only significant in
women developing pre-menopausal breast cancer 30
. A more recent meta-analysis found a clear
dose-response relationship between breastfeeding duration and breast cancer incidence whereby
the incidence of maternal breast cancer decreased 4.3% for each additional year of breastfeeding
(95%CI: 2.9-5.8).70
2.3.2.4 Ovarian Cancer
A meta-analysis compiling evidence from nine studies suggests that women who ever breastfed
their infants were significantly less likely to develop ovarian cancer over their lifetime
(OR=0.79, 95%CI: 0.68-0.91) compared to those who never breastfed19
. The protective effect of
breastfeeding on ovarian cancer was strengthened when the researchers compared women who
had at least 12 months cumulative experience breastfeeding to those with no experience
(OR=0.72, 95%CI: 0.54-0.97)19
.
2.3.2.5 Type II Diabetes
Data from the Nurses’ Health Study (NHS) and the Nurses’ Health Study II (NHS II)
demonstrate that each additional year of breastfeeding confers further protection against maternal
development of Type II diabetes in the following 15 years. In the NHS, the risk of diabetes was
0.96 (95% CI: 0.92-0.99) lower for each additional year a mother breastfed her infant. In the
NHS II the risk of diabetes was 0.88 (95% CI: 0.82-0.94) lower for each additional year of
breastfeeding.31
Further evidence from cross-sectional research suggests that any breastfeeding is associated with
a lower odds of metabolic syndrome (OR=0.79, 95%CI: 0.63-0.99)71
. These findings are
consistent with other biological evidence that indicate lactating improves maternal glucose
sensitivity during pregnancy72
.
2.3.2.6 Maternal Cardiovascular Disease
Evidence from two large cohort studies of women support the protective effect of maternal
breastfeeding on later development of cardiovascular disease. Using data from the Women’s
18
Health Initiative, Schwarz et al73
found that women who practiced breastfeeding for 24 months
or more over their lifetime were less likely to develop any cardiovascular disease (RR=0.89,
95%CI: 0.80-0.98) compared to women who did not initiate breastfeeding. Turning to specific
types of cardiovascular disease, breastfeeding women were less likely to develop hypertension
(RR=0.87, 95% CI: 0.82-0.92) and hyperlipidemia (RR=0.80, 95% CI: 0.73-0.87). Moreover,
there was a significant protective dose response trend (p<0.001) noted for longer duration of
breastfeeding against cardiovascular disease (any, and specific subtypes).73
Data from the Nurses’ Health Study suggest a protective effect of breastfeeding on the incidence
of myocardial infarction. Breastfeeding for two years or more over a lifetime was associated
with a decreased risk of myocardial infarction (RR=0.77; 95% CI: 0.62-0.94), although the
protective effect of breastfeeding appeared to decrease over time. Stratifying for time since
giving birth, the relative effect of two years of breastfeeding compared to no years was 0.66
(95% CI: 0.49-0.89) for women who had given birth in the past 30 years and 0.90 (95% CI: 0.67-
1.19) for women who had given birth 30 years ago or more.32
2.3.2.7 Rheumatoid Arthritis
Using data from the Nurses’ Health Study, Karlson et al noted that breastfeeding an infant was
associated with a decreased risk of rheumatoid arthritis74
. The effect was evaluated using several
definitions of breastfeeding (i.e. ever vs. never; ≤ 3 months of any breastfeeding vs. never; 4 to
11 months vs. never; 12 to 23 months vs. never; and 24 months or more vs. never). Although the
relationship was only significant among women with at least 24 months cumulative
breastfeeding duration (RR=0.5, 95% CI: 0.3-0.8), there was a noted dose-response effect74
.
More research is needed to support this association.
2.4 Factors Associated with Breastfeeding Practices
The noted benefits of exclusive breastfeeding to six months have prompted many researchers to
investigate factors associated with breastfeeding initiation and exclusive breastfeeding as
recommended (i.e. to six months). Several studies, both nationally and internationally, have
examined factors that are associated with breastfeeding practices. The most recent population-
based survey in Canada is the Maternity Experiences Survey75
, a postcensal, cross-sectional
study of approximately 11% of women aged 15 and older who gave birth in Canada in late 2005
19
and early 200675
. To be eligible, the infant had to reside with the mother for at least one day per
month, be a singleton birth and be born in Canada. The survey was conducted when infants were
between the ages of five and fourteen months75
. Information on breastfeeding initiation, duration
and several both modifiable and non-modifiable risk factors were collected. Aboriginal identity
was not included in the questionnaire. Although the Aboriginal Children’s Survey, another
postcensal survey conducted in late 2006 amongst Aboriginal children in Canada and children
living in Canada’s north who were aged five or younger at the time of data collection76
did
collect information on Aboriginal identify, it did not collect information on the duration of
exclusive breastfeeding.
Knowledge of the determinants of breastfeeding practices can form the foundation for evidence
based health promotion programs to support breastfeeding. In this section, various maternal,
infant, social and contextual factors associated with breastfeeding practices in other studies are
examined.
2.4.1 Maternal Characteristics
2.4.1.1 Socioeconomic Position
Socioeconomic position (SEP) captures the social and economic factors that are related to an
individual’s position in society77
and is commonly measured using indicators from at least one of
four domains: income, wealth, occupation and educational attainment78,79
. The two most
common indicators of SEP used in breastfeeding research are income and education, although
occupational class is also common in research originating in the United Kingdom80
.
Most research in developed countries support the association between higher income and/or
education and breastfeeding as recommended (i.e. initiation15,80,81
and exclusive breastfeeding to
six months17,81,82
), particularly in crude models. However, the effect of income often disappears
after education is adjusted for17,81,83,84
, suggesting that more educated women may be better able
to locate and appraise the appropriate information regarding breast and formula feeding85
.
The remaining influence of education on breastfeeding practices, after adjusting for income, has
been noted in many populations (e.g. United States, Canadian), including Canadian Aboriginal
groups86
. Although education is likely the best indicator of SEP to use in research of
20
determinants of breastfeeding practices in developed countries, measures of income may be a
reasonable proxy variable in the absence of educational data.
Recent Canadian data from the Maternity Experiences Survey further support the relationship
between education and breastfeeding practices. The proportion of mothers with a high school
education who initiated breastfeeding was 82.5% while the proportion of mothers with a
university education who initiated breastfeeding was 95.9%15
. Moreover, the odds of practicing
exclusive breastfeeding to six months increased by 8% (95%CI: 1.05-1.12) for each additional
year of maternal education17
, after adjustment for other important covariates (e.g. maternal
employment status in the six months following delivery, age at first pregnancy, infant in neonatal
intensive care unit after birth).
In spite of the consistent relationship between higher SEP and breastfeeding as recommended in
developed countries, the relationship between social position and breastfeeding is often reversed
in developing countries. In this situation, women of higher social position may elect to bottle
feed their infants, as it conveys their higher status in society (i.e. they can afford formula,
nannies)87
. A discussion group with Inuit women (Appendix D) suggests that some Inuit women
may also feel that bottle feeding an infant communicates a higher status and position in society.
This suggests that the relationship between SEP and breastfeeding practices in Inuit Canadians
may not be consistent with that observed in other regions of the country.
2.4.1.2 Age
Studies have consistently demonstrated an effect of maternal age on breastfeeding practices,
wherein older mothers report a greater likelihood of both initiating81,88
and practicing any (and
exclusive) breastfeeding to six months81
. In a study of Cree Canadians, mothers who were aged
25 or younger were less likely to begin breastfeeding than older mothers (OR=0.41, 95% CI:
0.24-0.67)86
; the association between age and duration of breastfeeding was not investigated in
this study. In terms of breastfeeding duration, for each one year increase in age at first
pregnancy, mothers were 5% more likely (95% CI: 1.03-1.07) to practice exclusive breastfeeding
to six months in fully adjusted models using data from the Maternity Experiences Survey 17
.
21
2.4.1.3 Employment
The timing of return to work can impact a mother’s decision to both initiate89-91
and practice any
breastfeeding (or exclusive breastfeeding) at six months17,90,92,93
. Mothers who intend to return
to work shortly after giving birth are less likely to initiate breastfeeding compared to mothers
who do not expect to work. When the anticipated time period between giving birth and returning
to work is longer than six weeks, the effect of employment status on breastfeeding initiation
attenuates and is no longer significant in many population-based surveys89-91
.
Employment is also related to the duration of breastfeeding. Canadian research suggests that
mothers who were not employed in the previous six months before giving birth were 55% (95%
CI: 1.14-2.10) more likely to practice exclusive breastfeeding to six months17
. Parallel findings
have been published in the United States94
, Greece95
and Australia93
under various definitions of
breastfeeding (e.g. full breastfeeding92
, any breastfeeding93
). Women with full time employment
within six months of giving birth are more likely to wean before six months93,96,97
Employment policies can help to modify or mitigate the effects of employment on breastfeeding
practices. Longer maternity leave98
, part time employment96
or flexible work time99
and
employer sponsored programs100
have all been shown to either modify or reduce the effect of
employment on breastfeeding duration.
2.4.1.4 Ethnicity and Acculturation
The effect of specific ethnicities on breastfeeding practices vary depending on geographic
location. For example, population-based surveys suggest that black individuals were more likely
to initiate breastfeeding compared to white individuals when they lived in the United Kingdom
(Black African: OR=1.36, 95%CI: 1.31-1.42)101
, but less likely to initiate breastfeeding when
they lived in the United States (White: 74.3% (73.7%-74.9%); Black, non-Hispanic: 54.4%
(52.6%-56.2%)). Estimates from the United States were unadjusted for other important factors
(e.g. education) which may be driving these disparate findings. Alternatively, these data may
suggest a statistical interaction between ethnicity and contextual factors.
Population-based surveys conducted in Canada (2006) 17,102
, the United States (2002) 103
and the
United Kingdom (2000)101
, demonstrate that foreign born women were more likely to initiate
breastfeeding compared to other women. These surveys also suggest foreign born women are
22
more likely to practice exclusive breastfeeding to six months. Using data from the Maternity
Experiences Survey (2006), Al-Sahab et al noted a modest increase in the likelihood that foreign
born mothers would practice exclusive breastfeeding to six months in crude models (OR=1.36,
95% CI: 1.12-1.64), but this attenuated and was no longer significant once other important
factors (e.g. age, marital status, education) were accounted for (*adjusted OR not reported)17
.
Another factor that may influence infant feeding practices is cultural continuity. Cross-cultural
studies indicate that breastfeeding is a highly cultured behaviour – cultural attitudes about
breastfeeding and the breast (e.g. as a sexual organ vs. an organ with nutritional function) can
either directly (e.g. reduced vs. improved access to support systems) or indirectly (e.g. social
stigma and discrimination for exposing breasts in public vs. for failing to develop bonds between
mother and child) influence infant feeding practices 104-109
. Studies on Mexican, Cuban, and
Puerto Rican immigrants in the United States indicate that those with a stronger cultural
commitment to traditional values – or who are less acculturated - are more likely to initiate and
continue breastfeeding to six months than those with a stronger cultural commitment to
American values 110-112
.
Unadjusted data suggest that Canadian Aboriginals are less likely to initiate breastfeeding
compared to other Canadians, but may be more likely to continue any breastfeeding at six
months, particularly Inuit (See Section 2.2 for further details).113
2.4.1.5 Confidence, Intentions and Knowledge
Breastfeeding can be physically challenging for the new mother, causing pain and discomfort in
the nipples and breasts114
. Being confident in the ability to both breastfeed and to overcome
barriers to breastfeeding is strongly predictive of both initiation and duration of
breastfeeding115,116
. In 1992, O’Campo et al report that the risk of early weaning was three times
greater (95% CI: 1.39-6.70)115
in American mothers who scored lowest on a 10-item confidence
scale compared to those scoring highest. Similar findings have been reported elsewhere116
. In
response to the growing literature suggesting maternal confidence was a modifiable determinant
of breastfeeding behaviour, the 33-item Breastfeeding Self-Efficacy Scale (BSES) was
developed by Dennis and Faux117
, which applied Bandura’s Social Cognitive Theory to measure
maternal confidence in breastfeeding practices. This scale has since been reduced to 14-items
(BSES-SF).118
Studies from various countries (e.g. Canada119
, Australia120,121
, Japan122
) and
23
among various population subgroups (e.g. adolescent girls123
) using both the long and short
version of the BSES, consistently report that maternal confidence is related to breastfeeding
initiation and duration. Specifically, more confident mothers are more likely to initiate and
continue breastfeeding exclusively for the recommended six months.
Pre-pregnancy breastfeeding intentions also agree well with actual breastfeeding practices. In a
cohort of 300 pregnant women from Queensland Australia, those who intended to breastfeed for
six to 12 weeks or more than 12 weeks were 4.61 (95% CI: 2.04-10.39) and 9.75 (95% CI: 3.52-
26.96) times more likely, respectively, to report any breastfeeding to four months compared to
women who intended to breast feed for less than six weeks120
. Similar findings have been
reported elsewhere.115,124
Finally, positive attitudes and accurate knowledge about breastfeeding can improve
breastfeeding practices. The Iowa Infant Feeding Attitudes Scale (IIFAS)125
, a 17-item scale,
includes a series of questions about the mother’s attitudes towards breast and bottle feeding and
is the only scale that has demonstrated reasonable validity and reliability to measure mothers’
attitudes in this regard126
. Higher scores on the scale correspond to more positive attitudes about
breastfeeding and less favorable attitudes towards bottle-feeding. In a cohort of 382 women
from Perth, Australia each one unit increase on the IIFAS corresponded with a 0.97 times lower
risk (95% CI: 0.96-0.99) of discontinuing full breastfeeding at six months and 0.96 times lower
risk of discontinuing any breastfeeding at 12 months (95% CI: 0.94-0.98). Improving attitudes
towards breastfeeding can be accomplished with increased knowledge and education127
2.4.1.6 Smoking
Evidence from national and international population-based studies demonstrate mothers who
smoke during pregnancy are less likely to initiate and continue breastfeeding compared to other
mothers.102,128-131
A meta-analysis of 13 observational studies published between 1979 and 1997
found that infants of mothers who smoke are 71% more likely (95% CI: 1.34-2.18) to be weaned
by three months compared to infants of mothers who did not smoke.132
In Canada, maternal
smoking during pregnancy was associated with a lower likelihood of initiating breastfeeding
(RR=0.43, 95% CI: 0.28-0.66) and a greater hazard of weaning prior to six months (HR=1.56,
95% CI: 1.27-1.92) in a cohort of 858 infant-mother dyads from Southwestern Ontario.92
It is
unclear if this relationship persists in Aboriginal populations. Although this relationship was not
24
supported in a study of Australian Aboriginals133
, this may be related to statistical power,
inadequate adjustment of confounding factors or chance.
Biologically, smoking mothers produce less milk volume134
, which may explain at least some of
the differences in breastfeeding duration among smoking and non-smoking mothers. However,
there are likely psychological factors (e.g. maternal confidence, maternal self-efficacy)129
and
demographic factors (e.g. age, education) that are inadequately adjusted that account for at least
some of the relationship between smoking and breastfeeding practices135
.
2.4.1.7 Body Mass Index
A growing body of literature demonstrates the harmful effect of high maternal pre-pregnancy
body mass index on breastfeeding practices. A recent systematic review identified 16 studies
that examined the role of obesity on breastfeeding initiation and 18 that examined the role of
obesity on breastfeeding duration136
. The consensus across most studies is that obese women
report a lower likelihood of initiating breastfeeding and practicing any breastfeeding to six
months136
.
Differences in lactogenic capacity, wherein overweight mothers have a more difficult time
producing milk, may contribute to some of these observed variations.137
However, some
American research suggests the effect of obesity is modified by race: while black women who
were severely obese (BMI>35 kg/m2) report similar initiation (OR=1.08, 95% CI: 0.68-1.70) and
duration (HR=1.02, 95% CI: 0.69-1.50) practices to normal weight black women (BMI: 20
kg/m2 to 25 kg/m
2); white women who were very obese were less likely to initiate (OR=0.63,
95% CI: 0.42-0.94) and more likely to wean early (HR=1.89, 95% CI: 1.39-2.58), compared to
normal weight white women.138
. Modification of effect may be due to cultural variations in body
image, genetic differences in lactogenic capacity across races or other, presently unstudied
factors.
2.4.2 Infant Characteristics
2.4.2.1 Birth Weight
The indicator low birth weight is defined as weighing less than 2500 grams or 5.5 pounds at
delivery and captures two types of infants: those who are born prematurely (i.e. before 37 weeks)
25
and those who are carried to term but experience fetal growth restriction139
. Low birth weight
infants are less likely to begin and continue breastfeeding compared to other infants (i.e. at least
2500 grams or 5.5 pounds)102,140
, although it is unclear if there are different feeding patterns for
the two mechanisms by which low birth weight arises. This is related to the fact that most
breastfeeding studies either combine all low birth weight infants together, without consideration
for the separate pathways for which the outcome arose, or concentrate on only one subgroup of
low birth weight infants (e.g. pre-mature infants).141
Low birth weight infants are often separated from their mothers and placed in the neo-natal
intensive care unit142
. Many of these infants, particularly those born prematurely, are fed by oral
gavage since they have not yet developed a sucking reflex143,144
. Mothers need to establish and
maintain milk supply by pumping for as many as three months142,145
before the infant can feed
directly from the breast.
Among low birth weight babies, many factors predictive of beginning and continuing
breastfeeding are similar to those of healthy weight babies (e.g. socioeconomic position,
maternal age), although supportive hospital environments may be particularly important143,146
.
The role of hospitals in protecting and supporting breastfeeding is further elaborated upon in
section 2.4.4.1.
2.4.3 Supportive Environments
2.4.3.1 Supportive Partner and Extended Family
A systematic review of breastfeeding studies from Australia, Canada and the United States finds
that married women are consistently more likely to begin and continue breastfeeding to six
months compared to unmarried women147
. As the structure of relationships in Canada moves
away from the more traditional family unit of husband, wife and children towards more
common-law partnerships148
, more emphasis should be placed on exploring not just the presence
or absence of a husband or partner, but the type of support the partner provides to the mother. It
is likely not just the type of relationship that affects breastfeeding practices, but also the quality
of these relationships.
Sullivan et al report that women who experience more distress in their relationships are 1.75
times more likely to wean early compared to women with no relationship distress (95% CI: 0.95,
26
3.22)149
. On the opposing end of the spectrum, Scott et al finds the hazard of early weaning
before six months is lower amongst mothers whose partner preferred breastfeeding compared to
those who were indifferent or preferred formula feeding (HR=0.71; 95% CI: 0.55-0.91)124
.
Moreover, one randomized control trial of 280 partners report that teaching fathers to prevent
and manage lactation difficulties can increase the likelihood of breastfeeding to six months (25%
intervention group vs. 15% of control, p<0.05) 150
.
Other family members may also help shape breastfeeding practices. In certain ethnic groups
(e.g. Mexican Americans), grandmothers are an important source of breastfeeding support for
new mothers151
. Research on the historical breastfeeding practices of First Nations Canadians152
and Inuit Canadians also highlight the importance of the extended family on breastfeeding
practices, particularly from female figures (e.g. mother, sister, aunt).153,154
2.4.3.2 Supportive Health Care Professionals
The type of health care provider used in antenatal care can directly impact a new mother’s infant
feeding choices. Using data from an Ontario population-based registry, McDonald et al155
found that women who use midwives are approximately 4.5 times (95% CI: 4.16-4.85) more
likely to be breastfeeding at hospital discharge compared to women who use an obstetrician.
Parallel findings were documented in the population-based Maternity Experiences Survey
(2006)156
: breastfeeding initiation rates among mothers who use midwives was 97.1% (95% CI:
95.0%-99.2%) compared to 89.4% (95% CI: 88.4%-90.3%) for mothers who used obstetricians.
The prevalence of exclusive breastfeeding at six months in the Maternity Experiences Survey
was 30.1% (95% CI: 23.7%-36.5%) and 13.5% (95% CI: 12.3%-14.7%) for each group,
respectively156
.
Incomplete adjustment for possible confounding factors may explain at least some of the
variation in breastfeeding practices by health care provider (e.g. education level)157
158
; but
differences in the quality of care are also likely. A survey of 1983 women from Quebec,
approximately half of using an obstetrician for delivery and half using a midwife, found that
those using midwives for prenatal and antenatal care reported better quality and more frequent
prenatal care visits than those using physicians. Moreover, the quality of the provider-patient
relationship was perceived to be better and more personalized among mothers who relied largely
on midwives for their care 159
.
27
However, simple interventions by health care providers can help mitigate disparities in
breastfeeding practices between provider type. A recent Cochrane review found that health care
providers who offer extra breastfeeding support above the usual maternal care, such as reassuring
the mother and training or educating health care staff to support new mothers, all helped increase
breastfeeding duration160
.
Other types of health care providers can also improve breastfeeding outcomes. After
establishing that there were low breastfeeding initiation rates in a Manitoba First Nations
community, an educational and peer support strategy was implemented in an effort to improve
breastfeeding outcomes. Five years after the program’s implementation, breastfeeding initiation
increased from 38% to 60% (p<0.001), suggesting that breastfeeding outcomes can be improved
with prenatal instruction and postpartum peer counseling 161
.
2.4.4 Contextual Factors
2.4.4.1 Hospital Factors
The Baby Friendly Hospital Initiative (BFHI) was launched by the World Health Organization in
1991 and outlines ten steps that a hospital should take to promote, protect and support
breastfeeding162
. These steps are: 1) having a written breastfeeding policy that is routinely
communicated to all health care staff; 2) training all health care staff in the skills necessary to
implement this policy; 3) training all pregnant women about the benefits and management of
breastfeeding; 4) helping mothers initiate breastfeeding within the first 30 minutes after birth; 5)
showing mothers how to breastfeed and maintain lactation, even if they are separated from their
infants; 6) giving only breast milk to infants, unless otherwise medically indicated; 7) allowing
mothers and infants to remain together 24 hours of the day; 8) encouraging breastfeeding on
demand; 9) giving no artificial treats or pacifiers to the infant; 10) fostering the establishment of
breastfeeding support groups and referring mothers to them on discharge from the hospital or
clinic.162
Since the introduction of the BFHI, hospitals in Canada have substantially improved their
policies and practices towards breastfeeding. From 1993 to 2007, the proportion of hospitals
providing formula samples decreased, while the proportion of hospitals with written policies on
28
breastfeeding increased, as did the likelihood that new mothers would receive information on
breastfeeding support at discharge163
.
Research suggests that women giving birth in hospitals who adhere to BFHI recommendations
are more likely to initiate and continue breastfeeding. Using a longitudinal survey of 1085
women in the United States who were followed from an early prenatal state to 12 months post-
partum, DiGirolamo et al noted a clear dose-response relationship between the number of BFHI
practices in the delivery hospital, as reported by the mother, and the risk of stopping
breastfeeding at six weeks164
. Women who report birthing in hospitals with only two BFHI
practices were 4 times more likely to stop breastfeeding by six weeks (95% CI: 1.3-12.3), while
women who report birthing in hospitals with no BFHI practices were 7.7 times more likely to
stop breastfeeding by six weeks (95% CI: 2.3-25.8) compared to women who report birthing in
hospitals adhering to five BFHI practices.164
Similar benefits of the BFHI have been noted in
Manitoba, Canada where women who gave birth in hospitals that provided supplementation were
2.1 times more likely to wean early (95% CI: 1.03-4.36) compared to other mothers165
.
2.4.4.2 Neighbourhood Environment
There is a considerable amount of evidence that suggests where one lives influences
breastfeeding practices94,166
. As noted previously (Section 2.2), breastfeeding initiation and
duration varies substantially between Canadian provinces and territories, with mothers living in
Atlantic provinces being the least likely to adhere to recommendations, while mothers in
Western provinces and Northern Territories being the most likely to adhere to breastfeeding
recommendations15,17
. However, much of this research does not adjust for other potential
covariates (e.g. sociodemographic characteristics).
Other Australian research also implicates the role of environment on breastfeeding practices: at
discharge from hospital, those living in remote and very remote Australian communities report
initiation rates of 73.3% and 73.7% respectively, whereas those living in inner regional settings
report initiation rates of 82.1%167
. This finding may be particularly meaningful for Inuit
Canadian research, since all Inuit communities in Canada are remote or isolated (communities
are considered remote if they are 350 km away from the nearest acute care hospital but have year
round road access and isolated if they do not have year round road access but provide telephone
services for most homes and scheduled flights for travel needs of occupants)168
.
29
More recent epidemiologic research has started to consider the effect of neighbourhood on
breastfeeding practices. Two studies, one in Australia169
and the other in the United Kingdom170
report that breastfeeding initiation and duration are related to area-level deprivation (e.g. income
deprivation, employment deprivation, housing and service barriers). The Australian study used
data from three population-based surveys conducted in 1995, 2001, and 2005, and found a clear
dose-response effect of neighbourhood deprivation on breastfeeding initiation and duration -
those living in the most deprived neighbourhoods were the least likely to report initiating and
continuing breastfeeding for up to 12 months in all three time periods169
.
2.4.5 Factors Unique to an Inuit Setting
Many factors that are more unique to an Inuit setting may also influence infant feeding practices,
including household factors (i.e. crowding and food insecurity), health care access factors (e.g.
lack of primary care in the communities), and the high prevalence of adoption. There is no
Canadian research that has explored the effects of these factors on breastfeeding practices and
limited, if any, international research on these factors.
Food security exists when all people at all times have access to sufficient, safe and nutritious
foods for an active and healthy lifestyle171
. Inadequate support for hunters, community freezers
and hunting education programs as well as the expensive cost of commercial food products
purchased in the grocery store contribute to a high prevalence of food insecurity in Nunavut 172
.
The prevalence of household food insecurity is estimated anywhere from 50 % to 88% 173-175
,
much greater than the overall Canadian prevalence of approximately 7% to 15% 175,176
. It is
unclear if and how food security will impact infant feeding practices. Those mothers living in
food insecure households may be more likely to initiate and continue exclusive breastfeeding
because it is an affordable means to feed their infant, although this hypothesis was not supported
in the one study of food security and infant feeding practices conducted in Bangladesh 177
.
Household overcrowding exists when there are two or more persons per bedroom in the
household178
and is a major concern in Inuit communities179
. Homes with children are
disproportionately affected by crowding: 35.3% of all Inuit homes in Northern Canada are
crowded and 54.4% of Inuit homes with children are crowded179
. Overcrowding is even more
pervasive in Nunavut: 41% of all houses are crowded and 58.3% of houses with children are
crowded179
. Poor housing construction, the extreme cold climate and the high birth rate have all
30
contributed to the housing crisis in Northern communities. Only two studies have explored the
effect of crowding on breastfeeding practices and both use American data. The first reported that
living in compact quarters increases the intensity (e.g. more feedings per day) and duration of
breastfeeding180
. The other reported disparate findings: mothers living in crowded homes were
less likely to be breastfeeding at one month compared to mothers living in homes that were not
crowded (30% vs. 36%)33
. More research is needed, particularly in an Inuit context, to
understand if and how household crowding influences breastfeeding.
The high prevalence of adoption in Northern communities may place additional barriers on
breastfeeding initiation. Data suggest adoption rates in Inuit communities may range between
17% and 42% 181-183
. In a discussion group with Inuit women (Appendix D), I learned that Inuit
adoption often occurs between relatives, meaning that the birth mother can continue to have
contact with the child, providing an opportunity for breastfeeding. Yet, limited research suggests
that adopted children in the North are less likely to begin breastfeeding.184,185
. Moreover, no
research has explored the determinants of breastfeeding initiation in this unique population of
Inuit infants.
Several factors other than adoption may also influence infant feeding practices of Canadian Inuit.
One is the policy of evacuation. A shortage of skilled health care providers willing to work in
the North and the lack of local capacity has resulted in the almost universal evacuation of most
Inuit women for childbirth since the 1970’s 186
. Only women who live in one of three
communities (Iqaluit, Rankin Inlet, and Cambridge Bay) with traditional birthing centers or
hospitals are able to give birth within their communities. Inuit women living in communities
without hospitals or birthing centers are relocated to larger towns and cities (e.g. Ottawa,
Edmonton, Yellowknife) from 36 weeks gestation to about a week after giving birth 154,186,187
.
These women usually give birth alone, without their partners, other children, or family members
present, a departure from the traditional community centered Inuit birthing experience 154,187
.
Lack of instrumental and social support from family members, a direct result of evacuation
practices, has been cited by at least some Inuit women as a barrier to initiating breastfeeding 188
,
although a quantitative evaluation of the impact of evacuation on breastfeeding rates has not
been conducted.
31
Living in a community that advocates more primary prevention provides a critical context for the
acquisition of breastfeeding skills. Further, community participation in healthy baby initiatives,
such as the Canada Prenatal Nutrition Program (CPNP) may serve to increase initiation and
duration of breastfeeding. There were 15 communities in Nunavut (60%) participating in the
CPNP in 2005, a program which aims to support and promote breastfeeding among women 189
.
The CPNP is available to all Inuit communities, but communities must submit an application to
Health Canada before the program is implemented. In Nunavut communities, educational
materials about breastfeeding are distributed at participating health centers to all prenatal
women. Research in other populations suggests that providing educational material about
breastfeeding can positively impact breastfeeding initiation and duration8,190,191
, yet the effect of
CPNP on the prevalence of breastfeeding in Inuit communities has not been evaluated.
Given the unique barriers that Inuit Canadians face with access to health care, especially
accessing culturally compatible breastfeeding messages, it seems likely that beginning
breastfeeding requires some degree of personal resiliency. Resiliency, or positive adaptation
despite adversity192
, is an emerging theme in Aboriginal research192,193
. Higher degrees of
cultural resiliency have been associated with positive psychosocial outcomes in Aboriginal
Canadians194,195
. There are many different components to cultural resiliency and a recent review
of the literature suggests that participation in traditional activities serves as an indicator of
resilience in Aboriginal communities193
. Exploring how and if cultural resiliency affects
breastfeeding practices can provide insight into a potentially modifiable determinant to improve
breastfeeding practices in the North.
2.5 Motivation for Research Project
Despite some recent improvements, disparities between Inuit and other Canadians in infant
health indicators remain substantial196
. Table 1 presents the rates of nine health related events
that are more prevalent in Inuit compared to other Canadians, such as lower respiratory tract
infections197
, obesity198
and sudden infant death syndrome (SIDs)199
. Barriers with accessing
timely health care may further exacerbate health concerns, resulting in more severe illness when
presenting at medical clinics. Practicing exclusive breastfeeding as recommended by Health
Canada3 and the World Health Organization
6 is a cost-effective
200,201 way to ameliorate at least
some differences in health.
32
Although research on determinants of infant feeding practices has been conducted in the general
Canadian population17,92,155
and in First Nations communities152,161
, there have been few
evaluations of individual determinants18,184
and no evaluations of community level determinants
of infant feeding in Inuit populations. This seems inappropriate since the determinants of infant
feeding, and consequently, the probable types of successful intervention programs, can vary
across cultures. Furthermore, few (if any) studies have examined the effect of various factors
that may be important in an Inuit context (e.g. food security; birthing evacuation; adoption) on
breastfeeding practices. One particularly interesting factor is cultural resiliency. Given the
additional barriers that Inuit face around the birthing process (i.e. evacuation), individual and
cultural resiliency may be important determinants of infant feeding practices.
In 2004, Health Canada noted the need “to identify population-based, cost-effective strategies to
support the implementation of the recommendation [i.e. exclusive breastfeeding to 6 months]” 2.
The most effective strategies will be community based and evidence-driven 202
. The results of
this study provide a better understanding of the factors that are driving breastfeeding practices in
Inuit Canadian women so that appropriate population-based interventions that are rooted in
theory and evidence can be implemented for primary prevention.
33
Table 1: Prevalence of Specific Health Events for which breastfeeding is protective against in
Inuit Populations and Comparator Populations
Health Event Rate in Comparator Population Rate in Inuit Population
Infants
Sudden Infant Death
Syndrome (SIDS)
Quebec (English speakers): 0.4/1000203
Quebec (French speakers): 0.5/1000203
Canada wide: 0.51/1000199
Quebec: 6/1000 for Inuit Canadians203
Nunavut: 6.08/1000199
Baffin: 5.76/1000199
Kivalliq: 5.68/1000199
Kitikmeot: 7.80/1000199
Odds of SIDS is 12x higher in Inuit
compared to other Canadians199
Lower Respiratory
Tract Infection
(LRTI)
LRTI (0-11 months): Aboriginal
community (Saskatchewan): 200/1000
person years204
Hospital admission LRTI (<1 year):
Manitoba: 150 per 1000 person years205
LRTI: (age <6 months: Baffin) 484 per
1000 183
; 12% severe and required
intubation183
LRTI (0-11 months: Nunavik):
1385/1000 person years197
Hospital admission LRTI (<1 year):
305/1000 person years; 2x greater than
Manitoba low income community197
Gastrointestinal
Infection
Winnipeg: Norovirus Infection rates:
0.007/infant/year206
Winnipeg: Rotovirus infection:
0.36/child/year206
Alaska hospitalizations from diarrheal
disease (total population): 192 per
10,000207
Eskimo Point (Inuit): Norovirus
infection rates: 0.087/infant/year206
Eskimo Point: Rotovirus infection:
1.07/child/year206
Alaska hospitalizations from diarrheal
disease (American Indian & Eskimo
Population): 275 per 10,000207
Otitis Media No Canadian estimates
Prevalence of Chronic otitis media is
less than 1% in United Sates208
Chronic otitis media: 46% Alaskan
Inuit, 31% Canadian Inuit, 12%
Greenlandic Inuit208
.
78% any otitis media in Kuujjaraapik209
21% otitis media with effusion in
Sanililuaq210
16.7% chronic otitis media in Inukjuak
children41
Atopic Dermatitis Canada: approximately 12% (Aged 13
to 14 years)211
1/3 of patients visiting medical clinic in
Baffin Island have severe atopic
dermatitis212
Prevalence in Baffin medical clinic:
22%213
Table 1 continued next page
34
Health Event Rate in Comparator Population Rate in Inuit Population
Body Mass Index
(BMI)
BMI (World Health Organization
growth standards):
Canada (ages 5-11 years):
32.8% overweight or obese214
Girls (ages 5-11 years):
25.9% overweight or obese214
Boys (ages 5-11 years):
39.7% overweight or obese214
International Obesity Task Force:
Canada (ages 5-11 years):
14.9% overweight214
7.9% obese214
BMI (Centers for Disease Control
growth standards):
Nunavut (age 3-5 years):
50.8% overweight or obese63
Girls (age 3-5 years):
45.2% overweight or obese63
Boys (age 3-5 years):
57.2% overweight or obese63
International Obesity Task Force:
Nunavut (ages 3-5 years):
42.6% overweight198
26.9% obese198
Nunavik (ages 3-5):
42.5% overweight198
12.6% obese198
Type II Diabetes Canada (ages 35-39 years): 2.4%215
(ages 60-64 years): 15.7%215
Historically low: prevalence increasing
over time216
Nunavut (total Inuit population):
5.1%217
(Inuit: <50 years): 1.9%217
(Inuit: ≥50 years): 12.2%217
Nunavik (total Inuit population):
4.7%218
(males): 2.7%218
(females) 6.6%218
Mothers
Fertility Rate 1.5 children/woman219
Proportion of teenaged mothers (15-
19): 1.3%219
(Inuit): 3.4 children/woman219
Proportion of teenaged mothers (15-
19): 9.0%219
Type II Diabetes Under “infant” section Under “infant” section
35
2.6 Section References
1. World Health Organization. The World Health Organization's infant feeding
recommendation. 2001;
http://www.who.int/nutrition/topics/infantfeeding_recommendation/en/index.html.
Accessed June 20, 2009.
2. Health Canada. Exclusive Breastfeeding Duration: 2004 Health Canada
Recommendation. 2004; http://www.hc-sc.gc.ca/fn-an/alt_formats/hpfb-
dgpsa/pdf/nutrition/excl_bf_dur-dur_am_excl-eng.pdf. Accessed June 3, 2012.
3. Infant Feeding Joint Working Group. Nutrition for healthy term infants:
recommendations from birth to six months: a joint statement of Health Canada, Canadian
Paediatric Society, Dietitians of Canada, and Breastfeeding Committee for Canada. 2012;
http://www.hc-sc.gc.ca/fn-an/nutrition/infant-nourisson/recom/index-eng.php#a3.
4. American Academy of Pediatrics, Work Group on Breastfeeding. Breastfeeding and the
use of human milk. Pediatrics. 1997;100:1035 - 1039.
5. American Academy of Pediatrics. Breastfeeding and the Use of Human Milk. Pediatrics.
2012;129(3):e827-e841.
6. World Health Organization. Indicators for assessing infant and young child feeding
practices. 2008; http://whqlibdoc.who.int/publications/2008/9789241596664_eng.pdf.
Accessed December 12, 2012.
7. Kramer M, Kakuma R. Optimal Duration of exclusive breastfeeding. Cochrane Database
of Systematic Reviews. 2002;1.
8. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention
Trial (PROBIT): A Randomized Trial in the Republic of Belarus. JAMA.
2001;285(4):413-420.
9. Kramer MS, Guo T, Platt RW, et al. Infant growth and health outcomes associated with 3
compared with 6 months of exclusive breastfeeding. Am J Clin Nutr. 2003;78(2):291-
295.
10. Ministry of Health. Galactosemia.
http://www.health.gov.on.ca/english/providers/program/child/screening/pdf/fs_galact.pdf.
Accessed December 10, 2012.
11. Canadian Pediatric Society. Maternal infectious diseases, antimicrobial therapy or
immunizations: very few contraindications to breastfeeding. 2012;
http://www.cps.ca/documents/position/maternal-infectious-diseases-breastfeeding.
Accessed December 23, 2012.
12. Nathoo T, Ostry A. The one best way? Breastfeeding history, politics and policy in
Canada. Waterloo, ON: Wilfred Laurier University Press; 2009.
36
13. Nathoo T, A. O. The one best way? Breastfeeding history, politics and policy in Canada.
Waterloo, ON: Wilfred Laurier University Press; 2009.
14. Health Canada. Trends in breastfeeding practices in Canada (2001-2009-2010). 2012;
http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/prenatal/trends-tendances-
eng.php. Accessed Dec 8, 2012.
15. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
16. Statistics Canada. Low income cut-offs. 2009;
http://www.statcan.gc.ca/pub/75f0002m/2009002/s2-eng.htm. Accessed February 7,
2013.
17. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
18. Asuri S, Ryan A, Arbour L. Early Inuit Child Health in Canada: Breastfeeding among
Inuit in Canada. 2011; https://www.itk.ca/sites/default/files/2011-Report-Breastfeeding-
among-Inuit-in-Canada.pdf. Accessed December 12, 2012.
19. Ip S, Chung M, Raman G, Trikalinos TA, Lau J. A summary of the Agency for
Healthcare Research and Quality's evidence report on breastfeeding in developed
countries. Breastfeeding Medicine. 2009;4(S1):17-30.
20. Cindy-Lee Dennis. Breastfeeding Initiation and Duration: A 1990-2000 Literature
Review. Journal of Obstetric, Gynecologic, and Neonatal Nursing. 2002;31(1):12-32.
21. Chien P, Howie P. Breast milk and the risk of opportunistic infection in infancy in
industrialized and non-industrialized settings. Advances in nutritional research.
