Prophylaxis with whole gut irrigation and antimicrobials in colorectal surgery

SCIENTIFIC PAPERS

Prophylaxis With Whole Gut Irrigation and Antimicrobials in Coiorectal Surgery A Prospective, Randomized Double-Blind Clinical Trial

Finn Gottrup, MD, PhD, Aarhus, Denmark Poul Diederich, MD, Randers, Denmark Keld Sglrensen, MD, Randers, Denmark S@ren Vinther Nielsen, MD, Randers, Denmark Jglrgen Ornsholt, MD, Aarhus, Denmark Ole Brandsborg, MD, PhD, Aarhus, Denmark

Operation on the colon and rectum is associated with a high incidence of postoperative sepsis [1]. Attempts to reduce this high incidence have been directed toward improvement in preoperative preparation and use of antimicrobials. The colon lumen contains an abundant reservoir of pathogenic bacteria which has led surgeons to seek reliable control of the source of these infections by means of preoperative colon preparation. Mechanical preparation alone is effective in reducing the amount of solid feces in the colon, but has a minor influence on the number of fecal bacteria at the time of operation [2,3]. A newer ~clmique of bowel preparation, known as whole gut irrigation, consists of infusion of large quantities of saline solution through a nasogastric tube [4,5].This method requires a short time, is in most cases well tolerated by the patients, and results in excellent mechanical preparation. However, almost nothing is known about the clinicaleffect of whole gut irrigation as prophylaxis against postoperative infectious complications. The use of prophylactic agents containing antimicrobialshas for m a n y years been a controversial issue,and the aim of this study was to explore the usefulness of whole gut irrigation and to perform a prospective, randomized double-blind clinicaltrial comparing the effectof whole gut irrigationalone with two systemic Municipal Hospital, Aarhus County Hosplta}, and Randers County Hosplla~, Randers, Denmark. Requests for reprints shou~ be addressedto Finn ~ MD, D e ~ of Corme~Ne Tissue Bk~ogy, tnstit-~e of Anatomy C, University of AadnuS, DK-8000, Aarhus C, Denmark.

Volume149,March19aS

antimicrobial regimens. One regimen was directed against aerobic and anaerobic organisms, and the other regimen acted against anaerobes only. Material and Methods The study included consecutive patients undergoing electivecolorectalsurgeryat the Departments of Surgical Gastroenterologyof Aarhtm Municipal Hospital,Aarhus County Hospital,and Randers County Hospitalduring a 26 month period (1979 to 1981).Criteriaof exclusionwere ulcerativecolitis,Crohn's disease,severe stenosisin the gastrointestinaltract,uremia, cardiopulmonary insufficiency,inabilityto maintain a sittingposition,and patient age younger than 15 years. Patientswith allergyto metronidazoleand ampicillinor a recenthistoryof antibiotic or steroid therapy were also excluded. Of 148 patients meeting the criteria,13 were excludeddue to reasonsshown in Table I,which left135 patientsin the trial.The protocols and methods were acceptedby the Danish NationalHealth Service,and informed consentwas obtained from the patients. The day before operation, all patients received a careful whole gut irrigation as described by Hewitt et al [4] and Crapp et al [5]. A no. 14 nasogastric tube was placed in the stomach with the patient in a sitting position and infusion of body temperature electrolyte solution (isotonic saline solution) was begun at a rate of 3 to 4 liters/hour until the fluid passed was clear or until 15 liters had been given. Metoclopramide hydrochloride (10 mg intramuscularly) was given before irrigation, and furosemide (40 mg intramuscularly) was given after I hour of irrigation. Weight was measured before and after irrigation. After whole gut irrigation no solid food was given. The patients were randomized into the three groups shown in Table IL Group A followed the regimen usually employed by the institutions

317

Gottrup et a|

TABLE I

Reasons for Exclusions From the Study In Retrospect

Irrigation not completed Medication error Nonstudy antibiotic therapy within 2 days Death from noninfectlous causes within 4 days Total

