For outcomes where events are observed over a prolonged period, for example survival in cancer trials, it is important to check that follow-up is as up to date as possible and that it is consistent for each of the intervention groups. Producing a ‘reverse’ Kaplan Meier curve, based on just those patients who have not experienced the event of interest, with censoring then used as the event, can provide a useful check on the balance of follow-up across the groups.
For any individual study, the results of all these checks should be considered together to build up an overall picture of the study and the quality of the data that have been supplied, and any potential problems. Any concerns should be brought diplomatically to the attention of the researchers responsible. Usually, problems turn out to be simple errors or misunderstandings, which can be resolved through discussion. Major problems that cannot be resolved are rare.
A copy of the data as supplied should be archived before carrying out conversions or modifications to the data. Throughout the data checking processes, it is important that any changes and alterations made to the supplied data are properly logged.