2001;10:69.
22. Bachrach V, Schwarz E, Bachrach L. Breastfeeding and the risk of hospitalization for
respiratory disease in infancy: A meta-analysis. Archives of pediatrics & adolescent
medicine. 2003;157(3):237-243.
23. Gdalevich M, Mimouni D, David M, Mimouni M. Breast-feeding and the onset of atopic
dermatitis in childhood: A systematic review and meta-analysis of prospective studies.
Journal of the American Academy of Dermatology. 2001;45(4):520-527.
24. Owen CG, Martin RM, Whincup PH, Smith GD, Cook DG. Does breastfeeding influence
risk of type 2 diabetes in later life? A quantitative analysis of published evidence. Am J
Clin Nutr. 2006;84(5):1043-1054.
25. Harder T, Bergmann R, Kallischnigg G, Plagemann A. Duration of Breastfeeding and
Risk of Overweight: A Meta-Analysis. American Journal of Epidemiology.
2005;162(5):397-403.
37
26. Anderson JW, Johnstone BM, Remley DT. Breast-feeding and cognitive development: a
meta-analysis. Am J Clin Nutr. 1999;70(4):525-535.
27. Der G, Batty GD, Deary IJ. Effect of breast feeding on intelligence in children:
prospective study, sibling pairs analysis, and meta-analysis. BMJ. 2006;333(7575):945.
28. Kramer MS, Aboud F, Mironova E, et al. Breastfeeding and Child Cognitive
Development: New Evidence From a Large Randomized Trial. Arch Gen Psychiatry.
2008;65(5):578-584.
29. Kennedy KI, Visness CM. Contraceptive efficacy of lactational amenorrhoea. The
Lancet. 1992;339(8787):227-230.
30. Bernier MO, Plu-Bureau G, Bossard N, Ayzac L, Thalabard JC. Breastfeeding and risk of
breast cancer: a meta-analysis of published studies. Human Reproduction Update.
2000;6(4):374-386.
31. Stuebe AM, Rich-Edwards JW, Willett WC, Manson JAE, Michels KB. Duration of
lactation and incidence of type 2 diabetes. JAMA: the journal of the American Medical
Association. 2005;294(20):2601-2610.
32. Stuebe AM, Michels KB, Willett WC, Manson JE, Rexrode K, Rich-Edwards JW.
Duration of lactation and incidence of myocardial infarction in middle to late adulthood.
American Journal of Obstetrics and Gynecology. 2009;200(2):138.e131-138.e138.
33. Raisler J, Alexander C, O'Campo P. Breast-feeding and infant illness: a dose-response
relationship? Am J Public Health. 1999;89(1):25-30.
34. Scariati PD, Grummer-Strawn LM, Fein SB. A longitudinal analysis of infant morbidity
and the extent of breastfeeding in the United States. Pediatrics. 1997;99(6):e5-e5.
35. Duijts L, Ramadhani MK, Moll HA. Breastfeeding protects against infectious diseases
during infancy in industrialized countries. A systematic review. Maternal & Child
Nutrition. 2009;5(3):199-210.
36. Oddy WH, Kickett-Tucker C, De Maio J, et al. The association of infant feeding with
parent-reported infections and hospitalisations in the West Australian Aboriginal Child
Health Survey. Australian and New Zealand Journal of Public Health. 2008;32(3):207-
215.
37. Bulkow LR, Singleton RJ, Karron RA, Harrison LH. Risk Factors for Severe Respiratory
Syncytial Virus Infection Among Alaska Native Children. Pediatrics. 2002;109(2):210-
216.
38. Koch A, Mølbak K, Homøe P, et al. Risk Factors for Acute Respiratory Tract Infections
in Young Greenlandic Children. American Journal of Epidemiology. 2003;158(4):374-
384.
38
39. Hamosh M. Bioactive factors in human milk. Pediatric Clinics of North America.
2001;48(1):69-86.
40. MacIntyre EA, Karr CJ, Koehoorn M, et al. Otitis media incidence and risk factors in a
population-based birth cohort. Paediatrics & child health. 2010;15(7):437.
41. Ayukawa H, Bruneau S, Proulx J, Macarthur J, Baxter J. Otitis media and hearing loss
among 12-16 year old Inuit of Inukjuak, Quebec, Canada. Int J Circumpolar Health.
2004;63(S2):312-314.
42. Baxter JD. Otitis media in Inuit children in the Eastern Canadian Arctic-an overview-
1968 to date. Int J Pediatr Otorhinolaryngol. 1999;49(S1):165-168.
43. Kramer A, McCullough D. The prevalence of otitis media with effusion among Inuit
children. International Journal of Circumpolar Health. 1998;57(s1):265-267.
44. McVea KLSP, Turner PD, Peppler DK. The Role of Breastfeeding in Sudden Infant
Death Syndrome. Journal of Human Lactation. 2000;16(1):13-20.
45. Hauck FR, Thompson JMD, Tanabe KO, Moon RY, Vennemann MM. Breastfeeding and
Reduced Risk of Sudden Infant Death Syndrome: A Meta-analysis. Pediatrics. 2011.
46. Gdalevich M, Mimouni D, Mimouni M. Breast-feeding and the risk of bronchial asthma
in childhood: A systematic review with meta-analysis of prospective studies [doi: DOI:
10.1067/mpd.2001.117006]2001.
47. Brew BK, Allen CW, Toelle BG, Marks GB. Systematic review and meta‐analysis
investigating breast feeding and childhood wheezing illness. Paediatric and Perinatal
Epidemiology. 2011;25(6):507-518.
48. Yang YW, Tsai CL, Lu CY. Exclusive breastfeeding and incident atopic dermatitis in
childhood: a systematic review and meta-analysis of prospective cohort studies. British
Journal of Dermatology. 2009;161(2):373-383.
49. Stabell Benn C, Wohlfahrt J, Aaby P, et al. Breastfeeding and risk of atopic dermatitis, by
parental history of allergy, during the first 18 months of life. American Journal of
Epidemiology. 2004;160(3):217-223.
50. Kwan ML, Buffler PA, Abrams B, Kiley VA. Breastfeeding and the risk of childhood
leukemia: a meta-analysis. Public Health Reports. 2004;119(6):521.
51. Jain A, Concato J, Leventhal JM. How good is the evidence linking breastfeeding and
intelligence? Pediatrics. 2002;109(6):1044-1053.
52. Drane DL, Logemann JA. A critical evaluation of the evidence on the association
between type of infant feeding and cognitive development. Paediatric and Perinatal
Epidemiology. 2000;14(4):349-356.
39
53. Klement E, Cohen RV, Boxman J, Joseph A, Reif S. Breastfeeding and risk of
inflammatory bowel disease: a systematic review with meta-analysis. The American
Journal of Clinical Nutrition. 2004;80(5):1342-1352.
54. Akobeng AK, Ramanan AV, Buchan I, Heller RF. Effect of breast feeding on risk of
coeliac disease: a systematic review and meta-analysis of observational studies. Archives
of Disease in Childhood. 2006;91(1):39-43.
55. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
56. Taylor JS, Kacmar JE, Nothnagle M, Lawrence RA. A Systematic Review of the
Literature Associating Breastfeeding with Type 2 Diabetes and Gestational Diabetes.
Journal of the American College of Nutrition. 2005;24(5):320-326.
57. Young TK, Martens PJ, Taback SP, et al. Type 2 diabetes mellitus in children: Prenatal
and early infancy risk factors among native canadians. Archives of pediatrics &
adolescent medicine. 2002;156(7):651-655.
58. Martin RM, Ben-Shlomo Y, Gunnell D, Elwood P, Yarnell JWG, Smith GD. Breast
feeding and cardiovascular disease risk factors, incidence, and mortality: the Caerphilly
study. Journal of Epidemiology and Community Health. 2005;59(2):121-129.
59. Pettitt DJ, Forman MR, Hanson RL, Knowler WC, Bennett PH. Breastfeeding and
incidence of non-insulin-dependent diabetes mellitus in Pima Indians. Lancet.
1997;350(9072):166.
60. Health Canada. Diabetes among Aboriginal People in Canada: The evidence. 2000;
http://www.hc-sc.gc.ca/fniah-spnia/alt_formats/fnihb-
dgspni/pdf/pubs/diabete/2001_evidence_faits-eng.pdf. Accessed December 1, 2012.
61. Horta B, Bahl R, Martines JC, Victora CG. Evidence on the long term effects of
breastfeeding: a systematic review and meta-analysis. In: Department of Child and
Adolscent Health and Development, ed. Switzerland: World Health Organization; 2007.
62. Kramer MS, Matush L, Vanilovich I, et al. Effects of prolonged and exclusive
breastfeeding on child height, weight, adiposity, and blood pressure at age 6.5 y:
evidence from a large randomized trial. The American Journal of Clinical Nutrition.
2007;86(6):1717-1721.
63. Galloway T, Young TK, Egeland GM. Emerging obesity among preschool-aged
Canadian Inuit children: results from the Nunavut Inuit Child Health Survey.
International Journal of Circumpolar Health. 2010;69(2).
64. Findlay L, Janz T. Health of First Nations children living off reserve and Metis children
younger than age 6. Health Reports. 2012;23(1).
65. Kohen D, Brehaut J, Garner R, et al. Conceptualizing childhood health problems using
survey data: a comparison of key indicators. BMC Pediatrics. 2007;7(1):40.
40
66. Heinig MJ, Nommsen-Rivers LA, Peerson JM, Dewey KG. Factors related to duration of
postpartum amenorrhoea among USA women with prolonged lactation. Journal of
biosocial science. 1994;26:517-517.
67. World Health Organization. The World Health Organization multinational study of
breast-feeding and lactational amenorrhea. II. Factors associated with the length of
amenorrhea. Fertility and sterility. 1998;70(3):461-471.
68. Eastwood JG, Phung H, Barnett B. Postnatal Depression and Socio-Demographic Risk:
Factors Associated with Edinburgh Depression Scale Scores in a Metropolitan Area of
New South Wales, Australia. Australian and New Zealand Journal of Psychiatry.
2011;45(12):1040-1046.
69. Davey HL, Tough SC, Adair CE, Benzies KM. Risk factors for sub-clinical and major
postpartum depression among a community cohort of Canadian women. Maternal and
Child Health Journal. 2011;15(7):866-875.
70. Möller T, Olsson H, Ranstam J. Breast cancer and breastfeeding: collaborative reanalysis
of individual data from 47 epidemiological studies in 30 countries, including 50302
women with breast cancer and 96973 women without the disease. Lancet.
2002;360(9328):187-195.
71. Ram KT, Bobby P, Hailpern SM, et al. Duration of lactation is associated with lower
prevalence of the metabolic syndrome in midlife—SWAN, the study of women’s health
across the nation. American Journal of Obstetrics and Gynecology.
2008;198(3):268.e261-268.e266.
72. Kjos SL, Henry O, Lee RM, Buchanan TA, Mishell Jr DR. The effect of lactation on
glucose and lipid metabolism in women with recent gestational diabetes. Obstetrics and
Gynecology. 1993;82(3):451.
73. Schwarz EB, Ray RM, Stuebe AM, et al. Duration of lactation and risk factors for
maternal cardiovascular disease. Obstetrics and Gynecology. 2009;113(5):974-982.
74. Karlson EW, Mandl LA, Hankinson SE, Grodstein F. Do breast‐feeding and other
reproductive factors influence future risk of rheumatoid arthritis?: Results from the
Nurses' Health Study. Arthritis & Rheumatism. 2004;50(11):3458-3467.
75. Statistics Canada. Definitions, data sources and methods: Maternity Experiences Survey
(MES). 2008;
http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=5019&lang=en
&db=imdb&adm=8&dis=2. Accessed December 12, 2012.
76. Statistics Canada. Definitions, data sources and methods: Aboriginal Children's Survey
(ACS). 2008;
http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=5108&lang=en
&db=imdb&adm=8&dis=2. Accessed December 15, 2012.
41
77. Krieger N, Williams DR, Moss NE. Measuring social class in US public health research:
concepts, methodologies, and guidelines. Annual Review of Public Health.
1997;18(1):341-378.
78. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Smith GD. Indicators of socioeconomic
position (part 1). Journal of Epidemiology and Community Health. 2006;60(1):7-12.
79. Lynch J, Kaplan G. Socioeconomic position. In: Berkman LF, Kawachi I, eds. Social
epidemiology. New York: Oxford University Press; 2000.
80. Power C, Matthews S. Origins of health inequalities in a national population sample. The
Lancet. 1997;350(9091):1584-1589.
81. van Rossem L, Oenema A, Steegers EAP, et al. Are Starting and Continuing
Breastfeeding Related to Educational Background? The Generation R Study. Pediatrics.
2009;123(6):e1017-e1027.
82. Braun MLG, Giugliani ERJ, Soares MEM, Giugliani C, De Oliveira AP, Danelon CMM.
Evaluation of the impact of the baby-friendly hospital initiative on rates of breastfeeding.
Journal Information. 2003;93(8).
83. Heck K, Braveman P, Cubbin C, Chavez G, Kiely J. Socioeconomic status and
breastfeeding initiation among California mothers. Public Health Rep. 2006;121(1):51-
59.
84. Dubois L, Girard M. Social Determinants of initiation, duration and exclusivity of
breastfeeding at the population level: the results of the Longitudinal Study of Child
Development in Quebec (ELDEQ 1998-2002). Can J Public Health. 2003;94(3):300-305.
85. Blakely T, Subramanian SV. Multilevel Studies. In: Oakes J, Kaufman J, eds. Methods in
Social Epidemiology. San Francisco: John Wiley & Sons, Inc; 2006.
86. Black R, Godwin M, Ponka D. Breastfeeding among the Ontario James Bay Cree: a
retrospective study. Can J Public Health. 2008;99(2):98-101.
87. Rogers IS, Emmett PM, Golding J. The incidence and duration of breast feeding. Early
human development. 1997;49:S45-S74.
88. Merewood A, Brooks D, Bauchner H, MacAuley L, Mehta SD. Maternal Birthplace and
Breastfeeding Initiation Among Term and Preterm Infants: A Statewide Assessment for
Massachusetts. Pediatrics. 2006;118(4):e1048-e1054.
89. Kimbro R. On-the-Job Moms: Work and Breastfeeding Initiation and Duration for a
Sample of Low-Income Women. Maternal and Child Health Journal. 2006;10(1):19-26.
90. Fein SB, Roe B. The effect of work status on initiation and duration of breast-feeding.
Am J Public Health. 1998;88(7):1042-1046.
42
91. Noble S, The AST. Maternal employment and the initiation of breastfeeding. Acta
Pædiatrica. 2001;90(4):423-428.
92. Clifford TJ, Campbell MK, Speechley KN, Gorodzinsky F. Factors Influencing Full
Breastfeeding in a Southwestern Ontario Community: Assessments at 1 Week and at 6
Months Postpartum. J Hum Lact. 2006;22(3):292-304.
93. Cooklin AR, Donath SM, Amir LH. Maternal employment and breastfeeding: results
from the longitudinal study of Australian children. Acta Pædiatrica. 2008;97(5):620-623.
94. Li R, Zhao Z, Mokdad A, Barker L, Grummer-Strawn L. Prevalence of breastfeeding in
the United States: the 2001 National Immunization Survey. Pediatrics. 2003;111(5 Part
2):1198-1201.
95. Ladomenou F, Kafatos A, Galanakis E. Risk factors related to intention to breastfeed,
early weaning and suboptimal duration of breastfeeding. Acta Paediatrica.
2007;96(10):1441-1444.
96. Ryan AS, Zhou W, Arensberg MB. The effect of employment status on breastfeeding in
the United States. Women's health issues : official publication of the Jacobs Institute of
Women's Health. 2006;16(5):243-251.
97. Guendelman S, Kosa JL, Pearl M, Graham S, Goodman J, Kharrazi M. Juggling work
and breastfeeding: effects of maternity leave and occupational characteristics. Pediatrics.
2009;123(1):e38-e46.
98. Baker M, Milligan K. Maternal employment, breastfeeding, and health: Evidence from
maternity leave mandates. Journal of Health Economics. 2008;27(4):871-887.
99. Hawkins SS, Griffiths LJ, Dezateux C. The impact of maternal employment on breast-
feeding duration in the UK Millennium Cohort Study. Public Health Nutrition.
2007;10(09):891-896.
100. Oritz J, McGilligan K, Kelly P. Duration of breast milk expression among working
mothers enrolled in an employer-sponsored lactation program. Pediatric Nursing.
2004;30(2):111-119.
101. Griffiths LJ, Tate AR, Dezateux C, The Millennium Cohort Study Child Health Group.
The contribution of parental and community ethnicity to breastfeeding practices:
evidence from the Millennium Cohort Study. International Journal of Epidemiology.
2005;34(6):1378-1386.
102. Simard I, O'Brien H, Beaudoin A, et al. Factors influencing the initiation and duration of
breastfeeding among low-income women followed by the Canada prenatal nutrition
program in 4 regions of Quebec. J Hum Lact. 2005;21(3):327-337.
103. Celi A, Rich-Edwards J, Richardson M, Kleinman K, Gillman M. Immigration,
race/ethnicity, and social and economic factors as predictors of breastfeeding initiation.
Archives of pediatrics & adolescent medicine. 2005;159(3):255-260.
43
104. Murphy E. Breast is best: infant feeding decisions and maternal deviance. Sociology of
health and illness. 1999;21(2):187-208.
105. Dettwyler KA. Breastfeeding and weaning in Mali: Cultural context and hard data. Social
Science & Medicine. 1987;24(8):633-644.
106. Bohler E, Ingstad B. The struggle of weaning: factors determining breastfeeding duration
in East Bhutan. Social Science & Medicine. 1996;43(12):1805-1815.
107. Bryant CA. The impact of kin, friend and neighbor networks on infant feeding practices.
Cuban, Puerto Rican and Anglo families in Florida. Soc Sci Med. 1982;16(20):1757-
1765.
108. Morse JM. Cross-cultural nursing: a unique contribution to medical anthropology. Med
Anthropol. Nov 1989;12(1):1-5.
109. Van Esterik P. Contemporary trends in infant feeding research. Annual Review of
Anthropology. 2002;31:257-278.
110. Kimbro R, Lynch S, McLanahan S. The Influence of Acculturation on Breastfeeding
Initiation and Duration for Mexican-Americans. Population Research and Policy Review.
2008;27(2):183-199.
111. Rassin D, Markides K, Baranowski T, Richardson C, Mikrut W, Bee D. Acculturation
and the initiation of breastfeeding. J Clin Epidemiol. 1994;47:739 - 746.
112. Gibson MV, Diaz VA, Mainous AG, Geesey ME. Prevalence of Breastfeeding and
Acculturation in Hispanics: Results from NHANES 1999–2000 Study. Birth.
2005;32(2):93-98.
113. Guevremont A, Kohen D. Inuit in Canada: Findings from the Aboriginal Peoples' Survey
- Survey of Living Conditions in the Arctic. The Health of Inuit Children: Fact Sheet.
2007; http://www.statcan.gc.ca/pub/89-627-x/89-627-x2007002-eng.pdf. Accessed June
3, 2012.
114. Gatrell CJ. Secrets and lies: breastfeeding and professional paid work. Social Science &
Medicine. 2007;65(2):393-404.
115. O'Campo P, Faden RR, Gielen AC, Wang MC. Prenatal Factors Associated with
Breastfeeding Duration: Recommendations for Prenatal Interventions. Birth.
1992;19(4):195-201.
116. Buxton KE, Gielen AC, Faden RR, Brown CH. Women intending to breastfeed:
predictors of early infant feeding experiences. American journal of preventive medicine.
1991;7(2):101-106.
117. Dennis CL, Faux S. Development and psychometric testing of the Breastfeeding Self‐Efficacy Scale. Research in Nursing & Health. 1999;22(5):399-409.
44
118. Dennis CL. The Breastfeeding Self-Efficacy Scale: Psychometric Assessment of the
Short Form. Journal of Obstetric, Gynecologic, & Neonatal Nursing. 2003;32(6):734-
744.
119. Semenic S, Loiselle C, Gottlieb L. Predictors of the duration of exclusive breastfeeding
among first-time mothers. Research in Nursing & Health. 2008;31(5):428-441.
120. Blyth RJ, Creedy DK, Dennis CL, et al. Breastfeeding duration in an Australian
population: the influence of modifiable antenatal factors. Journal of Human Lactation.
2004;20(1):30-38.
121. Baghurst P, Pincombe J, Peat B, Henderson A, Reddin E, Antoniou G. Breast feeding
self-efficacy and other determinants of the duration of breast feeding in a cohort of first-
time mothers in Adelaide, Australia. Midwifery. 2007;23(4):382.
122. Otsuka K, Dennis C-L, Tatsuoka H, Jimba M. The Relationship Between Breastfeeding
Self-Efficacy and Perceived Insufficient Milk Among Japanese Mothers. Journal of
Obstetric, Gynecologic, & Neonatal Nursing. 2008;37(5):546-555.
123. Mossman M, Heaman M, Dennis C-L, Morris M. The Influence of Adolescent Mothers'
Breastfeeding Confidence and Attitudes on Breastfeeding Initiation and Duration.
Journal of Human Lactation. 2008;24(3):268-277.
124. Scott JA, Binns CW, Oddy WH, Graham KI. Predictors of breastfeeding duration:
evidence from a cohort study. Pediatrics. 2006;117(4):e646-e655.
125. Scott JA, Shaker I, Reid M. Parental Attitudes Toward Breastfeeding: Their Association
with Feeding Outcome at Hospital Discharge. Birth. 2004;31(2):125-131.
126. Chambers J, Alder E, Hoddinott P, McInnes R. A systematic review of measures
assessing mothers' knowledge, attitudes, confidence and satisfaction towards
breastfeeding. Breastfeeding review. 2007;15(3):17-25.
127. Martens PJ. The effect of breastfeeding education on adolescent beliefs and attitudes: a
randomized school intervention in the Canadian Ojibwa community of Sagkeeng.
Journal of Human Lactation. 2001;17(3):245-255.
128. Leung GM, Lam T-H, Ho L-M. Breast‐Feeding and Its Relation to Smoking and Mode of
Delivery. Obstetrics & Gynecology. 2002;99(5, Part 1):785-794.
129. Donath SM, Amir LH, The Alspac Study Team. The relationship between maternal
smoking and breastfeeding duration after adjustment for maternal infant feeding
intention. Acta Pædiatrica. 2004;93(11):1514-1518.
130. Di Napoli A, Di Lallo D, Pezzotti P, Forastiere F, Porta D. Effects of parental smoking
and level of education on initiation and duration of breastfeeding. Acta Pædiatrica.
2006;95(6):678-685.
45
131. Weiser TM, Lin M, Garikapaty V, Feyerharm RW, Bensyl DM, Zhu B-P. Association of
Maternal Smoking Status With Breastfeeding Practices: Missouri, 2005. Pediatrics.
2009;124(6):1603-1610.
132. Horta BL, Kramer MS, Platt RW. Maternal smoking and the risk of early weaning: a
meta-analysis. American Journal of Public Health. 2001;91(2):304-307.
133. Gilchrist D, Woods B, Binns CW, Scotth JA, Gracey M. Aboriginal mothers,
breastfeeding and smoking. Australian and New Zealand Journal of Public Health.
2004;28(3):225-228.
134. Vio F, Salazar G, Infante C. Smoking during pregnancy and lactation and its effects on
breast-milk volume. The American Journal of Clinical Nutrition. 1991;54(6):1011-1016.
135. Amir LH, Donath SM. Does Maternal Smoking Have a Negative Physiological Effect on
Breastfeeding? The Epidemiological Evidence. Birth. 2002;29(2):112-123.
136. Turcksin R, Bel S, Galjaard S, Devlieger R. Maternal obesity and breastfeeding intention,
initiation, intensity and duration: a systematic review. Maternal & Child Nutrition.
2012:10.1111/j.1740-8709.2012.00439.x.
137. Amir LH, Donath S. A systematic review of maternal obesity and breastfeeding intention,
initiation and duration. BMC Pregnancy and Childbirth. 2007;7(1):9.
138. Liu J, Smith MG, Dobre MA, Ferguson JE. Maternal Obesity and Breast-feeding
Practices Among White and Black Women. Obesity. 2009;18(1):175-182.
139. Goldenberg RL, Culhane JF. Low birth weight in the United States. The American
Journal of Clinical Nutrition. 2007;85(2):584S-590S.
140. Flacking R, Nyqvist KH, Ewald U, Wallin L. Long-term duration of breastfeeding in
Swedish low birth weight infants. Journal of Human Lactation. 2003;19(2):157-165.
141. Flacking R, Nyqvist KH, Ewald U. Effects of socioeconomic status on breastfeeding
duration in mothers of preterm and term infants. The European Journal of Public Health.
2007;17(6):579-584.
142. Callen J, Pinelli J. A review of the literature examining the benefits and challenges,
incidence and duration, and barriers to breastfeeding in preterm infants. Adv Neonatal
Care. 2005;5(2):72-88.
143. Hill PD, Hanson KS, Mefford AL. Mothers of Low Birthweight Infants: Breastfeeding
Patterns and Problems. Journal of Human Lactation. 1994;10(3):169-176.
144. Verronen P. Breast Feeding of Low Birthweight Infants. Acta Pædiatrica.
1985;74(4):495-499.
46
145. Callen J, Pinelli J, Atkinson S, Saigal S. Qualitative analysis of barriers to breastfeeding
in very-low-birthweight infants in the hospital and postdischarge. Adv Neonatal Care.
2005;5(2):93-103.
146. Furman L, Minich N, Hack M. Correlates of Lactation in Mothers of Very Low Birth
Weight Infants. Pediatrics. April 1, 2002 2002;109(4):e57.
147. Callen J, Pinelli J. Incidence and duration of breastfeeding for term infants in Canada,
United States, Europe and Australia: a literature review. Birth. 2004;31(4):285-292.
148. Human resources and skills development Canada. Canadians in Context - households and
families. 2012; http://www4.hrsdc.gc.ca/[email protected]?iid=37. Accessed
December 26, 2012.
149. Sullivan ML, Leathers SJ, Kelley MA. Family characteristics associated with duration of
breastfeeding during early infancy among primiparas. Journal of Human Lactation.
2004;20(2):196-205.
150. Pisacane A, Continisio GI, Aldinucci M, D'Amora S, Continisio P. A Controlled Trial of
the Father's Role in Breastfeeding Promotion. Pediatrics. 2005;116(4):e494-e498.
151. Baranowski T, Bee DE, Rassin DK, et al. Social support, social influence, ethnicity and
the breastfeeding decision. Social Science & Medicine. 1983;17(21):1599-1611.
152. Martens P. Prenatal infant feeding intent and perceived social support for breastfeeding in
Manitoba first nations communities: a role for health care providers. International
Journal of Circumpolar Health. 1997;56(4):104.
153. Archibald L. Teenage pregnancies in Inuit communities: Issues and prespectives. 2004;
http://www.pauktuutit.ca/pdf/publications/pauktuutit/TeenPregnancy_e.pdf. Accessed
April, 2010.
154. Qinuajuak L. Inuit birth traditions. Midwifery Today and Childbirth Education.
1996;40(1):56.
155. McDonald SD, Pullenayegum E, Chapman B, et al. Prevalence and Predictors of
Exclusive Breastfeeding at Hospital Discharge. Obstetrics & Gynecology.
2012;119(6):1171.
156. O’Brien B, Chalmers B, Fell D, Heaman M, Darling EK, Herbert P. The Experience of
Pregnancy and Birth with Midwives: Results from the Canadian Maternity Experiences
Survey. Birth. 2011;38(3):207-215.
157. Wilson KL, Sirois FM. Birth attendant choice and satisfaction with antenatal care: the
role of birth philosophy, relational style, and health self‐efficacy. Journal of Reproductive
and Infant Psychology. 2010;28(1):69-83.
158. Howell-White S. Choosing a birth attendant: The influence of a woman's childbirth
definition. Social Science & Medicine. 1997;45(6):925-936.
47
159. De Koninck M, Blais R, Joubert P, Gagnon C. Comparing women’s assessment of
midwifery and medical care in Quebec, Canada. Journal of Midwifery & Women's
Health. 2001;46(2):60-67.
160. Renfrew MJ, McCormick FM, Wade A, Quinn B, Dowswell T. Support for healthy
breastfeeding mothers with healthy term babies. Cochrane Database Syst Rev. 2012;5.
161. Martens PJ. Increasing breastfeeding initiation and duration at a community level: an
evaluation of Sagkeeng First Nation’s community health nurse and peer counselor
programs. Journal of Human Lactation. 2002;18(3):236-246.
162. World Health Organization, UNICEF, Wellstart International. Baby-Friendly Hospital
Initiative: Revised, updated and expaned for intergrated care. 2009;
http://whqlibdoc.who.int/publications/2009/9789241594967_eng.pdf. Accessed
December 15, 2012.
163. Levitt C, Hanvey L, Kaczorowski J, Chalmers B, Heaman M, Bartholomew S.
Breastfeeding policies and practices in Canadian hospitals: comparing 1993 with 2007.
Birth. 2011;38(3):228-237.
164. DiGirolamo AM, Grummer-Strawn LM, Fein S. Maternity Care Practices: Implications
for Breastfeeding. Birth. 2001;28(2):94-100.
165. Martens PJP, Phillips SJS, Cheang MSM, Rosolowich VV. How Baby-Friendly are
Manitoba hospitals? The Provincial Infant Feeding Study. Breastfeeding Promotion
Steering Committee of Manitoba. Canadian journal of public health. Revue canadienne
de santé publique. 2000;91(1):51-57.
166. Hannan A, Li R, Benton-Davis S, Grummer-Strawn L. Regional Variation in Public
Opinion About Breastfeeding in the United States. J Hum Lact. 2005;21(3):284-288.
167. Taylor LK, Lim K, Neville SE. Newborn feeding practices at the time of discharge from
hospital in NSW in 2007: a descriptive study. New South Wales Public Health Bulletin.
2010;20(12):177-181.
168. Public Health Agency of Canada. Considerations for definitions of 'remote' and 'isolated'
in the context of Pandemic (H1N1) 20092009.
169. Amir LH, Donath SM. Socioeconomic status and rates of breastfeeding in Australia:
evidence from three recent national health surveys. Med J Aust. 2008;189(5):254-256.
170. Brown AE, Raynor P, Benton D, Lee MD. Indices of Multiple Deprivation predict
breastfeeding duration in England and Wales. The European Journal of Public Health.
2010;20(2):231-235.
171. Food and Agriculture Organization of the United Nations. Rome Declaration on World
Food Summit Plan of Action, Rome Italy. 1996;
http://www.fao.org/docrep/003/w3613e/w3613e00.HTM. Accessed April 8, 2013.
48
172. Chan H, Fediuk K, Hamilton S, et al. Food security in Nunavut, Canada: barriers and
recommendations. International Journal of Circumpolar Health. 2006;65(5):416-431.
173. Lawn J, Harvey D, Dialogos Educational Consultants Inc. Nutrition and Food Security in
Kugaaruk, Nunavut: baseline survey for the Food Mail pilot project. 2003; http://dsp-
psd.pwgsc.gc.ca/Collection/R2-265-2003E.pdf. Accessed March 3, 2009.
174. Ford JD, Berrang-Ford L. Food security in Igloolik, Nunavut: an exploratory study. Polar
Record. 2009;45(03):225-236.
175. Ledrou I, Gervais J. Food Insecurity. Health Reports. 2005;16(3):47-51.
176. Health Canada. Canadian Community Health Survey, Cycle 2.2, Nutrition (2004):
Income-related household food security in Canada. Ottawa, 2007.
177. Saha KK, Frongillo EA, Alam DS, Arifeen SE, Persson LA, Rasmussen KM. Household
Food Security Is Associated with Infant Feeding Practices in Rural Bangladesh. J. Nutr.
2008;138(7):1383-1390.
178. Wanyeki I, Olson S, Brassard P, et al. Dwellings, crowding, and tuberculosis in Montreal.
Social Science & Medicine. 2006;63(2):501-511.
179. Minich K, Saudny H, Lennie C, et al. Inuit housing and homelessness: results from the
International Polar Year Inuit Health Survey 2007–2008. International Journal of
Circumpolar Health. 2011;70(5).
180. Stuart MacAdam P, Dettwyler KA. Breastfeeding: biocultural perspectives. New York:
Transaction Publishers; 1995.
181. Jenkins A, Gyorkos T, Joseph L, et al. Risk factors for hospitalization and infection and
Canadian Inuit infants over the first year of life - a pilot study. Int J Circumpolar Health.
2004;63(1):61-70.
182. Muckle G, Ayotte P, Dewailly E, Jacobson S, Jacobson J. Determinants of
polychlorinated biphenyls and methylmercury exposure in Inuit women of childbearing
age. Environ Health Perspect. 2001;109(9):7.
183. Banerji A, Bell A, Mills EL, et al. Lower respiratory tract infections in Inuit infants on
Baffin Island. CMAJ. 2001;164(13):1847-1850.
184. Dufour R. Breast feeding and adoption among the Inuit population of Northern Quebec.
Can J Publ Health. 1984;75(2):137-140.
185. Stewart P, Steckle J. Breastfeeding among Canadian Indians on-reserve and women in
the Yukon and NWT. Canadian journal of public health. Revue canadienne de santé
publique. 1987;78(4):255.
186. Tedford S, O'Neil J, van Wagner V. Examining Midwifery-based options to improve
continuity of maternity care in remote Nunavut communities. 2005;
49
http://www.naho.ca/inuit/midwifery/documents/2006-03-
09Tetfordarticlefinalshortform.pdf. Accessed March 15, 2009.
187. Douglas V. Childbirth among the Canadian Inuit: a review of the clinical and cultural
literature. International Journal of Circumpolar Health. 2006;64(2):117-133.
188. Chamberlain M, Barclay K. Psychosocial costs of transfering indigenous women from
their community for birth. Midwifery. 2000;16:116-122.
189. Thomas T, Greeley G, Guyot M, Loubert K, Banfield W. Inuit Wellness Programs in
Nunavut: 2004-2005. 2006; http://pubs.aina.ucalgary.ca/health/61935.pdf. Accessed May
24, 2012.
190. Long DG, Funk-Archuleta MA, Geiger CJ, Mozar AJ, Heins JN. Peer Counselor
Program Increases Breastfeeding Rates in Utah Native American WIC Population. J Hum
Lact. 1995;11(4):279-284.
191. Cohen R, Mrtek M. The impact of two corporate lactation programs on the incidence and
duration of breast-feeding by employed mothers. American Journal of Health Promotion.
1994;8(6):436-441.
192. Fleming J, Ledogar R. Resillience, an evolving concept: a review of literature relevant to
Aboriginal research. Pimatisiwin. 2008;6(2):7-23.
193. Kirmayer L, Sehdev M, Whitely R, Dandeneau S, Isaacs C. Community resilience:
models, metaphors and measures. Journal of Aboriginal Health. 2009;November:62-117.
194. Kirmayer L, Simpson C, Cargo M. Healing traditions: culture, community and mental
health promotion with Canadian Aboriginal peoples. Australasian Psychiatry.
2003;11(s1):S15-S23.
195. Chandler MJ, Lalonde C. Cultural Continuity as a Hedge against Suicide in Canada's
First Nations. Transcultural Psychiatry. 1998;35(2):191-219.
196. Luo Z-C, Senecal S, Simonet F, Guimond E, Penney C, Wilkins R. Birth outcomes in the
Inuit-inhabited areas of Canada. CMAJ. 2010;182(3):235-242.
197. Dallaire F, Dewailly E, Vézina C, Bruneau S, Ayotte P. Portrait of outpatient visits and
hospitalizations for acute infections in Nunavik preschool children. Canadian Journal of
Public Health. 2006;97(5):362-368.
198. Galloway T, Niclasen BVL, Muckle G, Young K, Egeland GM. Growth measures among
preschool-age Inuit children living in Canada and Greenland. Scandinavian Journal of
Public Health. 2012;40(8):712-717.
199. Collins SA, Surmala P, Osborne G, et al. Causes and risk factors for infant mortality in
Nunavut, Canada 1999-2011. BMC Pediatrics. 2012;12(1):190.
50
200. Weimer JP. The economic benefits of breastfeeding: A review and analysis: United States
Department of Agriculture, Economic Research Service;2001.
201. Bartick M, Reinhold A. The burden of suboptimal breastfeeding in the United States: a
pediatric cost analysis. Pediatrics. 2010;125(5):e1048-e1056.
202. Kay F. Applications of social data analysis to program planning and evaluation. Journal
of Evaluation and Program Planning. 1978;1(1):65-78.
203. Luo Z-C, Wilkins R, Platt RW, Kramer MS, for the F, Infant Health Study Group of the
Canadian Perinatal Surveillance S. Risks of adverse pregnancy outcomes among Inuit
and North American Indian women in Quebec, 1985–97. Paediatric and Perinatal
Epidemiology. 2004;18(1):40-50.
204. Harris SB, Glazier R, Eng K, McMurray L. Disease patterns among Canadian aboriginal
children. Study in a remote rural setting. Canadian Family Physician. 1998;44:1869.
205. Kozyrskyj AL, Hildes-Ripstein GE. Assessing health status in Manitoba children: acute
and chronic conditions. Canadian journal of public health. Revue canadienne de santé
publique. 2002;93:S44.
206. Gurwith M, Wenman W, Gurwith D, Brunton J, Feltham S, Greenberg H. Diarrhea
among Infants and Young Children in Canada: A Longitudinal Study in Three Northern
Communities. The Journal of Infectious Diseases. 1983;147(4):685-692.
207. Holman RC, Parashar UD, Clarke MJ, Kaufman SF, Glass RI. Trends in Diarrhea-
associated Hospitalizations Among American Indian and Alaska Native Children, 1980–
1995. Pediatrics. 1999;103(1):e11.
208. Acuin J. Chronic suppirative otits media: burden of illness and management options.
2004; http://www.who.int/pbd/deafness/activities/hearing_care/otitis_media.pdf.
Accessed Jan 8, 2008.
209. Julien G, Baxter J, Crago M, Illecki H, Therien F. Chronic otitis media and hearing
deficit among native children of Kuujjuaraapik (Northern Quebec): a pilot project.
Canadian Journal of Public Health. 1987;78(1):57-61.
210. Kramer A, McCullough D. The prevalence of otitis media with effusion among Inuit
children. International Journal of Circumpolar Health. 1998;57:265-267.
211. Beasley R. Worldwide variation in prevalence of symptoms of asthma, allergic
rhinoconjunctivitis, and atopic eczema: ISAAC. The Lancet. 1998;351(9111):1225-1232.
212. Jackson R. Whither dermatology: Thoughts generated after holding a dermatology clinic
on Baffin Island. Archives of Dermatology. 1998;134(11):1347-1348.
213. Jackson R. My eleven years treating skin disease on Baffin Island, Nunavut Canada -
what I saw, what I did. Circumpolar Health Supplements. 2010;7:508-510.
51
214. Roberts KC, Shields M, de Groh M, Aziz A, Gilbert JA. Overweight and obesity in
children and adolescents: results from the 2009 to 2011 Canadian Health Measures
Survey. Health reports/Statistics Canada, Canadian Centre for Health Information=
Rapports sur la santé/Statistique Canada, Centre canadien d'information sur la santé.
