Risk factors for infection in NDM-colonized patients and applicability of the Giannella Risk Score

Carbapenem-resistant Enterobacterales (CRE) represent a major global health challenge. New Delhi metallo-β-lactamase (NDM)- producing Enterobacterales have recently emerged as a serious threat due to their rapid worldwide spread and the limited therapeutic options available. The objective of this study was to analyse risk factors for infection in NDM carriers and the applicability of the Giannella Risk Score (GRS). The secondary objective was to evaluate the management and outcome of infected patients.

We conducted a retrospective study of adult patients with NDM-producing Enterobacterales colonisation or infection admitted to a secondary hospital between November 2023 and October 2025. Epidemiological characteristics, comorbidities, infectious episodes, and antimicrobial therapy were collected.

A total of 94 patients with NDM-producing Enterobacterales were included: 51 remained colonised without infection throughout follow-up (NDM-NI) and 43 developed infection (NDM-I). Of the latter, 22 already met the infection case definition at the time of first NDM detection (prevalent infection), while 21 were colonised at cohort entry and progressed to infection during follow-up (incident infection), corresponding to an incidence of infection among colonised patients of 29% (21/72). Restricting the analysis of risk factors to these 21 incident cases versus the 51 NDM-NI patients (n = 72), multisite colonisation (≥ 1 additional site: RR 9.43, 95% CI 3.06-29.08, p < 0.001) and a GRS ≥ 7 (RR 2.71, 95% CI 1.29-5.68, p = 0.006) were both associated with a higher risk of progression to infection. On multivariable analysis, multisite colonisation remained the only independent predictor (adjusted OR 24.60, 95% CI 6.04-100.18, p < 0.001), while GRS ≥ 7 did not reach independent significance (p = 0.797). Regarding treatment options, 19 patients were treated with cefiderocol (CEF), while 22 were treated with other therapy (OT). Alternative therapies consisted of aminoglycoside (n = 7), fosfomycin (n = 7), nitrofurantoin (n = 3), aztreonam monotherapy (n = 1), aztreonam-avibactam (n = 1), trimethoprim-sulfamethoxazole (n = 1), and a combination of ceftazidime-avibactam plus aztreonam (n = 1); with this single exception, all alternative-therapy regimens were monotherapy (Fig. 2). The OT group predominantly presented with a urinary source (CEF 10 (53%) vs. OT 20 (91%), p = 0.012). The CEF group had more patients with bacteremia (CEF 9 (47%) vs. OT 0 (0%), p = 0.001) and included more severely ill patients (SOFA ≥ 3, CEF 9 (47%) vs. OT 3 (14%), p = 0.037; INCREMENT ≥ 7, CEF 9 (47%) vs. OT 3 (14%), p = 0.037). There were no significant differences in clinical cure, recurrence, or mortality between the two groups.

Multisite colonisation was strongly and independently associated with progression to infection in NDM carriers, while the Giannella Risk Score, as a continuous score, showed acceptable discrimination. Given the substantial overlap between these measures, larger studies are needed to directly compare the predictive performance of colonisation burden with the full GRS in NDM-predominant settings.