Opioid Use and Out of Pocket Spending on Opioids among US Adults Aged 18+ by Serious Psychological Distress Status

Tuesday, June 12, 2018
Lullwater Ballroom - Garden Level (Emory Conference Center Hotel)

Presenter: Priscilla Novak

Co-Authors: Jie Chen; Sandra Chao


Opioid use disorder is a growing concern in the United States. Furthermore, serious psychological distress (SPD) has been shown to be correlated with higher health care utilization. The extent to which SPD may be a driver of opioid utilization has been underexplored.


To examine the distribution of the number of opioid scripts in a large, nationally representative dataset; to calculate mean number of opioid scripts filled among US adults by SPD status; and, to fit a model to predict number of scripts filled, out-of-pocket expenditures for opioid medications in 2015.


Using data from the Medical Expenditure Panel Survey, we examined utilization and out-of-pocket expenditures on opioids among people by SPD status, as indicated by a Kessler-6 score of 13 or greater. The Andersen model was used to select co-variates; survey weights were used.


The range of script numbers in 2015 varied from 1 (45% of those who had at least 1 script received only one script in the entire calendar year) to 108 scripts. In an attempt to understand higher utilization behavior, we divided the sample into 2 groups: those who had received 12 or fewer scripts (96.26% of observations) and those who had received >=12 scripts (3.74% of observations). People without SPD filled an average of 3.62 opioid scripts (95% CI 3.43 - 3.94), while those with SPD filled an average of 5.40 scripts (95% CI 4.45-6.95). A t-test for unequal distributions in the number of scripts by SPD status was significant (p<0.001). Mean OOP expenditures among people without SPD were $67.37; mean OOP among people with SPD was $92.25.

We fit a negative binomial model to predict number of scripts filled; African-Americans, Asians and Hispanics filled fewer scripts compared with non-Hispanic Whites. Medicare beneficiaries and those over 65 who were uninsured filled fewer opioid scripts compared with the privately insured. Decreases in self-rating of health were associated with higher number of scripts filled. In our sensitivity analysis, we calculated sub-domains of the Andersen model; when only predisposing, or predisposing and enabling characteristics where included, SPD was statistically significant as a driver of greater number of scripts (p<0.01). When needs characteristics were added, SPD became attenuated to statistical nonsignificance.

In a GLM model, OOP expenditures of opioids were lower for African Americans, Native Americans, Hispanics, and non-Hispanic people of mixed race. Compared with those who completed the survey entirely in English, those who responded to the survey in Spanish had lower OOP. Poor self-rated health was associated with a statistically significant increase in OOP. The association between SPD and OOP was not significant after controlling for clinical needs factors.


Results show that people with SPDs used more scripts of opioid and encountered higher opioid costs. However, the difference was mainly attributed to the presence of co-existing chronic conditions and other clinical needs. SPD alone is not a significant driver of opioid consumption. Findings also suggest that race/ethnicity, along with poor physical health, are important factors in the use of opioids and OOP expenditures.