News and Updates
Anmeldung zur Tagung
Eröffnung des Kongresses
Index of Authors
Interim Analysis of Controlled Clinical Trials - Breaking with a Taboo?
1Kersten-Gomez I., 2Krummenauer F.,
1Johannes-Gutenberg-Universität, Klinik und Poliklinik für Augenheilkunde (Mainz)
2Johannes-Gutenberg-Universität Mainz, Koordinierungszentrum für Klinische Studien (KKS) (Mainz)
Purpose: To summarize approaches for planned and exploratory interim analyses of controlled clinical trials with respect to GCP (Good Clinical Practice) to maximize patient safety and predictability of study budget and time lines; to illustrate the use of planned interim analyses to save a clinical trial from a non-significant result, which would otherwise result only due to underestimated sample sizes.
Methods: Phase III or IV clinical trials usually require tremendous sample sizes to ensure sufficient statistical and clinical significance of the trials results after analysis. However, this often affords several years of waiting, until these results can finally be obtained. Therefore it would be of great interest - although not necessarily legal - to perform interim analyses during the overall trial time. To achieve this, two approaches can be distinguished: First safety boards, operating independently of the trial sponsor and its principal investigator, will always be allowed to perform unblinded interim analyses on all study data at hand to ensure maximum patient safety. However, sponsors may as well wish to perform interim analyses on their own, to prolong the overall sample size; this is deemed necessary to avoid finishing the trial with a non-significant result, where the latter would only be due to an underestimated sample size. Furthermore sponsors would be interested, whether the trial could be stopped with a significant result before achieving the primarily intended overall sample size - both the ethical and economical gain in this positive earlier stopping would be obvious. Based on the planning considerations of a controlled phase III trial for the medical prevention of glaucoma progression, these procedures will be illustrated with regard to the interim analyses logistical impacts on the study performance as well as to their possible consequences. The group sequential design of Pocock and the adaptive design of Bauer will be compared concerning advantages and limitations concerning predictability of budget and time lines of the overall trial.
Results: Whereas the groups sequential approach allows for a priori computation of a fixed overall sample size (which can be severely reduced by stopping the trial based on an interim analysis), the overall sample sizes of adaptive designs can hardly be predicted, since the sample size of a study phase after an interim analysis crucially depends on the latters results. Therefore the increase in design flexibility during the trial is accomplished by a decrease in budget and time predictability. However, both designs allow for early stopping of the overall trial with a significant result, as soon as study effects turn out larger than expected in the planning phase. In particular the adaptive design even allows for early stopping with a confirmatory negative result, if study effects turn out smaller than previously suggested.
Conclusion: GCP recommends continuous safety monitoring of clinical trials by flexible interim analyses of independent safety boards. However, planned interim analyses by the sponsor are also possible, as long as they have been sensitively incorporated into both the planning of overall sample sizes and the clinical trial protocol in advance.