Skip to main content
. Author manuscript; available in PMC: 2019 Aug 21.
Published in final edited form as: Psychiatr Serv. 2017 Dec 15;69(4):448–455. doi: 10.1176/appi.ps.201700196

Table 2:

Results of Logistic Regression Models Predicting Adoption of Opioid Use Disorder Medications

Oral Naltrexone (n=383) Injectable Naltrexone (n=383) Buprenorphine (n=397)
AOR 95% CI p-value AOR 95% CI p-value AOR 95% CI p-value
State Policy
SSA block grant funding for medication 3.14 1.49-6.60 .004 1.44 .66-3.16 .350 .54 .27-1.11 .093
SSA subsidizes buprenorphine with state funds -- -- -- -- 2.47 1.31-4.67 .006
SSA level of technical support 1.13 .88-1.46 .332 1.25 .94-1.65 .115 1.18 1.00-1.39 .049
Control Variables
Organizational Characteristics
Program ownership
 Private non-profit (reference: private for-profit) 3.59 1.40-9.20 .009 1.17 .34-4.06 .789 2.56 1.09-6.02 .032
 Public ownership (reference: private for-profit) 3.07 .74-12.69 .117 .60 .10-.71 .573 2.87 .88-9.36 .078
Program type
 Inpatient/Residential (reference: outpatient) 3.14 1.07-9.22 .038 2.22 1.01-4.91 .049 1.51 .87-2.63 .138
Accredited by JC/CARF 1.41 .60-3.32 .424 1.58 .51-4.93 .420 1.16 .49-2.76 .732
Program size (number of clients served, log) 1.24 .91-1.69 .176 1.06 .74-1.50 .759 1.53 1.21-1.94 .001
Staff professionalism .53 .11-2.65 .428 3.74 .65-21.52 .135 1.18 .27-5.09 .817
% private insurance revenues 1.02 .99-1.04 .167 1.01 .99-1.04 .274 1.01 .99-1.03 .171
Client socio-demographic characteristics
 % Black .98 .96-1.01 .135 .99 .97-1.01 .343 .99 .98-.00 .183
 % Hispanic .99 .96-1.01 .394 1.00 .97-1.03 .854 1.00 .97-1.03 .893
 % women 1.01 .99-1.02 .259 1.00 .98-1.03 .692 1.00 .99-1.01 .903
Market Factors
Perceived increase in competition 2.33 1.34-4.03 .004 .54 .28-1.04 .064 1.08 .51-2.31 .829
% heroin clients 1.00 .98-1.02 .661 1.02 1.01-1.04 .013 1.01 .99-1.02 .057
% prescription opioid clients 1.00 .98-1.03 .689 1.00 .97-1.02 .784 1.01 1.00-1.02 .146

Note: The number of programs and states included in multivariate analyses varies by medication, based on the amount of missing data.