World J. Surg. 21, 811– 815, 1997
WORLD Journal of
SURGERY © 1997 by the Socie´te´ Internationale de Chirurgie
Antibiotic Prophylaxis and Open Groin Hernia Repair Eric W. Taylor, F.R.C.S.,1 Derek J. Byrne, F.R.C.S.,2 David J. Leaper, F.R.C.S., Ch.M.,3 Stephan J. Karran, F.R.C.S., M.Ch.,4 M. Kennedy Browne, F.R.C.S., M.D.,5 Kenneth J. Mitchell, F.R.C.S.6 1
Vale of Leven District General Hospital, Alexandria, Dunbartonshire G83 OUA, Scotland, UK Ninewells Hospital, Dundee DD2 1UB, Scotland, UK 3 Southmead Hospital, Westbury-on-Trym, Bristol BS10 5N, England, UK 4 Southampton General Hospital, Tremona Road, Southampton S09 4XY, England, UK 5 Monklands Hospital, Monkscourt Avenue, Airdrie ML6 OJS, Scotland, UK 6 Royal Alexandra Hospital, Corsebar Road, Paisley PA2 9PL, Scotland, UK 2
Abstract. Antibiotic prophylaxis is not routinely given for nonimplant, clean operations, although this view has recently been challenged. We have conducted a randomized multicenter, double-blind prospective trial to compare co-amoxiclav with placebo in 619 patients undergoing open groin hernia repair. Altogether 563 (91%) patients fulfilled the protocol; 283 received co-amoxiclav and 280 placebo. There was no difference between the groups in the number of patients receiving local or general anesthetic, the type of repair performed, the use of a subcutaneous fat suture, the type of skin closure used, the use of wound analgesia, or the use of a wound drain. Patients were given a card to return to the hospital in the event of their wound discharging or their needing to see their general practitioner. All patients were reviewed at approximately 6 weeks after operation. Fifty (8.9%) patients sustained a wound infection, 25 in the co-amoxiclav group and 25 in the placebo group. We conclude that antibiotic prophylaxis is of no benefit to patients undergoing open groin hernia repair.
Open groin hernia repair is regarded as clean surgery, and it is not normal practice in the United Kingdom to administer prophylactic antibiotic cover for such operations. Prophylactic antibiotic administration for clean surgery is not uncommon in other countries; and with the widespread adoption of the Lichtenstein mesh repair or insertion of mesh during laparoscopic hernia repair, the use of prophylactic antibiotics in the United Kingdom may become more popular. Platt and his colleagues [1], in a study of 1218 patients undergoing open groin hernia repair or nonimplant breast operations, showed a significant overall benefit to patients given cefonicid: They had fewer postoperative infections than those who received placebo. Few other studies have reported such a large number of clean operations, but the incidence of wound infection in the Platt et al. study was low after hernia repair: 10 (1.6%) of 612 patients. Patients usually leave hospital either the same day or the day after this type of surgery, and the true incidence of wound infection can be difficult to determine. Bailey and colleagues [2] reviewed hospital notes and found a 3% incidence of wound Correspondence to: E.W. Taylor, M.D.
infection after inguinal hernia repair, but the true incidence of infection was found to be 9% when these patients were followed up closely in the community. Such a level of infection may not disturb hospital bed occupancy or threaten the lives of patients, but it necessitates district nurse time, general practitioner consultations, and considerable inconvenience for the patient, who is often unable to return to work. This incidence of infection is the level usually associated with clean-contaminated or contaminated operations, which would normally be covered by a prophylactic antibiotic. We have sought to verify the findings of the Platt study in a prospective, randomized, double-blind, multicenter study of 619 patients undergoing open groin hernia repair. The organism most commonly isolated from wound infections after groin surgery is Staphylococcus aureus, but other organisms, particularly the gram-negative aerobic bacilli (GNAB) can be significant contaminants in this area of the body. For this reason we selected co-amoxiclav as the antibiotic with appropriate pharmacokinetics and antimicrobial spectrum and therefore most likely to show benefit.
