A comparison of the prophylactic efficacy of ceftriaxone and cefotaxime in abdominal surgery

Methods 

A total of 1,013 patients admitted into a general surgical unit for acute and elective open abdominal surgery were studied over a 3-year period in a prospective, randomized, double-blind study. Patients aged 16 years and over who met the conventional standards of antibiotic prophylaxis for their surgery were consecutively recruited. A clinical hospital pharmacist, using a blocked randomized technique (cell size of 10) with stratification, independently randomized the patients to antibiotic A or antibiotic B. The “noncolorectal” stratification included hepatobiliary, esophageal-gastricduodenal surgery, small bowel surgery, appendicectomy, splenectomy and vascular surgery. The “colorectal” stratification included all colorectal surgery and stomal surgery.

Intravenous cephalosporin, 1 g, was given at induction of anesthesia for all abdominal surgery; IV metronidazole, 500 mg, was also given at induction for those in the “colorectal” stratification. The identity of antibiotic A and antibiotic B was not known until after the completion of the study.

Exclusion criteria included active infection requiring treatment before or at the time of surgery, death within 30 days of surgery when no wound infection had developed, cephalosporin or metronidazole allergy, and deviation from the trial protocol such as incorrect antibiotic administration. An interim analysis was performed after 400 patients had been recruited and then at intervals of 200. The institution’s ethics committee approved the study and support staff obtained written consent.

The major endpoint was wound infection with pus formation, or cellulitis, defined as erythema and induration greater than 1 cm laterally, for at least two thirds of the wound length or 6 cm, whichever was the lesser. The minor endpoints included the following: deep peritoneal infection required radiological or operative demonstration of a collection with clinical evidence of sepsis or the culture of a pathogen on aspiration. If the infection was due to an anastamotic leak then this was classified as anastamotic morbidity. If there was no demonstrable anastamotic problem then this was classified as a deep infection. Chest infection required at least two of the following: clinical signs of chest infection, purulent sputum with a recognized pathogen, radiological demonstration of an infective process. Urinary tract infection required a urine with >108 bacteria per L or >107 white blood cells per L and a positive culture or correlation with clinical symptoms and signs. Febrile morbidity was defined as an oral temperature greater than 38°C on two occasions, more than 6 hours apart, and greater than 24 hours postoperatively. Other infections included septicemia, infective diarrhea with testing for Clostridium difficile toxin, yeast superinfection, intravenous line sepsis, and drain site infection.

A trained research nurse in conjunction with the surgical team performed clinical evaluation daily from surgery until hospital discharge. Subsequently a multidisciplinary team comprising of two surgeons, a medical microbiologist, a clinical pharmacist and the research nurse confirmed the assessment of each case on a weekly basis. All patients were reviewed at a minimum of 30 days postoperatively, either as an inpatient, as an outpatient at a surgical or wound clinic, or telephoned by the research nurse using a standardized questionnaire. The surgical audit was checked to ensure that all infections were included in the final study results.

Wound swabs were cultured aerobically only, unless the Gram stain was consistent with an anaerobic or polymicrobic infection. Sputum specimens were screened for “acceptability” before being processed. Specimens from “deep” or normally sterile sites were cultured aerobically and anaerobically. Identification of Gram negative aerobic and facultative anaerobic bacilli was done using a modified agar dilution, replica-plating method [12]. Other organisms were identified by standard methods [13]. Sensitivity testing of most aerobic pathogens was done by agar-dilution method, using NCCLS break-points [14]. Iso-Sensitest agar (Oxoid) was used for the replica-plating method, except for methicillin sensitivity when Mueller-Hinton agar (Gibco) was used. Other aerobic organism sensitivity testing was performed by disc sensitivity [15]. For fastidious organisms Iso-Sensitest agar was supplemented with 5% defibrinated sheep blood. Anaerobic sensitivity testing was not performed routinely [13]. All sensitivity results were recorded as sensitive, intermediate or resistant. Cefotaxime pure substance was supplied by Roussel. Desacetylcefotaxime was not tested. Ceftriaxone pure substance was supplied by Roche. Breakpoint sensitivity concentrations for both antibiotics were 8 μg/mL and 64 μg/mL.

The trial was designed to test for a 5% difference in infection rates between the two treatment groups, with a confidence of 95% and a power of 80%. Using a chi-square analysis with a two-tailed alternative hypothesis a sample size of 434 was required in both arms of the trial. The chi-square test was used to test for differences in endpoint and microbiology results. Yates’ correction and a two-tailed Fisher’s exact test were used to correct for small numbers. The analysis was performed on an intention-to-treat basis.

