Analysis of Factors Predictive of Gastrointestinal Tract Leak in Laparoscopic and Open Gastric Bypass

Hypothesis Patient characteristics and comorbidities, payer type, and operative technique (laparoscopic vs open) predict the risk of gastrointestinal (GI) tract leak in patients with morbid obesity undergoing gastric bypass.

Design Retrospective database analysis.

Setting Nationwide Inpatient Sample.

Patients Between January 1, 2006, and December 31, 2008, patients who underwent open or laparoscopic gastric bypass to treat morbid obesity.

Main Outcome Measures Factors predictive of GI tract leak using multivariate regression analyses.

Results A total 226 452 patients underwent laparoscopic (81.2%) or open (18.8%) gastric bypass during the 3-year period. Most patients were female (80.5%) and of white race/ethnicity (73.6%). The mean age of patients was 43.6 years; 30.0% of patients were older than 50 years. The overall prevalence of in-hospital GI tract leak was 0.7%. The GI tract leak rate was significantly lower in laparoscopic compared with open gastric bypass (0.3% vs 2.0%, P < .01). Using multivariate regression analysis, factors associated with higher risk of GI tract leak were open gastric bypass (adjusted odds ratio [aOR], 4.85), congestive heart failure (aOR, 3.04), chronic renal failure (aOR, 2.38), age older than 50 years (aOR, 1.82), Medicare payer (aOR, 1.54), male sex (aOR, 1.50), and chronic lung disease (aOR, 1.21). The GI tract leak rate was unaffected by race/ethnicity, hypertension, diabetes mellitus, sleep apnea, hyperlipidemia, liver disease, peripheral vascular disease, or smoking.

Conclusions We identified multiple factors associated with the higher risk of GI tract leak after gastric bypass. Surgeons should use this knowledge to counsel patients and possibly alter operative plans in high-risk patients to minimize this risk.

Gastrointestinal (GI) leak is a serious postoperative complication after Roux-en-Y gastric bypass that contributes to significant morbidity and mortality. The prevalence of GI tract leak after gastric bypass has been reported to be as high as 5.6% and has been shown to be an independent predictor of perioperative death.^1-8 In patients with postoperative GI tract leak, the overall leak-associated mortality can be as high as 17%.^4,9 High morbidity and mortality associated with GI tract leak are related to the difficulty in the diagnosis and treatment of this condition. Diagnosis of GI tract leak often requires a high index of suspicion, as early recognition and treatment are crucial.

Knowledge of factors associated with the higher risk of GI tract leak is important to aid surgeons in counseling patients, selecting surgical approaches, and developing high clinical suspicion for detection of leak in this high-risk patient population. One study³ to date has specifically examined preoperative factors predictive of GI tract leak. Because the prevalence of GI tract leak after gastric bypass can be low, analysis of factors that may predict leak requires a large series of patients, which frequently is unavailable from most single institutions. In this study, we used the Nationwide Inpatient Sample (NIS) database (http://www.hcup-us.ahrq.gov/nisoverview.jsp) to evaluate the prevalence of GI tract leak among patients undergoing bariatric surgery between January 1, 2006, and December 31, 2008, and to identify factors that may be predictive of GI tract leak following gastric bypass.

The NIS is the largest inpatient care US database, in which approximately 1000 hospitals participate, and contains information from almost 8 million hospital stays each year across the country. It is composed of a nationally representative sample of approximately 20% of US community hospitals, resulting in a sampling frame that comprises about 90% of all hospital discharges in the United States. Data elements within the NIS are drawn from hospital discharge abstracts, which allow determination of all procedures performed during a given hospital admission. The NIS also contains discharge information on inpatient hospitalizations, including patient characteristics, length of hospital stay, specific postoperative morbidity, and observed in-hospital mortality. The NIS database has no information available on complications occurring after discharge. Approval for the use of NIS patient-level data in this study was obtained from the Institutional Review Board of the University of California Irvine Medical Center and from the NIS.

Quiz Ref IDUsing NIS data from January 1, 2006, to December 31, 2008, we analyzed discharge data on patients with morbid obesity who underwent laparoscopic or open Roux-en-Y gastric bypass using International Classification of Diseases, Ninth Revision (ICD-9) procedure codes for laparoscopic gastric bypass (code 44.38) and open gastric bypass (codes 44.31 and 44.39) with principal diagnosis codes of obesity and morbid obesity (codes 278.0, 278.01, 278.8, and 278.1). Smokers were identified by ICD-9 codes 305.1 and V15.82. Gastrointestinal tract leak complication was identified using ICD-9 diagnosis codes for postoperative intra-abdominal abscess (code 998.59) or persistent fistula (code 998.6). Preoperative factors that were analyzed included patient characteristics and comorbidities, payer type, and operative technique (laparoscopic vs open). For clinical interpretation of the results, all factors were analyzed as dichotomous variables, categorized as follows: (1) patient-specific factors (sex, race/ethnicity [white vs nonwhite], payer type [private vs other], and age [≤50 vs >50 years]), (2) comorbidities (yes or no), and (3) operative technique (laparoscopic vs open). Multivariate regression analyses were performed to identify independent predictors of postoperative GI tract leak.

