Table 2.

Multilevel logistic regression models estimating odds of non-adherence to biguanide medication amongst Type 2 Diabetes patients enrolled in Medicaid over two years (2017–2018).

		‘Null’ model (n = 24,387)	Model 1 (n = 24,387)	Model 2 (n = 24,387)	Model 3 (n = 24,387)
		OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Individual-level variables
Sex	Male		1.01 (0.96–1.06)	1.02 (0.96–1.07)	1.02 (0.96–1.07)
Age	18–34		Referent	Referent	Referent
	35–49		0.58** (0.52–0.65)	0.58**(0.52–0.65)	0.58**(0.52–0.65)
	50–64		0.33** (0.30–0.37)	0.33**(0.30–0.37)	0.34**(0.30–0.37)
	≥65		0.29** (0.24–0.35)	0.28**(0.24–0.34)	0.28**(0.24–0.34)
Comorbidity a	0		Referent	Referent	Referent
	1		0.69** (0.64–0.75)	0.69**(0.63–0.75)	0.69**(0.63–0.74)
	2		0.68** (0.61–0.75)	0.68**(0.61–0.75)	0.68**(0.61–0.75)
	3		0.76** (0.67–0.87)	0.76**(0.67–0.87)	0.76**(0.67–0.87)
Contextual-level variables
Rural-Urban (RUCA) b	Urban core			Referent	Referent
	Suburban			0.90 (0.80–1.01)	0.90 (0.80–1.01)
	Large rural			0.78**(0.72–0.84)	0.78**(0.72–0.84)
	Small town/rural			0.75**(0.67–0.85)	0.75**(0.66–0.85)
Social Vulnerability Index (quintiles)	Q1 (Least)			Referent	Referent
	Q2			1.29**(1.08–1.53)	1.29**(1.08–1.53)
	Q3			1.34**(1.14–1.68)	1.34**(1.14–1.58)
	Q4			1.46**(1.24–1.72)	1.47**(1.25–1.73)
	Q5 (Greatest)			1.66**(1.42–1.95)	1.67**(1.42–1.95)
Distance to utilised pharmacy (km)	<1				Referent
	1–2.9				1.07 (0.98–1.16)
	3–4.9				1.06 (0.97–1.15)
	5–9.9				1.05 (0.96–1.14)
	>10				1.05 (0.96–1.15)
ICCb tract		0.0224	0.0249	0.0170	0.0167
AICb		33713.34	32777.19	32671.29	32676.56
BICb		33729.55	32850.11	32800.92	32838.59

*p value < 0.05, **p value < 0.01 OR Odds Ratio, CI Confidence Interval^.

^aCount of comorbid conditions including hypertension, coronary heart disease and coronary heart failure.

^bRUCA Rural-Urban Commuting Area codes, ICC intra-level correlation coefficient, AIC Akaike Information Criterion, BIC Bayesian Information Criteria.