Patients and Methods All patients over the age of 18 undergoing inguinal or femoral hernia repair were considered eligible for this study unless they were known to be hypersensitive to penicillin or cephalosporins, had taken antimicrobial agents within the preceding 72 hours, were undergoing a bilateral hernia repair, were pregnant or lactating, had known renal or liver impairment, or had previously been entered into the study. Patients for whom antibiotic prophylaxis was indicated for some other reason, such as valvular heart disease, were also not considered eligible. All patients gave written informed consent. The study was approved by the ethics committee of each participating hospital. If the wound infection rate proved to be 5.5% in the co-amoxiclav group and 11.0% in the placebo group, tables showed that 380 patients per group would be required to show a statistical difference in a study with 80% power.
812
World J. Surg. Vol. 21, No. 8, October 1997
Table 1. Demographic data, risk factors, and type of operation. Parameter Age, mean 6 SD (years) Weight, mean 6 SD (kg) Male/female Hernia Femoral Inguinal Inguinal hernia repair Darn Bassini Shouldice Lichtenstein Other Fat suture used Skin closure Subcuticular Interrupted sutures Staples Wound analgesia Wound drain
Co-amoxiclav group
Placebo group
56.7 6 17.4 72.4 6 10.6 268:15
56.6 6 16.5 74.3 6 12.1 265:15
10 273
6 274
73 143 36 16 15 87
66 147 43 17 7 86
204 31 42 140 4
190 34 45 130 4
Patients received co-amoxiclav 1.2 g (amoxicillin 1 g with clavulanic acid 200 mg) in 20 ml of sterile saline or were given sterile saline alone (placebo) by intravenous bolus injection at induction of anesthesia. Randomization was from a computergenerated code in blocks of four. Those patients having the operation performed under local or regional anesthesia received the antibiotic at commencement of the operation. Patients were discharged from hospital at the surgeons’ discretion. They were given a diary card on which their general practitioner or district nurse was asked to record any suggestion of a wound or other infection and whether therapeutic intervention, including the use of antibiotics, was required. Patients were also given a bacteriology culture swab to take home and, in the event of the wound discharging, were asked to return this swab to the laboratory for culture. All patients were reviewed at 4 to 6 weeks after operation. Infection was defined, based on clinical criteria, as a purulent wound discharge or spreading erythema indicative of cellulitis, wound breakdown, or dehiscence with clinical evidence of infection [3]. Success or failure of the antibiotic prophylaxis was assessed by the incidence of postoperative wound or other infections. Statistical analysis was made using the chi-square test. Results Six hospitals, four in Scotland and two in England, participated in the study, and 619 patients were recruited between September 1993 and January 1995. No patient refused to enter the study. In 34 patients no study medication was given, 3 had previously entered the study, 2 had prophylaxis administered at the wrong time, 5 had bilateral hernia operations, 3 had no hernia operation, 4 were found to be receiving an antibiotic preoperatively, and in 5 the antibiotics were continued postoperatively in error. These patients were withdrawn from the analysis of outcome, which left 283 evaluable patients who received co-amoxiclav and 280 who received placebo. The demographic data, risk factors for infection, and type of operation performed are shown in Table 1. The two groups were comparable.
Table 2. Total patients and infected patients, by center. Co-amoxiclav group
Placebo group
Center
No. entered
No. infected
No. entered
No. infected (%)
1 2 3 4 5 6 Total
64 90 40 23 48 18 283
6 (9.4) 11 (12.2) 2 (5.0) 3 (13.0) 2 (4.2) 1 (5.6) 25 (8.8)
61 97 33 22 48 19 280
6 (9.8) 11 (11.3) 0 3 (13.6) 3 (6.1) 2 (10.5) 25 (8.9)
Table 3. Analysis of criteria of infection.
