Estimating treatment effect on medical cost and examining medical cost trajectory using splines and change point techniques.

Abstract

In the world of growing medical needs, other than the clinical outcomes, the cost of healthcare is one of the important aspects to evaluate. The cost of treatment could act as a decisive factor on which one to choose from two equally likely effective treatment options. In literature, the most used quantity for the cost of treatment is cumulative lifetime cost since the diagnosis of a disease. While it provides a bird' eye view of the treatment cost, it fails to capture the underlying pattern of the treatment cost trajectory. We developed a marginal structural functional model (MSFM) using an I-spline basis to examine the accumulative cost trajectory over time. Further, to obtain a valid average treatment effect (ATE) estimator, we used the inverse probability of treatment weighting (IPTW) to control the confounding between the cost and the treatment groups. Penalized spline regressions were used to estimate the cost trajectory and ATE. We carried out extensive simulation studies to examine the performance of the proposed method. We also applied our proposed method on gastric cancers patients in SEER-Medicare 2005-2014 database and illustrated the cost pattern over time under different treatments. Another important aspect of healthcare cost is to identify the underlying pattern of the cost due to a disease. The estimation of healthcare cost and locating the change points across the cost trajectory is important to policymakers and clinicians, given the increasing costs of healthcare delivery, budgetary constraints, and the aging population. While in the literature the lifetime cost was often studied, the estimation of cost patterns and change points for cost patterns are important to policymakers and insurance companies. We develop a piece-wise linear mixed effect change point model as well as a I-spline based non-parametric model to estimate the cost trajectory over time and evaluate the change points for cost. We model the patient-level cost trajectory as well as population-level cost trajectory by using the patient-level regression parameters, which depend on patient-level characteristics and treatment choices. We applied our proposed methods on pancreatic cancer patients in SEER-Medicare 2005-2014 database and concluded that both models capture the cost trajectory as well the change points.