Figure. Relationship of model for end-stage liver disease (MELD) score to 30-day mortality in elective umbilical hernia repair (n = 241).
Cho SW, Bhayani N, Newell P, Cassera MA, Hammill CW, Wolf RF, Hansen PD. Umbilical Hernia Repair in Patients With Signs of Portal HypertensionSurgical Outcome and Predictors of Mortality. Arch Surg. 2012;147(9):864-869. doi:10.1001/archsurg.2012.1663
Author Affiliations: Hepatobiliary and Pancreatic Surgery Program, Providence Portland Medical Center, Portland, Oregon.
Objectives To compare the outcomes of umbilical hernia repair in patients with and without signs of portal hypertension, such as esophageal varices or ascites; to assess the effect of emergency surgery on complication rates; and to identify predictors of postoperative mortality.
Design Database search from January 1, 2005, through December 31, 2009.
Setting North American hospitals participating in the American College of Surgeons National Surgical Quality Improvement Program initiative.
Patients We studied patients who underwent umbilical hernia repair. Those with congestive heart failure, disseminated malignant tumor, or chronic renal failure while undergoing dialysis were excluded.
Main Outcome Measures Preoperative variables and perioperative course were analyzed. Main outcome measures were morbidity and mortality after umbilical hernia repair.
Results A total of 390 patients with ascites and/or esophageal varices formed the study group, and the remaining 22 952 patients formed the control group. The overall morbidity and mortality rates for the study group were 13.1% and 5.1%, whereas these rates were 3.9% and 0.1% for the control group, respectively (P < .001). For the study group, the mortality after elective repair among patients with a model for end-stage liver disease (MELD) score greater than 15 was 11.1% compared with 1.3% in patients with a MELD score of 15 or less. The patients with ascites and/or esophageal varices underwent emergency surgery more frequently than the control group (37.7% vs 4.9%; P < .001). Emergency surgery for the study group was associated with a higher morbidity than elective surgery (20.8% vs 8.3%; P < .001) but not a significantly higher mortality (7.4% vs 3.7%; P = .11). However, logistic regression analysis showed that age older than 65 years, MELD score higher than 15, albumin level less than 3.0 g/dL (to convert to grams per liter, multiply by 10), and sepsis at presentation were more predictive of postoperative mortality.
Conclusions Umbilical hernia repair in the presence of ascites and/or esophageal varices is associated with significant postoperative complication rates. Emergency surgery is associated with higher morbidity rates but not significantly higher mortality rates. Elective repair of umbilical hernia should be avoided for those with adverse predictors, such as age older than 65 years, MELD score higher than 15, and albumin level less than 3.0 g/dL.
Esophageal varices and ascites are common manifestations of portal hypertension secondary to liver cirrhosis. Approximately 20% of patients with liver cirrhosis develop umbilical hernia.1 Strangulation, incarceration, and hernia skin breakdown with leaking ascites are well-known complications. However, umbilical hernia repair in the presence of liver cirrhosis has largely been avoided in view of high operative risk, including risk of postoperative liver failure. The watchful waiting approach, on the other hand, may result in emergency surgery. There is therefore a need for a risk-stratifying strategy by which patients with acceptable operative risk can be identified and undergo umbilical hernia repair in an elective setting.
Studies2 on this subject have mostly been retrospective case series and are based on single institutional practices. These studies have the inherent limitation of a small number of patients and reflect practice patterns of individual institutions. Although abdominal operations in patients with liver cirrhosis are known to be higher risk than those in patients without liver cirrhosis, complication rates for patients with cirrhosis undergoing hernia repair have not been clearly delineated separately for elective and emergency settings. This information may be important when counseling these patients regarding elective umbilical hernia repair. Furthermore, there is a lack of national outcome studies that specifically examine what factors are associated with adverse outcome after umbilical hernia repair in patients with signs of portal hypertension, such as ascites and esophageal varices.