Results 

Of 1,013 patients randomized, 93 (9%) were excluded from analysis, 44 in the ceftriaxone group and 49 in the cefotaxime group. The numbers for each exclusion criteria were comparable in each group. The two most frequent were the use of therapeutic antibiotics (49) and death within 30 days without a wound infection (33). Patient and demographic data (Table 1) were comparable except that by chance cefotaxime was used more in both nonelective surgery and intermediate surgery. This difference was mainly due to appendectomies, which are analyzed separately and so do not confound the overall results of the study.

Table 1. Patient demographic and operative data

		Ceftriaxone	Cefotaxime
	Number	462	458
	Age (years ± SD)	51.6 (21.3)	49.6 (21.5)
	Sex, M/F	212/250	226/232
	Duration of surgery (minutes ± SD)	93 (66)	92 (68)
	Timing of surgery*
	Nonelective	223	261
	Elective	239	197
	Operation category
	Major	333	306
	Intermediate	129	152
	ASA score
	Grade 1	114	129
	Grade 2	105	102
	Grade 3	64	58
	Grade 4 and 5	13	10
	Wound category
	Clean	105	99
	Possibly contaminated	128	124
	Contaminated	169	180
	Dirty	60	55
	Gastrointestinal anastomoses	187	177
	Drains	184	160
	Surgeon
	Consultant	196	188
	Resident	266	270

Statistics: two-tailed chi-square test and Student’s t test. ASA = American Society of Anesthesiologists score.

* P <0.05.

The overall wound infection rate (Table 2) with ceftriaxone was 8% compared with 12% for cefotaxime, P <0.05. At the interim analysis for the first 800 patients a significant difference was already observed. This was due to appendicectomies being performed without metronidazole, with wound infection rates of 6% with ceftriaxone and 18% with cefotaxime. The antibiotic code was not broken and subsequent appendicectomies were entered into the “colorectal” strata. After exclusion of the initial appendicectomy bias the wound infection rates were similar (R 8%, C 10%), with infection rates of 6% and 10% for “noncolorectal” surgery and of 10% for “colorectal” surgery for both agents.

Table 2. Wound infection

		Ceftriaxone	Cefotaxime	P value
	All cases
	Noncolorectal	17/276	36/281	<0.01
	Colorectal	19/186	18/177	NS
	Total	36/462	54/458	<0.05
	Appendectomy
	Noncolorectal	5/83	18/100	<0.05
	Colorectal	2/23	3/27	NS*
	Adjusted
	Noncolorectal	12/193	18/181	NS*
	Total	31/379	36/358	NS

Adjusted: wound infection results with all noncolorectal appendectomies excluded. Statistics: two-tailed chi-square. NS = not significant.

* Yates’ correction used.

The difference in anastamotic morbidity (Table 3) approached significance at 3% with ceftriaxone and 8% with cefotaxime, P = 0.054 after Yates’ correction. Deep infection, chest infection, urinary tract infection, and septicemia were less frequent with ceftriaxone. Superinfections with Clostridium difficile diarrhea or a yeast infection were less frequent with cefotaxime (Table 3). Most cases of febrile morbidity were associated with another infection or a clinical diagnosis of pulmonary atelectasis.

Table 3. Peritoneal and Distant Infections

		Ceftriaxone	Cefotaxime
	Number	462	458
	Peritoneal Infection
	Deep infection	3	8
	Anastamotic morbidity*	6/187	15/177
	Distant infection
	Chest infection	11	19
	Urinary infection	14	23
	Clostridium difficile	4	0
	Yeast infection	12	7
	Febrile morbidity	23	25
	Septicemia	2	6
	IV line infection	6	9
	Drain site infection	3/184	5/160

Statistics: two-tailed chi-square test, with Yates’ correction and two-tailed Fisher’s exact test as indicated by small numbers. IV = intravenous.

* P = 0.054, Yates’ correction used.

The noncolorectal appendectomies were excluded from the analysis of the combined endoints (Table 4) as cefotaxime without metronidazole administration did not appear to give adequate protection for appendectomy. The number of patients having either or both a chest or a urinary infection were significantly reduced with ceftriaxone (R 6% C 11%, P = 0.01). The number of patients developing “any infection,” which included all endpoints except febrile morbidity, was also significantly reduced with ceftriaxone (R 20% C 27%, P = 0.03).

Table 4. Combined infections

	Infection	Ceftriaxone	Cefotaxime	P value
	Number	379	358
	Chest and urinary	23	40	0.01
	CLD and yeast	12	6	NS
	Peritoneal and distant*	46	61	NS
	Any†	77	97	<0.05

Statistics: two-tailed chi-square test. CLD = Clostridium difficile diarrhea; NS = not significant.

* Peritoneal infection includes deep infection and anastamotic morbidity; distant infection includes all distant infection except febrile morbidity. † Any infection is wound infection, distant infection, and peritoneal infection.

