Different formats are available for reporting the results of clinical trials. In an article just published in the British Medical Journal (BMJ), employees at the German Institute for Quality and Efficiency in Health Care (IQWiG) investigated to what extent journal publications, reports posted in trial results registries (registry reports), and clinical study reports submitted to regulatory authorities provide sufficient information to evaluate clinical trials. For this purpose they extracted information from benefit assessments of drugs; these assessments were finalized by IQWiG between 2006 and 2011.
Formats show different strengths and weaknesses
The IQWiG authors found that clinical study reports submitted by manufacturers to regulatory authorities during the drug approval process provide the most complete information to evaluate a clinical trial, but such reports are in general not publicly available.
The two other formats investigated performed considerably worse. These were journal publications, which generally present study results with some delay, and registry reports (several trial registries have been established worldwide). These formats show different strengths and weaknesses. Information on study methods, such as the allocation of participants to treatment groups or the statistical analysis of data, was more complete in journal publications than in registry reports, which on the other hand provided more information on study outcomes.
Registry reports are only available for about a third of studies
Registry reports and journal publications may therefore supplement each other. To a certain extent, the combination of these two formats can compensate the fact that clinical study reports are often accessible only to regulatory authorities. In the preparation of systematic reviews, researchers should always consider registry reports, but such reports are often unavailable: in the benefit assessments used by IQWiG, registry reports were only provided for about a third of the studies analysed.
Worldwide adequate standards for registry reports required
The IQWiG authors also identified a second problem. The international harmonization of drug approval processes also led to the establishment of standards for clinical study reports. However, such worldwide standards are lacking for trial results registries, especially since many registries are maintained by pharmaceutical companies and the legal obligation to register results only exists in a few countries.
The most comprehensive legislation came into effect in the United States in 2007 (www.clinicaltrials.gov). Although similar legislation has also been passed in Europe (in 2004) and Germany (in 2011) it has not yet been implemented.
In addition, the requirements for reports posted in the German trial registry (www.pharmnet-bund.de) are not stringent enough to provide sufficient information for a benefit assessment. This was clearly shown by IQWiG article in the BMJ. In order to render reports from trial registries comparable internationally and to ensure that the information contained in them is sufficient, the IQWiG authors urgently call for the mandatory implementation of worldwide uniform standards.
Clinical study reports are an indispensable source, at least for older studies
Legislation prescribing the reporting of study results is relatively recent, covering studies retrospectively at most until 2004, and even this coverage is incomplete. However, most drugs currently used in clinical practice were tested in clinical trials before 2004, so that registry reports on these drugs are not available, or available only to a limited extent. In order nevertheless to assess drugs using the most complete evidence base possible, the IQWiG authors propose making clinical study reports submitted to regulatory authorities publicly available, at least for older studies. As these reports are kept by authorities, the additional effort required would be manageable.
Selective publication of study data is the main focus of the BMJ special issue
The current BMJ special issue includes several other articles investigating the problem of publication bias and limited access to study results. This special issue was planned after the publication of another article by IQWiG authors investigating the problem of the biased reporting of study results on the benefit of the antidepressant reboxetine. This article was widely discussed both in Germany and internationally, also in the non-specialist press.