Background Systematic reviews are an integral component of evidence based health care, however little is known on how well they report the potential harms of interventions. We assessed the reporting of harms in recent systematic reviews of interventions relevant to clinical gastroenterology. Methods: We identified all systematic reviews of randomized trials of gastroenterology interventions published from 2008 to 2012 in ten gastroenterology journals and five general medical journals with the highest impact factors. We adapted the CONSORT reporting guidelines for harms and used this to assess qualitative and quantitative parameters of harms reporting. Regression analyses were undertaken to determine predictors of more comprehensive harms reporting. Results In total, 78 systematic reviews were identified, with 72 published in gastroenterology journals and 6 published in general medical journals. Overall, 1 in 3 systematic reviews (26/78, 33%) did not refer to harms of the intervention anywhere in the manuscript. Less than half of the studies included adverse events as an outcome measure, and data on absolute rates of adverse events were only provided in 28% of studies. The majority (65%) did not include any figures/tables on adverse events, however, all included figures/tables on efficacy outcomes with a mean of 3 (1-7). The absence of reporting on harms was often misinterpreted as safety in the discussion sections. Regression analyses indicated that the use of reporting guidelines was significantly associated with more comprehensive harms reporting (p=0.04). Conclusion: The coverage of harms in gastroenterology systematic reviews is largely inadequate, and is highly asymmetrical compared with the reporting of benefits. It is likely that existing systematic reviews have underestimated potential harms, and may be misleading for clinical decision making and evidence based guidelines. We suggest review authors consider adverse events as primary outcome measures equal in status with efficacy outcomes, and that an extension of the PRISMA reporting guidelines, specifically targeting harms reporting, be developed.
- Systematic reviews
- Harms of interventions