How Big Is the Right Size For a Clinical Trial?
When a drug company embarks on a clinical trial, one of the risks it runs is the nonadherence of participants to the rules. In simplest terms, that means if a patient is to take a pill in the morning and another at night, deviating from the schedule is nonadherence. If too many participants deviate, the trial could fail.
An article in Applied Clinical Trials discusses the statistical implications of nonadherence. With an estimated 20% of patients never even taking the "investigational medical product," the risks of a complete failure of a trial are high. To offset that, sponsors attempt to enroll many more participants than they might otherwise need.
In the article, writer Moe Alsumidaie examines three types of nonadherence: IMP nonadherence (not taking the product or doing so in a way different than prescribed); dropping out of the study; and fraud. The latter includes intentional efforts to deceive the investigators and may run as high as 28%.
To account for all these situations, Alsumidaie offers three mathematical models to calculate the number of participants a trial must have in order to report a statistically significant result. To obtain realistic results, he says, "requires precise data around IMP adherence, dropout rates and fraud."
Precision in obtaining that data can be difficult. However detecting nonadherence and adjusting for it is becoming more sophisticated, he points out, mentioning some of the methods sponsor have adopted. Nevertheless, Alsumidaie, cautions that, "The use of imprecise methods to capture nonadherence data as well as strict definitions of what constitute nonadherence present risk to drug development."