Kolakowski S, Kirkland ML, Schuricht AL. Routine Postoperative Upper Gastrointestinal Series After Roux-en-Y Gastric BypassDetermination of Whether It Is Necessary. Arch Surg. 2007;142(10):930-934. doi:10.1001/archsurg.142.10.930
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
To evaluate the clinical utility of the routine use of postoperative barium swallow to diagnose postoperative complications in patients undergoing open or laparoscopic Roux-en-Y gastric bypass.
A total of 417 consecutive patients undergoing Roux-en-Y gastric bypass at our institution between January 1, 2001, and December 31, 2002, were included. We performed 341 open procedures and 76 laparoscopic gastric bypasses. All patients received a limited postoperative fluoroscopic upper gastrointestinal series, except for the patients who exceeded the weight limitation of the radiologic equipment. Radiologic findings of anastomotic complications were anastomotic leak, delayed gastric emptying, gastric outlet obstruction, and gastrogastric fistula. We evaluated clinical signs and symptoms to obtain a list of criteria suggesting these complications. Patients were stratified into 2 groups: those with and those without radiographic anastomotic complications. Clinical and radiologic criteria were compared using univariate and multivariate logistic regression analysis.
We noted 42 radiologic abnormalities during a routine postoperative barium swallow evaluation. Among our 417 patients, we documented 12 leaks (2.9%), 19 cases of delayed gastric emptying (4.6%), 4 gastric outlet obstructions (1.0%), and 7 gastrogastric fistulas (1.7%). The combination of fever, tachycardia, and tachypnea was the most specific indicator of a leak, at 0.99 (95% confidence limit, 0.99, 1.01). Nausea with vomiting was the most predictive indicator of delayed gastric emptying and gastric outlet obstruction, with a specificity of 0.99 (95% confidence limit, 0.98, 0.99) and 0.97 (95% confidence limit, 0.96, 0.99), respectively.
Postoperative complications after Roux-en-Y gastric bypass surgery are predictable based on the patient's symptoms. The use of routine postoperative fluoroscopic upper gastrointestinal series is unnecessary in asymptomatic patients.
During a 30-year interval, from 1960 to 1990, the incidence of overweight Americans increased from 13% to 35%.1 With morbid obesity reaching epidemic levels in the United States, with greater than 30% of the adult population affected by this disease, overweight and obesity are the number 1 nutritional disorders in the United States.2 Obesity is a risk factor for many chronic conditions, including diabetes mellitus, hypertension, hypercholesterolemia, stroke, heart disease, sleep apnea, gastroesophageal reflux disease, certain cancers, and arthritis. A reduction in the prevalence of obesity has long-term health and economic benefits.
Bariatric surgery is being performed in increasing numbers in an era when reimbursements are being reduced. Bariatric surgery is a cost-effective alternative to no treatment and provides substantial health benefits to severely obese patients.3 The cost of the procedure ranges from $17 000 to $30 000 and is usually covered by most insurance companies. The use of clinical pathways for this subset of patients has been shown to lower expenses and improve the quality of care.4 A rate-limiting step in the discharge of a postoperative patient is the advancement of the patient's diet. This can be accomplished quicker if a swallowing study is not necessary, which also lowers the cost of the hospital stay by approximately $500, based on the price of the study alone.
The goal of this study was to review outcomes of patients who underwent gastric bypass. In particular, we were interested in the value of radiologic swallowing studies postoperatively. We wanted to determine whether the studies are a necessary step in the postoperative course or whether we can feel comfortable advancing the patient's diet solely on the basis of clinical signs and symptoms.
