Population-level intervention and information collection in dynamic healthcare policy

Abstract

We develop a general framework for optimal health policy design in a dynamic setting. We consider a hypothetical medical intervention for a cohort of patients where one parameter varies across cohorts with imperfectly observable linear dynamics. We seek to identify the optimal time to change the current health intervention policy and the optimal time to collect decision-relevant information. We formulate this problem as a discrete-time, infinite-horizon Markov decision process and we establish structural properties in terms of first and second-order monotonicity. We demonstrate that it is generally optimal to delay information acquisition until an effect on decisions is sufficiently likely. We apply this framework to the evaluation of hepatitis C virus (HCV) screening in the general population determining which birth cohorts to screen for HCV and when to collect information about HCV prevalence.

Optimal Allocation of Research Funds under a Budget Constraint

“…These 2 features, that c c s (n s ) and ENBS s (n s ) are or are well-approximated to be continuously differentiable, are present in EVSI examples in the literature (e.g., see realproblem examples in figure 9 of McKenna et al, 34 figure 5 of Stevenson et al, 35 figure 4 of Cipriano and Weber, 36 and Kunst et al, 37 where variable cost functions are linear and EVSI is a smooth Gaussian approximation). Therefore, features of the necessary optimality conditions for OPT2 provide important insight into many applied research funding allocation problems.…”

Optimal Information Collection Policies in a Markov Decision Process Framework

“…At each time, the policy maker can choose to screen the current cohort of 50-year olds and to purchase sample information about parameters influencing the decision. Using a stochastic dynamic programming approach, we have previously shown that it may be optimal to delay information collection about the time-varying parameter (HCV prevalence) until the information is more likely to result in a change in policy [26].…”

Dynamic Learning in Markets: Pricing, Advertising, and Information Acquisition

“…Optimal information acquisition at time t = 1.Remark 2 (Information-Cost Threshold). The expression for κ * 1 implies the cost thresholdβ d (1 + α)σ6 d /4, so that nontrivial market intelligence is economical if and only if the firm's information-acquisition cost is such that β ∈ (0,β d ).…”

“…Some applications in healthcare are of this form; see, e.g., Cipriano and Weber[6] who determine the amount of information to collect (using public health screening) to inform a planner's decision to discontinue a public health screening program for hepatitis C.…”

“…The firm also updates the variance of its belief, fromσ 2 d to σ 2 d /(1 + κ 1 σ 2 d ), independent of s 1 6. Here and in what follows we use the abbreviation [x] + max{0, x} for all x ∈ R.…”