abstract
Simulation models of opioid use disorder (OUD) aim at evaluating the impact of different treatment strategies on population-level outcomes. Researching Effective Strategies to Prevent Opioid Death (RESPOND) is a dynamic population state-transition model that simulates Massachusetts (MA) OUD population synthesizing data from the MA Public Health Data Warehouse, published survey studies, and randomized trials. We implement an empirical calibration approach to fit RESPOND to multiple calibration targets, including yearly counts of fatal overdoses and detox admissions in 2013-2015, and 2015 OUD population counts in MA. We used capture-recapture analysis to estimate the OUD population and to quantify uncertainty around calibration targets. 1 The empirical calibration approach involves Latin hypercube sampling for a parameter search of the multidimensional space, comprising demographics of “arrivals”, overdose rates, treatment transition rates, and substance use transition probabilities. The algorithm accepts proposed parameter values when the respective model outputs are “close” to the observed calibration targets based on uncertainty ranges of targets. Calibration provided an excellent fit to the model calibration targets. The flexibility of the algorithm also allowed us to identify certain “questionable” parts of the model structure and explore the underlying relationships between the model parameters in an efficient manner. The calibrated model also provided a good fit to validation targets: non-overdose related deaths, percentage of active OUDs, and all types of overdose counts (fatal and non-fatal). In addition, the resulting set of values for the calibrated parameters will inform the priors of a more comprehensive Bayesian calibration. The calibrated RESPOND model will be employed to improve shared decision-making for OUD.
