Ia

K,

a

b

c

d

e

f

g

h

i

a

ARRAA

KBAS2

1

ole

at

AT

0d

Vaccine 29 (2011) 8267– 8274

Contents lists available at SciVerse ScienceDirect

Vaccine

jou rn al h om epa ge: www.elsev ier .com/ locate /vacc ine

mpact of birth weight at term on rates of emergency room visits and hospitaldmissions following vaccination at 2 months of age�

umanan Wilsona,b,c,d,∗, Steven Hawkenb,c, Jeffrey C. Kwongd,e,f,g, Shelley L. Deekse,f, Doug G. Manuelb,c,h

Kirsten Holdt Henningsena, Natasha S. Crowcrofte,f,i

Department of Medicine, Ottawa Hospital Research Institute, University of Ottawa, CanadaICES@Uottawa, Ottawa Hospital Research Institute, University of Ottawa, CanadaDepartment of Clinical Epidemiology, Ottawa Hospital Research Institute, CanadaInstitute for Clinical Evaluative Sciences, Toronto, Ontario, CanadaPublic Health Ontario, Toronto, Ontario, CanadaDalla Lana School of Public Health, University of Toronto, CanadaDepartment of Family and Community Medicine, University of Toronto, CanadaDepartment of Family Medicine, University of Ottawa, CanadaLaboratory Medicine and Pathobiology, University of Toronto, Canada

r t i c l e i n f o

rticle history:eceived 8 July 2011eceived in revised form 11 August 2011ccepted 26 August 2011vailable online 9 September 2011

eywords:irth weightdverse events following immunizationelf-controlled case series design

month vaccination

a b s t r a c t

Background: Birth weight of children born at term may theoretically be associated with risk of adverseevents from immunization.Methods: We analyzed data on children born between April 1st 2002 and March 31st 2009 in the provinceof Ontario. Using the self-controlled case series design, we examined the risk of the combined endpoint ofemergency room visit and hospital admission in the immediate three days post vaccination at 2 monthsof age compared to a control period 9–18 days after vaccination. In term children, we conducted 4 com-parisons of relative incidence (RI) of events: (1) 4 lower birth weight quintiles compared to the largestquintile (2) SGA10 infants compared to non SGA10 infants, (3) low birth weight infants (<2500 g) com-pared to non low birth weight infants and (4) SGA10 infants vaccinated before 60 days compared to thosevaccinated after 60 days.Results: There was a significant trend towards increasing relative incidence of the combined endpointwith decreasing birth weight quintile (p = 0.016). There was an increased relative incidence of events in

SGA10 versus non SGA10 infants (RI 1.25 (95% CI 1.09–1.44)) and in SGA10 children vaccinated before60 days of age compared to after 60 days of age (RI 1.57 (95% CI 1.14–2.18)). No significant effect wasobserved in low birth weight children. The impact of birth weight was primarily mediated through anincrease in ER visits in the 24 h following vaccination.Conclusion: Lower birth weight appears to be correlated with an increased risk of emergency room visitswithin 24 h of vaccination. The absolute risk is small and there was no impact on admissions or death.

. Introduction

Premature and low birth weight children are at increased risk

f complications from vaccine preventable diseases, may be lessikely to receive vaccination on time and are therefore consid-red a priority to be vaccinated [1]. Several studies conclude

� The opinions, results and conclusions reported in this paper are those of theuthors and are independent from the funding sources. No endorsement by ICES orhe Ontario MOHLTC is intended or should be inferred.∗ Corresponding author at: Ottawa Hospital, Civic Campus, 1053 Carling Avenue,dministrative Services Building, Room 1009, Box 684, Ottawa, ON K1Y 4E9, Canada.el.: +1 613 798 5555x17921.

E-mail address: kwilson@ohri.ca (K. Wilson).

264-410X/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.oi:10.1016/j.vaccine.2011.08.107

© 2011 Elsevier Ltd. All rights reserved.

that the immune response to vaccinations in premature andlow birth weight children is adequate [2–4]. Consequently, theAmerican Advisory Committee on Immunization Practices and theCanadian National Advisory Committee on Immunization both rec-ommend that medically stable premature and low birth weightchildren should be vaccinated with all routinely recommendedchildhood vaccines at the same chronological age as full terminfants [5,6]. The only exception to this recommendation is thehepatitis B vaccine that is to be deferred if the infant weighsless than 2000 g and has a confirmed HBsAg negative mother

[5,7].

