Article type
Year
Abstract
Background: In the approval process for drugs, biologics, and devices in the USA, the sponsor of a therapy must present the Food and Drug Administration (FDA) with clinical trials data to establish efficacy and safety. These sponsor-provided data are not publicly available, but for approved therapies the report on these data by a FDA statistician in the premarket approval document is available on the FDA web site (www.fda.gov). This provides a potential source of unpublished clinical trials or additional outcome data for published trials. It has been shown that documents provided by the FDA under the Freedom of Information Act can serve as a useful supplement to a meta-analysis [1] and we speculated that the web site might provide the same utility.
Objectives: We set out to evaluate if efficacy data from premarket approval documents on the FDA web site would be a useful supplement to a meta-analysis.
Methods: For evaluation, we chose two groups of therapies for pain relief in osteoarthritis: hyaluronic acid (HA) and cyclooxygenase-2 (COX-2) inhibitors. The FDA Center for Devices and Radiological Health has approved 3 formulations of HA, and the FDA Center for Drug Evaluation and Research has approved 3 COX-2 inhibitors. Premarket approval documents for the therapies were located on the FDA web site and evaluated for controlled clinical trials of osteoarthritis subjects. Using only documents from the web site, we evaluated the reporting of these trials using the items of the Jadad scale, specification of intention-to-treat (ITT) results, presence of variability measures (standard deviations, standard errors, confidence intervals, or exact p-values) to permit calculation of a weighted or standardized mean difference, and the number of primary and secondary outcomes presented.
Results: Reports of 23 controlled clinical trials were obtained for the 6 therapies. 16 (70%) trials were characterized as randomized, but the method of randomization was not described for any trial. All trials were characterized as blinded, but descriptions of blinding methods were presented for only 5 (22%). Information on withdrawal was provided for 19 (83%) trials. Average Jadad scores by therapy are shown in Table 1. ITT results were given for 14 (61%) trials, and variability was reported for 17 (74%). Only 12 (52%) trials had both ITT results and variability estimates, and 6 of these had Jadad scores of 3 or more. The average number of outcomes presented was 5.4 with a range from 1 to 12.
Conclusions: The randomization status of trials is probably under-reported in these documents, and reporting of methods for randomization and blinding is uncommon. However, about 25% of the clinical trials were reported with Jadad scores of 3 or more, with ITT analysis, and adequate variability. In addition, several outcomes were reported from most trials. A portion of the Clinical trials data available on the FDA web site is suitable for inclusion in meta-analysis and the web site provides a resource for meta-analytic research.
References: 1. MacLean CH, Morton SC, Ofman JJ, Roth EA, Shekelle PG; Southern California Evidence-Based Practice Center. How useful are unpublished data from the Food and Drug Administration in meta-analysis? J Clin Epidemiol. 2003;56:44-51.
Objectives: We set out to evaluate if efficacy data from premarket approval documents on the FDA web site would be a useful supplement to a meta-analysis.
Methods: For evaluation, we chose two groups of therapies for pain relief in osteoarthritis: hyaluronic acid (HA) and cyclooxygenase-2 (COX-2) inhibitors. The FDA Center for Devices and Radiological Health has approved 3 formulations of HA, and the FDA Center for Drug Evaluation and Research has approved 3 COX-2 inhibitors. Premarket approval documents for the therapies were located on the FDA web site and evaluated for controlled clinical trials of osteoarthritis subjects. Using only documents from the web site, we evaluated the reporting of these trials using the items of the Jadad scale, specification of intention-to-treat (ITT) results, presence of variability measures (standard deviations, standard errors, confidence intervals, or exact p-values) to permit calculation of a weighted or standardized mean difference, and the number of primary and secondary outcomes presented.
Results: Reports of 23 controlled clinical trials were obtained for the 6 therapies. 16 (70%) trials were characterized as randomized, but the method of randomization was not described for any trial. All trials were characterized as blinded, but descriptions of blinding methods were presented for only 5 (22%). Information on withdrawal was provided for 19 (83%) trials. Average Jadad scores by therapy are shown in Table 1. ITT results were given for 14 (61%) trials, and variability was reported for 17 (74%). Only 12 (52%) trials had both ITT results and variability estimates, and 6 of these had Jadad scores of 3 or more. The average number of outcomes presented was 5.4 with a range from 1 to 12.
Conclusions: The randomization status of trials is probably under-reported in these documents, and reporting of methods for randomization and blinding is uncommon. However, about 25% of the clinical trials were reported with Jadad scores of 3 or more, with ITT analysis, and adequate variability. In addition, several outcomes were reported from most trials. A portion of the Clinical trials data available on the FDA web site is suitable for inclusion in meta-analysis and the web site provides a resource for meta-analytic research.
References: 1. MacLean CH, Morton SC, Ofman JJ, Roth EA, Shekelle PG; Southern California Evidence-Based Practice Center. How useful are unpublished data from the Food and Drug Administration in meta-analysis? J Clin Epidemiol. 2003;56:44-51.