TABLE III

Group A

Group B

Group C

3

1

I

2 0

2 2

0 0

2

0

0

7

5

1

TABLE II

Regimens of the Three Groups"

Preoperative day 1 Day of operation

Group A

Group B

Group C

WGI Saline

WGI Saline Metrontdazole

WGI Ampicillln Metronldszole

Saline

Saline Matronl0azole

Amptclllln Metronldazole

Postoperative day

1, 2, and 3

• Metronidazole (500 rag) was administered a half hour preoperatively and repeated every 8 hours. Amplclllln (1 g) was administered a half hour preoperativelyand repeated every 8 hours. Anttmtcrobtal agents and saline (100 ml each) were administered intravenously. WGI = whole gut irrigation.

Pallent Age and Preoperative Diagnosis

Age (yr) Mean Range Preoperative diagnosis Carcinoma of rectum Carcinoma of slgmold Carcinoma of colon Transverse Descending Ascending and cecum Sigmold diverticulitis Other Total

Group A

Group B

Group C

67 33-83

63 37-81

64 37-87

17 13

21 7

17 6

1 2 7 0 1 41

3 2 7 3 3 46

4 3 12 4 2 48

TABLE IV

Type of Surgery"

Primary colon anastomosis Right hemicolectomy Transverse colectomy Left hemlcolectomy Stgmold resection Rectum resection (low anterior resection) No colonic anastomosis Abdomtnoperinealresection Total proctocolectomy Colostomy Total

Group A

Group B

Group C

8 (1) 1 (1) 4 (1) 11 (7) 6 (3)

7 (3) 3 (0) 3 (1) 8 (3) 8 (4)

13 (0) 2 (0) 6 (0) 10 (0) 7 (2)

9 (6) 0 2 (1) 41 (20)

14 (5) 1 (0) 2 (0) 46 (16)

10 (3) 0 0 48 (5)

• Number of infectious complications are indicated in paren-

thoses.

involved. In Group B, metronidazole was added, and in Group C, rnetronidazoleand ampicillinwere included. During operation,the surgeon registeredthe efficacyof whole gut irrigation,designatingthe bowel content as clean, containing fecallystained fluid,or fecalloaded. Observations during the postoperativeperiod included length of hospitalstay,clinicalsignsof anastomotic leakage,wound infection, abdominal abscess, and nonwound related complications. Infections were classifiedin accordance with earlier studies from our surgicaldepartments [6,7].W o u n d infectionindicatedsuperficialaccumulation of pus requiring surgical drainage. Intraabdominal infectionwas present when two of the followingcriteriawere fulfilled:(1)pyrexia postoperativelyfor more than 72 hours for which allother causes uould be excluded; (2) abdominal tenderness,distension, or a palpable mass; or (3) discharge of pus per rectum. Ifabscess developed, itwas drained ifindicated, and specimens were taken for cultivationof bacterialorganisms. All patients were operated on in uniformly equipped theaters.The skin preparationregimens were identicaland wound protectingplasticdrapes were routinelyemployed. All anastomoses were performed in an open manner, employing a double-layered closure of interrupted 4-0 polyglycolic(Dexone) sutures.Topical antibioticsolution was not used. Primary closurewas carriedout in allcases.

318

The wound was closed with absorbable sutures in the peritoneum and fasciaand nonabsorbable sutures in the skin. Statisticalevaluation included the chi-squaretest applied with Yates correction,the Fisher exact testfor comparison of two rates,and the Student's t testfor unpaired data. Significancewas stated for p <0.05. Results

T h e characteristics of the patient groups are summarized in Tables III and IV. T h e groups were comparable in regard to age, diagnosis, and type of surgery performed. W o m e n were in the majority (79 w o m e n versus 56 men). Approximately two thirds of those in each group underwent a colonic operation involving an anastomosis, and the rest had a terminal stoma. Regarding the characteristics of whole gut irrigation, bowel preparation was completed without any discomfort in 129 patients (87 percent), and 14 patients (10 percent) had slight complaints, such as epigastric fullness,nausea, and one or two episodes of vomiting. In five patients (3 percent), bowel irri-

