The rate of surgical site infection (SSI) after liver transplantation (LT) is generally higher than that following other types of solid-organ transplantation, and SSI remains a major cause of morbidity and mortality.1–4 The high incidence of post-LT infection is related to the technical complexity of the procedure itself, the fact that it is performed in a potentially infected milieu within the abdominal cavity, and the markedly poor medical condition of many recipients.5 SSIs, including incisional wounds, intra-abdominal abscesses, and peritonitis, are frequent post-LT bacterial infections.5–6
The purpose of this study was to determine the incidence, timing, location, and risk factors, specifically antibiotic prophylaxis, for SSIs after LT.
CI, confidence interval; LT, liver transplantation; RBC, red blood cell; RESITRA, Red de Estudio de la Infección en el Trasplante; SSI, surgical site infection.
PATIENTS AND METHODS
A prospective study was performed that included a population of all consecutive patients who underwent deceased-donor whole LT in 11 Spanish hospitals belonging to the Red de Estudio de la Infección en el Trasplante (RESITRA) research network between August 2003 and September 2005. RESITRA is a research network financed by the Spanish Ministry of Health and formed by 16 transplant centers and 3 reference laboratories that prospectively includes all SOT and hematopoietic stem cell transplant recipients from these hospitals. It was created in April 2003. Each hospital has a research team (medical doctor, nurse, and technician) that manages the research tools. An online database and a serum, plasma, microorganism, and DNA bank are used as research tools. Pretransplant, peritransplant, and follow-up (days 0, 7, 14, 30, 60, 90, 180, 270, 360, and 720 after transplantation) data are prospectively included in the online database, as well as all infections (diagnostic workup, clinical presentation, therapy, and outcome) and rejection episodes, always according to the predefined definitions published on the RESITRA Web site (www.resitra.retics.net/web/index.asp)7 and based on universally accepted criteria. Data collection is made through PDF e-forms, which are sent to an SQL-server database located on the Web site.
Data exploitation is made through managerial and statistical databases generated from the SQL-server database after a validating process managed by each hospital coordinator.
Definitions of Infections
Centers for Disease Control and Prevention criteria definitions for SSI were adopted and were slightly modified to include infections that developed up to 6 months after surgery, including both wound incisional and organ/space infection, hepatic and intra-abdominal abscess, and peritonitis.8
Risk Factors for SSIs
In order to assess risk factors for SSI, only patients who developed SSI within the first 30 days after surgery were included. Only the most severe episode of SSI was included for patients who developed more than 1 SSI.
The following variables were assessed as risk factors for SSIs: (1) pretransplant variables including age, gender, diabetes mellitus, chronic renal failure, Child-Pugh class, elective or emergency surgery, previous transplantation, cause of LT, and cytomegalovirus donor and recipient status and (2) operative and posttransplant variables including duration of transplant surgery, quantity of red blood cell (RBC) transfusions (dichotomized by the median volume transfused), cytomegalovirus disease, type of biliary reconstruction, immunosuppression regimen, and antibiotic prophylaxis regimen. The number of LTs performed in each hospital during the study period was also included as a risk factor. Participant institutions used fluconazole (100 mg/day) as fungal prophylaxis during the period of hospitalization. Forty-one patients (belonging to a single hospital) received intestinal decontamination with norfloxacin (400 mg orally every day).
Continuous variables were compared with the Student t test or, when a normal distribution could not be assumed, the Mann-Whitney test. Categorical data were compared with the chi-square, chi-square for trends, or Fisher exact test when appropriate. Odds ratios and 95% confidence intervals (CIs) were computed for associations. When more than 2 categories were analyzed, a global chi-square, associated P value, and odds ratios comparing with the category with lowest risk were computed. A logistic regression model was used to evaluate variables found to be associated with SSIs by univariate analysis. To examine the effect of antimicrobial prophylaxis on the risk of SSIs, a final model was selected by a backward strategy from a maximal model including all variables associated in the univariate analysis with a P value < 0.29 and adjusting by Child-Pugh class and center. All analyses were performed with Stata SE version 8.0 computer software. All statistical tests were 2-tailed, and a P value < 0.05 was considered statistically significant.
Between August 2003 and September 2005, a total of 1222 patients underwent LT, and 70 of these (5.7%) were receiving a second or third transplant. Eight hundred ninety patients (73%) were male, with an age range of 18 to 71 years (mean, 53 years). The predominant underlying diseases were alcoholic and hepatitis C cirrhosis, which accounted for more than 50% of the cases. The third most frequent cause of LT was hepatocellular carcinoma (20%); less frequent causes were hepatitis B cirrhosis, autoimmune liver disease (including autoimmune cirrhosis, primary biliary cirrhosis, and primary sclerosing cholangitis), fulminant hepatitis, and toxic or deposit liver disease (Table 1).
