The All Our Babies pregnancy cohort: design, methods, and participant characteristics.

0.4 MB·English

by  McDonald, Sheila W

McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2
http://www.biomedcentral.com/1471-2393/13/S1/S2
RESEARCH Open Access
The All Our Babies pregnancy cohort: design,
methods, and participant characteristics
SheilaWMcDonald1*,AndrewWLyon2,KarenMBenzies3,DeborahAMcNeil4,StephenJLye5,SiobhanMDolan6,
CraigEPennell7,AlanDBocking8,SuzanneCTough1,9
Abstract
Background: The prospective cohort study design is ideal for examining diseases of public health importance, as
its inherent temporal nature renders it advantageous for studying early life influences on health outcomes and
research questions of aetiological significance. This paper will describe the development and characteristics of the
All Our Babies (AOB) study, a prospective pregnancy cohort in Calgary, Alberta, Canada designed to examine
determinants of maternal, infant, and child outcomes and identify barriers and facilitators in health care utilization.
Methods: Women were recruited from health care offices, communities, and through Calgary Laboratory Services
before 25 weeks gestation from May 2008 to December 2010. Participants completed two questionnaires during
pregnancy, a third at 4 months postpartum, and are currently being followed-up with questionnaires at 12, 24, and
36 months. Data was collected on pregnancy history, demographics, lifestyle, health care utilization, physical and
mental health, parenting, and child developmental outcomes and milestones. In addition, biological/serological and
genetic markers can be extracted from collected maternal and cord blood samples.
Results: A total of 4011 pregnant women were eligible for recruitment into the AOB study. Of this, 3388 women
completed at least one survey. The majority of participants were less than 35 years of age, Caucasian, Canadian
born, married or in a common-law relationship, well-educated, and reported household incomes above the Calgary
median. Women who discontinued after the first survey (n=123) were typically younger, non-Caucasian, foreign-
born, had lower education and household income levels, were less likely to be married or in a common-law
relationship, and had poor psychosocial health in early pregnancy. In general, AOB participants reflect the pregnant
and parenting population at local and provincial levels, and perinatal indicators from the study are comparable to
perinatal surveillance data.
Conclusions: The extensive and rich data collected in the AOB cohort provides the opportunity to answer
complex questions about the relationships between biology, early experiences, and developmental outcomes. This
cohort will contribute to the understanding of the biologic mechanisms and social/environmental pathways
underlying associations between early and later life outcomes, gene-environment interactions, and developmental
trajectories among children.
Background of development. Pregnancy and birth cohort studies are
Population-based cohortstudiesareimportantsourcesof particularlysalientforstudyingearlyoriginsofhealthand
data to investigate life course processes and to identify diseasethatbegininfetallifeandinfancy.Indeed,thecau-
aetiologicaldeterminantsofhealthanddiseaseoutcomes salunderpinningsofmanycommondiseasesinadulthood
inlaterlife[1].Astheyarenotspecifictoadiseasedpopu- (e.g., cardiovascular disease, obesity, psychopathology)
lation, they provide insight on what constitutes typical have roots in utero and the early postnatal phase [2-8].
trajectoriesandminorvariationswithinthenormalrange Early identification of threats to well-being is important
forthe developmentofpreventive andearlyintervention
strategies to optimize health and health care for indivi-
*Correspondence:[email protected] duals and communities. Cohort studies can provide
1DepartmentofPaediatrics,UniversityofCalgary,Calgary,AB,Canada
important aetiological, descriptive and surveillance
Fulllistofauthorinformationisavailableattheendofthearticle
©2013McDonaldetal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycited.
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page2of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
information about early risk factors for disease that can that was incorporated approximately one year after the
informresearch,policy,programs,andpractice. startofrecruitmentwasto examinebiologicalandenvir-
Advantagesofcohortstudiesforexaminingdevelopment onmentaldeterminantsofadversebirthoutcomes,specifi-
and links between early and later life outcomes are well callyspontaneouspretermbirth,forwhichapproximately
established[9-11].Theprospectivecohortstudydesignis halfoftheAOBsample(n=1862)providedbloodsamples
especially suited for examining associations that require at two time points during pregnancy, and cord blood,
considerationoftemporalityandarelesssubjecttorecall when retrievable, was collected at birth (n=1399). The
bias and reverse-causality bias compared to other epide- biological data collection and storage provides whole
miological study designs [1,9]. An important strength of blood, plasma, and serum samples from which lympho-
longitudinal studies is their potential for investigating cytes, cytokines, and proteins may be isolated and RNA
trajectoriesofdevelopmentandidentifyingsensitiveperi- and DNA will be extracted for micro-array analysis and
odsofriskorresilience[9,12].Furthermore,inlongitudinal futuremeasurement.Cordbloodsampleswillbeusedfor
research, there is a higher probability ofdiscovering true futurestudies.Biologicaldatacollectionmethodologyhas
exposureoutcomerelationships(i.e.,causalrelationships) beenpreviouslydescribed[19].Currently,theAOBstudy
when one exists [12]. An additional advantage relates to iscollectingobservationaldatabeyondtheperinatalperiod
efficiency gained through the breadth of data collection at 12 months, 24 months, and 36 months. Future data
and ability to assess a range of possible causes and out- collectionsatkeydevelopmentaltimepointsareplanned.