2012;23(3):37.
215. Public Health Agency of Canada. Report from the National Diabetes Surveillance
System: Diabetes in Canada, 2009. 2009; http://www.phac-
aspc.gc.ca/publicat/2009/ndssdic-snsddac-09/pdf/report-2009-eng.pdf. Accessed
December 28, 2012.
216. Young T, Bjerregaard P, eds. Health Transitions in Arctic Populations. Toronto:
University of Toronto Press; 2008.
217. Egeland GM, Cao Z, Young TK. Hypertriglyceridemic-waist phenotype and glucose
intolerance among Canadian Inuit: the International Polar Year Inuit Health Survey for
Adults 2007–2008. Canadian Medical Association Journal. 2011;183(9):E553-E558.
218. Chateau-Degat ML, Dewailly É, Louchini R, et al. Cardiovascular burden and related risk
factors among Nunavik (Quebec) Inuit: insights from baseline findings in the circumpolar
Inuit health in transition cohort study. Canadian Journal of Cardiology. 2010;26(6):e190-
e196.
219. O’Donnell V, Wallace S. First Nations, Metis and Inuit women. 2012;
http://www.statcan.gc.ca/pub/89-503-x/2010001/article/11442-eng.htm#a14. Accessed
Dec 5, 2012.
52
Chapter 3 :Study Methods
3 Methods Overview
This chapter describes the overall study methods that were used in this thesis. Additional
manuscript-specific methods are provided in the relevant chapter specific to various thesis
objectives.
3.1 Data Sources
Data for this research come primarily from the IPY (International Polar Year) Inuit Health
Survey and the Nunavut Inuit Child Health Survey, two cross-sectional, population-based
surveys conducted concurrently in the Inuit Nunangat (i.e. Inuit regions of Canada) in 2007 and
2008 (Principal Investigator: Grace Egeland; Steering committee members: Lynda Gunn,
Laakkuluk Williamson, Isaac Sobel, Geraldine Osborne and Kue Young). The surveys were
developed to better understand the factors contributing to Inuit health and the Inuit spirit of
thriving in the face of social and environmental changes1. Self-reported information on many
different aspects of health such as: health behaviours, health conditions, household and
neighbourhood characteristics, and dietary habits, as well as some clinical measures (e.g. blood
pressure, bone density) were included in the surveys. The IPY Inuit Health Survey was
distributed to non-pregnant adults aged 18 years and older and the Nunavut Inuit Child Survey
was distributed to caregivers of pre-school aged children (ages 3 to 5 years). The two surveys
were linked together using identifiers unique to each household.
These surveys were conducted during the International Polar Year 2007-20082 , an international
program of science and research in Arctic and Antarctic regions of the world and were funded
jointly by the Canadian Institutes of Health Research and the Government of Canada Program
for the International Polar Year.
53
3.2 Setting and Design of the IPY Inuit Health Survey and the Nunavut Inuit Child Health Survey
The IPY Inuit Health Survey is a cross-sectional, population-based survey conducted in Northern
Canada in the summer of 2007 and 2008. All of the 36 communities from three out of four Inuit
settlement regions (i.e. Nunangat) were included in the IPY Inuit Health Survey (see Table 2)1.
These communities came from the Inuvialuit Settlement Region, Northwest Territories;
Nunavut; and Nunatsiavut, Northern Labrador. There were no communities from Nunavik,
Quebec.
The Nunavut Inuit Child Health Survey is a cross-sectional, population-based survey conducted
in the Canadian territory of Nunavut in the summer and fall months of 2007 and 2008. The
territory has 26 communities contained within three administrative regions: Baffin, Kitikmeot
and Kivalliq3. Sixteen of these communities were selected, with certainty, to participate in the
Nunavut Inuit Child Health Survey and are listed in Table 2 (sampling procedures fully
described in section 3.3.1).
3.2.1 Demographic Characteristics of Regions Included in the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey
The 2006 census found that of the 50,475 Canadians self-identifying as Inuit4, 60% (n=29,910)
live in one of the three Inuit settlement regions sampled for the IPY Inuit Health Survey: 6%
(n=3,115) in the Inuvialuit Settlement Region; 49% (n=24,635) in Nunavut; and 4% (n=2,160) in
Nunatsiavut4. Moreover, 84% of the total population of Nunavut (n=29,325), 89% of the total
population of Nunatsiavut (n=2,415), and 54% of the total population of the Inuvialuit
Settlement Region (n=5,705) self-identify as Inuit4.
The use of Inuit languages at home varies by Nunangat. Nunavut has the highest proportion of
households speaking primarily Inuit languages at home (64%), followed by Nunatsiavut (7%)
and the Inuvialuit Settlement Region (4%)5.
Inuit women have a 2.3 times greater fertility rate compared to other Canadian women (3.4
children vs. 1.5 children) and the highest fertility rate of any Aboriginal Canadian identity group
(Metis: 2.2 children; First Nations: 2.9 children).6 Moreover, the territory of Nunavut has the
youngest population of any Canadian province or territory: approximately one third (34%) of
54
residents are under the age of 14 and the median age of residents is 23 years7 . In comparison,
fewer than 20% (17%) of Canadians are under the age of 14 and the median age of Canadian
residents is 39 years.
Table 2: Communities participating in the IPY Inuit Health Survey and Nunavut Inuit
Child Health Survey
Community Region Participate in
Nunavut Inuit Child
Health Survey
Arctic Bay Baffin, Nunavut No
Aklavik Inuvialuit Settlement Region, NWT No
Arviat Kivalliq, Nunavut Yes
Baker Laker Kivalliq, Nunavut Yes
Cambridge Bay Kitikmeot, Nunavut Yes
Cape Dorset Baffin, Nunavut No
Chesterfield Inlet Kivalliq, Nunavut Yes
Clyde River Baffin, Nunavut Yes
Coral Habour Kivalliq, Nunavut Yes
Gjoa Haven Kitikmeot, Nunavut No
Grise Fiord Baffin, Nunavut No
Hall Beach Baffin, Nunavut No
Hopedale Nunatsiavut, Labrador No
Igloolik Baffin, Nunavut Yes
Inuvik Inuvialuit Settlement Region, NWT No
Iqaluit Baffin, Nunavut Yes
Kimmirut Baffin, Nunavut Yes
Kugaaruk Kitikmeot, Nunavut Yes
Kugluktuk Kitikmeot, Nunavut Yes
Makkovik Nunatsiavut, Labrador No
Nain Nunatsiavut, Labrador No
Pangnirtung Baffin, Nunavut Yes
Paulatuk Inuvialuit Settlement Region, NWT No
Table 2 continued on following page
55
Community Region Participate in
Nunavut Inuit
Child Health
Survey
Pond Inlet Baffin, Nunavut Yes
Postville Nunatsiavut, Labrador No
Qikiqtarjuaq Baffin, Nunavut No
Rankin Inlet Kivalliq, Nunavut Yes
Repulse Bay Kivalliq, Nunavut No
Resolute Bay Baffin, Nunavut No
Rigolet Nunatsiavut, Labrador No
Sachs Harbour Inuvialuit Settlement Region, NWT No
Sanikiluaq Baffin, Nunavut Yes
Taloyoak Kitikmeot, Nunavut No
Tuktoyaktuk Inuvialuit Settlement Region, NWT No
Ulukhaktok Inuvialuit Settlement Region, NWT No
Whale Cove Kivalliq, Nunavut Yes
56
3.3 Methods for the Nunavut Inuit Child Health Survey
3.3.1 Participants and recruitment strategies
Inuit children aged 3 to 5 years, who were living in one of the 16 participating communities at
the time of data collection were the target population for the Nunavut Inuit Child Health Survey.
Inclusion and exclusion criteria were specific for each study objective and can be found in the
appropriate manuscript (Objective 1 and 2: Chapter 4; Objective 3 and 4: Chapter 5; Objective 5
and 6: Chapter 6).
Eligible children were recruited in one of two ways. In the first way, children of caregivers who
had taken part in the Inuit Health Survey (described in more detail in section 3.4.1) and who met
the eligibility criteria were invited to participate. All children who met the age eligibility criteria
in a household were extended an invitation to participate (less than 5% of this sample). This
approach recruited only 36% of the desired sample; as such, another sampling method was
needed. To supplement this initial sample, other children were randomly selected from
Community Medical Centre patient rosters. Because there is only one medical facility in each
community and children would be expected to visit these facilities for vaccinations and other
health care needs, patient rosters are expected to have close to, if not 100% coverage. This high
coverage rate reduces any bias related to sampling and enables a true random selection of study
participants.
The survey aimed to sample approximately 20% of eligible children per community, although a
higher sampling fraction was applied in smaller communities (e.g. Chesterfield Inlet) and a lower
sampling fraction was applied in larger communities (e.g. Iqaluit). Census data (2001), adjusted
for projected population growth, were used to estimate the number of children to sample per
community to achieve the desired sampling fractions.
Invitations to participate were extended to caregivers first by telephone. If this approach was
unsuccessful, members of the research team visited the home (time permitting); pamphlets were
left for caregivers with the research team’s contact information when no one was available. At
least three attempts to contact caregivers were made.
57
3.3.2 Data collection strategies
Members of the research team travelled to participating communities from August to November
2007 and August to September 2008 to administer questionnaires to caregivers (complete data
collection schedules in Appendix A). In 2007, most interviews were conducted by a bilingual
Inuk nurse, although others were conducted by one of two English speaking research assistants
from the Centre for Indigenous Peoples Nutrition and Environment at McGill University. In
2008, a non-Inuk northern nurse with previous experience working in Nunavut administered
most of the surveys. The interviewer administered questionnaire was usually conducted at the
Community Health Centre, but other community facilities were used if needed.
Interviews were conducted with the child’s caregiver. There were no established criteria for
caregiver, but approximately 80% of interviews were conducted with a biological parent (either
mother or father) and approximately 20% of interviews were conducted with an adoptive parent.
After receiving informed consent2, six survey modules were administered to the primary
caregiver in the following order: a home demographic identification chart; a home-based
questionnaire; a 24 hour dietary recall; a food frequency questionnaire; a medicine and
supplement use questionnaire; and the child individual questionnaire. When a member of the
household had participated in the IPY Inuit Health Survey, neither the demographic nor home
based questionnaires were repeated. After the caregivers had completed the six questionnaires, a
trained nurse took clinical measurements from the child. Clinical information was verified
through chart extraction.
A summary of information in the child individual questionnaire and medicine and supplement
use questionnaires is provided in Table 3. Other survey modules are as described for the IPY
Inuit Health Survey (see Table 4, section 3.4.2).
2 Consent forms were distributed to all eligible caregivers. Caregivers also had the opportunity to view an
informational DVD which contained a spoken version of the consent form. Eligible caregivers were required to sign
the consent form.
58
Table 3: Information collected in select survey modules: Nunavut Inuit Child Health
Survey
Survey Module Information Collected
Child Individual Questionnaire Demographic characteristics of child (e.g. age, gender)
Infant birth weight
Breastfeeding practices (initiation, duration of exclusive
breastfeeding, duration of any breastfeeding)
Maternal behaviours during pregnancy (e.g. drinking)
Health status and co-morbidities (e.g. ear infections, physical
activity, injuries)
Health care utilization (e.g. hospitalizations)
Medicine and Supplement Use Current prescription and over the counter drug use
Current vitamin and nutritional supplement use
59
3.3.3 Study variables: Measurement and classification
The Nunavut Inuit Child Health Survey (child component) is found is Appendix B.
3.3.3.1 Outcome variables
Breastfeeding initiation (Objective 1, 2 and 5)
Breastfeeding initiation was measured in response to the closed-ended question (Yes/No/Don’t
Know), “Was your child ever breastfed or fed breast milk” and was treated as a dichotomous
variable in all analyses, excluding those who were missing or responded ‘don’t know’. The
validity of this instrument is not known; however it is a widely used instrument and similar to
those on questionnaire modules from the World Health Organization8
Biological mothers’ recall of breastfeeding initiation at fifteen years postpartum has been shown
to agree well with medical records (85%) and has a sensitivity of 82% and a specificity of 93%9 .
Moreover, the two year test re-test reliability was 89% among American nurses, more than 50
years (on average) after giving birth 10
. In this study, the average child age was 4.4 years11
, a
much shorter duration to recall breastfeeding initiation.
The validity of breastfeeding initiation recall among biological fathers or other caregivers has not
been evaluated. Moreover, while a systematic review of breastfeeding initiation recall in a
variety of populations further corroborates mothers’ recall of breastfeeding initiation tends to
agree with ‘gold-standard’ measures12
, there may be important differences in recall in Inuit
Canadian populations.
Exclusive breastfeeding duration (Objective 3, 4 and 6)
Exclusive breastfeeding duration was measured in response to the open-ended question, “How
many months was your child exclusively breast fed or exclusively fed breast milk”.
The validity of exclusive breastfeeding duration varies depending on the type of infant feeding
transition: switching from breast milk to formula, and/or introducing complementary foods into
the diet while continuing to feed the infant breast milk.
60
In terms of the first type of transition -the movement from breast milk to formula feeding - it has
been suggested that approximately 58% of mothers recall within one month the transition from
breast milk to formula feeding13,14
. These data come from two studies, one following infant
feeding transitions prospectively and assessing maternal recall within the first year of an infant’s
life13
, and the other relying on retrospective recall of initiation up to ten years postpartum14
and
comparing data to medical records.
In terms of the second type of transition – the introduction of complementary or solid foods into
the infant’s diet – data suggests that anywhere from 65%9 to 88%
13 of mothers accurately recall
this transition within one month. Accuracy of recall is related to the amount of time passing
between the feeding transition and the date of recall. Specifically, higher accuracy (88%) is
noted when timing of introduction of complementary foods is solicited less than one year after
the feeding transition, and lower accuracy (65%) is noted when information was solicited up to
14 and 15 years later.
The World Health Organization recommends that exclusive breastfeeding duration be measured
by 24 hour dietary recall, wherein mothers are interviewed when the infant is between the ages
of 0 and 6 months and the previous days’ diet is recalled8. Infants who only receive breast milk
(drops and syrups of vitamins and medicines are permitted) are classified as exclusive
breastfeeding at that time; infants who receive anything else (e.g. water, fruit juice, infant
formula) are classified as not being exclusively breastfed. This method, however, tends to over-
estimate the proportion of infants who are exclusively breastfed. A longitudinal study following
508 infant-mother dyads in Sweden compared the infant’s diet over 14 days with one 24 hour
dietary recall measure found that the 24 hour recall measure tended to over-estimate the duration
of exclusive breastfeeding, as much as 40% in the first four months of the infant’s life15
.
Research in California also highlights difficulties with exclusive breastfeeding measurement16
.
This research compared exclusive breastfeeding rates in California at hospital discharge as
measured by the Nationwide Immunization Survey with rates as reported on the California
Newborn Screen. The Nationwide Immunization Survey is a population based survey of
children aged 19 to 35 months which relies on maternal recall of duration of exclusive
breastfeeding16
. The California Newborn Screen is a program implemented by the California
Department of Public Health that requires hospital staff to collect infant feeding information
61
from birth until the infant is discharged from hospital. Exclusive breastfeeding at hospital
discharge (defined as four days after birth) was estimated at 60.4% using the Nationwide
Immunization Survey and 41.6% using the California Newborn Screen.16
.
The measurement of exclusive breastfeeding in the Nunavut Inuit Child Health Survey did
depend on maternal recall; however, the method used to measure the practices was fairly
consistent with approaches on other population based Canadian Surveys (i.e. the Canadian
Health Measures Survey)17
.
The duration of exclusive breastfeeding was used in several different ways in the analyses. In
some analyses (i.e. descriptive), breastfeeding duration was treated as a continuous variable. In
other analyses, breastfeeding duration was treated as a categorical variable (i.e. exclusive
breastfeeding as recommended vs. not; short, long or as recommended duration of exclusive
breastfeeding).
3.3.3.2 Potential predictor variables
Primary caregiver of child
The primary caregiver for the child was measured in response to the partially closed-end
question “What is the respondent’s relationship to the child”. In all analyses, this was treated as
a binary variable: either the child was raised by their biological parent or was not. For some
analyses, only responses from the biological mother were of interest. When the respondent
indicated they were a biological parent and also indicated they were female, in response to the
closed-ended question “what is your gender”, they were classified as the biological mother.
Language
Language was measured in response to the partially closed-ended question “What is the
predominate [sic] language spoken in your home”. The five response options were: Inuktitut,
Inuinnaqtun, English, French, and other, where the other category invited respondents to identify
their predominant language. Respondents who indicated they spoke an Inuit dialect at home (i.e.
Inuktitut, Inuinnaqtun, or identified an Inuit dialect in the ‘other’ category) were classified as
speaking predominantly Inuit dialects at home in all analyses, while other respondents were
classified as not speaking Inuit dialects at home.
62
The language one speaks at home has been used previously among various ethnic and cultural
groups, including First Nations Canadians, as an indicator of their level of engagement with
traditional culture18,19
.
Household overcrowding
Household overcrowding was operationalized from two questions on the Nunavut Inuit Child
Health Survey. The first question was open-ended and asked how many bedrooms there were in
the respondents’ home. The second question was also open-ended and asked about the number
of inhabitants in each household. The number of bedrooms in the home was divided by the
number of inhabitants in each home to get a ratio of the number of rooms to number of
inhabitants. When there was more than one person per bedroom, the household was considered
over-crowded. This definition is consistent with other Canadian research20
.
Income support
The variable “receiving income support” was operationalized using a closed-ended question
from the Nunavut Inuit Child Health Survey: “In the past month, did anyone in your household
receive income support”. Traditional measures of socioeconomic position (i.e. income, wealth,
education, occupation)21
, which have been strongly related to breastfeeding practices in other
populations22,23
, were not available for the entire sample of participants in the Nunavut Inuit
Child Health Survey (i.e. only for parents who also completed the adult survey). Income support
is given to Canadian families who are unable to provide adequately for themselves and/or their
dependents with their unassisted income24
, and may be a reasonable proxy indicator for low
income families and households. As such, the measure ‘receiving income support’ was used to
indicate if a family would be considered low income.
Household smoking restrictions
Household smoking restrictions was measured in response to the question “Are there any
restrictions against smoking cigarettes inside your home?”. Household smoking has been
associated with an increased risk of respiratory infection in infants25
and Inuit infants and
children seem particularly susceptible26,27
. Conversely, exclusive breastfeeding to six months
has been shown to reduce infant respiratory infections in some28
, but not all29
, research.
63
Identifying if smoking restrictions are associated with breastfeeding duration may identify a high
risk population that can be targeted in public health campaigns.
Household and Childhood Food Security
Household and child food security were measured using a modified United States Department of
Agriculture (USDA) 18-item Household and Food Security Survey. The scale was modified by
Aboriginal Affairs and Northern Development Canada, formerly known as Indian and Northern
Affairs Canada, to improve its cultural acceptability among Aboriginal populations 30
. No
existing food security scales have been validated in Inuit populations 30
, but the USDA scale has
high internal validity, construct validity and test-retest reliability in other populations (e.g.
general United States population as well as subpopulations (e.g. ethnic minorities)) 31,32
.
Consistent with the original scale, the first ten questions are adult-specific and the remaining
eight questions are child-specific for food security 33
.
In analyses where child and adult food security were treated separately, a three-level food
security variable was created for each, consistent with scoring recommendations from Health
Canada34,35
. When an individual indicated an affirmative response to no more than one of the
ten adult-specific questions or eight child-specific questions, they were classified as food secure;
an affirmative response on two to five of the adult-specific questions or two to four of the child-
specific questions resulted in a classification of moderate food insecurity; lastly, an affirmative
response to more than five adult specific questions or more than four child-specific questions
resulted in a classification of severe food insecurity.
In some analyses, an overall measure of household food security was used. A household was
classified as food secure when both the childhood and adult food security scores were within the
range of food secure. A household was classified as moderately food insecure when one (or
both) of the childhood or adult food security scores were within the range of moderately food
insecure, but neither was severely food insecure. A household was classified as severely food
insecure when one (or both) of the childhood or adult food security scores were within the range
of severely food insecure. This classification system was selected to be consistent with other
researchers using the Nunavut Inuit Child Health Survey11
.
64
Maternal behaviours during pregnancy: drinking and smoking
Maternal drinking and smoking were measured in response to two closed-ended questions “Did
you or the biological mother drink alcohol/smoke during pregnancy?” All caregivers, regardless
of their relationship to the child, were asked to report the number of cigarettes smoked by the
biological mother each day or month during pregnancy. This was measured as an open-ended
question as was the number of times the biological mother consumed greater than five drinks in
one sitting during pregnancy, consistent with Health Canada’s binge drinking guidelines for
women. Due to a high proportion of missing data on the open-ended questions (approximately
40%), only the binary measures (e.g. number of drinks or cigarettes smoked during pregnancy)
were used in the analyses.
Information on substance abuse was not collected for different periods of the pregnancy (e.g. by
trimester), although a few caregivers indicated (< 3%), in an open-ended question, drinking or
smoking only before the biological mother was aware of the pregnancy. Drinking and smoking
during pregnancy may reflect the woman’s general feelings and/or circumstances around the
pregnancy (e.g., if the pregnancy was wanted or intended). The association between drinking
and smoking during pregnancy and breastfeeding practices may be driven, at least in part, by
feelings of resentment towards an unwanted or unintended pregnancy, manifesting in substance
abuse. If this is the case, grouping together women who stopped substance abuse immediately
after discovering they were pregnant with women who continued to abuse substances throughout
pregnancy, may under-estimate the association between substance abuse during pregnancy and
subsequent breastfeeding.
The reliability and validity of maternal recall of drinking and smoking practices during
pregnancy have not been evaluated in an Inuit specific population. However, in other
populations, maternal self-reported recall of smoking during pregnancy (any compared to never)
tends to have a high agreement, adjusted for chance, with medical records and/or earlier surveys
conducted during the pregnancy, up to 15 years after birth (κ=0.85) 36
(κ=0.77)37
. In terms of
maternal self-reported recall of alcohol consumption during pregnancy, among inner-city
mothers, agreement tended to be much lower, and women were more likely to over-estimate the
amount of alcohol they consumed during pregnancy as compared to earlier medical records or
self-reported practices (κ=0.08) 37
(r=0.60)38
(r=0.61)39
.
65
Infant Birth Weight
Infant birth weight was measured in two ways on the Nunavut Inuit Child Health Survey. In the
first way, respondents were asked to indicate the birth weight of their child in an open-ended
question. In the second way, respondents were asked to categorize their infant’s birth weight in a
closed-ended question: low: < 5.5 pounds or <2500 grams; high: > 9 pounds or >4100 gram; or
average: 5.5 to 9 pounds or 2500 to 4100 grams. Because of a high proportion of missing data,
only the closed-ended infant birth weight variable was used in these analyses.
Maternal recall of infant birth weight has high agreement with medical records at birth/delivery,
regardless if the question is asked recently (i.e. 6 months after birth)(κ=0.99)40
or more distantly
(22 years after birth)(κ=0.88; Sensitivity=100%; Specificity=99%)41
. Agreement is also high for
maternal reported categorical measures of birth weight (low, average, high) when compared to
medical records (ICC=0.76)41
(r= 0.89)37
.
Region
Region of residence was recorded by the research team at the time of data collection. Previous
research has indicated that breastfeeding rates can vary by place.22,42
A recent analysis using the
Maternity Experiences Survey found that Canadians in the Northern Territories (i.e. Northwest
Territories, Nunavut, and Yukon) were less likely to begin breastfeeding42
, but three times more
likely to practice exclusive breastfeeding to six months (95% CI: 2.21-4.12).43
Community hospital or birthing centre
There are presently three communities in Nunavut that have the capacity to support child birth,
although only two were in operation at the time of the Nunavut Inuit Child Health Survey: Iqaluit
and Rankin Inlet (The Kitikmeot Health Centre in Cambridge Bay began operations in 2010).
Iqaluit is home to the Qikqitani General Hospital (formerly known as the Baffin Regional
Hospital), which has a low to medium risk obstetrics program staffed by primary care physicians,
nurses and midwives. Rankin Inlet is home to a birthing centre that has been in operation since
1992, where midwives and other health care professionals are available to deliver low risk
babies44
. These communities were the only communities noted as having a hospital or birthing
66
centre in all analyses. Individuals who did not live in a community with a birthing centre or
hospital likely were evacuated and gave birth in Iqaluit or hospitals outside of the territory.
Canada Prenatal Nutrition Program
The Canada Prenatal Nutrition Program aims to improve access to culturally appropriate
programming for pregnant women, new mothers and their infants. Components of the program
include: providing education and support, food for the mother and infant, and referrals to outside
agencies. All women who are pregnant or have infants up to one year of age are eligible to
participate.45
Administrative information distributed publicly by the Government of Nunavut, Department of
Health and Social Services, was used to ascertain if a community participated in the Canada
Prenatal Nutrition Program46
. Data were extracted for the years 2002-2005 to correspond with
the time each child survey participant was born. Data on whether individual’s participated in the
program was not available, but research suggests 70% of women with access to the program
elect to participate.47
Preventive health care practices
Health centre utilization records by clinic (i.e. community) and visit type (e.g. preventive care,
prenatal care, chronic condition care) were obtained to determine the extent of preventive health
practices in each community (unpublished data collected by the Nunavut Health Profile Project,
PI: G. Marchildon, University of Regina). This project recorded the reason for a patient’s visit
to a medical centre, which could fall into one of seven categories: management of chronic
condition(s), management of tuberculosis, prenatal care, management of an acute illness, and a
well-woman, well-man, or well child visit. A visit was classified as a ‘well’ visit if it was for
preventive care purposes (e.g. an annual check-up).
Two measures were used: one specific to prenatal care and the other, to preventive care in
general. To determine the extent of preventive care in the community, the number of well
woman, man, and child visits in each community were divided by the total population of that
community. To determine the value of prenatal care in the community, the number of prenatal
67
visits was divided by the total number of live births in that community. These variables were
labeled: frequency of preventive care and frequency of prenatal care respectively.
Although measures of preventive care have not been included in studies of breastfeeding
previously, there is evidence from studies of risk behaviours (e.g. drinking and smoking) in
youth that suggest environmental context plays on important role in shaping future health
behaviours (e.g. parental smoking, school smoking rates, peer smoking rates influence individual
behaviours)48,49
.
3.4 Methods for the IPY Inuit Health Survey
Study methodology for the IPY Inuit Health Survey have been published previously1. An
overview of study methods and relevant details that pertain to this thesis are provided below.
3.4.1 Participants and recruitment strategies
Non-pregnant individuals aged 18 years or older, living in one of the 36 communities at the time
of data collection, were the target population for the IPY Inuit Health Survey. Anyone meeting
these criteria was eligible to participate in the survey. Additional details regarding the study
population used for each thesis objective are provided in the appropriate manuscripts.
A list of household addresses was provided to the research team by each community. These
households were randomly selected to participate using either a computerized random number
generator or a random digit table. The aim was to sample approximately 12% of households in
each of the sampled communities.1 Once selected, a nurse from the research team visited each
household to invite them to participate in the survey. Anyone living in the household was eligible
to participate in the survey. When individuals were not home, an informational pamphlet was left
at the household with the contact information of the research team. At least three attempts were
made to visit each sampled household. Refusals were recorded.
3.4.2 Data collection strategies
Members of the research team travelled to participating communities in 2007 and 2008 to
administer questionnaires to selected households. Interviews were conducted by trained,
bilingual members (English, Inuktitut) of the research team. The data collection schedule is
located in Appendix A.
68
After receiving informed consent, each of six questionnaire modules were interviewer-
administered in the following order: a home demographic identification chart (completed by one
member of the household only); a home-based questionnaire (completed by one member of the
household only); the individual questionnaire; a community and personal wellness questionnaire;
a 24 hour dietary recall; and a food frequency questionnaire. These questionnaires can be found
at: http://www.inuithealthsurvey.ca/?nav=questionnaire. A summary of the information
collected in each questionnaire can be found in Table 4.
After questionnaires were completed, one member of the household was invited to provide
clinical measurements (e.g. anthropometric, bone density, blood pressure, toenail samples).
These data were not required for these analyses.
69
Table 4: Information collected in each survey module: IPY Inuit Health Survey
Module Information Collected
Demographic identification chart Age, Sex, and the relationships of everyone in the
household
Household Questionnaire Social determinants of health50
(e.g. housing conditions,
income and government support, food security)
Traditional activities and behaviours (e.g. hunting,
languages)
Household smoking practices
Household diet
Individual Questionnaire Health status and co-morbidities (e.g. diabetes,
hypertension)
Individual health behaviours (e.g. smoking, physical
activity)
Individual sociodemographic characteristics.
Reproductive histories (females only)
Community and Personal Wellness Mental health (e.g. depression, suicide ideation,
substance abuse, violence in the home, gambling)
Community violence
Food frequency questionnaire Frequency of foods consumed in past year
24-hour dietary recall Detailed recall of diet in past 24 hours
70
3.4.3 Study variables: Measurement and classification
Information on participation in traditional activities was extracted from the IPY Inuit Health
Survey. Indicator variables were selected in consultation with Inuit Canadians as well as other
experts in the field of Aboriginal health research.
Language Use
Please see the description for this indicator in section 3.3.4.
Hunting Activities
Hunting is viewed as an important traditional activity for many Inuit and has been used in First
Nations research to measure “traditionalism”51
. The importance of hunting is reflected in the
Inuit philosophy Avatittinnik Kamatsiarniq (i.e. respect and caring for the land, animals and
environment)52
.
Hunting activities were measured in response to the question: “In the past 12 months, what sorts
of activities have you carried out on the land”. A list of activities was provided to respondents,
including hunting on the land and hunting on water and fishing; individuals were asked to
answer either affirmatively or negatively (i.e. yes/no). If an individual did not include a
response, but indicated they had not participated in any activities on the land in the previous
year, they were treated as ‘no’ for these activities in the analyses.
Traditional Skill Development
Traditional skill development was measured in response to the question: “In the past 12 months,
what sorts of activities have you carried out on the land”. A list of activities was provided to the
respondents and included: building a cabin, using a dog team and building an igloo. Individuals
were asked to answer either affirmatively or negatively (i.e. yes/no). If an individual did not
include a response, but indicated they had not participated in any activities on the land in the
previous year, they were treated as ‘no’ for these activities in the analyses.
71
Community participation
Inuit sense of health and well-being is strongly linked to the community, an idea which is
reflected in the Inuit philosophies of Pijitsirniq and Piliriqatigiinniq (i.e. serving and providing
for family and community members)52
. The IPY Inuit Health Survey contained several indicators
of serving family and community members; three were included in these analyses. The first
indicator measured the distribution of country food to other community members in response to
the question “do you or your household distribute country food to other members of the
community?” This was a dichotomous variable (yes/no). The second indicator measured
participation in community activities to benefit the community “In the last month, how often
have you participated in activities where people came together for the benefit of the community”.
The third indicator measured general participation in community activities through the question
“In the past month, how often have you gotten together with people to play games, sports or
recreational activities”. Both of the latter questions were measured on a five point Likert scale
(i.e.: very often, often, sometimes, rarely, never) and were binned into a two level variable (very
often and often vs. not) for the latent class analysis.
3.5 Statistical Analyses
All statistical analyses were conducted in SAS v9.2 and v9.3. Two-tailed tests with an alpha of
0.05 were used to account for type I error. Some general statistical considerations are
highlighted here. Specific analysis methods can be found in the appropriate manuscripts.
3.5.1 Complex survey methodology
The IPY Inuit Health Survey and Nunavut Inuit Child Health survey were both population-based
surveys of Canadian Inuit. For the Nunavut Inuit Child Survey, the majority of children were
selected using a stratified random sampling strategy, wherein children were randomly selected to
participate in the survey from the sixteen pre-selected communities (i.e. 16 strata). More detailed
sampling information is provided in section 3.3.1. For the IPY Inuit Health Survey, a stratified
cluster sampling technique was used. In the first stage, all communities were divided in separate
strata. Then, household clusters were randomly selected to participate within each stratum. All
individuals in the household meeting the eligibility criteria were invited to participate in the IPY
Inuit Health Survey (see section 3.4.1 for eligibility and sampling information).
72
The probability based nature of the sampling strategy provides an ideal opportunity to make
inferences about the entire target population – not just those sampled - if design weights are
used53
. Employing population weights can provide unbiased estimates of characteristics and
associations within a population, assuming all individuals have an equal probability of inclusion
and that non-responders are missing completely at random54
The survey weights used were provided by the IPY Inuit Health Survey Steering Committee
directly and were adjusted for the probability of participant selection and survey non-response.
Below is an overview for how these weights were calculated.
As previously noted, the sampling fractions varied across communities, particularly in the
Nunavut Inuit Child Health Survey, meaning that the individual probability of being sampled,
(i.e. the sampling fraction) was conditional upon the community in which the individual lived.
The initial design weight (i.e. based on the probability of participant selection) was calculated as
the inverse of the number of individuals selected in the community divided by the total number
of eligible individuals in that community53
. For example, if 20 children between the ages of 3
and 5 years were sampled from a total population of 100, 3 to 5 year olds, each of these 20
children was assigned an initial weight of 5 (i.e. the reciprocal of 20 out of 100). These weights
were further adjusted for total non-response, such that if 16 of the 20 children selected each
participated, they were assigned a non-response adjustment factor of 1.25 (i.e. the reciprocal of
16 out of 20). The total adjusted weight was calculated by multiplying the initial design weight
by the non-response adjustment factor. In this particular example, the design weight would be
6.25 (i.e. 1.25 x 5 = 6.25), meaning that each child in the community represented 6.25 children in
the overall study.
In manuscript 2 (Chapter 5), only a sub-set of the total survey population was analyzed (i.e. only
those who initiated breastfeeding). Failing to account for this sub-setting may result in
inappropriately small variance estimators55
if at least one stratum in the primary sampling unit
does not have an individual in the subpopulation of interest55
. All of the communities (i.e. PSU:
strata) had at least one participant who had initiated breastfeeding, suggesting that variances
could be correctly estimated without adjustment. Nonetheless, the DOMAIN statement was used
in SURVEYLOGISTIC, which employs the Taylor Series Expansion method, to test this
assumption.56
73
Most procedures in SAS presuppose a simple random sample. This assumption is inappropriate
for complex survey sampling procedures, since there is a greater likelihood of similarity within
clusters as opposed to between clusters, violating the assumption of independence. Moreover, the
unequal sampling fractions that were used to select participants in each community cannot be
easily incorporated in non-survey based SAS procedures. To incorporate sampling design and
strategy in these analyses, complex survey methodology procedures were used in SAS (e.g.
SURVERYFREQ, SURVEYLOGISTIC).57
3.5.2 Model building strategies
Binary (Objective 2, 5 and 6) and multinomial (Objective 4), also known as polytomous, logistic
regression models for complex survey methods were used to estimate the log odds of
breastfeeding initiation and exclusive breastfeeding duration.
The breastfeeding initiation variable is inherently dichotomous, lending itself well to a binary
logistic regression analysis. After reviewing the distribution of the duration of exclusive
breastfeeding in the sample, four distinct patterns of breastfeeding women were observed. The
first type of woman initiated breastfeeding but stopped breastfeeding exclusively within the first
month; the second type initiated breastfeeding and went beyond one month but stopped
breastfeeding exclusively at some point before six months; the third type practiced exclusive
breastfeeding as recommended (i.e. to six months); and the fourth type were still practicing
exclusive breastfeeding beyond the recommended six months. Multinomial logistic regression
enabled the modeling of each of these four outcomes, providing a more comprehensive
understanding of factors related to various types of exclusive breastfeeding practices.
Variable selection procedures for both the binary and multinomial regression models were
consistent with published recommendations for exploratory predictive models58
. A forwards
selection procedure with a test for backwards elimination was used to determine the order in
which variables should be entered into the model. The entry criteria was p<0.25 and the exit
criteria was p<0.30. These entry and exit criteria are thought to be fairly conservative, allowing
for the identification of all potentially important factors that would be missed at the p<0.05
level58
. This conservative level was also selected due to the relatively smaller sample sizes of
the population and subpopulations of interest.
74
To summarize, first the associations between each potential factor and the outcome of interest
were examined in crude models (i.e. univariable models). Factors that were significant at the
p<0.25 level using a likelihood ratio test were candidates for the multivariable (or multivariate)
model. The variable that was most strongly associated with the outcome of interest, as indicated
by the likelihood ratio test comparing k and k-1 models, was first entered into a logistic
regression model along with each of the other candidate variables, one at a time (i.e. a two
predictor model). The variable that produced the greatest change in the likelihood ratio test from
the previous model (i.e. the outcome variable and the strongest predictor variable), while still
meeting the entry criteria, was entered next into the model. The same procedure was followed to
determine the third variable that should be entered into the model. Once there was a three
variable model, variables were eliminated, one at a time, to verify their continued importance.
The forwards procedure with a backwards elimination continued until all variables that remained
in the model met entry (p<0.25) and exit (p<0.30) criteria.
As previously mentioned, model building strategies were in accordance with published
recommendations for exploratory predictive models58
. Using p-value driven algorithms for
variable selection is generally considered antithetical to the change-in-effect estimate algorithms
that are advocated for epidemiologic studies of causality59
. The goals in each analysis, however,
are disparate. When building an exploratory predictive model, the objective of modeling is to
identify which variable or sub-set of variables best predict the occurrence of an outcome58
.
When building a model to assess the causal role of a factor on an outcome, the aim is to provide
an unbiased estimate of the effect of an exposure variable on the outcome of interest, controlling
for important covariates and accounting for effect modifiers59
. Using p-value driven algorithms
is an acceptable strategy to identify important variables in exploratory model building,
recognizing that the reliability of variables selected, may be different in different samples60
.
3.5.3 Multilevel modeling
Multilevel modeling was used to explore the associations of various community factors (e.g.
having a hospital or birthing centre in the community; having access to the Canada Prenatal
Nutrition Program) and breastfeeding practices (Objective 2 and 4). Multilevel modeling is an
ideal analytic approach for such a research question, as it allows for the identification of: the
independent effect of community factors on individual breastfeeding outcomes; the independent
75
effect of individual factors on breastfeeding practices; and the joint effects of any combination of
1) two (or more) individual level variables, 2) two (or more) community level variables, or 3)
one (or more) variables at each level (i.e. cross-level) 61,62
. Moreover, multilevel models can
account for clustering within communities, producing appropriate standard errors.63,64
Multilevel models were selected in lieu of population averaged models because although the
latter do generate standard errors and parameters estimates corrected for clustering within a
community 54,65
, these models average out effects over the clusters and do not generate
community specific parameter estimates. Additionally, unlike population averaged models,
multilevel models partition out the covariance structure within and between community clusters
65,66. Identifying how much variability exists within and between communities is particularly
important for targeted public health strategies; an intervention aimed at high risk individuals
would be most appropriate when a high proportion of variability in breastfeeding was observed
within clusters, whereas an intervention aimed at high risk communities would be most
appropriate when a high proportion of variability in breastfeeding was observed between
clusters 62,67
.
Multilevel modeling requires a relatively large number of clusters at level two (i.e. the
community level) to make meaningful conclusions. The Nunavut Inuit Health Survey only
sampled 16 communities – considered quite sparse for multilevel models68
. As such, all
community variables were initially treated as individual level variables in initial logistic
regression models. Standard approaches were used to determine the model of best fit58
. If
community variables were important in these models, all of the pre-determined variables were
entered into the multilevel model.