This study was undertaken to quantify operative risks of umbilical hernia repair for the patients with signs of portal hypertension, such as esophageal varices and/or ascites, and to compare them with those of the general population. The effect of emergency status of umbilical hernia repair on surgical outcome was also assessed. In addition, we sought to identify predictors of postoperative mortality.
The American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) is a nationally validated database in which patient demographic characteristics, preoperative variables, perioperative course, and surgical outcomes are captured and recorded prospectively by trained clinical reviewers. We used a deidentified data set available to participating hospitals known as the ACS-NSQIP Participant Use Data File.
Current Procedural Terminology codes 49585 and 49587 were used to identify patients who underwent umbilical hernia repair from January 1, 2005, through December 31, 2009. Patients with disseminated malignant tumor, congestive heart failure, and renal failure requiring dialysis were excluded to ensure that ascites was not caused by conditions other than portal hypertension from liver cirrhosis. The inclusion criteria for the study group were the presence of esophageal varices or ascites within 30 days of surgery. Those patients without ascites and esophageal varices formed the control group. Demographic data, preoperative variables, perioperative course, and postoperative complications were analyzed.
We first examined demographic details and preoperative variables of the patients who underwent umbilical hernia repair. Overall morbidity and mortality rates after umbilical hernia repair were determined for those with ascites and/or esophageal varices and those without to assess the effect of ascites and esophageal varices on the surgical outcome.
Next, morbidity and mortality rates were compared between elective and emergency operations for the main study group to determine the effect of emergency surgery on complication rates. In addition, specific details of postoperative complications were assessed to determine whether some complications were more common in emergency situations, as opposed to elective situations.
The χ2 test was used for nonparametric variables. The Fisher exact test was used for variables with numbers less than 5. Predictors of postoperative mortality were identified. Univariate analysis was performed for preoperative variables to identify predictors of mortality within 30 days of surgery. Variables with P < .10 were included in the logistic regression analysis for predictors of postoperative mortality. P < .05 was considered significant.
During the study period, the ACS-NSQIP database contained 23 667 cases of umbilical hernia repair, 325 of which met the exclusion criteria and were excluded from the study. Therefore, 390 patients with ascites and/or esophageal varices were included in the final study group, and the remaining 22 952 patients formed the control group. In the study group, 81 patients had both esophageal varices and ascites, 35 patients had esophageal varices only, and 274 patients had ascites only. Demographic characteristics and preoperative variables are listed in Table 1. Compared with the general population undergoing umbilical hernia repair, the patients with ascites and/or esophageal varices were more likely to have comorbidities, such as older age, smoking history, chronic obstructive pulmonary disease, functional dependence, coronary artery disease, diabetes mellitus, hypertension, corticosteroid use, and higher American Society of Anesthesiologists class. The mean model for end-stage liver disease (MELD) score was significantly higher in the study group than in the control group (13 vs 8; P < .001). In addition, the patients in the study group were more likely than those in the control group to have a lower sodium level, lower platelet count, lower albumin level, higher creatinine level, higher bilirubin level, and higher international normalized ratio. Furthermore, patients with ascites and/or esophageal varices required emergency surgery for umbilical hernia more often than those without ascites or esophageal varices (37.7% vs 4.9%; P < .001). The presence of ascites and/or esophageal varices was also associated with longer operative time, longer hospital stay, and more frequent requirement for blood transfusion (Table 2).
The overall morbidity and mortality rates after umbilical hernia repair for the main study group were 13.1% and 5.1%, respectively. This finding contrasts to 3.9% morbidity and 0.1% mortality in the control group (P < .001).
For the study group, emergency umbilical hernia repair was associated with higher morbidity than the elective surgery (20.8% vs 8.3%; P < .001) but not significantly higher mortality (7.4% vs 3.7%; P = .11). In contrast, for the control group, emergency surgery was clearly associated with higher mortality than the elective surgery (1.2% vs 0.1%; P < .001).
