Conventional Search Strategies Cannot Easily Identify Those Trials Of Drug Therapy Which Provide Quantitative Adverse Effects Data

Loke Y.K.. Edwards J., Deny S. University of Oxford, UK.

Introduction:
Objectives: Adverse effects (AEs) are regarded as secondary outcome measures and are reported in detail in only a proportion of therapeutic trials. Systematic reviewers evaluating AEs would find it helpful if trials containing such data could be easily recognized. Our study looks at whether search strategies relying on information from the abstract and Medical Subject Heading (MeSH) index can reliably identify those trials where AEs are reported.
Methods: We identified 58 analgesic and 13 hypertension trials which had numerical data on adverse effects in both the treated and control groups. The abstracts were checked for identifying terms such as "side" or "adverse" effects. We also looked at whether MeSH indexing of the trials contained the term "adverse-effects".
Results: Only 16 of the analgesic trials were MeSH indexed as having AEs while abstracts were more useful, with 28 containing some reference to AEs. A combined search covering both areas would still have railed to find the 23 trials where no mention was made about AEs in the abstract or MeSH index. Results were similar for the hypertension trials, with 3/13 MeSH indexed for AEs and only 2/13 commenting on AEs in the abstract.
Discussion: Trials which contain AE data often make no mention of this in the abstract and may not be indexed as such under the MeSH categories. Given the absence of any other useful identifying markers, reviewers searching for high-quality AE information will need to manually check the full-text articles of all controlled trials involving the therapeutic agent of interest.

Rome 1999 PA19