Table 1. Main Diagnosis and Risk of Surgical Site Infection in the Liver Transplant Cohort
One hundred twenty-seven episodes of SSIs were recorded in 107 transplant patients. Eighty-one patients developed SSI within the first 30 days after surgery. The risk of developing at least 1 SSI was 8.8 (107/1222) per 100 patients, and the cumulative incidence of SSI was 10.3 episodes per 100 transplanted patients (95% CI, 8.4–12.2%). The predominant infection sites were incisional SSI (42%) and peritonitis (39%), with intra-abdominal abscess (16%) and hepatic abscess (10%) occurring less frequently (Table 1). The SSI rate in Child-Pugh class A, B, and C patients showed a linear trend, with SSI rates of 9.8%, 12.8%, and 14.3%, respectively, but they did not reach statistical association (chi-square for trend, 2.44; P = 0.119).
Case fatality risk ranged from 8.3% (95% CI, 0.2%–38.5%) for intra-abdominal abscess to 16.1% (95% CI, 5.5%–33.7%) for peritonitis, with no significant differences by SSI type.
Time of Occurrence
Seventy-six percent of the SSIs occurred during the first 4 weeks after surgery. The incidence declined progressively after the second postoperative week. Seventy-six percent (40/53) of the incisional SSIs occurred within 3 weeks after surgery. The timing of the organ/space SSIs was slightly delayed in comparison to that of incisional SSIs (Fig. 1), organ/space infections being 43% and 75% of the infections that developed during and after the first 3 postoperative weeks, respectively (chi square, 13.08; P < 0.001).
One hundred twenty-four organisms were recovered from 85 episodes of SSI, with 21 (17%) of the episodes being polymicrobial infections. Thirty-eight percent of pathogens recovered were gram-positive cocci, 52% were gram-negative rods, and 8% were Candida spp. Enterococcus spp., Escherichia coli, Staphylococcus aureus, and Acinetobacter baumannii were the most frequent pathogens isolated (Table 2). No differences in the site frequencies of microorganisms were found for the majority of pathogens, with the exception of S. aureus, which was more frequent in incisional SSIs (28% versus 8%), and Enterococcus spp, which was more frequent in the deepest infections (28% versus 20%; Table 2).
Table 2. Organisms Recovered from Incisional and Organ/Space Surgical Site Infections (SSIs) After Liver Transplantation
Eighty-one patients who developed SSI within 30 days after surgery were included and compared by univariate analysis. Patients who had received a previous liver or kidney transplant were found to have a significantly greater risk of SSI. Surgical reconstruction by choledochojejunostomy was also found to be associated with a greater risk of SSI than choledochocholedochostomy anastomosis. Patients receiving transplants in low-volume LT hospitals, which were defined as hospitals in which fewer than 50 LTs were performed per year, also had an increased risk of SSI (Table 3).
Table 3. Risk of Surgical Site Infection (SSI) by Potential Risk Factors: Univariate Analysis
By chi-square test except for age and duration of surgery (Student t test).
Age (mean ± standard deviation)
52.8 ± 11.6
Patients without SSI
53.5 ± 10.2
Type of surgery
Chronic renal failure
Duration of surgery (mean ± standard deviation)
Patients with SSI
6.0 ± 2.2
Patients without SSI
5.8 ± 2.2
Amount of blood transfused
>4 red blood cell units
≤4 red blood cell units
Number of liver transplants in hospital
Less than 50 per year
50 or more per year
The most frequently used immunosuppressive regimens were double therapy with steroids and calcineurin inhibitors: tacrolimus (44%; 95% CI, 41%–47%) or cyclosporine (18%; 95% CI, 15%–20%). The use of mycophenolate mofetil in triple-association prophylaxis was associated with a higher risk of SSI in comparison with the double regimen (7.9% versus 4.7% and 10.7% versus 4.8%; crude odds ratio, 2.1; 95% CI, 1.2–3.5).
More than 8 types of antibiotics or antibiotic combinations were used for surgical prophylaxis, with the most common being amoxicillin-clavulanate (19%) and combinations of glycopeptide and antipseudomonal penicillin (16%) and glycopeptide and aztreonam (10%). The risk of SSI associated with different antibiotics ranged from 1.7% for the combination glycopeptide/aztreonam to 17.1% for cefazolin alone (odds ratio, 12.3%; CI, 2.6–114; Table 3). No other variables showed statistical association in the univariate analysis.
After we controlled by Child-Pugh class and center, antibiotic prophylaxis no longer remained associated with SSI risk. Nevertheless, choledochojejunal reconstruction, previous solid organ transplant, and RBC transfusion greater than 4 units were found to be independently associated with the development of SSI by the unconditional multiple logistic regression model (Table 4).
Table 4. Risk Factors for Surgical Site Infection: Multivariate Analysis Controlling by Child-Pugh Class and Center Effect
95% Confidence Interval
Patients who developed incisional, peritoneal, or intra-abdominal abscess infection within the first month after transplantation were included as cases.
