Welcome to Chinese Journal of Clinical Pharmacology and Therapeutics,Today is Chinese

Chinese Journal of Clinical Pharmacology and Therapeutics ›› 2026, Vol. 31 ›› Issue (10): 1363-1376.doi: 10.12092/j.issn.1009-2501.2026.10.007

Previous Articles    

External validation of population pharmacokinetic models for linezolid in adult patients

Qiu FANG1,2(), Shuangmiao JIANG1, Jing WANG3, Xuemei LUO3, Lu JIN3, Huaijun ZHU1,3,4,*()   

  1. 1. Department of Pharmacy, Nanjing Drum Tower Hospital Clinical College, Nanjing University of Chinese Medicine, Nanjing 210008, Jiangsu, China
    2. Department of Pharmacy, the Affiliated Hospital of Xuzhou Medical University, Xuzhou 221004, Jiangsu, China
    3. Department of Pharmacy, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing 210008, Jiangsu, China
    4. Nanjing Medical Center for Clinical Pharmacy, Nanjing 210008, Jiangsu, China
  • Received:2025-09-25 Revised:2025-11-13 Online:2026-10-26 Published:2026-11-11
  • Contact: Huaijun ZHU E-mail:fangqiu5027@163.com;huaijun.zhu@gmail.com

Abstract:

AIM: To evaluate the predictive performance of published population pharmacokinetic models for linezolid and explore their applicability and limitations. METHODS: A systematic search was conducted in databases such as PubMed and CNKI (up to June 2025), and model information was extracted. Steady-state blood concentration data from 223 samples of 111 adult infected patients at Nanjing Drum Tower Hospital were collected for external validation using goodness-of-fit tests, prediction error tests, and visual predictive checks. RESULTS: A total of 25 population pharmacokinetic models for linezolid were included in this study. Validation results showed that the population prediction performance was suboptimal, with the best model having an R2 of only 0.28. The prediction error within the ±30% range (F30) was below 35%, and none of the models met the established standards. However, individual prediction capability was relatively good, with the best model achieving an R2 of 0.92 and an IF30 value as high as 82.11%, indicating that most models demonstrated good prediction ability. Subgroup analysis suggested that specific clinical environments could improve population predictions, and high-precision individual prediction models showed strong stability. CONCLUSION: The existing population pharmacokinetic models for linezolid exhibit large prediction errors and limited extrapolation ability in patients at this center, with insufficient clinical applicability. Individual predictions can help improve model accuracy. Future research should incorporate more covariates, expand the sample size of special populations, and enhance model generalizability through prospective studies.

Key words: linezolid, population pharmacokinetic models, external validation, therapeutic drug monitoring

CLC Number: