clindamycin in patients with penetrating abdominal trauma

Superiority of Aztreonaml Clindamycin Compared With Gentamicin/ Clindamycin in Patients With Penetrating Abdominal Trauma Timothy C. Fabian, MD, Mary M. Hess, PhD, Martin A. Croce, MD, Memphis,Tennessee, Robert s. Wilson, MD, Detroit, Michigan, Samuel E. Wilson, MD, Irvine, California, &Ott L. Charland, PhD, Philadelphia, Pennsylvania, John H. Rodman, PhD, Bradley A. Boucher, PhD, Memphis, Tennessee

There were 73 evaluable patients entered into a prospective, double-blinded trial comparing aztreonam/clindamycin (A/C) to gentamicin/clindamycin (G/C) for the prevention of infection after penetrating abdominal trauma. Aztreonam was administered at a dosage of 2 g every 8 hours and gentamicin at 5 mg/kg for the 6rst 24 hours and then adjusted by serum monitoring to a peak of 6 to 8 gg/mL and a trough of less than 2 pg/mL; all patients received 900 mg of clindamycin every 8 hours. Patients with colon wounds received 4 days of antibiotics, and the remaining patients received a 24-hour course. Gunshot wounds occurred in 69% of patients: 74% of all patients had some hollow viscus injury, and 26% had only solid viscus injury. The groups were well matched according to abdominal trauma index, percentage with colon injury, and transfusion requirements. Failures occurred in eight patients (11%): two wound infections, five intra-abdominal infections, and one case of necrotizing fasciitis. Seven infections occurred in 36 (19%) G/C patients compared with 1 in 37 (3%) A/C patients (p < 0.03). The hospital stay was 12 2 11 days for G/C patients and 8 + 7 for A/C patients (p < 0.12). The superiority of the A/C regimen may be partially attributable to relative underdosing of gentamicin (approximately half of the patients had inadequate levels after 24 hours) combined with a favorable pharmacokinetic profile (signiikantly prolonged half-life) of aztreonam in this clinical setting.

From the Departments of Surgery (TCF, MAC) and Clinical Pharmacy (MMH, JHR, BAB), University of Tennessee, Memphis, Tennessee, the Department of Surgery (RSW), Wayne State University, Detroit, Michigan, and the Department of Surgery, Universitv of California, Irvine (SEW), Irvine, California, and the Department of Pharmacy Practice and Pharmacy Administration, Philadelphia College of Pharmacy and Science (SLC), Philadelphia, Pennsylvania, and St. Jude Childrens’ Research Hospital (JHR), Memphis, Tennessee. This study was sponsored in partby a grant from Bristol-Myers Squibb Company and the Center for Pediatric Pharmacokinetics and Therapeutics, University of Tennessee, Memphis, Tennessee. Requests for reprints should be addressed to Timothy C. Fabian, MD, 956 Court Avenue, Room G210, Memphis, Tennessee 38163. Manuscript submitted June 18, 1993, and accepted in revised form November 2,1993.

W

hat has been

referred to as “an epidemic of violence” over the past 15 years in the United States has shown no recent signs of abatement. There were 17,670 deaths resulting from firearms and sharp instruments in 1991 in the United States compared with 14,255 in 1987 [I]. The surge of violence has been precipitated by complex socioeconomic circumstances, but it seems clear that the culture of illicit drug use is intimately entwined. For every death, there are approximately five patients who survive and require hospitalization. Thus, there are roughly 88,350 hospitalizations for penetrating trauma annually in this country. In addition to the resulting social impact, the financial burden is enormous. A substantial amount of the required care is uncompensated. Efforts to reduce morbidity and consequently decrease lengths of hospital stay are warranted. Infection remains the most serious threat to patients who survive the initial insult. Septic complications occur in about 10% to 15% of those patients [2]. Antibiotics continue to be the main adjunct to prompt, definitive surgery for minimizing sepsis after abdominal injury. Regimens are required that are effective against both anaerobic organisms and Enterobacteriaceae since 20% of patients will experience colonic injury. The combination of gentamicin and clindamycin has been a standard combination for treatment of intra-abdominal infection as well as for prevention of infection after abdominal injury. Aztreonam has been approved by the Food and Drug Administration as an injectable antibiotic for the treatment of abdominal sepsis. It is a monocyclic p-lactam with a half-life of 1.7 hours in normal volunteers and a spectrum that is quite similar to aminoglycosides for gram-negative bacteria. There are limited studies to date evaluating aztreonam pharmacokinetic disposition in the critically ill patient population. This trial was undertaken to evaluate the safety, efficacy, and pharmacokinetics of aztreonam in combination with clindamytin (A/C) compared with gentamicin and clindamycin (G/C) in preventing septic complications after penetrating abdominal trauma. PATIENTS

