I review a lot of papers using observational datasets (think large secondary national databases) and many report survival. Risk tables often show substantial censoring (e.g. out of 10,000 patients, only 3300 are at risk at 48 months and median survival has not been reached). Often, these datasets have followup in some patients up to 10 years or more. One paper stated that 'median survival was >10 years' when only 1/3rd of the patients initially at risk were available for analysis at that time point but survival was 75%. This is both incorrect (it should state median not reached) and misleading because the vast majority of the patients in the study are randomly censored by 48 months. Are there guidelines for when the data should be right truncated based on random censoring or some other way to ensure that conclusions appropriately reflect that we don't really know what happened to a large proportion of the patients after a certain timepoint?


Your Answer

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

Browse other questions tagged or ask your own question.