Correlation of antimicrobial fraction unbound and sieving coefficient in critically ill patients on continuous renal...

Fraction unbound has been used as a surrogate for antimicrobial sieving coefficient (SC) to predict extracorporeal clearance in critically ill patients on continuous renal replacement therapy (CRRT), but this is based largely on expert opinion.

To examine relationships between package insert-derived fraction unbound (Fu-P), study-specific fraction unbound (Fu-S), and SC in critically ill patients receiving CRRT.

English-language studies containing patient-specific in vivo pharmacokinetic parameters for antimicrobials in critically ill patients requiring CRRT were included. The primary outcome included correlations between Fu-S, Fu-P, and SC. Secondary outcomes included correlations across protein binding quartiles, serum albumin, and predicted in-hospital mortality, and identification of predictors for SC through multivariable analysis.

Eighty-nine studies including 32 antimicrobials were included for analysis. SC was moderately correlated to Fu-S (R2 = 0.55, P < 0.001) and Fu-P (R2 = 0.41, P < 0.001). SC was best correlated to Fu-S in first (<69%) and fourth (>92%) quartiles of fraction unbound and above median albumin concentrations of 24.5 g/L (R2 = 0.71, P = 0.07). Conversely, correlation was weaker in patients with mortality estimates greater than the median of 55% (R2 = 0.06, P = 0.84). SC and Fu-P were also best correlated in the first quartile of antimicrobial fraction unbound (R2 = 0.66, P < 0.001). Increasing Fu-P, flow rate, membrane surface area, and serum albumin, and decreasing physiologic charge significantly predicted increasing SC.

Fu-S and Fu-P were both reasonably correlated to SC. Caution should be taken when using Fu-S to calculate extracorporeal clearance in antimicrobials with 69%-92% fraction unbound or with >55% estimated in-hospital patient mortality. Fu-P may serve as a rudimentary surrogate for SC when Fu-S is unavailable.