The Ame¢lclmJournal of ~=rgery

Prophylaxis With Whole Gut Irrigation

TABLE V

Infectious C~nplicailona and Clinical AnastornaUc Leakage

~oupA Abdominal wound Infection Intraabdominal abscess Porlneal Infection Anastomatlc leakage

TABLE VII

TABLE Vl

Group

~ B

n

~'o

n

"~

n

7o

13

32

10

22

1

2

5

12

4

9

1

2

2

5

2

4

3

6

5

12

2

4

2

4

Assessment of Bowel Content at Operation

Clean bowel Fecal stained fluid In the bowel Fecal loaded bowel Total

Group

Group

A

B

Group

C

23 15

28 15

23 18

3 41

3 46

7 48

gation was stopped because of severe complaints, and these patients were excluded from the study. The time required for bowel irrigation was 239 minutes (range 60 to 465 minutes), and the average amount of fluid was 9.4 liters (range 4 to 15 liters). Only minor changes in body weight were found. The most prenounced weight gain was 5.1 kg, but most patients showed a negligible change of a few hundred grams, Postoperative infection developed in 41 patients (30 percent). The distribution in relation to groups and types of infection is shown in Tables IV and V. Abdominal wound infection occurred in 13 patients (32 percent) in Group A, in 10 patients (22 percent) in Group B, and in 1 patient (2 percent) in Group C. There was a significant difference between Groups A and C (p <0.001) and Groups B and C (p <0.01), although the wound infection rates of Groups A and B were not statistically different. Intraabdominal abscesses developed in five patients (12 percent) in Group A, in four patients (9 percent) in Group B, and in one patient (2 percent) in Group C (p = 0.06 when rates in Groups A and C were compared). The rates of postoperative infection in the perineal wound were identical in the three groups. Clinical signa of anastomotic leakage occurred in five patients (12 percent) in Group A, in two patients (4 percent) in Group B, and in two patients (4 percent) in Group C. No significant differences were found. All patients with anastomotic leakage had a left side resection, in seven for carcinoma of the rectum and in two for sigmoid carcinoma. The bacteriologic pattern demonstrated in the wound swabs was dominated by mixed anaerobic and

Voklme 140, Ma¢ch 19SS

Nonwound-Releted Complica'tlons

Pneumonia Urinary tract infection Urine retention Phlebitis Others

~o~

Grow

A

B

C

3 1 O 0 1

4 1 0 0 3

1 1 1 1 0

aerobic organisms in Group A. Material from 11 wound infections showed a mixed culture in 5, pure anaerobic growth in 2, and pure aerobic growth in 4. In Group B, all infections but one were caused by aerobic organisms (primarily Escherichia toll). The infection with anaerobic organisms was a mixed growth containing many E. coli and some anaerobic streptococci that were sensitive to ampicillin but not to metronidazole. In Group C, no anaerobic organisms were cultivated from the wounds, and the predominating aerobic organism was E. coll. Three patients died of a postoperative infectious complication, (one patient from each group). The patients from Groups A and B did not have abdominoperineal resection. Wound infection as well as intraperitoneal infection developed in the patient from Group A, and pneumonia, intraabdominal infection, and ileus developed in the patient from Group B. The patient from Group C whose rectum was resected had cardiac decompensation and anastomotic leakage developed. Nonwound-related complications are tabulated in Table VI. No pattern among the three groups of patients could be detected. The mean postoperative hospit~ stay for patients with no complications was 12.8 days (range 6 to 30 days). If infectious complication occurred, the stay was 29.2 days (range 10 to 95 days) (p <0.001). No difference among the groups was found for uncemplicated cases (group A 11.6 days, Group B 14.3 days, and Group C 12.3 days) or for cases of infectious complications (Group A 32.5 days, Group B 26.6 days, and Group C 25 days). No difference between men and women in any group was found. The results of the surgeons' assessments of bowel content during surgery are shown in Table VII. In 74 patients (55 percent), a clean bowel was found, whereas 47 patients (35 percent) had fecal fluid in the bowel, and 14 patients (10 percent), a fecal mass. No significant difference in distribution of bowel content between the groups was demonstrated. In Group A, the relation between bowel content and infectious complications (wound infection, intraabdominal infection, and pelvic wound infection) showed no difference in the postoperative infection rate between fecal loaded bowel (two of three patients or 67 percent, became infected) and bowel containing fecal fluid (10 of 15 patients, or 67 percent became in-

