Evaluation of Oral Tebipenem as a Step-Down Therapy following Intravenous Ertapenem against Extended-Spectrum ß-Lactamase-Producing Escherichia coli in a Hollow-Fiber In Vitro Infection Model.

Tebipenem is an orally bioavailable carbapenem in development for the treatment of patients with complicated urinary tract infections. Herein, we describe the results of studies designed to evaluate tebipenem's potential as an oral (p.o.) transition therapy from intravenous (i.v.) ertapenem therapy for the most common uropathogen, Escherichia coli. These studies utilized a 7-day hollow-fiber in vitro infection model and 5 extended-spectrum ß-lactamase-producing E. coli challenge isolates. Human free-drug serum concentration-time profiles for tebipenem 600 mg p.o. every 8 h and ertapenem 1 g i.v. every 24 h were simulated in the hollow-fiber in vitro infection model. Samples were collected for bacterial density and drug concentration determination over the 7-day study period. Generally, ertapenem monotherapy resulted in a greater reduction in bacterial density than did tebipenem monotherapy. In the treatment arms in which ertapenem dosing was stopped following dosing for 1 or 3 days, immediate bacterial regrowth occurred and matched that of the growth control. Finally, in the treatment arms in which ertapenem dosing was stopped following dosing for 1 or 3 days and tebipenem dosing was initiated for the remainder of the 7-day study, the intravenous-to-oral transition regimen reduced bacterial burdens and prevented regrowth. Given that transition from intravenous to oral antibiotic therapy has been shown to reduce hospital length of stay, nosocomial infection risk, and cost, and improve patient satisfaction, these data demonstrate tebipenem's potential role as an oral transition agent from intravenous antibiotic regimens within the antibiotic stewardship paradigm.

Retrospective review of the use of daptomycin for diabetic foot infections.

The increase in antibiotic resistance has led to a search for alternative treatments for diabetic foot infections. This retrospective review outlines the clinical outcomes reported for a lipopetide for these infections.

Effect of serum from different patient populations on the serum bactericidal test.

We evaluated the effect of serum from normal and uraemic volunteers, neutropenic patients and burn patients on the serum bactericidal test. Serum samples were spiked with ceftazidime to mimic in vivo peak (75 mg/L) and trough (5 mg/L) concentrations. Serum inhibitory and bactericidal titres (SIT and SBT) were performed in triplicate using Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853. For E. coli, the trough SIT and SBT were significantly higher in serum from burn patients compared with normal volunteers (P < or = 0.024). The trough SIT was significantly higher in serum from burn patients compared with neutropenic patients (P = 0.022) and in uraemic patients compared with normal volunteers (P = 0.04). No significant differences between subject populations were found for P. aeruginosa.

Effect of removal of duplicate isolates on cumulative susceptibility reports.

The objective of our study is to assess the impact of different methods of duplicate isolate removal on cumulative susceptibility reports. Over a 1-year period, we studied the effect of 3 methods of duplicate isolate removal on the cumulative percentage susceptibility of 9 Gram-negative bacilli to 15 antimicrobials. Raw data from which no duplicate isolates were removed (NR) were generated by the Sensititre breakpoint susceptibility testing system. D3 and D7 were methods of duplicate isolate removal defined as follows: same patient, bacterial species, irrespective of susceptibility within either three (D3) or seven (D7) calendar days of the date of the previous culture. The third method evaluated was an algorithm utilized by Cerner, a laboratory management program that defines duplicate isolates as follows: same patient, bacterial species, and NCCLS susceptibility category to an individual antimicrobial. Differences in percentage susceptibility between the three methods of duplicate isolate removal and NR were assessed. The number of isolates studied ranged from 80 (E. aerogenes) to 681 (P. aeruginosa). Of the methods of duplicate isolate removal, the highest percentage susceptibility occurred most frequently with Cerner followed by D7 and D3. Differences in percentage susceptibility between methods of removal and NR ranged from -11 to 25%, -5 to 8%, and -3 to 10%, with Cerner, D3, and D7, respectively. The percentage susceptibility was at least 5% higher than NR with a method of removal for 15 individual organism/antimicrobial combinations in which susceptibility was > or = 70% by at least one of the methods. These occurred most frequently with Enterobacter species and Cerner. Although there is no consensus on the ideal method of duplicate isolate removal, one should be cognizant that these manipulations may produce different cumulative susceptibility reports.

