Pediatric Specialist Care Is Associated with a Lower Risk of Bowel Resection in Children with Intussusception: A Population-Based Analysis

Transcription

1 Pediatric Specialist Care Is Associated with a Lower Risk of Bowel Resection in Children with Intussusception: A Population-Based Analysis Jarod P McAteer, MD, Steve Kwon, MD, MPH, Cabrini A LaRiviere, MD, MPH, Keith T Oldham, MD, FACS, Adam B Goldin, MD, MPH, FACS BACKGROUND: Although previous studies have shown that radiologic intussusception reduction is more likely at children s s, no study to date has compared outcomes among children advancing to surgical intervention. We hypothesized that rates of bowel resection would differ between s with and without pediatric surgeons. STUDY DESIGN: We conducted a population-based retrospective cohort study using Washington State discharge records. All children younger than 18 years undergoing operative intussusception reduction between 1999 and 2009 were included (n ¼ 327). Data were collected on demographics, disease severity, comorbidities, and concomitant gastrointestinal pathology. Multivariate logistic regression was used to estimate odds of intestinal resection during operative intussusception reduction. RESULTS: Pediatric s treated a smaller proportion of children older than 4 years of age (12.1% vs 44.4%), as well as a greater proportion of Medicaid patients (50.9% vs 42.6%). Patients at pediatric s had a lower prevalence of underlying intestinal anomalies or identifiable mass lesions (14.3% vs 16.7%). Severe disease (perforation, ischemia, acidosis) was more common at pediatric s (17.6% vs 9.3%). Overall, bowel resection was more commonly performed at nonpediatric s (59.3% vs 33.0%). On multivariate analysis, the odds of bowel resection were significantly lower at pediatric compared with nonpediatric s (odds ratio [OR] 0.20, p < 0.001), and this association was strongest in younger patients. Adjusted odds of postoperative complications were greater for bowel resection patients (OR 2.83, p < 0.001). CONCLUSIONS: Bowel resection during operative intussusception reduction is more likely at s without pediatric surgeons, and is associated with increased complications. Improved outcomes may be achieved by efforts aimed at standardizing care and decreasing variability in the treatment of pediatric intussusception. (J Am Coll Surg 2013;217:226e232. Ó 2013 by the American College of Surgeons) Intussusception is the most common cause of intestinal obstruction in the first 2 years of life, with a peak incidence in children ages 5 to 7 months. 1 Annual rates of CME questions for this article available at Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Received January 15, 2013; Revised February 25, 2013; Accepted February 25, From the Division of Pediatric General and Thoracic Surgery, Seattle Children s Hospital (McAteer, Goldin) and the Department of Surgery, University of Washington School of Medicine (McAteer, Kwon, Goldin), Seattle, WA; the Department of Surgery, Louisiana State University, New Orleans, LA (LaRiviere); and the Division of Pediatric Surgery, Children s Hospital of Wisconsin, Milwaukee, WI (Oldham). Correspondence address: Jarod P McAteer, MD, 4800 Sand Point Way NE, Seattle, WA jarodmc@u.washington.edu intussusception-related ization in infants are as high as 56 per 100,000, and incidence rates in older children may be higher than previously thought. 2,3 Given the frequency of the diagnosis and the importance of prompt treatment, the condition has historically been treated by a broad range of surgical providers, from adult surgeons in community s to fellowship-trained pediatric surgeons in tertiary children s centers. 4 Reduction of the intussuscepted bowel can be accomplished either by manual reduction in the operating room or by radiologic reduction using air-contrast or liquid contrast enema under fluoroscopic or ultrasound guidance. 5,6 Radiologic reduction has become the preferred first-line treatment for intussusception in stable children, although studies have shown radiologic reduction techniques are underused at nonchildren s s ª 2013 by the American College of Surgeons ISSN /13/$36.00 Published by Elsevier Inc

