Background General practise data provide large population-based cohorts of individuals with prospectively collected medical information with promising potential for studying the causes and consequences of congenital anomalies. We sought to validate these data through comparison with congenital anomaly registries.

Methods Our study population was 794,209 children in The Health Improvement Network (THIN) primary care database, born between 1990 and 2009 with a median follow-up of 6.7 years. We compared the birth prevalence of any major and system-specific congenital anomalies with the European Surveillance of Congenital Anomalies (EUROCAT) United Kingdom registries.

Results The birth prevalence of any major congenital anomaly for children in THIN diagnosed before one year of age was 198 per 10,000 (95% CI 195 – 201) which was slightly higher than the EUROCAT prevalence of 167 per 10,000 (Relative Risk 1.18, 95% Confidence Interval 1.16 – 1.20). Absolute differences in prevalence between THIN and EUROCAT were small across 16 system-specific anomaly groups. The majority of children in THIN with major congenital anomalies had recorded diagnoses before 1 year of age (72%), but including children diagnosed at any age increased the overall prevalence to 277 per 10,000 births.

Conclusions The prevalence of congenital anomalies in THIN was consistent with EUROCAT for early diagnoses, demonstrating THIN to be a valid source for investigating congenital anomalies. Age of diagnosis is an important factor in explaining a higher overall prevalence in THIN; the inclusion of diagnoses made after one year of age substantially improves capture of diagnoses making THIN more complete than registry data.