Methicillin-resistant Staphylococcus aureus in the ICU: risk factors and a predictive model to detect it at ICU admission

Abstract

Introduction Being capable of predicting MRSA on ICU admission is crucial to enhance infection control and to avoid inappropriate empirical treatment. Two objectives were studied: to describe risk factors for MRSA colonization/infection (MRSA-C/I) once admitted to the ICU; and to develop a predictive model at ICU admission, based on easy-to-obtain admission factors. Methods Data were collected prospectively from 69,894 patients admitted consecutively (stay >24 hours) to 147 Spanish ICUs participating in the National Surveillance Study of Nosocomial Infections in ICU registry (ENVIN) during April to June 2006 to 2010. Univariable and multivariable analysis was performed for both objectives but we used only easy-to-obtain variables for the predictive model exclusively from those admitted in 2010 (n = 16,950, 2/3 for analysis and 1/3 for subsequent validation). Results In the 2006 to 2010 period, 1,046 were C/I by MRSA (note that relative risks are not included due to space limitations). First objective: previous antibiotic, APACHE II score >18, skin-soft tissue or postsurgical superficial skin infections, trauma or medical patient, age >65 (especially >75), urinary catheter and admitted from a longterm care facility were independent risk factors for MRSA-C/I in ICU. Multicolonization increased significantly the risk of MRSA-C/I, and immunodeficiency and gender male emerged as protective factors. Second objective: independent risk factors on ICU admission were male gender, trauma critical patient, urgent surgery, admitted from other ICU, community or long-term facility, being immunosuppressed and skin-soft tissue infection. All configured the risk model for which, although showing good discrimination (AUC-ROC, 0.77; 95% CI, 0.72 to 0.82), sensitivity (67%) and specificity (76.5%) were insufficient for the ICU setting. Afterwards validation with the remaining 4,952 (1/3) showed AUC-ROC = 0.72 (95% CI, 0.65 to 0.79) and P value on the Hosmer-Lemeshow goodness of fit test = 0.539. The model did not improve even after including more complex variables (AUC-ROC = 0.82; 95% CI, 0.77 to 0.86, sensitivity 63.64%, specificity 78.48%). Conclusion Independent risk factors for MRSA-C/I in the ICU and at ICU admission are described. To predict MRSA-C/I at ICU admission we should not rely on clinical-demographic risk factors alone. Its combination with a rapid laboratory test could be the way to proceed in future studies.