The GLIMMIX procedure in SAS was used with a logit link to estimate parameter estimates in
multilevel models. This procedure allowed for the incorporation of the complex survey design
methodology using the ‘weight’ and ‘random’ statements while exploring both community and
individual effects. The GLIMMIX procedure implements a partial likelihood algorithm
developed by Wolfinger and O’Connell. This algorithm uses a First Order Taylor Series
Expansion to estimate fixed and random effects 69,70
. This partial likelihood is similar to the
penalized quasi-likelihood method used in MLWin 71
. Although there is some evidence to
suggest parameter estimates may be severely downward biased in GLIMMIX and other software
76
using first order Taylor series expansion 72
, these downward biases tend to be mitigated when the
variances do not exceed 0.5 and/or when the number of observations per cluster are large 71
. In
comparison with NLMIXED, another SAS procedure, GLIMMIX assumes that random effects
are normally distributed 70,73
. Tests for significance are always two sided for fixed effects and
one sided for random effects in these SAS procedures.
3.5.4 Latent class analysis
To classify various classes of traditional lifestyles in Inuit Canadians (Objective 5), Latent Class
Analysis (LCA) was used74
. LCA identifies mutually exclusive subgroups in a population by
uncovering heterogeneity in a population and grouping together those individuals who share
similar typologies into latent classes75
. All variables in the traditional LCA are categorical76
.
LCA has similarities to both cluster analysis and factor analysis. The goals of both cluster
analysis and LCA are to identify different classes of persons, but in the former, classes are
assigned on an absolute basis based on more arbitrary criteria, whereas LCA provides individual
probabilities of class membership based on statistical distributions77
. In terms of factor analysis,
both LCA and factor analysis generate latent (or unmeasured) constructs, but in the former, the
latent construct is developed to classify an individual whereas in the later, the latent factor is
developed to create a latent variable. Because of this difference, LCA is then considered a
‘person-centered’ approach whereas factor analysis is considered a ‘variable-centered’
approach76
.
In LCA, latent (i.e. unmeasured) variable(s) are ultimately identified, conditional on responses
to manifest variables76
. LCA produce two types of estimates: class membership probabilities
(i.e. the overall prevalence of that class in the population of interest) and item-response
probabilities which are conditional on class membership (i.e., the probability that an individual in
class of interest responded in a certain way to an indicator variable).76,78
The appropriate number of classes (i.e. class enumeration) can be determined in multiple ways.
The parametric bootstrap method, which uses bootstrap iterations to estimate the distribution
between k and k-1 models (i.e. nested models), is currently the recommended method75
. The
LCA bootstrap method79
, the adjusted Bayesian information criteria (adjusted BIC), Akaike’s
Information Criteria (AIC), and theory were used to decide the appropriate number of latent
classes for the data.
77
A more detailed description of LCA can be found in Manuscript #3 (Chapter 6).
3.6 Ethics
Ethical approval for these secondary data analysis was granted by the University of Toronto’s
Health Sciences Research Ethics Board (Protocol Reference #24845) in 2010 with annual
renewals occurring in 2011 and 2012. Initial ethical approval can be found in Appendix C.
Ethical approval for the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey were
obtained from the McGill Faculty of Medicine Institutional Review Board in 2007 with annual
renewals until 20121. Research licenses to conduct research in the North were granted from the
Aurora Research Institute (Northwest Territories) and the Nunavut Research Institute
(Nunavut)1.
78
3.7 Section References
1. Saudny H, Leggee D, Egeland G. Design and methods of the Adult Inuit Health Survey
2007–2008. International Journal of Circumpolar Health.
2012;71:10.3402/ijch.v3471i3400.19752.
2. International Polar Year 2007-2008. About IPY. 2010;
http://www.ipy.org/index.php?/ipy/about/. Accessed March 27, 2013.
3. Government of Nunavut. Nunavut Bureau of Statistics, Community estimates.
http://www.gov.nu.ca/eia/stats/index.html.
4. Statistics Canada. Total non-Aboriginal and Inuit population, by residence, Canada,
2006. 2012; http://www.statcan.gc.ca/pub/82-003-x/2012003/article/11695/tbl/tbl1-
eng.htm. Accessed October 23, 2012.
5. Statistics Canada. 2006 Census: Aborignal Peoples in Canada in 2006: Inuit, Metis and
First Nations, 2006 Census: Inuit. 2009; http://www12.statcan.ca/census-
recensement/2006/as-sa/97-558/p9-eng.cfm. Accessed October 23, 2012.
6. O’Donnell V, Wallace S. First Nations, Metis and Inuit women. 2012;
http://www.statcan.gc.ca/pub/89-503-x/2010001/article/11442-eng.htm#a14. Accessed
Dec 5, 2012.
7. Martel L, Caron-Malenfant E. 2006 Census: Portrait of the Canadian Population in 2006,
by age and sex. 2011; http://www12.statcan.gc.ca/census-recensement/2006/as-sa/97-
551/index-eng.cfm. Accessed March 24, 2013.
8. World Health Organization. Indicators for assessing infant and young child feeding
practices: part 2 measurment. 2010;
http://whqlibdoc.who.int/publications/2010/9789241599290_eng.pdf. Accessed
December 15, 2012.
9. Tienboon P, Rutishauser IHE, Wahlqvist ML. Maternal recall of infant feeding practices
after an interval of 14 to 15 years. Australian Journal of Nutrition and Dietetics.
1994;51:25-25.
10. Tomeo CA, Rich-Edwards JW, Michels KB, et al. Reproducibility and Validity of
Maternal Recall of Pregnancy-Related Events. Epidemiology. 1999;10(6):774-776.
11. Egeland GM, Pacey A, Cao Z, Sobol I. Food insecurity among Inuit preschoolers:
Nunavut Inuit Child Health Survey, 2007-2008. CMAJ. 2010;182(3):243-248.
12. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
13. Quandt SA. Material recall accuracy for dates of infant feeding transitions. Human
Organization. 1987;46(2):152-160.
79
14. Eaton-Evans J, Dugdale A. Recall by mothers of the birth weights and feeding of their
children. Human nutrition. Applied nutrition. 1986;40(3):171.
15. Aarts C, Kylberg E, Hörnell A, Hofvander Y, Gebre-Medhin M, Greiner T. How
exclusive is exclusive breastfeeding? A comparison of data since birth with current status
data. International Journal of Epidemiology. December 1, 2000 2000;29(6):1041-1046.
16. Flaherman VJ, Chien AT, McCulloch CE, Dudley RA. Breastfeeding rates differ
significantly by method used: a cause for concern for public health measurement.
Breastfeeding Medicine. 2011;6(1):31-35.
17. Statistics Canada. Canadian Health Measures Survey (CHMS). 2012;
http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=5071&Item_Id=
129548&lang=en. Accessed April 8, 2013.
18. Bjerregaard P, Curtis T. Cultural change and mental health in Greenland: the association
of childhood conditions, language, and urbanization with mental health and suicidal
thoughts among the Inuit of Greenland. Social Science & Medicine. 2002;54(1):33-48.
19. Foggin PM, Aurillon N. Respiratory health indicators and acculturation among the Inuit
and Cree of Northern Quebec: A regional approach using geographic seriation analysis.
Social Science & Medicine. 1989;29(5):617-626.
20. Canadian Tuberculosis Committee. Special report: Tuberculosis among the Aboriginal
Peoples of Canada, 2000 to 2004. 2004; http://www.phac-
aspc.gc.ca/publicat/2007/tbcan04/tbaboriginal-eng.php. Accessed September 21, 2012.
21. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Smith GD. Indicators of socioeconomic
position (part 1). Journal of Epidemiology and Community Health. 2006;60(1):7-12.
22. Li R, Darling N, Maurice E, Barker L, Grummer-Strawn LM. Breastfeeding rates in the
United States by characteristics of the child, mother, or family: the 2002 National
Immunization Survey. Pediatrics. 2005;115(1):e31-37.
23. Jacknowitz A, Novillo D, Tiehen L. Special supplemental nutrition program for women,
infants, and children and infant feeding practices. Pediatrics. 2007;119(2):281-289.
24. Federal-Provincial-Territorial (FPT) Directors of Income Support. Social Assistance:
Statistical Report. 2009;
http://www.hrsdc.gc.ca/eng/publications_resources/social_policy/fpt/hs_25-2e.pdf.
Accessed Janary 31, 2013.
25. Strachan DP, Cook DG. Health effects of passive smoking. Parental smoking and lower
respiratory illness in infancy and early childhood. Thorax. 1997;52(10):905-914.
26. Orr P, McDonald S, Milley D, Brown R. Bronchiolitis in Inuit children from a Canadian
central arctic community, 1995-1996. Int J Circumpolar Health. 2001;60(4):649-658.
80
27. Orr PH. Respiratory tract infections in Inuit children: "Set thine house in order". CMAJ.
2007;177(2):167-168.
28. Bachrach VR, Schwarz E, Bachrach LR. Breastfeeding and the risk of hospitalization for
respiratory disease in infancy: a meta-analysis. Archives of pediatrics & adolescent
medicine. Mar 2003;157(3):237-243.
29. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention
Trial (PROBIT): A Randomized Trial in the Republic of Belarus. JAMA.
2001;285(4):413-420.
30. Lawn J, Harvey D, Dialogos Educational Consultants Inc. Nutrition and Food Security in
Kugaaruk, Nunavut: baseline survey for the Food Mail pilot project. 2003; http://dsp-
psd.pwgsc.gc.ca/Collection/R2-265-2003E.pdf. Accessed March 3, 2009.
31. Carlson SJ, Andrews MS, Bickel GW. Measuring Food Insecurity and Hunger in the
United States: Development of a National Benchmark Measure and Prevalence
Estimates. J. Nutr. 1999;129(2):510S-516S.
32. Radimer KL. Measurement of household food security in the USA and other
industrialised countries. Public Health Nutrition. 2002;5(6a):859-864.
33. Bickel G, Nord M, Prince C, Hamilton W, Cook J. Measuring Food Security in the
United States. 2000; http://www.fns.usda.gov/fsec/files/fsguide.pdf. Accessed June 2,
2012.
34. Nord M, Hopwood H. A comparison of household food security in Canada and the
United States: US Department of Agriculture, Economic Research Service; 2008.
35. Health Canada. Canadian Community Health Survey, Cycle 2.2, Nutrition (2004):
Income-related household food security in Canada. Ottawa, 2007.
36. Yawn BP, Suman VJ, Jacobsen SJ. Maternal recall of distant pregnancy events. Journal
of clinical epidemiology. 1998;51(5):399-405.
37. Jacobson SW, Jacobson JL, Sokol RJ, Martier SS, Ager JW, Kaplan MG. Maternal recall
of alcohol, cocaine, and marijuana use during pregnancy. Neurotoxicology and
Teratology.13(5):535-540.
38. Bunin GR, Gyllstrom ME, Brown JE, Kahn EB, Kushi LH. Recall of diet during a past
pregnancy. American Journal of Epidemiology. 2001;154(12):1136-1142.
39. Bosco JL, Tseng M, Spector LG, Olshan AF, Bunin GR. Reproducibility of reported
nutrient intake and supplement use during a past pregnancy: a report from the Children's
Oncology Group. Paediatric and Perinatal Epidemiology. 2010;24(1):93-101.
40. Troude P, L’Hélias L, Raison-Boulley A-M, et al. Perinatal factors reported by mothers:
do they agree with medical records? European Journal of Epidemiology. 2008;23(8):557-
564.
81
41. Buka SL, Goldstein JM, Spartos E, Tsuang MT. The retrospective measurement of
prenatal and perinatal events: accuracy of maternal recall. Schizophrenia research.
2004;71(2-3):417.
42. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
43. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
44. Victoria C. Effect of breastfeeding on infant and child mortality due to infectious diseases
in less developed countries: a pooled analysis. Lancet 2000;355(9202):451-455.
45. Public Health Agency of Canada. The Canadian Prenatal Nutrition Program. 2002;
http://www.phac-aspc.gc.ca/dca-dea/programs-mes/cpnp_goals-eng.php. Accessed April
15, 2010.
46. Thomas T, Greeley G, Guyot M, Loubert K, Banfield W. Inuit Wellness Programs in
Nunavut: 2004-2005. 2006; http://pubs.aina.ucalgary.ca/health/61935.pdf. Accessed May
24, 2012.
47. Simard I, O'Brien H, Beaudoin A, et al. Factors influencing the initiation and duration of
breastfeeding among low-income women followed by the Canada prenatal nutrition
program in 4 regions of Quebec. J Hum Lact. 2005;21(3):327-337.
48. Frohlich KL, Potvin L, Chabot P, Corin E. A theoretical and empirical analysis of
context:: neighbourhoods, smoking and youth. Social Science & Medicine.
2002;54(9):1401-1417.
49. Novak SP, Clayton RR. The influence of school environment and self-regulation on
transitions between stages of cigarette smoking: a multilevel analysis. Health Psychology.
2001;20(3):196.
50. Inuit Tapiriit Kanatami. Social determinants of Inuit health in Canada: a discussion paper.
2007; http://ahrnets.ca/files/2011/02/ITK_Social_Determinants_paper_2007.pdf.
Accessed June 20, 2012.
51. Wilson K, Rosenberg MW. Exploring the determinants of health for First Nations
peoples in Canada: can existing frameworks accommodate traditional activities? Soc Sci
Med. 2002;55(11):2017-2031.
52. Government of Nunavut. Pinasuaqtavut 2004-2009: Our committment to building
Nunavut's Future. Working to improve the health, prosperity, and self-reliance of
Nunavummiut. 2004;
http://www.nunavuteconomicforum.ca/public/files/library/POLITICA/PINASUAQ.PDF.
53. Statistics Canada. Survey methods and practices. Ottawa: Minister of Industry; 2010.
82
54. Vittinghoff E, Glidden D, Shiboski S, McCulloch C. Regression methods in biostatistics:
linear, logistic, survival and repeated measures models. New York: Springer; 2005.
55. Graubard B, Korn E. Survey inference for subpopulations. American Journal of
Epidemiology. 1996;144(1):102-106.
56. Wang Z, Waldron W. Using the SAS Survey Procedures for Subpopulation Analysis with
Jacknife Repeated Replication Methods in SAS 9.2. SAS Global Forum 2010 2010;
http://support.sas.com/resources/papers/proceedings10/267-2010.pdf. Accessed
December 4, 2012.
57. SAS Institute Inc. Introduction to Survey Sampling and Analysis Procedures. SAS/STAT
9.2 User's Guide. Cary, North Carolina: SAS Institute Inc.; 2008.
58. Hosmer D, Lemeshow S. Applied Logistic Regression. 2nd ed. Toronto: John Wiley and
Sons, Inc; 2000.
59. Greenland S. Modeling and variable selection in epidemiologic analysis. American
Journal of Public Health. 1989;79(3):340-349.
60. Shmueli G. To explain or to predict? Statistical Science. 2010;25(3):289-310.
61. Hox J. Multilevel analysis: techniques and applications: Mahwah, N.J., Lawrence
Erlbaum Publishers; 2002.
62. Diez-Roux AV. Multilevel Analysis in Public Health Research. Annual Review of Public
Health. 2000;21(1):171-192.
63. Snijders T, Bosker R. Standard Errors and Sample Sizes for 2-Level Research. Journal of
Educational Statistics. 1993;18:237 - 259.
64. Snijders TAB, Bosker RJ. Multilevel analysis: an introduction to basic and advanced
multilevel modeling. Thousand Oaks, California: Sage Publications; 1999.
65. Hubbard AE, Ahern J, Fleischer NL, et al. To GEE or Not to GEE: Comparing
Population Average and Mixed Models for Estimating the Associations Between
Neighborhood Risk Factors and Health. Epidemiology. 2010;21(4):467-474.
66. Raudenbush S, Bryk A. Hierarchical Linear Models: applications and data analysis
methods. 2 ed. Thousand Oaks, California: Sage Publications; 2002.
67. Duncan C, Jones K, Moon G. Context, composition and heterogeneity: Using multilevel
models in health research. Social Science & Medicine. 1998;46(1):97-117.
68. Bell BA, Morgan GB, Kromrey JD, Ferron JM. The Impact of Small Cluster Size on
Multilevel Models: A Monte Carlo Examination of Two-Level Models with Binary and
Continuous Predictors. Paper presented at: JSM Proceedings, 2010.
83
69. Wolfinger R, O'Connell M. Generalized linear mixed models a pseudo-likelihood
approach. Journal of Statistical Computation and Simulation. 1993;48(3):233 - 243.
70. The SAS institute. The GLIMMIX procedure. 2006;
http://www.ats.ucla.edu/stat/sas/glimmix.pdf. Accessed March, 2010.
71. Guo G, Zhao H. Multilevel modeling for binary data. Annual Review of Sociology.
2000:441.
72. Rodriguez G, Goldman N. An Assessment of Estimation Procedures for Multilevel
Models with Binary Responses. Journal of the Royal Statistical Society. Series A
(Statistics in Society). 1995;158(1):73-89.
73. Witte JS, Greenland S, Kim L-L, Arab L. Multilevel Modeling in Epidemiology with
GLIMMIX. Epidemiology. 2000;11(6):684-688.
74. PROC LCA & PROC LTA (version 1.2.7) [computer program]: University Park: The
Methodology Centre, Penn State; 2011.
75. Nyland K, Asparouhov T, Muthen B. Deciding on the number of classes in latent class
analysis and growth mixture modeling: a Monte Carlo Simulation study. Structural
Equation Modeling. 2007;14(4):535-569.
76. Lanza S, Collins L, Lemmon D, Schafer J. PROC LCA: A SAS procedure for latent class
analysis. Structural Equation Modeling. 2007;14(4):671-694.
77. Magidson J, Vermunt J. Latent class models for clustering: A comparison with K-means.
Canadian Journal of Marketing Research. 2002;20(1):36-43.
78. Lanza S, Dziak J, Huang L, Xu S, Collins L. PROC LCA & PROC LTA users' guide
(Version 1.2.7). University Park: The Methodology Center, Penn State; 2011.
79. LCABootstrap SAS Macro (version 1.1.0) [computer program]. University Park: The
Methodology Center, Penn State; 2011.
84
Chapter 4 : Manuscript 1
4 Study Overview
This first manuscript fulfills the first and second thesis objectives: to describe the
prevalence of breastfeeding initiation among Canadian Inuit (Objective 1) and to identify factors
associated with breastfeeding initiation in this population (Objective 2) using data from the
Nunavut Inuit Child Health Survey.
85
4.1 Prevalence and characteristics associated with breastfeeding initiation among Canadian Inuit from the 2007-2008 Nunavut Inuit Child Health Survey
Kathryn E. McIsaac MSc1; Daniel W. Sellen PhD
1,2, Wendy Lou PhD
1, T. Kue Young PhD MD
1
1Dalla Lana School of Public Health, University of Toronto;
2 Department of Anthropology, University of Toronto
86
ABSTRACT
Background: Little population-based research has been done to explore the prevalence and
determinants of breastfeeding initiation in Inuit Canadians. The aim of this research is to
determine the prevalence of, and factors associated with, breastfeeding initiation in Inuit
Canadians.
Methods: We used data from the Nunavut Inuit Child Health Survey (2007-2008), a population-
based, cross-sectional survey of Inuit children (aged 3 to 5 years) conducted among randomly
selected caregivers of these children. Various household, maternal, infant and community
characteristics were identified as potential predictors of breastfeeding initiation. Weighted
logistic regression and random effects models were used to identify predictors of breastfeeding
initiation from three caregiver groups: biological mothers, biological fathers, and adoptive
parents.
Results: The weighted prevalence of breastfeeding initiation was estimated at 67.0 % (95% CI:
61.7-72.2) for the entire sample (n=361) and 76.3% (95% CI: 70.6-82.1) using a sub-sample of
reports from biological mothers only (n=228). Children raised by an adoptive parent had a lower
reported prevalence of breastfeeding initiation (33.5%) (OR=0.16, 95%CL: 0.08, 0.29). Factors
associated with breastfeeding initiation varied by caregiver subgroup. Among infants currently
raised by their biological mothers, low infant birth weight (OR: 0.16, 95%CI: 0.10-0.27) and
maternal smoking during pregnancy (OR: 0.34, 95%CI: 0.19-0.61) were associated with a lower
odds of breastfeeding initiation, while household crowding (OR=2.01, 95%CI: 1.39-2.94) was
associated with a higher odds of breastfeeding initiation. Living in a community served by the
Canada Prenatal Nutrition Program, though not statistically significant, showed a trend towards
being important for increasing breastfeeding initiation among infants being raised by adoptive
parents (OR=4.04, 95%CI: 0.99-17.54).
Interpretation: Increasing breastfeeding initiation of Inuit Canadians to levels observed in other
Canadian regions will require health interventions that effectively engage different types of
primary caregiver groups, with community health services seeming particularly important for
families with children likely to become adopted.
87
Breastfeeding is widely recognized as a normal and healthy infant feeding practice conferring
many benefits to the infant and mother 1. A recent national survey
2 found that 90% of Canadian
women delivering live births in 2006 initiated breastfeeding as recommended by the World
Health Organization and Health Canada 3,4
, but also that large variations exist between
provinces, socioeconomic and ethnic groups 2.
Canadian Inuit have a lower initiation rate (66%), based on responses to the Aboriginal
Children’s Survey (ACS) (2006)5. One concern is that, despite some recent reductions,
disparities between Inuit and other Canadians in infant health indicators remain substantial6.
Exclusive breastfeeding, particularly to six months, is protective against many conditions that
Inuit Canadians are disproportionately affected by7-12
, including respiratory tract infections13
,
gastrointestinal infections14
, otitis media15
and sudden infant death syndrome16
. Initiating
breastfeeding is the first step towards practicing exclusive breastfeeding to six months, a public
health measure that may help mitigate health disparities in Inuit Canadians.
Presently, no population based research exists on the determinants of breastfeeding initiation in
an Inuit-specific context and much of the research that does exist is more than twenty years
dated17,18
. Although research has been conducted in a Canadian context19,20
, determinants of
breastfeeding practices can vary across populations21,22
and results may not be directly
generalizable to Inuit Canadians. Moreover, several potentially important factors that are more
unique to an Inuit setting, such as the high prevalence of household food insecurity23,24
, and the
practice of relocating (i.e. evacuating) the majority of pregnant women to hospitals in more
southern regions of Canada to give birth 25
, may influence breastfeeding initiation and have not
been previously evaluated.
In order to improve breastfeeding initiation in Canada’s North, there needs to be a better
understanding of factors that are associated with the practice, so that public health campaigns can
be tailored to the most appropriate audiences. The aim of the present research is to determine the
prevalence of breastfeeding initiation and identify factors associated with this practice in a
sample of Inuit families living in Nunavut.
METHODS
We use data from the Nunavut Inuit Child Health Survey (N-ICHS), a cross-sectional,
population-based survey conducted in Nunavut between August and November of 2007 and
88
2008. The survey was designed to measure key aspects of health among Inuit children aged 3 to
5 years at the time of data collection. The Institutional Review Board of the McGill University
Faculty of Medicine approved the N-ICHS, and the University of Toronto’s Health Sciences
Research Ethics Board approved these secondary analyses (Protocol ID: 24845).
Sixteen out of 25 candidate Nunavut communities were selected to participate in the N-ICHS;
nine communities were excluded because of the small numbers of eligible participants or for
logistical reasons fully described elsewhere 26
. Potential proxy respondents were the primary
caregivers of the children and were identified for recruitment in one of two ways: (1) the children
themselves were randomly selected using either random number tables or random number
generators from medical records maintained at the community health centre, the sole source of
access to health care in communities (expected to have close to 100% coverage); or (2) primary
caregivers who were randomly selected to participate in the concurrent IPY Inuit Health Survey
(i.e. a survey of Inuit aged 18 and older) and had children of eligible age were invited to
participate. More detailed descriptions of participant selection for the IPY Inuit Health Survey
are published elsewhere27
. Approximately 20% of children living in each community were
selected to participate in the survey, although sampling fractions varied (i.e. from 12% to 45%)
depending on the number of children in the target population in each community.
Once participants were randomly selected, a member of the research team visited the home to
distribute an informational DVD, explained the survey and asked for caregiver consent. Three
attempts were made to contact potential participants. Those consenting to participate were then
invited to a community site to complete an interviewer-administered questionnaire in either
English or Inuktitut within two weeks.
Breastfeeding initiation was measured in response to the question “was your child ever breastfed
or fed breast milk?” Those responding “yes” were classified as initiating breastfeeding.
Maternal recall of breastfeeding initiation as far as 14 to 15 years earlier agrees well with
medical records (85%) according to a review of 11 studies from a variety of populations around
the world 28
. The average age of children in the N-ICHS was 4.4 years26
, a much shorter
duration to recall breastfeeding initiation, suggesting limited bias with respect to maternal recall
of breastfeeding initiation . The accuracy and precision of recalling maternal breastfeeding
initiation among other types of caregivers (i.e. adoptive parents, biological fathers) has not
previously been evaluated.
89
We selected factors potentially associated with breastfeeding initiation based on a review of the
literature including infant birth weight 29
, maternal drinking and smoking during pregnancy 20
,
place of residence 30,31
, socioeconomic status 32
and participation in preventive care practices33
.
We also explored potential social determinants of breastfeeding initiation that may be of
particular interest for Canadian Inuit communities, such as access to health care, household
overcrowding, cultural engagement with traditional knowledge and food security. Measures for
most of these factors were extracted directly from the N-ICHS. Information on some factors that
have previously been identified as important determinants of breastfeeding initiation, such as
maternal age19
and education were only available for those caregivers who had also completed
the IPY Inuit Health Survey and were not included in these analyses.
Primary language spoken at home was used as an indicator for cultural engagement with
traditional knowledge 34,35
. Receiving income support was used as an indicator for low
socioeconomic status. Food security was measured using a version of the national 18-item
Household Food Security Survey module 36
, as modified by Aboriginal Affairs and Northern
Development Canada 37
, and divided into two categories (secure and insecure) that are described
elsewhere26
. Household crowding was defined as more than one person per bedroom in the
household38
.
We were also interested in different aspects of community conditions. Access to health care was
assessed in terms of living in a community with a birthing facility and participating in the
Canada Prenatal Nutrition Program39
. Frequency of preventive health practices in each
community was assessed by two measures. Prenatal preventive care frequency was calculated by
dividing the number of prenatal visits by the number of live births per year in each community.
General preventive care frequency was calculated by dividing the number of preventive care
visits by total population size per year in each community. These health-care utilization data
were obtained from a recent inventory of health care resources in Nunavut [personal
communication: Gregory Marchildon]. When possible, community records corresponded to the
year the child was born (i.e. 2002-2005).
Statistical Analysis
We conducted all statistical analyses in SAS v.9.3. To estimate the breastfeeding initiation
practices of the entire target population (i.e. the Nunavut communities sampled), we assigned
sampling weights. These weights reflect the probability of participant selection within each
90
community, such that responses from individuals in communities with a 10% sampling fraction
would represent the responses of 10 community members40
. We used complex survey
methodology procedures in SAS (e.g. SURVEYLOGISTIC) to incorporate these sampling
weights in our analyses and to account for the stratified sampling design 41
.
For most analyses, we divide our sample into three distinct subgroups based on reports from
different categories of caregiver- biological mothers, biological fathers and adoptive parents.
Membership into each subgroup was based on response patterns to two questions: “what is the
respondent’s relationship to the child?” and “what is the respondent’s gender?” Individuals
indicating they were a female and a biological parent were classified as the biological mother;
individuals indicating they were a male and a biological parent were classified as the biological
father; all other caregivers were classified as an adoptive parent.
The prevalence of breastfeeding initiation (weighted and unweighted) was calculated in the
combined sample (i.e. biological mothers, fathers and adoptive parents) and in each of the
separate caregiver subgroups. Chi-square tests of association identified any differing
breastfeeding initiation patterns by primary caregiver.
Crude and multivariable logistic regression models were fit for each of our three subgroups using
only participants who provided full information on each potential predictor. We present the
weighted crude prevalence odds ratios and 95% confidence limits for each factors’ association
with breastfeeding initiation. To identify factors for our multivariable model, we used a
modified forwards-stepwise logistic regression approach 42
wherein all variables exhibiting a
crude association of p<0.25 were considered candidates for the model. Model building strategies
were consistent with published recommendations for exploratory prediction models42
.
Specifically, the most significant factor meeting the entry criteria (p<0.25), as identified by a
likelihood ratio test, was entered first into the model. Factors were added and removed from the
model, in order of significance (exit criteria: p<0.30)42
.
After determining which characteristics to include in our fully adjusted multivariable model, we
assessed these variables’ significance using a random effects multilevel regression model. This
strategy affords a balance between being able to account for the hierarchal structure of the data
while considering the sample size requirements of hierarchal model building.
91
RESULTS
Response rates are presented in Figure 1. There were 361 (67.2%) caregivers retained for
prevalence estimates and 278 retained for crude and multivariable analyses (i.e. 199 biological
mothers, 32 biological fathers and 47 adoptive parents).
Table 1 and Table 2 show the unweighted and weighted distributions, respectively, of select
characteristics of survey respondents as well as the prevalence of breastfeeding initiation, by
caregiver. With the exception of breastfeeding initiation, there were no statistically significant
differences in characteristics across caregiver subgroups. Missing value patterns are also
presented in Table 1.
In the combined prevalence data set (n=361), there were 244 (67.6%, 95% CI: 62.8-72.4)
children who began breastfeeding. Weighted frequency estimates were similar (67.0%, 95% CI:
61.7-72.2). In the weighted sample, 76.3% (95% CI: 70.6-82.1) of biological mothers, 60.4%
(95% CI: 43.5-77.4) of biological fathers and 33.5% (95% CI: 21.8-45.2) of adoptive parents
report their child had been breastfed. The odds of breastfeeding initiation were lower amongst
adoptive parents (OR=0.16, 95% CI 0.08-0.29) but not significantly lower amongst reports from
biological fathers (OR=0.47, 95% CI: 0.22-1.03) compared to biological mothers.
Table 3 presents crude prevalence odds ratios of breastfeeding initiation for each characteristic of
interest, by the type of caregiver report. There were no factors that were predictive of
breastfeeding initiation in all three subgroups. Several household, maternal and infant
characteristics were related to breastfeeding initiation among children raised by their biological
mothers or fathers. Community factors were most strongly related to breastfeeding initiation
among children raised by adoptive parents.
Table 4 shows effect estimates from fully adjusted regression models that were used to inform
our final random effects model (Table 5). In our final random effects model, low infant birth
weight (OR=0.16, 95% CI: 0.10-0.27) and maternal smoking during pregnancy (OR=0.34,
95%CI: 0.19-0.61) were associated with a reduced likelihood of breastfeeding initiation and
household crowding (OR=2.01, 95%CI: 1.39-2.94) was associated with an increased likelihood
of breastfeeding initiation among biological mothers. Among biological fathers, household food
insecurity (OR=0.02, 95%CI: 0.001-0.14) and maternal drinking during pregnancy (OR=43.33,
95%CI: 24.39, 98.70) were associated with breastfeeding initiation. Living in a community with
92
the Canada Prenatal Nutrition Program showed a trend towards significance in terms of
increasing the odds of beginning breastfeeding amongst adoptive parents (OR=4.04, 95%CI:
0.99-17.54).
INTERPRETATION
We estimate 67% of Inuit children ages 3 to 5 years living in Nunavut during 2007 and 2008
began breastfeeding as infants. This is consistent with Inuit-specific estimates from the ACS
(2006) (67%)5 and the Aboriginal Peoples’ Survey (2001) (67%)
31. The close similarity in
breastfeeding initiation rates between the Aboriginal Peoples’ Survey31
, which collected
breastfeeding data from Inuit children up to 15 years of age in 2001 (i.e. born between 1986 and
2001), the Aboriginal Children’s Survey (2006)5, which collected data from children up to 6
years of age, and our data, seems to suggest that breastfeeding initiation has remained fairly
stable over time in the Inuit population. Information from individual birth cohorts through
continued surveillance are necessary to more clearly illustrate if temporal trends in initiation are
stagnating or if they are following similar increases to the general Canadian population 43
.
We note considerable variation in breastfeeding initiation depending on the type of caregiver
providing the report, and this may reflect the influence of different family conditions and
processes on children that remain with at least one biological parent versus others. Children
staying with their biologic mothers at the time of the interview were the most likely to begin
breastfeeding, while children raised by adoptive parents were the least likely to begin
breastfeeding. Our estimates among biological mothers are consistent with estimates from the
ACS (2006) (76%)44
when such estimates also considered only those children being raised by
their biological mother. Notwithstanding, Inuit biological mothers still have a substantially
lower rate of breastfeeding initiation compared to biological mothers in the general Canadian
population (76% vs. 90%) over a similar time period 2.
The appreciable difference in breastfeeding initiation between adoptive parents and biological
mothers echoes findings from earlier, community based studies 17,45
. Adoption is quite common
among Inuit (18% in N-ICHS, 17-42% 45-47
previously published estimates), and literature
suggests many of these adoptions are traditional (i.e. within the family and informal) although no
formal estimates exist48
. Traditional adoption, particularly if it occurs within the context of close
93
family members, affords an opportunity for the infant to receive breast milk directly from the
biological mother, although the acceptability of this approach is not known. Some adopted
children did receive breastfeeding in our sample and it is unclear, given our data, how this
occurred. Timing of adoption was not collected on the N-ICHS, and it may be that these infants
who began breastfeeding were raised by their biological mother during early infancy and then
subsequently given up for adoption. Alternatively, children may have been raised by an adopted
parent from birth and received breast milk by other means, perhaps directly or indirectly (e.g.
pumping) from the biological mother. Protecting, promoting and supporting breastfeeding among
all Inuit children, not just those raised by a biologic parent, will entail the exploration of feasible,
safe and culturally acceptable alternatives (e.g. breast milk banks). Discussions with adoptive
parents who report their child was breastfed could help to inform these alternatives.
We noted that factors associated with breastfeeding initiation differed depending on the child’s
primary caregiver, supporting our decision to divide our sample into caregiver subgroups.
Among the predictors tested in these models, birth weight was the most significantly associated
with breastfeeding initiation for children raised by their biological mothers, supporting other
Canadian49
and international research 50,51
. Similarly, finding that maternal smoking during
pregnancy is associated with a lower likelihood of breastfeeding initiation is also consistent with
other research.52,53
. Our finding that household crowding was associated with an increased
likelihood of beginning breastfeeding among biological mothers is more novel. Information on
parity was not available in the survey, and it could be that crowded households were a marker for
previous breastfeeding experience, a factor that has previously been associated with
breastfeeding practices54-56
. Including measures of household crowding with other measures of
parity in multivariable models may help to further elucidate the mechanism by which this
association arose.
Community factors, such as having access to the Canada Prenatal Nutrition Program and living
in a community where mothers are more likely to seek prenatal care, were not related to
beginning breastfeeding in either biological mothers or fathers. The finding that prenatal
nutrition programs have no effect on breastfeeding history of infants remaining with biological
parents may seem contrary to a previous program review 57
. However, we lacked information
about individual participation in these programs and data show 70% of women with access to
the program elect to participate49
. Community factors did demonstrate a trend towards
94
association with breastfeeding history among children raised by adoptive parents, both in
multivariable logistic and random effects models. These results suggest that public education and
nutrition programs may be a positive influence on breastfeeding initiation in this vulnerable
subgroup, and more research to confirm this is warranted. Because we did not have information
on the timing of adoption, and the Canada Prenatal Nutrition Program’s clientele focuses on
‘high-risk’ mothers and infants58
, it is not clear if these infants and mothers participated in the
program before or after being adopted.
In terms of children raised by a biological father, we find that children in food insecure homes
were less likely to begin breastfeeding. This finding was not replicated in other caregiver
subgroups and is contrary to findings reported in research of biological mothers only 24
. Specific
Aboriginal groups in Canada, including the Inuit, are disproportionately vulnerable to being food
insecure 23,26
; confirming our finding in our larger sample may identify a group of infants
particularly at risk for not being breastfed. We also note that maternal drinking during
pregnancy was positively related to beginning breastfeeding, a finding that seems to be relatively
disparate with research in biological mothers 20
. This finding may be spurious – we only have
nine biological fathers reporting any maternal drinking during pregnancy. Large confidence
intervals and high magnitude estimates question the stability of these measures. We recommend
further research with larger sample sizes to correctly interpret these findings.
Finally, while the validity of maternal recall of breastfeeding initiation has been studied
previously28
, these data also relied on recall of the biological father and adoptive parents which
has not been validated. Because the majority of Inuit women give birth away from their
communities and families25
, we cannot discount the possibility that the mother did begin
breastfeeding at the hospital but stopped before or at discharge and biological fathers may not be
able to accurately speak to breastfeeding histories. The same concern would arise for adopted
parents, and the problem is even further compounded if the child was adopted well after birth.
These adoptive parents would have to rely on the accuracy and veracity of the biological mothers
claims. However, it should be noted that the overall combined prevalence estimates of
breastfeeding initiation from the N-ICHS are consistent with those reported in the ACS (2006)5
and APS (2001)31
, which also relied on recall of breastfeeding histories from the primary
caregiver (i.e. not necessarily the biological mother).
95
Limitations
Previous research in Canadian populations indicate several factors consistently associated with
initiating breastfeeding (e.g. education 2,59-61
, age 2,60,61
, marital status 60
). Since this
information was only available in the few instances where the mother also completed the IPY
Inuit Health Survey (i.e. the concurrent health survey for adults aged 18 years or older), we were
not able to include these variables in these analyses. Also, many of the individual characteristics
of interest (e.g. household crowding) reflect the status of participants at the time the survey was
conducted rather than the prenatal and perinatal period most relevant to breastfeeding initiation.
No data on the age at which children were adopted by other caregivers was available. A final
limitation is sample size restrictions in the community, which may have reduced our study
power.
Conclusions
Beginning breastfeeding is the first step to achieving exclusive breastfeeding to six months and
continued breastfeeding into the second year of life. Placing the results of this study in the
context of population-based surveys conducted in 2001 (APS) and 2006 (ACS), suggest that
breastfeeding initiation in Canadian Inuit is not following similar increases to the general
Canadian population. To improve breastfeeding initiation among Canadian Inuit, public health
interventions should protect, promote and support breastfeeding not only amongst biological
caregivers, but also explore feasible and acceptable options to increase the breastfeeding of
children who are adopted sooner or later after birth. Our research suggests different strategies
may be required to improve breastfeeding initiation, depending on the child’s primary caregiver.
96
LITERATURE CITED
1. Kramer MS, Guo T, Platt RW, et al. Infant growth and health outcomes associated with 3
compared with 6 months of exclusive breastfeeding. Am J Clin Nutr. 2003;78(2):291-
295.
2. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
3. World Health Organization. The World Health Organization's infant feeding
recommendation. 2001;
http://www.who.int/nutrition/topics/infantfeeding_recommendation/en/index.html.
Accessed June 20, 2009.
4. Health Canada. Exclusive Breastfeeding Duration: 2004 Health Canada
Recommendation. 2004; http://www.hc-sc.gc.ca/fn-an/alt_formats/hpfb-
dgpsa/pdf/nutrition/excl_bf_dur-dur_am_excl-eng.pdf. Accessed June 3, 2012.