The tolerance to vaccines is generally considered to be similar inpremature and low birth weight children compared with childrenborn at term [8,9]. However, there is a higher risk of apnea and

8 cine 2

cmtt

vbiccrbrc[

g2b2ohct

2

2

tt(v3dt1gtBlrpoavwtatritfmh

2

1fam

268 K. Wilson et al. / Vac

ardiorespiratory problems following immunization of very pre-ature infants, particularly those of extremely low birth weight or

hose that are medically unstable [10]. Appropriate monitoring ofhese infants is often recommended.

To our knowledge all studies on low birth weight children andaccination to date have examined the effects of vaccination in lowirth weight and premature children combined, while no stud-

es have examined the effects of vaccination in low birth weighthildren born at term. This, however, is an important physiologi-al distinction to make. Although premature children are at highisk of having low birth weight, prematurity itself, independent ofirth weight, may impact on the likelihood of having an adverseeaction. Prematurity could theoretically reduce reactions to vac-ines because of inadequate physiological responses to vaccination9,11].

Using historical data we sought to examine patterns of emer-ency room (ER) visits and hospitalizations in the immediate post-months of age vaccination period in term children in differentirth weight strata. In the province of Ontario, Canada, between002 and 2009 it was recommended that infants at 2 monthsf age be vaccinated against diphtheria, pertussis, tetanus, polio,aemophilus influenzae type b (Hib) (as a single pentavalent vac-ine). In 2005, 7 valent pneumococcal conjugate vaccine was addedo the schedule at this age.

. Methods

.1. Design

The study was conducted using the Vaccine and Immuniza-ion Surveillance in Ontario (VISION) system. The objective ofhis study was to examine whether rates of emergency roomER) visits and hospital admissions immediately post-vaccinationaried by birth weight in term children, those born at or after7 weeks, vaccinated in Ontario at 2 months of age with stan-ard pediatric vaccines. We also sought to examine whetherhe risk of an event in term children born in the lowest0th percentile of weight for a given gestational age (small forestational age [SGA10]) was increased in those vaccinated prioro 2 months of age versus those vaccinated after 2 months of age.ecause of the possibility that children of low birth weight, particu-

arly SGA10 children, would be expected to have a higher baselineate of events than normal birth weight children, a simple com-arison of rates would likely produce a biased result in favourf an increased effect in smaller children. Therefore we chose tossess the relative incidence of an event in the immediate post-accination period compared to a baseline rate across differenteight strata. Furthermore, to conduct this analysis we utilized

he self-controlled case-series design, described by Farrington andssociates [12,13]. In this design, individuals serve as their own con-rol and the rate of events in an “at risk” period is compared to theate of events in a control period(s) defined as a period during whicht would be unlikely that the exposure produced the outcome. Forhis analysis the at risk period was defined as the immediate 3 daysollowing vaccination, as an acute reaction to the vaccine would

ost likely occur within 48 h resulting in either emergency visit,ospitalization or death within 3 days of vaccination [14].

.2. Ascertainment of population, exposure and outcome

The study included all children born in Ontario between April

st, 2002 and March 31st, 2009, who were present in the Instituteor Clinical Evaluative Sciences (ICES) Registered Persons Database,nd were eligible for the Ontario Health Insurance Plan (OHIP) at 6onths of age. At least 4 months of follow-up data was required for

9 (2011) 8267– 8274

all subjects, allowing sufficient post-vaccination follow-up to iden-tify admissions, ER visits and deaths. Pediatric vaccinations wereidentified using billing codes for general vaccination in the OHIPdatabase. To identify the 2 months of age vaccination, we targetedvaccination occurring on the exact due date (60 days) as well asany vaccinations occurring up to14 days before, or 40 days after thedue date. If infants had more than one vaccination in the databaseduring the two month target period the first vaccination was usedas the index vaccination. If another vaccination occurred withinthe observation period (0–18 days after the index vaccination) fora given infant, then this individual was excluded from analysis.Only subjects with both vaccinations and events in the observa-tion period contribute to the conditional self-controlled case seriesanalysis.

