Case-control study of leukemia and environmental radiation fallout from the Chernobyl accident: Selection of hospital cases and multistage area probability sampling (MAPS) of controls in urban and rural areas of Ukraine

Case-control studies often involve cases selected from hospital or disease registries and controls selected from a computerized roster that has complete coverage of the study base. In our matched case-control study of Chernobyl radiation and cancer in the Ukraine, there are no computerized population rosters with complete coverage of the population and of those computerized case registries that do exist, most are incomplete and do not provide information on all cancer cases. In rural areas of the Ukraine with the highest levels of radionuclide contamination, low telephone coverage makes random digit dialing for control selection impractical. We describe a method for finding cases and controls based on searches in hospital records and existing disease registries. We also introduce a multistage area probability sampling (MAPS) scheme for control finding based on random sampling of "physician's districts" of similar size within raions (counties) from which cases originated. The MAPS method for control selection in this study assures low variation in access to and utilization of health care services between cases and controls and reduces the potential for overmatching on radiation exposure expected with neighborhood controls.