Research methodology and analytic techniques used in the Journal Obstetrics & Gynecology.

Abstract

To quantify trends over time in the research methodology and statistical reporting of published articles in the journal Obstetrics & Gynecology between the years 1991 and 2001. We reviewed 226 articles reporting original clinical research in Obstetrics & Gynecology, including 75 consecutive articles from 1991 and 1996 and 76 consecutive studies from 2001. Articles were classified by research methodology and level of evidence: I for randomized trials; II for controlled trials and observational, analytic studies; and III for case reports or case series. We also noted whether the investigators reported a research hypothesis, P values, effect sizes, 95% confidence intervals, and sample size/power calculations. We calculated chi2 for linear trend to evaluate changes over this 10-year period. Seventy-four percent of the 226 studies evaluated were level II. The percentage of randomized trials did not increase from 1991 to 2001. However, level II studies increased from 71% in 1991 to 88% in 2001 (P = .01). During the same period, level III studies (case reports and case series) declined from 19% in 1991 to 1% in 2001 (P < .001). We also noted that a clearly stated research hypothesis, effect measures (and respective confidence intervals), and sample size calculations were more common in more recent years. In the journal Obstetrics & Gynecology, we noted that observational, analytic studies increased in frequency, anecdotal reports decreased, and statistical reporting has improved. However, there still exists considerable opportunities for improvement.