Comparative in vitro pharmacodynamics of imipenem and meropenem against ATCC strains of Escherichia coli, Staphylococcus aureus and Bacteroides fragilis.

We evaluated the pharmacodynamics of imipenem and meropenem, utilizing time-kill studies over a concentration/MIC (C/MIC) range of 0.0625-1024 for E. coli ATCC 35218 (EC) and S. aureus ATCC 29213 (SA) and from 0.125-512 for B. fragilis ATCC 25285 (BF). Area under the time-kill curves were converted to percent response (%R). AUCs were calculated from drug concentrations corrected for degradation and %R vs. C/MIC and AUC/MIC were fit to a sigmoidal Emax model. Emax was similar for both agents for all organisms. Meropenem was 4x more potent than imipenem against EC based on the MIC and required 7 and 13.5-fold lower AUCs to achieve E50 and E90, (50% and 90% of the maximal response, respectively) respectively, whereas imipenem was 8x more potent than meropenem against SA based on the MIC and required 8 and 13-fold lower AUCs to achieve E50 and E90, respectively. The C/MIC and AUC/MIC to achieve E50 and E90 with imipenem were 2- and fourfold higher, respectively than meropenem against EC. There was less than a twofold difference in C/MICs and AUC/MIC between imipenem and meropenem for both E50 and E90 against SA. Against BF, concentrations and AUCs at E50 were similar for both agents; however, imipenem required 4 to 6-fold higher concentrations and AUC to achieve E90. Although there are differences in the potency of these agents as assessed by the MIC or AUC vs. response, when normalized by the MIC and corrected for drug degradation, these agents displayed similar pharmacodynamic parameter-response relationships.

Evaluation of antibiotic synergy against Acinetobacter baumannii: a comparison with Etest, time-kill, and checkerboard methods.

Acinetobacter baumannii is becoming increasingly resistant to antibiotics, often requiring combination therapy. Numerous methods exist to detect the presence of in vitro synergy with the time-kill and checkerboard tests being widely used. The Epsilometer test (E test) is a new method that is less labor intensive, but has not been evaluated using a wide range of antimicrobials and organisms. We assessed synergy using the time-kill and checkerboard tests and compared the results to the E test method using 10 clinical isolates of A. baumannii. Antimicrobial combinations evaluated consisted of trovafloxacin or tobramycin in combination with cefepime or piperacillin. Synergy was detected with all combinations by either the checkerboard or time-kill method. Synergy was not detected by the Etest method. The agreement between the time-kill test and Etest method was 72% (range 42-97%); for the time-kill and checkerboard tests, agreement was 51% (range 30-67%). The Etest method appears promising although further testing should be performed with additional antimicrobial agents and organisms.

Assessment of the relationship between antimicrobial usage and susceptibility: differences between the hospital and specific patient-care areas.

Current evidence suggests that controlling antibiotic resistance requires the monitoring of both susceptibility trends and antimicrobial usage within specific patient-care areas of the hospital. To assess the differences between antimicrobial usage-versus-susceptibility relationships found in the hospital and those relationships found in specific patient-care areas, susceptibility and antimicrobial usage data collected over a 5-year period (1992-1996) at the Medical University of South Carolina were analyzed. For each area, the relationship between drug use and susceptibility was analyzed for 8 gram-negative organisms with respect to 19 different agents and for 3 staphylococci with respect to 10 agents with use of simple linear regression. The relationships found in the hospital had a poorer overall agreement with the relationships found in the intensive care units (ICUs; <20%) than they did with the relationships found in the non-ICUs ( approximately 65%). Surveillance should include both susceptibility and drug usage patterns in individual areas within an institution.

Pharmacokinetics of cefepime in patients with thermal burn injury.

