Epidemiology, antimicrobial resistance trends, and predictors of mortality among vancomycin-resistant enterococci: a six-year retrospective cohort study from Oman.

backgroundVancomycin-resistant enterococci (VRE) are an escalating cause of healthcare-associated infection, yet longitudinal data from the Arabian Peninsula remain scarce. We characterised the epidemiology, resistance patterns and predictors of in-hospital mortality of enterococci over six years at a tertiary hospital in Oman, with explicit attention to the effect of changing surveillance practice and to the distinction between colonisation and infection.

methodsWe analysed 1,626 non-duplicate Enterococcus isolates from 1,018 patients (2020-2025). VRE was defined a priori as a vancomycin-resistant result on susceptibility testing or a laboratory VRE flag; agreement between these two routes was quantified, and all trend analyses were repeated using a phenotype-only definition restricted to clinical (non-screening) specimens and excluding intrinsically resistant vanC species. A chart-reviewed cohort provided comorbidity, biomarker and outcome data. Mortality models were fitted on complete cases and after multiple imputation by chained equations (m = 20, Rubin's rules), and discrimination and calibration were assessed by repeated stratified five-fold cross-validation.

resultsVRE by the composite definition accounted for 395 of 1,626 isolates (24.3%). Only 196 (49.6%) were vancomycin-resistant on testing; 199 (50.4%) were flag-only with no vancomycin result, and a further 66 isolates were vancomycin-resistant but unflagged. Active rectal screening began only in 2024 and contributed 186 of 395 VRE isolates (47.1%). The composite VRE proportion therefore rose from 1.8% (2020) to 47.4% (2025) (Cochran-Armitage Z = 16.56, p < 0.001), whereas vancomycin resistance among tested clinical non-vanC isolates rose from 0.5% to 21.8% and plateaued after 2023 (21.8%, 20.7%, 21.8% in 2023, 2024 and 2025; Z = 8.32, p < 0.001). Of 193 VRE-positive patients, 96 (49.7%) were detected by rectal screening alone and 30 (15.5%) carried vanC species, leaving 67 (34.7%) with acquired resistance in a clinical specimen; 29 (15.0%) were bacteraemic. E. faecium accounted for 141 of 193 VRE-positive patients (73.1%). Linezolid (0.9%) and tigecycline (1.9%) retained near-complete activity. In the chart-reviewed cohort, 163 of 213 patients were microbiologically confirmed as VRE-positive; in-hospital mortality was 29.5% (44/149) among confirmed patients. Low albumin was an independent predictor on complete-case analysis (adjusted OR 0.86 per g/L, 95% CI 0.77-0.95, p = 0.004; n = 82, 24 events) but attenuated after multiple imputation (OR 0.94, 0.87-1.00, p = 0.066), whereas intensive-care admission was the most robust predictor (imputed OR 3.14, 1.36-7.24, p = 0.007). Cross-validated discrimination was modest and unstable (logistic AUC 0.775, SD 0.128; fold range 0.345-1.000), calibration slope was 0.81, and accuracy at a 0.5 threshold (73.8%) barely exceeded the 70.7% obtained by predicting survival for every patient.

conclusionsEnterococcal vancomycin resistance increased substantially at this centre and is concentrated in E. faecium, with preserved last-line susceptibility. However, much of the apparent rise after 2023 is attributable to the introduction of active rectal screening and to intrinsically resistant vanC species, and half of VRE-positive patients were colonised rather than infected. Prognostic models built on routine admission biomarkers were insufficiently discriminating and calibrated for clinical use and should be regarded as hypothesis-generating.