Abstract
Abstract
Background:
A need exists to select the most appropriate bariatric operation for a particular patient. One-year data comparing sleeve gastrectomy (SG) to Roux-en-Y gastric bypass (RYGB) and biliopancreatic diversion with duodenal switch (BPD/DS) are sparse.
Methods:
The Bariatric Outcomes Longitudinal Database was queried from June 2007 to September 2011 for 30-day and 1-year adverse events, and 1-year weight loss and comorbidity resolution. Propensity scores with inverse probability weighting were used to match for age, gender, body mass index (BMI), ethnicity, and select comorbidities. Multivariate linear and logistic regressions estimated differences and odds ratios (ORs), respectively, for each pairwise bariatric operation comparison.
Results:
Among 73,702 subjects, 5942 patients underwent SG, 66,324 patients underwent RYGB, and 1436 patients underwent BPD/DS. Compared with SG, decrease in BMI units was greater by 5.3 for BPD/DS and by 2.2 U for RYGB at 1 year. Resolution of gastroesophageal reflux disease (GERD) was best for RYGB (OR = 1.88, 95% confidence interval [CI]: 1.73–2.03) and still good for BPD/DS (OR = 1.57, 95% CI: 1.29–1.90). Hypertension and diabetes mellitus (DM) resolution were better after BPD/DS (OR = 2.12, 95% CI: 1.83–1.64, and OR = 2.53, 95% CI: 2.13–3.00, respectively) and for RYGB were (OR = 1.54, 95% CI: 44–1.64 and OR = 1.63, 95% CI: 1.51–1.75, respectively). Odds of serious adverse events at 1 year were: RYGB, OR = 1.70, 95% CI: 1.45–2.00; BPD/DS, OR = 4.31, 95% CI: 3.06–6.07.
Conclusions:
Using SG as reference, RYGB was associated with highest resolution of GERD, whereas BPD/DS was associated with highest resolution of DM and hypertension. These findings can guide decision making regarding choice of bariatric operation.
Introduction
B
Previous studies of patients undergoing adjustable gastric banding (AGB), SG, and RYGB from Michigan have found that SG was better than AGB for weight loss and had fewer complications with equivalent benefit in resolving comorbidities compared with RYGB. 2 A nationwide study from the Bariatric Outcomes Longitudinal Database (BOLD) compared four of the most commonly performed bariatric operations, including AGB, SG, RYGB, and BPD/DS, and used inverse propensity matching to compare risks and benefits of each of these four operations using AGB as a reference (the operation with the lowest risk profile). This study adjusted for important baseline characteristics so that the risks and benefits of the various operations could be compared in matched patients. Important findings of this study were that compared with AGB, RYGB was associated with highest resolution of gastroesophageal reflux disease (GERD), whereas BPD/DS was associated with highest resolution of diabetes mellitus (DM). However, the risks of AEs following the more complex operations were also higher. 3 More recently, AGB operations are being performed much less frequently despite its ease of insertion and low AE profile, likely because of inadequate resolution of comorbidities and unsatisfactory weight loss. In the meantime, SG has become the most common operation and its short and mid-term results are promising. The operation has been associated with lower long-term complication rates such as marginal ulcers and internal hernias, and thus has a wide appeal in patients who may be ulcer prone or in whom malabsorption or rerouting of the intestinal flow is not desirable such as transplant patients. On the other hand, some patients may not have adequate weight loss or resolution of comorbid conditions or may have de novo onset of GERD.4–9
To better understand the risk and benefit profile of the current weight loss operations, the purpose of this study was to compare outcomes of RYGB and BPD/DS using SG as a reference.
Methods
Subjects more than 18 years of age, with no previous bariatric operation, and having one of three different bariatric operations SG, RYGB, or BPD/DS were included. Data from the BSCOE Data File from June 2007 to September 2011 that had previously been analyzed and published 3 was reanalyzed by using SG as a reference and excluding AGB.
Baseline variables
Baseline variables included patient demographics (age, sex, and race), select comorbid medical conditions, body mass index (BMI), excess body weight (EBW), substance use, functional status, medications, and American Society of Anesthesiology classification.
