| Summary: | <p>Clinical trials should be based on well-constructed study protocols. In practice, however, trial protocols are variable in their content and quality. This thesis examined the design and planned methodological conduct of contemporary clinical trials using a cohort of clinical trial protocols. The DPhil candidate was granted access to 259 clinical trial protocols approved by the United Kingdom National Research Ethics Service in 2012. The cohort included non-randomized clinical trials as well as randomized controlled trials (RCTs) and there was no restriction by intervention, funding source or medical specialty. There were five broad findings. First, clinical trial protocols were variable in the quality of reporting of multiple methodological items, including items that have been associated with bias. Second, the majority of trial protocols included patient reported outcome measures (PROMs) but few discussed the validity of the PROM, the strategy for dealing with missing data and the sample size calculation. Third, the primary and secondary outcomes in the trial protocols were compared to the trial registry entry if available. In this analysis, agreement in the reporting of outcomes in the trial protocol and trial registry was modest. Notably, pharmacologic trials and multicentre trials were less likely to have discrepancies between the protocol and registry as compared to non-pharmacologic and single centre trials, respectively. Fourth, multi-arm trials made up just 8% of all clinical trial protocols. Methodological items that were important for multi-arm trials were also omitted in trial protocols. Finally, phase I trials were generally safe but recommended strategies for the identification of starting doses and the administration of the novel agents were infrequently followed. Also, phase I trials with serious adverse events were not disseminated. Taken together, this thesis has provided insight into different methodological issues in the design, conduct and dissemination of clinical trials.</p>
|
|---|