Ankur Pandya

Red blood cell alloimmunization and consequent delayed hemolytic transfusion reaction (DHTR) incidence and mortality in patients with sickle cell disease (SCD) are high. A shared transfusion resource has decreased both in other countries, while in the United States cost concerns persist. We conducted a Markov cohort simulation of a birth cohort of alloimmunized patients with SCD to estimate lifetime DHTR incidence, DHTR‐specific mortality, quality‐adjusted life expectancy (QALE), and costs with the implementation of a shared transfusion resource to identify antibody history versus without (i.e., status quo). We conducted our analysis using a lifetime analytic time horizon and from a United States health system perspective. Implementation of shared transfusion resource projects to decrease cumulative DHTR‐specific mortality by 26% for alloimmunized patients with SCD in the United States, relative to the status quo. For an average patient population of 32 000, this intervention would generate a discounted increment of 4000 QALYs at an incremental discounted cost of $0.3 billion, resulting in an incremental cost‐effectiveness ratio of $75 600/QALY [95% credible interval $70 200–81 400/QALY]. The results are most sensitive to the baseline lifetime medical expenditure of patients with SCD. Alloantibody data exchange is cost‐effective in 100% of 10 000 Monte Carlo simulations. The resource would theoretically need a minimum patient population of 1819 patients or cost no more than $5.29 million annually to be cost‐effective. By reducing DHTR‐specific mortality, a shared transfusion resource in the United States projects to be a life‐saving and cost‐effective intervention for patients with SCD in the United States.

There is increasing interest in moving away from “one size fits all (OSFA)” approaches toward stratifying treatment decisions. Understanding how expected effectiveness and cost‐effectiveness varies with patient covariates is a key aspect of stratified decision making. Recently proposed machine learning (ML) methods can learn heterogeneity in outcomes without pre‐specifying subgroups or functional forms, enabling the construction of decision rules (‘policies’) that map individual covariates into a treatment decision. However, these methods do not yet integrate ML estimates into a decision modeling framework in order to reflect long‐term policy‐relevant outcomes and synthesize information from multiple sources. In this paper, we propose a method to integrate ML and decision modeling, when individual patient data is available to estimate treatment‐specific survival time. We also propose a novel implementation of policy tree algorithms to define subgroups using decision model output. We demonstrate these methods using the SPRINT (Systolic Blood Pressure Intervention Trial), comparing outcomes for “standard” and “intensive” blood pressure targets. We find that including ML into a decision model can impact the estimate of incremental net health benefit (INHB) for OSFA policies. We also find evidence that stratifying treatment using subgroups defined by a tree‐based algorithm can increase the estimates of the INHB.