Background: The management and indication of empiric treatment in Klebsiella pneumoniae carbapenemase-producing K. pneumoniae (KPC-Kp)-colonized patients should be improved.
Methods: A prospective cohort of 94 patients colonized by KPC-Kp was followed for 90 days to validate (i) the Giannella risk score (GRS) to predict the development of any type of KPC-Kp infection and (ii) the INCREMENT-CPE score (ICS) to predict 30-day mortality in patients with infection. Both scores were combined to recommend appropriate empiric treatment. The predictive ability of the scores was measured by calculating the area under the receiver operating characteristic (AUROC) curve.
Results: The GRS showed an AUROC curve for infection due to KPC-Kp of 0.92 (95% confidence interval [CI], .87-.98). The optimal cutoff point was fixed at <7 and ≥7 (92.9% sensitivity, 84.8% specificity); infection developed in 6.3% patients in the 0-6 GRS group and in 84.8% patient in the ≥7 GRS group. According to the ICS, the severity of the infection was also significantly higher in the ≥7 GRS group. The ICS showed an AUROC of 0.78 (95% CI, .65-.91) for 30-day all-cause mortality among patients with infection. A classification and regression tree analysis confirmed the GRS cutoff point at 7, and selected ≥12 points to predict a KPC-Kp infection with a high ICS.
Conclusions: Our results validate the GRS and ICS for indicating empiric therapy in KPC-Kp-colonized patients.