Surgical site infections in liver transplant recipients in the model for end-stage liver disease era: An analysis of the epidemiology, risk factors, and outcomes


  • The authors declare no conflicts of interest.

  • This study received no financial support.

Address reprint requests to Maristela Pinheiro Freire, M.Sc., Infection Control Team, Clinical Hospital, University of São Paulo School of Medicine, Rua Dr. Eneas de Carvalho Aguiar, 255, São Paulo, SP, 05403-900, Brazil. Telephone: +55-11-2661-6444; FAX: +55-11-2661-6444; E-mail:


In recipients of liver transplantation (LT), surgical site infection (SSIs) are among the most common types of infection occurring in the first 60 days after LT. In 2007, the Model for End-Stage Liver Disease (MELD) scoring system was adopted as the basis for prioritizing organ allocation. Patients with higher MELD scores are at higher risk for developing SSIs as well as other health care–associated infections. However, there have been no studies comparing the incidence of SSIs in the pre-MELD era with the incidence in the period since its adoption. Therefore, the objectives of this study were to evaluate the incidence, etiology, epidemiology, and outcomes of post-LT SSIs in those 2 periods and to identify risk factors for SSIs. We evaluated all patients who underwent LT over a 10-year period (2002-2011). SSI cases were identified through active surveillance. The primary outcome measure was an SSI during the first 60 days after LT. Risk factors were analyzed via logistic regression, and 60-day survival rates were evaluated via Cox regression. We evaluated 543 patients who underwent LT 597 times. The SSI rates in the 2002-2006 and 2007-2011 periods were 30% and 24%, respectively (P = 0.21). We identified the following risk factors for SSIs: retransplantation, the transfusion of more than 2 U of blood during LT, dialysis, cold ischemia for >400 minutes, and a cytomegalovirus infection. The overall 60-day survival rate was 79%. Risk factors for 60-day mortality were retransplantation, dialysis, and a longer surgical time. The use of the MELD score modified the incidence and epidemiology of SSIs only during the first year after its adoption. Risks for SSIs were related more to intraoperative conditions and intercurrences after LT than to a patient's status before LT. Liver Transpl 19:1011–1019, 2013. © 2013 AASLD.


American Society of Anesthesiologists




gram-negative bacteria


Gram positive cocci


health care-associated infection


hepatitis B virus


hepatitis C virus


intensive care unit


liver transplantation




Model for End-Stage Liver Disease


methicillin-resistant Staphylococcus aureus


surgical site infection


vancomycin-resistant enterococci

Infections at the surgical site, known as surgical site infections (SSIs), are among the most common health care–associated infections (HAIs). The occurrence of SSIs has been associated with longer hospital stays, higher mortality, and increased health care costs.[1] Among patients undergoing liver transplantation (LT), SSIs are among the most common types of infection occurring within the first 60 days after the procedure.[2] In such patients, SSIs have been associated with an increased risk of intensive care unit (ICU) admission, higher rates of rehospitalization, an increased risk of graft loss, and higher hospitalization costs.[3-6] For LT recipients, the following risk factors for SSIs have been identified4-9: a surgeon's inexperience, biliary leaks, longer surgical times, human leukocyte antigen mismatching, reoperation, previous liver or kidney transplantation, choledochojejunal or hepaticojejunal reconstruction, the transfusion of more than 4 U of blood during the surgical procedure, preoperative antibiotic use, intraoperative vasopressor use, combined liver-kidney transplantation, the preoperative use of mechanical ventilation, and severe hyperglycemia during the surgical procedure.

In the last 10 years, there have been advances in surgical techniques, improvements in the postoperative management of LT recipients, and changes in immunosuppression therapy. Despite these changes, the reported rates of SSIs have remained unchanged.[6, 7, 10] Another important change was the adoption of the Model for End-Stage Liver Disease (MELD) scoring system as the basis for prioritizing organ allocation. It has been reported that patients now undergo transplantation with a greater degree of renal impairment and with high MELD scores, which are 2 major risk factors for early postoperative infections.[11-14] However, there have been no studies focusing on the variations in SSI incidence between the pre-MELD era and the period since its adoption or on possible changes in the epidemiology of these infections. There is also a lack of data on common risk factors for SSIs in both periods. Therefore, studies of the incidence of SSIs and risk factor analyses of SSIs over a long period of time could clarify whether the adoption of the MELD score has increased the risk of SSIs and changed the microbiology profile of these infections. Therefore, the objectives of this study were to compare the 2 periods in terms of the incidence, etiology, epidemiology, and outcomes of post-LT SSIs and to identify the risk factors associated with this type of infection.


