Hacer Ozturk Akin, Ozge Unlu
Background and Objectives: Non-fermenting Gram-negative bacilli (NFGNB) are frequently recovered from respiratory specimens in intensive care units (ICUs) and are increasingly carbapenem-resistant. We described their distribution and resistance profiles in adult ICUs over three years. Materials and Methods: Respiratory specimens submitted from adult ICUs of a foundation university hospital in Istanbul between 6 January 2023 and 27 December 2025 were analysed retrospectively using laboratory information system data. Identification and susceptibility testing used the BD Phoenix M50 system with year-specific EUCAST breakpoints (v13.0-v15.0); colistin MICs were determined by broth microdilution. The number of patients contributing more than one isolate was 268. Accordingly, the primary trend analysis was performed using generalised estimating equations (GEE) accounting for within-patient clustering. Chi-square and first-isolate analyses were used as secondary and sensitivity analyses. Results: A total of 1407 isolates were recovered from 654 of 1194 patients with a respiratory culture. Acinetobacter spp. predominated (56.9% of isolates; 54.3% of patients), followed by Pseudomonas spp. (29.4%) and Stenotrophomonas maltophilia (11.1%). Carbapenem resistance in Acinetobacter spp. was 96.7% (95% CI 95.3-97.8) for imipenem and 96.6% (95.1-97.7) for meropenem, without temporal trend. In P. aeruginosa, meropenem resistance rose from 44.7% to 71.4% (GEE odds ratio per year 1.45, 95% CI 1.09-1.92, p = 0.011), whereas imipenem resistance did not change significantly. Among the agents tested, colistin was found to be the most effective (22.4% resistance in Acinetobacter spp. and 13.1% in P. aeruginosa). Ceftazidime-avibactam resistance in P. aeruginosa was 24.3%. Of the S. maltophilia isolates, 85.8% were categorised as susceptible-increased exposure to trimethoprim-sulfamethoxazole. Conclusions: Carbapenem resistance is high and stable in Acinetobacter spp. but rising for meropenem in P. aeruginosa. These data do not distinguish colonisation from infection and cannot be linked to clinical outcomes. They should therefore be used to inform local empirical-treatment policy alongside clinical assessment, rather than as evidence of treatment failure.