How can a Register be Used?

Whether concerned with the identification of teratogenic exposures, or with planning and evaluation of health services, or both, registers can be used in two main ways:

As a basis for surveillance using routinely collected data. Every register routinely collects a core of standard information on each malformed child, and a core of information (often more limited) on non-malformed children in the population.
As a basis for special or ad-hoc studies, such as case-control studies, requiring further data collection. The presence of a register which has already done the work of identifying who in the population has a congenital anomaly, with details of diagnosis, can greatly facilitate the conduct of ad-hoc studies which seek to address specific hypotheses concerning teratogenic exposures or effectiveness of health services.

The decision as to which data should be included in the routine dataset of a registry, and which data should be collected only in ad-hoc studies is a difficult one. Each data item that is included in the routine dataset uses resources that must be balanced against other uses of those resources. Collection of incomplete and inaccurate data is generally a waste of resources. Depending on local circumstances, it may be justifiable for the registry to concentrate almost entirely on data about the baby and its diagnosis in its routine data collection, leaving most risk factor data for collection in ad-hoc studies. Some ad-hoc data collection will always be necessary to address new or more elaborate hypotheses. However, registers that do not record the identity of children for confidentiality reasons can experience difficulties in supporting ad-hoc studies effectively and efficiently.

In general, risk factor data must be present for both cases and controls (non-malformed children) in order to be interpretable. While all registers collect basic information about the number of births in their population by type of birth and maternal age, some registers in addition collect the same set of risk factor information on control babies as on case babies in their routine data collection. There are also useful approaches to analysing risk factor data among malformed cases only, where children with different malformations act as controls for each other in "proportionate" analyses (1). For example, specific associations between particular drugs and particular malformation types can be sought.