Population-based birth-defect and risk-factor surveillance: data from the Northern Netherlands.

In many countries, birth defect monitoring systems have been set up in order to identify new teratogens as soon as possible. The usual approach to monitoring involves analysis of the frequency of specific birth defects over time. This approach has been criticized as having poor statistical power to detect epidemics due to new rare teratogenic exposures. A proposed alternative approach is the on-going analysis of risk-factor data with a case-control approach. In this paper, we present birth-defects and risk-factor surveillance data from the Northern Netherlands (NNL). For years of birth 1981-1994, 4014 cases had been registered. We investigated combinations of 32 diagnostic categories and 77 risk factors. For 10 combinations a P value < 0.01 was found; for another 25, the P value was between 0.01 and 0.05. We then checked these positive associations against data from the Metropolitan Atlanta Congenital Defects Program (MACDP) and the MAternal DRug Exposure surveillance project (MADRE). In all three data sets, an association between maternal use of psychotropic drugs (psycholeptics) and cleft lip with or without cleft palate (CLP) was present. The highest odds ratio was found for CLP and maternal use of oxazepam in the NNL data (OR = 8.17, 95% CI 1.26-42.2). Both in the MACDP data and in the NNL data, an association between maternal smoking and clubfoot was found. Although the odds ratios were low, the attributable fraction derived from the NNL data was 11%. Methodologic issues that should be considered in this approach include exposure ascertainment and classification, outcome specificity, and type I errors. The strengths of this approach include its population-based nature and the ability of users to check results against results from other similar systems.