Factors Associated With In-Hospital Mortality and Length of Stay Among Hospitalised Patients Receiving Carbapenem Therapy: A Retrospective Cohort Study.

Abstract

Background: Patients receiving carbapenems are clinically heterogeneous and have severe infections and multiple comorbidities. Evidence on factors associated with mortality and length of stay (LOS) in this treatment-defined population remains limited.

Methods: This retrospective cohort study included adult hospital episodes involving at least one carbapenem dose at Hospital Shah Alam during 2025. Firth penalised logistic regression was used for in-hospital mortality, and median regression with 2000 bootstrap resamples was used for right-skewed LOS. Covariates were selected a priori. Carbapenem continuation beyond 72 h was treated as a postbaseline characteristic and was not included in outcome models.

Results: Among 112 eligible episodes involving 107 patients, 38 deaths occurred (33.9%); median LOS was 19.0 days (IQR: 10.8-28.0). Older age (adjusted OR: 1.41 per 10 years, 95% CI: 1.07-1.92) and definitive indication at carbapenem initiation (adjusted OR: 0.28, 95% CI: 0.10-0.70) were associated with in-hospital mortality. Hospital-acquired infection was associated with a longer adjusted median LOS (median difference 8.63 days, 95% bootstrap CI: 0.50-14.44). Definitive indication was not clearly associated with LOS after adjustment (median difference -5.54 days, 95% bootstrap CI: -12.00-4.62).

Conclusions: Older age and definitive indication at carbapenem initiation were associated with in-hospital mortality, while hospital-acquired infection was associated with longer median LOS. The treatment-indication association should not be interpreted causally because residual confounding, particularly by acute illness severity and indication, remains likely.