2 Vol. 217, No. 2, August 2013 McAteer et al Bowel Resection for Pediatric Intussusception 227 compared with children s s. 7,8 In those patients ultimately progressing to operative therapy, treatment may include segmental bowel resection if there is concern for a pathologic lead point or nonviable bowel, although there is some concern that practice may vary across s and providers. 9 A few studies have shown the risk of bowel resection to be higher in certain patient groups, but no investigation to date has analyzed the risk of progressing to bowel resection as it relates to the resources and characteristics of the treating among a large cohort of children undergoing operative intussusception reduction. 3,10 In order to address this question, we performed a population-based analysis of a large cohort of children undergoing operative intussusception reduction, comparing the risk of bowel resection according to the type of surgical personnel used by the. We hypothesized that the risk of bowel resection would be greater at s without fellowship-trained pediatric surgeons after controlling for patient- and disease-specific factors, and that the risk of postoperative complications would be greater for patients who underwent bowel resection. METHODS Study design We performed a population-based retrospective cohort study using the Washington State Comprehensive Hospital Abstract Reporting System (CHARS), a statewide inpatient discharge database that provides deidentified patient data regarding age, sex, payer status, diagnoses, procedures performed, length of stay, and discharge disposition. We hypothesized that the adjusted odds of bowel resection during operative intussusception reduction would be greater at s without pediatric surgeons ( nonpediatric s ) compared with s with pediatric surgeons ( pediatric s ). Further, we hypothesized that adjusted length of stay and postoperative complications would be greater for patients undergoing bowel resection. The study was approved by the University of Washington Institutional Review Board (IRB #39105). Study subjects Using discharge records from January 1, 1999 to December 31, 2009, we identified all patients less than 18 years of age who underwent an operation for reduction of intussusception. Patients were included if they had both a diagnostic code for intussusception (ICD-9 diagnosis code 560.0) and any code for operative reduction, intestinal manipulation, or intestinal resection (ICD-9 procedure codes 45.6 to 46.24, to 46.82, and 45.0 to 45.03). Children without both a diagnostic code for intussusception and a procedure code for operative intervention were excluded. In order to avoid including patients who were found to be already spontaneously reduced at exploration, we excluded those with codes for exploratory laparotomy or laparoscopy alone. Bowel resection was defined as any procedure code specifically indicating intestinal resection, reanastomosis, and/or creation of a stoma (45.6 to 46.24). This is similar to the approach taken in previous studies using similar data sources. 7,8 Covariates of interest Pediatric was defined as any employing full-time, fellowship-trained pediatric surgeons who were available 24/7 at that. Of the 5 s meeting these criteria, 3 were freestanding children s s employing only pediatric surgeons. Excluding the other 2 s (designated children s units in adult s) from the analysis did not substantively change our results, so all 5 were included in the analysis. All other s were considered nonpediatric s. None of the nonpediatric s in our study ever employed a fellowship-trained pediatric surgeon during the study period. Demographic factors including sex, age (0 to 4 years, >4 years, as in previous studies), and Medicaid insurance status 7 were extracted. Consistent with earlier studies, we defined severe disease using ICD-9 codes for sepsis, acidosis, bowel ischemia, peritonitis, and/or intestinal perforation. 8 Transfer status was defined as transfer from another inpatient as the source of admission, and was determined from the record. Information was also extracted regarding whether patients had coexisting gastrointestinal (GI) conditions that could potentially increase the likelihood of bowel resection. These conditions were determined from the discharge record and included intestinal duplications or diverticula (ICD-9 diagnosis code 751.5), intestinal ulcer (569.82), Meckel s diverticulum (751.0), Hirschsprung s disease (751.3), intestinal atresia (751.1, 751.2), malrotation (751.4), HenocheSchönlein purpura (287.0), inflammatory bowel disease (555.0 to 555.9, to 556.9), benign GI neoplasms (211.2, 211.3, 211.4, and 211.8), and malignant neoplasms (Appendix 1, online only). Chronic comorbid conditions were identified using a set of ICD- 9 codes selected a priori for adjustment because these variables are commonly considered as potential confounding factors in pediatric outcomes research (Appendix 2, online only). 11 Total length of stay (in days) was determined from the discharge record. Postoperative complications were defined using ICD-9 codes indicating complications attributed to medical or procedural treatment (Appendix 3, online only).