AND METHODS

The study was a multicenter trial conducted in Memphis, Detroit, and Los Angeles. The pharmacokinetic arm of the study was conducted exclusively at the Memphis location. It was performed in a prospective, double-blinded fashion. Randomization was accomplished by computer generated sheets kept in the hospital pharmacies. The study was initiated August 1991 and concluded April 1992. The population included adult

THE AMERICAN JOURNAL OF SURGERY

VOLUME 167 MARCH 1994

291

patients (older than 18 years) with penetrating wounds requiring laparotomy. Those patients with negative laparotomy and those who died within 48 hours of injury of hemorrhagic shock were excluded. Patients were also excluded from analysis of efficacy of antibiotics if there were major dosing violations (ie, missed doses) or if additional antibiotics were administered during hospitalization. The doses for gentamicin were 5 mg/kg in three doses (2 mg/kg first dose) over the first 24 hours; subsequent doses were based on serum levels to maintain a peak of 6 to 8 kg/mL and a trough of less than 2 kg/mL [3]. Aminoglycoside concentrations were measured before and after infusion using fluorescence polarization immunoassay (TDx, Abbott Laboratories). Subsequent doses were prescribed in patients with colon injury by the clinical pharmacy services assuming a one compartment model using the method of Sawchuk and Zaske [4]. These levels were not reported in the patient record in order to maintain investigator blinding. Aztreonam dosage was 2 g every 8 hours. All patients received 900 mg of clindamycin every 8 hours. Initial doses were administered shortly after hospital admission. Patients without colbn injury received antibiotics for 24 hours, and those with colon injury received them for 4 days. Blood samples for determination of aztreonam concentrations were collected prior to infusion and at 1, 3, and 7 hours after infusion. All patients were studied following one of the aztreonam maintenance doses within 48 hours after enrollment. Those patients with colon injury were also studied after 96 hours. Samples were obtained from indwelling arterial catheters whenever possible. Alternatively, samples were obtained from a central venous catheter or by direct venipuncture. Each sample was immediately centrifuged, and the harvested serum was stored at -70°C. Aztreonam concentrations were assayed using high pressure liquid chromatography (HPLC). After thawing, serum proteins were precipitated with 100 FL of acetonitrile, vortexed for 30 seconds at room temperature, and then centrifuged for 3 minutes. Fifty FL of the clear supernatant was then injected onto a reverse phase HPLC system. The system consisted of a Selectosil Cl8 lo-micron analytical column (Phenomenex, Torrance, California) with a Cl8 lo-micron guard column (Alltech, Deerfield, Illinois) and a System Gold ultraviolet 166 detector (Beckman Instruments, Fullerton, California) set at 293 nm. The mobile phase consisted of acetonitrile and tetrabutylammonium hydrogen sulfate plus ammonium sulfate buffer with pH adjusted to 3.0. The flow rate was set at 2.0 mL/min. Average elution time was approximately 6 minutes, with an average run time of 10 minutes. The quantitation limits of this assay were linear between 2 and 100 kg/mL with the between-day coefficient of variation ranging from 7.6% at 7.5 bg/mL to 2.9% at 75 pg/mL. The lowest detectable concentration was 0.5 pg/mL. Individual aztreonam pharmacokinetic parameters were determined using Bayesian estimation. A one-compart-

