In:
Journal of Clinical Oncology, American Society of Clinical Oncology (ASCO), Vol. 38, No. 29_suppl ( 2020-10-10), p. 204-204
Abstract:
204 Background: The rapidly growing number of cancer survivors in the US have substantial healthcare needs requiring surveillance and care for the late and long-term effects of cancer treatment and comorbidities. Lacking a clear system of care, experts recommend a personalized approach to survivorship care. The objective of this study was to test a clinical prediction algorithm to distinguish low-complexity breast cancer survivors who may be suited to self-manage their survivorship care and be followed by their primary care provider (PCP) from survivors who require specialty care. Methods: We used the Surveillance and Epidemiology End Results (SEER) registry – Consumer Assessment of Healthcare Providers and Systems (CAHPS) survey data to identify women diagnosed with stage 0-3 breast cancer between 2003 and 2011. Cross-validated random forest machine learning models separately estimated survivors’ independent risk of all-cause death, cancer-specific death, recurrence, or severe late effects within 3 years following treatment completion. The absence of these outcomes identified survivors as potentially eligible for self-management and PCP care. Predictors included measures of baseline health status and health care utilization, patient socio-demographic characteristics, cancer characteristics, and financial burden. Results: Among the 4,516 survivors in the primary cohort, 82% were white, and the mean (SD) age was 75.1 (7.8) years. Almost 50% were diagnosed with Stage I breast cancer, followed by 25.2% with Stage 2, 19.3% with Stage 0, and 5.6% with Stage III. Within the 3-year follow-up period, 372 (8.2%) survivors died (111 or 2.5% from cancer), 665 (14.7%) experienced recurrence, and 488 (10.8%) were hospitalized due to severe late effects. Predicting all-cause death resulted in 91.9% out-of-sample accuracy, a 37.6% improvement over an uninformed model. Important predictors across outcomes included age, geographic region, diagnosis year, financial burden, comorbidities, and cancer stage. Conclusions: Survivors requiring specialty care are characterized by higher comorbidity, lower educational attainment, and advanced age, suggesting that, in addition to cancer characteristics, personalized care pathways developed in response to our findings must account for social and contextual factors as well.
Type of Medium:
Online Resource
ISSN:
0732-183X
,
1527-7755
DOI:
10.1200/JCO.2020.38.29_suppl.204
Language:
English
Publisher:
American Society of Clinical Oncology (ASCO)
Publication Date:
2020
detail.hit.zdb_id:
2005181-5