Differences in wound microbiology were due to staphylococcal infections. Of the 66 wound infections with a cultured growth (Table 5) 29 of these were polymicrobial. 42% grew skin microbes, 35% enteric microbes, and 23% a mixture of both. Staphylococcus aureus was the most important pathogen, being isolated in 29 of 66 wound cultures. It was cultured from 7 of 462 patients with ceftriaxone and in 22 of 458 patients with cefotaxime (P <0.01). It was isolated in 29% and 52% of positive wound cultures for ceftriaxone and cefotaxime, respectively (P = 0.11). It was also the most frequently isolated organism with chest infection, being present in 7 of 28 positive ceftriaxone wound and chest cultures and in 28 of 50 cefotaxime cultures (P <0.05). In contrast the incidence of Staphylococcus epidermidis wound infection was not significantly different (8 of 462 versus 2 of 458), but it was isolated in 33% and 5% of positive wound cultures for ceftriaxone and cefotaxime, respectively (P <0.01). The organisms isolated from deep peritoneal and urinary tract infections were comparable. The antibiotic sensitivity profile was similar for both antibiotics. No MRSA were isolated.

Table 5. Microbiology

		Ceftriaxone	Cefotaxime
	Wound
	Wound infections	36	54
	Cultures	27	44
	Cultures with growth	24	42
	Microbes
	Staphylococcus aureus	7	22
	Staphylococcus epidermidis*	8	2
	Staphylococcus species	3	4
	Bacteroides fragilis	8	9
	Escherichia coli	4	11
	Enterococcus faecalis	3	3
	Streptococcus anginosus	1	6
	Klebsiella pneumoniae	0	1
	Klebsiella oxytoca	1	0
	Proteus mirabilis	0	4
	Pseudomonas aeroginosa	3	2
	Morganella morganii	0	1
	Other	1	3

Statistics: two-tailed chi-square test and two-tailed Fisher’s exact test.

* P <0.01.

Comments 

Both antibiotics were confirmed to be effective prophylactic agents with wound infection results similar to that previously reported [16].

An important difference between the two antibiotics was identified at the interim analysis when cefotaxime alone did not provide satisfactory prophylaxis for appendectomy. The effectiveness of ceftriaxone alone had previously been demonstrated [17], [18]. The use of cefotaxime was supported by the observation that its in vitro activity when combined with its desacetyl metabolite was superior to that of ceftriaxone [19], [20]. The better performance of ceftriaxone, in spite of differences in in-vitro activity, is likely due to its excellent TPI and long half-life.

Although the overall frequency of wound infection was similar once the “noncolorectal” appendicectomies were corrected for, further differences were observed in the frequency of organisms isolated and in the severity of infection. Staphylococcus aureus was more frequently isolated with cefotaxime and S epidermidis with ceftriaxone. Staphylococcus epidermidis is a skin commensal that is difficult to eradicate. Its presence in wound infections when broad-spectrum antibiotics are being used [18] suggests it may have the ability to colonize wounds when other bacteria, such as S. aureus, are inhibited. The issue of the severity of infection is studied in our paper on the cost of postoperative infections (submitted to the Annals of Surgery).

The observation that ceftriaxone was also effective in reducing the frequency of some infections remote to the wound, raises the possibility of prophylactic ceftriaxone, and possibly other prophylactic antibiotics, having a greater roll than the prevention of wound infection alone. The reduction in chest and urinary tract infection with ceftriaxone was consistent with previously reported studies on prophylactic antibiotic use where ceftriaxone decreased chest [22], [23], urinary tract [24], and both chest and urinary tract [17] infections.

The difference in anastomotic morbidity fell just short of significance. The possibility of an anastomotic infection resulting in an increase in collagenolytic activity and adversely influencing anastomotic integrity has been demonstrated [25], [26], but animal and clinical studies have given inconsistent results [27], [28], [29]. Although the differences in anastomotic morbidity was probably due to chance differences in blood supply and surgical technique, this study supports further evaluation of the role of anastomotic infection and of ceftriaxone use on anastomotic collagenase activity.

For optimal effectiveness prophylactic antibiotics should be at therapeutic tissue concentration at the time of bacterial contamination. The excellent TPI and long half-life of ceftriaxone results in a therapeutic tissue concentration, which is greater than that of cefotaxime [30] and is sustained for 24 hours [31]. This results in both effective bacterial eradication [28] and a minimization of bacterial synergism [32]. For wound infection the time of contamination is completed when the wound is closed. This is confirmed by the initial observations of Burke [1] and by studies comparing single and multiple doses of antibiotics [33], [34]. In contrast for infection remote to the wound, bacterial contamination may occur in the first 24 hours after the completion of surgery. A reduced cough reflex, shallow breathing, and atelectasis may contribute to early postoperative colonization of the respiratory tract. Impaired emptying of the bladder and catheter trauma may also contribute to postoperative urinary tract infection [35]. The sustained antimicrobial cover given by a single intravenous dose of ceftriaxone may therefore contribute to reducing the incidence of postoperative infection remote to the wound [21].