292

THE AMERICAN

JOURNAL OF SURGERY

ment model was fit to the data using the ADAPT II pharmacokinetic software package (University of Southern California, Los Angeles, California) and previously published mean data for the population priors (ie, volume = 0.24 L/kg, elimination rate constant = 0.42 hr-‘) [5]. Antibiotic failures were classified as wound infections, intra-abdominal infection, or necrotizing fasciitis of the abdominal wall. The skin and subcutaneous fat were left open in the face of significant gastrointestinal contamination. Data analyzed included age, mechanism of injury, organs injured, abdominal trauma index, transfusion requirements within 24 hours of injury, complete blood count, liver function, blood urea nitrogen and creatinine levels, length of hospitalization, and mortality. Statistical analyses compared the following: those patients with and without hollow viscus injury, those with and without infection, and the two antibiotic regimens. x2 analysis and Fisher’s exact test were used for discrete variables; continuous variables were evaluated by unpaired two-tailed Student’s t test. Mean individual pharmacokinetic parameter estimates were compared with those of normal volunteers using an unpaired Student’s t test. Intrasubject mean pharmacokinetic parameter estimates were compared by a paired Student’s t test for day 1 versus day 4. A simple linear regression of aztreonam clearance and elimination rate constant versus creatinine clearance as estimated using the Cockcroft-Gault equation was also performed [6]. The study was approved by the institutional review boards of the participating institutions. Consent was obtained for all patients entered into the trial. RESULTS There were 119 patients enrolled in the study: 58 G/C and 61 A/C. Forty-six were excluded: 29 patients had no intra-abdominal injury, 10 received adjunctive antibiotics, 4 died within 48 hours due to massive hemorrhage, and 3 had major dosing violations; there were 2 failures (1 in each study arm)-both received inadequate antibiotic doses. This left 36 evaluable G/C patients and 37 A/C patients. The average age was 29 years. Gunshot wounds occurred in 69%, and stab wounds in 31%. Hollow viscus injuries occurred in 74% of patients, whereas 26% had injuries confined to solid viscera. Table I lists the organs injured in those with and without hollow viscus injury. There were eight antibiotic failures (11%). Failures included two wound infections, four intra-abdominal abscesses, one diffuse peritonitis, and one necrotizing fasciitis of the abdominal wall. The organisms involved were Escherichia coli (3 cases), Klebsiella species (3), Staphylococcus species (3), Streptococcus viridans (2), Bacteroides species (l), Enterococcus species (l), Citrobacter fieundii (l), Enterobacter aerogenes (l), and Proteus mirabilis (1) . The comparison of patients with and without hollow viscus injury is shown in Table II. Significantly more patients with hollow viscus injury had gunshot wounds, and there was a consistent trend toward a greater degree of injury comparing abdominal trauma index scores,

VOLUME

167

MARCH 1994

PENETRATING

TABLE I Organ8 InJuredIn Patkmta Wtth and Without Hollow

w!Bcus In)ury Hollow Yiscus Injury

No Hallow viscus Injury

No. Colon (%) Small bowel (%) Stomach (%)

54 31 65 20

19 -

Lifer (%) Kidney (%f Pancreas (%) Spleen (94) Major vascular (%)

20 13 6 9 13

53 16 11 26 11

Hollow Viscus Injury

No Hollow Viscus Injury

29 76

29 47

162 12 48 19 31 1.1 + 2.6 fl k9 15

1329 42 11 0 0.4 L 0.9 829 0

Age 56 gunshot* ATI %ATI > 15 %ATI >25 % colon injury Blood during initial 24 hr (U) Hospital stay (df % infectlon ATI = abdominaltraumaindex. lp < 0.03.

TABLE III Comparlson of Petlenb~Wlth and Wlthout Infection Infection

No.

Age % gunshot ATI %ATl r15 %ATI >25 % &on injury Blood during initial 24 hr (U) Hospital stay (d)

a 36 88 31 + 12 88 50 50 2.1 + 4.2 242 14

No Infecfion

p-Value

65 28 66 14 + 9 40 12 20 0.7 k 2.0 a&7

NS 6.42 0.0001 0.02 0.02 0.08 0.13 0.0001

All = abdominaltraumaindex:NS = not significant.

