Introduction

The incidence of obesity among children, adolescents, and adults is rapidly rising in both developed and developing countries [1, 2]. By 2010, overweight and obesity were estimated to cause 3.4 million deaths, 3.9 % of years of life lost, and 3.8 % of disability-adjusted life-years worldwide [3]. In addition, obesity increases the risk of hypertension, dyslipidemia, diabetes, coronary heart disease, stroke, gallbladder disease, osteoarthritis, sleep apnea, and cancer [4, 5]. Bariatric surgery has shown to be effective in reducing and maintaining weight loss, with potential benefits in improving cardiovascular risk factors [6, 7]. Current guidelines recommend that all patients undergo upper gastrointestinal endoscopy (UGE) before bariatric surgery [8]. Several studies have shown that preoperative UGE can identify a wide range of abnormal findings, including hiatal hernia, esophagitis, and gastric and duodenal ulcers [9, 10]. Others have questioned the need for routine UGE, stating that most findings are clinically irrelevant and that selective endoscopy in patients with upper gastrointestinal symptoms could reduce the number of exams by over 80 % [11].

Nevertheless, it remains unclear which findings should delay or contraindicate surgery. The aim of this study was to evaluate potential predictors, including abnormal endoscopic findings, of postoperative complications.

Material and Methods

Study Design and Patients

We performed a retrospective analysis of the data of 613 patients undergoing UGE in our center from May 2004 to May 2015 as a preoperative evaluation. All patients had an indication for bariatric surgery (body mass index (BMI) > 40 kg/m2 or BMI > 35 kg/m2 and one or more significant obesity-related comorbid conditions). Preoperative exams were performed in a dedicated endoscopy unit by experienced endoscopists or by supervised residents. When deemed necessary, histologic samples were collected and analyzed by our pathology team. In patients with gastric inflammation, biopsies were taken from the antrum and corpus to assess the presence of Helicobacter pylori (Hp). Patients were offered treatment for Hp, but confirmation of eradication was not routinely assessed. Three hundred forty-two patients (55.8 %) underwent surgery (Fig. 1). The type of surgical procedure included gastric sleeve (229), banding (95), and bypass (18). All surgical procedures were performed by dedicated surgeons from our center. Two hundred and one patients did not undergo surgery by the end of the study period. The reasons included discontinuation of follow-up, psychological or medical contraindications, and patient refusal to comply with medical treatment or to undergo surgery. Data relative to demographics, preoperative BMI, endoscopic findings, and postoperative complications were collected and recorded by the authors. Patients lacking information required for the study were excluded.

Fig. 1
figure 1

Flow chart of procedures and results

Full size image

Statistical Analysis

We used a Kolmogorov-Smirnov test to evaluate normal distribution of continuous variables. Continuous variables were expressed in mean ± standard deviation (SD) or interquartile range (IQR), depending on normal or non-normal distribution. Normal distribution continuous variables were compared using Student’s t test for independent samples and Mann-Whitney test for continuous variables with non-normal distribution. Categorical variables were described using frequencies and percentages, and chi-square contingency tables were used to compare proportions between groups. To identify predictors of abnormal UGE, a logistic regression analysis with abnormal endoscopy as dependent variable was used. For analysis of predictors of postoperative complications, we ran several logistic regression analysis testing each complication as a dependent variable. Variables significant in univariate analysis were included in a multivariate analysis. As we did not confirm Hp eradication following treatment, this variable was excluded from the regression. Significance level was chosen at 0.05. Statistical analysis was performed using SPSS v. 21.0.

Results

Preoperative Cohort

We included 613 patients in our analysis, 77.8 % female, with mean age of 46.5 ± 11.6 years and mean BMI of 44.7 ± 4.2 kg/m2. UGE presented at least one abnormal finding in 345 patients (56.3 %). Esophageal, gastric, and duodenal findings were reported in 22.2, 41.8, and 11.4 % of patients, respectively (Table 1). Three hundred twenty-two biopsies from 295 patients (48.1 %) were collected for histopathological examination (Table 2). Most findings consisted of chronic gastritis (82.4 %). Hp infection was present in 48.3 % of gastric biopsies. Histology diagnosed 3 patients (1.0 %) with non-dysplastic Barrett’s esophagus and 1 patient (0.4 %) with low-grade gastric MALT lymphoma (which went in remission following Hp treatment). In a logistic regression analysis (Table 3), Hp was the strongest independent predictor of an abnormal endoscopy (OR 10.304, 95 % CI [4.577–23.195], p <0.001).