3 228 McAteer et al Bowel Resection for Pediatric Intussusception J Am Coll Surg Statistical analysis Descriptive statistics were used to compare characteristics according to type as well as bowel resection status. Multivariate logistic regression was used to quantify the association between type and odds of bowel resection. Age was analyzed both as a confounder and as a potential effect modifier by creating separate models for children 0 to 4 years and older than 4 years of age. Age was treated as a binary variable (0 to 4 years, >4 years) for the model on the full cohort and as a continuous variable in each of the age-stratified models. A multivariate linear regression model was used to estimate the adjusted total length of stay associated with bowel resection for children undergoing operation for intussusception. Multivariate logistic regression was also used to estimate the odds of postoperative complications according to bowel resection status. Risk estimates were adjusted for the covariates listed above. A p value < 0.05 was considered statistically significant. Regression models were adjusted for clustering at the level using a random-effects model in order to account for the nonindependence of sampling. Statistical analysis was performed using Stata 12. RESULTS Three hundred twenty-seven children underwent operative treatment for intussusception over the study period. A total of 31 s treated these patients (5 pediatric s, 26 nonpediatric s), and designations (pediatric/nonpediatric) for each of the s did not change over the study period. Eighty-three percent of patients were age 4 years or younger, and two-thirds were male. Eighty-three percent of patients received their operation at a pediatric. Overall, 12% of patients experienced postoperative complications. Compared with nonpediatric s, a smaller proportion of children treated at pediatric s were older than 4 years of age (12.1% vs 44.4%; Table 1). Medicaid insurance was more common at pediatric s (50.9% vs 42.6%). Although coexisting GI conditions were more common among patients treated at nonpediatric s (16.7% at nonpediatric, 14.3% at pediatric), markers of severe disease were more commonly seen in patients at pediatric centers (17.6% at pediatric, 9.3% at nonpediatric). Predictably, patients at pediatric s were more likely to have been transferred from another inpatient before receiving their operation. Bowel resection was more commonly performed at nonpediatric compared with pediatric s (59.3% vs 33.0%). Median total length of stay was 1 day longer at nonpediatric compared with pediatric s, and a higher proportion of patients at nonpediatric s experienced postoperative complications (16.7% vs 11.4%). In comparing children who did and did not undergo bowel resection, a greater proportion of bowel resection patients were older than age 4 years compared with nonbowel resection patients (27.0% vs 11.7%; Table 2). Additionally, both coexisting GI conditions and severe disease were much more common among patients who received a bowel resection. Transfer status was nearly twice as common among bowel resection patients as compared with nonbowel resection patients (13.1% vs 7.3%). Patients who underwent bowel resection had a median length of stay of 6 days, compared with 3 days for nonbowel resection patients. Postoperative complications were nearly 4 times as common among bowel resection patients as compared with nonbowel resection patients (23.0% vs 5.9%). Of those patients who underwent bowel resection, concomitant GI pathology was seen in a slightly greater proportion of patients at nonpediatric compared with Table 1. Patient Characteristics by Hospital Type Pediatric (n ¼ 273) Nonpediatric (n ¼ 54) Characteristic n % n % Age <12 mo e23 mo e4 y >4 y Sex Male Female Primary insurance Medicaid HMO Commercial Self-pay Other Coexisting condition* Severe disease y Transferred Bowel resection Length of stay, median (25th, 75th quartiles), d 3 (2, 5) 4 (3, 6) Complications *Intestinal anomalies, intestinal ulcer, Meckel s, Hirschprung s, intestinal atresia, malrotation, Henoch Schönlein purpura, benign or malignant neoplasm. y Perforation, peritonitis, bowel ischemia, acidosis, sepsis.

