If data is extracted from electronic health records and used for survival analysis, can outcome vars be considered missing or are they just considered right censored data points?


Quoting from Stef van Buuren's Flexible Imputation of Missing Data, Section 3.7.3:

Survival analysis is the method of choice if censoring is restricted to the single outcomes. The approach is, however, less suited for censored predictors or for multiple interdependent censored outcomes.

With only one type of event and no missing predictor values, the usual practice is to treat such a case as having a right-censored event time based on the last known follow-up time.

With missing predictor values or multiple types of outcomes, however, multiple imputation of the missing data could be a preferred approach. Van Buuren provides an example Algorithm 3.6 that uses predictive mean matching, Kaplan-Meier estimates, and bootstrapping for imputation. He also provides further references to literature on the topic.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.