The Canadian Institute for Health Information’s (CIHI) DischargeAbstract Database (DAD) was used to ascertain hospital admis-sion and discharge date. The DAD captures all hospital admissions,including children in both tertiary and community hospitals. CIHI’sNational Ambulatory Care Registration System (NACRS) was usedto ascertain emergency department visits. All emergency depart-ments in Ontario participate in NACRS. The Registered PersonsDatabase was used to ascertain death in a secondary analysis. Alldatasets needed for this study were housed at ICES, where the datawere individually linked using the encrypted health card number asa unique identifier. Ethics approval was obtained from the OttawaHospital Research Ethics Board.

2.3. Analysis

In the self-controlled case series (SCCS) model, the date of vac-cination served as the index date for exposure for each patient. Foranalytical purposes, we divided each individual follow-up periodinto 3 distinct intervals after the vaccination date: an initial 3-day interval classified as exposed, followed by a 6 day washoutperiod, and then an unexposed period 9–18 days post vaccina-tion (Appendix 1). This 3-day risk window was defined based onthe physiological properties of the vaccines [14]. Our choice ofthe control period was based on the fact that it would be highlyunlikely that an acute reaction from the vaccine would result inincreases in adverse events this far removed from the date of vac-cination and is supported by a previous study where we identifiedthat the event rate remained stable during this time period [15](Appendix 2). Further, we did not want to choose a control periodso far removed from the vaccination that it could be influenced bysubsequent vaccination events (i.e. the 4 and 6 months of age stan-dard pediatric vaccinations). Where multiple events occurred for agiven individual, the first occurrence of the composite outcome ofER visit or hospitalization in each of the exposed and unexposedpost-vaccination periods was used.

The relative incidence rate of the composite end point of ERvisit or hospitalization during the exposed period compared withthe unexposed period was analyzed using a SCCS model, a fixedeffects conditional Poisson regression model that included termsfor exposure periods and a term for patient, allowing each individ-ual to serve as his or her own control, implicitly adjusting for allfixed covariates. The relative incidence is the ratio of the rate ofevents in the defined exposure period to the rate of events in thenon-exposed period, accounting for the differing durations of theexposed and unexposed periods.

We conducted 4 comparisons of relative incidence of eventsin term children (gestational age of 37 or more weeks at birth):(1) 4 lower birth weight quintiles compared to the highest quin-

tile, (2) SGA10 infants compared to non SGA10 infants, (3) lowbirth weight infants (<2500 g) compared to non low birth weightinfants and (4) SGA10 infants vaccinated before 60 days com-pared to those vaccinated after 60 days. We examined for an

K. Wilson et al. / Vaccine 29 (2011) 8267– 8274 8269

0

10000

20000

30000

40000

50000

60000

70000

80000

100989694929088868482807876747270686664626058565452504846

Birth

Cou

ntOveral lSGA10

ince d

isiceso

rsvf(rmsdv

3

A

TC

Days Since

Fig. 1. Frequency of vaccination events from 46 to 100 days s

nteraction between category and relative incidence in each analy-is. Interactions between fixed covariates and exposure period werencluded in the model by calculating terms for the product of theovariate and exposure periods. We conducted secondary analysesxamining the impact on ER visits, hospital admissions and deatheparately and conducted a post hoc analysis examining the impactf birth weight quintile on ER visits in the first 24 h.

Interaction terms were interpreted as “relative incidenceatios” (RIR), representing the ratio of relative incidences ineparate subgroups. Point estimates and 95% confidence inter-als were derived from the fitted Poisson regression modelsor both relative incidences (IRs) and relative incidence ratiosRIRs). Data manipulation and fitting of SSCS models was car-ied out with the assistance of SAS macros implementing theethods in Whitaker, Farrington et al. [13] provided on their web-

ite (http://statistics.open.ac.uk/sccs/sas.htm) as well as methodsescribed by Xu et al. [16]. All analyses were conducted using SASersion 9.2 (SAS Institute, Cary, NC).

