Sammendrag
Cohort analyses can be very expensive and an alternative is a nested case-control design (NCC) where one (or more) controls are sampled for each case. This is statistically less efficient since information is only available for cases and controls. In particular, in a multiple outcome situation, the traditional partial likelihood estimator for NCC can not handle controls sampled for cases of one disease as controls for another disease. There has, however, been developed methods based on inverse probability weighting (IPW) that allow for reusing controls (and cases). Also maximum likelihood methods (MLE) for NCC have been developed. Furthermore, in addition to the information collected on cases and controls, some information is usually known for the entire cohort, like gender, age, etc. This information can be utilized to obtain more accurate estimates for MLE. A calibration technique that aims at finding more efficient weights by using information known for the entire cohort has been developed for case-cohort designs. We suggest a generalization to the NCC. We have a study of prostate cancer and vitamin D where the controls are matched on three variables in addition to follow-up time. Motivated by this, we try to generalize IPW to handle additional matching. Through simulations we compare MLE and IPW both with and without left truncation. The generalization of IPW to handle matching is evaluated on the data set.