319

Gottrup et al

fected). However, when there was a clean bowel, a significant decrease in the infection rate was found (8 of 23 patients or 35 percent, p <0.05). When antimicrobials were given, the infection rates decreased. The incidence in Group B was 44 percent for a bowel containing feces (fluid or masses) and 28 percent for a clean bowel. The corresponding figures for Group C were 12 percent and 9 percent. No significant difference between a bowel containing feces or a clean bowel was demonstrated in Groups B and C. No relation between bowel content and clinical signs of anastomotic leakage was found. In Group A, three anastomotic leaks were found in patients with clean bowels and two leaks in patients with bowels containing fecal fluid. In Groups B and C, one anastomotic leak was in a clean bowel, two in bowels with fecal fluid, and one in a bowel loaded with fecal masses. Comments

Whole gut irrigation is now widely used as preoperative bowel preparation and, compared with traditional mechanical enema preparation of the bowel, it is rapid, well tolerated, and preferred by patients. Therefore, it is remarkable that no studies on the antiinfectious effect of whole gut irrigation alone or in combination with antimicrobials have been published. 0nly Raahave et al [8] tried to evaluate the effect of whole gut irrigation on postoperative wound infections. However, they used ampicillin topically in the wounds. Andersen et al [9] have shown that this prophylaxis decreases the wound infection rate and may cover a possible positive effect of whole gut irrigation itself. Few studies have investigated the value of systemic antimicrobials for prophylaxis in colorectal surgery, and no studies comparing whole gut irrigation with different systemic prophylaxis have been published. The systemic route of administration has been used with success for other types of prophylaxis in our surgical departments and is advantageous in that it can be used in patients who cannot take drugs orally or cannot cooperate for other reasons [6,7].Furthermore, the attempt to sterilize the colon before operation has been demonstrated to be potentially dangerous [10,11]. Such a regimen is associated with the emergence of resistant organisms and any preparation that alters the normal intestinal flora will encourage the development of pseudomembraneous colitis and overgrowth of yeast. Consequently, the systemic route seems to give more predictable results, is more effective, and is more consistently effective in control of wound infection than locally employed antibiotics [12]. Antimicrobial prophyl~xls in colorectal surgery can be administrated as a specific agent for aerobic or anaerobic organisms or as two or more agents administered simultaneously to cover the organisms involved as adequately as possible. Specific aerobic cover has been especially promising using different 320

