Table 3.

Logistic regression AUC and AIC values: Palmetto SADI versus four alternative small-area deprivation measures

Model	At least one chronic condition				Two or more chronic conditions
	AUC	p	AIC	p	AUC	p	AIC	p
Palmetto SADI	0.5741		4080.450		0.5716		2192.305
Townsend	0.5472	0.0000	4101.994	0.0000	0.5416	0.0000	2200.699	0.0000
Poverty	0.5563	0.0000	4095.502	0.0000	0.5528	0.0000	2198.326	0.0000
Primary Care HPSA	0.5440	0.0000	4095.045	0.0000	0.5419	0.0000	2197.537	0.0000
MUA/MUP	0.5389	0.0000	4101.473	0.0000	0.5299	0.0000	2202.166	0.0000

The table shows omnibus statistics for single predictor logistic regression models of two chronic disease burden outcome indicators representing a random sample of 5,000 South Carolina Medicaid recipients (FY2012). Area under the Receiver Operating Characteristic curve (AUC) values close to 1 show near perfect discrimination. Corrected Akaike information criterion measure (AIC) values indicate the model’s deviance from a perfectly predicting model. Lower values of AIC indicate a preferable model. Tabled p values reflect the probability that AUC and AIC values associated with each of the four alternative deprivation models do not differ statistically from the Palmetto SADI model