|
Abstract:
|
Differential ascertainment in a case control study means that different exposure groups have different ascertainment rates. We propose a method to estimate the odds ratio in a case-control study with differential ascertainment. Specifically, when there is differential ascertainment, but multiple sources of information are available, we propose to use a capture-recapture approach to estimate the true number of cases in both groups. In preliminary work, we have applied our method to simulated data from a Rasch model where the differential ascertainment was artificially created among the exposed and unexposed subjects in the population. Then we also introduced correlation among the lists. The result was approximately unbiased in the first case. In the second case the approach did have some bias but was still significantly better than the naive approach of just considering the observed quantities. We plan to apply our approach to two datasets, the NVDRS RAD and one on the Syrian war. In the first, we will consider as cases the victims that died because of maltreatment, in the second, the civil casualties. Sex, race and age will be the risk factors.
|
