Table 3. Multivariable Logistic Regression Models With Dependent Variable of Clinician Willingness to Prescribe MOUD to Adolescents^a.

Variable	Standard model		SLX model
	Estimate	Probability χ2	Estimate	Probability χ2
Sociodemographic characteristics of clinician location and neighboring locations
Total population (every 10 000 people increase)	0.68 (0.50-0.91)	0.01	0.65 (0.49-0.87)	0.004
Total population (W)b	NA	NA	0.79 (0.47-1.35)	0.40
% Black or African American (every percentage increase)	0.99 (0.97-1.01)	0.42	1.01 (0.97-1.06)	0.52
% Black or African American (W)	NA	NA	0.97 (0.92-1.03)	0.38
% With incomes below the federal poverty line (every percentage increase)	1.02 (0.97-1.07)	0.44	1.04 (0.98-1.09)	0.18
% With incomes below the federal poverty line (W)	NA	NA	0.92 (0.84-1.01)	0.09
Rurality (every 1 point increase [1-10])c	1.15 (0.99-1.33)	0.07	1.27 (1.02-1.58)	0.03
Rurality (W)	NA	NA	0.75 (0.54-1.04)	0.09
Clinician characteristics
Clinician credentials
MD/DO	1 [Reference]	NA	1 [Reference]	NA
APC	0.62 (0.38-0.99)	0.04	0.58 (0.35-0.97)	0.04
Clinician training
Family medicine	1 [Reference]	NA	1 [Reference]	NA
Addiction medicine or psychiatry	1.07 (0.62-1.86)	0.81	1.13 (0.60-2.10)	0.71
Otherd	0.41 (0.20-0.83)	0.01	0.40 (0.18-0.89)	0.03

Abbreviations: APC, advanced practice clinician (ie, nurse practitioner, physician’s assistant); DO, doctor of osteopathic medicine; MD, medical doctor; NA, not applicable.

^a Both the standard model (QIC statistic, 470) and the SLX model (QIC statistic, 450) account for clustering of clinicians within zip code tabulation areas.

^b W, spatially lagged variable, generated by spatial weight matrices.

^c Rurality is described with Rural Urban Commuting Area, version 2 (RUCA2) codes, an index from 1-10 in which higher scores indicate a location is more rural.²⁹

^d Other, clinician had training in any category other than addiction medicine, psychiatry, or family medicine.