4 Vol. 217, No. 2, August 2013 McAteer et al Bowel Resection for Pediatric Intussusception 229 Table 2. Patient Characteristics by Bowel Resection Status No bowel resection (n ¼ 205) Bowel resection (n ¼ 122) Characteristic n % n % Age <12 mo e23 mo e4 y >4 y Sex Male Female Hospital type Pediatric Nonpediatric Primary insurance Medicaid HMO Commercial Self-pay Other Coexisting condition* Severe disease y Transferred Length of stay, median (25th, 75th quartiles), d 3 (2,4) 6 (4,8) Complications *Intestinal anomalies, intestinal ulcer, Meckel s, Hirschprung s, intestinal atresia, malrotation, Henoch Schönlein purpura, benign or malignant neoplasm. y Perforation, peritonitis, bowel ischemia, acidosis, sepsis. pediatric centers (Table 3), but among 0 to 4-year-old children, such lesions were more common among children who underwent resection at pediatric s. Severe disease at presentation was much more common among bowel resection patients at pediatric compared with nonpediatric s (44% vs 16%), as was either a coexisting GI condition or severe disease (64% vs 38%). In general, these differences were most prominent and statistically significant among children in the younger (0 to 4 years) age group. In our multivariate analysis, the adjusted odds of bowel resection were estimated for a number of clinical and demographic factors (Table 4). The adjusted odds of bowel resection was 80% less for children treated at a pediatric compared with those treated at a nonpediatric (odds ratio [OR] 0.20, p < 0.001). Other factors associated with bowel resection were severe disease (OR 21.23, p < 0.001) and presence of a coexisting GI condition (OR 4.83, p < 0.001). Table 3. Selected Characteristics of Bowel Resection Patients Only, by Hospital Type Pediatric (n ¼ 90) Nonpediatric (n ¼ 32) p Value Characteristic n % n % Age <12 mo e23 mo e4 y >4 y Coexisting condition* All patients e4 y(n¼74 pediatric, n ¼ 15 nonpediatric) >4 y(n¼ 16 pediatric, n ¼ 17 nonpediatric) Severe disease y All patients e4 y >4 y Either coexisiting condition or severe disease All patients e4 y >4 y *Intestinal anomalies, intestinal ulcer, Meckel s, Hirschprung s, intestinal atresia, malrotation, Henoch Schönlein purpura, benign or malignant neoplasm. y Perforation, peritonitis, bowel ischemia, acidosis, sepsis. When separate analyses were performed according to age group, the association between pediatric s and bowel resection was more protective and significant among 0 to 4-year-old children (OR 0.20, p < 0.001) than among children greater than 4 years of age (OR 0.46, p ¼ 0.33). Adjusted total length of stay was also analyzed in a multivariate model including these same variables. The adjusted length of stay for children who underwent bowel resection was 1.9 days more Table 4. Adjusted Odds of Bowel Resection According to Various Preoperative Risk Factors (All Patients) Risk factor Odds ratio 95% CI p Value Pediatric e0.41 <0.001 Age >4 y e Female e Medicaid primary e Severe disease e50.01 <0.001 Transfer from another facility e Coexisting condition e10.30 <0.001 Chronic comorbidity e

5 230 McAteer et al Bowel Resection for Pediatric Intussusception J Am Coll Surg than that for nonbowel resection patients, but this did not reach statistical significance (p ¼ 0.09). The adjusted odds of postoperative complications after bowel resection was more than 4 times that after procedures with no bowel resection (OR 4.69, p ¼ 0.001). DISCUSSION This is the most comprehensive study to date looking specifically at outcomes for children undergoing operative treatment of intussusception, and the first to investigate the risk of bowel resection according to the specialty training status of surgeons at certain s. Using a large, population-based state discharge database, we found that the risk of bowel resection during operative intussusception reduction was 80% less when it was performed at a employing full-time pediatric surgeons, after adjusting for patient- and disease-specific factors. We also found that receipt of bowel resection was associated with a significantly greater adjusted risk of postoperative complications. Although enema reduction is now considered first line therapy for intussusception in children, many children still require operative reduction. Progression to operative therapy may be due to a lack of radiologic resources or expertise, but in many instances it is due to concern for perforation and clinical deterioration, or in cases of failed radiologic reduction. 12 Operative therapy involves manual reduction of the intussusception, with bowel resection if necessary for ischemic bowel, pathologic lead points (mass lesions), or inability to manually reduce. Most cases of intussusception in children less than 4 years old are idiopathic in nature, commonly due to bowel wall lymphoid hyperplasia, which is self-limited and generally should not prompt resection. 13 Previous studies have shown the risk of bowel resection to be elevated with prolonged duration of symptoms, pathologic lead points, nonileocolic (eg, ileo-ileal) intussusception, and markers of severe disease, including fever and shock. 