All the values for continuous variables are expressed as the mean (SD). Multiple logistic regression analyses were performed, and the adjusted odds ratio (aOR) (95% confidence interval) was calculated to determine the combined effect of various preoperative factors on outcomes. Female sex, age 50 years or younger, white race/ethnicity, laparoscopic operative technique, and private payer were used as references. All statistical analyses for the NIS data were performed using commercially available software (SAS, version 9.2; SAS Institute, Cary, North Carolina), incorporating recommended discharge and hospital weights. Discharge weight was used to create national estimates based on sampling of the data. Statistical significance was set at P < .05 and an OR (95% confidence interval) that excluded 1.

Quiz Ref IDA total of 226 452 patients underwent laparoscopic (81.2%) or open (18.8%) gastric bypass during the 3-year period. Most patients were female (80.5%) and of white race/ethnicity (73.6%) (Table 1). The mean age of patients was 43.6 years; 30.0% of patients were older than 50 years. The most prevalent comorbidities were hypertension (54.7%), diabetes mellitus (30.4%), and hyperlipidemia (20.3%). Private insurance was the most common payer type (73.7%).

The overall prevalence of in-hospital GI tract leak was 0.7%. The GI tract leak rate was significantly lower in laparoscopic compared with open gastric bypass (0.3% vs 2.0%, P < .01). Compared with patients who did not have GI tract leak, patients who had GI tract leak experienced a significantly longer mean length of hospital stay (21.5 vs 2.6 days), higher mean hospital charges ($192 218 vs $39 706), and greater in-hospital mortality (6.8% vs 0.1%) (P < .01 for all) (Table 2).

Table 3 summarizes the multivariate logistic regression analyses for in-hospital GI tract leak. For patient characteristics, factors associated with higher risk of GI tract leak were age older than 50 years (aOR, 1.82]), Medicare payer (aOR, 1.54), and male sex (aOR, 1.50). Comorbidities that were independent factors associated with increased risk of GI tract leak were congestive heart failure (aOR, 3.04), chronic renal failure (aOR, 2.38), and chronic lung disease (aOR, 1.21). Open operative technique (aOR, 4.85) was associated with a higher rate of GI tract leak. There was no effect of race/ethnicity, diabetes, hypertension, liver disease, smoking, hyperlipidemia, peripheral vascular disease, or sleep apnea on the GI tract leak rate.

Quiz Ref IDUsing data from a large sample of patients who underwent gastric bypass surgery between January 1, 2006, and December 31, 2008, this study showed a low overall in-hospital GI tract leak rate of 0.7%. Our results demonstrated that GI tract leak is a significant risk factor for morbidity and mortality in gastric bypass, with a leak-associated mortality of 6.8%.Quiz Ref IDUsing multivariate logistic regression analyses, several factors were significantly associated with higher risk of GI tract leak in gastric bypass surgery, including open operative technique, age older than 50 years, Medicare payer, male sex, congestive heart failure, chronic renal failure, and chronic lung disease. Understanding these factors will aid surgeons in counseling patients, selecting the type of bariatric procedure, and developing a high clinical index of suspicion for detection of GI tract leak in these high-risk patients.

Factors associated with GI tract leak after bariatric surgery can be multifactorial. These include systemic variables (age, nutrition, and metabolic disease), local factors (infection, ischemia, and edema), and technical conditions.¹⁰ In our study, patient characteristics that were independently predictive of higher risk for GI tract leak were age older than 50 years, Medicare payer, and male sex. Using logistic regression analysis of 3073 open and laparoscopic gastric bypass cases, Fernandez et al³ similarly found that age older than 55 years and male sex were predictors of higher risk for GI tract leak. Studies^10-12 have shown that older patients are more likely than younger patients to have surgical wound dehiscence and delayed healing. Our study showed that male patients had a 1.5 times greater likelihood of GI tract leak than female patients. Male patients often have android body habitus and increased intra-abdominal obesity, contributing to technical difficulty during the procedure that may lead to increased risk of GI tract leak.¹³ Livingston et al¹⁴ and Livingston and Ko¹⁵ demonstrated that male sex was an independent risk factor for major complications in a multivariate analysis of 1067 patients who underwent gastric bypass. Medicare recipients undergoing bariatric surgery are high-risk patients and may incur greater morbidity and mortality. In our study, Medicare payer status was associated with higher risk of GI tract leak. Examining Medicare databases between 1997 and 2002, Flum et al¹⁶ similarly reported that Medicare beneficiaries undergoing bariatric surgery are at risk of early death.

