Reporting of harms by randomised controlled trials in ophthalmology

To evaluate the reporting of harms by randomised controlled trials investigating intravitreal therapies for diabetic macular oedema. A thorough literature search identified eligible reports. Two authors independently extracted data from these articles using a prospectively created checklist. The main outcome measure was compliance with the 10 recommendations of the 2004 Consolidated Standards of Reporting Trials statement extension for better harms reporting. Secondary outcomes were the predictors of the number of recommendations met and the amount of space devoted to harms reporting. Thirty-six reports involving 7246 eyes met the criteria for analysis. The fidelity of the data extraction was excellent, with Cohen's κ coefficient of 0.90 for all items extracted. The median number of recommendations met was six, IQR 5-7. Recommendation 4 (describe how harms-related information was collected) was met by 97% of articles and recommendation 8 (present the absolute risk of each adverse event) by 92%. The least frequently met recommendations were numbers 3 (list addressed adverse events with definitions of each), 31%, and 6 (describe participant withdrawals because of harms), 36%. The mean percentage of the results section devoted to harms-related data was 25.8%, SD 10.8%. Harms reporting in published reports of trials of intravitreal therapies for diabetic macular oedema is still not entirely adequate despite increased attention and efforts to standardise it.