10,14,15 Not surprisingly, severe disease (eg, acidosis, perforation, peritonitis) and concomitant GI pathology were both associated with a significantly increased risk of bowel resection in our study. We also found that even after adjusting for disease severity, presence of GI pathology, and chronic comorbid conditions, bowel resection was significantly less likely if a patient was treated at a pediatric center rather than a nonpediatric center. Although 1 previous study suggested that community s might perform resections at a higher rate than children s s, this was evaluated only in a descriptive fashion. 9 Hospital designation has been noted to be an important driver of outcomes for a number of surgical procedures in children, including pyloromyotomy and appendectomy, with the mechanism of the association thought to be related primarily to the more extensive training and experience of surgeons at specialized centers Hospitallevel differences in the treatment of intussusception have also been evaluated, but before this study only with regard to the use of enema reduction as an initial therapeutic option. Both children s s and high volume s were noted previously to be more likely to perform radiologic reduction. 7,8 In our study, patients treated at pediatric s would have been almost exclusively treated by fellowship-trained pediatric surgeons, and the significantly decreased adjusted odds of bowel resection at pediatric s may reflect a greater comfort level among pediatric surgeons in not performing bowel resections in certain situations. The fact that a previous study found no difference in the odds of bowel resection according to volume may well indicate that the decision to perform a bowel resection depends more on surgeon rather than characteristics. 7 Studies of other pediatric procedures have indicated that common procedures, including pyloromyotomy, inguinal herniorrhaphy, and cholecystectomy, have outcomes that are more dependent on surgeon rather than factors So we chose to define type according to the training of the surgeons at those centers. If surgeon training is the key exposure that is being captured at pediatric s, there are a few potential reasons why the risk of bowel resection is lower at those centers. Pediatric surgeons have more experience in operating on these patients and may be more aggressive in manual bowel reduction before deciding to proceed with resection. Alternatively, pediatric surgeons may be more familiar with certain intraoperative findings that necessitate resection vs those that are less concerning. It is noteworthy that bowel resection patients treated at pediatric s in our study tended to have a greater prevalence of generally accepted indications for resection (eg, coexisting GI conditions, severe disease) than those treated at nonpediatric centers, and these differences were most pronounced in the younger age group. When our regression results were stratified by age, the negative association between bowel resection and pediatric s was greatest and most significant in younger children, which may partially reflect a relatively greater comfort level on the part of pediatric surgeons (relative to adult surgeons) in treating the youngest patients. The importance of limiting bowel resection, specifically in cases in which it is not absolutely necessary, is

6 Vol. 217, No. 2, August 2013 McAteer et al Bowel Resection for Pediatric Intussusception 231 manifested by the results of postoperative outcomes in our cohort. Previous studies have shown that length of stay is longer and postoperative complications more common among children with intussusception who undergo bowel resection compared with those who do not. 5 These findings, however, may be confounded by the fact that children who undergo bowel resection are often sicker to begin with. In our study, even after controlling for other disease factors, total length of stay was longer and risk of in- complications was higher for patients who underwent bowel resection, although length of stay data did not quite reach statistical significance. Although these findings are somewhat expected, this is the largest study to show this association and one of the few to control for other factors associated with bowel resection. This study has some limitations. These are observational data, so associations should be interpreted with caution. The Washington State Comprehensive Hospital Abstract Reporting System database is an administrative discharge database, and is therefore subject to potential misclassification of exposure and outcomes, which we have attempted to minimize by using stringent inclusion and exclusion criteria and definitions consistent with those used in previous studies. There are also certain factors for which one cannot control, including location of intussusception and the inability to manually reduce intraoperatively. Delay in presentation is another factor associated with increased risk for bowel resection that cannot be gleaned from the data, although we have controlled for transfer status in our model. Also, it should be mentioned that although surgeon training may be the key factor responsible for the results of this study, other salient factors at nonpediatric