Criterion Surgeon’s diagnosis Antibiotics administered Purulent discharge Wound abscess/ breakdown Total
Co-amoxiclav group (n 5 283)
Placebo group (n 5 280)
Total (n 5 563)
2 (0.7%) 11 (3.9%)
1 (0.4%) 14 (5.0%)
3 (0.5%) 25 (4.4%)
3 (1.1%) 9 (3.2%)
4 (1.4%) 6 (2.1%)
7 (1.2%) 15 (2.7%)
25 (8.8%)
25 (8.9%)
50 (8.9%)
Overall 50 (8.9%) patients developed a wound infection, 25 in the co-amoxiclav group and 25 in the placebo group (p 5 0.97). The number of patients with infection in each participating center is shown in Table 2. The incidence of infection ranged from 0 to 13.6%, but in some centers the numbers enrolled were small and in no center was the difference between the co-amoxiclav and placebo groups statistically significant. Nine patients developed a chest infection requiring treatment, 6 in the co-amoxiclav group and 3 in the placebo group. Seven patients developed a urinary tract infection, 5 in the co-amoxiclav group and 2 in the placebo group. Wound infection was diagnosed by the participating surgeon for unspecified reasons in 3 patients, by the administration of antibiotics for wound erythema in 25 patients, by a purulent discharge in 7, and by a wound abscess or wound breakdown with infection in 15 patients. The details of these infections are shown in Table 3. In addition six patients sustained wound breakdown that was not considered to be due to infection. An analysis was undertaken to assess the significance of the following factors in the etiology of infection: age (above or below 60 years), weight (above or below 70 kg), sex, duration of operation, skin preparation, skin shaving, type of anesthesia, type of repair, type of suture used for repair, method of skin closure, use of subcutaneous fat stitch, use of wound analgesia, and the grade of surgeon performing the operation. The analysis showed that no specific risk factor was associated with an increased incidence of infection (Table 4). Discussion We have conducted a prospective, randomized, double-blind, multicenter study to compare the efficacy of co-amoxiclav with placebo as antibiotic prophylaxis in patients undergoing open groin hernia repair. We have found no benefit from administering a prophylactic antibiotic and have been unable to confirm the
Taylor et al.: Antibiotics for Open Groin Hernia Repair
813
Table 5. Bacterial isolation from wound infections.
Table 4. Risk factors and postoperative infection. Risk factor Age , 60 years . 60 years Weight , 70 kg . 70 kg Gender Male Female Duration of operation , 35 minutes . 35 minutes Grade of surgeon Consultant SR/Reg/Staff SHO Skin preparation Betadine Chlorhexidine Skin shaved Preoperative day Day of operation Anesthetic General Spinal Type of repair Darn Bassini Shouldice Lichtenstein Suture for repair Braided Monofilament Fat suture Yes No Skin closure Subcuticular Interupted suture Staples Wound analgesia Yes No
No.
No. infected
%
300 263
28 22
9.3 8.4
228 318
17 31
7.5 9.7
533 30
46 4
8.6 13.3
276 274
21 27
7.6 9.9
133 310 138
15 23 14
11.3 7.4 10.1
338 198
31 16
9.2 8.1
260 175
27 15
10.4 8.6
406 144
32 17
7.9 11.8
139 290 79 33
14 28 4 2
10.1 9.7 5.1 6.1
72 477
6 44
8.3 9.2
173 381
18 32
10.4 8.4
394 65 87
34 8 8
8.6 12.3 9.2
270 286
22 28
8.1 9.8
Not all data were available for all patients. SR: senior registrar; Reg: registrar; SHO: senior house officer; Staff: staff grade surgeon.
