Using machine learning to advance disparities research: Subgroup analyses of access to opioid treatment

Abstract

Objective: To operationalize an intersectionality framework using a novel statistical approach and with these efforts, improve the estimation of disparities in access (i.e., wait time to treatment entry) to opioid use disorder (OUD) treatment beyond race.

Data source: Sample of 941,286 treatment episodes collected in 2015-2017 in the United States from the Treatment Episodes Data Survey (TEDS-A) and a subset from California (n = 188,637) and Maryland (n = 184,276), states with the largest sample of episodes.

Study design: This retrospective subgroup analysis used a two-step approach called virtual twins. In Step 1, we trained a classification model that gives the probability of waiting (1 day or more). In Step 2, we identified subgroups with a higher probability of differences due to race. We tested three classification models for Step 1 and identified the model with the best estimation.

Data collection: Client data were collected by states during personal interviews at admission and discharge.

Principal findings: Random forest was the most accurate model for the first step of subgroup analysis. We found large variation across states in racial disparities. Stratified analysis of two states with the largest samples showed critical factors that augmented disparities beyond race. In California, factors such as service setting, referral source, and homelessness defined the subgroup most vulnerable to racial disparities. In Maryland, service setting, prior episodes, receipt of medication-assisted opioid treatment, and primary drug use frequency augmented disparities beyond race. The identified subgroups had significantly larger racial disparities.

Conclusions: The methodology used in this study enabled a nuanced understanding of the complexities in disparities research. We found state and service factors that intersected with race and augmented disparities in wait time. Findings can help decision makers target modifiable factors that make subgroups vulnerable to waiting longer to enter treatment.

Keywords: racial disparities; regression tree; subgroup analysis; virtual twins; wait time for opioid treatment.

FIGURE 1

Flowchart of the two‐step subgroup analysis method virtual twins [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 2

Regression tree for the national data. Left branch indicates that the condition in the splitting node is met or satisfied. The decimal number in the node shows the increased or decreased probability of waiting 1 day or more due to race. The percent value shows the percentage of episodes falling into that node. Nodes with high positive decimal numbers include episodes more subject to racial disparities [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3

Regression tree for the subgroup analysis of California episodes. Left branch indicates that the condition in the splitting node is met or satisfied. The decimal number shows the increased or decreased probability of waiting 1 day or more due to race for that subgroup. The percent value shows the percentage of episodes falling into that node. The most vulnerable subgroup is enclosed in the red circle [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 4

Regression tree for the subgroup analysis of Maryland episodes. Left branch indicates that the condition in the splitting node is met or satisfied. The decimal number shows the increased/decreased probability of waiting 1 day or more due to race for that subgroup. The percent shows the percentage of episodes falling into that node. The most vulnerable subgroup is enclosed in the green circle [Color figure can be viewed at wileyonlinelibrary.com]