Profiling prostate cancer patient preferences using latent models

Melissa Seward Yale, Purdue University

Abstract

The use of patient treatment preferences in medical decision making is becoming common practice. For prostate cancer treatment decision making, patient preferences are a driving factor in treatment choice as the primary treatments have similar survival outcomes. Two studies tested and validated a theoretical framework of prostate cancer treatment decision making by exploring the domains of patient preferences related to prostate cancer treatment and analyzing whether individual prostate cancer patients can be grouped according to their preferences. The first study assessed the measurement quality of the Values Insight and Balance Evaluation Scales for Prostate Cancer (VIBEs-PC) by examining the dimensionality and construct validity to provide further evidence of the measure's applicability for use as a patient preference index in prostate cancer treatment decision making. Analyses included item and person fit, category ordering, dimensionality, local independence, targeting, person reliability, and differential item functioning (DIF) using the Rasch rating scale model. The findings of this study suggest the VIBEs-PC displays preliminary evidence for reliability and validity as a measure of the multiple domains of prostate cancer patient preferences. The second study examined latent heterogeneity in prostate cancer patient preferences and its relationship to patient demographic characteristics and two distal outcome variables, treatment choice and decision regret, through latent profile analysis. Four latent classes of prostate cancer patients were identified based on the empirical results and substantive understanding. Specifically, patients expressed various degrees of importance within each domain (e.g., physical, social, and psychological) and different patterns of importance relative to the other domains within classes. However, this latent heterogeneity was not categorized by observed patient characteristics and had minimal impact on prior treatment choice and decision regret. Implications of the results are discussed and areas for future research are considered.

Degree

Ph.D.

Advisors

Christ, Purdue University.

Subject Area

Quantitative psychology|Oncology

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS