Use of multiple imputation in supersampled nested case‐control and case‐cohort studies
研究了在巢式病例对照和巢式队列研究中，当部分协变量昂贵时，如何通过多重插补利用超采样数据提高效率，模拟显示该方法比传统分析更有效，且效率损失不大。
Abstract Nested case‐control and case‐cohort studies are useful for studying associations between covariates and time‐to‐event when some covariates are expensive to measure. Full covariate information is collected in the nested case‐control or case‐cohort sample only, while cheaply measured covariates are often observed for the full cohort. Standard analysis of such case‐control samples ignores any full cohort data. Previous work has shown how data for the full cohort can be used efficiently by multiple imputation of the expensive covariate(s), followed by a full‐cohort analysis. For large cohorts this is computationally expensive or even infeasible. An alternative is to supplement the case‐control samples with additional controls on which cheaply measured covariates are observed. We show how multiple imputation can be used for analysis of such supersampled data. Simulations show that this brings efficiency gains relative to a traditional analysis and that the efficiency loss relative to using the full cohort data is not substantial.