Timing of Testing and Treatment of Hepatitis C and other Diseases
Georgia Institute of Technology, School of Industrial and Systems Engineering
∗ Corresponding author: jswann@isye.gatech.edu or (404) 385-3054
Many papers in the medical literature analyze the cost-effectiveness/savings of screening by simulating the disease and a limited number of testing policies specified a priori. However, this may be insufficient to determine the best timing of the tests or incorporate changes over time. In this study, we develop a Markov Decision Process (MDP) model for diseases where our goal is to determine the best timing for testing (and treatment) decisions when the presence of the disease is not known in advance, we apply the model to Hepatitis C, and we discuss the insights for healthcare practice.
Our study is motivated in part by improving healthcare response to Hepatitis C, a blood-borne liver disease. In the US, an estimated 3.9 million people are currently infected, and it is the leading cause for liver transplants. Most people are unaware they have the disease until they develop end stage liver disease but may spread the disease to others even when they are asymptomatic; alcohol consumption significantly increases the progression of the disease. There is currently no vaccine for HCV, although treatments exist that can cure with a 54% rate if applied early enough. The high cost of treating advanced Hepatitis C, combined with the infectivity, behavioral aspects and long asymptomatic period make HCV an ideal candidate for screening programs.
We develop a general model that allows for the awareness of a disease to change behavior, and we analyze the model for structural results. We focus initially on minimizing the cost from a societal perspective. We find that under certain assumptions, the immediate costs and rewards determine a condition that is sufficient to establish that testing (and treatment) is beneficial. Furthermore, when the disease can be modeled by two states (e.g. healthy and sick), we show the conditions under which it is cost-saving to test (and treat). We also develop a contrasting MDP model to produce an efficient frontier of timing decisions with respect to the cost per quality adjusted life year (QALY) measure. Our general model can be used to study the timing of testing and treating for many diseases where behavior or risk changes may impact disease transmission or cost, including generating optimal policies or ones on the efficient frontier.
We use the MDP model in the case of Hepatitis C to study a limited number timing of test and treatment actions for various populations. We use medical data to estimate the progression of the disease, prevalence, health costs, and infectivity. We find the most cost-saving policies for the populations, e.g., commercial sex workers who drink alcohol excessively should be tested up to three times, while the general population with no risk factors should not be tested at all. We also determine that uniform intervals between tests are not necessarily the best strategy when multiple testing is used; this can be driven by risk behaviors that change over time as well as the health costs. We determine the minimum thresholds such that testing (and/or treating) would be cost- saving and cost effective in various groups, as well as the age range in which testing and treating is cost-saving and cost effective. While our overall recommendations agree with those of the CDC for testing for Hepatitis C, we go beyond their recommendation by considering the significant effects that alcohol consumption and behavior change can have on the timing and frequency of tests.
