The Baby Hearts Study – A case-control protocol with data linkage to evaluate risk and protective factors for congenital heart disease

Introduction
The Baby Hearts study aimed to investigate risk and protective factors for congenital heart disease (CHD), and to investigate the health behaviours of a representative sample of pregnant women in Northern Ireland.

Objectives
We describe and evaluate the population-based case-control design enhanced with data linkage to administrative health data.

Methods
Cases (mothers of babies with CHD, n=286) were recruited following diagnosis prenatally or post- natally. Controls (mothers of babies without CHD, n=966) were recruited at 18-22 weeks gestation, from all women attending each maternity unit during a designated month. Hybrid data collection methods were used, including a self-administered iPad/postal questionnaire, and linkage to maternity and prescription records.

Results
Refusal rates were low (8%). iPad questionnaire completion at clinic or home visit had high acceptability whereas postal questionnaires were poorly returned leading to a further 9-10% loss of eligible cases/controls. In total, 61% of eligible cases and 68% of eligible controls were recruited, closely representative of the NI population, and with no evidence of selection bias. Of those recruited, 97% gave consent for linkage to medical records. Thirty-three percent of women had an unplanned pregnancy, and 76% suspected they were pregnant by 5 weeks gestation, with no significant differences between cases and controls. There was considerable discordance between self-report, maternity records, and prescription records regarding medications obtained/taken in the first trimester, but no evidence of differences between cases and controls that would indicate substantial recall bias. There was high concordance between self-report and maternity data regarding folic acid supplementation but maternity data was significantly less often confirmed by self-report for cases than controls.

Conclusions
Our results suggest that hybrid data collection approaches are a useful way forward for aetiological studies to reduce responder burden and address and estimate recall bias, and that the Baby Hearts study protocol is suitable for replication in other populations, modified to the local context.