types of cephalosporins [13]. A significant reduction of postoperative infections has been observed in colorectal surgery patients after cephaloridine and cefazolin prophylaxis. However, newer cephalosporins with greater in vitro activity against bacteria responsible for infections in these patients may be ineffective, especially against the. predominant anaerobic organism Bacteroides fragilis [11,14]. The use of a specific anaerobic agent is based on the increasing awareness of anaerobic organisms as the etiologic agents for postoperative infection in colorectal surgery and of appendectomy [15-17]. Metronidazole has been shown to have a selective and bactericidal activity against obligate anaerobic and facultative anaerobic bacteria. However, some sort of pathogenic synergy between bacteroides and aerobic organisms, especially E. coil, has been demonstrated experimentally [15,18-20] as well as clinically [6,15,20,21]. The effect of metronidazole was a reduction of gram-negative organisms as well as their related infections when the anaerobic bacteria were eliminated. Therefore, it is important to investigate the possible effects of metronidazole alone and in combination with an aerobic agent after whole gut irrigation. Our employment of ampiciUin was based on our knowledge of the sensitivity of the predominant organisms isolated from infections after colorectal surgery in our hospital departments. E. coil was isolated from most postoperative wound infections after this type of surgery, and the incidence of ampicillin resistance was about 15 percent constantly for several years. This sensitivity of E. coil to ampicillin was higher than that reported from England and the United States [22,23]. Other aerobic organisms had minor importance for postoperative infectious complications in our surgical departments. Streptococcus faecalis was isolated from three patients and Staphylococcus aureus from two patients and always in mixed cultures. Ampicillin was effective against the enterococci b u t not Staph. aureus. Ill this study, whole gut irrigation was completed without any discomfort in 87 percent of tim patients, 10 percent had slight complaints, and only in 3 percent did the procedure have to be stopped because of severe complaints. Abdominal cramps in four of five patients who were excluded was the main reason for ending whole gut irrigation, and epigastric fullness, mild nausea, and vomiting were the main complaints in most patients. The incidences were comparable to the results of Raahave ~t al [8], b u t lower ~b~n those reported by Levy et al [24] when the patients were asked to drink the fluids. Overloading was never observed, b u t we believe that administration of a diuretic drug during whole gut irrigation is necessary. Our experience with the patients' opinions of whole gut irrigation is comparable to that of Downing et al [25] who observed that most patients prefer whole gut irrigation to conventional mechanical preparations they had experienced. The~Jotmml~8~gery

Prophylaxis With Whole Gut Irrigation

Abdominal wound infection developed after colorectal surgery in 32 percent of the patients prepared with irrigation only. This incidence is comparable to the results of Raahave et al [8], but their slightly lower infection rate may be a result of the use of topical ampicillin. However, the wound infection rate in this study is lower than reported in placebo groups after conventional mechanical enema preparation (40 to 65 percent) [1,11,26-30], and this may be interpreted as whole gut irrigation itself has some decreasing effect on the incidence of wound healing which is probably caused by the higher incidence of collapsed bowel found after whole gut irrigation compared with conventional bowel preparation [31,32]. We found significantly fewer postoperative infectious complications with a clean bowel compared with a bowel containing fecal fluid or masses. No differences in anastomotic leakage between the clean and fecal loaded bowel were found. This may demonstrate that the content of the bowel is important when infectious complications occur, but it has no influence on healing and patency of the anastomosis itself. Probably, the surgical technique is of major importance when anastomotic leakage occurs. When systemic metronidazole was added to the whole gut irrigation preparations, the wound infection rate decreased to 22 percent, but this difference was not significant. One patient in this group had an anaerobic wound infection (mixed with aerobic organisms). This means, compared with the number of anaerobic organisms in the control group, that metronidazole was effective against anaerobic organisms. However, the earlier demonstrated synergistic effect on E. coil after metronidazole has eliminated B. fragilis was not demonstrated after whole gut irrigation. One explanation could be that the effect of whole gut irrigation on the postoperative infection rate could partly obscure any effect of metronidazole. When a combined regimen against anaerobic and aerobic organisms was used in the whole gut irrigation regimen, a remarkable decrease in the wound infection rate was found. The rates were lower than those found for most studies investigating conventional bowel preparation and antibiotics [1I,I3]. The incidence of intraabdominal abscess tended to be lower (p = 0.06), whereas no difference was found for the perineal wounds. In view of the results achieved in this study, it is concluded that whole gut irrigation is rapid, well tolerated, easily performed, and causes far less discomfort than conventional bowel preparation that calls for administration of a laxative and enemas for up to 3 to 4 days. However, whole gut irrigation alone is unlikely to provide any significant protection against postoperative infectious complications. Adding an antimicrobial agent against anaerobic organisms such as metronidazole did not provide significant protection. Combined anaerobic and aerobic antimicrobial prophylaxis together with a

