Diagnostic accuracy of carboxyhemoglobin saturation with pulse CO-oximetry in patients with carbon monoxide poisoning.

Abstract

Carboxyhemoglobin saturation (SpCO)with pulse CO-oximetry is an alternative method in CO poisoning; however, the correlation of it with blood carboxyhemoglobin level (COHb) is still debated.The study aimed to evaluate the correlation between SpCO and venous COHb level and factors associated with the diagnostic accuracy of SpCO. SpCO and venous COHb levels of patients with CO poisoning, according to a COHb level was 3%, were evaluated. The sensitivity, specificity, 95% CI (confidence interval), and the cut-off value of SpCO were calculated using ROC analysis at a 10% threshold for COHb. Agreement levels were calculated with Bland-Altman analysis. Risk factors affecting diagnostic accuracy were analyzed using logistic regression analysis. If the 10% threshold of COHb was accepted as the diagnostic threshold for CO poisoning, the sensitivity and specificity of SpCO were 98.4% and 100% (95% CI: 0.996-1.000) at the 6.85 cut-off point of SpCO. The scatter plot of COHb and SpCO showed a strong positive relationship at values of presentation and discharge (r=.979, p<0.001; r=.969, p<0.001). With a bias of 3.1% for the mean difference between-COHb and SpCO, the limits of agreement from Bland- Altman analysis were -0.7 to 7.1. For the 10% threshold, age and male gender have significantly increased false negativity rates (B=.074, p=0.010, and B=0.252, p=0.011, respectively). SpCO is a reliable method with high sensitivity and specificity; therefore, a lower cut-off points of SpCO compared to COHb level may be accepted to diagnose and follow-up CO poisoning.