Quality of reporting clinical trials published in five leading Sri Lankan medical journals

To assess the quality of reporting of clinical trials published in five leading Sri Lankan medical journals.

Six medical journals were hand searched for clinical trials published from 1982 to 2011. Eligible criteria were all randomized and non-randomized clinical trials conducted in humans. A checklist was developed based on CONSORT and TREND recommendations. Each study was independently evaluated by two reviewers. Outcome measures were presence of checklist items in published reports.

Fifteen randomized and 24 non-randomized parallel group trials from 724 studies in five journals met the inclusion criteria. Out of 39 trials, 39 (97%) clearly described the objectives, 16 (41%) defined the periods of recruitment, 7 (18%) reported how sample size was determined, 10 (25.6%) reported the methods to enhance the quality of measurements, 20 (51%) reported baseline demographic and clinical characteristics of each group, 4 (10%) showed flow diagrams, 23 (69%) reported statistical methods used to compare groups for primary outcomes, 21 (54%) reported effect size, 4 (10%) reported its precision, and 20 (51%) interpreted the results in the context of current evidence, and 5 (13%) described the generalizability of the findings. Of the 15 randomized trials, only one (7%) reported sequence generation, 5 (33%) allocation concealment, 9 (60%) reported blinding status of participants or investigators, and 2 (13%) reported intention to treat analysis.

Reporting of several essential recommendations remained suboptimal. Education and training of trial methods and awareness of the CONSORT and TREND statements and more attention to the quality of reporting may improve matters.