Estimating prevalence of Fetal Alcohol Syndrome in Oregon

Dual- and multiple-record system (DRS and MRS) estimation methods can be used to estimate population size, prevalence of disease, and completeness of census data. Two of the goals of FAS surveillance in Oregon are to identify children with FAS and estimate the prevalence of FAS. DRS and MRS estimation methods were applied to multiple sources of data to estimate the prevalence of children with FAS. Data from the FAS Surveillance Project (birth certificates, Medicaid Management Information System data, and medical record abstractions) were reviewed to identify children with an FAS diagnosis. Each set of records from a data source was deemed a separate sampling event. Event histories were constructed for each child, reflecting each child's FAS diagnosis from each data source. Counts associated with the event histories were arranged in a 2k contingency table and contained all the data available for estimating population size and prevalence. Statistical methods included the Chapman estimation method for dual data sources and log-linear regression modeling for multiple data sources. Preliminary estimates of the prevalence of FAS in Oregon varied widely depending on data source combinations. FAS prevalence ranged from 0.55 to 2.6 per 1,000 live births. Estimates relying on a single source of information are often incomplete and underestimate the true proportion of the affected population. DRS and MRS can be used to obtain a more complete estimate of the affected population. However, challenges in case ascertainment and violations of estimation assumptions should be thoroughly evaluated and estimates should be interpreted with caution.