Antibiotic dosing issues in lower respiratory tract infection: population-derived area under inhibitory curve is predictive of efficacy.

Several lower respiratory tract infection (LRTI) trials have documented a correlation between clinical response and area under the inhibitory curve (24 h AUC/MIC; AUIC). The AUIC values in these studies were based on measured MICs and measured serum concentrations. This study evaluates AUIC estimates made using population pharmacokinetic parameters, and MICs from an automated microbiological susceptibility testing system. A computer database review over 2 years yielded 81 patients at Millard Fillmore Hospital with a culture-documented gram-negative LRTI who had been treated with piperacillin and an aminoglycoside, ceftazidime, ciprofloxacin or imipenem. Their AUIC values were estimated using renal function, drug dosages and MIC values. Outcome groups (clinical and microbiological cures and failures) were related to the AUIC values using Kruskal-Wallis ANOVA, linear regression and classification and regression tree (CART) analysis. A significant breakpoint for clinical cures was an AUIC value at least 72 SIT(-1) x 24 h (inverse serum inhibitory titre integrated over time). All antibiotics performed significantly better above this value than below it. Clinical cure was well described by a Hill-type equation. Within the piperacillin/aminoglycoside regimen, most of the activity came from the piperacillin, which had a higher overall AUIC value than the aminoglycoside. AUIC estimations based upon MIC values derived from the automated susceptibility testing method differed from NCCLS breakpoint data and from tube dilution derived values in this hospital by as much as three tube dilutions. These automated methods probably overestimated the MIC values of extremely susceptible organisms. The lack of precise MIC estimates in automated clinical microbiology methods impairs the use of AUIC to prospectively optimize microbiological outcome. Even ignoring this limitation and using the values as they are reported, the results of this analysis suggest that AUIC targets between 72 and 275 SIT(-1) x 24 h are useful in predicting clinical outcome.