Abstract
BACKGROUND: Existing surveillance mechanisms may underestimate the incidence of carbapenem-resistant gram-negative infections (CRGNIs). Although carbapenem resistance increases the risk of death, the trend in mortality over time is unknown. METHODS: A retrospective cohort study was conducted at 40 academic medical centers using a discharge database to identify adult hospital admissions without cystic fibrosis in 2006-2012 and received intravenous colistin for >3 consecutive days or died during therapy (termed colistin cases). The primary outcomes were the number of colistin cases per 100,000 admissions per year and change in the hospital mortality rate over time compared with the rate of discharges to home. Secondary outcomes included median overall and intensive care unit lengths of stay. RESULTS: From 2006 to 2012, a total of 5011 unique patients were identified as colistin cases. The number per 100,000 admissions per year increased from 35.56 to 92.98 during the 7-year study (P < .001). The odds of in-hospital death among colistin cases (compared with discharge to home) decreased by a mean of 5.2%/y (P = .04), whereas discharge to an institution (P = .24) or hospice (P = .89) remained steady over time. The median overall and intensive care unit lengths of stay decreased by 7.5 and 6 days, respectively (P < .001). In a 4-hospital chart review, 81.6% of colistin cases were found to have culture-positive CRGNIs. Conversely, 53% of extensively drug-resistant bloodstream CRGNIs at 2 of these hospitals met colistin case criteria. CONCLUSIONS: Colistin cases represent a severely ill population with a high probability of having culture-confirmed CRGNIs. Colistin tracking is a novel strategy for monitoring the incidence and mortality of CRGNIs, particularly those caused by extensively drug-resistant bacteria. Although the incidence of colistin cases nearly tripled within 7 years, more of these patients are surviving hospitalization and going home.