DRAFT

PRACTICE MANAGEMENT GUIDELINES FOR PROPHYLACTIC ANTIBIOTICS IN PENETRATING ABDOMINAL TRAUMA

Evidentiary Tables

I. Recommendations

A. Level 1

There is sufficient Class I and II data to recommend a single preoperative dose of prophylactic antibiotics with broad spectrum aerobic and anaerobic coverage as a standard of care for trauma patients sustaining penetrating abdominal wounds. Absence of a hollow viscus injury requires no further administration.

B. Level 2

There is sufficient Class I and Class II data to recommend continuation of prophylactic antibiotics for only 24 hours in the presence of injury to any hollow viscus.

C. Level 3

There is insufficient clinical data to provide meaningful guidelines for reducing infectious risks in trauma patients with hemorrhagic shock. Vasoconstriction does not allow for normal distribution of antibiotics resulting in reduced tissue penetration. To circumvent this problem, the administered dose may be increased two- or threefold and repeated after every 10^th unit of blood product transfusion until there is no further blood loss. Once hemodynamic stability has been achieved, antibiotics with excellent activity against obligate and facultative anaerobic bacteria should be continued for periods that depend on the degree of wound contamination. Aminoglycosides with optimal activity in an alkaline environment should not be used because traumatized tissue is acidotic.

II. Statement of the Problem

Fullen et al. first described a thorough examination of the role for antibiotics in patients sustaining penetrating abdominal injuries.¹ They retrospectively reviewed 295 patients undergoing celiotomy after sustaining penetrating abdominal wounds and categorized patients according to timing of their first antibiotic dose: preoperatively (n=16), intraoperatively (n=98) and postoperatively (n=81). The reported rate of infections which were trauma-related (incisional and intra-abdominal abscess) were 7%, 33% and 30%, respectively. Further analysis of the data for individuals with colon injuries found the postoperative infection rate for each group to be 11%, 57% and 70%, respectively. These rates remained constant when the data were analyzed for additional risk factors including number of associated intra-abdominal organs injured, frequency of shock, and need for transfusion of blood products. The average time from hospital admission to laparotomy was the same for all three groups. Regardless of whether the observed difference was due to the intra-operative or postoperative groups having a longer interval between injury and antibiotic administration or that the preoperative group had antibiotics circulating at the time the incision was made, this was the first study to suggest that the timing of antibiotic administration in patients with penetrating abdominal injuries can impact the development of injury related infections.

The importance of broad spectrum antibiotic coverage for these patients was demonstrated by Thadapalli et al. in 1973.² This was a prospective randomized comparison of Kanamycin and Cephalothin against Kanamycin and Clindamycin. Both antibiotic combinations were administered pre operatively. The Clindamycin group had a significantly lower rate of infection in the postoperative period compared to the Cephalothin group (10% vs 27%). They further demonstrated that the difference was due to significantly more anaerobic infections in the Cephalothin group (21%) compared to the Clindamycin group (2%).

These two studies demonstrated a significantly lower rate of infection when antibiotics providing aerobic and anaerobic coverage are administered prior to operative treatment. Prophylactic antibiotics in penetrating abdominal injuries with intestinal contamination do have a role for reducing the rate of incisional wound infection which is subjected to gastrointestinal soiling. A single dose providing sufficient concentration within the wound during the vulnerable period is optimal. The other aspect of prophylactic antibiotic administration in trauma is the potential therapeutic role. The problem is to define the time period when contamination of the abdominal cavity becomes infection. At celiotomy, the intestinal wound is closed eliminating further contamination and soiling of the peritoneal cavity limited by copious irrigation. Thus, no further antibiotic should be necessary.

Surgeons have concluded that "prophylactic antibiotics" in penetrating abdominal trauma can reduce the incidence of post operative infectious complications. Since the mid 1970's, none of the studies have included a placebo control arm because of the high incidence of infectious complications after intestinal injury. However, many studies in the past two decades have compared various antibiotic regimens to evaluate single agents versus combination regimens, duration of administration, and more recently the pharmacokinetics and cost implications of single versus combination therapy.

