Multiple Imputation for the Comparison of Two Screening Test in Two Phase Alzheimer Studies
*Ofer Harel, University of Connecticut
Xiao-Hua (Andrew) Zhou, University of Washington
Keywords: missing data, multiple imputation, verification bias, two-phase studies
It is well documented that two-phase designs, which are common in epidemiological studies often suffer from verification bias. In two-phase design all subjects are screened and then a subset of them is verified using a gold standard. When comparing the accuracy of two screening tests, inferences are commonly made using only the verified sample. When the two screening tests have only two values and we are trying to estimate the differences in sensitivities and specificities of the tests, one is actually estimating a confidence interval for differences of binomial proportions. We suggest imputation procedures in order to correct the verification bias. This procedure deals with the difficulty of estimating the difference of two binomial proportions as well as the missing data. We compare different methods of estimation, and evaluate the use of multiple imputation in this case.