NCRM is funded by the Economic and Social Research Council Fertility histories and later life health Emily Grundy and Sanna Read Website

NCRM is funded by the Economic and Social Research Council

Fertility histories and later life healthEmily Grundy and Sanna Read

Website http://pathways.lshtm.ac.uk

Email [email protected]

Twitter @pathwaysNCRM


Determinants of health in mid and later life

• Life course influences are recognized to be important, but most attention paid to socio-economic (and early life) factors

• Largely separate literature has shown differences by marital and household status and social support, more recent attention to partnership and parenting histories

• This literature has examined associations between the fertility histories of women (and less usually men) and mortality or health measured at one point in time

• Several, but not all, studies show worse health/higher mortality for nulliparous and high parity women (and men).

• Early parenthood is associated with poorer later health/mortality (women) and poorer later mental health (women and men)

• Late fertility associated better health/lower mortality in both women and men (but some studies the reverse)


Childrearing and health:Health promoting:• Incentives towards healthy

behaviours and risk avoidance • More social participation and

activity• Role enhancement• Social support - in childrearing

phases and in later life

Health challenging:• Physiological demands of

pregnancy, childbirth and lactation (although reduced risk breast & some other hormonally related cancers)

• Potential role conflict/role overload

• Stress (and depression)• Economic strain• Increased exposure infections• Disruption of careers/education –

especially for young parents


Associations between fertilityhistories and mortality in later life – Selection and reverse causation– Direct effects e.g. physiological consequences of pregnancy

and childbirth.– Indirect effects e.g. costs/benefits of child rearing mediated by

factors such as social support and health related behaviours.


Possible selection effects

• Poor health/health behaviours may restrict opportunities for marriage and reduce fertility (obesity, excessive alcohol and smoking all associated with fecundity of both women and men).

• Antecedent disadvantage associated both with early parenthood and with later poorer health

• Late fecundity and fertility may be marker of slower ageing/better health


Gender, age and SES variations/interactions

• Physiological effects of pregnancy/childbirth: mainly applicable women but some evidence of hormonal changes in new fathers

• Early childbearing may disrupt attainment adult career/educational roles

• Late childbearing may be stressful if perceived as ‘off time

• Ages at which mortality/morbidity observed may be important because of associations between cause and age of death

• Stress: negative effects likely to be greater for lone parents, those on low incomes and those with more stressful parenting histories (many children, children early in life).

• Social support: Women have closer links adult children, but some evidence that social support from children more important for health of older men – especially poorly educated

• Upward transfers greater to lower SES parents: support of children therefore potentially more important

• Life style/behavioural: effects noted for women and men – modified by residence with children.


Fertility history and later life health: previous research:

Data sources and outcomes investigated: • All cause mortality: Norwegian population registers; ONS

Longitudinal Study (E&W): USA Health and Retirement Survey linked to mortality

• Cause specific mortality: Norwegian population registers• Health, health trajectories, mental health: USA HRS;

UK British Household Panel Study; English Longitudinal Study of Ageing, 1946 birth cohort (NSHD).

• Quality of life, loneliness, social contacts, receipt of help from children: ELSA


Fertility history and mortality ages ~45-69 comparing England & Wales, Norway & USA (controlling for age, marital & socio-economic status &, in

USA, race/ethnicity).E&W deaths 19802000 at ages 50-69

Norway deaths 19802003 at ages 45-68

USA deaths 19942000 at ages 53-69

ALL Women/Men: OR OR OR

0 1.28 1.50 1.47

1 1.10 1.31 1.34

2 (ref) 1.00 1.00 1.00

3 1.01 0.95 1.21

4 1.11 0.95 1.41

5+ 1.25 0.94 1.66

PAROUS

Birth before 20 (F)/23 (M) 1.30 1.21 1.55

Birth after 39 0.94 0.86 0.74

Number of deaths 2,212 23,241 329

Analysis of ONS LS data ; Norwegian register data & US HRS, Grundy 2009. P<0.05; P<0.10


Associations between parity and mortality by cause group, Norwegian women aged 45-68

0 1 2 3 4+0

0.5

1

1.5

2

2.5

Lung caOther caCirc disResp disAlcoholAccs & ViOtherAll

Odd

s R

atio

Controlling for age, year, education, marital status, region, log population size of municipality, Grundy & Kravdal 2010.


BHPS analysis: Measures

• Fertility history: Number of natural children (0, 1, 2, 3, 4+); for parous: young age at first birth (<20/23); any birth at age >35/39; for parents with 2+ births: any birth interval < 18 months.

• Co-variates: Education; marital status; housing tenure; smoking; emotional support; co-residence with children (parents only)- all time varying except emotional support.

• Variables hypothesised to be associated with sample retention- interviewers’ reports of problems with interview; recent mover; foreign born.