transfusions, and length of hospitalization. All of the failures occurred in patients with hollow viscus injury (p ~0.12). Four of 17 patients (24%) with colon wounds, 2 of 35 (6%) with small bowel injuries, and 2 of 11(18%) with gastric injuries developed infection. The comparison of patients who became infected versus the uninfected patients is shown in Table III, One third of the patients with an abdominal trauma index

TRAUMA

TABLE IV Comparlson of the Two Antlblotlc Reglmens and Exclusions

No. Age % gunshot ATI %ATI > 15* % ATI > 25t % colon injur# Blood during initial 24 hr (U) Hospital stay (d) % infections

TABLE II Comparison of Patlents With and Wlthout Hollow Vlrcus Injury

ABDOMINAL

Aztreonam and Clindamycin

Gentamicin and Clindamycin

Exclusions

37 31 65 15 * 10 41 ?6 22 0.9 2 2.4

36 28 72 16 2 12 50 17 25 0.8 2 2.3

46 30 48 9L 17 22 13 7 1.6 L 4.6

8~7 35

122 11 195

9-t

14 4

ATl= abdominaltraumaindex.

lAztreonam/clindamycin: ures (n = t8); exclusions:

0 failures (n = 15); gentamiciniclindamycin:

7 fail-

1 failure (n = 10).

tktreonam/clindamycin:Clfailures(n = 6): gentamicin/clindamycln: 4 failures (n = 6); exclusions:f failure(n = 6). k?treonamlclindamycin: 0 failures (n = 8); gentamiciniclindamycin: 4 failures (n = 9); exclusions:1failure(n = 3). ‘p < 0.03.

score above 25 developed infection. A reflection of the morbidity associated with septic complications is the fact that the average hospitalization was prolonged by 2 weeks in patients with infection (p < 0.0001). A comparison of the two antibiotic groups and the excluded patients is shown in Table IV. The study groups are well matched for mechanism and degree of injury, Similarly, incidence of colon wounds and transfusion requirements are not different. The excluded patients were less severely injured than those in the study groups, with the exception of the four early deaths. Seven of the eight failures occurred in G/C patients (p <0.03). A/C patients required an average of 4 fewer days of hospitalization (p <0.12). There ‘was one death in a patient (G/C) with necrotizing fasciitis who subsequently developed multiple organ failure. Using “intent to treat” analysis that includes excluded patients, there were 61 A/C and 58 G/C patients. The infection rate was 3% for A/C patients and 14% for G/C patients (p < 0.05). Twenty-three of the 37 A/C patients (22 male, 1 female) had blood sampIes obtained for estimation of aztreonam pharmacokinetic parameters. In this subset of patients, the mean age was 31 years, actual body weight was 81 kg, and estimated chromium chloride was 102 mL/min. The mean (-I- SEM) perioperative aztreonam concentration versus time data for the patients during the first 48 hours of therapy are illustrated in Figure 1. A mean peak concentration of aztreonam of 75 pg/mL was achieved, and, after 8 hours, a mean trough concentration of 15 pg/mL was found. The mean tr/2 estimate was significantly longer in the patients with penetrating abdominal trauma compared with normal volunteers, 2.2 versus 1.65 hours, respectively (p < 0.05). Mean V and Cl estimates were not significantly different

THE AMERICAN JOURNAL OF SURGERY

VOLUME

167

MARCH 1994

293

from normal volunteers (V = 0.26 versus 0.24 L/kg; Cl = 1.41 versus 1.67 mL/min/kg) (p >0.05). In the patients with colon injury studied on two separate occasions (n = 5) mean ti/2 increased from 1.84 to 2.17 hour, V increased from 0.25 to 0.32 l/kg, and Cl increased from 1.33 to 1.76 mL/min/kg. However, none of these changes were significantly different (p > 0.05). Additionally, no significant relationship was observed between aztreonam clearance and estimated creatinine clearance (r = 0.278, p >0.05) or the elimination rate constant (r = 0.004, p >0.05). The aztreonam failure occurred at one of the study sites where pharmacokinetic parameters were not estimated. In the G/C patients, 23 of 36 patients had evaluable concentration data sets. Eleven of the G/C patients (48%) were found to have subtherapeutic maximum gentamicin concentrations at 24 hours requiring an increase in dosage for colon injury patients; no patient required a decrease in dosage. The overall mean (*SD) peak gentamicin concentration was 5.8 (1.6) ug/mL (Table V). Six of the seven G/C treatment failures occurred in the 23 patients with evaluable data. Fifty percent of the patients who experienced a failure of therapy had subtherapeutic concentrations at 24 hours, and 47% of the patients in the success group were subtherapeutic at that time. No relevant differences were found in laboratory parameters between the two treatment groups. COMMENTS The majority of comparative able to demonstrate significant

