Impact of blood sampling time deviations on vancomycin therapeutic drug monitoring and clinical decision-making.

OBJECTIVE

To evaluate the impact of deviations in blood sampling time on the interpretation of measured trough concentrations in vancomycin therapeutic drug monitoring; to explore a reference time window for early blood sampling in clinical settings where trough concentration monitoring remains the primary approach.

METHODS

A retrospective analysis was conducted on 160 therapeutic drug monitoring (TDM) records from inpatients receiving vancomycin TDM in our hospital between March 2024 and March 2026. Among these, 137 early-sampled cases were included for analysis, with predicted trough concentrations (C_pred) obtained using the Pharmado platform. Using C_pred as a model-assisted reference, the relative standard deviation (RSD) of measured concentrations (C_obs) was calculated. The target range of 10-20 μg/mL was applied to assess misjudgments in target attainment. Spearman correlation and Logistic regression analyses were employed to evaluate the relationship between time deviation (Δt) and RSD.

RESULTS

C_obs exhibited a systematic tendency toward overestimation compared to C_pred, with a significant positive correlation between Δt and RSD (ρ = 0.62, p < 0.001). In the group with sampling within 1 h before scheduled dosing, the median RSD was 10.8%, with 81.3% of cases meeting |RSD| ≤ 20% and zero misjudgments in target attainment. For sampling 1-2 h prior, this proportion declined to 60.0%. In the >2 h group, it fell to less than 24.6%, with significantly elevated misjudgment risk. Logistic regression confirmed advancing sampling time negatively predicted |RSD| ≤ 20% probability (β₁ = -1.088, odds ratio = 0.337). The model estimated that when blood sampling occurred approximately 0.83 and 2.10 h in advance, this probability decreased to 80 and 50%, respectively; however, these time points and the ≤1 h reference window should be considered exploratory findings conditional on the predictive performance of the Pharmado model and the limited sample size, and external validation is required before broader clinical application.

CONCLUSION

Early blood sampling in vancomycin TDM can lead to systematic overestimation of trough concentrations, with greater deviations associated with longer early sampling times. When the blood sampling time is advanced by ≤1 h, the deviation between the measured concentration and the predicted trough concentration is minimal. Under the condition that the sampling time is accurately recorded, the measured value may be cautiously used for target concentration assessment. If the sampling time falls outside this window, either platform-based correction or re-sampling is recommended. The proposed 1-h threshold should be interpreted as an exploratory, model-dependent reference rather than a definitive clinical cutoff.