Outcomes: • Self rated health: Excellent,

Good, Fair, Poor, Very poor. Ordinal variable, higher=worse.

• Health limitation: “Does your health in any way limit your activities compared to most people of your age?”


BHPS analysis: Results for a) parous men & women and b) parous with 2+ childrenHealth limitations Self-rated health

Men Women Men Womena) Parous respondents:

Number of children:1 +34+ +++ +++ +++ +++Birth before 23/20 +++ +++ +++ ++Birth after 39/35

b) Parity 2+; spacing effects

Number of children:3 +4+ +++ +++ +++Birth before 23/20 ++ +++ +++ +++Birth after 39/35

Birth interval < 18 months ++ +++ +++ +++


Rate-of-change in health over 11 years: Predicted probability of health limitation by fertility history characteristics, British women born 1923-49

(reference group = women with 2 children born when mother 20-34)

42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 740.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

Pr(HLIM=1 for reference group)

Pr(HLIM=1|early first birth)

Pr(HLIM=1|short birth interval)

Pro

bab

ilit

y

Source: Analysis of BHPS data in Read, Grundy & Wolf, Population Studies 2011


BHPS analysis: key findings• High parity (4+ children) associated with

health limitation and worse self-rated health among women and men

• Also a slightly higher risk of health limitation for childless women

• Early parenthood for parous) and short birth intervals (among those with 2+ children) associated with higher risk of health limitation, worse self rated health and faster accumulation of health limitation


Are children the key to a health and happy old age?

Yes• More children increases

chance of social contacts ; for parents especially having a daughter

• More children associated with more help from children

• Parents (of smallish families) seem to have lower mortality and better health than the childless

No• High parity associated

higher mortality and worse health

• ‘Intensive’ family formation patterns – early parenthood and short birth intervals- associated with worse health and faster decline in health

BUT the context is very important – known variations and interactions by Gender, country, education etc AND we need to consider selection.


Limitations of previous work• Outcome measures –

mortality and ADL limitation- may be too far ‘upstream’ – need indicators of sub clinical morbidity observable earlier in life course

• Failure to identify PATHWAYs through which fertility histories influence later life health

• Limited consideration of early life influences on both fertility histories and later health

Addressing these limitations• Measures of allostatic load in

mid and later life

• SEM and path analysis to

identify pathways

• Modelling including early life indicators

15


Progress to date:• Identifying earlier health outcomes: derivation

of a measure of allostatic load using data from the English Longitudinal Study of Ageing (ELSA)

• Identifying pathways from fertility histories to later life health (via allostatic load).

• In progress; taking account of early life influences

16

Allostatic load• a multisystem dysregulation state resulting from

accumulated physiological ‘wear and tear’ • Allostasis = a process whereby organism maintains

physiological stability by adapting itself to environmental demands

- > health is a state of responsiveness and optimal predictive fluctuation to adapt to the demands of the environment -> dynamic biological process interacting with context

Factors associated with allostatic load in previous studies

Socioeconomics: education, income, occupational status, downward mobility, homelessness

Family: attachment, violence, single parent, separation, care-giving, demands/criticism, spouse

Individual: type A/hostility, locus of control, a polymorphism of ACE gene

Neigbourhoods: crowding, noise, lack of housing, rural/urban

Allostatic load

Ethnicity: Non-whites (U.S.)

Spirituality: religious attendance, sense of meaning/purpose

Social networks: emotional support, social position

Work: control, demands, decisions, career instability, effort-reward imbalance

Sample• English Longitudinal Study of Ageing (ELSA) waves 1 -

4 (2002-2008)• men and women (n = 5279) aged 50+ in 2002• Measures:

– Biomarkers available in waves 2 and 4– Health: self reported health, limitation in health, ADL and

IADL limitation– Fertility history: number of children, birth before age 20

(women) or age 23 (men), birth after age 34 (women) and 39 (men), coresidence with child

– Background factors: age, marital status, qualification, tenure status, net wealth quintile (non-pension wealth indicating financial, physical and housing wealth net of debts)

Selected biomarkers to measure allostatic load in ELSA

Neuro-endocrine

Immune Cardiovascular Respiratory Metabolic Body fat

DHEAS* (dehydroepiandrostorone sulphate)

C-reactive protein

Systolic blood pressure

Peak expiratory flow

Total blood cholesterol/HDL cholesterol ratio

Waist-hip ratio

Fibrinogen Diastolic blood pressure

Triglycerides

IGF-1* (insulin-like growth hormone)

Glycated HgB

* only in wave 4

Allostatic load scores in ELSA• Group allostatic load index: number of biomakers

indicating high risk (25th percentile) calculated separately for men and women, range 0 - 9