come variables, although in cases of rare but important OverallrecruitmentoftheAOBcohortaswellasobserva-
outcomes, collaboration with similar studies, or a more tionaldatacollectionproceduresduringtheperinatalper-
suitabledesign(i.e.,case-control)iswarranted[9]. iodandearlychildhoodaredescribedinturnbelow.
Theprospectivecohortstudyhasemergedasanimpor-
tantstudydesigntoinvestigategene-environmentinterac- Ethical approval
tions in diseases of major public health importance [1]. This study was approved by the Child Health Research
Although the case-control study remains a widely used Officeandthe ConjointHealth ResearchEthicsBoardof
methodforexamininggeneticandenvironmentaldetermi- theFacultiesofMedicine,Nursing,andKinesiology,Uni-
nants of complex disease, they are subject to significant versityofCalgary,andtheAffiliatedTeachingInstitutions
sources of bias that relate to subject selection and mea- (Ethics ID 20821 and 22821). Participants provided
surement of exposures and outcomes [1]. Prospective consent at the time of recruitment and were provided
cohortstudiesandtheirsubstudies(e.g.,nestedcase-con- copiesoftheconsentformfortheirrecords.
trol studies) can address some of these irremediable
sources of bias and offer complementary and innovative Recruitment
sources of information for studying early origins of later Aplannedapproximate3-yearrecruitmentstrategyforthe
diseaseandgene-environment interactions. Anumberof AOB study began in May, 2008 and was completed in
prospective pregnancy and birth cohorts studies exist in December, 2010. A total of 4011 pregnant women were
bothdevelopinganddevelopedcountries,andmanyhave assessed for eligibility from primary health care offices
contributed to understanding the role of the pre- and (n=573),communitypostersandwordofmouth(n=675),
postnatal environment on later life health, crucial for and through a city-wide single provider public health
aetiological and prevention research; examples include laboratory service (Calgary Laboratory Services;n=2763)
European cohorts such as The Avon Longitudinal Study (Figure 1). The AOB cohort is population-based and the
ofParentsandChildren(ALSPAC)[13],theGeneration-R largestproportionofrecruitedparticipants(69%)wascol-
study [14], the Danish National Birth Cohort study [15], lectedthroughCalgaryLaboratoryServices.Womenwere
the Millennium Cohort Study [16],and NorthAmerican eligible ifthey were less than 24 weeksand 6days gesta-
cohorts such as the National Children’s Study [17], and tionageatthetimeofrecruitment,atleast18yearsofage,
the Ottawa and Kingston Birth Cohort [18]. This paper receiving prenatal care in Calgary, and able to complete
will describe the development and characteristics of the thequestionnairesinEnglish.Eightwomenweredeemed
All Our Babies (AOB) study, a prospective pregnancy ineligibleattimeofrecruitmentduetoalanguagebarrier.
cohortstudyinCalgary,Alberta,Canada. The most common reason for discontinuation from the
study was active method of withdrawal (44%), including
Methods but not limited to: loss of interest, lack of time, reasons
Overview relatedtobloodcollectionsorlinkagetomedicalrecords
TheAOBstudy(n=3388)wasdesignedtoexaminemater- (althoughparticipantswerenotobligatedtoprovidecon-
nal and infant outcomes during the perinatal period and sentfortheseprocessestoparticipate),andlackofpartner
to identify current barriers and facilitators to accessing support.Passivewithdrawals(34%)includedgeographical
healthcareservicesinCalgary,Alberta.Afurtherobjective moves, lost to follow-up, or unknown reasons, while
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page3of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Figure1AOBparticipantrecruitment
ineligible(1%)includedthosewhoself-definedasEnglish Data collection (perinatal period)
as aSecondLanguage,as noted above. Babylosses(21%) Eligible participants (n=4003) were invited to complete
included both miscarriages and neonatal/infant loss threequestionnairesatseparatetimepointsacrosstheperi-
(Figure1). natalperiodand85%completedatleastonequestionnaire
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page4of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
(Figure 1). In the AOB study, both cross-sectional and accuracy. Data was exported and cleaned according to
longitudinal responses contribute valuable information, datacleaningguidelines,includingdatacoding,frequency
withresponseratesranging from76%to84%acrossthe editing,andcross-sectionalandlongitudinallogicalediting
threedatacollectiontimepoints(Figure1).Seventy-four [20].Informationacrossthethreetimepointswaslinked
percent completed all three questionnaires (Figure 1). accordingtoauniqueidentifierthatwasassignedtoeach
Thesecomprehensivequestionnairestookabout25min- participantatstudyentry,preservingparticipantconfiden-
uteseachto complete andincluded questionsabout the tiality. Information from medical charts was linked with
participant’spregnancyhistory,healthserviceutilization, questionnairedata bymeansofpersonalhealthnumbers.