Quiz Ref IDAmong patient comorbidities, congestive heart failure (aOR, 3.04), chronic renal failure (aOR, 2.38), and chronic lung disease (aOR, 1.21) were predictive of increased risk for GI tract leak. To our knowledge, no prior study has examined these comorbidities as factors predictive of GI tract leak. Heart failure and renal failure can contribute to impaired wound healing and may be a cofactor for development of other conditions. In patients with chronic lung disease, hypoxemia is a contributing factor to impaired wound healing. Low oxygen tension has been shown to have a profound deleterious effect on all aspects of wound healing.^10-12 In our study, smoking, hypertension, diabetes mellitus, hyperlipidemia, liver disease, sleep apnea, and peripheral vascular disease were not independent risk factors for GI tract leak. Similar to our study, Livingston and Ko¹⁵ found no effect of smoking, hypertension, diabetes mellitus, or sleep apnea on the risk of major complications following gastric bypass. Contrary to our findings, Fernandez et al³ reported that diabetes mellitus was a significant risk factor for GI leak; however, their results were specifically related to laparoscopic gastric bypass.

In our study, a procedure-specific factor that was associated with increased in-hospital GI tract leak was open operative technique. This finding agrees with recent study findings that compared outcomes of laparoscopic vs open gastric bypass. In a study of 22 422 patients undergoing laparoscopic and open gastric bypass surgery between 2004 and 2006, Nguyen et al¹ reported that the GI tract leak rate after laparoscopic gastric bypass (1.4%) was significantly lower than that after open gastric bypass (3.1%), with an OR of 2.2 (95% confidence interval, 1.8-2.7). Similarly, using 2005 data from the NIS, Weller and Rosati¹⁷ found that patients undergoing open gastric bypass surgery were more likely to experience reoperation and anastomotic GI tract leak complications (OR, 1.32). In contrast to our findings, Durak et al¹⁸ reported that the GI tract leak rate was considerably higher using laparoscopic gastric bypass compared with open gastric bypass (2.3% vs 0.7%, P < .05) based on data among 1133 patients who underwent primary gastric bypass between 1996 and 2006. However, their results are likely related to a learning curve in the laparoscopic gastric bypass procedure.

There are limitations to this study. The NIS database is compiled from abstracted discharge data, limited to in-hospital stays without outpatient follow-up information. For example, GI tract leak that occurs after discharge would not be captured in this database. Therefore, our calculation of the GI tract leak rate probably underestimates the actual leak rate. Durak et al¹⁸ reported that 6 of 17 GI tract leaks (35.3%) had onset symptoms later than 48 hours after surgery, and such late occurrences potentially would not be detected after discharge. Despite this limitation, underestimation of the leak rate in this study does not alter our findings because the primary objective of this study was to examine factors predictive of GI tract leak, not to determine the true prevalence of leak. The NIS database does not have information on body mass index, which could be a predictive factor for increased risk of GI tract leak. We were unable to distinguish gastric bypass performed as primary procedures vs revisional bariatric surgical procedures, most of which are performed as open operations and are associated with significantly higher risk of GI tract leak.³ There are no ICD-9 codes for laparoscopic cases that require conversion to open surgery, and these cases were likely included in the open operative technique group. Finally, unavailable for analysis within the NIS database are other potential factors predictive of GI tract leak, such as individual surgeon experience, annual case volume, and specific details about the operative technique. Despite these limitations, our study is the largest to date that evaluates factors predictive of GI tract leak after gastric bypass.

At 0.7%, the rate of GI tract leak following gastric bypass surgery was low herein. However, GI tract leak is a major contributor to morbidity and mortality after gastric bypass. Independent risk factors for development of GI tract leak were open operative technique, age older than 50 years, Medicare payer, male sex, congestive heart failure, chronic renal failure, and chronic lung disease. Knowledge of these factors may aid surgeons in counseling patients, selecting an operative technique that may reduce GI tract leak, and developing high clinical suspicion for detection of leak in these high-risk patients.

Correspondence: Ninh T. Nguyen, MD, Department of Surgery, University of California, Irvine Medical Center, 333 City Blvd W, Ste 850, Orange, CA 92868 (ninhn@uci.edu).

Accepted for Publication: May 2, 2011.

Author Contributions:Study concept and design: Masoomi, Kim, Reavis, Mills, Stamos, and Nguyen. Acquisition of data: Masoomi and Nguyen. Analysis and interpretation of data: Masoomi and Nguyen. Drafting of the manuscript: Masoomi, Kim, Stamos, and Nguyen. Critical revision of the manuscript for important intellectual content: Masoomi, Reavis, Mills, Stamos, and Nguyen. Statistical analysis: Masoomi and Nguyen. Administrative, technical, and material support: Masoomi and Kim. Study supervision: Reavis, Mills, Stamos, and Nguyen.

Previous Presentation: This paper was presented at the 82nd Annual Meeting of the Pacific Coast Surgical Association; February 19, 2011; Scottsdale, Arizona, and is published after peer review and revision.