III. Process

A. Identification of references

The recommended guidelines for prophylactic antibiotic usage for trauma patients sustaining penetrating abdominal wounds are evidenced based. A MEDLINE search from 1976 to 1996 was performed. The following subject words were used for the query: antibiotic prophylaxis, penetrating abdominal injuries, abdominal injuries-complications, peritonitis, wound infection-prevention and control, pharmacokinetics, trauma and cost analysis. This identified 55 English references. The bibliography of each article was reviewed for additional references which were not identified in the original MEDLINE query. Letters to the editor, case reports and review articles were deleted from further evaluation. Thirty nine articles were identified for this evidentiary review. Thirty two were felt to deal with comparison of various antibiotic regimens and put into the first table titled outcome. The remaining seven include six clinical studies and one meta-analysis. They deal with pharmacokinetics and cost and are listed in the table titled pharmacokinetics and cost. The articles were reviewed by five general surgeons and two pharmacologists with interest in pharmacokinetics and health care economics. These individuals then collaborated to produce the guidelines.

B. Quality of the references

The references were classified using methodology established by the Agency for Health Care Policy and Research (AHCPR) of the U.S. Department of Health and Human Services. Additional criteria and specifications were use for Class I articles from a tool described by Oxman et al.³

Thus, the classifications were:

Class I: Prospective, Randomized, Double-Blinded Study.

Class II: Prospective, Randomized, Non-Blinded Trial.

Class III: Retrospective Series of Patients or Meta-Analysis.

In the evidentiary section are two tables. The first includes 32 studies comparing various antibiotic regimens for infectious outcome. The second table is a summary of the 7 articles dealing with pharmacokinetics and cost.

IV. Scientific Foundation

A. Historical background

The work by Fullen and Thadepalli set the standard for usage of antibiotic prophylaxis in patients sustaining penetrating wounds to the abdomen.^1,2 Multiple studies through the 80's compared the various antibiotic regimens for efficacy of reducing infection rates, safety and duration of antibiotic administration. These articles have been nicely summarized by Dellinger.⁴ He discusses specific issues regarding choice of agent, duration of therapy and optimum dose for the various antimicrobial agents. Several studies since that review compare third generation cephalosporins and ß-lactam penicillin derivatives with combination therapy. There have been several studies evaluating the duration of therapy. Dellinger concluded in 1989 that the studies reported up to that time did not permit a definitive statement regarding the ideal antimicrobial agent for patients sustaining penetrating abdominal injuries and we would agree with this statement. The additional studies identified in this review address some of the concerns raised by Dellinger and will be the focus of this evidentiary review.

Many difficulties exist with interpreting the literature to date with regards to prophylactic antibiotics for penetrating abdominal wounds. Specifically, there continues to be a lack of standardization as suggested by Dellinger in study design and reporting of data and results. There is also the lack of discussion about the common risk factors identified by Nichols¹⁴ and Dellinger.¹¹ These include the importance of transfusion requirements, length of operation, age, and penetrating abdominal trauma index as significant risk factors for development of any postoperative infection are not mentioned in the majority of studies. With the trend towards primary repair of all colon injuries , the surgical management of the colon is rarely mentioned nor standardized in the study design. The frequency of colostomy versus primary repair particularly regarding left colon injury in high risk patients is not standardized in the design or discussed in the results. Thus, the goal of this evidentiary review is to evaluate the literature regarding mechanism of injury, choice of agent, duration of therapy, the unique pharmacokinetics of the trauma patient and its effect on dosing considerations, and cost analysis.

B. The risk factors for trauma-related infections (wound infection, intra-abdominal abscess, bacteremia, drain tract infection, and urinary tract infection)

1. Mechanism of injury

The majority of studies have combinations of patients sustaining stab wounds and gunshot wounds.^{5,6,8,11,12,14,15,20,22-27,30,33-36} Four reports did enroll patients controlling for types of penetrating forces. Moore et al. compared 3 antibiotic regimens in only firearm wounds.¹⁰ Of the 86 patients studied, less than half had colon injuries (N=39). The infection rate ranged from 13 to 23% with no significant difference between the three groups. Three studies evaluated patients primarily injured by knives.^21,30,31 Heseltine et al. only enrolled stab wounds to compared gentamicin and clindamycin (G+C) against cefoxitin. Both drugs were administered for 72 hours. The infection rates were 7% and 9%, respectively. Of the 75 patients studied, less than one third had colon injuries (N=21). Demetriades et al. studied 123 patients (76% stab) receiving 48 hours of ceftriaxone or cefoxitin and observed a 7-8% rate of infection.³⁰ Van Rensburg et al. observed a 1.4 % infection rate in 290 patients (89% stab wounds) receiving ceftriaxone and metronidazole for 24 hours.³¹ However, there were only 47 (16%) colonic injuries. None of these studies discusses the management of the colon injury. These three studies suggest that prophylactic antibiotics in abdominal stabbing injuries can be stopped after 24 hours.