demographics,lifestyle,mental,psychosocialandphysical Questionnaire and medical data were stored separately
health, life events, quality of life, work status, parenting from participant data, the latter which include personal
morale, and breastfeeding (see additional file 1 for a information such as name, address, and personal health
descriptionofvariablesassessedintheAOBstudybydata number. This separation acts to set up a central barrier
collectiontimepoint).Thequestionnairesweredeveloped between administrative data needed for conducting the
withinputfromhealthcareproviders,epidemiologistsand studyandanonymiseddataneededtoanswertheresearch
community program experts. Standardized tools were questions. Bothhardcopiesandelectroniccopiesofdata
includedaspartofthequestionnaireswhenavailable,and arestoredinasecureenvironmentandadheretosecurity
questionswerecreatedspecificallyforthestudywhenstan- andconfidentialityprotocolaspertheinstitutionalethics
dardized items or previously developed items were not boardandrecommendedguidelines[20].
suitable.Thequestionnaireswerepilottestedonapproxi-
mately10-12pregnantwomeninthecommunitytoensure Data collection (early childhood)
clarityandculturalsensitivity.Relevantresources,suchas Foreachfollow-updatacollectionwaveinearlychildhood
theMentalHealthHelpLine,wereprovidedintheques- (12 months, 24 months,and 36 months), the AOB study
tionnaireswheresensitivequestionswereasked.Inaddition team developed a 20 page questionnaire to measure
tothequestionnaires,allparticipantswereaskedtoprovide domains ofmaternalphysicaland mental health, parent-
consentfortheresearchteamtoaccesstheirprenataland ing,healthcareutilization,andfamilywell-being.Specific
birthrecorddata,includingpastpregnancyhistory,medical questionsandstandardizedtoolstoassesschilddevelop-
historyandcurrentconditions,pregnancycomplications, mentaloutcomesandmilestoneswerealsoadministered.
labour/birthoutcomes,andinfanthealthdata(Table1). In order to understand trajectories of development, the
The mailedquestionnairepackagesincludedaninforma- same construct (e.g., maternal depression) was assessed
tionletter,consentform,contactinformationform,ques- across time, using the same tool if appropriate. Further-
tionnaire, and postage pre-paid return envelope. The more,relevantdomainsoffunctioningateachtimepoint
participantswereaskedtocompletethefirstquestionnaire wereassessed.Forexample,questionsregardingwork-life
at recruitment (before 25 weeks gestation), the second balance/returntoworkandseparationanxietywereasked
between34-36weeksgestation,andthethirdat4months atthe12monthdatacollectiontimepoint,andquestions
postpartum. The questionnaires were returned to the regarding child behaviour and oral health were deemed
researchteambyregularpost.Trainedresearchassistants importantforthe36monthfollow-up.Outcomesofinter-
contactedtheparticipantsifdataweremissingorclarifica- est that will be measured in the AOB study across time
tionofresponseswasrequired.Participants whofailedto will include those relevant to population health such as
return their questionnaire within three weeks were obesity,injuries,recreation,chronic/inflammatorydisease,
contacted by telephone and/or e-mail and reminded to anddevelopmentaldisorders.Planneddomainsfora5and
completethequestionnaire;multipleattemptsweremade 8yearfollow-upalsoincluderecreation,screentime,sleep,
until the participant was contacted and provided the andoralhealth, amongothers. Detailedin-homeanthro-
opportunityforarepeatmail-outortocompletetheques- pometricanddevelopmentalassessments,aswellasDNA
tionnaireoverthetelephone.Aftercompletionandreturn collections are also planned for in subsequent follow-up
oftheirquestionnairesateachtimepoint,theparticipants datacollections.
wereprovidedwithatokenofappreciationsuchaslibrary
andgrocerystoregiftcards.Inordertokeepparticipants Results
engagedandupdated,congratulationcardsweresentafter Characteristics of the AOB participants
thebirthoftheirbaby,aswellasnewsletterssemi-annually Participantdemographics,pregnancycharacteristics,and
containingsuchinformationasprojectprogressandfind- labour andbirthoutcomesare presentedinTable 2and
ings(e.g.,mostpopular baby names),preliminary results Table 3.Psychosocialcharacteristicsduringthe prenatal
andresearchteammemberprofiles. andpostpartumperiodareshowninTable4.Themajor-
All raw data was scanned into Teleform (Version 10.1) ityofparticipantswerelessthan35yearsofageatdeliv-
and went through a verification process to improve ery (76%), Caucasian (79%), and Canadian born (78%).
