Candidate for the Lee B. Lusted Student Prize Competition
Purpose: The influence of future technology improvement on current medical technology adoption decisions is overlooked in many cost-effectiveness analyses. We ask how long a patient with a chronic disease should wait for more effective treatments before accepting the best available treatment. This decision involves a tradeoff between a patient's deteriorating health while waiting and the expected magnitude of future technological improvement. This is applicable to many chronic diseases characterized by ongoing health deterioration and rapid improvements in treatment technologies. The treatment decision of patients with chronic hepatitis C virus (HCV) provides a relevant example. Chronic HCV affects 3-4 million Americans and historically has been difficult to treat. New treatments show promise in achieving higher rates of cure.
Method: We model the treatment timing decision as an optimal stopping problem using a dynamic, discrete-time, finite-horizon Markov Decision Process. We derive structural properties of the model and solve a three-period problem analytically. We present a numerical example of optimal treatment decisions for chronic HCV with patients stratified by age, sex, race, life-style risk and initial liver damage.
Result: It is optimal for healthier patients to delay treatment longer than sicker patients; the threshold for treatment adoption (i.e., the minimum treatment effectiveness for which the patient should choose to adopt treatment now) is decreasing over time under certain conditions that are commonly satisfied in chronic diseases; as current treatment improves, the decision moves from waiting towards adopting treatment; and patients should wait longer when there is higher expected magnitude of future treatment improvement. Figure 1 shows an optimal treatment decision region resulting from our HCV example. Our analyses suggest that patients should be more willing to accept treatment when they are older, sicker or have higher co-morbidities.
Conclusion: We develop a framework to guide optimal treatment decisions for a chronic disease when treatment technologies are improving over time. Use of this framework could help maximize patients' lifetime health by informing decisions about treatment priority.
Figure 1. Decision space for 50-year-old white male in the three-period problem as a function of current treatment effectiveness (x-axis) and expected future treatment improvement (y-axis). Treatment is accepted under fewer parameter combinations in period 1 (black region) than in period 2 (additional striped region).