We evaluated all patients who underwent LT between January 2002 and December 2011 at the Clinical Hospital of the University of São Paulo School of Medicine. This study protocol was approved by institutional ethics committee. We excluded patients who died within the first 72 hours after transplantation. Follow-up started at hospital admission and continued until the end of posttransplant month 2. Infections were identified through active surveillance in the LT ward and through reviews of outpatient evaluation records. The criteria used for identifying and classifying HAIs were those outlined by the US National Healthcare Safety Network.[15] The criteria for cytomegalovirus (CMV) infections and disease were those defined by Ljungman et al.[16] We defined the category of multidrug-resistant (MDR) bacteria to include the following: carbapenem-resistant, nonfermentative, gram-negative bacilli; extended-spectrum, beta-lactamase–producing Enterobacteriaceae; enterobacteria resistant to at least 1 carbapenem; methicillin-resistant Staphylococcus aureus (MRSA); and vancomycin-resistant enterococci (VRE).

The standard surgical prophylaxis in use during the study period was ampicillin with cefotaxime for 48 hours. The prophylaxis protocol was modified for cases in which patients underwent LT during treatment for an infection and for cases in which the donor had a suspected or confirmed infection. In such cases, the antibiotic used for prophylaxis was the same as that used for the treatment of the infection in question. CMV prophylaxis was administered to all patients defined to be at high risk for CMV infections: recipients with a negative serology (immunoglobulin G) for CMV and those on immunosuppressive therapy with anti-lymphocyte immunoglobulin. Trimethoprim-sulfamethoxazole as prophylaxis for Pneumocystis jiroveci was administered to all patients and was begun in the first week after LT as soon as patients were able to receive oral medications. The standard immunosuppression was tacrolimus plus a corticoid; when combined immunosuppression therapy was necessary, mycophenolate mofetil was most frequently used with the standard immunosuppression.

Infection rates were calculated from the number of SSIs with respect to the total number of transplants performed during the study period. In our analysis of risk factors for infection and mortality, patients were included only once, regardless of the number of transplants that they underwent. We evaluated the following variables related to the surgical procedure: cold ischemia time, total ischemia time, number of units of blood transfused during LT, reoperation, retransplantation, level of experience of the lead surgeon, American Society of Anesthesiologists (ASA) physical status classification, surgical time, type of biliary reconstruction, donor type (living or cadaveric), extensiveness of the procedure (liver-only or combined liver-kidney transplantation), and type of surgical prophylaxis. We also evaluated variables related to the organ recipient: age; sex; presence or absence of fulminant hepatitis, hepatocellular carcinoma, and other underlying diseases; MELD score; and pretransplant serum creatinine level. In addition, we evaluated the following variables related to hospitalization: length of the hospital stay before LT, need or no need for dialysis, duration of an abdominal drain, occurrence or nonoccurrence of acute cellular rejection, occurrence or nonoccurrence of a CMV infection, occurrence or nonoccurrence of an infection within the week before transplantation, and type of immunosuppression therapy (monotherapy versus combined therapy). In the survival analysis, we also analyzed as a variable LT in the MELD era, which means LT performed in or after 2007 (when the MELD scoring system was adopted as the basis for prioritizing organ allocation).

The main outcome measure was an SSI during the first 60 days after transplantation. For dichotomous variables, we performed a univariate analysis with the chi-square test or Fisher's exact test as appropriate. For continuous variables, we used the Mann-Whitney test. Continuous variables were transformed into dichotomous variables through a cluster analysis. Multivariate analysis was performed via stepwise binary logistic regression. The criterion for inclusion in the multivariate analysis was P < 0.2 in the univariate analysis. Two-month survival was evaluated via Cox regression. To calculate P values, we used the log-rank test. All statistical analyses were performed with Statistical Package for the Social Sciences 15.0 (SPSS, Inc., Chicago, IL).


