Duration of Systemic Inflammatory Response Syndrome Influences Serum Vancomycin Concentration in Patients With Sepsis

Vancomycin (VCM) is used in the treatment of methicillin-resistant Staphylococcus aureus infection. The dosage of VCM must be adjusted by using therapeutic drug monitoring because of the drug's narrow therapeutic concentration window. Although optimal administration based on population pharmacokinetic (PPK) analysis and/or a Bayesian method has improved prediction accuracy, serum concentrations of VCM in patients with sepsis often deviate significantly from predicted values. We investigated factors influencing prediction errors with PPK analysis in VCM dosing.

This retrospective cohort study included patients treated with VCM. Their clinical data were recorded, and there were 27 nonseptic patients and 68 septic patients. VCM concentrations were predicted by using PPK analysis and data compared with observed concentrations.

Patients with sepsis had a higher mean absolute error than nonseptic patients, indicating a deviation of VCM concentrations from predicted values in the septic patients. To determine factors influencing prediction errors, we classified patients with sepsis into 3 subgroups according to the mean absolute error value (2.17) for the nonseptic patients: "lower" group (prediction errors, below -2.17), "upper" group (>2.17), and "no change" group (-2.17 to 2.17). In a comparison of clinical characteristics of the 3 groups, significant differences were found in the duration of systemic inflammatory response syndrome (SIRS), SIRS score, disseminated intravascular coagulation score, and levels of creatinine clearance (CrCl), hemoglobin, and diastolic blood pressure. Multiple logistic regression analysis identified SIRS duration and CrCl as factors associated with VCM concentrations in the lower and/or upper groups of septic patients. Shorter and longer SIRS duration were associated with VCM concentrations in the lower group and the upper group, respectively, compared with predicted values in patients with sepsis. According to receiver-operating characteristic curve analysis, the optimal cutoff value of SIRS duration for the lower group was 2 days; for the upper group, it was 6 days. VCM clearance in patients with an SIRS duration <2 days was higher than that for patients with an SIRS duration ≥6 days.

SIRS duration was identified as influencing VCM concentration in patients with sepsis. This study has 2 limitations. First, we performed blood sampling only for trough concentrations. Repeated blood sampling for both peak and trough concentrations should be performed for more accurate pharmacokinetic evaluation in critically ill patients. Second, we determined CrCl by using the Cockcroft-Gault formula, which may not be accurate in critically ill patients. Modifying VCM dosing according to SIRS duration will improve prediction accuracy of VCM concentration based on therapeutic drug monitoring.