01

I

0

8

I

antibiotic trials are not differences in the rela-

I

I

32

I

40

Figure 1. Mean (SEM) aztreonam concentrations (open dlamonds) at 1, 3, and 7 hours around the first study dose for 22 patients with intra-abdominal penetrating trauma injury. Aztreonam concentrations were obtained around dose #2 (n = 3), dose #3 (n = lo), dose #4 (n = 6) and dose #5 (n = 3). (One patient with concentrations around dose #7 was excluded.) The solld line is a simulation of aztreonam concentration versus time using study mean pharmacokinetic parameter estimates.

294

THE AMERICAN

JOURNAL OF SURGERY

TABLE V Mean (STD) and Range of Peak Gentamlcln Concentrations (pg/mL) in Evaluable Study Patients Target Concentration Status*

Outcome Success

Failure

Total

*Therapeutic

Therapeutic

Subtherapeutic

Total

7.1 (0.88) 6.2-9.1 n=9 6.9 (0.6) 6.2-7.4 n=3 7.0 (0.8) 6.2-9.1 n = 12

4.6 (0.91) 3.0-5.7 n=8 4.3 (1.4) 2.6-5.2 n=3 4.5 (1 .O) 2.6-5.7 n = 11

5.9 (1 S) 3.0-9.1 n = 17 5.6 (1.8) 2.6-7.4 n=6 5.8 (1.6) 2.6-9.1 n = 23

defined as a

equal to 6 pg/mL,

peak gentamicin concentration of greater than or

subtherapeutic

defined as peak gentamicin concentration

of

less than 6 kg/mL

tive efficacy of the regimens. This is largely because the advanced-generation cephalosporins, expanded penicillins, quinolones, carbapenems, and aminoglycosides achieve excellent coverage of the aerobic gram-negative rods that are so ubiquitous as pathogens in surgical infection. These agents alone or in combination with anaerobe-specific antibiotics (clindamycin or metronidazole) also eradicate the anaerobic organisms, which are the other major group of pathogens. Most trials contain small numbers of patients, which leads to two problems. First, those patients at greatest risk for infection constitute only a small percentage of the total [6]. Second, real differences can be missed by a p-error of statistical interpretation. Recognizing these facts, we were somewhat surprised at finding a significant difference in outcome in this relatively small trial. There are several possibilities that must be considered when a difference is detected: (1) investigator bias; (2) mismatched patient groups; (3) o-error of statistical interpretation; (4) differences in bacterial spectrum; or (5) inadequate dosing. Consideration of these possibilities will be the foundation for these comments. Investigator bias may occur when the study is conducted in an open fashion. This has the potential to influence randomization, affect adjunctive therapy, or influence the interpretation of results. This does not call into question the ethics of the investigation, but, given the complexities and multiple personnel involved in various stages of clinical trials, occasionally decisions can be made that provide an advantage to either the experimental or standard arm. An example would be in the interpretation of outcome in a pneumonia study. Clearly, the diagnosis is not always obvious due to confusion with atelectasis, pulmonary contusion, and adult respiratory distress syndrome, especially since bacterial colonization occurs in all of these. A doubleblinded study should eliminate this source of error in interpretation. There was a good deal of discussion among the investigators prior to initiation of this study concerning the logistical problems with a double-