Details of complications for elective and emergency cases for the main study group are given in Table 3. Emergency umbilical hernia repair was associated with higher risk of failure to wean ventilation (P = .001), sepsis (P = .001), and return to the operating room (P = .04).
We then performed statistical analysis to identify predictors of adverse outcomes after umbilical hernia repair in patients with ascites and/or esophageal varices. Univariate analysis revealed that a MELD score higher than 15 (P = .002) and sepsis at presentation (P < .001) were predictive of postoperative mortality (Table 4). Mortality after elective umbilical hernia repair among patients with a MELD score higher than 15 was 11.1% compared with 1.3% in patients with a MELD score of 15 or less. When variables with P < .10 on univariate analysis were included in the logistic regression analysis, the following 4 risk factors were identified: age older than 65 years (P = .002; odds ratio [OR], 2.5; 95% CI, 1.0-6.3), MELD score higher than 15 (P < .001; OR, 5.6; 95% CI, 1.9-16.7), preoperative sepsis (P < .001; OR, 10.9; 95% CI, 3.6-32.6), and albumin level less than 3.0 g/dL (to convert to grams per liter, multiply by 10) (P = .02; OR, 3.0; 95% CI, 1.1-8.3) (Table 5).
Umbilical hernia in the presence of liver cirrhosis is a common clinical dilemma faced by surgeons. Umbilical hernia is in large part due to the increased intra-abdominal pressure in the presence of ascites, which leads to development of hernia through the umbilical fascial defect. Commonly accepted indications for umbilical hernia repair are symptoms and development of secondary complications. One crucial variable in deciding for or against recommending umbilical hernia repair is surgical risk, which is known to be significantly higher in patients with evidence of portal hypertension.1 The overall operative mortality in this ACS-NSQIP database after umbilical hernia repair in patients with ascites and/or esophageal varices was 5.1%, whereas in the control group, this rate was 0.1%. Our findings are consistent with the literature. A large Danish cohort study3 reported a 5.5% 30-day mortality among 201 patients with liver cirrhosis. Choi and others4 also found an overall mortality of 6.5% after umbilical hernia repair among 31 patients with liver cirrhosis.
These high mortality rates may explain the general tendency among surgeons to avoid elective umbilical hernia repair in patients with signs of portal hypertension, and it may have contributed to our findings that patients with ascites and/or esophageal varices underwent emergency umbilical hernia repair more frequently than did the control group (37.7% vs 4.9%). This result is similar to the findings of a retrospective study of patients at Veterans Affair hospitals by Gray et al,5 who reported that patients with liver cirrhosis were more likely to undergo emergency surgery than those without liver cirrhosis (26.0% vs 4.8%). Carbonell et al6 also noted that surgical mortality was 7-fold higher in emergency surgery for umbilical hernia repair for patients with liver cirrhosis. As a result, some authors advocate for elective repair of all umbilical hernias for patients with liver cirrhosis. For instance, Marsman et al7 concluded from a retrospective study of 34 patients with liver cirrhosis that conservative management of umbilical hernia was successful in only 23% of patients and that 46.1% of patients required emergency umbilical hernia repair. In contrast, complications occurred in 3 of 17 patients whose umbilical hernia was repaired electively. Eker et al8 also recommend elective repair of all umbilical hernias by indicating that elective umbilical hernia repair was performed safely in 30 patients with liver cirrhosis and ascites, with 7% morbidity and no postoperative mortality. However, these are small retrospective studies that may not have been large enough to capture the full extent of surgical risk in patients with liver cirrhosis undergoing umbilical hernia repair. In our study, we found that emergency umbilical hernia repair is associated with a higher morbidity rate than elective surgery. Specifically, failure to wean ventilation, postoperative sepsis, and return to the operating room were more common in the emergency setting for the patients with ascites and/or esophageal varices. However, we found that mortality rates did not reach a significant difference between elective and emergency operations in the study group. This result contrasts with the finding that for the general population, emergency surgery was clearly associated with higher mortality than the elective surgery (1.2% vs 0.1%). This finding suggests that for the high-risk group of patients, such as those with ascites and/or esophageal varices, variables other than the emergency status of the surgery may be more strongly associated with postoperative mortality.
