Misleading reporting and interpretation of results in major infertility journals

Abstract

Objective: To evaluate the proportion of randomized controlled trials (RCTs) published in top infertility journals indexed on PubMed that reported their results with proper effect estimates and their precision estimation, while correctly interpreting both measures.

Design: Cross-sectional study evaluating all the RCTs published in top infertility journals during 2014.

Setting: Not applicable.

Patient(s): Not applicable.

Intervention(s): Not applicable.

Main outcome measure(s): Proportion of RCTs that reported both relative and absolute effect size measures and its precision.

Result(s): Among the 32 RCTs published in 2014 in the top infertility journals reviewed, 37.5% (95% confidence interval [CI], 21.1-56.3) did not mention in their abstracts whether the difference among the study arms was statistically or clinically significant, and only 6.3% (95% CI, 0.8-20.8) used a CI of the absolute difference. Similarly, in the results section, these elements were observed in 28.2% (95% CI, 13.7-46.7) and 15.6% (95% CI, 5.3-32.8), respectively. Only one study clearly expressed the minimal clinically important difference in their methods section, but we found related proxies in 53% (95% CI, 34.7-70.9). None of the studies used CIs to draw conclusions about the clinical or statistical significance. We found 13 studies where the interpretation of the findings could be misleading.

Conclusion(s): Recommended reporting items are underused in top infertility journals, which could lead to misleading interpretations. Authors, reviewers, and editorial boards should emphasize their use to improve reporting quality.

Keywords: P value; Reporting quality; absolute difference; confidence intervals; minimal clinically important difference.