Previous liver or kidney transplant
Red blood cell transfusion > 4 units
SSIs occurred in 8.8% of patients, and this figure is within the range of those previously reported but much lower than the risk found by Iinuma et al.10 in patients receiving living-donor liver transplant. Some of the differences found among studies could be due to the use of different SSI definitions, the duration of SSI surveillance, and technical improvements in recent years.10–14 Most SSIs occurred during the first 2-3 weeks, as noted in previous reports, and could be related to contamination of the abdominal cavity during the transplant procedure.12–16 Although SSI infections carried a high mortality in the 1980s,17 in the present series, fatality related to this complication occurred only in 11 patients (10%). Rapid advances in surgical techniques and antimicrobial therapy during recent years could have induced improved prognoses in these patients.18
Most SSIs were caused by gram-negative aerobic bacteria, which are inhabitants of the digestive tract, as has been reported in previous studies.2, 14 Fungal infection by Candida spp. occurred in 10 cases (8%). Invasive candidiasis after LT usually presents as an intra-abdominal abscess, peritonitis, catheter-related fungemia, or fungemia of unknown origin.10–12 A number of approaches toward antifungal prophylaxis have been proposed.4 The high incidence of candidiasis in this study is remarkable because many of the participant institutions used fluconazole (100 mg/day) prophylaxis during the period of hospital admission.16, 18 However, a recent meta-analysis of fungal prophylaxis studies in LT showed that a higher dose of fluconazole (400 mg/day) and a more prolonged duration of prophylaxis are associated with a lower incidence of invasive fugal infection.19, 20 Forty percent of isolated species were Candida glabrata, which could have emerged after fluconazole usage.
No prophylactic antibiotic regimen has been definitively established for LT in the medical literature.13 This could be due to the lack of randomized controlled studies comparing the efficacy of different antimicrobial prophylactic regimens. In fact, second-generation or third-generation cephalosporin alone is used in many institutions4, 19–21 and also in hospitals with a high frequency of SSI due to enterococci.10
An antibiotic prophylaxis regimen, specifically the use of cefazolin, was found to be associated with the risk of SSI in the univariate analysis; however, this association did not remain after we controlled by center effect and Child-Pugh class. The use of cefazolin was found to be associated with a specific center with high rates of SSI, so our study could not identify whether the use of cefazolin itself or some other unmeasured factor associated with that center was responsible for the high increase of SSI. Nevertheless, this study suggests that the use of cephalosporin as a unique prophylactic agent is inadvisable and highlights the importance of enterococci as the main microbiological agents. Furthermore, one study found a significantly higher incidence of enterococcal infections under cefotetan prophylaxis than under ampicillin-sulbactam administration.10 The rationale for this fact could be a difference in the pathogenicity of enterococci in LT compared to other abdominal surgical procedures. In addition, isolation of enterococci in intraoperative cultures in LT patients has been demonstrated to increase the risk of early postoperative enterococcal infection.12
Amoxicillin-clavulanate, or the combination of a third-generation cephalosporin plus amoxicillin, could be a reasonable antibiotic prophylaxis regimen in LT in institutions with a low incidence of penicillin-resistant and vancomycin-resistant enterococci.16, 21–24 In hospitals with a high incidence of methicillin-resistant S. aureus postoperative infections, vancomycin could be included in the prophylactic antibiotic regimen.25 However, the normally good prognosis of patients with SSI must be pointed out. Stool and urine cultures, performed prior to LT, may be useful in detecting multiresistant enterococci, which have become a nosocomial problem in transplant units during recent years and could justify an appropriate prophylactic regimen.
The greatest risk factor for SSI was choledochojejunal reconstruction. This risk factor has frequently been identified in previous studies.5, 12, 19, 21, 26 Opening the jejunum increased enteric organism contamination of the surgical field with a resultant increased risk of infection.12, 21 In fact, prolonged prophylaxis (more than 2 days) has been recommended in cases involving jejunostomy.12 Perioperative blood transfusion has also been considered a risk factor for bacterial infections but was weakly related to SSI in this study.5, 14, 27
The use of mycophenolate mofetil was associated with SSI at univariate analysis. This medication has been considered a risk factor for wound infections in kidney transplantation because it is a powerful immunosuppressive drug that inhibits the proliferation of T and B cells.28, 29 Nevertheless, no increased risk of bacterial infection in LT patients treated with mycophenolate mofetil has been observed in other studies.30, 31
It is also remarkable that no other known SSI risk factors, such as duration of the operation and repeat surgery,7, 11, 32 were detected in our study. Other risk factors, such as retransplantation, were detected only in the univariate analysis.5, 14
This study shows several limitations. No formal validation was performed on the assessment of infections, so differences in SSI between centers could be due to differences in the sensitivity and specificity of the criteria used. In addition, the number and timing of repeat doses during surgery were not recorded and therefore were unavailable for the analysis. Moreover, some important risk factors, such as ABO incompatibility, treatment of early rejection, use of monoclonal antibody, immune modulators, and evidence of infection prior to SSI, were not assessed, so their importance was not evaluated and could constitute a difficulty in its generalizability.
Despite these limitations, these results provide some insight into the risk of SSI in relation to previous transplantation, choledochojejunal reconstruction, and RBC transfusion, which could motivate new studies to aid in the understanding of the pathogenesis of SSI in LT.
The authors thank Martin Hadley-Adams for his assistance with the English language and preparation of the manuscript.