Standardizing abortion research outcomes (STAR): Results from an international consensus development study.

ObjectiveTo develop a minimum data set, known as a core outcome set, for future abortion randomized controlled trials.Study designWe extracted outcomes from quantitative and qualitative systematic reviews of abortion studies to assess using a modified Delphi method. Via email, we invited researchers, clinicians, patients, and healthcare organization representatives with expertise in abortion to rate the importance of the outcomes on a 9-point Likert scale. After 2 rounds, we used descriptive analyses to determine which outcomes met the predefined consensus criteria. We finalized the core outcome set during a series of consensus development meetings.ResultsWe entered 42 outcomes, organized in 15 domains, into the Delphi survey. Two-hundred eighteen of 251 invitees (87%) provided responses (203 complete responses) for round 1 and 118 of 218 (42%) completed round2. Sixteen experts participated in the development meetings. The final outcome set includes 15 outcomes: 10 outcomes apply to all abortion trials (successful abortion, ongoing pregnancy, death, hemorrhage, uterine infection, hospitalization, surgical intervention, pain, gastrointestinal symptoms, and patients' experience of abortion); 2 outcomes apply to only surgical abortion trials (uterine perforation and cervical injury), one applies only to medical abortion trials (uterine rupture); and 2 apply to trials evaluating abortions with anesthesia (over-sedation/respiratory depression and local anesthetic systemic toxicity).ConclusionUsing robust consensus science methods we have developed a core outcome set for future abortion research.ImplicationsStandardized outcomes in abortion research could decrease heterogeneity among trials and improve the quality of systematic reviews and clinical guidelines. Researchers should select, collect, and report these core outcomes in future abortion trials. Journal editors should advocate for core outcome set reporting.