VOLUME 167

MARCH 1994

PENETRATING

blinded study in this scenario. Many of the patients are enrolled during awkward hours for hospital staffing, and the aminoglycoside arm further compounded the difficulty by requiring the hospital pharmacy to monitor the levels and independently adjust dosages. Fortunately, both regimens were delivered every 8 hours. The decision to conduct the trial in a blinded fashion was justified by the excellent coordination provided by the pharmacies. The blinding was maintained until after interpreting the results, thus, eliminating investigator bias as an explanation for the differences observed. Whether open or blinded, there is always a chance that, during randomization, an unequal proportion of high-risk patients will be assigned to one group. Degree of injury, as measured by the abdominal trauma index, consistently correlates significantly with infection [7-91. Likewise, colon injury and transfusion requirements are significantly associated with septic complications [2,9]. The data from this trial further substantiate those parameters as high risk (Table III). Comparing these parameters between the two antibiotic groups demonstrated no differences in randomization (Table IV). Thus, the groups were well matched for risk, eliminating disproportionate randomization as an explanation for superior results with A/C. A significance level of p < 0.03 leaves open the possibility of an o-error of interpretation. A larger study would be required to further address that possibility. Turning to the issues relating specifically to the antibiotic, spectrum and dosing must be addressed. Both gentamicin and aztreonam have poor anaerobic coverage. However, the addition of an equal amount of clindamycin to each overcame this shortcoming. For several years, aminoglycosides have been the yardstick by which other antibiotics are measured for their activity against aerobic gram-negative rods. Enterobacteriaceae are the major aerobes involved in abdominal sepsis. The gram-negative spectrum of aztreonam has been shown to be similar to that of an aminoglycoside [IO]. A meta-analysis of 17 trials with 1,956 patients found no difference in outcomes between single p-lactam antimicrobials versus aminoglycoside combinations in treating penetrating trauma [II]. We doubt that the differences in outcome could be attributed to any added coverage provided by aztreonam. Could dosing have had an impact on the outcome of this study? Gentamicin was administered at 5 mg/kg for the first 24 hours, and the dose was subsequently adjusted to maintain therapeutic, nontoxic serum levels. The dose of 5 mglkglday is that recommended for serious or life-threatening infection. Nonetheless, with that dosage nearly 50% of patients had subtherapeutic concentrations after 24 hours, and no patient had toxic concentrations. That observation is consistent with a recent report by Reed et al [12] of the pharmacokinetic profile of amikacin in this clinical population. Those investigators found a significant decrease in the half-life from the expected (2.17 hours versus 2.94 hours, p ~0.01). In an experimental model of hemorrhagic

THE AMERICAN

ABDOMINAL

TRAUMA

shock and sepsis in rats, Livingston et al [13] found that abscesses were significantly reduced when cefazolin was administered at a mega-dose (200 mg/kg) compared with the standard dose (30 mg/kg). Others have suggested higher doses of prophylactic agents may be necessary in trauma patients [14]. It has also been suggested that early, high dosages of antibiotics that maintain the antibiotic concentration above minimum inhibitory concentration (MIC) will both enhance cure and decrease resistance [IS]. That concept may be germane to the decreased efficacy of gentamicin in this study, since, although therapeutic levels were maintained by monitoring serum concentrations after the first 24 hours, a substantial number of patients were underdosed in the initial hours after contamination within both the infected and uninfected groups. In addition to those drawbacks associated with gentamicin dosing, there were apparent advantages associated with aztreonam pharmacokinetics. As opposed to the decreased half-life of aminoglycosides in this scenario [12], the pharmacokinetic study of 23 patients receiving aztreonam in this trial demonstrated a significant increase in half-life compared with normal volunteers. The mean peak and trough serum concentrations of 75 and 15 ug/mL, respectively, provided a large area under the curve above MIC for most Enterobacteriaceae. These findings underscore the importance of further studies of pharmacokinetics in critically ill patients because different drugs are handled by varying metabolic pathways and routes of excretion, We believe one explanation for the improved outcome with aztreonam compared with gentamicin is based on these pharmacokinetic considerations. It might be argued that simply increasing the aminoglycoside dosage would eliminate this difference. However, one must consider the therapeutic indices of aminoglycosides versus B-lactams. The index is narrow for the former and wide for the latter. Increasing doses of aminoglycoside may lead to ototoxicity or nephrotoxicity, which is not a major concern with the p-lactam class. Considering the high serum concentrations achieved with B-lactams at conventional doses [16], these agents are more practical for empiric usage in prevention of infection following penetrating trauma. Putting this issue in proper perspective, it is clear that attainment and maintenance of adequate antimicrobial concentrations is not the only factor affecting therapeutic outcome. For example, therapy in three G/C patients in the present study failed despite therapeutic concentrations. However, selection of antibiotics with more favorable pharmacokinetic and safety profiles may reduce the influence of this source of therapeutic failure in penetrating trauma patients. Individualization of therapy by adjusting antimicrobial serum concentrations based on bacterial MICs might enhance the utility of these agents even further. This prospective, randomized trial demonstrates a clinically superior outcome with the antimicrobial combination

