MedicalResearch.com Interview with:
Victor Serebruany, MD, PhD
HeartDrug Research, Towson, Maryland
Department of Neurology Johns Hopkins University Baltimore, Maryland
Medical Research: What is the background for this study? What are the main findings?
Dr. Serebruany: Missing data are common challenges to the validity of trial results, yet it is unclear how to characterize the extent of missing data. We compared the published lost-to-follow-up rates to incomplete follow-up rates determined from subject records submitted to the FDA for major oral antithrombotic trials. The 21 trials having both sets of rates included 270,089 patients followed for a median duration of 20 months. The mean published lost-to-follow-up rates is 0.4% (median 0.3%, range 0.005% to 2%), consistently much lower than the FDA incomplete follow-up rates: mean 12% (median 13%, range 2% to 23%). There is no correlation between the publication and FDA-calculated rates (R 0.07, p = 0.76). The FDA rates exceed greatly the endpoint rate differences: mean 1.3% (median 1,0%, range 0.2% to 3.0%).
Medical Research: What should clinicians and patients take away from your report?
Dr. Serebruany: That the FDA incomplete follow-up rates greatly exceed the endpoint rate differences raises questions of whether the endpoint differences may be due to differential follow-up rather than drug effect. That they greatly exceed the measures routinely reported for trials, i.e., lost-to-follow-up rates, suggests that current trial reporting is inadequate. Completeness of follow-up and other indicators of trial data quality should be considered when interpreting trial results.
Medical Research: What recommendations do you have for future research as a result of this study?
Dr. Serebruany: Trials publications should report more details on follow-up rates and other measures of trial data quality. The FDA incomplete follow-up rates based on subject records would be a good start. They, and other measures of data quality, should be examined in trials of other conditions and compared to trial outcomes. We need measures to express the operational uncertainty of trial results just as we use P values to express the statistical uncertainty.
Marciniak TA, Cherepanov V, Golukhova E, Kim M, Serebruany V. Drug Discontinuation and Follow-up Rates in Oral Antithrombotic Trials. JAMA Intern Med. Published online January 11, 2016. doi:10.1001/jamainternmed.2015.6769.
Victor Serebruany, MD, PhD (2016). Incomplete Follow Up Rates Exceed 10% in Oral Antithrombotic Trials