. Results

Follow-up data was available for 980 987 children born betweenpril 1st 2002 and March 31st 2009. Of these, 859 839 were term

able 1ombined endpoints in different birth weight categories in term babies.

# of children Median BW Control events

Overall 713 372 3440 9868

Birth weight quintile1 (large) 140 094 4070 2010

2 143 220 3693 1924

3 143 071 3440 1908

4 142 134 3201 1947

5 (small) 144 853 2850 2079

SGA10c status in term babiesNO 643 815 3500 8858

YES 63 128 2735 944

Low birth weight (<2500 g) status in term babiesNO 699 176 3460 9617

YES 14 196 2360 250

Timing of vaccination in SGA10 children≥60 days 53 351 2715 803

<60 days 9777 2750 141

a Relative incidence in risk versus control period from SCCS model.b RIR is the ratio of relative incidences between subgroup category and the reference gc SGA10 = Small for gestational age (bottom 10% for birth weight stratified by gestation

ate of birth in all term children and in term SGA10 children.

births (gestational age of 37 weeks or more) with birth weight avail-able. Finally, 713 372 of these children had an index vaccination thatoccurred at 2 months (60 days) plus 40 or minus 14 days (Appendix3) (Fig. 1).

3.1. Association of birth weight with event

Birth weight was available for 713 372 term babies. 2643 chil-dren experienced one of the combined end-points during theimmediate 3 days post vaccination, compared to 9868 in the 9 daycontrol period for our primary analysis (Table 1). The overall rela-tive incidence of an event was 0.80 (95% CI 0.77, 0.84). Each quintilein our birth weight analysis consisted of approximately 143 000children. The median weights per quintile were 4070, 3693, 3440,3201 and 2850 g. Compared to the highest quintile, the relative inci-dence ratio was 1.07 (95% CI 0.93, 1.23), 1.07 (95% CI 0.93, 1.23), 1.21(95% CI 1.06, 1.39), and 1.21 (95% CI 1.06, 1.38) in declining orderof weight (Table 1). The p-value for the test for interaction was

p = 0.016. We estimate that there were 70 excess events/100 000vaccinated in the lowest quintile of weight compared to the highestquintile of weight. This translates into 1 extra event within 3 daysof vaccination for every 1426 infants in the lowest weight quintile

Risk events RIa (95% CI) Interaction RIRb (95% CI)

2643 0.80 (0.77, 0.84) –

483 0.72 (0.65, 0.80) 1 (ref)496 0.77 (0.70, 0.85) 1.07 (0.93, 1.23)491 0.77 (0.70, 0.85) 1.07 (0.93, 1.23)568 0.88 (0.80, 0.96) 1.21 (1.06, 1.39)605 0.87 (0.80, 0.96) 1.21 (1.06, 1.38) p (interaction) = 0.016

2313 0.78 (0.75, 0.82) 1 (ref)309 0.98 (0.86, 1.12) 1.25 (1.09, 1.44)p (interaction) = 0.0012

2570 0.80 (0.77, 0.84) 1 (ref)73 0.88 (0.68, 1.14) 1.09 (0.84, 1.42) p (interaction) = 0.51

242 0.90 (0.78, 1.04) 1 (ref)67 1.43 (1.07, 1.91) 1.57 (1.14, 2.18)p (interaction) = 0.0059

roup from the SCCS model with interaction terms for subgroups.al age).

8270 K. Wilson et al. / Vaccine 29 (2011) 8267– 8274

Table 2Emergency room visits and hospital admissions in different birth weight categories in term babies.

Quintile Emergency roomvisits RI (95% CI)

Emergency room visits RIR (95% CI) Hospitaladmissions RI(95% CI)

Hospital admissionsinteraction RIR (95% CI)

