NETWORK META-ANALYSIS USING DATA FROM PUBLISHED TRIALS AND DATA FROM THE FOOD AND DRUG ADMINISTRATION MEDICAL REVIEWS A CASE EXAMPLE OF FIRST LINE MEDICATIONS FOR OPEN ANGLE GLAUCOMA

Embargo until
Date
2015-05-18
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
Background Network meta-analysis (NMA) is a method that incorporates both direct evidence within trials and indirect evidence across trials. The impact of different data sources on NMAs is not widely examined. Objectives The objective of this study was to conduct and compare the results from NMAs using clinical trial data available in the Food and Drug Administration (FDA) medical reviews (FDA trials) and those published in the medical literature (published trials) on first line medications for open angle glaucoma. Methods We searched the Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE in March 2014 and the US Food and Drug Administration’s website in April 2014 for randomized, parallel group trials on the first line topical medications for primary open-angle glaucoma and ocular hypertension. Eligible trials had to compare a single active treatment with no treatment/placebo or another single active topical medical treatment. Two individuals independently assessed trial eligibility, abstracted data, and assessed the risk of bias. Using the FDA trials and the published trials, we performed pair-wise meta-analyses and Bayesian network meta-analyses on intraocular pressure at 3 months, our pre-specified primary outcome. We compared the results from these analyses. Results We included 16 FDA trials and 105 published trials. The network based on FDA trials had 10 nodes (nine active drugs and placebo/vehicle) with a total sample size of 6183. The network based on published trials had 14 nodes (13 active drugs and placebo/vehicle) with a total sample size of 16898. There were 9 common treatments between the FDA trials and the published trials in the network meta-analysis, resulting in 36 comparisons. We estimated mean difference in IOP of 36 pairs of these treatments. The median relative difference of the two networks was 14% (interquartile range: 7% to 42%). The relative differences were greater than 25% in 14 pairs. Point estimates from two pairs showed opposite directions. Generally, the results of the published trials had better precision than the results of the FDA trials because of the larger sample size of the published trials. In terms of relative rankings, bimatoprost, travoprost, and latanoprost are the best ranking drugs regardless of the data source, although 4/9 (44%) drugs had different ranking. Conclusions Even though the sample size is smaller than published literature reviews, systematic reviewers should consider FDA medical reviews because of the additional amount of information they can provide. The data source did not seem to change the inference of the relative effectiveness of most of the drugs studied. The relative rankings changed for the middle-ranked drugs but not for the top-ranked drugs. Whether reporting bias has a role in the differences we observed needs to be further evaluated.
Description
Keywords
Network meta-analysis, FDA medical reviews, Reporting bias
Citation