BACKGROUND: Standard population pharmacokinetic models for vancomycin rely on serum creatinine and body weight, but their performance in ultra-lean elderly patients with sarcopenia remains unclear. This study evaluated the internal consistency of the Japanese practical area under the concentration-time curve-guided therapeutic drug monitoring (TDM) model across patients stratified by age and body composition. METHODS: Adult patients receiving vancomycin with TDM were retrospectively analyzed and stratified into 3 groups: ultra-lean Elderly (age ≥75 years, body mass index <18.5 kg/m2, n = 24), standard elderly (age ≥75 years, body mass index ≥ 18.5 kg/m2, n = 68), and young controls (age <65 years, n = 24). The primary outcome was the mean error (ME) between a priori predicted and Bayesian posterior vancomycin clearance (CL) estimates. RESULTS: Although posterior vancomycin CL estimates were comparable between elderly groups (2.1 L/h), a priori predictions were significantly lower in ultra-lean elderly patients (1.7 versus 1.9 L/h, P < 0.001). The model significantly underestimated vancomycin CL in ultra-lean elderly patients (ME = -0.312 L/h, P = 0.038) but not in standard elderly individuals (ME = -0.151 L/h, P = 0.051) or young controls (ME = -0.075 L/h, P = 0.679). Despite lower absolute doses in ultra-lean patients (873 versus 1004 mg/d), weight-normalized doses were significantly higher (22.4 versus 19.1 mg/kg/d, P = 0.014), suggesting that clinical judgment may have partially compensated for conservative model predictions. CONCLUSIONS: The practical area under the concentration-time curve-guided TDM model systematically underestimated CL in ultra-lean elderly patients. Careful interpretation of initial predictions and early TDM remain important in such patients.
Kazumasa Nojima (Mon,) studied this question.