We identified all patients who underwent gastric bypass surgery between January 1, 2001, and December 31, 2002, at Pennsylvania Hospital. All procedures were performed according to standards provided by the American College of Surgeons for facilities performing bariatric surgery.5 In all patients, the indication for the procedure was a body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) of at least 40 or, if comorbidities existed, a BMI of at least 35, supported by National Institutes of Health consensus statement guidelines.6
The operations were performed either open or laparoscopically according to the surgeon's preference and patient habitus. The open procedures were all performed using an end-to-end anastomosis, and the laparoscopic bypasses were performed using an Endostich hand-sewn anastomosis. The gastric pouch with open procedures was performed by excluding the gastric remnant using 2 firings of an endoscopic stapler. The laparoscopic pouches were created by completely dividing the stomach. The open procedures did not involve any specific intraoperative measures to evaluate the anastomosis. The laparoscopic bypasses were performed with an endoscope stenting the anastomosis during the creation of the gastrojejunostomy. The anastomosis was then tested using the endoscope, inflating the pouch under isotonic sodium chloride solution to check for bubbles.
An upper gastrointestinal contrast study was performed within 48 hours on all patients undergoing Roux-en-Y gastric bypass during the study period. A standard protocol was used for the postoperative swallow study by the radiology department. The protocol consisted of an initial scout film of the abdomen. The patient was then positioned in a semiupright position on the fluoroscopic table. Approximately 60 mL of a combination of oral diatrizoate meglumine and diatrizoate sodium (Gastrografin; Bracco Diagnostics, Princeton, New Jersey) was administered. Fluoroscopy was performed as the patient swallowed the contrast, and images were obtained of the esophagus, gastric pouch, and jejunum. Fifteen-minute delayed films were obtained to evaluate the pouch's emptying capacity and to assess for leaks or reflux into the duodenum. The radiologist and surgeon (M.L.K. or A.L.S.) reviewed the films before the advancement of the patient's diet.
Patient demographic characteristics, clinical factors, radiologic studies, and outcomes were collected from hospital charts. Key subjective findings observed were nausea and vomiting. Key objective factors observed were fever (temperature, > 38.6°C), tachycardia (heart rate, > 100 beats/min), tachypnea (respirations, > 20/min), and elevated white blood cell count (> 11 000/μL).
Twelve categories of single or combined clinical indicators were used to evaluate the patient's clinical course. These categories included the following: (1) tachycardia; (2) elevated white blood cell count; (3) nausea; (4) nausea and vomiting; (5) fever; (6) tachypnea; (7) tachypnea and fever; (8) tachycardia and fever; (9) tachypnea and tachycardia; (10) fever and elevated white blood cell count; (11) nausea, vomiting, and tachycardia; and (12) fever, tachycardia, and tachypnea.
All patients underwent a postoperative swallowing study unless the radiology equipment was unable to accommodate the weight and/or girth of the patient. The key radiologic outcomes identified were normal postoperative findings, delayed gastric emptying, gastric outlet obstruction (GOO), anastomotic leak, and gastrogastric fistula.
After data collection, the patients were stratified into 2 groups: those with and those without radiographic anastomotic complications. Between-group analysis of clinical indicators was performed for the 4 radiologic findings using the unpaired t test for continuous variables and the χ2 test for discrete variables. Clinical and radiologic criteria were compared using univariate and multivariate logistic regression analysis between the 2 groups. Statistical significance was assumed for P < .05.
Three surgeons at Pennsylvania Hospital performed 417 gastric bypass surgical procedures between January 1, 2001, and December 31, 2002. A total of 341 patients underwent open Roux-en-Y gastric bypass, and 76 underwent laparoscopic Roux-en-Y gastric bypass. Nine of the laparoscopic Roux-en-Y gastric bypass procedures had to be converted to open procedures. The total population's preoperative demographic characteristics and comorbidities were reviewed and are presented in Table 1.
Forty-two radiologic abnormalities were documented for the 417 procedures. Twelve anastomotic leaks (2.9%) were identified. Nineteen cases (4.6%) of delayed gastric emptying were documented. Four GOOs (1.0%) and 7 gastrogastric fistulas (1.7%) were noted. A total of 4 patients were excluded because of limitations of the radiology equipment. None of these patients developed clinical symptoms of anastomotic complications.
