Using tramadol for acute pain as an example, where information on 3500 patients was provided, the results of two trials were explained, one showing that tramadol was a highly effective analgesic, the other showing that it was no different from placebo. The truth was somewhere in between. Although tramadol had been in use in some European countries for many years, meeting regulatory requirements in the United States required studies to meet contemporary requirements, the meta-analyses revealed useful results.
Meta-analysis of randomized trials before a new technology is commercially available is even rarer, although there are at least two examples. Meta-analyses are typically conducted several years after initial commercial availability because it takes time to publish randomized trials conducted for efficacy and/or safety reasons. The importance of meta-analyses in drug development and regulatory processes is increasingly recognized.
Meta-analysis results are undoubtedly important, both in the regulatory process and in the evaluation of rare but serious adverse events. For COX-2 inhibitors, a meta-analysis was planned before randomized trials to examine the association between treatments and rare events.
However, the greatest point of change is in the period immediately following commercial launch. Media interest can raise patient expectations at a time when healthcare professionals and organizations have the least knowledge and experience, and few have had the opportunity to consider the full impact of the new technology on budgets and services. For Sildenafil, for example, 85% of initial prescriptions occurred in the first 12 weeks of availability at a health care provider in New England. It is at this point of market approval when the greatest need for the best information exists. At best, only a few studies have been published, and although they can be large and usually demonstrate a difference from placebo or common practice, they are unlikely to accurately measure the magnitude of benefit.
We wanted to assess whether clinical trial reports submitted for marketing approval would form the basis of a systematic review at the time of launch if they were publicly available. We did this regarding the erectile dysfunction treatment sildenafil (Viagra), using clinical trials provided by Pfizer Ltd.
No search strategy was needed because this review was of material provided by Pfizer UK Ltd in the form of clinical trial reports in a September 1997 marketing application for sildenafil (Viagra). QUORUM guidelines were otherwise followed. Previous intent was to use studies relevant to the use of sildenafil in clinical practice. This required the setting to be home, not clinic, the use of sildenafil as needed rather than fixed dosing schedules (such as daily tablets), and studies of a minimal duration, which we arbitrarily set as four weeks.
Excluded were studies with laboratory measurements of penile tumescence or rigidity with single doses of sildenafil, studies that examined only erectile function in a clinical setting, studies that used fixed daily dosing rather than as needed, and studies that were shorter than four weeks. Included were randomized trials that examined sildenafil with efficacy or safety data, were longer than four weeks, and were conducted at home, using doses in the approved range of 25 mg to 100 mg, although lower and higher doses would be analyzed if there were sufficient information. Clinical trials in men with erectile dysfunction specifically due to single causes such as spinal cord trauma or diabetes were not included because they would represent clinical heterogeneity along with the other data.
Each report was scored for quality using a three-item, 1-5-point quality scale. Points were assigned to studies depending on whether they were randomized or double-blind, and failures or study discontinuations were cited. An additional point was assigned if the method of randomization or double-blinding was described and appropriate.
From each study, we extracted the number of patients treated per group, the dosing regimen, the study design, and the number of patients with efficacy and/or safety results. The denominator was the number of randomized patients, so the results were on an intention-to-treat basis. This analysis included all randomized patients regardless of diary completion, protocol concordance, or missing data. It was assumed that patients with missing or illegible diary data had an intercourse success rate of 0%.
|
|
|