In all, 561 patients underwent LT in the period, and 18 were excluded for death within 72 hours after LT. We evaluated 543 patients who underwent LT 597 times. Three hundred fifty of the 543 patients (64%) were male. The median age was 51 years. Table 1 shows the pretransplant diagnoses of the patients. The most common underlying diagnosis was cirrhosis due to hepatitis C virus (HCV; 37.0%), and the prevalence of hepatocellular carcinoma was 29.8%.

Table 1. Diagnoses of 543 Patients Undergoing LT
HCV infection16130
Alcoholic cirrhosis6712
Fulminant hepatitis5610
Cryptogenic cirrhosis5310
HBV infection418
HCV infection with alcoholic cirrhosis295
Autoimmune hepatitis275
Biliary primary cirrhosis153
Familial amyloid polyneuropathy132
HBV and HCV infections112
Primary sclerosing cholangitis81
Budd-Chiari syndrome71
Nonalcoholic steatohepatitis71
Biliary atresia61
HBV infection with alcoholic cirrhosis51
Nonhepatocellular carcinoma51
Wilson's disease41
Late graft dysfunction31
Alpha-1-antitrypsin deficiency31

Twenty-one of the 543 evaluated patients (3.9%) underwent liver-kidney transplantation. Forty-two of the 597 procedures (7.0%) were living donor transplants. In the first 60 days after LT, 50 patients (9.2%) required a second transplant, and 2 (0.4%) underwent 2 retransplants. In all, 142 patients underwent LT in the pre-MELD era. The SSI incidence rates were 30% and 24% in the 2002-2006 and 2007-2011 periods, respectively (P = 0.21).

HAIs were identified for 286 of the 597 LT procedures (47.9%), and SSIs occurred in 141 (23.6%) during the first 60 days after transplantation. The median time between transplantation and the diagnosis of SSIs was 11 days. In 117 of the 141 cases (83.0%), the SSIs were classified as organ-space SSIs. Surgical intervention after the diagnosis of SSIs was needed for 35% of the cases (n = 50). The median time for the duration of antibiotic therapy for SSIs was 19 days. The SSI incidence rate declined between 2002 and 2006 (P = 0.07), spiked in 2007 when the MELD score was adopted, and stabilized thereafter (Fig. 1).

Figure 1.

SSI rate.

The infectious agent was identified for 121 of the SSIs (85.8%). In 52 of the 141 cases (36.9%), the agent was isolated from a blood sample. In all, 174 microorganisms were associated with the SSIs, and 44 of the infections (31.2%) were polymicrobial. We identified a high incidence of gram-negative bacilli, a large proportion of which were MDR (Table 2). Fourteen of the 26 identified Klebsiella pneumoniae cases (54%) were carbapenem-resistant, and 8 were extended-spectrum and beta-lactamase–producing. The proportions of carbapenem-resistant Enterobacter aerogenes and Enterobacter cloacae cases were 45% (n = 5) and 11% (n = 1), respectively. Among the gram-positive bacteria, we observed a significant decrease in MRSA (P < 0.001) and an increase in the incidence of VRE species (P < 0.001; Fig. 2).

Table 2. Microorganisms Isolated From 141 SSIs (n = 174)
Microorganismn%MDR (%)
A. baumannii311894
K. pneumoniae261585
Enterococcus faecium231396
S. aureus211295
E. aerogenes11645
E. cloacae9511
Enterococcus faecalis950
Staphylococcus epidermidis850
Non-albicans Candida species85
Candida albicans63
Escherichia coli4275
Staphylococcus haemolyticus320
Pseudomonas aeruginosa3233
Morganella morganii320
Elizabethkingia meningoseptica21
Serratia marcescens210
Hansenula anomala11
Streptococcus sanguinis110
Bacteroides ovatus110
Alcaligenes faecalis110
Rhizopus species11
Figure 2.

Incidence of VRE species and MRSA among SSIs.

