This article is more than 5 years old. Information may no longer be current.
EMR data rapidly predict carbapenem-resistant infection
Click Here to Manage Email Alerts
In patients with Klebsiella pneumoniae bacteremia, a multiple logistic regression model using electronic medical record data can produce fast, sensitive predictions of carbapenem resistance, according to recent findings.
“The early initiation of effective antibiotic therapy is a critical step in the successful treatment of gram-negative bacteremia,” researchers wrote in Open Forum Infectious Diseases. “Identifying the bacterial species growing in a blood culture can help guide early antibiotic choices, and rapid species identification is now possible in many laboratories. However, antibiotic susceptibility results may not be available for several days.”
To develop a tool to rapidly detect carbapenem resistance, Timothy Sullivan, MD, assistant professor at Mount Sinai Health System, and colleagues evaluated all cases of K. pneumoniae bacteremia (n = 613) at Mount Sinai Hospital from September 2012 through September 2016. The researchers randomly separated cases into one of two categories: a “training set” (n = 460) and a “testing set” (n = 153). Both sets had an equal proportion of imipenem-resistant cases.
Using EMR data from the training set cases, the researchers constructed a multiple logistic regression model of carbapenem resistance. The performance of the model was assessed by repeated K-fold cross-validation, and by applying the training set model to the testing set.
The following risk factors for imipenem resistance were included in the logistic regression model: prior colonization with imipenem-resistant K. pneumoniae; hospital unit; total inpatient days in the previous 5 years; total days of oral or parenteral antibiotics in the past few years; and age older than 60 years.
Of the 613 cases of K. pneumoniae bacteremia, 61 (10%) were carbapenem resistant, which was defined in the study as an imipenem minimum inhibitory concentration of at least 2 g/mL.
The regression model constructed from the training set successfully predicted 73% of carbapenem-resistant cases and 59% of carbapenem-susceptible cases in the testing set (sensitivity, 73%; specificity, 59%; positive predictive value, 16%; negative predictive value, 95%). The mean area under the receiver operator characteristic curve of the K-fold cross-validation repeats was 0.731, the researchers said.
The researchers determined the time to active antibiotics for all 613 cases. Thirty-three percent of patients with imipenem-resistant K. pneumoniae never received active antibiotics vs. 3% of those with imipenem-susceptible infections (P < .0001). Among the patients who did receive active antibiotics, those with imipenem-resistant infections had a significantly longer time to active antibiotics vs. those with imipenem-susceptible infections (40.4 hours vs. 9.6 hours; P < .0001). Appropriate treatment was given on the first day for only 44% of imipenem-resistant infections vs. 84% of cases that were imipenem susceptible.
“Because effective treatment is often delayed in cases of carbapenem-resistant bacteremia, and this delay may contribute to the high mortality associated with these infections, our model could potentially help improve outcomes by quickly identifying patients at risk for [carbapenem-resistant Enterobacteriaceae] infection,” the researchers wrote. – by Jennifer Byrne
Disclosures: The authors report no relevant disclosures.
Click Here to Manage Email Alerts