Table 2.

Multivariate Regression Models^* Relating Intensity of Antiglycemic Medication Therapy to Glycemic Control

Change in HbA1c		Outcome HbA1c>8%
Variable	Coefficient	p-Value	Odds Ratio	95% CI
Intensity of therapy (per increase of 0.1)	−0.064	<.0001	0.91	0.88, 0.98
Index visit HbA1c	—	—	1.76	1.72, 1.81
Age (per increase of 10 years)	0.013	.39	0.78	0.75, 0.81
Male gender	0.147	.13	1.12	0.86, 1.45
Any diabetes complication	−0.131	<.0001	1.29	1.18, 1.41
Charlson index	0.026	.01	1.02	0.99, 1.05
Psychiatric diagnosis	0.156	.0002	1.03	0.92, 1.15
Alcohol/substance abuse diagnosis	−0.169	.06	0.79	0.62, 1.00
Oral steroid use	−0.022	.75	0.83	0.69, 1.00

^*A linear model was used when modeling outcome visit HbA1c minus index visit HbA1c; a negative coefficient indicates improved glycemic control. A logistic model was used to predict odds of an outcome HbA1c>8%; an odds ratio less than 1.0 indicates better control

HbA1c, hemoglobin A1c; CI, confidence interval