Twelve anastomotic leaks were identified by radiologic studies. Of the 12 radiographic leaks, 6 (50.0%) required subsequent operations, 2 (16.7%) were managed conservatively, and 4 (33.3%) were false-positive studies by the radiologist. By univariate analysis, radiographic leaks were seen in patients with the following clinical indicators: combination of fever, tachycardia, and tachypnea; combination of fever, tachypnea, nausea, and vomiting; combination of fever and tachycardia; and nausea (Table 2). After multivariate analysis, the combination of fever, tachycardia, and tachypnea was identified as a significant predictor of an anastomotic leak (Table 3). This combination was the most specific indicator of a leak, at 0.99 (95% confidence limit [CL], 0.99, 1.01), with a likelihood ratio positive of 298.37 (95% CL, 41.42, 2149.35) (Table 4). Statistically significant demographic characteristics and comorbidities identified were being female (10 of the 12 patients [83.3%]) (P < .001), having diabetes mellitus (6 of the 12 patients [50.0%]) (P < .02), and having obstructive sleep apnea (7 of the 12 patients [58.3%]) (P = .02). The mean length of stay for patients with leaks was 22.00 days compared with 3.77 days for the nonleak group (P < .03). The average BMI of the leak group was 55.1 (P = .20).
Nineteen cases of delayed gastric emptying were documented by radiologic studies. Nausea combined with vomiting was the most predictive indicator of delayed gastric emptying, with a specificity of 0.99 (95% CL, 0.98, 0.99) and a likelihood ratio positive of 72.53 (95% CL, 29.45, 178.63) (P < .001). Of statistical significance, 10 of the 19 delayed gastric emptying cases were found in patients undergoing an open procedure; the mean BMI of those experiencing delayed gastric emptying was 45.9 vs 50.8 for those with normal study results (P < .002).
Four GOOs were identified. Nausea combined with vomiting was the most predictive indicator of GOO, with a specificity of 0.97 (95% CL, 0.96, 0.99) and a likelihood ratio positive of 18.77 (95% CL, 6.00, 58.72) (P < .003). Hypertension was the only statistically significant comorbidity identified in 3 of the 4 patients with GOO (P = .04).
Gastrogastric fistulas were identified in 7 of the patients studied. No clinically significant indicators were identified for this finding; however, 2 of the 7 patients with this finding did have nausea.
This series represents more than 400 consecutive gastric bypass procedures performed by 3 bariatric surgeons at 1 institution over a 2-year period. Most of these patients underwent open Roux-en-Y gastric bypass because the laparoscopic approach was initiated during this time.
Our results show that an abnormal swallowing study was found in 42 patients (10.0%), of whom only 8 (6 with an anastomotic leak and 2 with GOO) required a subsequent operation because of the abnormal study result. Each patient requiring a subsequent operation exhibited the clinical criteria predicting an abnormal swallowing study result.
Our clinical leak rate of 2.9% is consistent with that of others reported in the literature.7 Anastomotic leak is the most severe complication that can occur after gastric bypass surgery and is associated with a high mortality rate. Of the 3 deaths observed in our series, 2 were because of anastomotic leaks. The other death recorded was secondary to a pulmonary embolism. The similar clinical factors that these 2 dangerous consequences have are tachycardia and tachypnea. Most of our bariatric patients postoperatively tend to have a heart rate that exceeded 100 beats/min. However, in this study, each patient who developed a leak had a heart rate of greater than 120 beats/min. The presence of these signs with or without fever is extremely worrisome in the postoperative course and requires 2 completely different strategies for therapy. If a fever is present, the first study would be a swallowing study; if it is absent, based on clinical information, a respiratory source must be evaluated.
