Reporting of harm in randomized controlled trials published in the urological literature

Evidence-based decision making seeks to balance potential benefits and harms (adverse effects) of health care interventions for an individual patient. We determined the prevalence and completeness of harm reporting in randomized controlled trials in the urological literature.

We performed a systematic literature search of all randomized controlled trials of therapeutic interventions published in The Journal of Urology, Urology, European Urology and BJU International in 1996 and 2004. Each article was reviewed by 2 independent investigators for 10 harm reporting criteria recommended by the CONSORT group. Discrepancies were settled by discussion and consensus.

A total of 152 randomized controlled trials met the inclusion criteria, of which 109 (72%) reported adverse event outcomes. The median number of harm reporting criteria satisfied improved marginally from 1996 to 2004 (2.8 to 3.3, p = 0.36). A large proportion of studies failed to address harm in the abstract (55, 36%), introduction (71, 47%) and discussion (52, 34%). Few studies specified which adverse events were evaluated (21, 14%), when harm information was collected (32, 21%) or how the harm was attributed to the intervention (5, 3%). Only 48 (32%) articles provided reasons for patient withdrawal and 1 in 5 (33, 22%) reported the severity of adverse events.

Randomized controlled trials published in the urological literature contain significant deficiencies in adverse event reporting. These findings suggest the need for reporting standards for harm in urological journals. Improvements in adverse event reporting would permit a more balanced assessment of interventions and would enhance evidence-based urological practice.