5. Smylie J, Adomako P. Indigenous Children's Health Report: Health assessment in action.
2009; http://www.stmichaelshospital.com/crich/wp-content/uploads/ichr_report-web.pdf.
Accessed January 31, 2013.
6. Luo Z-C, Senecal S, Simonet F, Guimond E, Penney C, Wilkins R. Birth outcomes in the
Inuit-inhabited areas of Canada. CMAJ. 2010;182(3):235-242.
7. Collins SA, Surmala P, Osborne G, et al. Causes and risk factors for infant mortality in
Nunavut, Canada 1999-2011. BMC Pediatrics. 2012;12(1):190.
8. Kramer A, McCullough D. The prevalence of otitis media with effusion among Inuit
children. International Journal of Circumpolar Health. 1998;57(s1):265-267.
9. Acuin J. Chronic suppirative otits media: burden of illness and management options.
2004; http://www.who.int/pbd/deafness/activities/hearing_care/otitis_media.pdf.
Accessed Jan 8, 2008.
10. Gurwith M, Wenman W, Gurwith D, Brunton J, Feltham S, Greenberg H. Diarrhea
among Infants and Young Children in Canada: A Longitudinal Study in Three Northern
Communities. The Journal of Infectious Diseases. 1983;147(4):685-692.
11. Dallaire F, Dewailly E, Vézina C, Bruneau S, Ayotte P. Portrait of outpatient visits and
hospitalizations for acute infections in Nunavik preschool children. Canadian Journal of
Public Health. 2006;97(5):362-368.
12. Holman RC, Parashar UD, Clarke MJ, Kaufman SF, Glass RI. Trends in Diarrhea-
associated Hospitalizations Among American Indian and Alaska Native Children, 1980–
1995. Pediatrics. 1999;103(1):e11.
97
13. Bachrach VR, Schwarz E, Bachrach LR. Breastfeeding and the risk of hospitalization for
respiratory disease in infancy: a meta-analysis. Archives of pediatrics & adolescent
medicine. Mar 2003;157(3):237-243.
14. Chien P, Howie P. Breast milk and the risk of opportunistic infection in infancy in
industrialized and non-industrialized settings. Advances in nutritional research.
2001;10:69.
15. Ip S, Chung M, Raman G, Trikalinos TA, Lau J. A summary of the Agency for
Healthcare Research and Quality's evidence report on breastfeeding in developed
countries. Breastfeeding Medicine. 2009;4(S1):17-30.
16. Hauck FR, Thompson JMD, Tanabe KO, Moon RY, Vennemann MM. Breastfeeding and
Reduced Risk of Sudden Infant Death Syndrome: A Meta-analysis. Pediatrics. 2011.
17. Dufour R. Breast feeding and adoption among the Inuit population of Northern Quebec.
Can J Publ Health. 1984;75(2):137-140.
18. Stewart P, Steckle J. Breastfeeding among Canadian Indians on-reserve and women in
the Yukon and NWT. Canadian journal of public health. Revue canadienne de santé
publique. 1987;78(4):255.
19. McDonald SD, Pullenayegum E, Chapman B, et al. Prevalence and Predictors of
Exclusive Breastfeeding at Hospital Discharge. Obstetrics & Gynecology.
2012;119(6):1171.
20. Clifford TJ, Campbell MK, Speechley KN, Gorodzinsky F. Factors Influencing Full
Breastfeeding in a Southwestern Ontario Community: Assessments at 1 Week and at 6
Months Postpartum. J Hum Lact. 2006;22(3):292-304.
21. Ryan AS, Zhou W, Gaston MH. Regional and Sociodemographic Variation of
Breastfeeding in the United States, 2002. Clinical Pediatrics. November/December 2004
2004;43(9):815-824.
22. Kelly YJ, Watt RG, Nazroo JY. Racial/Ethnic Differences in Breastfeeding Initiation and
Continuation in the United Kingdom and Comparison With Findings in the United States.
Pediatrics. November 2006 2006;118(5):e1428-e1435.
23. Huet C, Rosol R, Egeland G. The prevalence of food insecurity is high and the diet
quiality poor in Inuit communities. J Nutr. 2012;142(541-547).
24. Saha KK, Frongillo EA, Alam DS, Arifeen SE, Persson LA, Rasmussen KM. Household
Food Security Is Associated with Infant Feeding Practices in Rural Bangladesh. J. Nutr.
2008;138(7):1383-1390.
25. Douglas V. Childbirth among the Canadian Inuit: a review of the clinical and cultural
literature. International Journal of Circumpolar Health. 2006;64(2):117-133.
98
26. Egeland GM, Pacey A, Cao Z, Sobol I. Food insecurity among Inuit preschoolers:
Nunavut Inuit Child Health Survey, 2007-2008. CMAJ. 2010;182(3):243-248.
27. Saudny H, Leggee D, Egeland G. Design and methods of the Adult Inuit Health Survey
2007–2008. International Journal of Circumpolar Health.
2012;71:10.3402/ijch.v3471i3400.19752.
28. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
29. Panagiotakos DB, Papadimitriou A, Anthracopoulos MB, et al. Birthweight, breast-
feeding, parental weight and prevalence of obesity in schoolchildren aged 10-12 years, in
Greece; the Physical Activity, Nutrition and Allergies in Children Examined in Athens
(PANACEA) study. Pediatr. Int. 2008;50:563-568.
30. Kogan MD, Singh GK, Dee DL, Belanoff C, Grummer-Strawn LM. Multivariate analysis
of state variation in breastfeeding rates in the United States. Am J Public Health.
2008;98(10):1872-1880.
31. Guevremont A, Kohen D. Inuit in Canada: Findings from the Aboriginal Peoples' Survey
- Survey of Living Conditions in the Arctic. The Health of Inuit Children: Fact Sheet.
2007; http://www.statcan.gc.ca/pub/89-627-x/89-627-x2007002-eng.pdf. Accessed June
3, 2012.
32. Heck K, Braveman P, Cubbin C, Chavez G, Kiely J. Socioeconomic status and
breastfeeding initiation among California mothers. Public Health Rep. 2006;121(1):51-
59.
33. Bailey C, Pain R. Geographies of infant feeding and access to primary health-care.
Health Soc Care Comm. 2001;9(5):309-317.
34. Cabassa LJ. Measuring Acculturation: Where We Are and Where We Need to Go.
Hispanic J Behav Sci. 2003;25(2):127-146.
35. Lara M, Gamboa C, Kahramanian MI, Morales LS, Hayes Bautista DE. Acculturation
and Latino Health in the United States: A Review of the Literature and its Sociopolitical
Context. Annu Rev Publ Health. 2005;26(1):367-397.
36. Bickel G, Nord M, Prince C, Hamilton W, Cook J. Measuring Food Security in the
United States. 2000; http://www.fns.usda.gov/fsec/files/fsguide.pdf. Accessed June 2,
2012.
37. Indian and Northern Affairs Canada. Nutrition and food security in Kugaaruk, Nunavut:
baseline survey for the food mail pilot project. 2003; http://www.aadnc-
aandc.gc.ca/DAM/DAM-INTER-HQ/STAGING/texte-
text/kg03_1100100035822_eng.pdf. Accessed June 2, 2012.
99
38. O'Donnell V, Tait H. Aboriginal Peoples' Survey 2001 - initial findings: well-being of the
non-reserve Aboriginal Population. 2003; http://www.statcan.gc.ca/pub/89-589-x/89-589-
x2003001-eng.pdf. Accessed March 3, 2010.
39. Thomas T, Greeley G, Guyot M, Loubert K, Banfield W. Inuit Wellness Programs in
Nunavut: 2004-2005. 2006; http://pubs.aina.ucalgary.ca/health/61935.pdf. Accessed May
24, 2012.
40. Statistics Canada. Survey methods and practices. Ottawa: Minister of Industry; 2010.
41. SAS Institute Inc. Introduction to Survey Sampling and Analysis Procedures. SAS/STAT
9.2 User's Guide. Cary, North Carolina: SAS Institute Inc.; 2008.
42. Hosmer D, Lemeshow S. Applied Logistic Regression. 2nd ed. Toronto: John Wiley and
Sons, Inc; 2000.
43. Health Canada. Trends in Breastfeeding practices in Canada (2001-2008). 2010;
http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/prenatal/trends-tendances-
eng.php. Accessed May 25, 2012.
44. Asuri S, Ryan A, Arbour L. Early Inuit Child Health in Canada: Breastfeeding among
Inuit in Canada. 2011; https://www.itk.ca/sites/default/files/2011-Report-Breastfeeding-
among-Inuit-in-Canada.pdf. Accessed December 12, 2012.
45. Jenkins A, Gyorkos T, Joseph L, et al. Risk factors for hospitalization and infection and
Canadian Inuit infants over the first year of life - a pilot study. Int J Circumpolar Health.
2004;63(1):61-70.
46. Banerji A, Bell A, Mills EL, et al. Lower respiratory tract infections in Inuit infants on
Baffin Island. CMAJ. 2001;164(13):1847-1850.
47. Muckle G, Ayotte P, Dewailly E, Jacobson S, Jacobson J. Determinants of
polychlorinated biphenyls and methylmercury exposure in Inuit women of childbearing
age. Environ Health Perspect. 2001;109(9):7.
48. Pauktuutit Inuit Women of Canada. The Inuit way: A guide to Inuit culture. 2006;
http://www.pauktuutit.ca/pdf/publications/pauktuutit/InuitWay_e.pdf. Accessed
November 28, 2011.
49. Simard I, O'Brien H, Beaudoin A, et al. Factors influencing the initiation and duration of
breastfeeding among low-income women followed by the Canada prenatal nutrition
program in 4 regions of Quebec. J Hum Lact. 2005;21(3):327-337.
50. England L, Brenner R, Bhaskar B, et al. Breastfeeding practices in a cohort of inner-city
women: the role of contraindications. BMC Public Health. 2003;3(1):28.
51. Wagner C, Hulsey T, Southgate W, Annibale DJ. Breastfeeding Rates at an Urban
Medical University after Initiation of an Educational Program. Southern Med J.
2002;95(8):909-913.
100
52. Weiser TM, Lin M, Garikapaty V, Feyerharm RW, Bensyl DM, Zhu B-P. Association of
Maternal Smoking Status With Breastfeeding Practices: Missouri, 2005. Pediatrics.
2009;124(6):1603-1610.
53. Di Napoli A, Di Lallo D, Pezzotti P, Forastiere F, Porta D. Effects of parental smoking
and level of education on initiation and duration of breastfeeding. Acta Pædiatrica.
2006;95(6):678-685.
54. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
55. Martens PJ. Increasing breastfeeding initiation and duration at a community level: an
evaluation of Sagkeeng First Nation’s community health nurse and peer counselor
programs. Journal of Human Lactation. 2002;18(3):236-246.
56. Scott J, Binns C. Factors associated with the initiation and duration of breastfeeding: a
review of the literature. Breastfeed Rev. 1999;7:5 - 16.
57. Anderson N, Milne D, Martin T, et al. Evaluation of the Canada Prenatal Nutrition
Program in First Nations communities. 2003;
http://publications.gc.ca/site/eng/290477/publication.html. Accessed March 22, 2011.
58. Health Canada. First Nations and Inuit Health: Funded Health Programs and Services.
2008; http://www.hc-sc.gc.ca/fniah-spnia/finance/agree-accord/prog/index-eng.php.
Accessed February 15, 2013.
59. Black R, Godwin M, Ponka D. Breastfeeding among the Ontario James Bay Cree: a
retrospective study. Can J Public Health. 2008;99(2):98-101.
60. Dubois L, Girard M. Social Determinants of initiation, duration and exclusivity of
breastfeeding at the population level: the results of the Longitudinal Study of Child
Development in Quebec (ELDEQ 1998-2002). Can J Public Health. 2003;94(3):300-305.
61. Callen J, Pinelli J. Incidence and duration of breastfeeding for term infants in Canada,
United States, Europe and Australia: a literature review. Birth. 2004;31(4):285-292.
101
Figure 1: Survey response patterns, Nunavut Inuit Child Heath Survey
* Number of participants actually excluded may be less than those with missing data due to some overlap between categories
102
Table 1: Unweighted household, maternal, infant and community characteristics of study respondents,
by caregiver subgroup, Nunavut Inuit Child Health Survey (n=361)
Number affirmative/Number responding (% affirmative)a
Characteristics
Biological
Mothers
(n=257)
Biological
Fathers
(n=40)
Adoptive
Parents
(n=64)
p-value
Breastfeeding initiation 195/257 (76) 24/40 (60) 25/64 (39) p<0.01
Household Characteristics
Speaks Inuit dialect at home 121/256 (48) 14/39 (36) 31/64 (48) p=0.36
Household crowding 121/257 (53) 24/40 (50) 36/64 (52) p=0.96
Receives income support 101/203 (50) 14/34 (41) 19/52 (37) p=0.18
Household food insecurity 176/257 (69) 28/40 (70) 48/64 (75) p=0.59
Maternal Characteristics
Mother drank during pregnancy 55/247 (22) 9/40 (25) 15/48 (31) p=0.40
Mother smoked during pregnancy 216/256 (84) 31/39 (79) 49/60 (82) p=0.69
Infant Characteristics
Infant birth weight <2500 grams 22/251 (9) 2/38 (5) 7/59 (12) p=0.53b
Community Characteristics
Region – Baffin 112/257 (44) 25/40 (62) 27/64 (42) p=0.23b
- Kitikmeot 92/257 (36) 10/40 (25) 22/64 (34)
- Kiviliq 53/257 (21) 5/40 (12) 15/64 (23)
Canada Prenatal Nutrition Program 142/257 (55) 23/40 (58) 38/64 (59) p=0.83
Birthing Centre 36/257 (16) 8/40 (20) 13/64 (20) p=0.69
Frequency of Prenatal care – highc
121/257 (47) 16/40 (40) 38/64 (59) p=0.11
Frequency of Preventive care – highc
136/257 (53) 21/40 (52) 37/64 (58) p=0.77 a Percentages exclude missing values b Exact tests used to estimate p-values, due to small cell size (≤5) c(≥4 visits per live birth) d (≥ 0.2 visit per community member)
103
a Weighted population estimates may not be exactly equal to column totals, due to rounding b(≥4 visits per live birth) c (≥ 0.2 visit per community member)
Table 2: Weighteda household, maternal, infant and community characteristics of study respondents, by caregiver subgroup.
Nunavut Inuit Child Health Survey (n=1088)
Percent (%) with affirmative response
Characteristics
Biological
Mothers
(n=795)
Biological
Fathers
(n=126)
Adoptive Parents
(n=167) p-value
Breastfeeding initiation 76.3 60.4 33.5 p<0.01
Household Characteristics
Speaks Inuit dialect at home 54.5 62.4 53.5 p=0.65
Household crowding 48.5 48.3 48.8 p=1.00
Receives income support 51.8 57.2 67.5 p=0.17
Household food insecurity 67.0 75.8 74.8 p=0.36
Maternal Characteristics
Mother drank during pregnancy 22.3 29.5 27.7 p=0.56
Mother smoked during pregnancy 83.0 82.0 79.0 p=0.81
Infant Characteristics
Infant birth weight < 2500 grams 8.9 5.0 10.8 p=0.65
Community Characteristics
Region – Baffin 46.8 64.3 48.7 p=0.32
- Kitikmeot 37.7 25.9 32.1
- Kiviliq 15.4 9.8 19.1
Canada Prenatal Nutrition Program 44.7 46.7 47.6 p=0.92
Birthing centre 21.7 28.2 28.1 p=0.55
Frequency of Prenatal care – highb
34.5 23.4 46.7 p=0.05
Frequency of Preventive care – highc
40.5 36.2 47.9 p=0.48
104
Table 3: Weighted, bivariable prevalence odds ratios (pOR) and 95% confidence intervals (95% CI) for tests of association between selected household, maternal, infant and community characteristics
with breastfeeding initiation amongst Canadian Inuit, by caregiver subgroup. Nunavut Inuit Child Health Survey (n=1088)
Biological Mothers (n=795) Biological Fathers (n=126) Adoptive Parents (n=167)
Characteristics
Group size
na =795
Initiated breastfeeding
(n=604)
na, %b
Bivariable prevalence
Odds Ratios
(pOR, 95% CIc)
Group size
na =126
Initiated breastfeeding
(n=77)
na , %b
Bivariable prevalence
Odds Ratios
(pOR, 95% CIc)
Group size
na=167
Initiated breastfeeding
(n=56)
na , %b
Bivariable prevalence
Odds Ratios
(pOR, 95% CIc)
Household Characteristics
Primary language spoken at home
Inuktitut or Inuinnaqtun 429 322 (75) 1.00 79 48 (61) 1.00 89 32 (35) 1.00
Other 366 282 (77) 1.11 (0.59-2.10) 47 28 (61) 0.98 (0.21-4.52) 78 24 (31) 0.83 (0.26-2.60)
Household overcrowding
No 389 284 (73) 1.00 f 61 37 (60) 1.00 82 25 (31) 1.00
Yes 405 320 (79) 1.44 (0.76-2.60) 65 40 (61) 1.08 (0.23-5.01) 85 31 (36) 1.26 (0.43-3.71)
Receives income support
No 356 279 (78) 1.00 f 72 47 (65) 1.00 113 39 (34) 1.00
Yes 439 323 (73) 0.76 (0.39-1.55) 54 29 (53) 0.65 (0.15-3.13) 54 16 (29) 0.81 (0.26-2.76)
Household food security
Food secure 270 205 (76) 1.00 30 26 (86) 1.00d 42 15 (35) 1.00
Food insecure 525 399 (76) 1.03 (0.51-2.06) 98 52 (53) 0.19 (0.05-0.75) 126 41 (33) 0.83 (0.26, 2.64)
Maternal Pregnancy Behaviour
Drinking during pregnancy
No 621 481 (78) 1.00 f 89 42 (47) 1.00d 121 38 (31) 1.00
Yes 173 124 (72) 0.71 (0.34-1.50) 37 36 (97) 36.29 (4.14-317.7) 46 18 (39) 1.42 (0.30-6.59)
Smoking during pregnancy
No 137 121 (89) 1.00 e 23 14 (60) 1.00 35 16 (45) 1.00
Yes 657 484 (74) 0.36 (0.13-1.02) 103 64 (62) 1.13 (0.19-6.68) 132 41 (31) 0.56 (0.14-2.33)
Infant Characteristics
Infant birth weight
≥ 2500 grams 724 483(78) 1.00d 119 72 (60) 1.00 148 47 (32) 1.00
< 2500 grams 71 35 (49) 0.26 (0.10-0.67) 7 5 (73) 1.63 (0.09-30.8) 18 9 (50) 1.99 (0.52-7.56)
Community Characteristics
Region
Baffin 373 276 (74) 1.00 81 44 (54) 1.00 f 81 31 (38) 1.00
Kitikmeot 229 233 (78) 1.23 (0.60-2.51) 33 23 (71) 2.08 (0.31-14.13) 54 14 (26) 0.58 (0.16-2.02)
Kivalliq 122 95 (78) 1.23 (0.53-2.84) 13 9 (73) 2.31 (0.88-6.06) 32 11 (34) 0.84 (0.28-2.54)
Canada Prenatal Nutrition Program
No 437 331 (76) 1.00 67 40 (59) 1.00 87 14 (16) 1.00d
Yes 357 274 (77) 1.04 (0.56-1.95) 57 37 (62) 1.13 (0.26-4.82) 80 42 (52) 5.68 (1.74-17.15)
Community hospital or birthing center
No 621 477 (77) 1.00 90 59 (66) 1.00 120 48 (40) 1.00 e
Yes 173 128 (74) 0.81 (0.36-1.84) 36 17 (47) 0.46 (0.08-2.78) 47 8 (17) 0.31 (0.09-1.10)
Table 3 continues
105
Table 3 continued
Characteristics
Biological Mothers (n=795) Biological Fathers (n=126) Adoptive Parents (n=167)
Group
size
n=795*
Initiated
breastfeeding
(n=604)
N*, %b
Bivariable
prevalence
Odds Ratios (pOR, 95% CI‡)
Group
size
n=126*
Initiated
breastfeeding
(n=77)
N*, %†
Bivariable
prevalence
Odds Ratios (pOR, 95% CI‡)
Group
size
n=167*
Initiated
breastfeeding
(n=56)
N*, %†
Bivariable
prevalence
Odds Ratios (pOR, 95% CI‡)
Frequency of prenatal care in community
Low (<4 visits per live birth) 522 393 (75) 1.00 96 57 (59) 1.00 89 16 (18) 1.00d
High (≥4 visits per live birth) 273 211 (77) 1.10 (0.56-1.93) 30 19 (62) 0.91 (0.23-3.65) 78 40 (51) 4.16 (1.45-14.51)
Frequency of preventive care in community
Low (<0.2 visits per community member) 466 361 (77) 1.00 80 46 (57) 1.00 87 17(20) 1.00d
High (≥ 0.2 visit per community member) 329 244 (74) 0.87 (0.48-1.68) 46 30 (66) 1.43 (0.37-5.52) 80 39 (49) 3.84 (1.23-12.28)
a Group size may not equal 795 due to rounding of weighted estimates
b Percent calculations may not be equal to those calculated by hand, due to rounding c 95% Confidence limits were estimated using the Wald method d p<0.05, Wald-test; e p<0.10 Wald test, f p<0.25 Wald test
106
Table 4: Weightedd multivariable prevalence odds ratios (pOR) and 95% confidence intervals (95%CI) of factors
associated with breastfeeding initiation, by caregiver subgroup. Nunavut Inuit Child Health Survey (n=1088)
Biological mothers
(n=795)
Biological fathers
(n=126)
Adoptive parents
(n=167)
Characteristics
Multivariable
prevalence odds ratio
(pOR, 95% CIa)
Multivariable
prevalence odds ratio
(pOR, 95% CIa)
Multivariable
prevalence odds ratio
(pOR, 95% CIa)
Household Characteristics
Household crowding 1.81 (0.86-3.82)c
- -
Household food insecurity 0.10 (0.01-0.91)b
Maternal Pregnancy Behaviour
Drinking during pregnancy - 42.09 (4.13-428.58)b
-
Smoking during pregnancy 0.37 (0.10-1.38)c
- -
Infant Characteristics
Infant Birth weight <2500 grams 0.19 (0.07-0.55)b
Community Characteristics
Canada Prenatal Nutrition Program - - 4.13 (1.04-16.41)b
Frequency of Prenatal care – highe
- - 3.11 (0.72-13.35)c
a 95% Confidence limits were estimated using the Wald method bp<0.05 Wald test, cp<0.15, Wald test d weighted to the population aged 3 to5 years at the time of data collection e (≥4 visits per live birth)
107
a p<0.05, b p<0.10, c p<0.15
d (≥4 visits per live birth)
Table 5: Fixed, prevalence odds ratios (pOR), 95% confidence intervals (95%CI) and random effect estimates from
hierarchical linear model of breastfeeding initiation, by caregiver subgroup. Nunavut Inuit Child Health Survey (n=1088)
Biological mothers
(n=795)
Biological fathers
(n=126)
Adoptive parents
(n=167)
Characteristics
Multivariable
Fixed effects odds
ratio
(pOR, 95% CI)
Multivariable
Fixed effects odds
ratio
(pOR, 95% CI)
Multivariable
Fixed effects odds
ratio
(pOR, 95% CI)
Household Characteristics
Household crowding 2.01 (1.39-2.94)a - -
Household food insecurity - 0.02 (0.001, 0.14) a -
Maternal Pregnancy Behaviour
Drinking during pregnancy - 43.33 (24.39-98.70) a -
Smoking during pregnancy 0.34 (0.19-0.61) a - -
Infant Characteristics
Infant Birth weight <2500 grams 0.16 (0.10-0.27) a - -
Community Characteristics
Canada Prenatal Nutrition Program - 4.04 (0.99-17.54)b
Frequency of Prenatal care- highd - - 3.38 (0.81-14.14)
c
Random effects
Random intercept (SE) 0.27 (0.15) 9.36 (5.82) 1.19 (0.79)
108
Chapter 5 : Manuscript 2
5 Study Overview
The second manuscript fulfills the third and fourth objectives of this thesis: to describe
the distribution of exclusive breastfeeding duration among Canadian Inuit (Objective 3) and to
identify factors associated with exclusive breastfeeding as recommended (Objective 4) using the
Nunavut Inuit Child Health Survey.
109
5.1 Exclusive breastfeeding among Canadian Inuit: results from the Nunavut Inuit Child Health Survey
List of Contributing Authors: Kathryn E McIsaac, MSc1; Wendy Lou, PhD
1; Daniel Sellen,
PhD1,2
; T Kue Young, MD DPhil1
Author Affiliations: 1Dalla Lana School of Public Health, University of Toronto, Toronto
Canada; 2Department of Anthropology, University of Toronto, Toronto Canada
110
ABSTRACT
Background: Very little is known about the breastfeeding practices of Inuit Canadians. Our
research aims to identify the distribution of, and identify factors associated with, the duration of
exclusive breastfeeding in this population.
Methods: We use data from the Nunavut Inuit Child Health Survey, a cross-sectional,
population-based survey of Inuit children aged between 3 and 5 years. A series of multinomial
multivariate logistic regression models were run to identify factors associated with four different
exclusive breastfeeding durations (≤1 month, >1 to < 5.5 months, 5.5-6.5 months, and >6.5
months).
Results: Twenty three percent of infants received exclusive breastfeeding to six months, as
recommended (95% CI: 16.2-29.3). Many infants (61%) were exclusively breastfed for less than
5.5 months and 16% (95% CI: 10.9-22.0) of infants were exclusively breastfed for more than 6.5
months. Families receiving income support were less likely to stop exclusive breastfeeding
before the recommended duration (OR≤1month=0.39, 95%CI: 0.14-1.05; OR1-<5.5 months=0.34,
95%CI: 0.13, 0.85), relative to those not receiving income support. No other factors were
strongly related to exclusive breastfeeding duration.
Conclusions: There is substantial variation in exclusive breastfeeding amongst Canadian Inuit,
yet we were unable to identify many factors strongly predictive of exclusive breastfeeding
duration. Educating Inuit Canadians about optimal feeding practices, including messages about
the timely introduction of complementary foods into the diet, is recommended. At present, our
data do not provide strong enough evidence to support targeted breastfeeding campaigns to
particular population sub-groups.
111
BACKGROUND
Few Canadian mothers exclusively breastfeed their infants to six months, the current
recommendation of Health Canada and other health authorities1,2
. Population-based surveys
estimate between 14% 3 and 26%
4 practice exclusive breastfeeding to six months, yet,
differences exist across various Canadian population subgroups. Married women5, women with
more education5 and women living in the western-most provinces and some territories
3,5 are
more likely to meet the six month recommendation compared to other Canadians.
Presently, there are no population-based estimates of exclusive breastfeeding practices among
Inuit Canadians. We do know that Inuit mothers are more likely to practice any breastfeeding at
six months compared to other Canadians (58%6 compared to 53%
3), and the Canadian territory
of Nunavut, where approximately 84% of inhabitants are Inuit7, reports one of the highest rates
of exclusive breastfeeding to six months in all of Canada (19.7% in Nunavut vs. 14.4% in
Canada)3. However, with the current data, we cannot make conclusive comparisons about Inuit
and other Canadians’ exclusive breastfeeding practices.
Furthermore, no study has attempted to identify characteristics associated with exclusive
breastfeeding duration among Canadian Inuit. Living predominantly in the arctic and subarctic
regions of Canada8, the Inuit face unique challenges that may serve as barriers or promoters of
exclusive breastfeeding to six months. For example, a high prevalence of food insecurity9,10
may
or may not play a role in extending the duration of exclusive breastfeeding11-13
, while difficulties
accessing timely health care due to the remoteness of several communities may reduce the
duration of exclusive breastfeeding because of inadequate means of tangible support form health
care professionals14
. Exhibiting a more traditional way of life, could in itself, be more
predictive of exclusive breastfeeding duration in this population. Alternatively, factors
predictive of exclusive breastfeeding among Canadian Inuit may be similar to other Canadians.
The aim of the present research is to describe the distribution of, and to identify factors
associated with, exclusive breastfeeding duration among Canadian Inuit.
112
METHODS
This study uses data from the Nunavut Inuit Child Health Survey, a population-based, cross-
sectional survey conducted in Nunavut in the summer and fall of 2007 and 2008 (August to
November). The survey captured key aspects of health among preschool aged Inuit children
(aged 3 to 5 years), including information on maternal breastfeeding practices.
Nunavut is located in the northeastern-most region of Canada and covers a geographic area of
almost 2 million square kilometres15
. It has three administrative areas: Baffin, in the far north
and east; Kivalliq, in the southwest; and Kitikmeot, in the northwest. The territory has a
population of approximately 33,000 persons and is home to 50% of all of Canadian Inuit8.
There are 25 communities in Nunavut15
.
Sixteen of these 25 communities participated in the Nunavut Inuit Child Health Survey. The
remaining nine communities had a relatively small number of children relative to the costs of
study participation and were excluded for financial and logistical reasons fully described
elsewhere9. Inuit children, aged 3 to 5 years, speaking either English or Inuktitut, living in one
of these 16 communities were eligible to participate. For the present study, we included only
those children who received some breastfeeding from birth. We excluded children who were
adopted because previous research16-18
, including our own, as yet unpublished work, suggests
that adopted children are less likely to receive any breastfeeding.
Participants in the Nunavut Inuit Child Health Survey were selected in one of two ways. In the
first way, children of parents who were randomly selected for the IPY Inuit Health Survey, a
survey of Inuit adults occurring concurrently, and who met the eligibility criteria were invited to
participate. Methods for the IPY Inuit Health Survey are fully described elsewhere19
. In the
second way, children themselves were randomly selected using random number tables and
random number generators from patient rosters at the local community health centre. The survey
aimed to sample approximately 20% of children in each community, although higher sampling
fractions were applied in smaller communities and lower sampling fractions were applied in
larger communities.
About two weeks prior to survey administration, members of the research team traveled to
communities to begin recruitment. Caregivers of selected children were invited to participate.
Often times, invitations were extended by a bilingual member of the research team via a personal
113
home visit, but other methods were also employed to contact selected individuals (i.e. telephone
calls, hand delivered notification letters, community radio announcements). At least three
contact attempts were made, sometimes more when time permitted. Those who formally
consented to participate were invited to a local community centre (e.g. health facility) within two
weeks to complete an interviewer-led survey.
The duration of exclusive breastfeeding was measured in response to the question “how many
months was your child ‘exclusively’ breastfed or exclusively fed breast milk”. Approximately
58% of mothers are able to accurately recall the transition from exclusively breast feeding to at
least some bottle feeding within one month20,21
and between 65%22
and 88%21
are able to
accurately recall the timing of complementary food introduction into the diet within one month23
.
Accuracy of recall of timing of transition to solid foods was higher when the time period
between the behaviour and recall was shorter (i.e. less than one year21
compared to 14 to 15
years22
). On average, mothers were more likely to recall an earlier date of introduction of
formula (i.e. exclusive breastfeeding occurred for longer durations than reported) and a later date
of introduction to semi-solid foods (i.e. transition to complementary foods occurred sooner than
recalled)23
.
We identified several possible determinants of exclusive breastfeeding duration, most of which
were measured directly in the Nunavut Inuit Child Health Survey, including infant birth
weight24
, maternal practices during pregnancy (i.e. smoking, drinking) 25,26
, household
overcrowding 27
, receiving income assistance 28-30
, place of residence3,28,31
, access to health care
and other instrumental and instrumental support14
, and level of engagement with traditional
culture32-34
. We were also interested to learn whether scarcity of food resources affected
breastfeeding duration, so we included measures of household and childhood food security.
Lastly, we were interested in the effect of smoking restrictions in the home (yes or no);
identifying an association may have important public health implications relating to infant
respiratory infection 35,36
.
The single indicator variable, primary language spoken at home (Inuit dialects versus other), was
used as a proxy measure for level of engagement with traditional culture. This measure has been
used previously as an indicator of traditional knowledge in other Inuit research37
. Access to
health care was measured using four community level indicator variables: living in a community
with a birthing centre, living in a community with the Canada Prenatal Nutrition Program
114
(CPNP), the frequency of prenatal visits per live births per year, and the frequency of preventive
care visits per population per year. We obtained participation in the CPNP from reports released
on behalf of the Government of Nunavut38
. We obtained community-level health care utilization
data from a recent inventory of health care resources in Nunavut [personal communication:
Gregory Marchildon]. All other variables were extracted directly from the Nunavut Inuit Child
Health Survey.
Most of these indicator variables were collected as closed-ended questions and treated as
categorical measures in analyses. Household overcrowding was captured by dividing the
number of persons in a household by the number of rooms in that household. When there was
more than one person per room, the household was considered overcrowded, a definition
consistent with other Canadian research39,40
. Household and childhood food security were
measured using a modified version of the USDA Food Security Module41-43
, revised for
Canadian Aboriginal communities. Both household and childhood food security were
categorized as three-level variables (food secure, moderately food insecure, severely food
insecure), based on response patterns to the 18-item scale, fully described elsewhere 9. Lastly,
the proportions of prenatal and preventive care visits in the community each year were
dichotomized at the median.
Statistical analyses were conducted in SAS version 9.3 and were weighted to the probability of
participant selection. We used complex survey procedures (e.g. SURVEYFREQ,
SURVEYLOGISITC) to account for the sampling procedure, unless otherwise noted. We
accounted for the sub-population (i.e. only those who initiated breastfeeding) in our variance
estimation through use of the DOMAIN statement44,45
. Individuals with item non-response for
any predictor variable were removed from the data set.
We used multinomial logistic regression to model the odds of four different exclusive
breastfeeding durations: one month or less, more than one month to less than 5.5 months, 5.5
months to 6.5 months, and more than 6.5 months. We considered anyone reporting exclusive
breastfeeding between 5.5 and 6.5 months as exclusive breastfeeding as recommended. We
chose a range of values between 5.5 and 6.5 to account for variation in infant development that
would lead to either earlier or later introduction of complementary foods into the infant’s diet.
Health Canada’s indicators for readiness to receive complementary foods include: having
adequate head control, being able to sit up and lean forward, being able to turn the head away
115
from the breast or bottle when full, and picking up and putting food directly in the mouth46
.
Sensitivity analysis, modifying the boundaries of the outcome category ‘exclusive breastfeeding
as recommended’ were also conducted.
We first tested a series of bivariate regression models. Factors exhibiting at least a weak
association with exclusive breastfeeding (p<0.25) were candidates for the multivariate
multinomial model. All candidate variables were then tested in a multivariate model. We used a
forwards stepwise model building strategy, adding the most significant variables sequentially as
calculated by likelihood ratio tests, until all variables remaining in the model met our entry and
exit criteria (pentry=0.25, pexit=0.30), consistent with recommendations for building exploratory
predictive models47
.
RESULTS
Figure 1 illustrates the number of participants in the Nunavut Inuit Child Health Survey eligible
at each phase of the study and included in these analyses. Of the 388 participants who
completed the survey, 219 met the eligibility criteria. There were 188 participants who had
complete item response for factors of interest, representing 752 infant-mother dyads after
applying population weights.
Table 1 presents the unweighted and weighted characteristics of study respondents. Most of the
individual characteristics exhibited similar unweighted and weighted frequencies in our sample,
with the exception of childhood food security, which increased after applying population
weights. Community characteristics (e.g. having the prenatal nutrition program), tended to
decrease after applying population weights, suggesting these features were most common in
smaller communities.
The median duration of exclusive breastfeeding was 4.3 months. Figure 2 illustrates the
distribution of exclusive breastfeeding in our sample. Exclusive breastfeeding as recommended
(i.e. 5.5 to 6.5 months) was practiced by 22.7% of the respondents (95% CI: 16.2-29.3).
Moreover, while the overwhelming majority of respondents (60.8%) practiced exclusive
breastfeeding for a short duration (≤1 month: 22.6%, 95% CI:16.0, 29.2; >1-<5.5 months:
38.2%, 95% CI: 30.8, 45.7), a substantial percentage (16.4%, 95% CI: 10.9-22.0) went beyond
the recommended six months.
116
Table 2 displays the weighted prevalence of select characteristics of interest, by exclusive
breastfeeding duration and Table 3 shows their bivariate associations with exclusive
breastfeeding duration. Those receiving income support had a lower odds of breastfeeding for a
short duration as opposed to exclusive breastfeeding as recommended, relative to those not
receiving income support (OR≤1 month=0.36, 95%CI: 0.14-0.94; OR>1-<5.5 months=0.34; 95% CI:
0.14-0.81). We identified four additional characteristics as candidates for our multivariable
model (i.e., p<0.25): speaking Inuit dialects in the household, imposing household smoking
restrictions, place of residence and access to the Canada Prenatal Nutrition Program.
Table 4 presents our multivariate adjusted risk estimates. Point estimates suggest that those
receiving income support were less likely to have been exclusively breastfed for a short, as
opposed to the recommended, duration (i.e. 5.5-6.5 months) relative to those who did not receive
income support (OR≤1 month=0.39, 95%CI: 0.14-1.05; OR>1-<5.5 months=0.34; 95% CI: 0.13-0.85),
in fully adjusted models. Languages spoken at home and living in a community with the Canada
Prenatal Nutrition Program added to the predictive ability of our model, but were not statistically
significant.
DISCUSSION
Our study provides up to date population-based estimates of the duration of exclusive
breastfeeding among Canadian Inuit who live in Nunavut, and identifies factors associated with
exclusive breastfeeding duration. We found that almost 23% of Inuit women who initiated
breastfeeding met the exclusive breastfeeding recommendation; 61% of Inuit women stopped
exclusive breastfeeding early and another 16% of Inuit women surpassed the recommended six
months.
Almost one quarter of Inuit women who initiated breastfeeding did not practice exclusive
breastfeeding beyond one month, a trend characteristic of other Canadian and western
populations 48-50
. Although there were many Inuit women who reported exclusive breastfeeding
to at least 5.5 months (39%) in our sample, a substantial portion of these (40% or 16% of the
total sample) went beyond the recommended six months: only 22.7% of Inuit Canadians reported
practicing exclusive breastfeeding as recommended. On the surface, this seems fairly consistent
with practices in other Canadians (14% and 26%)3,4
. However, these analyses excluded those
117
who did not initiate breastfeeding. Re-analyzing the data, we found that the prevalence of
exclusive breastfeeding, as recommended, in the entire sample of participants in the Nunavut
Inuit Child Health Survey is 14.2% (95% CI: 10.2-18.1, n=364) and 18.3% in reports provided
by biological mothers (95% CI: 13.0-23.6, n=289). These data suggest that the duration of
exclusive breastfeeding, as recommended, in Inuit Canadians is consistent with other Canadians.