Overall 0.81 (0.78, 0.85) – 0.51 (0.45, 0.58) –1 (large) 0.72 (0.65, 0.80) 1 (ref) 0.51 (0.38, 0.67) 1 (ref)2 0.78 (0.70, 0.86) 1.07 (0.93, 1.24) 0.55 (0.42, 0.72) 1.08 (0.73, 1.60)3 0.78 (0.71, 0.86) 1.08 (0.94, 1.24) 0.45 (0.33, 0.60) 0.88 (0.59, 1.32)4 0.89 (0.81, 0.98) 1.23 (1.07, 1.41) 0.54 (0.41, 0.71) 1.06 (0.71, 1.51)5 (small) 0.88 (0.80, 0.96) 1.21 (1.06, 1.39) p (interaction) = 0.015 0.52 (0.40, 0.68) 1.03 (0.71, 1.51)

p (interaction) = 0.87SGA10 status in term babies

No 0.79 (0.75, 0.83) 1 (ref) 0.49 (0.43, 0.56) 1 (ref)Yes 1.00 (0.88, 1.14) 1.27 (1.10, 1.46)p (interaction) = 0.0008 0.63 (0.44, 0.89) 1.27 (0.88, 1.85)

p (interaction) = 0.20Low birth weight (<2500 g) status in term babies

No 0.81 (0.77, 0.84) 1 (ref) 0.51 (0.45, 0.58) 1 (ref)Yes 0.91 (0.70, 1.18) 1.12 (0.86, 1.47) p (interaction) = 0.39 0.55 (0.27, 1.12) 1.08 (0.52, 2.22)

p (interaction) = 0.84Timing of vaccination in SGA10 children

eracti

Tii(

STnwTei1lwcC

3

yhdrv21

TI

≥60 days 0.92 (0.79, 1.06) 1 (ref)

<60 days 1.46 (1.09, 1.95) 1.59 (1.14, 2.20) p (int

able 3. When examining end points separately, the increase in riskn lower birth weight children was primarily driven by increasesn emergency visits. There was no increase in hospital admissionsTable 2).

There were a total of 63 128 infants who were classified asGA10 in our analysis compared to 643 815 non SGA10 infants.he relative incidence ratio in the SGA10 children compared toon SGA10 children was 1.25 (95% CI 1.09, 1.44, p = 0.0012). Thereere 92 excess events for every 100 000 SGA10 children vaccinated.

his translates into 1 extra event within 3 days of vaccination forvery 1089 children vaccinated. The increase in risk was due to anncrease in emergency room visits (Table 2). There were a total of4 196 low birth weight children, defined as infants that weighed

ess than 2500 g in our analysis, compared to 699 176 non low birtheight children. The relative incidence ratio in low birth weight

hildren compared to non low birth weight children was 1.09 (95%I 0.84, 1.42, p = 0.51) which did not achieve statistical significance.

.2. Timing of vaccination

Of the 63 128 infants who were classified as SGA10 in our anal-sis, 9777 had vaccinations between 46 and 60 days and 53 351ad vaccinations between 60 and 100 days. Fig. 1 presents theistribution of vaccines from time of birth in these children. The

elative incidence ratio in the early vaccinators versus the lateaccinators was 1.57 (1.14, 2.18) (p = 0.0059). This translates into60 excess events in the 3 days following vaccination for every00 000 children vaccinated or 1 excess event for every 385 children

able 3ncreased risk of combined endpoints from vaccination.

Quintile Additional children experiencing at least oneevent (per 100 000 vaccinated)

1 Ref

2 24

3 244 71

5 70

SGA10 status in term babiesNo Ref

Yes 92

Timing of vaccination in SGA10 children≥60 days Ref

<60 days 260

0.61 (0.42, 0.89) 1 (ref)on) = 0.006 0.79 (0.29, 2.11) 1.29 (0.45, 3.71)

p (interaction) = 0.63

vaccinated earlier than 60 days. The primary reason for extra eventswas presentation to the emergency room.

3.3. Secondary analyses

We further explored the timing of the events post-vaccinationgraphically and identified that the effect of birth weight was medi-ated primarily through emergency room visits in the first 24 hfollowing vaccination. Table 4 demonstrates the relative incidenceof ER visits in the first 24 h following vaccination in different riskcategories which is also graphically apparent in Fig. 2. Children inthe lowest quintile of weight have an approximate 50% increase inrelative incidence of emergency room visits within 24 h of vaccina-tion compared to the highest quintile. This increased risk translatesinto 58 excess ER visits one day post vaccination per 100 000 chil-dren vaccinated or 1 excess visit for every 1730 children vaccinated.We observed no association between admissions to hospital andbirth weight quintile and we observed ≤5 deaths in our overallanalysis with no association with birth weight quintile.