results of Platt and colleagues [1] who reported a significant reduction in the overall incidence of infection in patients receiving a single dose of cefonicid. It should be noted that the difference in the incidence of wound infection alone in the Platt et al. study was not significant in either the hernia or the breast surgery groups, and we have not found the high incidence of other infections they reported. We initially sought to recruit 760 patients, 360 to each group. However sequential analysis by the sponsor after 15 months of the study (619 patients) showed that continuing to that endpoint was unlikely to produce any significant difference between the groups, and it was thought appropriate to discontinue the study at that point. We believe co-amoxiclav to be an appropriate choice of antibiotic for this study. It has appropriate pharmocokinetics and good activity against Staphylococcus aureus and Staphylococcus epidermidis, which are the major pathogens associated with this type of
Bacterium
No.
Staphylococcus aureus Staphylococcus albus Streptococci b-Hemolytic Anaerobic Unspecified Coliforms Enterobacter aerogenes
14 1 2 1 1 1 1
surgery. S. aureus was the organism isolated most frequently from postoperative infections in this study (Table 5). Co-amoxiclav is also active against the GNAB that frequently contaminate the groin and that may also be pathogens in this area. We have found no evidence to suggest that resistance to the antibiotic was a reason for failure of prophylaxis. We found it surprising that none of the commonly recognized risk factors for infection after clean operations proved to be linked to the high incidence in this study. It is particularly of note that the seniority and, by presumption, experience of the surgeon made no difference. Neither did the technique of repair, the use of staples or sutures, or other variations in the manner in which the operation was conducted. The criteria used to diagnose infection are crucial to studies such as this one. If only purulent wound discharge, wound abscess, or wound breakdown due to infection had been taken as the diagnostic criterion, the overall infection rate would have been 3.9%. There still would have been no benefit from prophylaxis. The incidence of other nosocomial infections was low (2.8%) and again shows no benefit from prophylaxis. We are disappointed and concerned to report such a high incidence of infection after allegedly clean operations, but, we note that this incidence is not dissimilar to that reported by others [2, 4, 5]. This incidence of infection is well into the range of infection that one would anticipate after clean-contaminated operations. Antibiotic prophylaxis is considered to be of proved value in clean-contaminated operations, and so we are at a loss to explain why we have not been able to show efficacy. These patients all underwent open groin hernia repair. It is perhaps significant that open cholecystectomy is associated with an incidence of infection in the region of 5% to 10% [6, 7], yet laparoscopic cholecystectomy has been reported to be complicated by a much lower incidence of infection, in the region of 0.5% [8, 9]. It will be of interest to see what incidence of infection is reported when sufficiently large studies of laparoscopic hernia repair are published [10, 11], and it may well be that the difference in infection rates could become a reason to move from an open to a laparoscopic technique. In view of the proximity of the groin to the perineum, it is arguable that open groin hernia repair should no longer be considered a “clean” procedure; but if it were reclassified as “clean-contaminated,” we would have anticipated some benefit from administration of a prophylactic antibiotic and we have shown none. This finding serves only to emphasize the importance of greater attention to other methods to prevent surgical wound infection, such as skin preparation and tissue handling, and to stress the importance of a strict audit of postoperative complications.