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page5of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Table 1Information retrieved fromhospital andmedical recordsin the AOB study
Maternalhistory
Smoking,drugdependent
Pre-existingdiabetes,heartdisease,hypertension
Chronicrenaldisease,otherchronicdisorder,autoimmuneconditions
Maternalpastpregnancyhistory
Previoustermbirths,pastpretermbirth,previouspretermdeliveries
Numberofpreviousc-sections
Abortion,stillbirth(s),neonataldeath,majorcongenitalanomaly
Historyofintrauterinegrowthrestriction,SGA,LGA
Indicatedpregnancy
Smoking,drinking
Dateadmittedtolabouranddelivery
Assistedconception
Gravidity,parity
Maternalheight<=152cm,maternalweight(<=45Kg,>=91Kg),poorweightgain
Antepartumriskscore
Pregnancycomplications/problems
Infectioninpregnancy(GBS,HIV,HepB,other),fever,UTI
Poly/oligo,ROM<37wks,bleeding
Pregnancyinducedhypertension,gestationaldiabetes
Proteinuria,anemia
Cerclage,pre-eclampsia,eclampsia,abruption,prolongedprematureruptureofmembranes,placentaprevia,
Intrauterinegrowthrestriction,polyhydramnios,chorioamnionitis
Delivery
Site,typeofdeliveryprovider
Multiplepregnancy,maternalageatdelivery,gestation,pregnancy>=41weeks
Admittedforelectc-section,reasonforoperativedelivery
Indicationforinduction,cervicaldilatationatpresentation,typeofdelivery,deliverymode
(Fetal)presentationinlabour,trialoflabour
Methodofinduction(oxytocin,artificialruptureofmembranes,other)
Narcoticsinlabour,epiduralinlabour,Antenatalsteroids,useofintrapartumantibiotics
Secondstage(minutes),thirdstage(minutes)
Fetal
Neonatalgender,birthweight,date/time,disposition
5minuteApgarscore
Meconium,resuscitation
NICUadmission,congenitalanomaly
Maternal
Maternaldischargedate,maternaldischargedisposition,lengthofstay
Breastfeedingatdischarge
Almost all were either married or living in a common- being pregnant (87%). Approximately 3% conceived
law relationship (94%). Eighty-nine percent had com- throughassistedreproductivetechnologies,includingferti-
pleted at least some post secondary education and 69% lity-enhancing drugs,artificial insemination, andin-vitro
reported an annual household income greater than fertilization.Forty-onepercentgainedweightthataligned
$80,000. The index pregnancy was the first for approxi- with the recommended guidelines for gestational weight
mately1/3ofthesampleandalmosthalfwerenulliparous, gain based on pre-pregnancy body mass index [21], and
whichsuggeststhatasignificantproportionofthesefirst- one-quarter delivered by caesarean-section. The study
timemothershadexperiencedfetallossinpreviouspreg- yielded 36 sets of twin births. The overall preterm birth
nancies (27%). The majority had been trying to become ratewas7.9%.Amongsingletonbirths,thepreterm birth
pregnant (80%) and most reported feeling happy about andSmallforGestationalAge(SGA)rateswere7.3%and
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page6of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Table 2Demographic characteristics ofthe AOB study Table 3Pregnancy andlabour/delivery characteristics of
participants the AOB studyparticipants
Characteristic n(%) Characteristic n(%)
Maternalageatdelivery(n=2670) Pregnancyintention(n=3355)
19-24 153(5.7) Tryingtogetpregnant 2698(80.4)
25-29 716(26.8) Nottryingtogetpregnant 657(19.6)
30-34 1156(43.3) Feelingsaboutpregnancy(n=3348)
35-39 553(20.7) Happy 2913(87)
40+ 92(3.4) Unhappy/notsure 435(13)
Maritalstatus(n=3354) Gravidity(n=3338)
Married/CommonLaw 3165(94.4) Nulligravida 1192(35.7)
Other 189(5.6) Primi/Multigravida 2141(64.3)
Education(n=3356) Parity(n=3340)
Highschoolorless 370(11) Nulliparous 1637(49)
Someorcompleteduniversity/college 2458(73.2) Primi/Multiparous 1703(51)
Someorcompletedgradschool 528(15.7) Weightgainduringpregnancya(n=3002)
Ethnicity(n=3354) Inadequate 895(29.8)
Caucasian 2636(78.6) Adequate 1239(41.3)
Non-Caucasian 718(21.4) Excessive 868(28.9)
Income(n=3252) Methodofdelivery(n=3055)
<$40,000 299(9.2) Vaginal 2297(75.2)
$40,000-$79,000 717(22) Caesareansection 758(24.8)
≥$80,000 2236(68.8) Gestationalage(n=3032)
BorninCanada(n=3360) <34wks 51(1.7)
Yes 2623(78.1) 34-36wks 190(6.2)
No 737(21.9) 37+wks 2791(92.1)
SmallforGestationalAge(singletons;n=2836)
SGA 300(10.6)
10.6%, respectively. Almost all mothers initiated breast-
NotSGA 2536(89.4)
feeding,ifonlyforashorttime(98%);ofthese,61%were
LargeforGestationalAge(singletons;n=2836)
exclusivelybreastfeedingat4monthspostpartum.