Many Inuit mothers surpassed the recommended duration of exclusive breastfeeding. This high
proportion of mothers (16%) is a public health concern in its own right51
. Between ages six and
23 months, an infant requires complimentary feeding in addition to breast milk. Failing to
introduce complimentary foods when the infant is developmentally ready (approximately six
months) deprives them of essential nutrition and is associated with stunting52
, wasting53
, delayed
psychomotor and cognitive development in early childhood as well as other deleterious health
outcomes 54,55
. Our finding could be an accurate representation of the exclusive breastfeeding
duration of Canadian Inuit or it could be spurious – perhaps the result of caregivers
misinterpreting the definition of, or intentionally over-stating the duration of, exclusive
breastfeeding for reasons such as social desirability bias. Anthropometric measurements of pre-
school aged Inuit children in the Nunavut Inuit Child Health Survey do not suggest any stunting
or wasting between ages 3 and 5 years56
, indicating that some misclassification of exclusive
breastfeeding duration may have occurred. A follow-up study to clarify the perceptions and
beliefs about recommended breastfeeding practices among Inuit would help identify if inaccurate
beliefs about exclusive breastfeeding duration exist in these communities.
Our weighted sample of 752 participants identified only one factor, receiving income support,
that was strongly associated (p<0.05), with exclusive breastfeeding duration. We also identified
two additional characteristics weakly associated (p<0.25) with the practice: speaking Inuit
dialects at home - a surrogate for traditionalism - and community access to the Canada Prenatal
Nutrition Program.
We used income support as a surrogate for household income, and our finding that infants in
households’ receiving income support were less likely to be breastfed exclusively for a short
duration is generally inconsistent with research in other western populations (i.e. lower income
tends to be associated with shorter exclusive breastfeeding duration) 28,30
. While some research
suggests income itself is related to exclusive breastfeeding duration only when other, highly
correlated factors (e.g. education) are not adequately adjusted for57
, it seems unlikely that this
118
would explain our finding: we were not able to adjust for other socioeconomic variables and the
direction of the effect is opposite to what one would expect.
One explanation is mothers in households not receiving income support returned to work shortly
after giving birth, a known risk factor for early cessation of exclusive breastfeeding48,58,59
.
Information on individual employment histories, which were not available to us, would be
necessary to further study this hypothesis. Another possible explanation is mothers who received
income support may practice exclusive breastfeeding for a longer duration because of the
personal cost-savings the behaviour affords60-62
. Grocery prices in the North tend to be much
greater than in other regions in Canada63
; yet, household food security was not associated with
exclusive breastfeeding duration in our sample, which would be expected if those receiving
income support were motivated to continue exclusive breastfeeding for its cost-savings. In a
somewhat related vein, Inuit women who do not receive income support, and are of a higher
socioeconomic position, may choose formula feeding to reflect their status in society (i.e. they
can afford to pay for infant food), a phenomena that is observed in developing countries64-66
.
Alternatively, our finding could be related to measurement error and misclassification, wherein
our dichotomous income support variable did not accurately measure income in our sample or
was not sensitive enough to identify an inverse association between exclusive breastfeeding
duration and income. However, it seems unlikely that the latter is exclusively true since other
studies tend to demonstrate more of a dose-response effect of income on exclusive breastfeeding
duration 5. Lastly, this finding may have arisen by chance alone. Regardless, these data suggest
that more research is needed, with better measures of socioeconomic position, to understand its
relationship with exclusive breastfeeding practices among Inuit in northern Canada.
Identifying factors associated with exclusive breastfeeding can help inform public health
campaigns by establishing population sub-groups that are particularly at risk for early cessation
of exclusive breastfeeding for targeted messaging. We used a multinomial logistic regression to
model our data, enabling the identification of characteristics not only associated with a shorter
duration, but also with a longer duration of exclusive breastfeeding. Although, we were not able
to find any evidence of factors associated with exclusive breastfeeding other than receiving
income support, multinomial regression methods require a larger sample than standard logistic
regressions47
. We also performed a binary logistic regression for comparative purposes. These
models also indicate that receiving income support was associated with exclusive breastfeeding
119
as recommended (i.e. 5.5-6.5 months vs. other: OR= 0.45, 95% CI: 0.21, 0.98), and also suggest
that other factors such as speaking Inuit dialects at home and living in crowded homes were also
important. Combining these data with more recent data could increase sample size, allowing the
detection of other, potentially important factors. From a statistical perspective, a multinomial
logistic regression model is more efficient than multiple binary logistic models and incorporating
all the data into one model instead of three separate models ensures that parameters are
accurately estimated67
. From a public health perspective, modeling four different outcomes
instead of two can help identify different subpopulations at risk for suboptimal exclusive
breastfeeding practices at various time points.
Limitations
This cross-sectional survey collected information on participants’ exclusive breastfeeding
practices up to five years after giving birth. The determinants of exclusive breastfeeding used in
this study most often reflect the status of the child at the time of survey administration, and not at
time the child was receiving exclusive breastfeeding. We assume that many of these
characteristics are stable over the 3 to 5 year period, but we are unable to verify this inference.
Because we cannot demonstrate a clear temporal sequence between our exposures and exclusive
breastfeeding duration, causal discussions are limited with these data. Moreover, many of these
factors were measured by self-report, and may be subject to a higher degree of measurement
error and misclassification.
Another limitation is the measurement of exclusive breastfeeding duration. Maternal recall of
exclusive breastfeeding duration, as stated earlier, is 58% accurate within one month, and is most
often overstated 23
. We conducted a sensitivity analysis (not shown) extending the limits of
appropriate breastfeeding duration in various ways and still were not able to find strongly
significant predictors of exclusive breastfeeding duration.
Finally, we did not have data on some factors that have been found to be associated with
exclusive breastfeeding duration in other studies, such as maternal age68,69
, maternal education69
and maternal work history48
. Moreover, we did not have a direct measure of maternal income
and used receiving income supplement as a surrogate measure. Failing to control for these
important factors may have biased some of our effect estimates.
120
Conclusions
Promoting exclusive breastfeeding in Canada’s North may potentially save thousands of dollars
in health care costs during the first year of life70
. Our study is the first to provide population-
based estimates of exclusive breastfeeding in Canadian Inuit. Findings of this study support
continued public health programs (e.g. education, Canada Prenatal Nutrition Programs, social
marketing) to Canadian Inuit which include messages regarding the optimal duration of
exclusive breastfeeding. Qualitative research can help inform which of these strategies may be
most efficacious. We recommend these programs not only include messages about the benefits
of continued breastfeeding to six months, but also include messages about the importance of
introducing nutrient rich foods in the diet at approximately six months. We did not find strong
evidence to support targeted public health messages to certain population subgroups. Although
receiving income support may be related to optimal exclusive breastfeeding duration, we suggest
more work be conducted to clarify the mechanisms of this association. Given the population-
based sampling of the Nunavut Inuit Health Survey and the few exclusion criteria imposed in the
present study, we believe our study findings can be cautiously extended to most Inuit Canadians.
121
LITERATURE CITED
1. Health Canada. Exclusive Breastfeeding Duration: 2004 Health Canada
Recommendation. 2004; http://www.hc-sc.gc.ca/fn-an/alt_formats/hpfb-
dgpsa/pdf/nutrition/excl_bf_dur-dur_am_excl-eng.pdf. Accessed June 3, 2012.
2. World Health Organization. The World Health Organization's infant feeding
recommendation. 2001;
http://www.who.int/nutrition/topics/infantfeeding_recommendation/en/index.html.
Accessed June 20, 2009.
3. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
4. Health Canada. Trends in breastfeeding practices in Canada (2001-2009-2010). 2012;
http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/prenatal/trends-tendances-
eng.php. Accessed Dec 8, 2012.
5. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
6. Guevremont A, Kohen D. Inuit in Canada: Findings from the Aboriginal Peoples' Survey
- Survey of Living Conditions in the Arctic. The Health of Inuit Children: Fact Sheet.
2007; http://www.statcan.gc.ca/pub/89-627-x/89-627-x2007002-eng.pdf. Accessed June
3, 2012.
7. Statistics Canada. Total non-Aboriginal and Inuit population, by residence, Canada,
2006. 2012; http://www.statcan.gc.ca/pub/82-003-x/2012003/article/11695/tbl/tbl1-
eng.htm. Accessed October 23, 2012.
8. Gionet L. Inuit in Canada: selected findings of the 2006 Census. 2008;
http://www.statcan.gc.ca/pub/11-008-x/2008002/article/10712-eng.htm. Accessed March
26, 2013.
9. Egeland GM, Pacey A, Cao Z, Sobol I. Food insecurity among Inuit preschoolers:
Nunavut Inuit Child Health Survey, 2007-2008. CMAJ. 2010;182(3):243-248.
10. Ford JD, Berrang-Ford L. Food security in Igloolik, Nunavut: an exploratory study. Polar
Record. 2009;45(03):225-236.
11. Saha KK, Frongillo EA, Alam DS, Arifeen SE, Persson LA, Rasmussen KM. Household
Food Security Is Associated with Infant Feeding Practices in Rural Bangladesh. J. Nutr.
2008;138(7):1383-1390.
12. Neault NB, Frank DA, Merewood A, et al. Breastfeeding and health outcomes among
citizen infants of immigrant mothers. Journal of the American Dietetic Association.
2007;107(12):2077-2086.
122
13. Gross RS, Mendelsohn AL, Fierman AH, Racine AD, Messito MJ. Food Insecurity and
Obesogenic Maternal Infant Feeding Styles and Practices in Low-Income Families.
Pediatrics. 2012;130(2):254-261.
14. de Oliveira M, Camacho L, Tedstone A. Extending breastfeeding duration through
primary care: a systematic review of prenatal and postnatal interventions. Journal of
human lactation: official journal of International Lactation Consultant Association.
2001;17(4):326.
15. Government of Nunavut. Nunavut Bureau of Statistics, Community estimates.
http://www.gov.nu.ca/eia/stats/index.html.
16. Dufour R. Breast feeding and adoption among the Inuit population of Northern Quebec.
Can J Publ Health. 1984;75(2):137-140.
17. Stewart P, Steckle J. Breastfeeding among Canadian Indians on-reserve and women in
the Yukon and NWT. Canadian journal of public health. Revue canadienne de santé
publique. 1987;78(4):255.
18. Asuri S, Ryan A, Arbour L. Early Inuit Child Health in Canada: Breastfeeding among
Inuit in Canada. 2011; https://www.itk.ca/sites/default/files/2011-Report-Breastfeeding-
among-Inuit-in-Canada.pdf. Accessed December 12, 2012.
19. Saudny H, Leggee D, Egeland G. Design and methods of the Adult Inuit Health Survey
2007–2008. International Journal of Circumpolar Health.
2012;71:10.3402/ijch.v3471i3400.19752.
20. Kark J, Troya G, Friedlander Y, Slater P, Stein Y. Validity of maternal reporting of breast
feeding history and the association with blood lipids in 17 year olds in Jerusalem. Journal
of Epidemiology and Community Health. 1984;38(3):218-225.
21. Quandt SA. Material recall accuracy for dates of infant feeding transitions. Human
Organization. 1987;46(2):152-160.
22. Tienboon P, Rutishauser IHE, Wahlqvist ML. Maternal recall of infant feeding practices
after an interval of 14 to 15 years. Australian Journal of Nutrition and Dietetics.
1994;51:25-25.
23. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
24. Pande H, Unwin C, Haheim L. Factors associated with the duration of breastfeeding:
analysis of the primary and secondary responders to a self-completed questionnaire. Acta
Paediatr. 1997;86:173 - 177.
25. Scott J, Landers M, Hughes R, Binns C. Factors associated with breastfeeding at
discharge and duration of breastfeeding. Journal of Paediatrics and Child Health.
2001;37(3):254-261.
123
26. Horta BL, Kramer MS, Platt RW. Maternal smoking and the risk of early weaning: a
meta-analysis. American Journal of Public Health. 2001;91(2):304-307.
27. Stuart MacAdam P, Dettwyler KA. Breastfeeding: biocultural perspectives. New York:
Transaction Publishers; 1995.
28. Li R, Darling N, Maurice E, Barker L, Grummer-Strawn LM. Breastfeeding rates in the
United States by characteristics of the child, mother, or family: the 2002 National
Immunization Survey. Pediatrics. 2005;115(1):e31-37.
29. Forste R, Hoffmann JP. Are US mothers meeting the Healthy People 2010 breastfeeding
targets for initiation, duration, and exclusivity? The 2003 and 2004 National
Immunization Surveys. Journal of Human Lactation. 2008;24(3):278-288.
30. Jacknowitz A, Novillo D, Tiehen L. Special supplemental nutrition program for women,
infants, and children and infant feeding practices. Pediatrics. 2007;119(2):281-289.
31. Datta GD, Subramanian SV, Colditz GA, Kawachi I, Palmer JR, Rosenberg L.
Individual, neighborhood, and state-level predictors of smoking among US Black
women: a multilevel analysis. Soc Sci Med. 2006;63(4):1034-1044.
32. Kimbro R, Lynch S, McLanahan S. The Influence of Acculturation on Breastfeeding
Initiation and Duration for Mexican-Americans. Population Research and Policy Review.
2008;27(2):183-199.
33. Gibson-Davis CM, Brooks-Gunn J. Couples' Immigration Status and Ethnicity as
Determinants of Breastfeeding. Am J Public Health. 2006;96(4):641-646.
34. Gibson MV, Diaz VA, Mainous AG, Geesey ME. Prevalence of Breastfeeding and
Acculturation in Hispanics: Results from NHANES 1999–2000 Study. Birth.
2005;32(2):93-98.
35. Banerji A, Bell A, Mills EL, et al. Lower respiratory tract infections in Inuit infants on
Baffin Island. CMAJ. 2001;164(13):1847-1850.
36. Kovesi T, MD, Gilbert NL, MSc, Stocco C, MSc, et al. Indoor air quality and the risk of
lower respiratory tract infections in young Canadian Inuit children. CMAJ.
2007;177(2):155-160.
37. Bjerregaard P, Curtis T. Cultural change and mental health in Greenland: the association
of childhood conditions, language, and urbanization with mental health and suicidal
thoughts among the Inuit of Greenland. Social Science & Medicine. 2002;54(1):33-48.
38. Thomas T, Greeley G, Guyot M, Loubert K, Banfield W. Inuit Wellness Programs in
Nunavut: 2004-2005. 2006; http://pubs.aina.ucalgary.ca/health/61935.pdf. Accessed May
24, 2012.
124
39. Canadian Tuberculosis Committee. Special report: Tuberculosis among the Aboriginal
Peoples of Canada, 2000 to 2004. 2004; http://www.phac-
aspc.gc.ca/publicat/2007/tbcan04/tbaboriginal-eng.php. Accessed September 21, 2012.
40. Wanyeki I, Olson S, Brassard P, et al. Dwellings, crowding, and tuberculosis in Montreal.
Social Science & Medicine. 2006;63(2):501-511.
41. Radimer KL. Measurement of household food security in the USA and other
industrialised countries. Public Health Nutrition. 2002;5(6a):859-864.
42. Bickel G, Nord M, Prince C, Hamilton W, Cook J. Measuring Food Security in the
United States. 2000; http://www.fns.usda.gov/fsec/files/fsguide.pdf. Accessed June 2,
2012.
43. Carlson SJ, Andrews MS, Bickel GW. Measuring Food Insecurity and Hunger in the
United States: Development of a National Benchmark Measure and Prevalence
Estimates. J. Nutr. 1999;129(2):510S-516S.
44. Wang Z, Waldron W. Using the SAS Survey Procedures for Subpopulation Analysis with
Jacknife Repeated Replication Methods in SAS 9.2. SAS Global Forum 2010 2010;
http://support.sas.com/resources/papers/proceedings10/267-2010.pdf. Accessed
December 4, 2012.
45. Graubard B, Korn E. Survey inference for subpopulations. American Journal of
Epidemiology. 1996;144(1):102-106.
46. Infant Feeding Joint Working Group. Nutrition for healthy term infants:
recommendations from birth to six months: a joint statement of Health Canada, Canadian
Paediatric Society, Dietitians of Canada, and Breastfeeding Committee for Canada. 2012;
http://www.hc-sc.gc.ca/fn-an/nutrition/infant-nourisson/recom/index-eng.php#a3.
47. Hosmer D, Lemeshow S. Applied Logistic Regression. 2nd ed. Toronto: John Wiley and
Sons, Inc; 2000.
48. Clifford TJ, Campbell MK, Speechley KN, Gorodzinsky F. Factors Influencing Full
Breastfeeding in a Southwestern Ontario Community: Assessments at 1 Week and at 6
Months Postpartum. J Hum Lact. 2006;22(3):292-304.
49. Dubois L, Girard M. Social Determinants of initiation, duration and exclusivity of
breastfeeding at the population level: the results of the Longitudinal Study of Child
Development in Quebec (ELDEQ 1998-2002). Can J Public Health. 2003;94(3):300-305.
50. Public Health Agency of Canada. What Mothers Say: The Canadian Maternity
Experiences Survey: Ottawa; 2009.
51. Reilly JJ, Wells JCK. Duration of exclusive breast-feeding: introduction of
complementary feeding may be necessary before 6 months of age. British Journal of
Nutrition. 2005;94(06):869-872.
125
52. Fawzi WW, Herrera MG, Nestel P, El Amin A, Mohamed KA. A longitudinal study of
prolonged breastfeeding in relation to child undernutrition. International Journal of
Epidemiology. 1998;27(2):255-260.
53. Simondon KB, Simondon F. Mothers prolong breastfeeding of undernourished children
in rural Senegal. International Journal of Epidemiology. 1998;27(3):490-494.
54. Morgan J, Taylor A, Fewtrell M. Meat consumption is positively associated with
psychomotor outcome in children up to 24 months of age. Journal of pediatric
gastroenterology and nutrition. 2004;39(5):493-498.
55. Krebs NF, Westcott JE, Butler N, Robinson C, Bell M, Hambidge KM. Meat as a first
complementary food for breastfed infants: feasibility and impact on zinc intake and
status. Journal of pediatric gastroenterology and nutrition. 2006;42(2):207-214.
56. Galloway T, Young TK, Egeland GM. Emerging obesity among preschool-aged
Canadian Inuit children: results from the Nunavut Inuit Child Health Survey.
International Journal of Circumpolar Health. 2010;69(2).
57. Heck K, Braveman P, Cubbin C, Chavez G, Kiely J. Socioeconomic status and
breastfeeding initiation among California mothers. Public Health Rep. 2006;121(1):51-
59.
58. Kimbro R. On-the-Job Moms: Work and Breastfeeding Initiation and Duration for a
Sample of Low-Income Women. Maternal and Child Health Journal. 2006;10(1):19-26.
59. Arlotti J, Cottrell B, Lee S, Curtin J. Breastfeeding among low-income women with and
without peer support. Journal of Community Health Nursing. 1998;15(3):163-178.
60. Montgomery DL, Splett PL. Economic Benefit of Breast-Feeding Infants Enrolled in
WIC. Journal of the American Dietetic Association. 1997;97(4):379-385.
61. Fok D, Mong T, Chua D. The economics of breastfeeding in Singapore. Breastfeeding
review. 1998;6(2):5-9.
62. Jarosz L. Breast-feeding versus formula: cost comparison. Hawaii Medical Journal.
1993;52(1):16-18.
63. Inuit Tapiriit Kanatami. Social determinants of Inuit health in Canada: a discussion paper.
2007; http://ahrnets.ca/files/2011/02/ITK_Social_Determinants_paper_2007.pdf.
Accessed June 20, 2012.
64. Pawloski L, Kodadek M, Davidson M, Sears W, Young A. Understanding cultural
differences when advising mothers about feeding choices. Pediatric Nursing.
2001;27(1):52-53.
65. Malhotra R, Noheria A, Amir O, Ackerson LK, Subramanian SV. Determinants of
termination of breastfeeding within the first 2 years of life in India: evidence from the
National Family Health Survey-2. Maternal & Child Nutrition. 2008;4(3):181-193.
126
66. Giashuddin M, Kabir M. Duration of breast-feeding in Bangladesh. Indian Journal of
Medical Research. 2004;119:267-272.
67. Kleinbaum DG, Kupper LL, Muller KE. Applied regression analysis and other
multivariable methods: Duxbury Press; 2007.
68. Kristiansen AL, Lande B, Øverby NC, Andersen LF. Factors associated with exclusive
breast-feeding and breast-feeding in Norway. Public Health Nutrition. 2010;13(12):2087-
2096.
69. Michaelsen KF, Larsen PS, Thomsen BL, Samuelson G. The Copenhagen cohort study
on infant nutrition and growth: duration of breast feeding and influencing factors. Acta
Paediatrica. 1994;83(6):565-571.
70. Ball TM, Wright AL. Health Care Costs of Formula-feeding in the First Year of Life.
Pediatrics. 1999;103(4):870-876.
127
Figure 1: Flow diagram of participants from the Nunavut Inuit Child Health Survey (N-ICHS)
(2007-2008) included in the present analysis
Participants contacted to complete
the N-ICHS
(n=537)
Participants completing the
N-ICHS
(n=388)
Participants initiating breastfeeding
(N=244)
Surveys completed by a biological
parent
(N=219)
Sample included in analyses
(N=188)
Survey non-responders (n=149)
Did not initiate breastfeeding (n=133)
Missing/unknown breastfeeding initiation (n=11)
Missing/unknown breastfeeding duration (n=22)
Not completed by biological parent (n=25)
Incomplete covariate information (n=33)
128
Figure 2: Exclusive Breastfeeding duration (n=752), Nunavut Inuit Child Health Survey (2007-
2008)
0
10
20
30
40
50
60
70
80
90
100
≤ 1 month >1 to <5.5 months 5.5-6.5 months > 6.5 months
Fre
qu
en
cy (
%)
Exclusive breastfeeding duration (months)
129
Table 1: Selected unweighted (n=188) and weighted (n=752) household, maternal, infant and community characteristics of respondents,
Nunavut Inuit Child Health Survey (2007-2008)
Characteristics Unweighted (n=188) Weighted (n=752)
N % Na
%
Household Characteristics
Speaks primarily Inuit dialects at home 96 51.1 390 51.9
Receives income support 83 44.2 330 43.8
Lives in an overcrowded conditions 97 51.6 387 51.4
Household smoking restrictions 172 91.5 690 91.8
Household food security - secure 61 32.4 252 33.5
- moderate insecurity 73 38.8 280 37.2
- severe insecurity 54 28.7 220 29.3
Child food security – secure 84 44.7 340 45.2
- moderate insecurity 63 33.5 232 30.8
- severe insecurity 41 21.8 180 23.9
Maternal Pregnancy Behaviours
Maternal drinking during pregnancy 47 25.0 194 25.8
Maternal smoking during pregnancy 155 82.4 606 80.6
Infant Characteristics
Infant birth weight : <2500 grams 12 6.4 43 5.8
Community Characteristics
Community has <1600 population 101 53.4 287 38.2
Community Region – Baffin 86 45.8 365 48.5
-Kitikmeot Region 65 34.6 268 35.7
-Kivalliq Region 37 19.7 118 15.7
Community has Canada Prenatal Nutrition Program 98 52.1 326 43.4
Frequency of Preventive care (≥ 3.5 visits per 100 persons) 97 51.6 304 40.4
Frequency of Prenatal care (≥1 visit per live birth) 99 52.7 308 41.0
a Weighted values were rounded to the nearest whole number
130
Table 2: Distribution of exclusive breastfeeding practices (weighted), by select household, maternal, infant and community characteristics
of respondents, Nunavut Inuit Child Health, (n=752) Survey (2007-2008)
Characteristics
Exclusive breastfeeding duration
As recommended
Suboptimal duration
p-value c
Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
na %
b n
a %
b n
a %
b n
a %
b
Household Characteristics
Languages spoken at home
Other dialects (n=361) 69 19.1 92 25.4 164 45.3 37 10.2 p=0.06
Inuit dialects (n=390) 102 26.1 78 19.9 124 31.8 87 22.2
Income support
No (n=422) 70 16.7 112 26.4 193 45.8 46 11.0 p<0.01
Yes (n=329) 100 30.4 58 17.6 94 28.6 77 23.4
Household crowding
≤1 person/room (n=365) 92 25.3 86 23.5 121 33.2 65 17.9 p=0.65
> 1 person/room (n=387) 79 20.3 84 21.7 166 43.0 58 15.0
Household smoking restrictions
Yes (n=690) 162 23.5 152 22.1 274 39.7 101 14.7 p=0.14
No (n=62) 9 14.2 17 28.1 14 22.0 22 35.6
Household food security
Secure (n=252) 60 23.7 55 21.6 99 39.4 38 15.2
p=0.99 Moderate insecurity (n=280) 62 22.1 67 23.9 104 37.3 47 16.7
Severe insecurity (n=220) 49 22.4 48 22.0 84 38.1 39 17.5
Childhood food security
Secure (n=340) 78 23.0 82 24.0 129 38.0 51 15.0
p=0.99 Moderate insecurity (n=232) 56 24.2 44 18.8 89 38.4 43 18.6
Severe insecurity (n=180) 36 20.2 45 24.8 70 38.6 29 16.2
Table 2 continued next page...
131
Table 2 continued from previous page
Characteristics
Exclusive breastfeeding duration
p-value c
As recommended
Suboptimal duration
Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
na %
b n
a %
b n
a %
b n
a %
b
Maternal Pregnancy Behaviours
Drinking
No (n=558) 123 22.0 132 23.7 205 36.8 98 17.6 p=0.80
Yes (n=194) 48 24.9 38 19.4 83 42.6 26 13.2
Smoking
No (n=146) 37 25.2 41 28.1 44 29.9 24 16.7 p=0.72
Yes (n=606) 134 22.1 129 21.3 244 40.2 99 16.4
Infant Characteristics
Infant birth weight
< 2500 grams (n=43) 7 15.7 19 44.9 12 27.5 5 11.9 p=0.35
≥ 2500 grams (n=708) 164 23.2 150 21.2 276 38.9 118 16.7
Community Characteristics
Community size
<1500 people (n=287) 65 22.5 45 15.6 124 43.4 53 18.4 p=0.37
≥1500 people (n=465) 106 22.9 125 26.9 163 35.1 71 15.2
Region
Baffin (n=365) 85 23.4 60 16.4 166 45.5 53 14.6
p=0.18 Kitikmeot (n=268) 52 19.6 87 32.3 73 27.4 56 20.7
Kivalliq (n=118) 33 27.8 23 19.5 48 40.4 14 12.2
Canada Prenatal Nutrition Program
No (n=426) 91 21.3 119 27.8 143 33.4 74 17.4 p=0.23
Yes (n=326) 80 24.6 51 15.7 145 44.5 49 15.1
Table 2 continued next page…
132
a Weighted frequencies rounded to the nearest whole number; may not equal column totals due to rounding
b Row percent may not total to 100 due to rounding
c p-values generated from Wald tests
Table 2 continued from previous page
Characteristics
Exclusive breastfeeding duration
p-value c
As recommended
Suboptimal duration
Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
na %
b n
a %
b n
a %
b n
a %
b
Frequency of Preventive care
< 3.5 visits per 100 persons (n=448) 106 23.6 115 25.7 153 34.2 73 16.4 p=0.51
≥ 3.5 visits per 100 persons (n=304) 65 21.4 55 18.0 134 44.1 50 16.4
Frequency of Prenatal care
< 1 visits per live birth (n=444) 117 26.3 91 20.5 175 39.4 61 13.8 p=0.39
≥ 1 visit per live birth (n=308) 54 17.6 78 25.6 113 36.6 62 20.2
133
Table 3: Weighted bivariate prevalence odds ratios (pOR) and 95% confidence interval of selected household, maternal, infant and
community characteristics from multinomial logistic regression models, by exclusive breastfeeding duration. Nunavut Inuit Child Health
Survey, 2007-2008 (nweighted = 752)
Characteristic
Exclusive Breastfeeding Duration
As recommended
Sub-optimal duration
Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
pOR pOR, 95% CIa
pOR, 95% CIa
pOR, 95%CIa
Household Characteristics
Speaking Inuit dialects c 1.00 0.57 (0.22-1.49) 0.51 (0.22-1.19) 1.60 (0.58-4.40)
Receiving income support a 1.00 0.36 (0.14-0.94) 0.34 (0.14-0.81) 1.17 (0.40-3.10)
Crowding : > 1 person/room 1.00 1.14 (0.44-2.97) 1.61 (0.69-3.72) 1.04 (0.38-2.90)
No smoking restrictionsd
1.00 2.10 (0.41-10.69) 0.92 (0.17-4.96) 4.00 (0.82-19.55)
Household food security (vs. secure)
Moderate insecurity 1.00 1.18 (0.38-3.65) 1.02 (0.37-2.77) 1.18 (0.36-3.85)
Severe insecurity 1.00 1.08 (0.32-3.65) 1.02 (0.36-2.87) 1.22 (0.34-4.41)
Childhood food security
Moderate insecurity 1.00 0.74 (0.24-2.28) 0.96 (0.38-2.43) 1.17 (0.38-3.66)
Severe insecurity 1.00 1.17 (0.35-3.97) 1.16 (0.39-3.40) 1.22 (0.32-4.75)
Maternal Pregnancy Behaviours
Drinking during pregnancy 1.00 0.72 (0.23-2.24) 1.02 (0.40-2.64) 0.66 (0.21-2.10)
Smoking during pregnancy 1.00 0.87 (0.29-2.62) 1.54 (0.52-4.50) 1.12 (0.33-3.76)
Infant Characteristics
Infant birth weight
< 2500 grams (vs. ≥ 2500 grams) 1.00 3.12 (0.42-22.85) 1.04 (0.13-8.50) 1.05 (0.11-9.64)
Table 3 continued on next page…
134
Table 3 continued from page before..
Characteristic
Exclusive Breastfeeding Duration
Sub-optimal duration
As recommended Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
pOR pOR, 95% CIa
pOR, 95% CIa
pOR, 95%CIa
Community Characteristics
Community size ≥1500 people 1.00 1.69 (0.68-4.21) 0.80 (0.36-1.76) 0.81 (0.31-2.10)
Region (vs. Baffin)d
Kitikmeot 1.00 2.35 (0.80-6.92) 0.72 (0.27-1.89) 1.70 (0.55-5.27)
Kivalliq 1.00 1.00 (0.24-4.14) 0.74 (0.25-2.25) 0.70 (0.19-2.56)
Has Prenatal Nutrition Programd
1.00 0.49 (0.19-1.26) 1.15 (0.50-2.62) 0.75 (0.28-2.03)
Frequency of preventive care
≥ 3.5 visits per 100 persons 1.00 0.77 (0.30-1.98) 1.42 (0.62-3.21) 1.10 (0.42-2.89)
Frequency of prenatal care
≥ 1 visit per live birth 1.00 1.86 (0.74-4.63) 1.38 (0.62-3.06) 2.18 (0.82-5.80)
a Confidence interval calculated using Wald method b p<0.05;
cp<0.10,
dp<0.20,
ep<0.25
135
Table 4: Multivariate adjusted weighted prevalence odds ratios and 95% confidence limits, by exclusive breastfeeding duration. Nunavut
Inuit Child Health Survey (2007-2008) (nweighted = 752)
a Confidence limits calculated using Wald method b p<0.05;
cp<0.10
Characteristic
Exclusive breastfeeding duration
As
recommended
Suboptimal duration
Shorter than recommended Longer than
recommended
5.5-6.5 months
(n=171, 22.7%)
≤ 1 month
(n=170, 22.6%)
>1-<5.5 months
(n=288, 38.2%)
> 6.5 months
(n=124, 16.4%)
pOR pOR (95% CIa) pOR (95% CI
a) pOR (95% CI
a)
Speaking Inuit dialectsc 1.00 0.74 (0.27-2.04) 0.50 (0.21-1.21) 1.77 (0.54-5.78)
Receiving income support b 1.00 0.39 (0.14-1.05) 0.34 (0.13-0.85) 1.06 (0.38-2.95)
Has Prenatal Nutrition Programc 1.00 0.61 (0.22-1.68) 1.70 (0.68-4.24) 0.62 (0.20-1.96)
136
Chapter 6 : Manuscript 3
6 Study Overview
The third manuscript addresses the fifth and sixth thesis objectives. In this paper, a latent model
that classifies heterogeneity in traditional activities amongst Inuit Canadians is developed. This
model is then used to examine if breastfeeding initiation and duration vary across latent classes.
137
6.1 Measuring level of engagement in traditional activities among Inuit Canadians by a latent class analysis and the application of this analysis to describe breastfeeding practices among Canadian Inuit: IPY (International Polar Year) Inuit Health Survey
List of Contributing Authors: Kathryn E McIsaac, MSc1; Wendy Lou, PhD
1; Daniel Sellen,
PhD1,2
; T Kue Young, MD DPhil1
Author Affiliations: 1Dalla Lana School of Public Health, University of Toronto, Toronto
Canada; 2Department of Anthropology, University of Toronto, Toronto Canada
138
Background: Evacuating Inuit women to urban centers for childbirth may introduce barriers for
timely breastfeeding initiation. Overcoming these barriers and beginning breastfeeding likely
requires a high degree of resiliency. Engaging in traditional activities is seen as an important
indicator of individual and community cultural resiliency.
Objective: The purpose of this study is to create latent categories of measure of participation in
traditional activities among Canadian Inuit, as an indicator of resiliency and use this measure We
also aim to examine if these classes predict breastfeeding initiation and appropriate exclusive
breastfeeding duration.
Design: We perform a latent class analysis using data from the IPY Inuit Health Survey (2007-
2008), a cross-sectional population-based survey of Inuit Canadians, to identify mutually
exclusive classes of engagement with traditional activities. We link these data to the Nunavut
Inuit Child Health Survey (2007-2008) to examine if latent membership class predicts
breastfeeding practices.
Results: We identify five different categories of participation in traditional activities. Between
classes, members varied in the extent to which they participated in land based activities and
community based activities: the most common class (32.7%) participated infrequently in all
types of traditional activities measured, while the two least common classes (7.5% each)
participated in various traditional activities (both land and community based) most frequently.
Being younger and female was associated with belonging to classes with lower likelihood of
engaging in traditional activities on the land and higher likelihoods of engaging in community
based activities. There was neither an association between breastfeeding initiation and class
membership, nor exclusive breastfeeding as recommended and class membership.
Conclusions: Although breastfeeding practices were not associated with classes of engaging in
traditional activities, age and sex were. These age specific trends may indicate a shift in the
central meaning of Inuit identity to a younger generation and should be explored further in both
qualitative and quantitative research.
Keywords: Resiliency; Breastfeeding; Inuit; Latent Class Analysis; Acculturation; Traditional
Activities
139
Much has been written about the cultural change in Canada’s North during the latter half of the
20th
century and its effects on the health of Inuit1-3
. As Inuit transitioned from a hunting (pre-
1910), to a trading (1910-1950) and finally to a more community-based wage economy (1950-
onwards)4, many physical and mental health concerns became increasingly prevalent. The
population started smoking more5, spending less time on the land (lowering physical fitness)
6,7
and relying more on less nutrient dense store food as opposed to country (i.e. traditional) food6,7
.
There have also been increasing rates of many chronic conditions (e.g. obesity, lung cancer)8,9
.
This economic shift also precipitated a variety of social problems: roles within the family needed
to be redefined to fit into the new economic model, creating an identity crisis for many Inuit,
manifesting in high rates of alcoholism, violence, self-harm and suicide4,10,11
The birthing process has also been affected by this cultural change. Beginning in the 1970’s,
most Inuit women have been evacuated from their families and communities at 36 weeks
gestation to give birth in urban hospitals in more southern regions of Canada12-14
. This practice
has been culturally disruptive to the birthing process2,15
which was historically viewed as a
communal responsibility16
. Although some regions have been successful in their efforts to
return birth to communities2,17
, the majority of Inuit women continue to be evacuated, in spite of
more recent research suggesting better psychosocial outcomes for the mother15,16
and equivalent
birth outcomes for low risk deliveries18
with community-based deliveries.
The practice of evacuation may have deleterious effects on breastfeeding practices. The
breastfeeding initiation rates in Canada’s North are some of the lowest in the country19
, even
after accounting for the high prevalence of traditional adoption20
. Some Inuit women have
expressed difficulties accessing culturally appropriate breastfeeding support in southern
communities15
. These women must wait until they return to their communities - days after
giving birth – to draw upon the resources of their family and community before they begin
breastfeeding15
. Yet, the longer the time interval between birth and breastfeeding initiation, the
more likely the mother is to rely on infant formula during the first six months of life21
. For most
Inuit women, it seems likely that beginning breastfeeding requires some degree of resiliency.
Resiliency, or positive adaptation despite adversity22
, is an emerging theme in Aboriginal
research22,23
. Higher degrees of cultural resiliency have been associated with positive mental
health outcomes in Aboriginal Canadians1,24
. There are many different components to resiliency,
140
and a recent review of the literature suggests that participation in traditional activities serves as
an indicator of resilience in Aboriginal communities23
.
For the present research, we use latent class analysis to identify classes of traditional activities
among Inuit Canadians based on their response patterns to a series of questions. This analytic
technique has been used previously to sub-type drinking behaviours25,26
, sexual behaviours27
and
even acculturation28
. Latent class analysis requires heterogeneity in the construct being classified
and presupposes an underlying class structure exists than can be identified with measured
variables.
It is reasonable to expect inter-individual heterogeneity in engagement with traditional activities
among Inuit Canadians. According to a recent inventory, some behaviours, such as speaking
traditional languages and sharing country food are almost universally embraced by Inuit (90%5
and 96%29
, respectively) while other behaviours, like harvesting country food and engaging in
cultural activities, are not as frequently practiced (68%29
and 50%30
, respectively).
The aim of this research was twofold. First, to identify various classes of engagement in
traditional activities among Inuit Canadians. Second, we evaluate how these resultant class
structures predict breastfeeding practices.
METHODS
We use data from both the International Polar Year (IPY) Inuit Health Survey (herein referred to
as the adult survey) and Nunavut Inuit Child Health Survey (herein referred to as the child
survey), two cross-sectional, population-based surveys conducted among Inuit Canadians in the
summer and fall of 2007 and 2008. Both Inuit Health Surveys intended to provide an overall
sense of the health and well-being of Inuit Canadians from self-administered surveys and clinical
assessments. For these present analyses, only data from the self-reported surveys were used.
The adult survey was administered in 36 communities, from one of three regions of Canada’s
North: the Inuvialuit Settlement Region, Northwest Territories; Nunavut; and Nunatsiavut,
Northern Labrador. The aim was to sample approximately 12% of households per community31
.
Community housing lists were the primary sampling frame; households were randomly selected
from these lists by members of the research team approximately two weeks prior to survey
141
administration. All Inuit individuals living in selected households, speaking English or select
Inuit dialects and aged 18 or older were invited to participate in the survey portion.
The child survey was administered in 16 communities, all located in Nunavut, Canada. Other
communities were excluded due to financial and logistical reasons that are described fully
elsewhere32
. Children in these communities were recruited in one of two ways: first, adults who
participated in the IPY Inuit Health Survey and had children meeting age eligibility criteria were
invited to participate. When that approach did not produce a sufficient number of participants,
the remaining children were randomly sampled from community health centre lists. Inuit
children were eligible to participate if they were aged 3 to 5 years at the time of data collection
and their caregiver (i.e. proxy respondent completing the survey) spoke English or Inuktitut.
Selected households and participants for the IPY Inuit Health Survey and the Nunavut Inuit
Child Health Survey, respectively, were contacted by a member of the research team who
delivered an informational DVD which described the study and provided the required
information for recruited participants to make informed consent. Those consenting to participate
were then asked to complete an interviewer-administered survey, either in their home or a
community facility, most often the community health centre.
