Table 1.

Predictors of Response to Treatment: P Value

	P value
Baseline	SAP Model 1	MDI Model 1	SAP Model 2	MDI Model 2
A1C	0.008	0.003	0.002	0.02
Documented SMBG/day	NA	NA	0.96	0.68
BMI	NA	NA	0.80	0.39
Age at randomization	0.01	NA	0.003	0.24
Duration of diabetes	NA	NA	0.37	0.33
Age at diagnosis	NA	NA	0.048	0.86
Smoking (yes/no)	NA	NA	0.87	0.44
Alcohol use (yes/no)	NA	NA	0.69	0.58
Sex (M/F)	NA	NA	0.50	0.60
SD during 1-week screening	NA	NA	0.96	0.22
Mean sensor glucose value	NA	0.01	0.88	0.10

Model 1 is a multivariate logistic regression model using backward selection keeping those variables with P < 0.05 that detects whether the characteristic predicts a −0.5% change in A1C without severe hypoglycemia. Model 2 is a univariate logistic regression model that detects whether the individual characteristic predicts a −0.5% change in A1C without severe hypoglycemia. P values represent correlation between baseline characteristic and 1 year −0.5% change in A1C without severe hypoglycemia.

BMI, body mass index; MDI, multiple daily injection; NA, not available (dropped from logistic backward selection procedure); SAP, sensor-augmented pump; SMBG, self-monitoring of blood glucose.