s, including rural location, limited resources, and later presentation for treated patients, may influence our findings, and results should therefore be interpreted with this in mind. Last, because this is a population-based sample from a single state, the findings may not be generalizable to all patient populations. CONCLUSIONS In this study, we found that bowel resection for children undergoing operative treatment for intussusception is less likely with care at a pediatric. Further, our results show that bowel resection is independently associated with adverse postoperative outcomes. Although bowel resection is required in certain cases for adequate treatment and further pathologic diagnosis, these data suggest that this practice is not performed with uniformity across different s. Given what is already known regarding the decreased likelihood of attempting enema reduction at nonchildren s s, our results further highlight the importance of experienced surgeons, experienced radiologists, and an experienced system in treating these children, especially those of younger age. Further efforts should be directed toward standardizing care for pediatric intussusception, emphasizing the importance of treatment pathways and, for those requiring operative intervention, the importance of surgeon comfort level with the procedure and with interpretation of intraoperative findings. Author Contributions Study conception and design: McAteer, Kwon, Oldham, Goldin Acquisition of data: McAteer, LaRiviere Analysis and interpretation of data: McAteer, Kwon, LaRiviere Drafting of manuscript: McAteer Critical revision: Kwon, LaRiviere, Oldham, Goldin REFERENCES 1. Fischer TK, Bihrmann K, Perch M, et al. Intussusception in early childhood: a cohort study of 1.7 million children. Pediatrics 2004;114:782e Parashar UD, Holman RC, Cummings KC, et al. Trends in intussusception-associated izations and deaths among US infants. Pediatrics 2000;106:1413e Cochran AA, Higgins GL, Strout TD. Intussusception in traditional pediatric, nontraditional pediatric, and adult patients. Am J Emerg Med 2011;29:523e Cosper GH, Hamann MS, Stiles A, et al. Hospital characteristics affect outcomes for common pediatric surgical conditions. Am Surg 2006;72:739e Kaiser AD, Applegate KE, Ladd AP. Current success in the treatment of intussusception in children. Surgery 2007;142: 469e475; discussion 475e Applegate KE. Intussusception in children: evidence-based diagnosis and treatment. Pediatr Radiol 2009;39: S140eS Bratton SL, Haberkern CM, Waldhausen JH, et al. Intussusception: size and risk of surgery. Pediatrics 2001; 107:299e Jen HC, Shew SB. The impact of type and experience on the operative utilization in pediatric intussusception: a nationwide study. J Pediatr Surg 2009;44:241e Shekherdimian S, Lee SL. Management of pediatric intussusception in general s: diagnosis, treatment, and differences based on age. World J Pediatr 2011;7:70e Johnson B, Gargiullo P, Murphy TV, et al. Factors associated with bowel resection among infants with intussusception in the United States. Pediatr Emerg Care 2012;28:529e Feudtner C, Christakis DA, Connell FA. Pediatric deaths attributable to complex chronic conditions: a populationbased study of Washington State, Pediatrics 2000;106:205e209.

7 232 McAteer et al Bowel Resection for Pediatric Intussusception J Am Coll Surg 12. Fike FB, Mortellaro VE, Holcomb GW, et al. Predictors of failed enema reduction in childhood intussusception. J Pediatr Surg 2012;47:925e Goldin A, Sawin RS. Intussusception. In: Myers J, Millikan K, Saclarides T, eds. Common Surgical Diseases: An Algorithmic Approach to Problem Solving. New York: Springer; 2008: 259e Chua JH, Chui CH, Jacobsen AS. Role of surgery in the era of highly successful air enema reduction of intussusception. Asian J Surg 2006;29:267e Justice FA, Auldist AW, Bines JE. Intussusception: trends in clinical presentation and management. J Gastroenterol Hepatol 2006;21:842e Ly DP, Liao JG, Burd RS. Effect of surgeon and characteristics on outcome after pyloromyotomy. Arch Surg 2005; 140:1191e Whisker L, Luke D, Hendrickse C, et al. Appendicitis in children: a comparative study between a specialist paediatric centre and a district general. J Pediatr Surg 2009;44: 362e Collins HL, Almond SL, Thompson B, et al. Comparison of childhood appendicitis management in the regional paediatric surgery unit and the district general. J Pediatr Surg 2010;45:300e Raval MV, Cohen ME, Barsness KA, et al. Does type affect pyloromyotomy outcomes? Analysis of the Kids Inpatient Database. Surgery 2010;148:411e Pranikoff T, Campbell BT, Travis J, et al. Differences in outcome with subspecialty care: pyloromyotomy in North Carolina. J Pediatr Surg 2002;37:352e Langer JC, To T. Does pediatric surgical specialty training affect outcome after Ramstedt pyloromyotomy? A population-based study. Pediatrics 2004;113:1342e Borenstein SH, To T, Wajja A, et al. Effect of subspecialty training and volume on outcome after pediatric inguinal hernia repair. J Pediatr Surg 2005;40:75e Chen K, Cheung K, Sosa JA. Surgeon volume trumps specialty: outcomes from 3596 pediatric cholecystectomies. J Pediatr Surg 2012;47:673e680.