Abstract
Background:
Epidemiologic studies commonly recommend the integration of harm reduction programs with health and social services to improve the well-being of persons who inject drugs (PWIDs). This study identified service utilization clusters for PWIDs attending a syringe exchange program (SEP) in 2017 to better understand in-house service usage.
Methods:
We applied Multiple Correspondence Analysis and Hierarchical Clustering on Principal Components to classify 475 PWIDs into clusters using anonymized, SEP records data from New York. Multinomial logistic regression was used to identify sociodemographic and program engagement correlates of cluster membership.
Results:
Only 22% of participants utilized at least one service. We identified three clusters of service utilization defined by 1) Nonuse; 2) Support, Primary Care, & Maintenance service use; and 3) HIV/STD, Support, Primary Care, & Maintenance service use. Cluster 2 members were less likely to be living alone compared to Cluster 1 (AOR = 0.08, 95% CI: 0.04, 0.17) while Cluster 3 members were less likely to be White (AOR = 0.19, 95% CI: 0.07, 0.50) or living alone (AOR = 0.16, 95% CI: 0.06, 0.44) and more likely to be Medicaid recipients (AOR = 2.89, 95% CI: 1.01, 8.36) compared to Cluster 1. Greater than one SEP interaction, lower syringe return ratios, and being a long-term client increased the odds of service utilization.
Discussion:
Overall, PWID clients had a low prevalence of in-house service use particularly those who live alone. However, higher service utilization was observed among more vulnerable populations (i.e., non-White and LGBT). Future research is needed to profile services used outside of the SEP.
Disclosure statement
No potential conflict of interest was reported by the authors.
Data availability statement
Due to the nature of this research, data is confidential and cannot be shared publicly.
Notes
1 Over-expression refers to variables where the frequency (or mean) for the indicated cluster has a statistically significant positive variation from the overall sample frequency (or mean). Under-expression refers to variables where the frequency (or mean) for the indicated cluster has a statistically significant negative variation from the overall sample frequency (or mean) (Lê et al., Citation2008).