4. Discussion

Our analysis suggests that there may be an increased risk ofpost vaccination events, which are primarily emergency room vis-

its, in full term lower birth weight children compared to higherbirth weight children. This risk amounts to 70 extra events in the3 days post vaccination for every 100 000 children vaccinated inthe lowest quintile of weight compared to the highest quintile of

Number vaccinated Number vaccinated per excess event

140 094 Ref143 220 4131143 071 4243142 134 1406144 853 1426

643 815 Ref63 128 1089

53 351 Ref9777 385

K. Wilson et al. / Vaccine 29 (2011) 8267– 8274 8271

Table 4Emergency room visits within 24 h according to birth weight quintiles.

Quintile Control events Risk events RI (95% CI) RIR (95% CI)

1 (large) 1930 153 0.71 (0.61, 0.84) 1 (ref)2 1847 182 0.89 (0.76, 1.03) 1.24 (0.99, 1.55)3 1834 182 0.89 (0.77, 1.04) 1.25 (1.00, 1.57)4 1872 230 1.11 (0.96, 1.27) 1.55 (1.25, 1.92)

p

wtsipvsWwc

5 (small) 1999 242

(interaction) = 0.0002.

eight. Further evidence for the potential impact of weight at theime of vaccination on events is provided by our analysis demon-trating increase in events in term SGA10 infants and in SGA10nfants vaccinated earlier. The impact of weight on combined end-oints appears to be mediated by an increase in emergency roomisits in the first 24 h after vaccination. Importantly there was no

tatistically significant impact on admissions to hospital or death.

e did not see a statistically significant effect in our lowest birtheight cohort (low birth weight children <2500 g) although this

ould be a consequence of these children having high baseline rates

Fig. 2. ER visits in first 24 h post-vaccination by birth w

1.09 (0.95, 1.25) 1.53 (1.24, 1.89)

of events which reduced the relative incidence of the event in the3 days post vaccination period.

We believe that this is the first analysis of health services uti-lization patterns following vaccination in term low birth weightchildren. Several previous studies have demonstrated the safety ofvaccines in low birth weight children but all of these studies have

included pre-term children and birth weight is highly correlatedwith degree of prematurity [17–19]. It is conceivable that prema-turity may protect against acute reactions. This is partly supportedby studies on immunogenicity after an acellular pertussis vaccine in

eight quintiles and SGA10 status in term babies.

8 cine 29 (2011) 8267– 8274

pitg

afabTsaacictvr

ticiaw

Fig. A1. Illustration of the self-controlled case series design. The observation periodfor each patient begins with pediatric vaccination date (leftmost upward arrow)

Fiu

272 K. Wilson et al. / Vac

remature infants showing that the immune response in pretermnfants is significantly lower compared with term infants but thathe incidence of adverse reactions is low and comparable in bothroups [11].

We expect that birth weight is strongly correlated with weightt the time of 2 months of age vaccination. A potential explanationor our observations is that the same dose of vaccine is presented toll children and those with lower weights may be expected to haveeen comparatively dosed higher than those with higher weights.his hypothesis has previously been tested in studies on animalshowing that the risk of a vaccine reaction decreased significantlys the body weight of the animal increased [20]. Our findings may be

consequence of increased parental anxiety in lower birth weighthildren post vaccination. However, we observed an increased riskn normal birth weight children compared to high birth weighthildren. We believe the examination of the physiological reactiono vaccines in term children with different weights at the time ofaccination and the possibility of a weight based dose–responseelationship merits further analysis.