We found that for the patients with ascites and/or esophageal varices, predictors of postoperative mortalities are MELD score higher than 15, age older than 65 years, albumin level less than 3.0 g/dL, and preoperative sepsis. It has been well documented in the literature that MELD score can predict postoperative outcome after abdominal surgery in patients with liver cirrhosis.9 Our study goes one step further in that we specifically quantify the operative risk, reporting that with a MELD score of 15 or lower, elective umbilical hernia repair can be undertaken with mortality rates of 1.3%. On the other hand, in patients with a MELD score greater than 15, elective umbilical hernia repair should be avoided because of the 11.1% mortality rate (Figure). Therefore, we recommend assessing MELD score, age, and albumin level of the patient with signs of portal hypertension, such as ascites and esophageal varices, who presents with umbilical hernia and recommend considering elective repair of umbilical hernia if those predictors of postoperative mortality are favorable. Careful calculation of the risk-benefit ratio in this high-risk patient group is required, as highlighted by the fact that the presence of preoperative sepsis was associated with an OR of postoperative mortality of 10.9, emphasizing the importance of preemptively repairing symptomatic umbilical hernia before complications develop in appropriately selected, medically optimized patients in an elective setting.
Optimizing the patients with liver cirrhosis before elective umbilical hernia repair is crucial to minimizing postoperative complications and reducing recurrence. Such optimization includes low salt intake, free water restriction, and use of diuretics, such as furosemide and spironolactone. Large-volume paracentesis and intravenous infusion of salt-poor albumin can help to control ascites. Another important consideration in treating patients with liver cirrhosis who develop ascites and present with umbilical hernia is to assess the patient for suitability for liver transplantation because new onset of ascites in patients with liver cirrhosis portends poor prognosis with a 50% mortality at 2 years.10 Therefore, we recommend referral of such patients to specialized centers with a multidisciplinary team who can assist in perioperative care of the patient undergoing umbilical hernia repair. They can also ideally participate in the discussion for suitability of the patient for liver transplantation. If the patient is a candidate for liver transplantation, umbilical hernia can be repaired at the time of transplantation surgery.
Risk of recurrence of umbilical hernia can be reduced by maximizing medical treatment of ascites. If ascites is refractory to medical management, several options are available, such as large-volume paracentesis and albumin replacement, the transjugular intrahepatic portosystemic shunt (TIPS) procedure, and temporary placement of peritoneal dialysis and peritoneovenous shunt. The TIPS procedure has been shown to successfully control ascites in 80% to 90% of patients who are refractory to medical management, and it has been shown to be superior to large-volume paracentesis in terms of survival.11 Several authors2,12 have used the TIPS procedure perioperatively to control refractory ascites. Telem et al2 note in their study of 21 patients that placement of a closed-suction drain did not control ascites-related wound complications, but the TIPS procedure was effective in controlling ascites when performed as a preoperative procedure or a treatment of ascites-related complications. Fagan et al12 described 3 patients who underwent the TIPS procedure perioperatively for umbilical hernia with impending rupture or leaking ascites to control ascites and allow healing of the umbilical hernia repair. Bajaj et al13 used the TIPS procedure as a rescue therapy in 5 patients who had a recurrent umbilical hernia after multiple unsuccessful attempts at repair to control ascites with good outcome. Slakey and others14 also reported a series of 8 patients in whom they used a temporary peritoneal dialysis catheter at the time of umbilical hernia repair to control ascites postoperatively. They found no wound complication and 1 recurrence during follow-up. Peritoneovenous shunt, once popular in the past, is no longer commonly used because of the effectiveness of other modalities and complications of the peritoneovenous shunt, such as disseminated intravascular coagulation and occlusion of the shunt.15
Use of synthetic mesh has been shown to reduce the recurrence rate in a randomized trial of patients without cirrhosis undergoing umbilical hernia repair.16 However, in patients with cirrhosis, there is a concern for impaired mesh ingrowth and wound infection in the presence of ascites.7 In one randomized trial of 80 patients with liver cirrhosis and umbilical hernia repair, the hernia recurrence rate was significantly less in the mesh group than in the primary repair group with at least 6 months of follow-up (14.2% vs 2.7%).17 However, in this study, the surgical site infection rate was higher in the mesh group than in the primary repair group, even though no patient required removal of the mesh, and it did not reach statistical significance (16.2% vs 8.5%). Therefore, one needs to balance risk of infection and risk of recurrence in deciding whether to use mesh in patients with liver cirrhosis who present with umbilical hernia.