Table 1 Findings at upper gastrointestinal endoscopy in the morbidly obese patients. Most findings were mild and benign
Full size table
Table 2 Abnormal findings at histopathological examination of tissue biopsies of endoscopies performed prior to surgery
Full size table
Table 3 Univariate and multivariate analyses for the prediction of abnormal endoscopic findings
Full size table

Postoperative Cohort

Three hundred forty-two patients (55.8 %) underwent bariatric surgery. Forty-three patients (12.6 %) developed a postoperative complication, and 2 patients (0.6 %) died (Fig. 1). In logistic regression analysis (Table 4), the only variable that significantly predicted postoperative complications was gastric ulceration (OR 11.10, 95 % CI [1.80–68.47], p < 0.001). Postoperative fistulas were associated with both gastric (OR 13.29, 95 % CI [2.07–85.24], p = 0.006) and duodenal (OR 19.94, 95 % CI [1.19–333.46], p = 0.037) ulceration. Postoperative sepsis was only predicted by gastric ulceration (OR 10.28, 95 % CI [1.03–102.63], p = 0.047). There was a slight influence of age in the advent of gastrointestinal bleeding (OR 1.10, 95 % CI [1.01–1.20], p = 0.036). We found no predictors of gastric outlet stenosis, band slippage, or death. The type of surgical procedure did not influence the rate of postoperative complications. Of note, all gastric and duodenal ulcers were found in patients with preoperative evidence of infection with Hp.

Table 4 Logistic regression analysis for the prediction of postoperative complications
Full size table

Discussion

The role of UGE before bariatric surgery remains controversial. The reasons supporting this recommendation include high prevalence of endoscopic findings in bariatric patients across studies, fear of missing a significant lesion before surgery (ex. cancer), inability to screen the stomach and duodenum with standard endoscopy after some procedures, and the potential to change the surgical procedure based on some findings (ex. large hiatus hernia). In accordance with our results, abnormal endoscopic findings have been reported in 10 to 90 % of patients across studies (Table 5). Nevertheless, excessive and unnecessary interventions have been brought forth by mild and insignificant endoscopic findings [9, 17, 19]. Zeni [9] performed a partial gastrectomy for a suspicious polypoid lesion later proved to be benign; Azagury [19] submitted 2 patients to a partial gastrectomy for fundic gland polyposis (for which prophylactic gastrectomy in the absence of a history of familial polyposis is not recommended) and delayed another surgery due to a submucosal esophageal lesion (which remained stable after 4.5 years). This had led many authors to limit endoscopy to patients with symptoms of gastrointestinal disease [22]. This may be inadequate, as several studies have reported pathologic endoscopic findings in asymptomatic patients [17, 19]. Other authors have abandoned routine UGE before surgery [14]. There is a paucity of data regarding which endoscopic findings should delay, change, or contraindicate surgery. Ours is one of the first studies addressing this matter. We found gastric and duodenal ulcers to be the only significant predictors of postoperative complications. Most surgeons still advocate crural tightening in patients with a significant hiatal hernia before gastric banding in order to prevent band slippage. Surprisingly, the evidence for this recommendation comes from a single study including 50 patients, in which 4 of the 5 patients with band slippage had a hiatus hernia [26]. The size of the hiatus hernia was not provided by the authors. We found no association between hiatus hernia size and band slippage. The role of Hp infection has been a matter of debate over the years. In some studies, Hp has been associated with an increased risk of postoperative marginal ulcers [17]. This has not been confirmed in other studies [27, 28]. In a study of asymptomatic patients scheduled for RYGB, patients with a positive rapid urease test were more likely to have an abnormal endoscopy than those who tested negative (94 versus 51 %) [21]. Likewise, we found Hp infection to be the most important predictor of an abnormal UGE. Furthermore, all patients presenting with gastroduodenal ulceration in the preoperative UGE had evidence of active Hp infection. As we did not confirm eradication of Hp, we can only hypothesize that these patients might have had incomplete ulcer healing or persisting Hp infection. As stated by other authors, the treatment for most lesions found on UGE includes proton pump inhibitors and/or Hp eradication. This strategy could potentially reduce the costs with endoscopy considerably [19]. Further investigation is required to assess if routine Hp eradication could decrease the rate of postoperative complications and potentially present an alternative to preoperative UGE.

Table 5 Studies addressing gastrointestinal screening before bariatric surgery. Most studies are retrospective and report variable results
Full size table

The main strengths of our study include its fairly large sample size, which represents one of the largest to date assessing bariatric patients. Limitations include its retrospective nature, with all the data retrieved from clinical charts, and the absence of confirmation of Hp eradication, which could have elucidated its role in decreasing postoperative complications. Finally, as this information was not routinely provided in the patient’s clinical charts, we could not assess how many patients had a change in their surgical procedure owing to abnormal endoscopic findings.

Conclusion

Preoperative endoscopy identifies a wide range of abnormal endoscopic findings, but only gastric and duodenal ulcerations were associated with postoperative complications. By reducing gastroduodenal ulceration, Hp eradication might potentially decrease the incidence of postoperative complications.