For the study period as a whole, the mean prevalence of fungal infections was 9%, although there was a transitory increase to 25% in 2007 (Fig. 3). Infections with Candida species were predominantly caused by non-albicans Candida species, the most common being Candida glabrata, Candida tropicalis, and Candida parapsilosis, each of which was identified in 2 cases.

Figure 3.

Distribution of microorganisms.

In all, 167 patients received surgical antibiotic prophylaxis other than the standard. Fifty-two of these patients had SSIs, and the proportion of MDR agents was 71% (n = 37); the proportion of MDR agents among patients with SSIs who received standard antibiotic prophylaxis was 55% (49/89; P = 0.06).

The multivariate analysis identified the following risk factors for SSIs: retransplantation, the transfusion of more than 2 U of blood during LT, dialysis, cold ischemia for >400 minutes, and CMV infection. A surgical time longer than the 75th percentile was marginally significant (Table 3).

Table 3. Risk Factors for SSIs After LT in 543 Patients
   Univariate AnalysisMultivariate Analysis
VariableSSI [n (%)]aNo Infection [n (%)]bRelative RiskP ValueOdds RatioConfidence IntervalP Value
  1. a

    n = 141.

  2. b

    n = 402.

  3. c

    Monotherapy versus combined therapy.

Retransplantation40 (28)12 (3)1.35<0.0018.343.99-17.43<0.001
Intraoperative blood transfusion > 2 IU82 (58)143 (36)1.28<0.0012.031.27-3.250.003
Dialysis89 (63)132 (33)1.45<0.0012.361.47-3.79<0.001
Cold ischemia time > 400 minutes95 (67)202 (50)1.520.0012.281.41-3.680.001
CMV Infection40 (28)48 (12)1.230.0012.071.17-3.680.01
Surgical time > 75th percentile60 (43)120 (30)1.230.0051.570.98-2.520.06
Hepatocellular carcinoma26 (18)136 (34)0.810.001   
ASA physical status classification > 386 (61)196 (49)1.10.41   
Acute cellular rejection27 (19)75 (19)1.010.90   
Abdominal drain for >5 days71 (50)152 (38)1.250.009   
HCV infection before LT41 (10)161 (40)0.890.02   
MELD score > 2191 (65)216 (54)1.310.03   
Serum creatinine > 2 mg/dL40 (28)82 (20)1.110.05   
Age > 60 years23 (16)95 (24)0.910.07   
Reoperation26 (18)56 (14)1.060.20   
Living donor14 (10)28 (7)1.030.26   
Fulminant hepatitis18 (13)38 (9)1.040.27   
Biliary reconstruction choledocho-choledochostomy reconstruction131 (93)383 (95)1.170.28   
Female sex55 (39)138 (34)1.10.32   
Senior surgeon88 (62)269 (67)0.860.33   
Vancomycin prophylaxis38 (27)94 (23)1.050.40   
Combined liver-kidney transplant7 (5)14 (3)1.020.43   
Infection from previous LT44 (31)41 (23)1.120.43   
Immunosuppression therapyc45 (32)136 (34)0.970.68   
Previous hospital stay for transplantation > 5 days45 (32)94 (23)1.120.46   
Standard prophylaxis89 (63)287 (71)0.780.67   

In the univariate analysis, the following risk factors for SSIs by MDR agents were identified: hepatocellular carcinoma before LT, fulminant hepatitis, retransplantation, reoperation, dialysis, CMV infection, pre-LT infection, use of nonstandard antibiotic prophylaxis, a pre-LT hospital stay > 5 days, MELD score, and a surgical time longer than the 75th percentile. However, the only variables that remained significant in the multivariate analysis were retransplantation (P < 0.001, odds ration − 8.34), the amount of blood transfused during LT (P = 0.003, odds ration = 2.03), dialysis after LT (P < 0.001, odds ratio 2.36), cold ischemia time duration (P = 0.001, odds ration 2.28), CMV infection (P = 0.01, odds ration 2.07).

The 60-day survival rate was 79%, and the median length of the post-LT hospital stay was 19 days. The multivariate analysis identified retransplantation, dialysis, and a surgical time longer than the 75th percentile as risk factors for 60-day mortality (Table 4).

