Table 3.

Risk-Adjusted Rates of 30-Day Readmissions and Emergency Department Visits After Hospital Discharge, by Homeless Status and Site of Care

		Model 1*			Model 2†
	No. of patients	Adjusted rate, % (95% CI)	Adjusted OR (95% CI)	p value	Adjusted rate, % (95% CI)	Adjusted OR (95% CI)	p value
Readmission rate
Homeless status
Homeless	134,755	27.3 (25.0, 29.7)	1.93 (1.69, 2.21)	< 0.001	28.2 (26.3, 30.1)	2.06 (1.84, 2.31)	< 0.001
Non-homeless	3,392,628	17.5 (17.2, 17.9)	Reference		17.5 (17.3, 17.8)	Reference
Site of care
Homeless-serving hospitals	377,654	17.9 (16.6, 19.1)	0.99 (0.89, 1.09)	0.83	17.8 (16.9, 18.6)	0.98 (0.91, 1.06)	0.63
Non-homeless-serving hospitals	3,149,729	18.0 (17.7, 18.3)	Reference		18.1 (17.7, 18.3)	Reference
Emergency department visit rate
Homeless status
Homeless	134,755	37.1 (34.3, 39.8)	1.98 (1.74, 2.25)	< 0.001	37.7 (35.1, 40.2)	2.04 (1.82, 2.30)	< 0.001
Non-homeless	3,392,628	23.6 (23.3, 23.9)	Reference		23.6 (23.3, 23.9)	Reference
Site of care
Homeless-serving hospitals	377,654	22.9 (21.9, 23.9)	0.92 (0.86, 0.98)	0.01	23.3 (22.3, 24.3)	0.94 (0.88, 1.00)	0.07
Non-homeless-serving hospitals	3,149,729	24.4 (24.0, 24.7)	Reference		24.3 (24.0, 24.6)	Reference

OR odds ratio, CI confidence interval

We used multivariable logistic regression models. Standard errors were clustered at the hospital level to account for the potential correlation of patients treated at the same hospital. ORs of all-cause 30-day readmissions and emergency department visits are shown for each model. Adjusted rates were calculated using marginal standardization for each model

*Model 1 adjusted for patients’ characteristics (the primary diagnosis for the admission, age, sex, race/ethnicity, expected primary payer, and patient comorbidities), state dummies, and quarter dummies

†Model 2 additionally adjusted for discharge disposition, length of stay, and hospital characteristics