Table 4

Negative Binomial Regression Models of Location of Practice and Physician Characteristics and Association With Buprenorphine Treatment

	Model 1 Patients in Past Year With Heroin Use Disorder IRR (95% CI)	Model 2 Patients in Past Year With Prescription Opioid Use Disorder IRR (95% CI)	Model 3 Patients in Past Year Who Received Counseling From Clinicians in the Respondent’s Practice IRR (95% CI)	Model 4 Patients in Past Year Who Received No Counseling IRR (95% CI)
Rural-urban continuum
Large metropolitan area	Reference	Reference	Reference	Reference
Small metropolitan area	0.753 (0.651, 0.869)^a	1.076 (1.000, 1.158)^c	1.219 (1.021, 1.455)^c	0.732 (0.570, 0.942)^c
Nonmetropolitan area	0.641 (0.506, 0.812)^a	1.227 (1.118, 1.347)^a	1.094 (0.886, 1.351)	0.675 (0.447, 1.020)
Medical specialty
Primary care	Reference	Reference	Reference	Reference
Addiction or psychiatry	1.113 (0.965, 1.284)	0.917 (0.872, 0.966)^b	1.927 (1.591, 2.333)^a	0.383 (0.308, 0.475)^a
Other	0.804 (0.587, 1.100)	1.117 (1.000, 1.248)	1.177 (0.895, 1.548)	0.933 (0.755, 1.152)
Practice
Individual medical practice	0.879 (0.767, 1.007)	1.105 (1.020, 1.197)^c	0.486 (0.418, 0.566)^a	1.045 (0.819, 1.332)
Other	Reference	Reference	Reference	Reference
Years prescribing buprenorphine	0.989 (0.973, 1.005)	1.006 (0.998, 1.015)	0.985 (0.968, 1.003)	1.003 (0.978, 1.029)
Waiver type
30-patient limit	Reference	Reference	Reference	Reference
100-patient limit	1.206 (1.037, 1.402)^c	0.939 (0.887, 0.993)^c	1.094 (0.971, 1.232)	1.218 (0.968, 1.534)
Payment type for office visits
Cash only	Reference	Reference	Reference	Reference
Private insurance but not Medicaid	0.837 (0.682, 1.028)	1.115 (1.023, 1.215)^c	1.114 (0.799, 1.553)	0.960 (0.720, 1.279)

Medicaid (with or without private insurance)	1.246 (1.045, 1.484)^c	0.873 (0.803, 0.950)^b	1.409 (0.981, 2.024)	0.998 (0.751, 1.326)
Other	0.954 (0.696, 1.306)	0.940 (0.799, 1.104)	1.330 (0.956, 1.851)	1.394 (0.713, 2.724)
Age	0.999 (0.994, 1.004)	1.001 (0.999, 1.003)	0.998 (0.993, 1.003)	0.988 (0.977, 0.999)^c
Sex
Male	Reference	Reference	Reference	Reference
Female	1.099 (0.969, 1.246)	0.978 (0.916, 1.044)	0.992 (0.853, 1.154)	0.752 (0.556, 1.016)
Race
White	Reference	Reference	Reference	Reference
Asian	0.898 (0.767, 1.051)	1.026 (0.969, 1.087)	0.893 (0.716, 1.113)	1.119 (0.848, 1.477)
Other	1.107 (0.872, 1.406)	0.976 (0.877, 1.086)	0.770 (0.615, 0.964)c	0.894 (0.630, 1.268)
Census division
New England	Reference	Reference	Reference	Reference
Middle Atlantic	1.057 (0.796, 1.403)	1.024 (0.850, 1.233)	0.910 (0.706, 1.173)	0.937 (0.662, 1.327)
East North Central	0.943 (0.684, 1.299)	1.160 (0.910, 1.478)	1.074 (0.849, 1.359)	0.852 (0.477, 1.521)
West North Central	0.668 (0.492, 0.908)^c	1.400 (1.156, 1.695)^a	0.900 (0.694, 1.167)	0.891 (0.546, 1.452)
South Atlantic	0.683 (0.476, 0.979)^c	1.313 (1.035, 1.667)^c	1.082 (0.883, 1.326)	1.359 (0.879, 2.101)
East South Central	0.580 (0.332, 1.015)	1.429 (1.107, 1.845)^b	1.365 (0.975, 1.913)	1.057 (0.488, 2.288)
West South Central	0.592 (0.435, 0.807)^b	1.527 (1.278, 1.826)^a	1.237 (0.803, 1.904)	1.517 (1.037, 2.219)^c
Mountain	0.918 (0.633, 1.332)	1.212 (0.992, 1.482)	0.923 (0.729, 1.168)	1.050 (0.672, 1.640)
Pacific	0.993 (0.705, 1.400)	1.168 (0.963, 1.415)	0.923 (0.773, 1.102)	1.380 (0.961, 1.981)
Constant	29.374 (20.616, 41.853)	41.332 (33.698, 50.696)	30.001 (18.583, 48.433)	30.427 (12.756, 72.573)

IRR = incidence rate ratio.
Note: These models present adjusted analyses in which locations of practice remained significant after controlling for other physician characteristics (ie, variables in Table 1). Models represent the pooled estimates from 20 imputed data sets (n = 1,174). Robust standard errors were used to adjust for physician clustering within states.
↵a P = <.001 (2-tailed tests).
↵b P = <.01 (2-tailed tests).
↵c P = <.05 (2-tailed tests).