Albumin infusion in hospitalized patients with acute heart failure: a retrospective cohort study

Table 2 Associations between the use of albumin and composite end-point in the propensity-score, multivariable, and crude analyses

Analysis	Composite end point
No. of events/no. of patients at risk (%)
Albumin	149 /337(44.2)
No albumin	208/701 (29.7)
Crude analysis—hazard ratio (95% CI)	1.43(1.16–1.76)
Multivariable analysis—hazard ratio (95% CI) *	1.00(0.75–1.32)
Propensity-score analyses—hazard ratio (95% CI)
With matching**	1.11 (0.64–1.57)
With inverse probability weighting***	1.05 (0.75–1.47)

^*Shown is the hazard ratio from the multivariate Cox proportional-hazards model adjusted for past diagnoses, sex, clinical signs and symptoms, laboratory tests, and medications. All 1038 participants were included in the analysis
^** Shown is the hazard ratio from a multivariate Cox proportional-hazards model with identical strata and covariates matching on the basis of the propensity score. 674 patients were involved in the analysis (337 were treated with albumin and 337 were not)
^***Shown is the primary analysis with a hazard ratio from the IPW multivariate Cox proportional hazards model with identical strata and covariates on the basis of the propensity score. All patients were involved in the analysis

ISSN: 1471-2261