Upper 25th percentile Lower 25th percentile

Systolic blood pressure Diastolic blood pressure

Fibrinogen Peak expiratory flow

Triglycerides

C-reactive protein

Glycated HgB

Waist-hip ratio

Total/HDL cholesterol ratio

Allostatic load change in ELSAIs change associated with any of the following factors? • Age• Qualification, tenure status, net wealth quintile• Being married• Perceived support and critique received from family

and friends• Number of children• Coresidence with child, early child birth, late child

birth (among parents only)

Allostatic load change in ELSAComparison between wave 2 (2004) and wave 4

(2008):• Low allostatic score (score 0-1) and high allostatic

score (2+)

High High

Low Low

2004 2008

Does allostatic load predict later disability?:Allostatic load measures in 2004 and ADL limitations in 2006 in men in ELSA

Systo

lic BP

Diasto

lic BP

Fibrin

ogen

C-reac

tive pro

tein

Triglys

erides

Glycate

d HgB

Wais

t-hip ra

tio

HDL choleste

role ra

tio

Peak exp

irato

ry flow

0

5

10

15

20

25

1234

ADL

prob

lem

%

Lowest 25%

Highest25%

Allostatic load change in ELSA

Men Women0

10

20

30

40

Low -> LowLow -> HighHigh -> LowHigh -> High

AD

L lim

itatio

n %

Allostatic load change between 2004 and 2008 in ELSA

Women Men0%

20%

40%

60%

80%

100%

High -> HighHigh -> LowLow -> HighLow -> Low


Women Men56.00

60.00

64.00

68.00

72.00

Low -> lowLow -> HighHigh -> LowHigh -> HighA

ge in

yea

rs


Men Women0

10

20

30

40

50


No

qual

ifica

tion

%


Men Women0

10

20

Low -> LowLow -> HighHigh -> LowHigh -> HighCh

ildle

ss %

Results from fully adjusted model


Men Women0

10

20


4+ c

hild

ren

%

Results from full adjusted model

Allostatic load change in ELSAIs change associated with any of the following factors? • Age -> Yes• Qualification, tenure status, net wealth quintile ->

Yes• Being married -> No• Perceived support and critique received from family

and friends -> Yes• n of children -> Yes• coresidence with child, early child birth, late child

birth (among parents only) -> No

Fertility history Allostatic load

Health

Education

Is the association between fertility history and health mediated by allostatic load?Does SEP influence this association?

The model to be tested

Wealth

4+ children vs. 2 children

Allostatic loadHealth

Education

All men

Wealth

Model adjusted for age. Unstandardized estimate and standard error shown.

-0.282 (0.070)

0.897 (0.070)

-0.499 (0.088)

-0.186 (0.023)

0.144 (0.015)

-0.097 (0.022)

Early birth Allostatic loadHealth

Education

Parous men

Wealth

Late birth

Model adjusted for age, n of children and coresidence with child. Unstandardized estimate and standard error shown.

1.238 (0.400)

-0.555 (0.130)

-0.239 (0.087)

-2.884 (0.765)

-0.209 (0.032)

0.238 (0.400)

-0.157 (0.016)

0.136 (0.022)

7.691 (2.310)

0 children vs. 2 children Allostatic load

Health

Education

All women

Wealth

4+ children vs. 2 children

Model adjusted for age. Unstandardized estimate and standard error shown.

0.779 (0.254)

1 child vs. 2 children

-3.637 (0.995)

-0.124 (0.060)

0.136 (0.026)

0.784 (0.253)

0.290 (0.079)

-0.461 (0.075)

-0.239 (0.087)

-0.095 (0.034)

Early birth Allostatic loadHealth

Education

Parous women

Wealth

Model adjusted for age, n of children, late birth and coresidence with child. Unstandardized estimate and standard error shown.

0.631 (0.060)

0.103 (0.015)-0.168

(0.023)0.174 (0.046)-0.225

(0.063)

-0.248 (0.079)

-0.064 (0.012)

-0.735 (0.112)

Is the association between fertility history and health mediated by allostatic load? - Yes, it is in men and to some extent also in women. In women there are also direct paths to health suggesting that there are other potential mediators.

Does SEP influence this association?- In men, and to some extent in women, SEP mediates the association between fertility history and later allostatic load and health.

Fertility history, allostatic load and health in ELSA

Work in progress

• Refinement of measures of allostatic load including medication use

• Further modelling of Pathways including life style variables

• Investigation of Pathways including early life influences

• Analyses taking account of missing data


- Several time-ordered exposuresintermediate outcomes

Early adulthood

SEP

Health in late

adulthood

fertility

Childhood SEPStress/support

Smoking,Diet/BMI Smoking

Diet/BMI

Documents

NCRM is funded by the Economic and Social Research Council Fertility histories and later life health Emily Grundy and Sanna Read Website