Patients with diabetes mellitus and sleep apnea had a higher incidence of anastomotic leak. After reviewing our patients with sleep apnea and an anastomotic leak, 5 of 7 patients were using bilevel positive airway pressure devices in the postoperative period. Bilevel positive airway pressure is used to prevent apneic arrest in patients with obstructive sleep apnea. Our concern with this device is that pressured air can overinflate the stomach and proximal intestines if improperly fitted and may contribute to anastomotic disruption. Our institution recommends that all patients who use bilevel positive airway pressure devices preoperatively bring their own devices to use in the hospital postoperatively.
Of the 12 anastomotic leaks, 6 (50.0%) required surgery for deterioration in clinical status and 2 (16.7%) could be managed with Jackson-Pratt drainage of the anastomotic site and parenteral nutrition. The remaining 4 patients (33.3%) with radiographic leaks were false positives in this study because they never progressed to a clinical leak. The 4 patients with false-positive leaks were clinically asymptomatic and delayed initiation of oral intake an extra 48 hours to rule out the possibility of a leak by performing serial blood work and a physical examination. These patients were also closely monitored in the hospital for an extra 24 hours after the initiation of oral intake. In our series, no patient developed a clinically significant anastomotic leak after postoperative day 2.
The findings of GOO and delayed gastric emptying had common clinical predictors of nausea and vomiting. The most likely reason for this is that they both have a common point of defect. The gastrojejunal anastomosis can develop edema after the operation. Based on the severity of the edema, the patient may or may not be able to pass oral and gastric secretions through the anastomosis. With the information of the swallowing study, we were able to conservatively treat all of the delayed gastric emptying cases and 2 of the 4 GOO cases via conservative management, consisting of nasogastric tube decompression of the gastric pouch. The 2 GOOs requiring surgery did not respond to conservative management and required revisional surgery.
Our gastrogastric fistula rate of 1.7% is well below the reported rate for divided gastric bypass surgical procedures.8 The fistulas that we documented did not need any further management. The value of knowing whether a fistula is present may be useful if the patient is not losing the appropriate amount of weight. In the early setting, this finding is of little value unless the staple line has a large defect, which would require a subsequent operation.
Based on our results, we believe radiographic swallowing studies may be an unnecessary step in the routine postoperative care of patients undergoing gastric bypass. Clinical signs and symptoms are excellent surveillance tools in the postoperative period to identify complications. By eliminating this step in the postoperative clinical pathway, the patient is able to begin an oral diet quicker and able to leave the hospital sooner.
This study does not eliminate the need for postoperative swallowing studies. A swallow study still has a role in known complicated cases, such as subsequent gastric bypass procedures and intraoperative known complicated cases.9 Another key indication for a swallowing study would be for surgeons and institutions just beginning to perform bariatric surgery. For example, this may have a role in the first 50 cases performed by the surgeon.
In conclusion, gastric bypass surgery is a safe and effective treatment for morbid obesity. Postoperative complications after Roux-en-Y gastric bypass surgery are predictable when based on the patient's symptoms. We believe routine postoperative fluoroscopic upper gastrointestinal series may be unnecessary in asymptomatic patients and uncomplicated cases. Further prospective trials are warranted to definitively rule out the need for postoperative swallowing studies. However, 2 highly predictive red flags that would warrant a postoperative study include the following. First, patients with a combination of fever, tachycardia, and tachypnea may have an anastomotic leak. Second, nausea with vomiting is suggestive of either GOO or delayed gastric emptying. If neither of these clinical scenarios exists, the patient's diet should be advanced without a postoperative swallowing study.
Correspondence: Stephen Kolakowski Jr, MD, Department of Surgery, Pennsylvania Hospital, 301 S Eighth St, Floor 4, Philadelphia, PA 19106-4000 (Skolakowskijr@yahoo.com).
Accepted for Publication: February 25, 2006.
Author Contributions:Study concept and design: Kolakowski, Kirkland, and Schuricht. Drafting of the manuscript: Kolakowski, Kirkland, and Schuricht. Administrative, technical, and material support: Kolakowski, Kirkland, and Schuricht.
Financial Disclosure: None reported.