Table 4. Risk Factors for 60-Day Mortality After LT
VariableUnivariate Analysis: P ValueMultivariate Analysis
P ValueHRConfidence Interval
Post-LT HAI<0.001   
SSI with MDR agent<0.001   
Pre-LT serum creatinine > 2 mg/dL0.001   
LT surgical time > 75th percentile0.0040.021.011.00-1.02
Infection from previous LT0.004   
MELD score > 210.008   
ASA physical status classification > 30.01   
Length of hospital stay for previous LT0.02   
Fulminant hepatitis0.03   
LT in MELD era0.10   
HCV infection before LT0.12   
Intraoperative blood transfusion > 2 IU0.16   
CMV infection0.25   
Age > 60 years0.26   
Combined liver-kidney transplant0.28   
Living donor0.41   
Female sex0.45   
Hepatocellular carcinoma0.47   
Acute cellular rejection0.55   
Cold ischemia time > 400 minutes0.99   


The reported incidence of SSIs ranges from 9% to 38%, with more recent studies typically reporting rates less than 20%.[3, 4, 6, 7, 9, 10] In our study, the rates ranged from 14% to 39% over the course of the study period. As previously mentioned, we observed a decline in the incidence of SSIs before 2007 (when the MELD score began to be used as the allocation score for patients on the LT waiting list). At the beginning of the MELD era, a large number of transplants were performed in patients with advanced hepatic impairment. Although there was an increase in the SSI rate immediately after the MELD score was adopted, a new downward trend was observed in 2008. Thereafter, the profile of patients selected to undergo LT changed, with those selected for the procedure having MELD scores that were lower than those observed for patients selected at the beginning of the MELD era.

Patients undergoing LT in the MELD era have been described as having higher MELD scores, more often having a serum creatinine level > 2 mg/dL, and more often being on dialysis, and all of these are factors that increase the risk of infection. However, with the new allocation criterion (the MELD score), an increase in infections has not been associated with decreased survival during the first 60 days after LT.[13, 14]

Sun et al.[17] evaluated 240 LT procedures for which the patients had been selected with MELD criteria and compared them with 37 others for which the patients had been selected with pre-MELD criteria. The authors found that the incidence of SSIs was higher among the patients selected with the MELD criteria, although the difference was not statistically significant. In addition, the total number of major infections was not found to differ between the 2 periods.

One major factor that increased the risk of SSIs immediately after the allocation criteria for LT were changed was the fact that patients who had been hospitalized for long periods or had a history of multiple hospitalizations came to account for a large proportion of the patients who underwent LT. This implied greater exposure to antimicrobial agents and an increased risk of HAIs. Antibiotic use has been described as a risk factor for post-LT SSIs, possibly because it is a marker of patients who undergo the procedure while they are infected with or colonized by MDR bacteria.[4]

In late 2006 and in 2007, the proportion of post-LT fungal infections increased. The use of the MELD score has been implicated in increasing the risk of invasive fungal infections.[18] Possible explanations for this are, again, greater exposure to antimicrobial agents, the use of invasive devices (eg, urinary catheters, abdominal drains, central venous catheters, and mechanical ventilation) before LT, and the need for dialysis, all of which are known risk factors for fungal infections.[19-21] When Sun et al.17 compared the pre-MELD and MELD eras in terms of the infectious agents involved in the etiology of post-LT infections, they identified an increase in the incidence of invasive candidiasis.

Among the isolated microorganisms, there was a predominance of gram-negative bacilli; the most frequent agents were Acinetobacter baumannii and K. pneumoniae, both of which had a high proportion of MDR strains. Among the agents of early post-LT infections, A. baumannii has been described as an emerging threat. Its incidence ranges from 6% to 10% among total infections and is up to 13% among SSI agents.[22-24] Kim et al.[22] described 37 infectious episodes with A. baumannii in the early post-LT period, and they reported that the primary site of infections was intra-abdominal (the biliary tract in 57% of the cases and the peritoneal cavity in 16%). In SSIs, K. pneumoniae is often isolated. The incidence of infection with K. pneumoniae carbapenemase–producing K. pneumoniae is increasing in several countries.[25] Patel et al.[25] identified transplantation as a risk factor for such infections. In a study of infections in solid organ transplant recipients, SSIs accounted for 17% of all posttransplant infections with K. pneumoniae carbapenemase–producing K. pneumoniae.[26]

