Table 4:

Multivariable comparison of treatment outcomes

n = 2535	Ratio (95% CI)^*	P^*	Power (%)
Neurosurgical/Endovascular
Adverse outcome	1.7 (1.2, 2.5)	<.05	82
In-hospital death	2.3 (0.97, 5.7)	.06	45
Length of stay^†	1.8 (1.6, 2.0)	<.0001	100
Hospital charges^†	1.3 (1.2, 1.5)	<.0001	97
Emergency department admissions only
Adverse outcome^‡	2.7 (0.8, 8.5)	.10	39
In-hospital death^§
Length of stay^†	1.3 (0.9, 1.7)	.12	43
Hospital charges^†	1.2 (0.9, 1.7)	.24	19
Excluding emergency department admissions
Adverse outcome	1.6 (1.1, 2.4)	<.05	71
In-hospital death	2.0 (0.7, 5.8)	.18	26
Length of stay^†	1.9 (1.7, 2.0)	<.0001	100
Hospital charges^†	1.3 (1.2, 1.5)	<.0001	97

* Results are derived from generalized estimating equations and are adjusted for age, sex, admission source, region, year of treatment, insurance status, comorbidity score, and hospital treatment volume. Odds ratios are reported for dichotomous outcomes. Significant P values are grouped into 4 categories: <.05, <.01, <.001, <.0001.
† Natural log values of lengths of stay and hospital charges were used in models; results are ratio of geometric mean days and ratio of geometric mean dollars, respectively.
‡ Results are derived from generalized estimating equations and are adjusted for age, sex, admission source, region, insurance status, and hospital treatment volume.
§ Multivariable model not feasible due to zero deaths in the endovascular group.