CAN DECISION TREE ANALYSIS BE USED TO IDENTIFY THE OPTIMAL EVENT PATHWAY TO REACH AN ACCURATE DIAGNOSIS? THE MAZE OF CARE STUDY IN OVARIAN CANCER

Tuesday, October 25, 2011
Grand Ballroom AB (Hyatt Regency Chicago)
Poster Board # 57
(ESP) Applied Health Economics, Services, and Policy Research

Lisa M. Hess, PhD, MA, MS, BA1, Michael W. Method, MD, MPH, MBA2, Frederick B. Stehman, MD1, Tess D. Weathers, MPH1, Paridha Gupta, MPH1 and Jeanne M. Schilder, MD1, (1)Indiana University School of Medicine, Indianapolis, IN, (2)Northern Indiana Cancer Research Consortium, South Bend, IN

Purpose: There are no screening or early detection tests for ovarian cancer, representing a major challenge to accurate diagnosis. The current study explored the use of capturing and analyzing health care (HC) events within the pathways followed by individual patients leading to a definitive diagnosis.

Method: Women who were referred to gynecologic oncology for a suspected ovarian cancer (OC) were enrolled to this study and recalled their HC experiences in a semi-structured interview. Complete medical records (MRs) from all HC providers were obtained from the time the patient recalled suspecting a health issue through the time of diagnosis. We conducted a preliminary analysis using decision tree analysis to identify the optimal path to a rapid diagnosis incorporating both time and HC events. HC cost will be incorporated in the full model.

Result: Of the 105 women who were enrolled to the study and provided medical record access, 92 eligible patients completed the interview about their experience in the HC system. Most of the HC encounters during this process were with primary care physicians (24.4%) and gynecologists (18.1%). More than 19% of patients did not report seeing a gynecologic oncologist prior to the diagnosis of ovarian cancer. The pathways originating from PCPs averaged 3.7 HC visits in 73 days prior to diagnosis (range 2-8 visits, 0-435 days). Time was significantly different between treatment pathways (p=0.003). Other common pathways averaged 2.8 visits in 62 days (for visits initiated in the emergency room or urgent care), and 3.2 visits and 64 days for care originating with a gynecologist. Women diagnosed with ovarian cancer had an average of 3.3 HC encounters over an average of 83 days whereas women with benign conditions had an average of 3.7 encounters over an average 104 days prior to diagnosis. Medical records will be used to validate patient-reported events and for additional information about tests and procedures completed.

Conclusion: Although most women present to their PCPs with symptoms and concerns that ultimately lead to referral to gynecologic oncology for a suspected OC, the PCP pathway was associated with the longest time to reach diagnosis, but not with additional health care visits. Additional tasks occurring at each event (e.g. blood work, scans) will be included in the full analysis to explore the expected costs of each pathway.