. Author manuscript; available in PMC: 2011 Dec 1.

Published in final edited form as: Med Care Res Rev. 2010 Jun 16;67(6):657–675. doi: 10.1177/1077558710369911

Table 3.

Cox Regressions Predicting Time until Final Loss of Coverage for Males and Females

	Females		Males
	Hazard Ratio	p	Hazard Ratio	p

Race^a
White	1.771	<.001	1.048	.356
African American	1.066	.373	1.031	.540
Rural County^b	.949	.448	.998	.964
Diagnosis^b
Major Depressive	1.049	.617	1.046	.448
Bipolar	1.143	.392	.884	.190
Schizophrenia	.729	.093	.524	<.001
Attention Deficit/Hyperactivity Disorder	1.363	.015	1.236	<.001
Conduct Disorder/Oppositional Defiant Disorder	1.093	.356	1.095	.093
Substance Use	1.224	.072	1.060	.323
Mental Retardation/Dev. Disability	.511	<.001	.377	<.001
Other only	1.024	.830	1.031	.649
Eligibility category^b
Foster care	.894	.314	.583	<.001
Supplemental Security Income	.372	<.001	.152	<.001
Pregnancy, <185% of poverty	.634	<.001	-	-
Med. Assistance to Families	1.101	.189	.803	.003
Aid to Families with Dependent Children/Temporary Assistance for Needy Families	.641	<.001	1.032	.658
Child under 19	.971	.750	.359	<.001

Model Statistics

χ² statistic		250		2168
df		17		16
Model significance		<.001		<.001
−2 Log Likelihood		15668		36839

Race included as a deviation contrast, with each category compared to the overall effect of race including “other” and “missing” race

Included as indicator contrast, with their respective absence as indicator category