LGA 251(8.9)
Psychosocial characteristics in the AOB cohort were
NotLGA 2585(91.1)
assessed using standardized tools (see additional file 1).
Breastfeedinginitiation(n=3057)
Prenatalpsychosocialhealthwasoperationalizedasscor-
Yes 2993(97.9)
ingintheexcessivesymptomrange(highorlowdepend-
No 64(2.1)
ing on the construct) at one or bothofthe prenatal data
Exclusivebreastfeedingat1wk(n=2969)b
collection time points. Women in the AOB cohort
Yes 1786(60.2)
reportedprevalences of prenatal depression,anxiety,and
No 1183(39.8)
stress of 12%, 28%, and 31%, respectively. At 4 months
Exclusivebreastfeedingat4-months(n=2976)b
postpartum, the rates were lower, at 5% for depression,
Yes 1809(60.8)
15% for anxiety, and 24% for stress. Perceived social
No 1167(39.2)
supportremainedhighatbothtimepoints(>80%)andthe
majority of women reported high optimism (80%) and aDifferenceinweightbetween34-36wksandpre-pregnancy
bAmongthosewhoinitiatedbreastfeeding
parentingmorale(83%)(Table4).
Characteristics of discontinued participants after the first questionnaire for reasons other than preg-
Inordertogainabetterunderstandingofthevariablesthat nancylossweremorelikelytobeyounger,non-Caucasian
maybeassociatedwithstudyattrition,whichwouldinform and foreign born, and to report lower education and
the extent of possible selection bias, we compared the householdincomelevels.Comparedtothosewhocontin-
demographic characteristics between those women who ued,discontinuerswerelesslikelytobemarriedorliving
dropped out of the study after the first questionnaire, in a common-law relationship, and reported poorer
excluding pregnancy losses, and those who continued to psychosocial health in early pregnancy (Table 5). There
thesecondand/or thirddata collection (Table 5).Results werenosignificantdifferencesbetweenthetwogroupsin
in Table 5 show that women who stopped participation terms ofgravidity, or feelings about pregnancy. We were
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page7of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Table 4Psychosocial characteristics ofthe AOBstudy data, the MES is a cross-sectional sample survey that
participants servesasthetargetpopulationofwomenandfamilieswho
Characteristic n(%) becomeparentsinCanada. AstheMESwasrestrictedto
Prenatal womenwithsingletonbirths,weinvokedthiscriterionfor
Depression,EPDSa≥13(n=3384) the AOB sample to facilitate comparisons. A greater
Yes 405(12) proportion of women in the AOB sample were older
No 2979(88) (≥35years)andreportedahigherhouseholdincomecom-
Anxiety,SAIb≥40(n=3363) pared to MES participants (Table 6). In terms of preg-
Yes 924(27.5) nancy characteristics, women in the AOB sample were
No 2439(72.5) more likely to have received a first ultrasound before 18
Stress,PSSc80thpercentile(n=3376) weeks gestational age and to have attended prenatal or
Yes 1041(30.8) childbirtheducationclasses.Percentagesfortheremaining
No 2335(69.2) demographicandpregnancycharacteristicswere,ingen-
Socialsupport,MOSdtotal≤69(n=3379) eral,similarbetweenAOBandMESparticipants.Thepre-
Inadequate 645(19.1) termbirthrate(singletons)forAOBwashigherthanthat
Adequate 2734(80.9) reported in the MES, and AOB participants reported a
Optimism,LOT-Re20thpercentile(n=2925) shorterlengthofstayforbothvaginalandcaesarean-sec-
Lowoptimism 582(19.9) tiondeliveries.ComparedtoMESparticipants,AOBparti-
Highoptimism 2343(80.1) cipantswerelesslikelytoreporttheirphysicalpostpartum
4monthspostpartum health as very good or excellent, yet were less likely to
Depression,EPDSa≥13(n=3041) score13oraboveonawidelyusedpostpartumdepression
Yes 152(5) scale(Table6).Onaverage,theremainingpregnancyand
No 2889(95) postpartum characteristics compared between the two
Anxiety,SAIb≥40(n=2942) samplesweresimilar.