Our study has several strengths and limitations. Strengths ofhis study include the examination of the impact of weight in termnfants, the large sample size and the use of the self-controlled

ase series design. Another strength of our study is the compar-son of relative incidence across different categories of risk. Suchn approach controls for the variations in baseline risk whichould have created a bias towards an effect in lower birth weight

ig. A2. Number of combined endpoints versus days before/after 2 month vaccination (frzation and death. Days = number of days before or after vaccination, day 0 being the day ose of the control period for the analyses.

and continues for a total of 18 days. In the primary analyses, the first three dayspost vaccination is the risk interval, and days 9–18 comprise the control interval. Theintervening days represent the wash-out period.

categories. It also allowed us to overcome the bias created by thehealthy vaccine effect [21–23]. We had previously demonstratedthat the time of vaccination is preceded by low rates of emer-gency room visits and admissions, suggesting that vaccinations

were deferred if the child was unwell [15]. This effect resulted ina reduced risk in the immediate post-vaccination period and couldhave masked any vaccine related adverse event. By comparing therelative incidences we have adjusted for this bias. However, the

om 15). Count = number of combined endpoints of emergency room visit, hospital-f vaccination. Combined end point events remain stable after Day 5 permitting the

K. Wilson et al. / Vaccine 29

Fa

cunaitccci

laia

A

iKM

[

[

[

[

[

[

ig. A3. Flow chart for identifying cohort to be included in self-controlled case seriesnalysis.

hoice of this design prevented us from conducting an analysissing birth weight as a continuous variable. A strength and weak-ess of this study include the use of aggregate health services datas an outcome. Adverse events that are not severe enough to resultn an emergency visit would be missed. Another potential limita-ion is that the recommended vaccines given at 2 months of agehanged over the time period examined, with pneumococcal vac-ine introduced during the study period. This, however, is a fixedovariate and should not influence our analysis unless there wasnteraction with birth weight quintile.

While we observed an increase in relative incidence in events inower birth weight children, it is reassuring that we did not identifyn increase in hospital admissions or death and that the absolutencreased risk of events is low. We therefore would not recommendny change to vaccination practice based on our findings.

cknowledgements

Dr. Wilson holds the Canada Research Chair in Public Health Pol-cy. Dr. Manuel holds the CIHR Chair in Applied Public Health. Dr.wong is supported by a Career Scientist award from the Ontarioinistry of Health and Long-Term Care and a Clinician Scientist

[

(2011) 8267– 8274 8273

award from the Department of Family and Community Medicine,University of Toronto. This study received ethics approval fromthe Ottawa Hospital Research Institute’s Research Ethics Board. Wewould like to thank Paddy Farrington, Heather Whitaker and BartSpiessens for methodological advice and provision of SAS code toimplement the SCCS methodology for the analysis of our data.

Funding: This study was supported by the Canadian Foundationfor Innovation and by the Institute for Clinical Evaluative Sciences(ICES), which is funded by an annual grant from the Ontario Min-istry of Health and Long-Term Care (MOHLTC). The opinions, resultsand conclusions reported in this paper are those of the authors andare independent from the funding sources. No endorsement by ICESor the Ontario MOHLTC is intended or should be inferred.

Appendix 1.

See Fig. A1.

Appendix 2.

See Fig. A2.

Appendix 3.

See Fig. A3.

References

[1] Davis RL, Rubanowice D, Shinefield HR, Lewis N, Gu D, Black SB, DeStefano F,et al. Immunization levels among premature and low-birth-weight infants andrisk factors for delayed up-to-date immunization status. Centers for DiseaseControl and Prevention Vaccine Safety Datalink Group. JAMA: The Journal ofthe American Medical Association 1999;282(6):547–53.

[2] Conway S, James J, Balfour A, Smithells R. Immunisation of the preterm baby.The Journal of Infection 1993;27(2):143–50.

[3] Pullan CR, Hull D. Routine immunisation of preterm infants. Archives of Diseasein Childhood 1989;64(10):1438–41.

[4] D’Angio CT, Maniscalco WM, Pichichero ME. Immunologic response ofextremely premature infants to tetanus, haemophilus influenzae, and polioimmunizations. Pediatrics 1995;96(1 Pt 1):18–22.

[5] Saari TN. Immunization of preterm and low birth weight infants. Amer-ican Academy of Pediatrics Committee on Infectious Diseases. Pediatrics2003;112(1 Pt 1):193–8.

[6] Public Health Agency of Canada. Canadian immunization guide. 7th ed. Part3. Recommended immunization-immunization of infants born prematurely;2006.

[7] Lau Y-L, Tam AYC, Ng KW, Tsoi NS, Lam B, Lam P, et al. Response of preterminfants to hepatitis B vaccine. The Journal of Pediatrics 1992;121(6):962–5.