In several studies the mechanism of injury was not reported ^{12,15,23,32,35} Yet others also included non-penetrating injuries. Erickson compared the duration and dose of amikacin and clindamycin in 150 randomized patients.²⁸ Bowel injuries occurred from gunshot wounds (N=76), blunt trauma (N=40) or knives (N=24). Hoffstetter also included blunt bowel wounds (20%) and observed no difference in infection rates for patients treated with cefoxitin or triple antibiotics for 24 hours (14% vs 18%, respectively)^.13

2. Choice of antimicrobial agents

The 32 articles included in the outcomes table compared various antibiotics for differences in infectious complications. Twelve of these studies were class I data (prospective, randomized, double-blinded). Three series which controlled for mechanism of injury at the time of enrollment were discussed in section 1.^10,21,32 Another study evaluated duration of therapy.⁸ The other six articles included mixed populations of patients (blunt, gunshot, stab) comparing various single agents against combination therapy.^{7,14,17,23,32,33,35,36} Fabian et al. compared various classes of cephalosporins and limited duration to 24 hours. The number of colon injuries in each of the three arms was less than 20% of the patients enrolled in each. The trauma-related infections did not differ (range 9 to 17%).⁷ Nichols and colleagues compared cefoxitin to C+G. Both groups had an infection rate of 24% despite 5 days of therapy.¹⁴ Jones et al. compared tobramycin plus clindamycin (T+C) to cefamandole and cefoxitin in 257 patients. Ninety six (37%) had colon injuries and were equally distributed between the three arms. He concluded cefoxitin and T+C were superior to cefamandole in reducing infections (18%, 29% versus 36%, respectively) with only 48 hours of therapy.¹⁷ In a comparison of moxalactam to combination (tobramycin plus clindamycin), Nelson observed no difference in postoperative infections (19% versus 23%).²³ Cefoxitin and cefotetan were compared to each other with two different durations (1 versus 5 days).³² There was no difference in trauma-related infection rates with either agent or duration of prophylaxis. The ß-lactam penicillin, aztreonam, was compared to gentamycin when both agents were used in combination with clindamycin.³⁵ The authors concluded aztreonam was superior to gentamycin (3% versus 13%) because of the under dosing of the gentamycin and subsequent subtherapeutic levels compared to the more stable pharmacokinetics of aztreonam. Unfortunately, of the 63 patients enrolled in this trial only 17 (27%) had colon injuries.³⁵ The role of enterococcus in abdominal infectious complications was evaluated by Sims et al.³⁶ Using cefoperazone which has no enterococcal coverage or combination therapy with entercoccus coverage. Both regimens were administered for at least five days. Trauma-related infections ranged from 2 to 8%. They concluded that coverage for entercoccus is not necessary in penetrating abdominal wounds.

In these six studies, the trauma-related infectious complication rate ranged from 3% ³⁵ to 36%.¹⁷ The only study with a significant reduction in infections was based on a small study population (N=63) with only 17 colonic wounds.³⁵ It does not justify any definitive statement. None of the other studies showed superiority of any agent compared to an aminoglycocide in combination with clindamycin or metronidazole (Class I data table). Finally, one Class I study evaluated duration of therapy using penicillin G plus doxycycline administered for either 12 hours or 5 days.⁸ There was no difference in the trauma-related infections between the two durations. These Class I data indicate that single and combination therapy are equally effective in minimizing trauma-related infections following penetrating abdominal wounds. The antibiotics need not be continued for more than 24 hours following injury.

3. Duration of therapy

There is a small amount of Class I data regarding duration of therapy. Griswold et al. Stated that the injury severity, as measured by the abdominal trauma index, should dictate therapy duration.³³ While there were no significant differences in abdominal abscess rates for the antibiotic groups (cefoxitin, ceftriaxone, and mezlocillin), there were differences in abscess rates for those with lower abdominal trauma index (ATI). The authors concluded that antibiotics should be given longer than 12 hours for high risk patients. Fabian et al. analyzed 515 patients, randomized to receive either cefoxitin or cefotetan for either 1 or 5 days.³² There were no differences in abdominal infection rates for the different antibiotics or for duration of therapy. When duration was compared in the high risk population (colon wounds or ATI > 25) there was also no statistical difference in infection rates. In fact, infections tended to be more frequent in the 5 day group. These authors concluded that 24 hours of therapy was adequate therapy for all penetrating abdominal wounds.