The pharmacokinetics of cefepime following administration of a single 2-g dose were evaluated for 12 adult patients with thermal burn injury and suspected or documented infection. Serial blood and urine samples for cefepime concentration determination were obtained for 24 h following drug administration. Serum and urine cefepime concentrations were determined by high-performance liquid chromatography and serum concentrations were fit to a two-compartment pharmacokinetic model. Mean (standard deviation [SD]) age, actual body weight (ABW), percent total body surface area burned, and days postburn at the time of study were 41 (13) years, 84 (22) kg, 36 (17)%, and 9 (3) days, respectively. Mean (SD) measured creatinine clearance (CL(CR)), total clearance (CL(T)), renal clearance (CL(R)), alpha phase half-life, beta phase half-life, and volume of distribution at steady state (V(SS)) were 135 (31) ml/min, 8.8 (2.4) liters/h, 8.1 (2.0) liters/h, 0.33 (0.14) h, 2.8 (0.6) h, and 0.43 (0.10) liters/kg ABW, respectively. Cefepime CL(T) and CL(R) in burn patients were similar to previously reported values for healthy volunteers when normalized by CL(CR). Stepwise multiple regression was used to associate CL(T) with CL(CR) and days postburn (r(2) = 0.861), CL(R) with CL(CR) and days postburn (r(2) = 0.773), nonrenal clearance with percent third-degree (% 3 degrees ) burn and albumin concentration (r(2) = 0.550), and V(SS) only with % 3 degrees burn (r(2) = 0.624). Simulated steady-state serum concentrations obtained by using the patients' pharmacokinetic parameters exceeded the susceptibility interpretive standard (breakpoint) of cefepime for at least 60% of the dosing interval with dosing regimens of 1 g every 8 h (q8h), 2 g q8h, and 2 g q12h. Despite differences in pharmacokinetic parameters between our patients and healthy volunteers, it appears that these dosing regimens may be adequate in similar burn patients.

Impact of use of multiple antimicrobials on changes in susceptibility of gram-negative aerobes.

Evaluation of antimicrobial usage vs. susceptibility relationships typically involves single agents. However, susceptibility profiles may be affected by multiple drugs. From 1992 through 1996, we studied relationships between drug usage and the susceptibility (only susceptibility rates of > or = 70%) of Acinetobacter anitratus (baumannii), Enterobacter aerogenes, Escherichia coli, Enterobacter cloacae, Klebsiella pneumoniae, Pseudomonas aeruginosa, Proteus mirabilis, and Serratia marcescens to 22 agents. Linear regression was used to assess usage of each agent vs. susceptibility to it and to all agents. Only relationships with a coefficient of determination of > or = 0.5 and a negative slope were evaluated and classified as increasing drug use and decreasing susceptibility (increasing D, decreasing %S) or decreasing drug use and increasing susceptibility (decreasing D, increasing %S). The mean numbers (range) of drugs associated with a change in susceptibility were 1.7 (0-14) and 0.6 (0-7), respectively, for increasing D, decreasing %S and decreasing D, increasing %S relationships. Multiple antimicrobials are associated with susceptibility to other drugs; thus, surveillance of these relationships should not be limited to single drugs.

Comparative bactericidal activity of ceftazidime against isolates of Pseudomonas aeruginosa as assessed in an in vitro pharmacodynamic model versus the traditional time-kill method.

Bactericidal activity, historically assessed by in vitro tests which employ fixed drug concentrations, may also be evaluated in in vitro pharmacodynamic models in which in vivo pharmacokinetics and bacterial growth conditions can be simulated. However, systematic comparisons between the two methods are lacking. We evaluated the bactericidal activities of ceftazidime, at two different concentration/MIC ratios (C/MICs), against 10 clinical isolates of Pseudomonas aeruginosa in a two-compartment model with continuous-infusion conditions and a 2-h half-life. These values were compared to those determined by traditional 24-h time-kill (TTK) methods at the same C/MICs. Bactericidal activities were compared by using area under the colony count-time curves. Antibiotic exposure (area under the drug concentration-time curve) was also evaluated. Although bactericidal activity appeared greater by the TTK method (P = 0.05), when it was normalized for drug exposure, these differences disappeared (P = 0.2). This disparity was likely due to differences in drug exposure in the TTK method and in the peripheral compartment of the model (site of bacteria) over the first 8 h of the experiment, during which the antibiotic accumulated to target concentrations. This suggests that the bactericidal effects with constant antibiotic concentrations are similar in the two methods; however, this may not hold true with fluctuating drug concentrations. Further, results from the pharmacodynamic model may theoretically be more relevant, as in vivo pharmacokinetics and bacterial growth conditions call be more faithfully simulated.

In vitro pharmacodynamics of ceftazidime against Pseudomonas aeruginosa isolates from cystic fibrosis patients.

