Seyeon Hwang 1, Donghwan Lee 1, Minji Kwon 2, Jongdae Han 3, Sumin Chae 4, Won Gun Kwack 5, Bo-Hyung Kim 2
1 Department of Statistics, Ewha Womans University (Seoul, Republic of Korea), 2 Department of Clinical Pharmacology and Therapeutics, Kyung Hee University College of Medicine, Kyung Hee University Hospital (Seoul, Republic of Korea), 3 Department of Statistics and Data Science, Korea National Open University (Seoul, Republic of Korea), 4 Department of Surgery, Kyung Hee University College of Medicine, Kyung Hee University Hospital (Seoul, Republic of Korea), 5 Department of Internal Medicine, Kyung Hee University College of Medicine, Kyung Hee University Hospital (Seoul, Republic of Korea)
Introduction / Objectives
Antibiotics are among the most frequently prescribed medications in hospitalized patients, and several agents have been associated with an increased risk of acute kidney injury (AKI). In particular, broad-spectrum antibiotics used in severe infections are often suspected of nephrotoxic potential, making renal safety an important clinical consideration [1]. However, AKI risk is not determined by antibiotic exposure alone; it is also influenced by patient-specific factors such as baseline renal function, comorbidities, concomitant medications, and clinical interventions. This study aims to evaluate differences in AKI risk according to antibiotic type using real-world clinical data and to assess potential effect heterogeneity by patient-level characteristics, including comorbidities and concomitant medications.
Methods
This observational study was conducted using integrated electronic medical record (EMR) and order communication system (OCS) data collected from 2013 to 2025. Overall, 18,793 hospitalized adult patients who received antibiotics commonly used for severe infections during hospitalization were included in the analysis. Commonly used agents with adequate sample size, such as vancomycin, teicoplanin, meropenem, ertapenem, and imipenem, were selected for the analysis.
The primary outcome was a binary indicator of AKI, defined as meeting either of the following criteria: an increase in serum creatinine of ≥0.3 mg/dL within 48 hours or ≥1.5-fold from baseline within 7 days [2]. Demographic characteristics, selected comorbidities, indicators of clinical severity, and concomitant nephrotoxic medications were considered as potential confounders.
AKI incidence was summarized descriptively across antibiotic groups. Associations between antibiotic exposure and AKI were estimated using time-varying pooled logistic regression models that incorporated time-updated antibiotic exposure and clinical status during hospitalization [3].
To mitigate time-dependent confounding, stabilized inverse probability weights were estimated and incorporated into marginal structural models [4]. Balance of baseline and time-varying covariates after weighting was evaluated using standardized mean differences (SMDs) [5].
Effect modification was assessed by including interaction terms between antibiotic type and selected patient characteristics. Corresponding subgroup-specific effect estimates were obtained, and sensitivity analyses were conducted to evaluate the robustness of the results.
Results
A total of 18,793 hospitalized adult patients receiving at least one of the selected antibiotics were included in the analysis. In the time-fixed analysis based on the index antibiotic at treatment initiation, AKI incidence ranged from 3.8% to 47.3% across antibiotic groups, indicating substantial variability in crude risk.
After accounting for demographic characteristics and time-varying clinical factors, differences in AKI risk across antibiotic types remained evident. Estimates obtained using weighted causal models were consistent with primary adjusted analyses, supporting the robustness of the findings.
Interaction analyses indicated potential heterogeneity in treatment effects, particularly by baseline renal function and selected comorbidities. However, subgroup findings should be interpreted cautiously due to limited statistical power in certain strata.
Conclusions
In our real-world observational study, differences in AKI risk across commonly used antibiotics suggest that these agents may not be interchangeable with respect to renal safety in routine clinical practice. Differences in nephrotoxic potential imply that antibiotic selection should extend beyond antimicrobial spectrum alone and incorporate considerations of renal risk.
The potential effect modification by baseline renal function and comorbid conditions further supports incorporating patient-specific vulnerability into antibiotic selection. Although these observations warrant cautious interpretation, they highlight the importance of considering patient-specific characteristics when selecting antibiotic therapy.
Further methodological refinement and external validation are warranted to strengthen causal interpretation and to inform more individualized antibiotic selection strategies for hospitalized patients.
References:
[1] Campbell, R. E., Chen, C. H., & Edelstein, C. L. (2023). Overview of antibiotic-induced nephrotoxicity. Kidney International Reports, 8(11), 2211-2225.
[2] Khwaja, A. (2012). KDIGO clinical practice guidelines for acute kidney injury. Nephron Clinical Practice, 120(4), c179-c184.
[3] Hernán, M. Á., Brumback, B., & Robins, J. M. (2000). Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology, 11(5), 561-570.
[4] Robins, J. M., Hernan, M. A., & Brumback, B. (2000). Marginal structural models and causal inference in epidemiology. Epidemiology, 11(5), 550-560.
[5] Austin, P. C. (2009). Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity‐score matched samples. Statistics in medicine, 28(25), 3083-3107.
Reference: PAGE 34 (2026) Abstr 12000 [www.page-meeting.org/?abstract=12000]
Poster: Drug/Disease Modelling - Safety