The interviewer-administered survey obtained information on many indicators of health as well
as other individual, household and community characteristics (e.g. age, marital status). Manifest
variables believed to be suitable indicators of traditional activity were selected from the IPY
Inuit Health Survey. These variables were chosen based on consultations with Inuit Canadian
and other experts in the field. A summary of these indicator variables is provided in Table 1.
Information on breastfeeding initiation and exclusive breastfeeding duration were extracted from
the Nunavut Inuit Child Health Survey. Breastfeeding initiation was measured in response to the
question “was your child ever breastfed or fed breast milk” and was treated as a binary outcome
variable (yes vs. no) in analyses. A review of studies in various populations found that maternal
recall of breastfeeding initiation agrees well with external validation standards (Sensitivity: 82%;
Specificity: 93%).33
Exclusive breastfeeding duration was measured in response to the question
“how many months was your child exclusively breast fed or exclusively fed breast milk”. This
was treated as a binary variable in all analyses: those caregivers reporting their infant received
exclusive breastfeeding for between 5.5 and 6.5 months were classified as ‘exclusive
142
breastfeeding as recommended’. All other responses were coded as ‘suboptimal exclusive
breastfeeding’. Maternal recall of timing of introduction of complementary foods into the diet
can range from as high as 88%34
to as low as 65%35
, depending on the period of recall (one
year34
to 14-15 years). We chose the dates between 5.5 and 6.5 months of exclusive
breastfeeding to account for different developmental trajectories in infants that would make them
more or less ready to transition to other complementary foods.36
Because our own previous
research (Chapter 5) found that many Inuit report exclusive breastfeeding for much longer than
the recommended duration, we did not want to simply dichotomize this outcome at six months.
Several variables that we had previously identified to be associated with breastfeeding initiation
(Chapter 4) and exclusive breastfeeding duration (Chapter 5), including who the child’s primary
caregiver is, if the family received income support and the infant’s birth weight, were also
extracted to use as covariates in these analyses.
We linked the measures of engagement with traditional activities from the IPY Inuit Health
Survey with breastfeeding initiation and exclusive breastfeeding duration from the Nunavut Inuit
Child Survey using identifiers that were unique to each household. Thus, in order to be included
in the breastfeeding histories analyses, a caregiver had to have completed both the IPY Inuit
Health Survey and the Nunavut Inuit Child Health Survey as well as have complete data on the
outcome variables and other covariates of interest. When more than one of the child’s caregivers
completed the IPY Inuit Health Survey, the responses from the biological mother were used,
since the validity of recalling breastfeeding histories has not been evaluated for other types of
caregivers.
All of the variables included in these analysis were categorical, and most were dichotomous
(yes/no). In the IPY Inuit Health Survey, we reduced two items (i.e. How often have you
participated in activities where people get together to work for the benefit of the community;
how often have you gotten together with people to play games, sports or recreational activities)
that were measured on a five point Likert scale (i.e. very often, often, sometimes, rarely, never)
to dichotomous variables (often and very often versus not) to aid in interpretability of the classes.
We also elected to dichotomize these variables because some methods of class enumeration (e.g.
bootstrap likelihood test, described in further detail below) require binary variables.
143
Statistical analysis
Latent class analysis (LCA) was used to identify separate classes of engagement in traditional
activities among study participants. This analytic technique uncovers heterogeneity in a
population and groups together those individuals who share similar characteristics into latent
classes37
. In the present analysis, LCA was used to identify groups of individuals who engage in
similar types of traditional activities. Latent (i.e. unmeasured) classes are ultimately identified,
conditional on responses to manifest variables38
.
The LCA procedure shares similarities with both cluster and factor analysis. The underlying
analytic goals of LCA and cluster analysis are the same: to identify subgroups or classes of
individuals in a population but the methods used to identify classes differ39
. In a traditional
cluster analysis, the number of classes is driven by the researcher, while in the LCA, the number
of classes is decided by statistical probabilities using maximum likelihood estimations.
Fundamentally, traditional cluster analysis assigns individuals absolutely to classes, while in
LCA, assignment to classes are based on posterior probabilities.39
LCA also shares similar methodologies to factor analysis, although the latter requires manifest
variables to be measured on an interval or ratio scale to meet model assumptions 40
, whereas
LCA places no distributional assumptions on manifest variables. The only assumption in LCA is
conditional independence within a class41
. Like factor analysis, missing data are permitted and
assumed to be missing at random, and latent classes are assumed to be free of measurement
error41
.
Two separate types of parameters are generated in LCA (like factor analysis): class probabilities
and item response probabilities. Class probabilities indicate the prevalence of a particular class
in the study population, whereas item response probabilities indicate the proportion of
individuals within a certain class who responded a particular way to a measured variable (i.e.
item response probabilities are conditional upon class membership). Class and item probabilities
were reported for the final model only.
Like factor analysis, the number of appropriate classes to include in a model is decided
successively, starting with one class and then added more classes successively until the model of
best fit model is produced37
. There are several different approaches to class enumeration, but
144
information criteria and modified likelihood ratio tests are used most often to decide on the final
number of classes 37
.
Although latent class models with k-1 classes are technically tested in models with k classes,
their difference does not follow the standard chi-square distribution; as such, standard likelihood
ratio tests for model fit are not appropriate37
. Some modifications of the likelihood ratio test
have been proposed37,42
. Simulation studies provide evidence to support the use of the bootstrap
likelihood ratio test, which uses bootstrap iterations to estimate the distribution between k and k-
1 models. These studies show that the bootstrap likelihood ratio tests identifies the correct
number of classes most often, even compared to the adjusted Bayesian Information criteria, the
recommended information criteria for latent class models37
.
We relied largely on the parametric bootstrap likelihood ratio method as well as substantive
theory to decide the number of classes to include in the final model (i.e. class enumeration)37,43
,
although we also recorded changes in Akaike’s Information Criteria (AIC), and the adjusted
Bayesian Information Criteria (aBIC)44
for each model considered.
After determining the appropriate number of classes of traditional activities from our LCA, we
assigned individuals to latent classes based on posterior probabilities. The assignment approach
used is described in detail elsewhere45
; briefly individuals were assigned to the class they were
most likely to belong to, based on the class showed the greatest loading. We used then used
logistic regression models on a reduced sample (i.e. those who had completed both the IPY Inuit
Health Survey and the Nunavut Inuit Child Health Survey) to evaluate if breastfeeding initiation
and exclusive breastfeeding as recommended (i.e. 5.5 months to 6.5 months) were associated
with the posterior probabilities of latent class membership. Only individuals with full covariates
were included in these models, meaning that different samples were used for our breastfeeding
initiation and breastfeeding duration analysis. These models were built in accordance with
published recommendations46
.
All statistical analyses were conducted in SAS v.9.3. We used the SAS LCA (v.1.27)
procedure47
for the latent class analyses and the SAS LCA bootstrap (v1.1.0) macro 48
for
bootstrap likelihood ratio tests. The SAS GENMOD procedure with a logit link function was
used to explore how class membership predicted breastfeeding outcomes. Fully adjusted models
145
only included factors that were identified from previous analyses of these data (see Chapters 4
and 5).
RESULTS
There were 2796 households randomly selected to participate in the IPY Inuit Health Survey. Of
these, 841 immediately refused and 54 did not attend their scheduled appointments, resulting in a
total sample of 1901 households (68% response rate). From these 1901 homes, 2595 individuals
participated in the survey and were included in the LCA (Sample 1). For the breastfeeding
initiation analysis, all children whose biological caregivers completed the IPY Inuit Health
Survey and had complete information on breastfeeding initiation and other covariates of interest
were included (Sample 2; n=135). The breastfeeding duration analysis excluded those who did
not report initiating breastfeeding and/or had incomplete information on breastfeeding duration
and covariates of interest (Sample 3; n=108). A flow diagram of participation is provided in
Figure 1.
Table 2 presents the demographic characteristics of the three study samples. The average age of
study participants was 42.1 years in Sample 1(i.e. all respondents in the IPY Inuit Health Survey)
and approximately 35 years in Samples 2 and 3 (i.e. breastfeeding initiation and duration
analyses, respectively). There were significantly more males, married couples and high school
educated respondents in Samples 2 and 3 compared to Sample 1.
Table 3 displays the distribution of the samples’ participation in the various traditional activities
that were included in the measurement model. In the IPY Inuit Health Survey, the prevalence of
traditional activities ranged from 30% for hunting on water to 76% for distributing country food
to members of the community. The prevalence of engaging in traditional activities was higher in
Samples 2 and 3, with the exception of participating in community activities (both recreational
and for the benefit of the community), which were more common in Sample 1.
Table 4 summarizes the fit statistics for varying numbers of latent classes. Using all criteria
(AIC, adjusted BIC and the bootstrap likelihood ratio test), the five class model best
characterized the varying degrees of traditional activities in the study sample. The results of the
five class model are shown in Table 5.
146
Class membership ranged from 7.5% to 32.7%. An interpretation of these classes is provided in
Table 6. Classes were grouped according to their time spent on land to collect food, their time
spent engaging in other traditional activities on the land and their participation in community
activities. The most common class was characterized by low frequencies of traditional activities
in each of the three domains (spending time on land to get food, spending time on land for other
traditional activities and participating in community activities). The least common class was
characterized by high frequencies of spending time on land to get food and participating in
community activities and low frequencies of other activities on the land. The odds of latent class
membership varied by sex and age. Specifically, females were more likely to belong to Class 2,
characterized by high participation in community activities but low participation in other types of
traditional activities, and older individuals (aged 42 and older, divided at the mean) more likely
to belong to classes characterized by any participation in traditional activities (Class 3, 4 and 5)
Table 7 presents the association between class membership and breast feeding initiation and
Table 8 presents the association between class membership and exclusive breastfeeding duration.
Class membership was not associated with either breastfeeding practices in crude or adjusted
models.
DISCUSSION
Acculturation has been recognized as an important determinant of Inuit health49
. This is the first
analysis that has attempted to create a latent model to operationalize a domain of resiliency to
acculturation: engaging in traditional activities23
.
In these analyses, we identify five mutually exclusive classes of engagement with traditional
activities among Inuit. We noted that certain behaviours tend to cluster together within classes in
three separate areas: using traditional methods to acquire food (e.g. hunting); spending time on
the land in other capacities (e.g. building an igloo), and engaging in community activities.
Members of the most common class engaged infrequently in all traditional activities and were
more likely to be younger and female compared to those in other classes.
Class interpretation is one of the noted challenges of LCA. There are many statistical measures
that indicate the model of best fit (e.g. AIC, BIC, adjusted BIC)37
, but the identified classes
147
should align with current theory and knowledge in a meaningful way37,38
. Most LCA report
behaviours along a continuum (e.g. most severe to least severe class)25,50,51
, which we did not
find. Rather, we found that Inuit chose to participate in various traditional activities clustered
within each of three domains identified. This maybe an issue of heterogeneity in constructs
included in our model.
Scales are only useful if they are valid and reliable measures of the underlying construct of
interest.52
Ideally, we want to test the validity of our model in another population of Inuit
Canadians. One method would be to assess construct validity of the scale by conducting cross-
correlations with measures that are known to be associated with traditional activities.
Our finding that younger individuals are generally less engaged in activities on the land opens
avenues for further research. While this may be an issue of model validity, the prevalence of
traditional activities in our survey was lower than reported in a previous population-based survey
of Inuit Canadians53
, and the most common latent class was characterized by infrequent
participation in any traditional activities. This may be the consequence of the shifting beliefs in
the Inuit population about identity. Specifically, while an older cohort may feel spending time
on the land is central to being Inuit, the younger cohort may feel building community
partnerships and connections are of greater import. Moreover, the younger generation may have
stronger connections to other domains of resiliency, such as social capital23
, which were not
included in these analyses. Qualitative analysis may be able to speak to these patterns.
We intended to include indicators of language usage in our latent class model as other research
suggests these indicators are a measure of resiliency in other populations24,54,55
. However,
speaking primarily Inuit dialects at home did not exhibit substantial variability across classes in
any of the latent models we evaluated (not shown) even though we noted differences across
classes in the probabilities of engaging in other traditional behaviours. This suggests that
language use has poor discriminant validity for resiliency to acculturation in an Inuit context.
We caution against using languages as a single indicator of traditional knowledge in quantitative
research, as has been done with previous Aboriginal research in Canada56
.
We found that class membership predicted neither breastfeeding initiation nor exclusive
breastfeeding duration. As previously noted, even though our five class model fit the data best,
further research is needed to confirm this class structure. Moreover, because breastfeeding
148
information was only available in a substantially reduced data set (n=108), modeling associations
within the latent class framework was not possible (i.e. the data would not converge). Instead,
individuals were assigned to classes using posterior probabilities45
, introducing at least some
measurement error in the results. While this finding could be real, we cannot discount the
possibility that an inappropriate class structure or measurement error is driving the association
between breastfeeding practices and engagement in traditional activities – a marker of resiliency.
We did identify some important differences in demographic characteristics between the full
sample used to construct the latent classes and the reduced samples used to model the association
between latent class membership and breastfeeding histories. We also found that the odds of
class membership varied by both age and gender, two characteristics that were differentially
distributed in our full and two reduced samples. Assuming that the five class LCA model could
be validly applied to the reduced sample may or may not have been an appropriate approach.
Indeed, re-analyzing the latent class structure in the reduced model, only four latent classes were
identified.
Limitations
We performed an opportunistic analysis using data collected in the IPY Inuit Health Survey to
develop a measure of traditional activities, an indicator of resiliency, amongst Inuit Canadians.
The survey was designed to measure health outcomes and did not include comprehensive
measures of traditional activities. Other indicators that may be important include: wearing
traditional dress, throat singing, sharing oral histories and production of painting and sculpture23
.
Expanding latent models to include these, or other indicators of resiliency (e.g. social capital)23
,may further elucidate various classes of traditional activities in Inuit society.
We also cannot discount the possibility of selection bias, which would arise if participating in the
survey was related to an individual’s likelihood of engaging in traditional activities and/or if the
association between traditional class membership and breastfeeding practices differed between
participants and non-participants. The IPY Inuit Health Survey was conducted in the summer
and fall months, a time when families may participate in land-based activities away from their
communities. If in fact, the resultant sample was less likely to engage in traditional activities
than all Inuit, this may have affected our prevalence estimates of traditional activities; the
149
underlying class structure; and/or the association between traditional activities and various
breastfeeding practices.
Strengths
This is the first study that has attempted to operationalize an indicator of individual resiliency in
Inuit communities - engaging in traditional activities – using latent variable models. We use a
large, population-based, representative sample of Inuit Canadians and demonstrate that
measuring culture in this population is possible. Moreover, this research suggests how various
indicators of traditional activities may relate to each other which can be used to further our class
structure.
We feel this research is particularly timely. Amongst many Inuit, there is a belief that returning
to more traditional ways of life will improve health and well-being49,57,58
. With the creation of
Nunavut on April 1, 199959
came the simultaneous adoption of both the Nunavut Land Claims
Agreement (land settlement claim) and the Nunavut Act, the latter of which granted inhabitants
of the territory the right to self-governance. Because the majority of those living in Nunavut self-
identify as Inuit, by extension, the newly formed government was largely an Inuit government,
providing an ideal opportunity to protect, promote and preserve Inuit culture. Creating an
empirical measure of resiliency can be used in future research to substantiate or refute the idea
that traditional activity participation is associated with health outcomes.
Conclusions
Understanding how cultural identity and resiliency shapes health and health behaviours is
particularly important among Inuit Canadians, who have experienced rapid colonization in recent
history. The birthing process has been affected by this cultural change. The extent to which
resiliency and identity affect breastfeeding practices as well as other indicators of health cannot
be empirically evaluated without tools to measure these constructs. We identified a five class
structure of engaging in traditional activities; although there are many various domains of
resiliency, including social support60
, we elected to only explore only this one. Breastfeeding
practices were not associated with class membership, however, we note that age and sex were,
with older individuals and males more likely to belong to classes characterized by spending time
on the land and younger individuals and females more likely to belong to classes characterized
by spending time in the community. These age specific trends may indicate a shift in the central
150
meaning of Inuit identity to a younger generation and should be explored further in both
qualitative and quantitative research.
151
LITERATURE CITED
1. Kirmayer L, Simpson C, Cargo M. Healing traditions: culture, community and mental
health promotion with Canadian Aboriginal peoples. Australasian Psychiatry.
2003;11(s1):S15-S23.
2. Van Wagner V, Epoo B, Nastapoka J, Harney E. Reclaiming Birth, Health, and
Community: Midwifery in the Inuit Villages of Nunavik, Canada. The Journal of
Midwifery & Women’s Health. 2007;52(4):384-391.
3. Bjerregaard P, Young TK, Dewailly E, Ebbesson SO. Indigenous health in the Arctic: an
overview of the circumpolar Inuit population. Scandinavian Journal of Public Health.
2004;32(5):390 - 395.
4. Legare A. Canada's experiment with Aboriginal self-determination in Nunavut: from
vision to illusion. International Journal on Minority and Group Rights. 2008;15:335-367.
5. Tait H. Aboriginal Peoples' Survey, 2006: Inuit health and social conditions. 2008;
http://www.statcan.gc.ca/pub/89-637-x/89-637-x2008001-eng.pdf. Accessed Jan 13,
2013.
6. Rode A, Shephard R. Physiological consequences of acculturation: a 20-year study of
fitness in an Inuit community. European Journal of Applied Physiology and
Occupational Physiology. 1994;69(6):516-524.
7. Rode A, Shephard R. Acculturation and loss of fitness in the Inuit: the preventive role of
active leisure. Arctic medical research. 1993;52(3):107.
8. Kuhnlein HV, Receveur O, Soueida R, Egeland GM. Arctic Indigenous Peoples
Experience the Nutrition Transition with Changing Dietary Patterns and Obesity. The
Journal of Nutrition. 2004;134(6):1447-1453.
9. Kelly J, Lanier A, Santos M, et al. Cancer among the circumpolar Inuit, 1989-2003. II.
Patterns and trends. International Journal of Circumpolar Health. 2008;67(5):408.
10. Young T, Bjerregaard P, eds. Health Transitions in Arctic Populations. Toronto:
University of Toronto Press; 2008.
11. Steenbeek A, Tyndall M, Rothenberg R, Sheps S. Determinants of Sexually Transmitted
Infections Among Canadian Inuit Adolescent Populations. Public Health Nursing.
2006;23(6):531-534.
12. Jasen P. Race, Culture, and the Colonization of Childbirth in Northern Canada. Social
History of Medicine. 1997;10(3):383-400.
13. Kaufert PA, Gilbert P, O'Neil J, et al. Obstetric care in the Keewatin. Changes in the
place of birth 1971-1985. Arctic Medical Research. 1988;47:481.
152
14. Kaufert PA, O'Neil JD. Cooptation and Control: The Reconstruction of Inuit Birth.
Medical Anthropology Quarterly. 1990;4(4):427-442.
15. Chamberlain M, Barclay K. Psychosocial costs of transfering indigenous women from
their community for birth. Midwifery. 2000;16:116-122.
16. Douglas V. Childbirth among the Canadian Inuit: a review of the clinical and cultural
literature. International Journal of Circumpolar Health. 2006;64(2):117-133.
17. Tedford S, O'Neil J, van Wagner V. Examining Midwifery-based options to improve
continuity of maternity care in remote Nunavut communities. 2005;
http://www.naho.ca/inuit/midwifery/documents/2006-03-
09Tetfordarticlefinalshortform.pdf. Accessed March 15, 2009.
18. Van Wagner V, Osepchook C, Harney E, Crosbie C, Tulugak M. Remote Midwifery in
Nunavik, Québec, Canada: Outcomes of Perinatal Care for the Inuulitsivik Health Centre,
2000–2007. Birth. 2012;39(3):230-237.
19. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
20. Asuri S, Ryan A, Arbour L. Early Inuit Child Health in Canada: Breastfeeding among
Inuit in Canada. 2011; https://www.itk.ca/sites/default/files/2011-Report-Breastfeeding-
among-Inuit-in-Canada.pdf. Accessed December 12, 2012.
21. DiGirolamo AM, Grummer-Strawn LM, Fein S. Maternity Care Practices: Implications
for Breastfeeding. Birth. 2001;28(2):94-100.
22. Fleming J, Ledogar R. Resillience, an evolving concept: a review of literature relevant to
Aboriginal research. Pimatisiwin. 2008;6(2):7-23.
23. Kirmayer L, Sehdev M, Whitely R, Dandeneau S, Isaacs C. Community resilience:
models, metaphors and measures. Journal of Aboriginal Health. 2009;November:62-117.
24. Chandler MJ, Lalonde C. Cultural Continuity as a Hedge against Suicide in Canada's
First Nations. Transcultural Psychiatry. 1998;35(2):191-219.
25. Reboussin BA, Song EY, Shrestha A, Lohman KK, Wolfson M. A latent class analysis of
underage problem drinking: Evidence from a community sample of 16− 20 year olds.
Drug and Alcohol Dependence. 2006;83(3):199.
26. Bucholz KK, Heath AC, Reich T, et al. Can we subtype alcoholism? A latent class
analysis of data from relatives of alcoholics in a multicenter family study of alcoholism.
Alcoholism: Clinical and Experimental Research. 2006;20(8):1462-1471.
27. Lanza ST, Collins LM. A new SAS procedure for latent transition analysis: transitions in
dating and sexual risk behavior. Developmental psychology. 2008;44(2):446-452.
153
28. Coatsworth JD, Maldonado‐Molina M, Pantin H, Szapocznik J. A person‐centered and
ecological investigation of acculturation strategies in Hispanic immigrant youth. Journal
of Community Psychology. 2005;33(2):157-174.
29. Tait H. Harvesting and country food: fact sheet. 2007; http://www.statcan.gc.ca/pub/89-
627-x/89-627-x2007001-eng.htm. Accessed Nov 12, 2012.
30. Findlay L, Kohen D. Inuit Children's Leisure Time activities: fact sheet. 2008;
http://www.statcan.gc.ca/pub/89-627-x/89-627-x2008005-eng.htm. Accessed January 13,
2013.
31. Saudny H, Leggee D, Egeland G. Design and methods of the Adult Inuit Health Survey
2007–2008. International Journal of Circumpolar Health.
2012;71:10.3402/ijch.v3471i3400.19752.
32. Egeland GM, Pacey A, Cao Z, Sobol I. Food insecurity among Inuit preschoolers:
Nunavut Inuit Child Health Survey, 2007-2008. CMAJ. 2010;182(3):243-248.
33. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
34. Quandt SA. Material recall accuracy for dates of infant feeding transitions. Human
Organization. 1987;46(2):152-160.
35. Tienboon P, Rutishauser IHE, Wahlqvist ML. Maternal recall of infant feeding practices
after an interval of 14 to 15 years. Australian Journal of Nutrition and Dietetics.
1994;51:25-25.
36. Infant Feeding Joint Working Group. Nutrition for healthy term infants:
recommendations from birth to six months: a joint statement of Health Canada, Canadian
Paediatric Society, Dietitians of Canada, and Breastfeeding Committee for Canada. 2012;
http://www.hc-sc.gc.ca/fn-an/nutrition/infant-nourisson/recom/index-eng.php#a3.
37. Nyland K, Asparouhov T, Muthen B. Deciding on the number of classes in latent class
analysis and growth mixture modeling: a Monte Carlo Simulation study. Structural
Equation Modeling. 2007;14(4):535-569.
38. Lanza S, Collins L, Lemmon D, Schafer J. PROC LCA: A SAS procedure for latent class
analysis. Structural Equation Modeling. 2007;14(4):671-694.
39. Magidson J, Vermunt J. Latent class models for clustering: A comparison with K-means.
Canadian Journal of Marketing Research. 2002;20(1):36-43.
40. Hatcher LI. A step by step approach to using SAS for factor analysis and structural
equation modeling. Cary, NC: SAS Institute Inc.; 1994.
41. Lanza S, Dziak J, Huang L, Xu S, Collins L. PROC LCA & PROC LTA users' guide
(Version 1.2.7). University Park: The Methodology Center, Penn State; 2011.
154
42. Lo Y, Mendell NR, Rubin DB. Testing the number of components in a normal mixture.
Biometrika. 2001;88(3):767-778.
43. Dziak JJ, Coffman DL, Lanza ST, Li R. Sensitivity and specificity of information criteria.
2012.
44. Yang C-C. Evaluating latent class analysis models in qualitative phenotype identification.
Computational Statistics & Data Analysis. 2006;50(4):1090-1104.
45. Goodman LA. On the assignment of individuals to latent classes. Sociological
Methodology. 2007;37(1):1-22.
46. Hosmer D, Lemeshow S. Applied Logistic Regression. 2nd ed. Toronto: John Wiley and
Sons, Inc; 2000.
47. PROC LCA & PROC LTA (version 1.2.7) [computer program]: University Park: The
Methodology Centre, Penn State; 2011.
48. LCABootstrap SAS Macro (version 1.1.0) [computer program]. University Park: The
Methodology Center, Penn State; 2011.
49. Inuit Tapiriit Kanatami. Social determinants of Inuit health in Canada: a discussion paper.
2007; http://ahrnets.ca/files/2011/02/ITK_Social_Determinants_paper_2007.pdf.
Accessed June 20, 2012.
50. Stevens GWJM, Pels TVM, Vollebergh WAM, Crijnen AAM. Patterns of psychological
acculturation in adult and adolescent Moroccan immigrants living in the Netherlands.
Journal of Cross-Cultural Psychology. 2004;35(6):689-704.
51. Notelaers G, Einarsen S, De Witte H, Vermunt JK. Measuring exposure to bullying at
work: The validity and advantages of the latent class cluster approach. Work & Stress.
2006;20(4):289-302.
52. Streiner D, Norman G. Health Measurement Scales: a practical guide to their
development and use, 3rd ed. Toronto: Oxford University Press; 2003.
53. Statistics Canada. Aboriginal Peoples in Canada in 2006: Inuit, Metis and First Nations,
2006 census2009.
54. Abraido-Lanza AF, Armbrister AN, Florez KR, Aguirre AN. Toward a Theory-Driven
Model of Acculturation in Public Health Research. Am J Public Health.
2006;96(8):1342-1346.
55. Hallett D, Chandler MJ, Lalonde CE. Aboriginal language knowledge and youth suicide.
Cognitive Development. 2007;22(3):392-399.
155
56. Wilson K, Rosenberg MW. Exploring the determinants of health for First Nations
peoples in Canada: can existing frameworks accommodate traditional activities? Soc Sci
Med. 2002;55(11):2017-2031.
57. Kral MJ, Adams E, Akoomalik L, et al. Unikkaartuit: Meanings of well-being, sadness,
suicide, and change in two Inuit communities. Final Report to the National Health
Research and Development Programs, Health Canada, Project. 2003:6606-6231.
58. Healey G, Meadows L. Tradition and culture: an important determinant of Inuit women's
health. Journal of Aboriginal Health. 2008:25-33.
59. Government of Nunavut. Tunngasugitti --Welcome to Nunavut! Nunavut, Our Land.
2012; http://www.gov.nu.ca/files/Nunavut,%20Our%20Land.pdf. Accessed October 12,
2012.
60. Richmond CAM, Ross NA, Egeland GM. Social Support and Thriving Health: A New
Approach to Understanding the Health of Indigenous Canadians. Am J Public Health.
2007;97(10):1827-1833.
156
Figure 1: Flow diagram of participants in the IPY Inuit Health Survey (2007-2008) and Nunavut Inuit Child Health Survey (2007-
2008) included in these analyses
157
Table 1: Manifest variables that were considered for Latent Class Analysis of traditional behaviours amongst Canadian Inuit, IPY
Inuit Health Survey (2007-2008)
Label Variable Description Number of
categories
Hunt land In the past 12 months, have you hunted on the land? 2~
Hunt water In the past 12 months, have you hunted on the water? 2~
Fishing In the past 12 months, have you gone fishing? 2~
Cabin In the past 12 months, have you built a cabin on the land? 2~
Dog team In the past 12 months, have you used a dog team on the land? 2~
Igloo In the past 12 months, have you built an igloo on the land? 2~
Distribute Country food Do you or your household distribute country food to other members of the
community? 2
~
Language What is the predominate language spoken in your home? 2~
Community recreation In the past month, have you participated in activities with community members? 5*
Community activity In the past month, have you participated in activities to benefit the community? 5*
~ Question offered a yes or no response option
*Changed to two each: very often, often together vs. not
158
Table 2: Demographic Characteristics of Study Respondents. IPY Inuit Health Survey (2007-2008) and Nunavut Inuit Child Health
Survey (2007-2008)
1 Percent may not equal 100 due to missing data
2 Sample 1: All individuals completing the IPY Inuit Health Survey
3 Sample 2: Individuals completing both the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey with complete information on breastfeeding
initiation and other covariates 4 Sample 3: Individuals completing both the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey with complete information on breastfeeding
duration and other covariates
Characteristic Sample 1
2 (n=2595) Sample 2
3 (n=135) Sample 3
4 (n=108)
%1 %
1 %
1
Age (mean, SD) 42.1 (15.2) 35.6 (11.8) 35.1 (10.4)
Sex
Male 38.5 (36.6-40.3) 47.7 (39.0-55.8) 50.0 (40.6-59.4)
Female 61.5 (59.7, 63.4) 52.6 (44.2, 61.0) 50.0 (40.6-59.4)
Marital Status
Single 24.4 (22.6-26.2) 16.3 (9.9-22.6) 11.7 (5.4-17.8)
Married 64.1 (62.1-66.1) 79.8 (72.9-86.8) 84.5 (77.5-91.5)
Separated/Divorced 4.9 (4.0-5.8) 2.3 (0.0-4.9) 2.9 (0.0-6.2)
Widowed 6.6 (5.6-7.7) 1.6 (0.0-3.7) 1.0 (0.0-2.9)
Educational Attainment
No Certificate or Diploma 28.7 (27.0-30.5) 12.9 (7.2-18.6) 11.3 (5.3, 17.4)
Elementary Diploma 39.0 (37.2-40.9) 54.6 (46.0-63.0) 53.8 (44.3-63.3)
High school Diploma or GED 16.1 (14.6-17.5) 19.7 (12.9, 26.5) 21.7 (13.8-29.5)
At least some College or University 16.2 (14.8,17.6) 12.9 (7.2-18.6) 13.2 (6.7-19.6)
159
Table 3: Traditional Behaviours in Study Respondents. IPY Inuit Health Survey (2007-2008)
and Nunavut Inuit Child Health Survey (2007-2008)
Variable Prevalence in
Sample 12
n=2595
(95% CI)1
Prevalence in
Sample 23
n=135
(95% CI)1
Prevalence in
Sample 34
n=108
(95% CI)1
Hunt land 41.8 (40.0-43.7) 49.6 (40.9, 58.4) 50.0 (40.2, 59.8)
Hunt water 30.6 (28.9-32.4) 42.2 (33.8, 51.0) 39.8 (30.5, 49.7)
Fishing 36.5 (34.6-38.3) 40.0 (31.7, 48.8) 37.8 (28.8, 47.8)
Cabin 40.0 (38.1-41.9) 48.9 (40.2, 57.6) 48.2 (38.4, 57.9)
Dog team 46.8 (44.9-48.8) 47.1 (38.7, 56.2) 45.4 (35.7, 55.2)
Igloo 31.3 (29.5-33.2) 39.2 (31.0, 48.0) 37.9 (28.9, 47.8)
Language 56.5 (54.3-58.8) 73.2 (63.2, 81.7) 71.6 (60.5, 81.1)
Distribute CF 76.3 (74.3, 78.2) 83.8 (75.8, 90.0) 82.2 (73.3, 89.1)
Community recreation 45.0 (43.0, 46.9) 42.6 (33.1, 52.5) 42.6 (33.1, 52.5)
Community activity 45.9 (43.9, 47.8) 35.2 (26.3, 45.0) 35.2 (26.3, 45.0) 1 95% Confidence Interval
2 Sample 1: All individuals completing the IPY Inuit Health Survey
3 Sample 2: Individuals completing both the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey with
complete information on breastfeeding initiation and other covariates 4 Sample 3: Individuals completing both the IPY Inuit Health Survey and Nunavut Inuit Child Health Survey with
complete information on breastfeeding duration and other covariates
Table 4: Fit Statistics for Latent Class analysis of traditional behaviours amongst Canadian Inuit.
IPY Inuit Health Survey (2007-2008) (n=2595)
1Akaike’s Information Criteria
2Adjusted Bayesian Information Criteria
3Bootstrap Likelihood Ratio Test
Number of
Classes
AIC Value1
Adjusted BIC Value2
BLRT3
One 3308 3332 -
Two 1330 1376 p=0.01
Three 397 467 p=0.01
Four 277 372 p=0.01
Five 231 350 p=0.01
Six 234 377 p=0.87
160
Table 5: Class and item membership probabilities for traditional behaviours amongst Canadian Inuit and odds of class membership by age
and sex, IPY Inuit Health Survey (2007-2008) (n=2595)
1 Item response probabilities indicate the probability of responding affirmatively to any of the characteristics of interest
Class One Class Two Class Three Class Four Class Five
% in class (SE)
32.7 (1.4) 26.2 (1.1) 26.0 (3.6) 7.5 (1.7) 7.5 (3.0)
Item response probabilities1: % (SE)
Hunt land 13.6 (1.9) 0.9 (0.7) 89.0 (2.0) 98.2 (1.7) 87.5 (5.3)
Hunt water 4.0 (1.1) <0.1 (0.2) 71.9 (2.7) 86.7 (4.0) 54.1 (6.9)
Fish 29.4 (2.0) 1.9 (0.8) 88.7 (1.8) 98.0 (1.5) 82.9 (5.0)
Cabin 1.8 (0.6) <0.1 (0.0) 20.4 (2.2) 55.8 (5.8) 14.7 (4.9)
Dog team 1.1 (0.4) 0.3 (0.2) 3.2 (1.0) 43.2 (4.4) 0.1 (1.3)
Igloo 0.7 (0.4) 0.2 (0.2) 6.6 (1.9) 63.6 (10.5) 2.6 (3.3)
Community recreation 18.1 (2.1) 98.6 (1.9) 12.7 (7.8) 51.0 (5.6) 92.2 (14.2)
Community activity 17.2 (2.3) 99.4 (1.1) 17.7 (3.2) 55.5 (5.6) 79.6 (14.6)
Odds ratiofemale
95% CI 1.00
189.0
(0.0-∞)
0.03
(<0.00, 0.57)
0.04
(0.02, 0.07)
0.38
(0.29, 0.51)
Odds ≥42 1.00 0.78 2.67 2.09 1.65 95% CI (0.50, 1.20) (1.67, 4.32) (1.26, 3.45) (1.32, 2.06)
161
Table 6: Interpretations of classes from Latent Class Analysis of traditional behaviours amongst Canadian Inuit. IPY Inuit Health Survey
(n=2595)
Latent Class Class interpretation
Class One Lowest participation in all types of traditional activities
Class Two Low participation in land based activities, high participation in community based activities
Class Three High participation in land based activities to acquire food, low participation in other traditional activities
Class Four High participation in land based activities to acquire food, moderate participation in other traditional activities
Class Five High participation in land based activities to acquire food and community activities, low participation in other
activities on the land
162
Table 7: Prevalence Odds (pOR) of breastfeeding initiation, by class membership of traditional
behaviours. IPY Inuit Health Survey (2007-2008) and Nunavut Inuit Child Health Survey (2007-
2008) (n=135)
Latent Class Crude pOR1, 95%
CI2
(n=135)
Adjusted3 pOR
1,
95% CI2
(n=135)
Class One 1.00 1.00
Class Two 1.03 (0.21, 4.92) 1.67 (0.31, 9.14)
Class Three 0.49 (0.14, 1.71) 0.38 (0.15, 2.28)
Class Four 0.60 (0.13, 2.67) 0.88 (0.24, 4.52)
Class Five 0.98 (0.28, 3.44) 0.95 (0.22, 4.16)
1 pOR is the prevalence odds ratio, calculated from logistic regression models using breastfeeding initiation as the
outcome 2 95% CI is the 95% confidence limits calculated using the Wald method
3 Models are adjusted for infant birth weight, maternal smoking during pregnancy and household overcrowding
(defined as more than one person per bedroom)
Table 8: Prevalence Odds (pOR) of exclusive breastfeeding as recommended1, by class
membership of traditional activities. IPY Inuit Health Survey (2007-2008) and Nunavut Inuit
Child Health Survey (2007-2008) (n=108)
Latent Class Crude pOR2, 95%
CI3
(n=108)
Adjusted4 pOR
2,
95% CI3
(n=108)
Class One 1.00 1.00
Class Two 0.85 (0.13, 5.49) 0.78 (0.10, 6.20)
Class Three 0.56 (0.15, 2.04) 0.44 (0.10, 1.94)
Class Four 1.82 (0.27, 12.43) 1.16 (0.14, 9.31)
Class Five 3.78 (0.40, 36.25) 2.36 (0.23, 24.79)
1 Exclusive breastfeeding as recommended defined as exclusive breastfeeding for 5.5 to 6.5 months
2 pOR is the prevalence odds ratio calculated from logistic regression models, using appropriate exclusive
breastfeeding as the outcome 3 95% CI is the 95% Confidence limits calculated using the Wald method
4 Models are adjusted for receiving household income support
163
Chapter 7 : Discussion and Conclusions
7 Discussion Overview
The study results have been discussed in each of the specific manuscripts. Chapter 4
(Manuscript #1), discusses the findings relevant to the first and second thesis objectives; Chapter
5 (Manuscript #2) discusses the findings relevant to the third and fourth thesis objectives; and
Chapter 6 discusses the findings relevant to the fifth and sixth thesis objectives (Manuscript #3).
This chapter discusses some of the most important overall findings, addresses some of the
methodologic issues of the research, suggests how this research applies to public health and
recommends avenues for future research.
7.1 Main findings
Taken together, this research reveals that there is substantial room for improvement in the
breastfeeding practices of Inuit Canadians based on two standard indicators1: breastfeeding
initiation and exclusive breastfeeding, as recommended.
The prevalence of breastfeeding initiation was 67% in the combined sample of all caregivers (i.e.
biological mothers, biological fathers and non-biological parents) and 76% in children raised by
their biological mothers, as reported in the Nunavut Inuit Child Health Survey. This is
substantially lower than the 90% reported among all Canadians2,3
. Among those who initiated
breastfeeding, the proportion of Inuit women who practiced exclusive breastfeeding to six
months, as recommended was 23%, seemingly more comparable to Canada wide estimates
(13.5% to 23.5%) over the same time period (2006-2007)2-4
, yet Canada wide estimates include
all children, and not just those who initiated breastfeeding. Among all biological mothers (i.e. not
just those initiating breastfeeding) in the North, the proportion who practiced exclusive
breastfeeding as recommended was 18%. These findings seemingly contradict other findings that
suggest Canadians in the Northern Territories are three times more likely to be exclusively
breastfeeding at six months3, but these data are limited. These previous studies have grouped
together both Inuit and non-Inuit, combined all three territories together despite considerable
variation between them (e.g. exclusive breastfeeding duration in Yukon: 34.2% and in Northwest
164
Territories: 18.2%)2 and failed to distinguish between those practicing exclusive breastfeeding
beyond six months3.