In the present study, we identified polymicrobial infections in 31% of the cases, and this rate is comparable to the rates of 29% to 33% reported in the literature.[3, 10] We observed a high proportion of intra-abdominal infections and a large number of cases of secondary bacteremia. Previous studies have described SSIs as severe infections with a high proportion of organ-space infections (73%-96% of cases).[3, 4, 9, 10]

The risk factors for SSIs identified in our study were related to the complexity of the procedure (an intraoperative transfusion of more than 2 U of blood, retransplantation, and cold ischemia for >400 minutes) and to the degree of patient immunosuppression (CMV infection and dialysis). Multiple transfusions during LT have been reported to be a risk factor not only for SSIs but also for most early post-LT infections.[9, 11, 27] Transfusion itself plays an immunosuppressive role and, in the case of SSIs, is also an indirect marker of intraoperative complications. Other studies of LT have employed a large number of transfused blood units as the cutoff point for SSI risk.[9, 11] The relatively low transfusion cutoff point identified in our study shows how sensitive this parameter is as a marker of surgical complications. Retransplantation has also been identified as a risk factor for early post-LT infections. In a review of 1222 patients who underwent LT in Spain, Asensio et al.[9] found that retransplantation increased the risk of infection by 2.6 times. In our study, we included only retransplants that occurred within 60 days after the first LT procedure, and this makes it likely that the affected individuals constituted a group of patients for whom surgical complications were more common, liver function was more severely impaired, ICU stays were longer, and dialysis was required by a greater proportion. As previously mentioned, all of these factors are known to increase the risk of infection.[11, 12]

To our knowledge, there have been no studies associating the cold ischemia time with the SSI risk. However, the cold ischemia time has been associated with post-LT survival and with the incidence of nonanastomotic biliary strictures.[28-30] In general, it can be assumed that a longer cold ischemia time is associated with slower post-LT recovery of the systemic inflammatory response and is indirectly associated with longer ICU stays, an increased use of invasive devices, and a delayed recovery of renal function.[29, 31, 32]

As previously mentioned, we identified dialysis and posttransplant CMV infections as risk factors for SSIs. Even though dialysis after LT is a well-known risk factor for early postoperative bacterial infections, there have been, to our knowledge, no studies in which a multivariate analysis has identified dialysis as a specific risk factor for post-LT SSIs.[12] Patients requiring dialysis are more often subjected to the use of invasive devices and are more often admitted to the ICU, so they have a higher incidence of bacterial infections.[12, 24, 33, 34] Renal failure itself plays an immunosuppressive role and can be an indirect marker of graft malfunction. Acute kidney injury can also be seen in LT recipients with severe preoperative hepatic impairment. Therefore, the incidence of SSIs would be expected to be high among patients with renal failure.

CMV infections have been shown to increase the post-LT rate of bacterial infections.[35, 36] That can be explained, in part, by the fact that CMV has an immunomodulatory effect.[37, 38] However, we hypothesized that the most plausible explanation for the association between SSIs and CMV infections is that a CMV infection is a marker of a high degree of immunosuppression in the early post-LT period.

We did not find SSIs to be a risk factor for 60-day mortality after LT. Although the association between SSIs and low graft survival is well established, there are conflicting data regarding the association between SSIs and patient survival.[3, 4, 6, 10] In a study of patients undergoing LT between 2000 and 2010, Cockbain et al.[27] identified an association between SSIs and reduced patient survival. However, various other studies designed to assess post-LT SSIs found no such association.

In conclusion, the use of the MELD score as the allocation criterion for patients on the LT waiting list temporarily modified the incidence and epidemiology of SSIs, both of which returned to their previous profiles after 1 year under the new system. Risk factors for SSIs after LT were related more to surgical procedure conditions and intercurrences after transplantation than to a patient's status before LT.