Doctor decision making with time inconsistent patients

Non-adherence to treatments is prevalent. The aim of this paper is to model how doctors should adapt their medical treatment decisions if non-adherence is due to present-bias in the patient population, and to test the predictions of this model in a lab experiment. Under certain conditions, a rational doctor should adapt to non-adherence by choosing a treatment all patients complete (though less effective) when the probability of a patient being present-biased is sufficiently large. This is explored in a lab experiment where we test whether students in the doctor role adapt their behaviour as they learn about the distribution of non-adherence (due to present bias) in the patient population over the rounds of the experiment. We test the model prediction when we align in-dividual incentives with the goal of maximising overall patient welfare. The results show that, on average, participants adapt to non-adherence as they learn about the probability of non-adherence (due to present-bias). However, a proportion of participants do not adapt to the optimal choice. The rate of adaptation was similar for the first 5 rounds under both individual incentives and salary. However, participants continued to adapt after round 5 under individual incentives whilst adaptation plateaued under salary. The adaptation to non-adherence may indicate that adherence can be improved by providing doctors with information about the probability of non-adherence (due to present-bias) in their patients.

Automated summary

This paper models and tests how doctors should adapt their medical treatment decisions to accommodate patient non-adherence. The results show that doctors adapt to non-adherence as they learn about the probability of non-adherence in patients.