Table 2.

Associations Between HOUSES Quartile and Outcomes Using Unadjusted and Multivariable Regression Models

Outcome	Unadjusted Statistics	Unadjusted Estimate	Unadjusted p-value	Adjusted Estimate†	Adjusted p-value†
Hospital Mortality	N Deaths	Hazard Ratio	0.31	Hazard Ratio	0.10
HOUSES Q1, n=838	87 (10.4%)	1.00 (ref)		1.00 (ref)
HOUSES Q2-Q4, n=2540	234 (9.2%)	0.88 (0.69, 1.12)		0.81 (0.63, 1.04)
ICU-Free Days	Mean (SD)	Mean Estimate	0.67	Mean Estimate	0.10
HOUSES Q1, n=838	23.4 (8.2)	0.00 (ref)		0.00 (ref)
HOUSES Q2-Q4, n=2540	23.5 (7.9)	0.14 (−0.48, 0.76)		0.50 (−0.10, 1.09)
Hospital-Free Days	Mean (SD)	Mean Estimate	0.57	Mean Estimate	0.03
HOUSES Q1, n=838	19.7 (8.8)	0.00 (ref)		0.00 (ref)
HOUSES Q2-Q4, n=2540	19.9 (8.3)	0.19 (−0.47, 0.85)		0.70 (0.08, 1.32)*
ICU Readmission	N Readmissions	Odds Ratio	0.31	Odds Ratio	0.051
HOUSES Q1, n=838	69 (8.2%)	1.00 (ref)		1.00 (ref)
HOUSES Q2-Q4, n=2540	182 (7.2%)	0.86 (0.64, 1.15)		0.74 (0.55, 1.00)

Unadjusted models consisted of just HOUSES as the predictor.

^†Multivariable models were additionally adjusted for the effects of age, sex, race, and APACHE.

Missing data were handled using multiple imputation using 10 independent, imputed datasets.