JOURNAL

OF SURGERY

VOLUME

167

MARCH

1994

295

of aztreonam lclindamycin when compared with gentamicinlclindamycin and is one of the few studies ofpenetrating abdominal trauma that demonstrates a difference between antibioticregimens. Th&studycomes from a strong researchproductive trauma center and is both well designed and carefully reported.

REFERENCES 1.Uniform Crime Reports for the United States 1991. Washington, D.C.: Federal Bureau of Investigation, 1992: 168. 2. Dellinger EP. Antibiotic prophylaxis in trauma: penetrating abdominal injuries and open fractures. Rev Infect Dis 1991; 13 (suppll0): S847-57. 3. Zaske DE. Aminoglycosides. In: Evans WE, Schentag JJ, Juskowi, editors. Applied pharmacokinetics: principles of therapeutic drug monitoring. Vancouver: Applied Therapeutics, Inc., 1992: 14-1-14-47. 4. Sawchuk RJ, Zaske DE. Pharmacokinetics of dosing regimens which utilize multiple intravenous infusions: gentamicin in burn patients. J Pharmacokinet Biopharm 1976; 4: 183-95. 5. Swabb EA, Singhvi SM, Ieitz MA, et al. Metabolism and pharmacokinetics of aztreonam in healthy subjects. Antimicrob Agents Chemother 1983; 24: 394-400. 6. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976; 16: 31-41. 7. Croce MA, Fabian TC, Stewart RM, Pritchard FE, Minard G, Kudsk KA. Correlation of abdominal trauma index and injury

296

THE AMERICAN JOURNAL OF SURGERY

severity score with abdominal septic complications in penetrating and blunt trauma. J Trauma 1992; 32: 380-8. 8. Moore EE, Dunn EL, Moore JB, ef al. Penetrating abdominal trauma index. J Trauma 1981; 21: 43945. 9. Fabian TC, Croce MA, Payne LW, Minard G, Pritchard FE, Kudsk KA. Duration of antibiotic therapy for penetrating abdominal trauma: a prospective trial. Surgery 1992; 112: 788-95. 10.Henry SA. Overall clinical experience with aztreonam in the treatment of intraabdominal infections. Rev Infect Dis 1985; 7: S729-33. 11. Hooker KD, DiPiro JT, Wynn JJ. Aminoglycoside combinations versus beta lactams alone for penetrating abdominal trauma: a meta-analysis. J Trauma 1991; 8: 115-60. 12. Reed RL, Ericsson CD, Wu A, Miller-Crotchett P, Fischer RP. The pharmacokinetics of prophylactic antibiotics in trauma. J Trauma 1992; 32: 21-7. 13. Livingston DH, Shumate CR, Polk HC, Malangoni MA. More is better: antibiotic management after hemorrhagic shock. Ann Surg 1988; 208: 451-9. 14. Ericsson CD, Fischer RP, Rowlands BJ, Hunt C, MillerCrotchett P, Reed L. Prophylactic antibiotics in trauma: the hazards of underdosing. J Trauma 1989; 29: 135661. 15.Shentag JJ, Ballow CH, Paladin0 JA, Nix DE. Dual individualization with antibiotics: integrated antibiotic management strategies for use in hospitals. In: Evans WE, Schentag JJ, Jusko WJ, editors. Applied pharmacokinetics: principles of therapeutic drug monitoring. Vancouver: Applied Therapeutics, Inc., 1992: 17-l-1720. 16. Kunin CM. Dosage schedules of antimicrobial agents: historical review. Rev Infect Dis 1981; 3: 4-11.

VOLUME 167 MARCH 1994