The Class II data (prospective, randomized, non blinded) include several trials evaluating various lengths of therapy.^{15, 18, 19, 25, 30, 34,} Rowlands and associates incorporated two independent studies in their review comparing various antibiotics for 3 or 5 day courses. Unfortunately, they did not identify colon or other hollow viscus injuries in any treatment.¹⁵ Although rates of infection were lowest with the five day therapy, the lack of knowledge about organ injuries does not allow any convincing conclusion. Dellinger evaluated 116 patients with penetrating wounds of the colon and/or small bowel randomized to receive 12 hours or 5 days of antibiotics. There was no statistical difference in the rate of trauma-related infections.¹⁹ In contrast, a second study by Rowlands used intra operative findings to define patients as high or low risk for infection.²⁵ The high risk group included one or more of the following injuries: penetration of the GI tract, major liver or pancreatic injury, close range shotgun wounds, patients in whom complete hemostasis was not obtained, patients in whom nonviable tissue was present at the time of wound closure and following major splenic repair. Patients without any of these injuries were considered low risk and received antibiotics for less than 24 hours where as prophylaxis for the high risk was for 72 hours. Each group had a second stratification for one of two antibiotic combinations. Despite this randomization by operative findings, their data reported one colon and one small bowel injury in the low risk group receiving less than 24 hours of antibiotics with a 6% rate of infection. The high risk group included 53 colon injuries in 103 patients. The infection rates for the two antibiotic regimen were 16 and 25%.²⁵ Lou compared mezlocillin to G+C. Both therapies were continued for 5 to 10 days if there was a colon injury. The trauma-related infection rate was similar between the two groups (9 versus 10%).²⁷ Moore et al. compared the same two antibiotic regimens as Lou but limited therapy to a 5 day course for patients sustaining a colon injury and reported a 13 to 15% incidence of infections.²⁹ Even when antibiotic therapy is continued for up to 15 days, the trauma-related infection rate remains 7- 8% as reported by Sims et al.³⁴ These Class II data further support limiting prophylaxis to 24 hours or less since none documented any benefit with a longer course.

4. Pharmacokinetics (Optimal Dosage)

The majority of studies today evaluating efficacy of various antibiotic regimens have used standard drug doses recommended for healthy patients undergoing elective procedures. This is a particular concern in patients sustaining penetrating wounds where hemodynamic instability is common. It is this group of patients that present with shock which receive inappropriate low doses of antibiotics when standard dosing is followed and not adjusted for the reduced circulating volume. The ideal dose of antibiotics has not been established for these patients. Ericsson et al.²⁸ was the first to raise the question of adequate dosing because of large fluid shifts and the hyperdynamic physiologic response seen in trauma patients. He compared clindamycin 600 mg every six hours against 1200 mg dosed every 12 hours. Each was continued for a 72 hour course. He saw no difference in infection rates. However, early in the study they measured the serum concentration of a second agent, amikacin, and found the average peak level was less than 20µg/ml. They subsequently increased the dose of amikacin to 11 mg/kg and saw an inverse correlation between increasing dose of amikacin and rate of infection. Similarly, Townsend and colleagues reported the volume of distribution for aminoglycosides in trauma patients was greater than predicted and serum levels were subtherapeutic (not in table).³⁷ In order to maintain adequate levels of aminoglycosides in the serum, wound fluid, or target tissue, they recommended an initial loading dose of 3 mg/kg.

The relationship of volume of distribution being altered in trauma patients by massive fluid resuscitation was further elucidated by Reed et al.⁴¹ They compared standard amikacin dosing to dosing according to pharmacokinetic analysis of serum levels. Both groups had a significant expansion of the volume of distribution (71%) during the first 24 hours post injury. For study days 2 and 3, the volume expansion was only 43% over the expected values. In contrast, the elimination rates remained elevated on these days. Because of the expanded volume of distribution during resuscitation, they concluded that empiric prophylactic antibiotic dosing should be high rather than low. They also recommended frequent dosing during the fluid resuscitation phase. Once bacterial contamination of tissues has ceased, the antibiotic should be stopped. Because of the relatively narrow therapeutic index associated with aminoglycocides and the highly variable volumes of distribution in trauma patients they concluded that standard dosing of these agents would result in subtherapeutic serum and tissue levels and thus inadequate prophylaxis. Rosemurgy and associates evaluated the effects of volume of distribution on serum concentrations of ceftizoxime and subsequent infection rates.⁴² They demonstrated that early peak serum concentrations were significantly lower in patients who experienced infections because of the increased volume of distribution of the drug compared to patients who did not develop infection.

These four studies, although with small numbers of patients raise serious questions about the adequacy of dosing of antibiotics in previous trials which included patients in shock. Some of the high infection rates reported in earlier studies could be explained by inadequate dosing.