[8] Slack MH, Schapira D, Thwaites RJ, Andrews N, Schapira D. Acellular pertussisvaccine given by accelerated schedule: response of preterm infants. Archivesof Disease in Childhood Fetal and Neonatal Edition 2004;89(1):F57–60.

[9] Schloesser RL, Fischer D, Otto W, Rettwitz-Volk W, Herden P, Zielen S. Safetyand immunogenicity of an acellular pertussis vaccine in premature infants.Pediatrics 1999;103(5):e60.

10] D’Angio CT. Active immunization of premature and low birth-weight infants:a review of immunogenicity, efficacy, and tolerability. Paediatric drugs2007;9(1):17–32.

11] Khalak R, Pichichero ME, D’Angio CT. Three-year follow-up of vaccine responsein extremely preterm infants. Pediatrics 1998;101(4 Pt 1):597–603.

12] Farrington P, Pugh S, Colville A, Flower A, Nash J, Morgan-Capner P,et al. A new method for active surveillance of adverse events fromdiphtheria/tetanus/pertussis and measles/mumps/rubella vaccines. Lancet1995;345(8949):567–9.

13] Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in bio-statistics: the self-controlled case series method. Statistics in Medicine2006;25(10):1768–97.

14] Siegrist CA. Mechanisms underlying adverse reactions to vaccines. Journal ofComparative Pathology 2007;137(Suppl. 1):S46–50.

15] Wilson K, Hawken S, Potter BK, Chakraborty P, Kwong J, Crowcroft N, et al. Pat-terns of emergency room visits, admissions and death following recommendedpediatric vaccinations – a population based study of 969,519 vaccination

events. Vaccine 2011;29(21):3746–52.

16] Xu S, Garguillo P, Mullooly J, McClure D, Hambridge SJ, Glanz J. Fitting para-metric and semi-parametric conditional Poission regression models with Cox’spartial likelihood in self-controlled case series and matched cohort studies.Journal of Data Science 2010;8:349–60.

8 cine 2

[

[

[

[

[

[

[23] Davis RL, Marcuse E, Black S, Shinefield H, Givens B, Schwalbe J, et al.

274 K. Wilson et al. / Vac

17] Linder N, Handsher R, German B, Sirota L, Bachman M, Zinger S, et al. Con-trolled trial of immune response of preterm infants to recombinant hepatitisB and inactivated poliovirus vaccines administered simultaneously shortlyafter birth. Archives of Disease in Childhood Fetal and Neonatal Edition2000;83(1):F24–7.

18] Slack MH, Schapira D, Thwaites RJ, Burrage M, Southern J, Andrews N,et al. Immune response of premature infants to meningococcal serogroupC and combined diphtheria-tetanus toxoids-acellular pertussis-Haemophilusinfluenzae type b conjugate vaccines. The Journal of Infectious Diseases

2001;184(12):1617–20.

19] Esposito S, Pugni L, Bosis S, Proto A, Cesati L, Bianchi C, et al. Immunogenicity,safety and tolerability of heptavalent pneumococcal conjugate vaccine admin-istered at 3, 5 and 11 months post-natally to pre- and full-term infants. Vaccine2005;23(14):1703–8.

9 (2011) 8267– 8274

20] Moore GE, Guptill LF, Ward MP, Glickman NW, Lewis HB, et al. Adverse eventsdiagnosed within three days of vaccine administration in dogs. Journal of theAmerican Veterinary Medical Association 2005;227(7):1102–8.

21] Virtanen M, Peltola H, Paunio M, Heinonen OP. Day-to-day reactogenicity andthe healthy vaccine effect of measles-mumps-rubella vaccination. Pediatrics2000;106(5):E62.

22] Fine PE, Chen RT. Confounding in studies of adverse reactions to vaccines.American Journal of Epidemiology 1992;136(2):121–35.

MMR2 immunization at 4 to 5 years and 10 to 12 years of age: a com-parison of adverse clinical events after immunization in the Vaccine SafetyDatalink project. The Vaccine Safety Datalink Team. Pediatrics 1997;100(5):767–71.