Scaling up overdose education and naloxone distribution in Kentucky: adoption and reach achieved through a “hub with many spokes” model

Table 2 Negative binomial regression model of naloxone reach

Variable	Incidence rate ratio coefficient (95% CI)	p-value
Rural location (vs. urban location)	0.82 (0.50, 1.36)	0.44
Organizational Type
Medication for opioid use disorder (MOUD) clinics	Reference
Non-MOUD addiction treatment and recovery services	1.41 (0.76, 2.61)	0.27
Healthcare organizations	0.43 (0.20, 0.90)	0.03
Social services (e.g., Department of Community-Based Services, homeless shelters)	0.83 (0.37, 1.85)	0.64
Jails	2.13 (0.77, 5.89)	0.14
Drug courts and private alternatives to incarceration	0.57 (0.21, 1.56)	0.27
Emergency response (ambulance or fire)	0.60 (0.20, 1.80)	0.37
Health departments and syringe service programs embedded in health departments	4.52 (1.70, 12.04)	0.003
Months of implementation	1.15 (1.11, 1.20)	< 0.001
Constant	17.42 (7.43, 40.85)
Alpha	1.61 (1.31, 1.99)

The likelihood ratio test for the alpha statistic was significant (p < 0.001), indicating that negative binomial regression was more appropriate than Poisson regression. Healthcare organizations included outpatient clinics not providing MOUD, hospitals, and dental clinics; these dental clinics were owned by a healthcare organization that largely provides primary care

ISSN: 1940-0640