Yes 440(15) AlthoughtheMESmaybealessthanidealcomparison
No 2502(85) forrepresentativeness, giventhat AOBand MES employ
Stress,PSSc80thpercentile(n=3004) different sampling strategies (i.e., stratified sampling in
Yes 714(23.8) MES,non-stratifiedsamplinginAOB),therangeoffactors
No 2290(76.2) assessedinthe MES allows for a wide range of compari-
Socialsupport,MOSdtotal≤69(n=3012) sons, beyond sociodemographic characteristicsand birth
Inadequate 412(13.7) indicators.Furthercomparisonswithotherdatasourcesat
Adequate 2600(86.3) the local and provincial levelsuch as administrative data
ParentingMoraleIndex,PMIf20thpercentile(n=2931) onperinatalhealthandCensuscommunityprofilesduring
Lowparentingmorale 491(16.8) or close to the study time period suggest that the AOB
Highparentingmorale 2440(83.2) participantsaregenerallyrepresentative ofthepregnancy
andparentingpopulationatthelocal(city)andprovincial
aEdinburghPostnatalDepressionScale(EPDS)[40]
bState-TraitAnxietyInventory(stateanxietyscale;SAI)[41] levels.Forexample,theaverageageofwomeninCalgary
cPerceivedStressScale(PSS)[42] and Alberta giving birth in 2010was30.8and 29.5 years
dMedicalOutcomesStudy(MOS)SocialSupportScale[38] [24]. In the AOB study, the average age at delivery was
eLifetimeOrientationTest-Revised(LOT-R)[43] 31.2 (SD=4.4). Approximately one-quarter of women in
fParentingMoraleIndex(PMI)[44]
Calgarywereforeign-bornandone-quarterwereavisible
minority according to the Canadian Census [25], with
unable to carry out an assessment of characteristics of
similar percentages seen in the AOB study (Table 2).
womenwhoagreedtoparticipatebutthenfailedtoreturn
Furthermore, 53% of women in the AOB study report a
aquestionnairebecauseethicallywewereunabletocollect
household income of over 100K, which aligns with the
any information about data about these women at
medianincomeofcouplefamiliesaccordingtorecentsta-
recruitment.
tistics from Statistics Canada for 2010 (approximately
97K)[26].
Comparison to the target population
Wecomparedthedemographicandpregnancycharacter-
Comparison to perinatal surveillance data
istics,aswellas thedeliveryandpostpartumexperiences
Recentdataonperinatalindicators[27]reportasingleton
ofthe AOBstudy participants to provincial and national
preterm birth rate of 7.9% and 8.8% in Canada and
statistics drawn from the Maternity Experiences Survey
Alberta,respectively.TheAOBpretermbirthrateforsin-
(MES)[22,23].Usingpost-census(2006CanadianCensus)
gletonsof7.3%fallsbelowboththeprovincialandnational
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page8of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Table 5Comparison between AOB discontinuersa andAOB continuersb
Characteristic Drop-outs(n=123) Continuers(n=3208) p-value
Maternalage
<30years 56(49.1) 1094(36.5) 0.006
30+years 1904(63.5)
Maritalstatus
Married/common-law 102(83.6) 3030(94.8) <0.001
Education
Highschoolorless 32(26.0) 336(10.5) <0.001
91(74.0)
Ethnicity
Caucasian 85(69.7) 671(21.0) 0.014
2525(79.0)
Income
<$40K 29(24.4) 266(8.6) <0.001
90(75.6)
Gravidity
Nulligravida 43(35.0) 1138(35.8) 0.85
80(65.0)
BorninCanada
Yes 87(70.7) 2508(78.4) 0.045
Depressioninearlypregnancy
Yes 27(22.1) 239(7.5) <0.001
Anxietyinearlypregnancy
Yes 37(32.2) 506(16.3) <0.001
Stressinearlypregnancy
Yes 50(41.3) 660(20.9) <0.001
Socialsupportinearlypregnancy
Inadequate 25(20.8) 412(13.0) 0.013
Feelingsaboutpregnancy
Happy 100(81.3) 2783(87.2) 0.057
rates; on the other hand, the AOB SGA rate of 10.6% is studyratestoperinatalsurveillancedata.Finally,mothers
greater than the corresponding provincial and national intheAOBcohorthadmuchhigherbreastfeedinginitia-
rates.Takentogether,thissuggestspossiblemisclassifica- tion rates than those reported for both Canada and
tionofbothbirthweightandgestationalagedataaccord- Alberta(98%vs.87%and91%,respectively).
