PATIENT‐CENTERED RESEARCH

Abstract

Research ObjectiveFor over a century, appendectomy has been the first‐line approach to managing acute appendicitis, but several European trials suggest antibiotics alone can be efficacious. While randomized controlled trials (RCTs) provide rigor for generating evidence, the acute nature of appendicitis, differences in risk and cost of interventions, and deeply held beliefs about treatment options pose challenges to feasibility for randomizing between surgical and medical treatments. To design a RCT, we previously assessed willingness to randomize to appendectomy or antibiotics through surveys administered through Amazon’s Mechanical Turk (MTurk)—an online crowdsourcing platform.Study DesignTo evaluate this engagement approach for study design, we assessed crowdsourced willingness to randomize for a hypothetical trial of surgery versus antibiotics for appendicitis compared with the true willingness to randomize among patients with acute appendicitis approached for enrollment in the Comparison of Antibiotic Drugs and Appendectomy (CODA) study. The first survey occurred in December 2014 and invited MTurk members to review information on appendicitis and the two treatment options, and then indicates willingness to participate in a theoretical study randomizing them to “surgery” or “antibiotics first.” Those unwilling to randomize could provide a reason. Participants were compensated $0.25. CODA participants were individuals >18 years presenting to the emergency department with their first diagnosis of image‐confirmed appendicitis. Electronic health record review confirmed eligibility criteria. Patients willing to participate were randomized to appendectomy or 10 days of antibiotics. Patients declining participation were asked to provide a reason. We compared rates of willingness to randomize among MTurk and CODA participants and qualitatively analyzed reasons for declining randomization and treatment preferences.Population StudiedMTurk participants represented a sample of US‐based individuals registered on MTurk in December 2014. CODA participants included eligible individuals approached for study consent between July 2016 and May 2019.Principal FindingsAmong 557 individuals surveyed on MTurk, 253 (45%) reported willingness to randomize. Among those unwilling to randomize, 80 (26%) preferred appendectomy, 42 (14%) preferred antibiotics, and 182 (60%) indicated other reasons (e.g., lack of interest, lack of time, and desire for autonomy). Of the 3,829 eligible CODA participants, 1,207 (32%) consented for the RCT. Among eligible patients unwilling to randomize, 1,777 (75%) preferred appendectomy, 471 (20%) preferred antibiotics, and 134 (5.6%) indicated other reasons for not randomizing.ConclusionsMTurk provided a unique platform to obtain input on the feasibility for a trial comparing appendicitis treatments. While MTurk overestimated the willingness to randomize, the results informed study design, ultimately resulting in successful recruitment of trial participants. Assuming stated and revealed preferences may differ, we discounted the rate reported in MTurk and projected a randomization rate of 30% for CODA, which reflected actual recruitment outcomes during study conduct.Implications for Policy or PracticeExpanded outreach to populations of interest during study development is an important step in understanding the feasibility of study conduct. Our experience suggests crowdsourcing offers valuable information when designing RCTs and may be an efficient adjunct to conventional study development. Using exclusion criteria that mirrors the targeted study population may further improve insight to strengthen study design including selection of outcomes.Primary Funding SourcePatient‐Centered Outcomes Research Institute.