This research suggests that although some factors that were associated with breastfeeding
initiation in Inuit Canadians were similar to those among other Canadians, variability was
observed. This research showed that low birth weight infants were less likely to begin
breastfeeding, which is consistent with other research5,6
. Our finding that household crowding
was associated with increased initiation was novel and it could be that household crowding was a
proxy for other variables associated with breastfeeding histories, such as parity and previous
breastfeeding experience7. The pathways to which this association arises should be investigated
more closely. Also of note, the factors related to breastfeeding initiation varied across primary
caregiver group, with community factors seeming particularly important for children currently
being raised by adoptive parents.
In terms of breastfeeding duration, receiving income support was the only factor significantly
associated with exclusive breastfeeding as recommended in multinomial models among
Canadian Inuit (discussed in further detail in Chapter 5). This is likely related to study power
(discussed further in 7.3.4). A recent examination of factors associated with exclusive
breastfeeding duration among Canadians3 identified twelve characteristics that were predictive of
exclusive breastfeeding as recommended. Most of these factors were not measured for all
participants in the Nunavut Inuit Child Health Survey (e.g. maternal employment status,
maternal BMI and self-rated health), and many novel factors which were included in these
analyses (e.g. food security, household crowding, language use at home) were not considered in
the Canada-wide sample. The disparate methods and factors included in this thesis study and the
Canada-wide study makes comparisons challenging.
One of the objectives of this thesis was to explore the effect of engaging in traditional activities -
an indicator for cultural resiliency - on breastfeeding practices and these data suggest there was
insufficient evidence to suggest an association. As noted in Chapter 6, failing to find an
association does not rule out the fact that one exists. Some possible reasons for this finding
include: misspecification of the latent model; assigning individuals to classes based on posterior
probabilities instead of modeling within the latent class framework; conducting the analysis on a
165
reduced sample that is fundamentally different in class structure compared to the original
sample; or chance alone.8
Overall, this thesis provides some insight into the breastfeeding practices of Inuit Canadians, yet
simply considering the proportion of Inuit infants who meet the exclusive breastfeeding
recommendation will not provide a comprehensive understanding of breastfeeding histories.
Continued breastfeeding up to two years and beyond is recommended by both the World Health
Organization and Health Canada1. Two additional recommended indicators are: the proportion
of Inuit children who continue to be breastfed at two years and age-appropriate breastfeeding1.
The first indicator is calculated by dividing the number of infants aged 20 to 23 months who
received breast milk in the previous day by the total number of infants aged 20 to 23 months1.
The second indicator considers both the proportion of infants who were exclusively breastfeed to
six months as well as the proportion of infants who received continued breastfeeding to two
years1.
Exploring other breastfeeding practices were not objectives for this thesis; however, data from
the Nunavut Inuit Child Health Survey suggest that 32% of infants raised by their biological
mother received continued breastfeeding to one year and 27% to two years, and a total of 9% of
the Inuit population received age-appropriate breastfeeding. Estimates on age-appropriate
breastfeeding in Canada do not exist nor do population-based estimates of continued
breastfeeding up to two years. However, in the United States, where the observed proportion of
women initiating and practicing exclusive breastfeeding to six months is similar9, 25.5% of
American women practiced continued breastfeeding to one year.9 Assuming Canadian estimates
are similar, these data suggest that while Inuit women were less likely to practice optimal
exclusive breastfeeding, they were more likely to continue breastfeeding to at least one year and
beyond. Evidence from the Aboriginal Peoples’ Survey and the Canadian Community Heath
Survey, which reported rates of any breastfeeding at six months, also suggest that Inuit
Canadians continue any breastfeeding for a longer duration than other Canadians10,11
.
7.2 Strengths
The decision to breastfeed is driven, at least partially, by cultural norms12
, underscoring a
fundamental need to examine factors related to both breastfeeding initiation and exclusive
166
breastfeeding duration in well-defined cultural and ethnic groups. This thesis provides a
population-based exploration of breastfeeding practices among Inuit Canadians.
Using data from the Nunavut Inuit Child Health Survey allowed for the observation of
breastfeeding practices among Inuit Canadians that would not have been possible otherwise.
Statistics Canada sponsored surveys are limited in that they either do not gather information
about Aboriginal identity (i.e. Maternity Experiences Survey2) or do not specifically capture the
duration of exclusive breastfeeding (i.e. Aboriginal Peoples’ Survey10
; Aboriginal Children’s
Survey11
). That a substantial proportion of Inuit women are actually breastfeeding their infants
exclusively beyond six months would not have been observed from any survey other than the
Nunavut Inuit Child Health Survey.
This thesis is not only the first to examine determinants of breastfeeding histories in an Inuit-
specific population, but also the first to examine the effects of certain social determinants of
health (e.g. crowded housing; food security) on breastfeeding histories. These social
determinants, as well as many of the community factors that were included (e.g. access to a
birthing centre or hospital in a community) are important concerns in remote and isolated Inuit
communities. Moreover, although it has been previously noted that Inuit children who are
adopted are less likely to begin breastfeeding11,13
; there has been no research about breastfeeding
practices in this unique, though prevalent (18%) segment of the population.
Conducting research in Canada’s North can be particularly challenging. Logistically, all
Northern communities are remote or isolated; because no roads link communities together, travel
between is either by plane or boat amounting to a costly endeavor14
. Many epidemiologic studies
conduct research in only a few northern communities at most15
, presumably to overcome at least
some of these financial constraints. Such an approach will minimize the overall cost of research,
yet Northern communities are sparsely populated (Iqaluit is home to the largest population in
Nunavut with approximately 6,700 residents, as per the 2011 census16
). The result of these
single or few community studies is that the sample sizes are too small to make powerful
conclusions about health associations14
. This is further amplified by the deep chasm that can
often exist between Aboriginal researchers and the Aboriginal peoples’ themselves. These
differences stem from historical marginalization and colonization of Inuit17
, disparate views
about research priorities (i.e., questions are developed by non-Aboriginals and may not reflect
167
and/or benefit community/Inuit interests)18
, and previous distrust with research teams
disseminating their results only to the scientific community and not the Inuit themselves18
.
The creation of the self-governing territory, Nunavut, on April 1, 199919
, provided an ideal
opportunity to protect, promote and preserve Inuit culture. The effects of this cultural
empowerment on health have not been explored, in spite of the belief that these cultural changes
will improve health and well-being 20-22
. This thesis is the first to create a latent model of
traditional activities in Inuit communities, which was used as a marker for the complex construct
- cultural resiliency.
7.3 Limitations
One of the central goals of an observational study (e.g. cross-sectional surveys) is to make causal
inferences about the effect of various exposures on the outcome of interest. Unlike randomized
trials, exposures are self-selected (including intensity and duration) and not assigned on a chance
basis. This likely introduces confounding and other biases that need to be addressed before
moving to discussions of causality. Various limitations of this thesis related to selection bias,
information bias, confounding and statistical error are discussed below.
7.3.1 Selection Bias
Selection bias arises when there are systematic differences between those who participate and
those who do not participate in a study23
and these differences subsequently distort the true
association between the exposure and outcome.24
Of the 644 households sampled in the Nunavut Inuit Child Health Survey, 107 (16.6%) were
away or could not be contacted during recruitment, and an additional 147 (22.8%) did not
complete the survey for reasons listed in Chapter 4. The remaining sample (n=388, 62.2%) was
further reduced by inclusion and exclusion criteria, specific to each manuscript. Collecting the
basic demographic characteristics of those who refuse to participate and comparing these to the
characteristics of those who actually participate can sometimes, but not always25,26
be useful for
examining the extent to which participant selection may be biasing study results. However, there
no data were collected on non-responders in the Nunavut Inuit Child Health Survey.
168
A specific example of how selection bias from survey non-response may be affecting results
from this thesis relates to the timing of participant selection and data collection – in the summer
and fall months of 2007-2008. During the summer and fall months, many Inuit would have been
spending time on the land. Had they been on the land during the two weeks of recruitment, they
would not be able to participate in the survey. Indicators of spending time on the land were
important components of the latent class model. Including these missed individuals may have
changed the overall class structure identified in Chapter 6. Even if the class structure was stable
after including non-responders, the relative measure of effect of resiliency and breastfeeding
practices may have been misestimated.
Another way in which selection bias may have arose is using the reduced sample of respondents
who completed both the IPY Inuit Health Survey and the Nunavut Inuit Child Health Survey for
the analysis of the relationship between latent class membership and breastfeeding histories. It is
possible that the latent class structure in the smaller sample of participants completing both
surveys was different from the latent class structure in the full IPY sample which was used to
construct latent models. Re-running the latent class analysis on only those participants
completing both the Nunavut Inuit Child Health Survey and the IPY Inuit Health Survey
revealed that a four class structure was more likely in this group.
Missing Data
In an epidemiologic study, missing data can arise in three ways27,28
. First, the data can be missing
completely at random, meaning there are no systematic differences between observed and
missing data28
; when item non-responders are removed from the analysis (i.e. list-wise deletion),
the underlying assumption is that data are missing completely at random28
. Second, the data can
be missing at random, meaning there are differences between observed and missing data, but the
values of missing data can be predicted from other observed variables. This is an underlying
assumption for most imputation methods27
. Finally, data can be not missing at random, meaning
there are systematic differences between observed and missing data and these differences cannot
be predicted by other observed variables27,28
.
In these analyses, all missing values were excluded from the analyses (i.e. list-wise deletion),
effectively inferring that data were missing completely at random. The variable with the highest
169
degree of non-response was receiving income support: 54 mothers (21%), 12 adopted parents
(19%) and 6 fathers (15%) provided complete information on breastfeeding initiation but did not
respond to the question “In the past month did anyone in your household receive income
support” (Chapter 4). It is possible that individuals who received income support would be less
likely to respond to this particular survey question because of social desirability29
, shame or
embarrassment. It is also likely that other variables (e.g. education, income, marital status) could
predict missing-ness, yet these variables were not available on all surveys, limiting indicators for
multiple imputation procedures. Although income support was associated with exclusive
breastfeeding as recommended (Chapter 5), it is not clear if non-responders would have been
different in their breastfeeding practices than responders. Because of small sample sizes,
subgroup analyses using only those with full information on socioeconomic position were not
very informative.
Sampling Bias
Community medical records were used to select caregivers in the Nunavut Inuit Child Health
Survey. This is the sole source of entry to medical care in the community, and is expected to
have close to, if not, 100% coverage. Housing records were maintained by the community and
were used for the IPY Inuit Health Survey sampling frame. These are also expected to have
close to 100% coverage.
Incidence-Prevalence Bias
Incidence-Prevalence bias is a common concern in cross-sectional research29
and occurs when
effects of exposures are evaluated in prevalent instead of incident cases, and survivorship (i.e.
being a prevalent case) is associated with the exposure(s) of interest, resulting in misestimation
of effects.
In order to be included in the Nunavut Inuit Child Health Survey, children had to have been alive
and in the community at the time of data collection. Infant mortality is 3.6 times higher in Inuit
inhabited regions of Canada compared to other areas of Canada: 16.5 deaths per 1,000 live
births vs. 4.6 deaths per 1,000 live births30
. If the Inuit Health Survey was conducted with the
same cohort of children but started shortly after they were born, an estimated 10.5 children out of
644 (i.e. 16.5/1000 x 644) would have been included that had died by the time of the 2007-2008
170
survey. After applying population-based weights, these children may have represented anywhere
from 13 to 70 Inuit infants.
Exclusive breastfeeding is associated with reduced infant mortality31
and morbidity (e.g.
infections)32
. Selection bias would arise if the relationship between any of the independent
variables of interest and breastfeeding practices were different among those eligible to sampled
(i.e. alive and in the community when they were aged 3 to 5) compared to those who were not
eligible (i.e. those who died before and were not in the community 2007 and 2008). To
illustrate, one of the independent variables of interest was low birth weight. Low birth weight
has been associated with a lower likelihood of breastfeeding initiation5 and also with an
increased risk of infant mortality33-35
. Although a strong inverse association between infant birth
weight and breastfeeding initiation (Chapter 4) was observed, the magnitude of effect may have
been greater had the data been collected shortly after the child’s birth.
7.3.2 Information Bias
Measurement Error
Breastfeeding initiation and exclusive breastfeeding duration were the primary outcomes of
interest in this thesis. Recall of breastfeeding initiation among biological mothers is high
compared to medical records (r=0.77 to r=94) even after lengthy periods of time (i.e. 15 years)36
.
But, this thesis also explored the determinants of breastfeeding initiation in biological fathers and
adoptive parents for which the validity of recall of breastfeeding initiation is uncertain. Subgroup
analyses suggest the effects of various exposures (e.g. low birth weight) were different across
caregiver groups. It is possible that some of this may be related to misclassification, the
magnitude of which cannot be predicted. Stratifying and conducting analysis by caregiver sub-
type (Chapter 4) helps to ensure that our analysis of factors related to breastfeeding in biological
mothers were not biased by differential validity of recall in various caregiver groups.
Recall of exclusive breastfeeding duration is a bit more complicated: even amongst biological
mothers, retrospective recall of the timing of the transition from breast milk to infant formula
agrees 58%36
of the time with medical records within one month. These estimates come from a
population of mothers who recently stopped breastfeeding (within the past 6 months): it is
unclear how longer time periods (up to five years in our study) would affect the validity of
171
recalled breastfeeding duration. Moreover, the agreement between maternal recall of the timing
of introduction of complementary foods into the diet and other external validation measures can
range from as high as 88%37
to as low as 65%38
, depending on the period of recall (one year37
to
14-15 years).
The World Health Organization recommends using 24 hour dietary recall for the measurement of
exclusive breastfeeding, but such an approach requires a young infant (aged 0 to 6 months)1.
The methods used in the Nunavut Inuit Child Health Survey to measure exclusive breastfeeding
are fairly compatible to those used in the Canadian Health Measures Survey (i.e. “how long was
your infant feed only breast milk?)39
. Other Canadian Surveys (i.e. the Canadian Community
Health Survey and the Maternity Experiences Survey) ask about timing of introduction of
formula into the diet as well as timing of introduction of other complementary foods into the
diet, which may or may not be more accurate than asking about “exclusive” breastfeeding.
Temporal Bias
Sir Bradford Hill’s article “The environment and disease: association or causation”40
presents
nine considerations for assessing causality in observational studies. Although Hill never
intended for these considerations to be applied universally as checklist for causal inference, they
have been widely used as such. However much criticism this practice has generated28,41
,
epidemiologists’ largely agree on one of Hill’s central ideas – in order to infer causation, a clear
temporal sequence between exposure and outcome needs to be demonstrated28,42,43
.
The Nunavut Inuit Child Heath Survey was conducted in 2007 and 2008 and evaluated the past
breastfeeding histories of Inuit children who were aged 3 to 5 years at the time of data collection.
Many of the other measures used, with the exception of infant birth weight and maternal
behaviours during pregnancy, represent the participants’ present status (in 2007 and 2008).
It is unclear if participants’ status changed from the time the child was born (e.g. 2002-2005) to
the time the data were collected. Research indicates few Inuit migrate out of their
communities44,45
, suggesting little misclassification of community measures. However,
individual measures may be subject to greater misclassification. It is likely that household
crowding increased over the three to five year period given the high birth rate46
and housing
shortages in the North47
. Food security may have also increased over time48
.
172
These migrations across exposure categories may have impacted overall effect estimates. For
example, if food insecure homes were more likely to practice exclusive breastfeeding to six
months compared to food secure homes, the migration of food secure individuals to the insecure
category over the time period would have diluted the true relative effect. Yet, this could not be
assessed with the current data.
7.3.3 Confounding
Residual Confounding
There is some debate as to whether residual confounding is a variable or a variable and model-
based concept24,49
. In the first instance (i.e. a variable-based concept), residual confounding is
confounding that persists after unsuccessful attempts to adjust for it49
and can arise from
imperfect measurement of confounding factors or misclassification of confounding variables. In
the second instance (i.e. a variable and model-based concept), residual confounding can arise not
only from issues of measurement and misclassification, but also from failing to include relevant
confounding variables in statistical models24
. The variable- and model-based concept definition
was used in this discussion.
Many indicators that have been associated with breastfeeding initiation and exclusive
breastfeeding duration in other populations were not available in the full data set. These include
demographic indicators (e.g. age)7; indicators of socioeconomic position (e.g. education,
income)7 and psychosocial indicators (e.g. maternal self-efficacy)
50. Failing to include these
potentially important factors in these analysis may have resulted in measures of association that
were under- or over-estimated.
7.3.4 Statistical Concerns
Type 1 and Type 2 Error
All statistical tests were two-tailed with an alpha of 0.05 to address type 1 error.
A priori power calculations indicated that a difference of at least 12% in exclusive breastfeeding
duration (assuming approximately 30% of the population practices appropriate exclusive
breastfeeding) across various exposure categories would be needed for at least 70% power
(Appendix E). The proportion of women initiating breastfeeding did not exhibit substantial
173
variability across most of our exposure categories: the breastfeeding initiation rate was the same
in both food secure and insecure mothers (76%); within 2% for speaking Inuit languages at home
(75% and 77%); and within 3% for living in a community with a birthing centre (74% and 77%).
Although there was greater variability in the four exclusive breastfeeding duration categories (i.e.
≤1 month; >1 to < 5.5 months; 5.5 to 6.5 months; > 6.5 months), power calculations were
performed before the distributional patterns of exclusive breastfeeding were known and reflect
the power needed for a binary outcome variable.
Wide confidence limits echo the substantial amount of variability in these data (i.e. low
precision). Given the population of Inuit Canadian is sparse (approximately 50,000 total) of
which 12% are under the age of 551
, conducting research that can make robust conclusions in
age-specific samples is challenging. Multi-level research is further limited by the number of
Inuit communities (n=36) in Canada.
Variable Definitions: Exposures
All of the predictor variables in these analyses were categorical and most were binary. While
some of these categorical variables were inherently nominal (e.g. languages spoken at home),
others (e.g. maternal smoking behaviours during pregnancy; household food security) were at
least ordinal. Grouping together ordinal or continuous data may mask the true shape of the
relationship between exposure and outcome variables (e.g. J-shape; U-shape; dose-response,
threshold).
When data with higher resolution were available (e.g. household food security; number of drinks
per week during pregnancy), the original groupings were retained. If the relationship did not
appear to exhibit a pattern that would benefit from multiple levels of the variable, the number of
levels of the variable were reduced to increase statistical power and reduce the total number of
degrees of freedom in the model. For example, originally a three-level food security variable
was used to estimate associations with breastfeeding initiation, but this was reduced to a two
level variable, since risk estimates were relatively homogeneous in the two food insecure
categories (i.e. there was no effect). Continuous or ordinal data were not available for all
possible indicators (e.g. number of cigarettes smoked per day during pregnancy); therefore some
important relationships may be missed. Yet, when continuous data were available, there tended
174
to be a high proportion of missing values. Using binary variables increased the overall sample
size used in these analyses.
Effect Estimates
This research uses prevalence odds ratios to communicate the effect of various factors on
predicting breastfeeding histories of Inuit children. There was much debate in the epidemiologic
literature in the mid- to late 1990’s and early 2000’s about the appropriate measures of effect to
use in cross-sectional research52-58
. The overwhelming consensus reached was that prevalence
ratios (also referred to as prevalence rate ratios) are preferred to prevalence odds ratios,
particularly when outcomes are common. Prevalence odds ratios always overestimate the
prevalence ratio and can either over or underestimate the incidence rate ratio, depending on the
duration of follow-up57
. The prevalence ratio, however, is always a more conservative measure
of effect compared to the incidence rate ratio57
. The extent to which prevalence odds ratio
overestimates the prevalence ratio depends on the frequency of the outcome (in this thesis,
breastfeeding histories) and also the frequency of the exposure (e.g. infant birth weight,
household food security), though to a lesser extent58
. The more common both the outcome and
exposure are in the source population, the more the prevalence odds ratio will overestimate the
prevalence ratio58
. Notwithstanding, odds ratios are still valid measures of association, as long
as they are interpreted correctly (i.e. as a measure of relative odds and not relative risk)
Although neither of the breastfeeding outcomes used in this thesis could be classified as rare, the
odds ratio was selected as the measure of effect. Reviewing the Canadian research on predictors
of breastfeeding histories reveals that odds ratios are the most commonly reported measure of
association3,59
, even when prospective research data are used5,60
. Reporting prevalence odds
ratios can more easily facilitate comparisons with other Canadian breastfeeding studies.
Moreover, electing to report a more conservative measure of association (i.e. prevalence ratio)
may result in the undervaluing of the true importance of factors associated with breastfeeding
histories if this research is read by those who are not aware of the nuances of different measures
of association.
175
7.3.5 External Validity
The population-based nature of the IPY Inuit Health Survey and the Nunavut Inuit Child Health
Survey suggests that these results can be extended cautiously to the population of Nunavut and
likely other Inuit inhabited regions of Canada.
7.4 Public Health Significance
Adhering to the recommended breastfeeding practices may prevent many health related events
which are prevalent in the North. For example, promoting exclusive breastfeeding may reduce
the high rates of infectious illness amongst Inuit infants61,62
and may also help to improve birth
spacing in Inuit women. The latter may be particularly important for younger women – it has
been estimated that 20%of Inuit women aged 15 to 19 years are mothers63
. Difficulties accessing
contraceptives in Northern communities, as reported by Inuit women and youth to the Pauktuutit
Inuit Women’s Association, may be partially responsible for these high fertility rates64
.
Improving breastfeeding practices in this community may help prevent unintended pregnancies
occurring in rapid succession among teenaged Inuit mothers.
This research indicates much room for improvement in Canada’s North with respect to
breastfeeding practices: initiation rates, even amongst biological mothers, and optimal exclusive
breastfeeding as recommended are the same or lower than other Canadians.2 This supports
continued public health education measures. Regular evaluations of these measures will help
ensure culturally appropriate public health interventions are delivered effectively. These
measures can be simple health promotion activities, educational messages or continued support
through programs like the Canada Prenatal Nutrition Program.
Public health campaigns can target high risk individuals (i.e. causes of cases)65
. The relative
effect estimates can be used to inform which segments of the population are particularly
vulnerable to not initiating or practicing exclusive breastfeeding as recommended. For example,
this research reinforces the need to support mothers of low birth weight babies to improve
breastfeeding initiation. These supports may be most effective if administered in the hospital as
soon as possible after giving birth (e.g. continued pumping while infant in ICU).
176
Public health campaigns can also be delivered in high risk communities. This research was also
able to clarify the role of community factors (e.g. prenatal nutrition programs) in promoting
appropriate breastfeeding practices, especially for parents with adopted children.
Two of the most important findings for public health relate to the breastfeeding of adopted
children and the extended duration of exclusive breastfeeding in a substantial portion of infants.
1) Adopted parents are more likely to report their infant began breastfeeding when
supportive services were available in the community (e.g. the Canada Prenatal Nutrition
Program). Continued funding of these programs may help improve breastfeeding
outcomes in a fairly prevalent segment of the Inuit population (i.e. adopted children).
2) A substantial portion of Inuit Canadians continued practicing exclusive breastfeeding
beyond six months - a public health concern in its own right. Public health campaigns in
Northern Canada need to emphasize not only the benefits of breastfeeding exclusively to
six months, but also emphasize the importance of timely introduction of nutrient rich
complementary foods into the diet.
7.5 Future Directions for Further Research
Based on the results of this thesis, six key areas for further research are recommended:
1) This research created a latent classification system for participating in Inuit activities,
which was used as an indicator of resiliency. This latent variable needs to be tested for
validity and reliability and further modified, as necessary.
2) Ensuring all Inuit infants begin and continue breastfeeding to six months, and not just
those infants raised by their biological mother, will require further exploration of viable
alternatives, such as breast milk banks. Given the high prevalence of adoption in Inuit
communities, addressing this issue has important implications to the population
attributable risk of not breastfeeding. I encourage qualitative research among biological
mothers who are still living with their adopted children, to discover what supports they
need to initiate and continue breastfeeding to six months. I also encourage interviews
with adoptive parents to determine acceptable viable alternatives to infant formula (e.g.
breast milk banks).
177
3) This study was limited by sample size. Combining Nunavut Inuit Child Health Survey
data with other population-based survey data, or more recent local and regional data, can
increase the statistical power and improve the precision of effect estimates. Moreover,
extending these analyses to the entire Inuit Nunangat and not just Nunavut would
increase the total number of communities, making multilevel models a more viable
analytic strategy. This would help clarify any independent and joint effects of
community on breastfeeding practices.
4) I recommend prospective studies beginning before the infant is born. These studies will
not only provide stronger evidence with respect to factors associated with breastfeeding
initiation and exclusive breastfeeding as recommended, but can also provide insight about
risk factors for other health outcomes (e.g. infant mortality) that disproportionately affect
Inuit Canadians.
5) I suggest ongoing surveillance of the breastfeeding practices in Inuit Nunangat.
Understanding temporal, demographic and geographic trends through ongoing
surveillance can help manage public health programs and direct public policy with
respect to breastfeeding initiation and exclusive breastfeeding duration.
6) Understanding how to most effectively deliver public health messages regarding
breastfeeding to Inuit women will be necessary to improve breastfeeding initiation and
exclusive breastfeeding practices to rates equal to or better than other areas of Canada.
This will require focused research with Inuit women and may also require public health
measures be implemented among the urban health care providers who are responsible for
delivering the infants of Inuit women who are evacuated from their communities to give
birth.
178
7.6 Section References
1. World Health Organization. Indicators for assessing infant and young child feeding
practices. 2008; http://whqlibdoc.who.int/publications/2008/9789241596664_eng.pdf.
Accessed December 12, 2012.
2. Chalmers B, Levitt C, Heaman M, et al. Breastfeeding Rates and Hospital Breastfeeding
Practices in Canada: A National Survey of Women. Birth. 2009;36(2):122-132.
3. Al-Sahab B, Lanes A, Feldman M, Tamim H. Prevalence and predictors of 6-month
exclusive breastfeeding among Canadian women: a national survey. BMC Pediatrics.
2010;10(20):doi:10.1186/1471-2431-1110-1120.
4. Health Canada. Trends in breastfeeding practices in Canada (2001-2009-2010). 2012;
http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/prenatal/trends-tendances-
eng.php. Accessed Dec 8, 2012.
5. Simard I, O'Brien H, Beaudoin A, et al. Factors influencing the initiation and duration of
breastfeeding among low-income women followed by the Canada prenatal nutrition
program in 4 regions of Quebec. J Hum Lact. 2005;21(3):327-337.
6. England L, Brenner R, Bhaskar B, et al. Breastfeeding practices in a cohort of inner-city
women: the role of contraindications. BMC Public Health. 2003;3(1):28.
7. Scott J, Binns C. Factors associated with the initiation and duration of breastfeeding: a
review of the literature. Breastfeed Rev. 1999;7:5 - 16.
8. Lanza S, Collins L, Lemmon D, Schafer J. PROC LCA: A SAS procedure for latent class
analysis. Structural Equation Modeling. 2007;14(4):671-694.
9. Centers for Disease Control and Prevention. Breastfeeding Report Card - United States,
2012. 2002; http://www.cdc.gov/breastfeeding/pdf/2012BreastfeedingReportCard.pdf.
Accessed December 12, 2012.
10. Guevremont A, Kohen D. Inuit in Canada: Findings from the Aboriginal Peoples' Survey
- Survey of Living Conditions in the Arctic. The Health of Inuit Children: Fact Sheet.
2007; http://www.statcan.gc.ca/pub/89-627-x/89-627-x2007002-eng.pdf. Accessed June
3, 2012.
11. Asuri S, Ryan A, Arbour L. Early Inuit Child Health in Canada: Breastfeeding among
Inuit in Canada. 2011; https://www.itk.ca/sites/default/files/2011-Report-Breastfeeding-
among-Inuit-in-Canada.pdf. Accessed December 12, 2012.
12. Stuart MacAdam P, Dettwyler KA. Breastfeeding: biocultural perspectives. New York:
Transaction Publishers; 1995.
13. Dufour R. Breast feeding and adoption among the Inuit population of Northern Quebec.
Can J Publ Health. 1984;75(2):137-140.
179
14. Young T, Bjerregaard P, eds. Health Transitions in Arctic Populations. Toronto:
University of Toronto Press; 2008.
15. Young TK. Review of research on aboriginal populations in Canada: relevance to their
health needs. BMJ. 2003-08-21 00:00:00 2003;327(7412):419-422.
16. Statistics Canada. Focus on Geography Series, 2011 Census: Census subdivision of
Iqaluit, CY - Nunavut. 2012; http://www12.statcan.gc.ca/census-recensement/2011/as-
sa/fogs-spg/Facts-csd-eng.cfm?LANG=Eng&GK=CSD&GC=6204003. Accessed
February 15, 2013.
17. Møller H, Universitet K. A Problem of the Government?: Colonization and the Socio-
cultural Experience of Tuberculosis in Nunavut: Institut for Antropologi, Københavns
Universitet; 2005.
18. Smith HS, Bjerregaard P, Chan HM, et al. Research with Arctic peoples: Unique research
opportunities in heart, lung, blood and sleep disorders. International Journal of
Circumpolar Health. 2006;65(1).
19. Government of Nunavut. Tunngasugitti --Welcome to Nunavut! Nunavut, Our Land.
2012; http://www.gov.nu.ca/files/Nunavut,%20Our%20Land.pdf. Accessed October 12,
2012.
20. Kral MJ, Adams E, Akoomalik L, et al. Unikkaartuit: Meanings of well-being, sadness,
suicide, and change in two Inuit communities. Final Report to the National Health
Research and Development Programs, Health Canada, Project. 2003:6606-6231.
21. Healey G, Meadows L. Tradition and culture: an important determinant of Inuit women's
health. Journal of Aboriginal Health. 2008:25-33.
22. Inuit Tapiriit Kanatami. Social determinants of Inuit health in Canada: a discussion paper.
2007; http://ahrnets.ca/files/2011/02/ITK_Social_Determinants_paper_2007.pdf.
Accessed June 20, 2012.
23. Last J. A dictionary of epidemiology (4th ed.). Toronto: Oxford University Press; 2001.
24. Szklo M, Nieto FJ. Epidemiology : Beyond the Basics. Sudbury, Mass. Toronto: Jones
and Bartlett; 2004.
25. Søgaard AJ, Selmer R, Bjertness E, Thelle D. The Oslo Health Study: The impact of self-
selection in a large, population-based survey. International Journal for Equity in Health.
2004;3(1):3.
26. der Wiel A, Van Exel E, De Craen A, et al. A high response is not essential to prevent
selection bias: results from the Leiden 85-plus study. Journal of clinical epidemiology.
2002;55(11):1119-1125.
180
27. Sterne JAC, White IR, Carlin JB, et al. Multiple imputation for missing data in
epidemiological and clinical research: potential and pitfalls. BMJ: British Medical
Journal. 2009;338.
28. Rothman K, Greenland S, Lash T. Modern Epidemiology. 3 ed. Philadelphia, PA:
Lippincott Williams and Wilkins; 2008.
29. Delgado-Rodríguez M, Llorca J. Bias. Journal of Epidemiology and Community Health.
2004;58(8):635-641.
30. Luo Z-C, Senecal S, Simonet F, Guimond E, Penney C, Wilkins R. Birth outcomes in the
Inuit-inhabited areas of Canada. CMAJ. 2010;182(3):235-242.
31. Victoria C. Effect of breastfeeding on infant and child mortality due to infectious diseases
in less developed countries: a pooled analysis. Lancet 2000;355(9202):451-455.
32. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention
Trial (PROBIT): A Randomized Trial in the Republic of Belarus. JAMA.
2001;285(4):413-420.
33. Buehler JW, Kleinman JC, Hogue C, Strauss LT, Smith JC. Birth weight-specific infant
mortality, United States, 1960 and 1980. Public Health Reports. 1987;102(2):151.
34. McCormick MC. The contribution of low birth weight to infant mortality and childhood
morbidity. The New England Journal of Medicine. 1985.
35. McIntire DD, Bloom SL, Casey BM, Leveno KJ. Birth Weight in Relation to Morbidity
and Mortality among Newborn Infants. New England Journal of Medicine.
1999;340(16):1234-1238.
36. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of
breastfeeding practice. Nutr Rev. 2005;63(4):103-110.
37. Quandt SA. Material recall accuracy for dates of infant feeding transitions. Human
Organization. 1987;46(2):152-160.
38. Tienboon P, Rutishauser IHE, Wahlqvist ML. Maternal recall of infant feeding practices
after an interval of 14 to 15 years. Australian Journal of Nutrition and Dietetics.
1994;51:25-25.
39. Statistics Canada. Canadian Health Measures Survey (CHMS). 2012;
http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=5071&Item_Id=
129548&lang=en. Accessed April 8, 2013.
40. Hill A. The environment and disease: Association or causation. Proc Royal Soc Med.
1965;58:295-300.
41. Phillips C, Goodman K. The missed lessons of Sir Austin Bradford Hill. Epidemiologic
Perspectives & Innovations. 2004;1(1):3.
181
42. Goodman KJ, Phillips CV. Hill's criteria of causation. Encyclopedia of Statistics in
Behavioral Science. 2005.
43. Rothman K, Greenland S. Causation and causal inference in epidemiology. Am J Public
Health. 2005;95(Suppl 1):S144 - S150.
44. Kirmayer L, Malus M, Boothroyd L. Suicide attempts among Inuit youth: a community
survey of prevalence and risk factors. Acta Psychiatrica Scandinavica. 1996;94(1):8-17.
45. Norris M, Cooke M, Beavon D, Guimond E, Clatworthy S. Registered Indian mobility
and migration in Canada. Mobility and indigenous peoples in Australasia and Northern
America. 2004:136-160.
46. Aboriginal Peoples in Canada in 2006: Inuit, Metis and First Nations, census 2006. 2008.
http://www.statcan.gc.ca/daily-quotidien/080115/dq080115a-eng.htm. Accessed April
2010.
47. The Government of Nunavut, Nunavut Tunngavik Inc. Nunavut ten-year Inuit housing
action plan. 2004; http://www.tunngavik.com/documents/publications/2004-09-00-Ten-
Year-Inuit-Housing-Plan-English).pdf. Accessed December 12, 2012.
48. Lawn J, Harvey D. Change in nutrition and food security in two Inuit communities, 1992
to 1997: Minister of Public Works and Government Services Canada; 2001.
49. Olsen J, Basso O. RE: Residual Confounding. American Journal of Epidemiology.
1999;149(3):290.
50. Blyth R, Creedy DK, Dennis C-L, Moyle W, Pratt J, De Vries SM. Effect of Maternal
Confidence on Breastfeeding Duration: An Application of Breastfeeding Self-Efficacy
Theory. Birth. 2002;29:278-284.
51. Statistics Canada. Aboriginal Peoples in Canada in 2006: Inuit, Metis and First Nations
2006 census. 2008; http://www12.statcan.ca/census-recensement/2006/as-sa/97-
558/pdf/97-558-XIE2006001.pdf. Accessed November 1, 2012.
52. Axelson O, Fredriksson M, Ekberg K. Use of the prevalence ratio v the prevalence odds
ratio as a measure of risk in cross sectional studies. Occup Environ Med. 1994;51:574.
53. Behrens T, Taeger D, Wellmann J, Keil U. Different methods to calculate effect
estimates in cross-sectional studies. A comparison between prevalence odds ratio and
prevalence ratio. Meth Inform Med. 2004;43:505 - 509.
54. Lee J. Odds ratio or relative risk for cross-sectional data? Int J Epidemiol. 1994;23:201 -
203.
55. Lee J, Chia K. Use of the prevalence ratio v the prevalence odds ratio as a measure of risk
in cross sectional studies. Occup Environ Med. 1994;51:841.
182
56. Stromberg U. Prevalence odds ratio v prevalence ratio. Occup Environ Med. 1994;51:143
- 144.
57. Thompson M, Myers J, Kriebel D. Prevalence odds ratio or prevalence ratio in the
analysis of cross sectional data: what is to be done? Occup Environ Med. 1998;55:272 -
277.
58. Zocchetti C, Consonni D, Bertazzi PA. Relationship between prevalence rate ratios and
odds ratios in cross-sectional studies. International Journal of Epidemiology. February 1,
1997 1997;26(1):220-223.
59. McDonald SD, Pullenayegum E, Chapman B, et al. Prevalence and Predictors of
Exclusive Breastfeeding at Hospital Discharge. Obstetrics & Gynecology.
2012;119(6):1171.
60. Dennis C-L, Gagnon A, Van Hulst A, Dougherty G, Wahoush O. Prediction of Duration
of Breastfeeding among Migrant and Canadian-Born Women: Results from a Multi-
Center Study. The Journal of Pediatrics. 2012.
61. Banerji A, Bell A, Mills EL, et al. Lower respiratory tract infections in Inuit infants on
Baffin Island. CMAJ. 2001;164(13):1847-1850.
62. Orr PH. Respiratory tract infections in Inuit children: "Set thine house in order". CMAJ.
2007;177(2):167-168.
63. Guevremont A, Kohen D. The phyiscal and mental health of Inuit children of teenage
mothers. Health Reports. 2012;23(4):3-10.
64. Archibald L. Teenage pregnancies in Inuit communities: Issues and prespectives. 2004;
http://www.pauktuutit.ca/pdf/publications/pauktuutit/TeenPregnancy_e.pdf. Accessed
April, 2010.
65. Rose G. Sick individuals and sick populations. International Journal of Epidemiology.
2001;30(3):427-432.
Appendices
Appendix A: Data Collection Schedule
Table 5: Child Inuit Health Survey 2007-2008 Data Collection Schedule
Community Data Collection Dates
2007
Sanikiluaq August 8 -August 10
Pont Inlet September 26 - September 30
Igloolik October 2 - October 5
Clyde River October 8 -October 11
183
Pangnirtung October 13 -October 18
Iqaluit October 19 – October 28
Kimmirut October 29-November 1
Rankin Inlet November 2 – November 9
Coral Harbour November 9-November 13
Chesterfield Inlet November 14-November 16
Arviat November 17-November 21
Whale Cove November 21-November 23
2008
Cambridge Bay August 21-August 23
Kugluktuk August 25-August 28
Kugaaruk September 1-September 3
Baker Lake September 3-September 9
Table 6: Adult Inuit Health Survey 2007-2008 Data Collection Schedule
Community Data Collection Dates
2007
Sanikiluaq August 6-8
Arviat August 18-21
Whale Cove August 22
Rankin Inlet August 22-25
Chesterfield Inlet August 26
Coral Harbour August 27-28
Repulse Bay August 29
Hall Beach August 31-September 1
Igloolik September 1- September 3
Cape Dorset September 5- September 6
Kimmirut September 7
184
Iqaluit September 8 – September 12
Pangnirtung September 14- September 15
Qikiqtarjuag September 17
Clyde River September 18 – September 19
Grise Fiord September 21
Pond Inlet September 22 – September 24
Arctic Bay September 25
2008
Kugluktuk August 17-18
Cambridge Bay August 19-21
Gjoa Heaven August 23-25
Kugaaruk August 30
Resolute Bay September 3- 8
Baker Lake September 17-18
185
Appendix B: Child Questionnaire
186
187
188
189
190
191
192
193
194
Appendix C: Ethics Approval
195
Appendix D: Summary of Discussion Group at the National Aboriginal Health
Organization
196
197
Appendix E: Power Calculations
198