814
Re´sume´ Bien que cette position ait ´ete´ de´battue dernie `rement, l’antibioprophylaxie n’est pas syste´matique en cas d’interventions propres, en l’absence d’implant. Nous avons mene´ une ´etude multicentrique randomise´e `a double aveugie pour comparer l’association amoxycilline-acide clavulanique `a un placebo chez 619 patients ayant ´ete´ ope´re´s d’une hernie inguinale par voie traditionnelle. Cinq cent soixante-trois des patients (91%) ayant satisfait aux conditions du protocole, 283 ont rec¸u de l’association amoxycilline-acide clavulanique et 280, du placebo. Il n’y avait aucune diffe´rence entre les groupes en ce qui concerne le nombre de patients ayant eu une anesthe´sie locale ou ge ´ne´rale, le type de hernie ope´re´e, l’utilisation d’un plan sous-cutane ´ ou pas, l’utilisation d’un anesthe´sique dans la plaie ou l’utilisation d’un drain. On a demande´ aux patients de revenir `a l’ho ˆpital en cas de suintement de la plaie et s’ils souhaitaient voir leur me´decin. Tous les patients ont ´ete´ revus environ six semaines apre`s l’ope´ration. Cinquante patients (8.9%) ont eu une infection de leur plaie, 25 (8.8%) dans le groupe amoxycilline-acide clavulanique et 25 (8.9%) dans le groupe placebo. Nous concluons que l’antibioprophylaxie n’est pas ne´cessaire chez le patient ayant ´ete´ ope´re´ d’une hernie inguinale par voie traditionnelle. Resumen No se administra profilaxis antibio ´tica en forma rutinaria en pacientes sometidos a operaciones limpias, aunque esta polı´tica ha sido recientemente cuestionada. Hemos realizado un estudio multice´ntrico randomizado prospectivo doble ciego, con el fin de comparar co-amoxiclav con placebo en 619 pacientes sometidos a reparacio ´n abierta de hernia inguinal. Quinientos sesenta y tres pacientes (91%) cumplieron el protocolo, 283 recibieron coamoxiclav y 280 placebo. No hubo diferencia en los grupos en cuanto al nu ´mero de casos que recibieron anestesia local o general, el tipo de reparacio ´n, el uso de suturas en la grasa subcuta´nea, el tipo de cierre de la piel, el uso de analgesia de la herida o el uso de un dren. Se entrego ´ una tarjeta a cada paciente con instruccio ´n de retornar al hospital en el caso de presentarse drenaje por la herida o de que requiriera servisto por el me´dico general. Todos los pacientes fueron vistos en control aproximadamente a las 6 semanas despue´s de la operacio ´n. Cincuenta (8.9%) de los pacientes presentaron infeccio ´n, 25 (8.9%) en el
Invited Commentary Christos A. Pissiotis, M.D. Department of Surgery, University of Athens School of Medicine, Athens, Greece
This meticulously designed, carefully conducted study shows a high incidence of wound infection associated with groin hernia repair (close to 9%) that a rationally chosen antibiotic could not prevent. The authors showed that the well known surgical infection risk factors played no role in this high incidence of wound infection. It is not understood why co-amoxiclav did not prevent postop-
World J. Surg. Vol. 21, No. 8, October 1997
grupo co-amoxiclav y 25 (8.9%) en el grupo placebo. Nuestra conclusio ´n es que la profila´xis antibio ´tica no aporta beneficio en pacientes sometidos a reparacio ´n abierta de hernia inguinal. Acknowledgments The authors express their gratitude to the staff of the Medical Department of SmithKline Beecham Pharmaceuticals Ltd. for a grant for the study, for supplying the co-amoxiclav, and for assistance with the statistical analysis of the study. References 1. Platt, R., Zaleznik, D.F., Hopkins, C.C., Dellinger, E.P., Karchmer, A.W., Bryan, C.S., Burke, J.F., Wikler, M.A., Marino, S.K., Holbrook, K.F.: Perioperative antibiotic prophylaxis for hernia and breast surgery. N. Engl. J. Med. 322:153, 1990 2. Bailey, I.S., Karran, S.E., Toyn, K., Brough, P., Ranaboldo, C., Karran, S.G.: Community surveillance of complications after hernia surgery. BMJ 304:46, 1992 3. Peel, A.L.G., Taylor, E.W.: Proposed definitions for the audit of post-operative infection: a discussion paper. Ann. R. Coll. Surg. Engl. 73:385, 1991 4. Simchen, E., Wax, Y., Pevsner, B.: The Israeli study of surgical infections. II. Initial comparisons among hospitals with special focus on hernia operations. Infect. Control Hosp. Epidemiol. 