Our study is not without limitations. From the data available, it is not possible to assess how many patients with umbilical hernia repair were observed expectantly and had no complications from umbilical hernia. This denominator is required to discuss indications for repair of umbilical hernia in this high-risk group. However, our findings add valuable information to the current literature in that the risk of surgery is clearly delineated for elective and emergency settings, as well as factors associated with worse outcome. These are important variables in assessing risk-benefit ratio of umbilical hernia repair in this high-risk group of patients.
Because the ACS-NSQIP data set does not capture liver cirrhosis as one of the preoperative variables, we used ascites and esophageal varices as signs of portal hypertension and liver cirrhosis. However, to ensure homogeneity of the study population, we excluded patients with causes of ascites other than portal hypertension. When comparing patients with ascites and/or esophageal varices with the control population, they had a higher MELD score, lower platelet count and albumin level, and higher bilirubin level and international normalized ratio, thereby suggesting that these patients had signs of portal hypertension from liver cirrhosis. On the other hand, the strength of our study is derived from the fact that we analyzed a nationally validated set of outcome data, with solid audit process and data integrity. Therefore, we believe our outcomes data can be generalized.
In conclusion, umbilical hernia repair in the presence of signs of portal hypertension, such as esophageal varices and ascites, is associated with significant morbidity and mortality. Emergency surgery for umbilical hernia repair is more commonly required for this group of patients than for the general population. Compared with elective surgery, emergency umbilical hernia repair is associated with higher morbidity rates. However, even elective surgery is associated with significant risk of adverse outcome in this high-risk group. Therefore, careful selection of surgical candidates is crucial. Elective repair of an umbilical hernia should be avoided in patients with predictors of postoperative mortality, such as age older than 65 years, MELD score higher than 15, and albumin level less than 3.0 mg/dL.
Correspondence: Paul D. Hansen, MD, Hepatobiliary and Pancreatic Surgery Program, Providence Portland Medical Center, 4805 Glisan St NE, Ste 6N50, Portland, OR 97213 (firstname.lastname@example.org).
Accepted for Publication: May 21, 2012.
Author Contributions:Study concept and design: Cho, Bhayani, Cassera, Hammill, Wolf, and Hansen. Acquisition of data: Cho, Bhayani, Cassera, and Hansen. Analysis and interpretation of data: Cho, Bhayani, Newell, Cassera, Hammill, Wolf, and Hansen. Drafting of the manuscript: Cho, Bhayani, Cassera, Hammill, and Hansen. Critical revision of the manuscript for important intellectual content: Cho, Bhayani, Newell, Cassera, Hammill, Wolf, and Hansen. Statistical analysis: Cho, Bhayani, Newell, Cassera, and Hammill. Obtained funding: Hansen. Administrative, technical, and material support: Cho, Bhayani, Cassera, Hammill, and Hansen. Study supervision: Wolf and Hansen.
Financial Disclosure: None reported.
Previous Presentation: This paper was presented at the 83rd Annual Meeting of the Pacific Coast Surgical Association; February 18, 2012; Napa Valley, California, and is published after peer review and revision.