. Author manuscript; available in PMC: 2010 Jun 4.

Published in final edited form as: Arch Phys Med Rehabil. 2008 Oct;89(10):1863–1872. doi: 10.1016/j.apmr.2008.03.013

Table 3.

Illustration of Balance for Key Clinical Characteristics Before and After the Application of Propensity Score Quintiles

Covariate	Before Propensity Adjustment OR (95% CI)	After Propensity Adjustment OR (95% CI)
Admission from home vs extended care	1.05 (1.03–1.08), P=.001	1.00 (0.98–1.03), P=.85
Admission from hospital vs extended care	1.02 (0.99–1.05), P=.15	1.00 (0.98–1.03), P=.85
Transtibial vs transfemoral	1.06 (1.03–1.08), P<.001	1.00 (0.98–1.03), P=.72
Paralysis	0.92 (0.87–0.98), P=.005	0.99 (0.93–1.05), P=.73
Systemic sepsis	0.92 (0.89–0.96), P<.001	0.99 (0.95–1.04), P=.78
Hypertension	1.07 (1.04–1.10), P<.001	1.00 (0.97–1.03), P=.88
ICU admission	0.93 (0.90–0.95), P<.001	1.00 (0.97–1.03), P=.88

NOTE. Values are ORs and CIs. ORs, CIs, and P values associated with treatment group in models predicting key baseline clinical characteristics before and after adjustment for the 5 propensity score quintiles. The treatment variable is structured such that not receiving inpatient rehabilitation forms the reference group.

Abbreviation: ICU, intensive care unit.