5. Cost analysis

In the past ten years there have been four studies evaluating cost of antibiotic therapy in trauma patients with penetrating abdominal wounds. Crots compared moxalactam against G+C in 50 patients.³⁷ The strength of this study is the well performed cost analysis which included the hospital cost for drug, laboratory test, personnel time and supplies. They observed no symptomatic trauma-related infections in either treatment group. There were also no direct toxic effects from either agent. The mean drug cost for each regimen did not differ. However, when laboratory test, personnel time and supply cost were added to the drug cost, the mean cost of therapy per patient was 38% greater with G+C compared to moxalactam. This study demonstrates the importance of considering all treatment costs when performing cost-effectiveneess analysis of combination therapy. In a similar study design, Bivins compared cefotaxime, cefoxitin and G+C.³⁸ Twenty-five patients were entered into each treatment arm and the septic complications were 8%, 4%, and 8%, respectively. The cost analysis included the same four categories (drug cost, laboratory test, personnel time and supply cost). The mean cost of therapy per patient was significantly less with the cefotaxime. Unfortunately, they did not specify solid versus hollow organ injuries. In a subsequent report, using the same study design, Bivins and colleagues used the same three antibiotic regimens for a 3-5 day course of prophylaxis.³⁹ This study included 129 patients. Only 17 patients had colon injuries. The infection rate for cefotaxime was 6.9%, cefoxitin 2.3% and G+C 6.9%. There was no statistical difference between groups. As in their previous study, the mean cost of therapy per patient was significantly lower for the cefotaxime group. The last study deals with pharmaco-economics and is by Fabian et al.⁴³ He compared aztreonam plus clindamycin against G+C in 85 trauma victims with suspected penetrating intra-abdominal injury. There were 34 colon injuries. They further analyzed the hospital cost by stratifying patients as infected versus non infected. They concluded that despite a lower infection rate in the Aztreonam group, neither hospital nor pharmacy cost were significantly different compared with those in the G+C group. These studies of cost analysis for antibiotic therapy would suggest that consideration for single agent (cefotaxime) therapy with aerobic and anaerobic coverage may be a cost effective choice compared to more the traditional combination antibiotic regimen (G+C).

V. Evidentiary Table

This review includes three tables. The first table is labeled outcome and includes 32 articles arranged in chronologic order by year of publication. The conclusion section list the: antibiotic, number of patients, days of therapy, organs injured (when identified) and percent of trauma-related infections. These infections included: wound infection, intra-abdominal abscess, drain tract wound infection, urinary tract infection or bacteremia. The rate of infection was determined by reviewing each result section for the specific infections and dividing this by the number of patients. The table includes 11 Class I articles (prospective, randomized, double blinded) and 21 Class II (prospective, randomized, nonblinded). The second table is labeled Pharmacokinetics and Cost Analysis. It includes 7 articles listed in chronologic order by year of publication (one Class I, 5 Class II, and one Class III). Three of the articles deal with cost of antimicrobial therapy, three evaluate pharmacokinetics and the last article is a meta-analysis of safety of ß-lactam penicillin derivatives against aminoglycocide combinations. The use of quinolones has not been tested for prophylaxis in penetrating abdominal wounds with GI contamination.

VI. Future Studies

Future studies need to be double-blinded in their design and clearly define the criteria for trauma-related infections. Other risk factors such as time to administration, shock, short versus long half life antibiotic, duration and organ injuries should be evaluated. More studies need to be conducted evaluating the interaction of hemodynamic status with volume of distribution. The specific organisms responsible for trauma-related infections need further study.

REFERENCES

1. Fullen WD, Hunt J, Altemeier WA. Prophylactic antibiotics in penetrating wounds of the abdomen. J Trauma 1972; 12: 282.

2. Thadepalli H, Gorbach SL, Broido PW, Norsen J, Neyhus L. Abdominal trauma, anaerobes and antibiotics. S,G&O 1973; 137: 270.

3. Oxman AD, Sackett DL, Guyatt GH. User’s guide to the medical literature. JAMA 1993; 270: 2093.

4. Dellinger EP. Antibiotic prophylaxis in trauma: Penetrating abdominal injuries and open fractures. Rev Inf Dis 1991; 13: 847.

Additional References Pending. See Evidentiary Tables.

37. Townsend PL, Fink MP, Stein KL, Murphy SG. Aminoglycoside pharmacokinetics: Dosage requirements and nephrotoxicity in trauma patients. Crit Care Med 17;154: 1989.

Evidentiary Tables