In the critical care setting, augmented renal clearance (ARC) is increasingly recognized as one of the leading causes of subtherapeutic antibiotic exposure [1]. However, commonly used formulas for estimating glomerular filtration rate (GFR) are inaccurate in patients with ARC and the 24-h urinary creatinine clearance (CrCL) remains the best available approach for optimizing empirical antimicrobial therapy [2]. On the other hand, no study has evaluated the clinical and prognostic value of the kinetic estimated GFR (KeGFR) in this context. We thus aimed to determine whether KeGFR could be a reliable alternative to measured CrCL in critically ill patients needing early initiation of an appropriate piperacillin dosing regimen.

For this purpose, we retrospectively analyzed 60 consecutive patients who underwent 24-h urinary CrCL measurements and therapeutic drug monitoring during the first 3 days of antimicrobial therapy of piperacillin administered 16 g/day continuously. The protocol pertaining to this substudy has been published elsewhere [3]. As previously described, the corresponding KeGFR was calculated as follows: \( \frac{\mathrm{Baseline}\ \mathrm{sCr}\ \mathrm{x}\ \mathrm{eGFR}}{\mathrm{Mean}\ \mathrm{sCr}}\times \left(1-\frac{24\ \mathrm{x}\ \Delta \mathrm{sCr}}{\Delta \mathrm{t}\ \mathrm{x}\ \mathrm{Max}\Delta \mathrm{sCr}/\mathrm{Day}}\right) \) with eGFR derived from the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation using serum creatinine (sCr) before admission, Δt fixed at 24 h between two sCr measurements, and maximal sCr increase per day approximated to 133 μmol/L [4]. ARC was defined by a measured CrCL ≥ 130 mL/min/1.73 m2. Piperacillin underdosing was arbitrarily defined by a free drug concentration ≤ 32 μg/ml at steady state.

Among the 180 samples analyzed, the incidence of ARC was 48% (median CrCL values = 124 [83–170] ml/min/1.73 m2) and the incidence of piperacillin underdosing was 51% (median piperacillin concentrations = 32 [22–47] μg/ml). The diagnostic agreement between KeGFR and CrCL was only moderate (κ = 0.48 [95% confidence interval 0.4–0.55]) (Fig. 1). Comparison between KeGFR and CrCL showed a mean bias of − 8.7 ml/min/1.73 m2 and limit of agreement from − 99 ml/min/1.73 m2 to 82 ml/min/1.73 m2. Finally, the area under the ROC curve generated for KeGFR was significantly lower than the one generated for measured CrCL for prediction of piperacillin underdosing (0.76 [0.68–0.83] vs 0.85 [0.79–0.91], p = 0.03; Fig. 2).

Fig. 1
figure 1

Correlation between measured CrCL and KeGFR (r2 = 0.54, p < 0.0001) and repartition of samples with (white circles) or without (black circles) piperacillin underdosing, defined by an unbound concentration ≤ 32 μg/ml. ARC was defined by a measured CrCL or a KeGFR ≥ 130 ml/min/1.73 m2. Samples in the gray shaded area are considered to be well classified

Full size image
Fig. 2
figure 2

Receiver operating characteristics (ROC) curves evaluating the ability of KeGFR and measured CrCL to predict piperacillin underdosing. Areas under ROC curves between KeGFR and measured CrCL were compared using the Handley approach. Piperacillin underdosing was defined by a free drug concentration ≤ 32 μg/ml

Full size image

In conclusion, KeGFR is not interchangeable with measured CrCL for prediction of piperacillin underexposure in critically ill patients with ARC. Also, scarce data may suggest a better predictive value of Cockcroft-Gault compared to MDRD (Modification of Diet in Renal Disease Study) or CKD-EPI for identifying patients with ARC [5]; a measured CLCR should be performed to accurately guide drug dosing. This study emphasizes the need for dosing adjustment and therapeutic drug monitoring in patients with ARC.