Table 5.

Independent Risk Factors for Breakthrough Invasive Fungal and Invasive Mold Infections Identified by Multivariate Logistic Regression

	Univariate, P	Multivariate
		P	Odds Ratio (95% CI)
Breakthrough IFI
African-American race	.02	.18
Receipt of >7 RBC units transfusion	.001	.01	7.5 (1.5–36.6)
Mold-positive respiratory culture at transplant (recipient or donor)	.025	.04	7.1 (1.1–44.4)
Breakthrough IMI
Receipt of >7 RBC units transfusion	.004	<.0001	8.6 (2.7–26.9)
Mold-positive respiratory culture at transplant (recipient or donor)	.02	<.0001	18.5 (3.7–91.7)
Bronchial necrosis extending >2 cm from the anastomosis	.03	.008	7.05 (1.7–29.7)
Alemtuzumab induction	.05	.01	0.17 (.04–.68)

An exact logistic regression model, which produces more accurate inference in small numbers of breakthrough IFI and breakthrough IMI, was used to compute P values and odds ratios with 95% CIs. A Bayesian regression simulation using 1000 iterative samples was also used to estimate the odds ratio, and similar results were obtained.

Abbreviations: CI, confidence interval; IFI, invasive fungal infection; IMI, invasive mold infection; RBC, red blood cell.