Table 1. Characteristics of Sample Groups.

		Prior to Propensity Class Implementation				After Propensity Class Implementation

		% with Characteristic				p-value Within Propensity Classes¹

		Enrollee Insurance Status

		New N=1124	Returning N=1587	Continuous N=5126	p-value X²	1 N=1961	2 N=1689	3 N=1233	4 N=1239	5 N=1715
Sex: Female		47.5	53.0	60.0	<0.001	0.09	0.73	0.26	0.02	0.09

Age Category	Less than 30	15.3	14.6	14.5	<0.001	0.12	0.32	0.51	0.86	0.01
	30 to <50	46.3	54.8	45.9
	50 to 65	38.4	30.6	39.6

Race/Ethnicity	Hispanic	4.0	3.8	3.5	0.109	0.34	0.86	0.46	0.89	1.00
	Non-Hispanic White	84.3	84.8	83.0
	Other minority	8.8	9.1	11.1
	Unknown	2.8	2.3	2.4

Child on Case		12.2	17.6	22.5	<0.001	0.19	0.05	0.85	0.45	0.18

Income (%FPL)	less than 10%	53.0	52.2	44.7	<0.001	0.25	0.68	0.83	0.02	0.752
	10% to less than 50%	19.0	18.8	19.8
	50% to less than 65%	7.9	7.5	11.0
	65% to less than 85%	10.1	12.4	13.9
	85% to less than 100%	9.9	9.1	10.6

Rural zipcode^b		42.8	46.5	42.5	0.019	0.09	0.80	0.19	0.29	0.07

Nearest hospital^c	More than 10 miles	18.3	19.3	16.8	0.151	0.23	0.81	0.35	0.11	0.43

Chronic Condition	Hypertension	36.5	35.9	37.4	0.505	0.18	0.18	0.11	0.42	0.41
	Diabetes	13.8	13.7	16.7	0.003	0.99	0.72	0.48	0.86	0.69
	Heart Disease	11.4	12.5	11.5	0.546	0.29	0.47	0.09	0.95	0.65
	Obstructive Pulmonary Disease	17.4	23.3	20.1	0.001	0.83	0.62	0.70	0.57	0.42
	Gastrointestinal disorder	20.8	24.6	20.2	0.001	0.97	0.23	0.06	0.59	0.26
	Anxiety	21.2	25.0	22.0	0.024	0.61	0.11	0.46	0.40	0.75
	Depression	27.8	31.1	25.9	<0.001	0.09	0.28	0.85	0.86	0.94
	Other psychiatric disorder	22.4	26.0	22.3	0.008	0.35	0.78	0.21	0.02	0.56
	Drug dependency	8.4	16.4	10.7	<0.001	0.31	0.81	1.00	0.49	0.03
	Alcohol dependency	12.9	17.3	8.9	<0.001	0.11	0.01	0.21	0.08	<0.01

Bold face indicates residual imbalance (p<.05) within propensity classes.

Propensity classes were created to adjust for the between group difference in covariates. To do this, propensity scores for membership in the newly insured and returning insured groups were generated from a multinomial regression of group membership on patient covariates. Propensity scores were categorized as: 1) low probability of being either newly or returning insured, 2) moderate probability of being either, 3) high probability of being newly insured and low probability of returning insured, 4) low probability of being newly insured and high probability of returning insured, and 5) high probability of being either. Details of the propensity class categorization are given in eMethods in the Supplement. Variables with residual imbalance were included with propensity class as covariates in all analyses.

Based on classification by the Oregon Office of Rural health

Based on the centroid of the patient's zip code

P-value(χ²): χ² test of between group difference