Advancing Patient-Centered Outcomes In Surgery: Development Of A Research Agenda On Patient-Reported Outcomes In Surgery And Creation Of A Patient-Centered Decision Aid For Stoma Formation During Low Anterior Resection.

Abstract

Background: Greater integration of patient-centered outcomes in surgery remains paramount in the current medical climate. Patient-reported outcomes have emerged as a useful tool to assess subjective outcomes such as pain, function, and quality of life. In order to determine areas of crucial research on patient-reported outcomes and advance cross-disciplinary collaboration, a research agenda was created through a Delphi survey of stakeholders in collaboration with the inaugural Patient-Reported Outcomes (PROS) in Surgery Conference. In the second part of this study, a decision model was created to simulate patient-centered outcomes after low anterior resection for rectal cancer. Randomized controlled trials support the use of defunctioning ileostomies as it reduces the morbidity and mortality of anastomotic leakage; however, they are associated with greater risk for readmission, complications (bowel obstruction, stomal hernia) and impact patient quality of life. Assimilating these data can be difficult for clinicians. To this end, a decision analysis was developed to evaluate the impact of defunctioning ileostomy on patient quantity and quality of life. Methods: To create a research agenda, an iterative Web-based interface was used to create a conference-based, modified Delphi survey. Participation was limited to PROS conference registrants, which included surgeons, PRO researchers, payers and other stakeholders. In the first round, research items were generated from qualitative review of responses to open-ended prompts. In the second and third rounds, items were ranked using a Likert scale. The top 20 items by mean rating were selected for the research agenda. In the decision analysis, a decision tree compared defunctioning ileostomy creation to no defunctioning ileostomy creation after low anterior resection for rectal cancer. The base case for model analysis was a 65-year-old man with resectable, stage II-III rectal cancer at 8cm from the anal verge status post neoadjuvant chemoradiation therapy. Long term health states after surgery were stoma, no stoma or death. The primary outcome measure was quality-adjusted life-years (QALYs). Model probabilities and health-state utilities were obtained from the literature using a priori search criteria. The probabilities and utilities were varied over their plausible ranges through sensitivity analyses of one, two and three variables. Results: In round one of the Delphi Survey, participants submitted 459 items, which were reduced through qualitative review to 53 distinct items across seven themes of PROs research. A research agenda was formulated after two successive rounds of ranking. The research agenda identified three themes important for future PROs research in surgery: (1) PROs in the decision-making process, (2) integrating PROs into the EHR and (3) measuring quality in surgery with PROs. For the decision analysis, defunctioning ileostomy creation was the preferred strategy (8.81 vs. 8.73 QALYs). In one-way sensitivity analyses, defunctioning ileostomy remained the preferred strategy over the plausible range of all variables with four exceptions: if the risk of clinical leak was less than 8.9%, if the leak-associated mortality was less than 4.1 %, if the one-year ileostomy reversal rate was less than 70% or if the utility of a stoma was less than 0.69. Conclusion: A research agenda on PROs in surgery was created using a modified Delphi survey of stakeholders that will help researchers, surgeons, and funders identify crucial areas of future PROs research in surgery. In the decision analysis, low anterior resection with defunctioning ileostomy is the preferred treatment strategy, however for a subset of patients, no ileostomy may be the preferred if patient perception and risk are accounted. The quality of life with a stoma, the probability of a clinical leak and the rate of stoma reversal significantly impacted model preferences, and should be considered in decisions on ileostomy formation during rectal resection.