Table 2. . Study outcomes by blood group.

Outcomes	A (n = 143)		AB (n = 18)		B (n = 45)		O (n = 267)		Main effect	p-value (A is reference)			Adj‡ p-value
	n (%)	OR† (95% CI)	n (%)	OR† (95% CI)	n (%)	OR† (95% CI)	n (%)	OR† (95% CI)	p-value†	AB	B	O
ICU	61 (42.7%)	Ref	9 (50%)	1.34 (0.5–3.6)	15 (33.3%)	0.67 (0.3–1.3)	122 (45.7%)	1.13 (0.8–1.7)	0.424				0.643
Intubation	47 (32.9%)	Ref	7 (38.9%)	1.3 (0.5–3.5)	10 (22.2%)	0.58 (0.3–1.2)	85 (31.8%)	0.95 (0.6–1.5)	0.477				0.590
Death	34 (23.8%)	Ref	7 (38.9%)	2.04 (0.7–5.6)	4 (8.9%)	0.31 (0.1–0.9)	55 (20.6%)	0.83 (0.5–1.4)	0.037	0.172	0.038	0.457	0.197

	Median§ (IQR)	HR¶ (95% CI)	Median§ (IQR)	HR¶ (95% CI)	Median§ (IQR)	HR¶ (95% CI)	Median§ (IQR)	HR¶ (95% CI)	LOS§p-value	Surv¶p-value	Adj#p-value
LOS#	9 (5.0–17.0)	Ref	9 (5.3–15.3)	0.66 (0.4–1.2)	8 (4.0–17.0)	1.42 (1.0–2.0)	10 (6.0–21.0)	0.95 (0.8–1.2)	0.139	0.0695	0.1446

^†OR and 95% CI calculated from independent, univariable logistic regression to identify effects of blood type on ICU admissions, intubation and death.

^‡Adjusted p-values are from tests of conditional associations of blood type on each outcome in a multivariable logistic regression model including Hispanic ethnicity, age and CKD.

^§Median LOS was compared using a Kruskal–Wallis test.

^¶HR and 95% CI’s result from a univariable Cox proportional hazards model with blood type as an independent predictor of survival to discharge, censoring on in-hospital death, with p-value indicated.

^#The adjusted p-value is from a multivariable Cox proportional hazards model to test conditional associations of blood type accounting for Hispanic ethnicity, age and CKD.

CKD: Chronic kidney disease; HR: Hazard ratio; ICU: Intensive care unit; IQR: Interquartile range; LOS: Length of hospital stay; OR: Odds ratio.