Vokamet49, Me~rch19aS

clean bowel after whole gut irrigation seems to be an ideal preoperative bowel preparation. In our surgical department the combination of metronidazole and ampicillin is effective, but selection of antimicrobials always has to be based on the present sensitivity of the dominant organisms isolated from the infections. Summary In a prospective, randomized double-blind trial,

the efficacy of whole gut irrigation as preoperative bowel preparation for elective colorectal surgery was evaluated alone and in combination with two antimicrobial agents in 148 patients. The antimicrobial regimens were metronidazole alone or metronidazole and ampicillin administered systemically preoperatively and continued for 3 days. Whole gut irrigation was completed without any discomfort in 87 percent of the patients. In 3 percent, the irrigation was stopped and the patients were excluded from the study. Abdominal wound infection developed in 32 percent of the patients after whole gut irrigation, and the addition of metronidazole decreased this incidence to 22 percent (not significant). The incidence in wound infections in the group receiving metronidazole as well as ampicillin was 2 percent, a n d this difference was highly significant compared with both other groups. No significant difference was found for the incidence of intraabdominal abscesses (p = 0.06), infection of the perineal wound, or anastomotic leakage. No difference in the postoperative infection rate was found between a bowel containing fecal fluid or fecal masses, but when a bowel was clean, significantly fewer infectious complications were found. Whole gut irrigation is a rapid, well-tolerated, easily performed, and safe form of preoperative bowel preparation in elective colorectal surgery if combined with systemic antimicrobial prophylaxis consisting of antimicrobial agents effective against anaerobic and aerobic organisms. Acknowledgment: We are indebted to the staff of the wards involved in the investigations for their cooperation and support, which made the trial possible. Furthermore, we thank A/S Rhone-Poulanc for supplying metronidazole and A/S Astra for supplying ampicillin. References 1. Bzxton RC. Postoperativewound infection in colonic and rectal surgery. Br J Surg 1973;60:363-5. 2. Nichols RL, Gorbaoh SL, Condon RE. Alteration of Intestinal mtcroflora following preoperative mechanical preoperatlon of the colon. Dis Colon Rectum 1971;14:123-7. 3. Nichols RL, Condon RE, Gorbach SL, Nyhus L. Efficacy of preoperative antimlcroblal preoperatlon of the bowel. Ann Surg 1972;176:227-32. 4. Hewltt J, Rlgby J, Reeve J, Cox AG. Whole-gut Irrigation in preparation for large-bowel surgery. Lancet 1973;2:33740. 5. Crapp AR, Powls SJA, TIllotson P, Cooke WT, Alexander-Williams J. Preparation of the bowel by whole-gut Irrigation.

321

Go~rupetal

Lancet 1975;2:1239-40. 6. Gottxup F, Proohyiactlc mettonldazoleIn prevention of infection after appendlcectomy: report of a double-blind trial. Acts Chlr Scand 1980;146:133-6. 7. G o , u p F, Saks¢ P. Metronldazole and appendicitis. Can a preoperative prophylaxis be chenged to a peroperatlve treatment In high risk patients? Acts Ch:r Scand 1981; "~47:445-6. 8. RaeJ'~veD, H&,I Hansen O, Cartensen HE, Frlls-~ller A. Septic wound complications after whole bowel Irrlgatlon before colorectal operations. Acts Chlr Scand 1981;147:215-8. 9. Andersen B, Komer B, (~sterg~-d AH. Topical amplclllln against wound Infection after colorectal surgery. Ann Surg 19)2; 176:129-32. 10. Kelghley MRB, Arabl Y, Alexander-Williams J, Young SD, Burdon DW. Comparison between systemic and oral antimicrobial prophylaxis in colorectal surgery. Lancet 1979; 1:894-7, 11. Kelghley MRB. Prevention and treatment of Infection In colorectal surgery. World J Surg 1982;6:312-20, 12. Polk HC. Prevention of surgical wo4~,,1Infection. Ann Intern IVied 1978;89:770-3. 13. Polk HC, Finn MP. Chemoprophylaxls of wound Infections. In: Simmons RL, Howard RJ, eds. Surgical Infectious disease, New York: Appleton-Century-Crofts, 1982:473-85. 14. Slams TG, Carey LC, Fass RJ. Comparative efficacy of prophylactic cephalothln and cefamandole for elective colon surgery. Am J Surg 1979;137:593-6. 15. Arenholz DH, Simmons RL. Mixed end synergistic infections. In: Simmons RL, Howard RJ, eds. Surgical infectious disease. New York: Appleton-Century-Crofts, 1982:119-33. 16. Jones FE, Condon RE. Infections of the colon. In: Simmons RL, Howard RJ, eds. Surgical infectious disease. New York: Appleton-Century-Crofts, 1982:921-73. 17. GoWup F, Hunt TK. Antlmlcrobial prophylaxis in appendectomy patients. World J Surg 1982;6:306-11. 18. Ingham HR, Stsson PR, Tharagonnet D, Selkon JB, Codd AA, Inhibition of phagocytosis in vitro by obligate anaerobes. Lancet 1977;2:1252-4, 19. Onderdonk AB, Louie TJ, Tally FP, Bartlett JG. Activity of m~. tronldazota against Escherlchia colt In experimental Intra° abdominal sepsis. J Antlmlcrob Ctmmother 1979;5:201. 20. Kelly MJ. Wound infection: a controlled clinical and experl-

