Economic Burden of Low-value Healthcare on Patients with Localized Prostate Cancer: Statistical & Machine Learning Approaches
Date of Graduation
School of Medicine
Malcolm D. Mattes
Adults with incident localized prostate cancer represent a large, medically complex population at risk for low-value care. Evidence-based guidelines recommend conservative management (CM) for localized prostate cancer patients with multimorbidity and limited life expectancy, however, 2 in 3 still choose treatment. This dissertation pursued three Aims to address research gaps related to healthcare practices associated with significant morbidity and economic burden on older men with incident localized prostate cancer: 1) examine the leading predictors of low-value healthcare practice of prostate cancer treatment for low-risk prostate cancer; 2) assess the role of patient‐reported experience with care on high-value prostate cancer management; and 3) estimate the association of high-value care on non-cancer related healthcare expenditures using machine learning and statistical approaches. In this study, 2 in 3 adults received low-value prostate cancer treatment. Multimorbidity and care fragmentation were among the leading predictors of low-value prostate cancer treatment and contrary to expectations, life expectancy was a weak predictor of treatment receipt. Social determinants of health were highly ranked predictors of treatment. Higher “timeliness of care” patient reported experience scores were associated with high-value CM use. Other forms of low-value care before incident prostate cancer diagnosis were associated with higher non-cancer related healthcare expenditures while high-value CM was associated with lower costs. In summary, this dissertation highlights the negative effect of multimorbidity and care fragmentation on overtreatment, high-value care, and cost outcomes. Perceptions of timely care with healthcare providers and systems have significant impact on high-value CM use among older men with localized prostate cancer. This dissertation reports strong independent predictive associations of incremental low-value healthcare use before incident prostate cancer diagnosis to have significant increases on long-term non-cancer related costs.
Fiano, Ryan, "Economic Burden of Low-value Healthcare on Patients with Localized Prostate Cancer: Statistical & Machine Learning Approaches" (2021). Graduate Theses, Dissertations, and Problem Reports. 8118.