The concentration/MIC (C/MIC) ratio maximizing the bactericidal activity of ceftazidime against 10 Pseudomonas aeruginosa isolates from cystic fibrosis patients was identified. Bactericidal activity was assessed by determining the percent difference in the area under the killing curve at each C/MIC ratio for all of the isolates from that of their growth control. The percent effect at each C/MIC ratio was fitted to a sigmoidal Emax model with maximum bactericidal activity defined as the C/MIC ratio that produced an effect that was 90% of the Emax. Our results suggest that at least some isolates may require higher C/MIC ratios than previously reported for maximal activity.

The costs of burn care: an analysis with an emphasis on the use of parenteral antimicrobials.

Because infection is a common cause of morbidity and mortality in patients with burns and intensive antibiotic therapy is often required, the focus of this study was to describe the patterns of use and costs of parenteral antibiotics in a burn unit. The study also evaluated the overall economics of burn care in our population. Forty-one percent of the study group received parenteral antimicrobial agents; the specific agents, indications, and costs are described. The costs of parenteral antibiotics made up a negligible 1.2% of hospital costs and only 12.4% of pharmacy costs. This study revealed the continued hospital losses related to the reimbursement system used by diagnosis-related groups for patients with Medicare. If all patients studied (n = 61) were reimbursed under diagnosis-related groups the unit would have experienced an annual loss of approximately 1.2 million dollars. If specialized burn care facilities are to remain, it may be necessary to reevaluate the appropriateness of the diagnosis-related group reimbursement system for burn-related injury. This is especially important if all third-party reimbursement sources consider conversion to this system of compensation.

Drug degradation during HPLC analysis of aztreonam: effect on pharmacokinetic parameter estimation.

OBJECTIVE: To assess the impact of degradation of aztreonam, a beta-lactam antibiotic, during HPLC analysis on pharmacokinetic parameter estimates. METHODS: The baseline (B) serum concentration-time data from a published pharmacokinetic study of aztreonam were degraded using first-order equations and a degradation rate constant (0.014 h-1) determined from a preliminary degradation study. Samples were mathematically degraded for autosampler run times of 8-13 h (D1) to approximate a normal work day and for autosampler run times of 16-17 h (D2) and compared with B data. It was assumed that B data were nondegraded and that changes in chromatography were the result of degradation of azetreonam and not to any changes in chromatographic conditions. A two-compartment model was used to fit the data and pharmacokinetic parameters were calculated using standard equations. Statistical significance between all pharmacokinetic parameters for B and D1 and B and D2 was determined using the paired, two-tailed Student's t-test. RESULTS: Increased variability was noted for all pharmacokinetic parameters for D1 and D2 compared with B. Statistically significant differences were found for clearance (B < D1, p = 0.0095 and B < D2, p = 0.0194), steady-state volume of distribution (B < D2, p = 0.0392), and area under the serum concentration-time curve (B > D1, p = 0.0497). CONCLUSIONS: Aztreonam degradation resulted in increased variability in pharmacokinetic parameter estimates. Investigators should be cognizant of this and preliminary studies should be performed to characterize degradation for the length of the expected autosampler run.

Impact of different statistical methodologies on the evaluation of the in-vitro MICs for Bacteroides fragilis of selected cephalosporins and cephamycins.

The purpose of this study was to assess the effect of different MIC statistics on the in-vitro evaluation of cefoxitin, cefotetan, ceftizoxime, cefotaxime, desacetylcefotaxime, and cefotaxime:desacetylcefotaxime (1:1 ratio) against clinical isolates of Bacteroides fragilis from the Medical University of South Carolina. MICs were determined by the agar dilution method following NCCLS guidelines. Statistical analyses included arithmetic and geometric means, median, mode, MIC50, MIC90, and range. Differences between antimicrobial agents were determined using parametric and nonparametric methods. Consistent with previous in-vitro studies on anaerobes with these agents, a wide range in the MICs was observed. The arithmetic mean MIC was lowest for cefotetan, but the geometric mean MIC was lowest for ceftizoxime. Using the MIC90, cefotetan was at least a two-fold dilution more active than the other agents. After natural logarithmic transformation of the MIC data, analysis of variance with the Scheffé post-hoc test demonstrated that the MICs of ceftizoxime were significantly lower than those of cefoxitin (P < 0.001), cefotetan (P < 0.05), cefotaxime (P < 0.05), and desacetylcefotaxime (P < 0.001). Median and mode MICs were lowest for ceftizoxime and cefotaxime:desacetylcefotaxime. Using published breakpoints, cumulative percent susceptibility was similar for all agents studied. In this analysis of the antimicrobial susceptibility of B. fragilis in our institution, the in-vitro activity of the cephalosporins and cephamycins tested appeared to be very similar when all statistical data were evaluated. Since apparent in-vitro activity may be influenced by the statistic evaluated, we suggest that all MIC data be reported to allow for a more complete analysis of microbiological potency.