322

21. 22. 23.

24.

25. 26. 27.

28. 29.

30. 31. 32.

mental demonstration of syncrgy between aerobic (Eschedchia coil) and anaerobic (Bacteroldes fragllls) bacteria. Ann R Coil Surg Engl 1980;62:52-9. Rosenkllde Olson P, Andersen HH, HebJ~rnM, MOiler Pedersen V, Kuld Hansen L. The prophylaxis of metronldazole in colorectal surgery. Dan Med Bu!l 1983;30:345-8. Eykyn SJ. The microbiology of postoperative bacteremia. World J Surg 1982;6:268-72. Sabath LD, Simmons RL, Howard RJ, Canafax DM. Antlmlcroblal agents. In: Simmons RL, Howard RJ, eds. Surgical Infectious disease. New York: Appleton-Century-Crofts, 1982:359-416. Levy AG, Benson JW, Hewlett EL, Herdt JR, Doppman JL, Gordon RS. Saline Lavage: a rapid effective and acceptable method for cleansing the gastrointestinal tract. Gastroenterology 1976;70:157-61. Downing R, Dorrlcott NJ, Kelghley MRB, Oatas GO, Alexander-Williams J. Whole gut Irrigation: a survey of patient oplnlon. Br J Surg 1979;68:201-2. GolckthgJ, Scott A, McNaught W, G;gesple G. Prophylactic ors! antlmlcrobial agents In elective colonic surgery. A conlrolled trial. Lancet 1975;2:997-9. Clarke JS, Condon RE, Bartlett JG, Gorbach SL, Nichols RL, Ochl S. Preoperative oral antibiotics reduce septic complications of colon operations, Results of prospective, ran* domlzed, double-blind clinical study. Ann Surg 1977;186: 251-8. Willis AT, Ferguson IR, Jones PH, et aL Metronldazole in pre. vention and treatment of bacteroldes lnfecUons In elective colonic surgery. Br Med J 1977; 1:607-10. Matheson DM, Arabl Y, Baxter-Smith D, Alaxander-Wllliams J, Keigh!ey MRB. Randomized multicentre trial of oral bowel preparation and antlmlcrobials for elective colorectal operations. Br J Surg 1978;65:597-600. Eykyn SJ, Jackson BT, Lockhart-Mumrnery HE, Phillips I. Pro* phyiactlc peroperatlve intravenous metronldazole In elective colorectal surgery. Lancet 1979;2:761-4. Chung RS, Gurll NJ, Berglund EM. A controlled clinical trial of whole gut lavage as a method of bowel preparation for co* Ionic operations. Am J Surg 1979;137:75-80. Chrlstensen PB, Kronborg O. Whole gut irrigation versus enema in elective colorectal surgery. A prospective randomized study. Dis Colon Rectum 1981;24:592-5.

The knr..lcan J ~ m l of Surgery