How often should trial results be updated for publication

For all trials with time-to-event outcomes there remains the possibility of additional events being reported after the main study publication. When is it appropriate to update the results? Ideally one should pre-specify a frequency with which data will be examined post-publication and some guidelines as to what would warrant publication. These should generally be based on either the accumulation of specific durations of follow-up (e.g. there may be interest in examining the results, when all surviving patients have a minimum of five or ten years follow-up) or accumulation of a given number of additional events. There should be no need for frequent reviews - the dangers of too-frequent analyses may be just as great after the closure of a trial as they are during its accrual phase, and similarly the timing of any analysis should be decided independently of the results. For example, one maybe prompted to update results by the publication of other, relevant, papers, or may receive a request to provide updated data for an individual patient data meta-analysis. In the absence of other 'prompts,' the accumulation of perhaps 15-20 per cent more events, or 5-year intervals of follow-up, may be a rule of thumb to trigger an updated analysis. Bear in mind though, that with long-term follow-up, particularly in trials with reasonably high overall survival rates, an increasing proportion of late events will not be disease-related. Generally this is simply a consequence of the age of the trial population. However, it may also represent late effects of treatment. For example, with respect to radiotherapy for early breast cancer, treatment-related mortality did not become apparent until 20-25 years after treatment, and meant that analysis of mature data found no overall benefit to radiotherapy with respect to all-cause mortality [9].

0 0

Post a comment