Pharmacoeconomic evaluation of treatment of penetrating abdominal trauma.

The hospital, pharmacy, and antibiotic costs for patients with penetrating abdominal trauma were compared with reimbursement received; these costs were also analyzed to assess the potential impact of a total prospective pricing system (PPS). During a four-year period, 46 patients admitted solely for penetrating abdominal trauma were retrospectively evaluated: their discharge summaries indicated that, for 9 patients, reimbursement was based on diagnosis-related groups (DRGs) under the PPS; 9 patients had private insurance; and 28 were classified as "self-paying/no insurance." All costs, corrected for inflation, were reported in 1989 dollars. Antibiotics represented 22.5%, 1.7%, and 0.5% of pharmacy, hospital, and DRG reimbursement, respectively; pharmacy costs were 8.5% of hospital costs and 2.3% of DRG reimbursement. For all 46 patients, a net loss of $295 per patient was incurred. Four patients accounted for 43% of the hospital costs. If the hospital had been reimbursed for all of these patients by prospective pricing and DRGs, it would have had a median profit of $9730 in 42 of 46 patients. Costs exceeded DRG reimbursement in the remaining four patients by a median of $8210. Antibiotic costs and pharmacy costs represent a small portion of hospital costs and DRG reimbursement for patients with penetrating abdominal trauma; thus, cost containment efforts in these patients should be directed at other ancillary services and length of stay.

Evaluation of cephalosporins/cephamycins with antianaerobic activity by integrating microbiologic and pharmacokinetic properties.

When evaluating antimicrobial agents, in vitro microbiologic activity and pharmacokinetics are important factors, but these data are usually not assessed simultaneously. The purpose of the study was to compare cefoxitin, cefotetan, ceftizoxime, cefotaxime (CT), desacetylcefotaxime (DACT), and CT/DACT (1:1 ratio) by integrating their microbiologic activity against clinical isolates of Bacteroides fragilis with their pharmacokinetic properties. Minimal inhibitory concentrations (MIC) were determined by the agar dilution method. Steady-state serum concentration--time profiles were simulated for 2-gm doses in a 70-kg patient using ADAPT software and pharmacokinetic data from published studies. Serum protein binding (%) of each agent was also obtained from published studies and used to calculate the unbound serum concentration--time profiles. As estimates of pharmacodynamic activity, time below the MIC (T less than MIC) and percentage of the dosing interval below the MIC (% INT less than MIC) were calculated for individual isolates using total and unbound serum concentrations. Data analysis included MIC50, MIC90, range, breakpoint susceptibility, and analysis of variance for T less than MIC and % INT less than MIC (Scheffé post-hoc test, P less than 0.05). The MIC90 of cefotetan was at least a twofold dilution lower than the other agents. However, using unbound (pharmacologically active) serum concentrations, T less than MIC and % INT less than MIC for ceftizoxime (at a simulated eight-hour dosing interval) were significantly smaller than with the other antibiotic regimens. Integration of in vitro and pharmacokinetic data may provide additional information to assist in the evaluation of antimicrobials. For B fragilis from our institution, the pharmacodynamic profile of unbound ceftizoxime is superior to the other antianaerobic cephalosporins/cephamycins tested.

Pseudoresistance of Pseudomonas aeruginosa resulting from degradation of imipenem in an automated susceptibility testing system with predried panels.

