A System Dynamics Study to Identify Leverage Points to Sustainably Reduce Illicit Drug Use Disorder

Background and Purpose:

Illicit drug use disorder (DUD) is a persistent problem, and treatment receipt and perceived need are both low. Policymakers and researchers have devoted considerable effort to reducing this treatment gap by developing policies that address accessibility, affordability, and quality. However, perceived need, affordability, accessibility, and quality of treatment are not independent of each other. Dynamic simulation modeling using system dynamics (SD) can provide insight into how changes in one part of a system can lead to effects in another part of the system. Understanding these effects is important for developing sustainable policy. 

The objectives of this study are to: 1) Build an SD model of adult (18+) illicit non-marijuana DUD, treatment seeking, and treatment receipt from 2002-2014; 2) Assess the sensitivity of the model to changes in key parameters; 3) Identify potential places to intervene to sustainably reduce the prevalence of illicit non-marijuana DUD.

Methods:Multiple data sources were used to parameterize the model. The primary data source was the annual National Survey on Drug Use and Health (NSDUH). Extant literature was used to estimate parameters related to incidence, probability of remission, and time to remission. Meta-analytic reviews of clinical trials and research on implementation of evidence-based practice were used to estimate quality, which in the model influenced acceptability, accessibility, and affordability. The ratio of persons receiving services relative to capacity influenced quality as well as efforts to increase quality.

The model results were compared to NSDUH data from 2003-2014. Model experimentation extended the model time horizon to 2030. Current estimated drug-related mortality rates were held constant. 

Results:

The model was able to reproduce DUD, treatment receipt, and acceptability (perceived need and treatmetn seeking) patterns with reasonable accuracy (2014 numbers were within 10 percentage points). However, levels of accessibility and affordability in the model that produced these results are lower than reported in NSDUH data.

Model experimentation focused on reducing DUD prevalence and the proportion of all persons with DUD getting quality care, holding constant incidence and mortality rates. Even at 100% affordability by 2030, to reflect policy changes, there was little appreciable effect on DUD, though the proportion seeking services increased. Sensitivity tests on quality (the average treatment effect size, the proportion of addiction professionals providing EBP, and reducing the time to implement EBPs) resulted in at most 50% of persons with DUD getting quality care by 2030, but no change in DUD prevalence. Allowing quality to have a larger effect on treatment outcomes, reducing time to treatment, and increasing the lifetime proportion who remit all had negligible effect. The only parameters that significantly decreased illicit DUD prevalence were the average time to remission, whether among those who never had treatment, were waiting for treatment, or were in treatment.

Conclusions and Implications:

Increasing effectiveness and implementation of treatments, and addressing affordability and accessibility barriers will have little effect on the overall prevalence of DUD if the time to remit is not decreased.