ingtoself-report.Validationworkwithmedicalchartsfor
important labour and delivery outcomes has been com- Conclusion
pleted and is described elsewhere in this issue [28]. Significance
Although relatively high agreement was found between Emergingevidencerecognizestheimportanceofprenatal
the two data sources for select perinatal indicators [28], andearlylifeeventsonthelongtermdevelopmentofchil-
misclassification cannot be ruled out when comparing dren[29,30].TheAOBcohorthastheuniqueopportunity
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page9of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
Table 6Comparison ofAOBparticipants toMESa participants
Characteristic AOB% Alberta% Canada%
Demographiccharacteristics
≥35years 24.1 15.6 17.5
Postsecondarycompleted 76.3 69.5 72.1
>$40K 92.3 77.8 72.6
Primiparousb 48.9 46.0 44.7
Pre-pregnancyBMI(mean) 24.3 24.4 24.4
Pregnancycharacteristicsc
Numberofprenatalcarevisits(mean) 12.8 13.0 12.9
Gestationalageatfirstprenatalcarevisit(mean) 9.1 7.2 7.5
Initiatedprenatalcareinfirsttrimester(<14weeks) 93.1 94.9 94.9
Firstultrasound<18weeks 85.6 63.4 66.8
Attendedprenatalorchildbirtheducationclasses 41.2 33.4 32.7
Satisfiedwithtimingofpregnancy 52.6 50.9 49.5
Feelinghappyduponrealizationofpregnancy 87.0 90.8 93.0
Intendedtobreastfeed 96.2 93.8 90.0
Deliveryandpostpartumexperiences
Pretermbirthrate 7.3 6.3 6.2
Caesareansectiondelivery 24.5 27.3 26.3
Shortlengthofmaternalstayinhospital
Vaginal(<2days) 66.8 60.7 33.6
Caesareansection(<4days) 79.9 59.1 53.0
Initiatedbreastfeeding 97.8 94.6 90.3
Scoring≥13onEdinburghPostnatalDepressionScale 5.1 6.5 7.5
Ratedpostpartumhealthasverygoodorexcellent 53.9 73.6 72.5
PostpartumBMI(mean) 25.6 25.5 25.4
aMaternityExperiencesSurvey2006-2007;comparisonsinvolvesingletonsonly
baccordingtostatusatbirth
cassessedduringpostpartuminMES(retrospectiverecall);assessedduringpregnancyinAOB
d“happy”derivedfromcollapsingresponsesof“somewhathappy”and“veryhappy”
to inform complex questions about the relationship other established longitudinal cohorts (e.g., ALSPAC,
between biology, early experiences, and developmental Generation-R),isthatitsprospectivedatacollectionbegan
outcomes,andtocontributetoabetterunderstandingof in pregnancy. Although birth cohorts and cohorts that
the current circumstances of importance to families for begin in early childhood are important sources for life
stakeholders, policy and decision makers. An informed courseresearch,pregnancycohortsarewellpositionedto
picture of the early determinants of childhood develop- overcome methodological limitations such as recall bias
mentandfamilyoutcomesispotentiallyimportantfornot for exposures and confounding variables in pregnancy.
only prevention of disability and ill-health but also in Commontoallcohortstudies,sampleattritionovertime
developing an understanding of mechanisms underlying may be a source of selection bias for the AOB cohort
associations between early and later life outcomes (see below). Although the AOB cohort demonstrated a
(e.g., early socioeconomic status (SES) as a predictor of retentionrateof90%ofparticipantsbetweenthefirstand
childhoodintelligence anditsrole inexplainingtheasso- thirdquestionnaire,therewasan86%responserateforthe
ciationbetweenchildhoodintelligence and risk foradult 12monthdatacollection.Althoughthislatterrate isstill
disease; [31]). Future studies examining associations high, the decrease across time serves as a reminder that
betweenriskfactors andlater life outcomesmust ensure intensiveparticipantengagementisanimportantcompo-
adequatecontrolforpotentialconfounders.Suchearlylife nentforongoingcohortmaintenanceandfollow-up.
determinants of such risk factors, that are outcomes in Trackingtypicalandatypicaltrajectoriesofchilddevel-
themselves, require elucidation and adequate measure- opment as well as risk factors and effect modifiers is
ment. A key advantage of the AOB cohort, like some importantforthedevelopmentofpreventativestrategies.
McDonaldetal.BMCPregnancyandChildbirth2013,13(Suppl1):S2 Page10of12
http://www.biomedcentral.com/1471-2393/13/S1/S2
We have incorporated assessment tools to screen for possible,wewillutilizemedicalrecordsandadministrative
atypicaldevelopmentaspartofthe12,24,and36month sourcesofinformationand/orconductvalidationanalyses
follow-updatacollections.Forexample,theMacAurthur- betweendifferentdatasourcestomaintaininternalvalidity.