9:241, 1988 5. Kingsnorth, A.N., Gray, M.R., Nott, D.M.: Prospective randomised trial comparing with Shouldice technique and plication darn for inguinal hernia. Br. J. Surg. 80:403, 1993 6. Wells, G.R., Taylor, E.W., Lindsay, G., Morton, L., West of Scotland Surgical Infection Study Group: The relationship between bile colonisation, high risk factors and post-operative sepsis in patients undergoing biliary tract operations whilst receiving a prophylactic antibiotic. Br. J. Surg. 76:374, 1989 7. Hambraeus, A., Laurell, G., Nybacka, O., Whyte, W.: Biliary tract surgery: a bacteriologic and epidemiologic study. Acta Chir. Scand. 156:155, 1990 8. Southern Surgeons Club: A prospective analysis of 1518 laparoscopic cholecystectomies. N. Engl. J. Med. 324:1073, 1991 9. Litwin, D.E., Girotti, M.J., Poulin, E.C., Mamazza, J., Nagy, A.G.: Laparoscopic cholecystectomy; trans Canada experience with 2201 cases. Can. J. Surg. 35:291, 1992 10. Davies, N.M., Dunn, D.C., Appleton, B., Bevington, E.: Experience with 300 laparoscopic inguinal hernia repairs with up to 3 years follow up. Ann. R. Coll. Surg. Engl. 77:409, 1995 11. Phillips, E.H., Arregui, M., Carroll, B.J., Corbitt, J., Crafton, W.B., Fallas, M.J., Filipi, C., Fitzgibbons, R.J., Franklin, M.J., McKernan, B.: Incidence of complications following laparoscopic hernioplasty. Surg. Endosc. 9:16, 1995
erative wound infection, since Staphylococcus aureus was the organism isolated most frequently from postoperative infections in this study, and normally, co-amoxiclav prevents S. aureus postoperative wound infections. Even if one considers the criteria for wound infection in this study austere and accepts only the purulent wound discharge and wound abscess as diagnostic criteria for wound infection, still the overall infection rate for this type of surgery is high. What is most important is that there is no difference between the co-amoxiclav and placebo groups (their Table 3). Perhaps the answer to the question of the high incidence of wound infection in this study lies in the reported differences of infection rates among centers. Why in center 3 was the wound infection rate 0 and in center 4 13.6% (their Table 2)?
Taylor et al.: Antibiotics for Open Groin Hernia Repair
This study brings up another important point. Perhaps it is time to change the present classification system of operations according their degree of contamination to another classification system based on the infection risk or the actual infection rate. Infection rates at the level of 8% to 10% characterize clean-contaminated and some contaminated procedures. Yet we are aware of cleancontaminated operations that have a less than 8% wound infection rate (cholecystectomy). Recent Publications on clean operations have reported infection rates higher than 8%. Maybe it is time to redefine well known infection risk factors according to their degree of importance for predicting wound infection. For example, groups of patients , 60 and those . 60 years of age show no difference in wound infection, but groups , 40 and those . 70 years of age perhaps do. Also the duration of operations (, 35 and . 35 minutes) show no difference in infection rates, but groups of operations with durations of , 35 and . 70 minutes perhaps do.
815
The conclusion that can be drawn from this interesting study is that each center, through a strict audit of wound infection rates, should identify the operations that carry a high incidence of wound infection and apply all the necessary precautions (i.e., skin preparation, meticulous tissue handling, and surgical technique) to avoid unpleasant consequences. The authors should be commended for measuring infection rates 4 to 6 weeks postoperatively, as it has been established that almost 50% of wound infections occur during the 4-week period after discharge from the hospital. We are going through an era of enthusiasm for laparoscopic surgery, but reported wound infection rates of 0.5% for laparoscopic cholecystectomy or inguinal hernia repair should be viewed with cautious optimism if not with criticism. Comparison of infection rates between open and laparoscopic procedures would be of value if they are performed in the same center under a strict audit for postoperative complications.