Carboxyhemoglobin levels in medical intensive care patients: a retrospective, observational study

Abstract

Introduction: Critical illness leads to increased endogenous production of carbon monoxide (CO) due to the induction of the stress-response enzyme, heme oxygenase-1 (HO-1). There is evidence for the cytoprotective and anti-inflammatory effects of CO based on animal studies. In critically ill patients after cardiothoracic surgery, low minimum and high maximum carboxyhemoglobin (COHb) levels were shown to be associated with increased mortality, which suggests that there is an 'optimal range' for HO-1 activity. Our study aimed to test whether this relationship between COHb and outcome exists in non-surgical ICU patients.

Methods: We conducted a retrospective, observational study in a medical ICU at a university hospital in Vienna, Austria involving 868 critically ill patients. No interventions were undertaken. Arterial COHb was measured on admission and during the course of treatment in the ICU. The association between arterial COHb levels and ICU mortality was evaluated using bivariate tests and a logistic regression model.

Results: Minimum COHb levels were slightly lower in non-survivors compared to survivors (0.9%, 0.7% to 1.2% versus 1.2%, 0.9% to 1.5%; P=0.0001), and the average COHb levels were marginally lower in non-survivors compared to survivors (1.5%, 1.2% to 1.8% versus 1.6%, 1.4% to 1.9%, P=0.003). The multivariate logistic regression analysis revealed that the association between a low minimum COHb level and increased mortality was independent of the severity of illness and the type of organ failure.

Conclusions: Critically ill patients surviving the admission to a medical ICU had slightly higher minimum and marginally higher average COHb levels when compared to non-survivors. Even though the observed differences are statistically significant, the minute margins would not qualify COHb as a predictive marker for ICU mortality.

Figure 1

Study Flow Chart.

Figure 2

Mean carboxyhemoglobin during the stay on the ICU according to the admission category. P < 0.0001 in univariate ANOVA.

Figure 3

Correlation between minimal carboxyhemoglobin during the stay on the ICU with SAPS II score. Pearson's correlation coefficient with 95% confidence intervals: 0.21 (0.14 to 0.27), P < 0.0001. The line and error bars (95% confidence intervals) are derived from linear regression.

Figure 4

Mean levels of COHb during the first 72 hours after admission to the ICU. Empty and full triangles represent non-survivors and survivors, respectively. In order to prevent overlap black triangles were shifted to the right by 1.2 hours.