During the use of a single lot of custom breakpoint panels (Sensititre; Radiometer America Inc., Westlake, Ohio), imipenem susceptibility declined from 70 to 44% for clinical isolates of Pseudomonas aeruginosa. With a new lot, susceptibility increased to 73%. Subsequent evaluations with P. aeruginosa ATCC 27853 revealed a similar susceptibility pattern and an increase in the MIC of imipenem when determined in panels with increasing ages. Imipenem concentrations were determined by high-performance liquid chromatography by using 11 different lots of MIC and breakpoint panels (139 to 893 days of age). The amount of imipenem remaining declined from 94.4 to 13.8% (r = 0.9225) over the age range of the panels. These data suggest that imipenem in Sensititre MIC and breakpoint panels degrades over time and that the decrease in imipenem may be largely responsible for the decline in P. aeruginosa susceptibility.

Aztreonam pharmacokinetics in burn patients.

The pharmacokinetics of aztreonam in eight adult patients with severe burn injuries (total body surface area burn, 49% +/- 21% [mean +/- standard deviation]) were studied. The time of initiation of study following burn injury was 7.0 +/- 1.4 days. Four patients at first dose and at steady state were studied. Aztreonam concentrations were measured by high-performance liquid chromatography, and a two-compartment model was used to fit the data. No significant differences in any pharmacokinetic parameters between first dose and steady state were observed. Volume of distribution of the central compartment after first dose (0.14 liters/kg) and volume of distribution at steady state (0.31 liters/kg) were approximately 30% higher than those reported for other patient populations. Total drug clearance and renal drug clearance when normalized to creatinine clearance (CLCR) were similar to those previously reported for other critically ill patients. CLCR was strongly correlated with renal drug clearance (r = 0.94) and total drug clearance (r = 0.95). The extent and degree of burn (percent second or third degree burn) were poorly correlated with all pharmacokinetic parameters with the exception of the volume of distribution at steady state, which was correlated with both total body surface area burn (r = 0.95) and percent second degree burn (r = 0.83). Aztreonam pharmacokinetics are altered as a result of thermal injury; however, CLCR can be used to assess the clearance of aztreonam in burn patients.

Antibiotic prophylaxis and tourniquet inflation in total knee arthroplasty.

Twenty-four patients receiving total knee arthroplasty (TKA) were randomized into one of three groups based on tourniquet inflation one, two, or five minutes after administration 1 g cefazolin. Simultaneous serum, soft-tissue, and bone samples were obtained at regular intervals during surgery. All soft-tissue and bone samples were corrected for cefazolin content. The percentage of cefazolin penetration into soft tissue and bone was calculated using the area under the concentration time curve. Adequate cefazolin concentrations for soft tissue and bone were defined as greater than or equal to 4 x minimum inhibitory concentration90 (MIC90 = 1 microgram/ml) of cefazolin to Staphylococcus aureus and coagulase-negative staphylococci. Patients were similar in age, actual body weight, creatinine clearance, and length of tourniquet inflation. The median percentage of cefazolin penetration into soft tissue and bone for the five-, two-, and one-minute groups was 14.5% and 4.6%, 6.7% and 3.0%, and 5.9% and 4.6%, respectively; the percentage of penetration into soft tissue between the five- and one-minute groups was statistically significant. A higher percentage of patients achieved the desired cefazolin concentration (greater than or equal to 4 micrograms/g) if a five-minute interval was selected. The five-minute group achieved the highest mean ratios of concentration to MIC compared with the two- and one-minute groups, although the differences were not statistically significant. The standard 1 g of cefazolin with a five-minute interval between administration and tourniquet inflation resulted in adequate mean soft-tissue and bone concentrations for prophylaxis during TKA with a tourniquet time less than two hours. Additional doses are not warranted after tourniquet release.

Effects of supraphysiologic temperature and broth dilution on serum protein binding.

Pooled human serum, unheated (UU) and heated at 56 degrees C for 30 (HU-30) and 60 (HU-60) min, and these media diluted 1:1 with broth (UD, HD-30, and HD-60) were used to assess the effects of supraphysiologic temperature and broth dilution on the serum protein binding of ceftriaxone. Protein binding was determined by ultrafiltration and subsequent high-performance liquid chromatography analysis. Significant differences in protein binding between UU and HU-60 (P less than 0.05), UD and HD-60 (P less than 0.01), HD-30 and HD-60 (P less than 0.01), and all undiluted and diluted media (P less than 0.01) were found. These alterations in protein binding may influence the in vitro microbiologic testing of highly protein-bound antibiotics.