Bates Communicative Development Inventories [32] are Although vulnerable women may be at higher risk of
includedduringfollow-uptoidentifythosechildrenatrisk discontinuation,variabilityinethnicity,SESetc.ispresent,
forlanguagedelay.Toourknowledge,nopreviouspopula- and tends to reflect the urban Calgary parenting popula-
tion-basedcohortexistsofthissizethatincorporatesthree tion,whichallowsforexaminingassociationsforthesefac-
assessments of atypical child development coupled with tors, maintaining internal validity at the expense of
richmaternaldataandothergoldstandardtools.Follow- externalvalidity(generalizability).
up data collections will also allow for examining typical
andatypicaltrajectoriesofmaternalandfamilywell-being Summary
after the birth of a new baby. Longitudinal data analyses TheAOBcohort,ingeneral,isrepresentativeofthepreg-
willbeperformedtoexamineprecursorsandoutcomesof nantandparentingpopulationinaCanadianurbansetting,
trajectories.Wewillalsotrackoutcomesaspartofsurveil- Important research and policy questions are currently
lance undertaking for the AOB cohort. Some specific underexamination,resultswhichhavethepotentialtoadd
projectsthatwilluselongitudinaldatainclude:examining totheevidencebaseandinformdecisionmakersaboutthe
early risk factors for language delay; intergenerational health and well-being of pregnant women and their
transmissionofpsychosocialrisk;andlong-termoutcomes families.TheAOBcohortwillcontinuetobeasignificant
forlate-preterminfantsandtheirfamilies. Albertaresourcethatwillhaveimplicationsfarbeyondits
localroots.
Threats to validity
Amainsourceofpotentialbiasforlongitudinalstudiesis
Additional material
thatduetonon-response;pregnancyandbirthcohortsare
no exception. Non-response canaffect bothexternal and
Additionalfile1:VariablesassessedbyquestionnaireintheAOB
internal validity. In general, non-response cantake three studybydatacollectiontimepoint
forms:unitnon-response,orabsenceofthetargetsample
atstudyoutset;temporaryorwavenon-response;andper-
manent non-response, commonly referred to as attrition
Listofabbreviationsused
[33].Ananalysisofunitnon-responsegenerallycomprises AOB:AllOurBabies;ALSPAC:AvonLongitudinalStudyofParentsand
acomparisonofthestudypopulationtotheeligibleortar- Children;SES:SocioeconomicStatus;MES:MaternityExperiencesSurvey
getpopulation,andmayderivefrompreviouscollectionof Authors’contributions
minimaldatasetsonindividualswhoeitherrefusedtopar- SCTisresponsiblefortheoverallintegrity,progressandtimelycompletion
ticipate or were missed [34], or the use of administrative oftheAOBstudy.AWLisresponsibleforalllab-basedqueries.AWL,KMB,
DAM,SJL,SMD,CEP,ADB,andSCTparticipatedinthedesignofthestudy.
data sources with total population coverage of births or
SWMdraftedthemanuscript,performeddatalinkage,andconductedall
pregnancies [14,18]. Temporary and permanent non- statisticalanalyses.Allauthorshavereadandapprovedthefinalmanuscript.
response can be assessed if baseline information is
Competinginterests
collectedbeforedrop-out;ourcomparisonbetweenconti-
Theauthorsdeclarethattheyhavenocompetinginterests.
nuersanddiscontinuersisanexampleofanassessmentof
this type of non-response and threat to validity. In line Acknowledgements
WeareextremelygratefultotheparticipantsinvolvedintheAllOurBabies
withothercohortstudies,non-continuersintheAOBwere
cohort,andtotheAllOurBabiesstaffandresearchteam.Weareextremely
more likely to report poorer mental health and lower gratefultotheinvestigators,co-ordinators,researchassistants,graduateand
socioeconomicstatus[35-37].Wewillcontinuetoexamine undergraduatestudents,volunteers,clericalstaffandmanagers.Alberta
Innovates-HealthSolutions,formerlytheAlbertaHeritageFoundationfor
thecharacteristicsofdiscontinuersacrosstimeasselection
MedicalResearch,aspartofthePretermBirthandHealthyOutcomesTeam
bias due to attrition may become an increasing threat to InterdisciplinaryTeamGrant(#200700595),ThreeCheersfortheEarlyYears,
validity, inparticular when examining lifecourse associa- AlbertaHealthServicesandtheAlbertaChildren’sHospitalFoundationhave
providedsupportforthestudy.TheUniversityofCalgaryhasprovided
tions. In the AOB cohort, other potential sources of bias
traineesalarysupport.AlbertaInnovatesHealthSolutionsprovidedfunding
suchasinformationbias(e.g.,misclassificationbias,recall towardsthiscohortandsalarysupportforSuzanneTough.Additional
bias)andbiasduetoconfoundingarekepttoaminimum fundingfromtheAlbertaCentreforChild,Family,andCommunityResearch
(postdoctoralfellowship)forSheilaMcDonaldassistedwiththeanalysisof
due to the prospective nature of data collection, use of
datapresentedinthismanuscript.
standardized tools, and assessment across a range of
variables including different data sources. However, we Declarations
ThisarticlehasbeenpublishedaspartofBMCPregnancyandChildbirth
cannotdiscountthepossibilitythatreportingbiasdueself-
Volume13Supplement1,2013:PretermBirth:InterdisciplinaryResearch
reportwillremainapotentialthreattovalidity,and,where fromthePretermBirthandHealthyOutcomesTeam(PreHOT).Thefull