Objective: To estimate the excess length of stay in an intensive care unit (ICU) due to a central line-associated blood stream infection (CLABSI), using a statistical model that accounts for the timing of infection.
Methods: A prospective cohort, active CLABSI surveillance study was conducted in adults admitted to tertiary-care ICU s in three Latin American countries (Argentina, Brazil and Mexico). The protocol, forms, and methodology implemented were developed by the International Nosocomial Infection Control Consortium (INICC). Data collection was performed in the participating ICUs. Data uploading and data analysis were conducted at INICC headquarters. CLABSI were recorded by applying the definitions provided by the CDC NNIS system. The study included all patients admitted to the ICUs during a defined time period with a central line place for more than 24 hours. We used a multi-state approach to model the excess length of stay due to CLABSI while accounting for the competing risks of death and discharge.
Results: A cohort of 3560 patients followed for 36,806 days in 11 ICUs was studied. The average excess length of stay due to a CLABSI in Argentina varied between 2.31 days to 4.69 days; in Brazil it varied between 0.78 days to 4.28 days; and in Mexico it varied between –1.23 days to 4.06 days. A reduction in length of stay in Mexico was caused by an increased risk of death due to a CLABSI, leading to shorter times to death. Adjusting for patient age and Acute Severity of Illness Score tended to increase the estimated excess length of stay due to CLABSI.
Conclusions:
The strength of multi-state models is their ability to incorporate time-dependent exposures. In this study, CLABSI is the key time-dependent exposure as it can happen at any time during a patient’s stay. CLABSIs are associated with an excess length of ICU stay and increased risk of death. The average excess length of stay varies between ICUs, most likely due to the case-mix of admissions and differences in the ways that hospitals deal with infections.