This report details the development, validation, and implementation of a digital health decision platform focused on guideline- and policy-congruent pain management pathways in the US Military Health System (MHS).
A discrete-event, patient-level model based on the Defense Health Agency Stepped Care Model for Pain included 11 nodes (eg, care escalation, emergency encounters) parameterized with cross-validated methods and verified through configuration-controlled tests. Validation incorporated clinician review, guideline and policy alignment, and population statistic comparisons. A synthetic cohort with common pain conditions was simulated to demonstrate model interpretability and policy relevance.
Many statistical approaches were incorporated into the simulation-based decision support platform. A patient generator produced simulated patients representative of the population based on 2016 to 2019 data. Statistical models determined the next encounter type (eg, primary care, physical therapy), system of care (eg, civilian versus MHS facilities), primary care encounters until secondary or tertiary care, days between appointment request and completion, procedural pain intervention receipt (eg, injections), prescription receipt, and end of pain episode. Several interrelated outcomes were captured, including opioid prescription receipt, emergency room utilization, and pain episode recurrence. Next, the capabilities necessary for modeling counterfactuals (hypothetical conditions) were developed to simulate outcomes relevant for individual and health system decision support.
The resulting simulation-based digital decision support platform enables testing for counterfactual policy and resource allocation decisions as it relates to chronic pain management in the MHS. Future work is needed to apply and further validate the platform.