Estimating the prevalence of malformation of the heart in the first year of life using capture-recapture methods.

The authors estimated the prevalence of heart malformation during the first year of life, using five data sets with varying degrees of completeness from two English regional health authorities. These areas covered a total population of 6,872,000. Analysis was carried out using capture-recapture methods, including log-linear modeling, on data collected between June 1993 and August 1994. A large number of cases in the community were unrecorded by any of the current sources of information. In South East Thames, where an antenatal training screening program for detecting heart malformations had been implemented in the late 1980s, the estimated prevalence rate varied from 5.5 per 1,000 births (95% confidence interval (CI): 3.5, 10.8) to 9.0 per 1,000 births (95% CI: 6.4, 14.2), depending on the assumptions in the model and the number of sources used in the analysis. In the Wessex region, which did not have a formal training program, prevalence was lower and varied little, from 4.3 per 1,000 (95% CI: 3.4, 6.0) to 5.1 per 1,000 (95% CI: 4.0, 7.2), according to assumptions. These two estimates were reasonable rates in comparison with reports in the literature. This analysis was helpful in demonstrating that the training program designed to identify severe heart malformations during the antenatal period in one of these regions had no lasting impact on prevalence.