Comparison of estimators in nested case–control studies with multiple outcomes