Outcomes
Postoperative outcomes included BMI and EBW at 1 year, resolution of select comorbid medical conditions at 1 year, and AEs at 30 days and 1 year. Weight loss was calculated as the difference from baseline to 1-year follow-up for both BMI and EBW.
Comorbidities were treated as binary variables, indicating whether a specific comorbidity was present or absent at baseline and present or resolved at follow-up. Common comorbidities included hypertension, musculoskeletal disease, obstructive sleep apnea syndrome (OSAS), GERD, and DM. Definitions of these comorbidities have been previously published, but the absence of use of medication or device was used to indicate resolution of comorbid conditions. 3
The BSCOE Data File included data on 134 different AEs, and a subset of these AEs were designated as serious AEs (SAEs), such as death, anastomotic leakage, cardiac arrest, venous thromboembolic events, major organ failure, myocardial infarction, and/or bleeding requiring blood transfusion.
Analysis
Continuous demographic and weight variables were summarized with mean and standard deviations and medians and interquartile ranges, whereas categorical demographic and comorbidity variables were summarized with frequencies and percentages. P values for comparisons across operations were calculated with ANOVA or chi-square tests, where appropriate. A two-sided P value <.05 was considered statistically significant.
Given the significant differences in baseline covariates across the three operations, inverse propensity weighting was used to balance the pairwise bariatric operation groups at baseline.
Linear regression models were used to assess a 1-year change in BMI and EBW, and logistic regression models were used to assess resolution of comorbidities and presence of AE and SAE at 1 year. Groups were compared using SG patients as the reference group. A post hoc multiple comparisons matched P value <.008 was considered statistically significant. All analyses were performed in SAS version 9.3.
Results
Characteristics of the cohorts
The data on bariatric surgery patients reported in this study were submitted by 1029 surgeons and 709 hospitals from June 2007 through September 2011. The cohort for weight and comorbidity change at 1 year included 73,702 patients.
Among 73,702 subjects, 5942 patients underwent SG, 66,324 patients underwent RYGB, and 1436 patients underwent BPD/DS. Analyses of baseline characteristics of patients undergoing the three bariatric operations before matching demonstrated that the mean age of patients was equivalent, but more males underwent either SG (24.75%) or BPD/DS (27.79%) compared with RYGB (21.06%) (Table 1). Hispanics were disproportionately less likely (1.6%) to undergo BPD/DS compared with other operations. Also, BPD/DS patients had a higher average BMI at baseline (52.2 kg/m2 [SD: 9.67]) compared with patients seeking SG or RYGB. The BMI of patients seeking RYGB and SG were equivalent (47.9 kg/m2 [SD: 9.58] and 47.8 kg/m2 [SD: 8.13] respectively). Patients with greater comorbidities, such as hypertension, GERD, OSAS, and DM, were more likely to undergo RYGB or BPD/DS compared with SG (Table 1).
BMI, body mass index; BPD/DS, biliopancreatic diversion with duodenal switch; EBW, excess body weight; GERD, gastroesophageal reflux disease; IQR, interquartile range; RYGB, Roux-en-Y gastric bypass; SD, standard deviation; SG, sleeve gastrectomy.
Matched weight change at 1 year
In the matched cohorts, BPD/DS had the greatest BMI change at 1 year, followed by RYGB and SG, respectively (Table 2). Compared with patients undergoing SG, RYGB had a 2.2 U reduction in BMI; and BPD/DS had a 5.3 U reduction. The EBW results reflected the relative reductions in BMI at 1 year, with BPD/DS experiencing the greatest reduction in EBW compared with RYGB or SG patients.
BPD/DS, biliopancreatic diversion with duodenal switch; RYGB, Roux-en-Y gastric bypass; SD, standard deviation; SG, sleeve gastrectomy.
Matched resolution of comorbid conditions at 1 year
Compared with patients undergoing SG (Table 3), patients undergoing the other two operations were more likely to have their hypertension remit with the greatest odds for BPD/DS (odds ratio [OR] = 2.12, 95% confidence interval [CI]: 1.83–1.64), and RYGB (OR = 1.54, 95% CI: 44–1.64). Patients were more likely to have their OSAS remit at 1 year after undergoing BPD/DS (OR = 1.76, 95% CI: 1.46–2.13) or RYGB (OR = 1.46, 95% CI: 1.34–1.58). GERD was more likely to remit for patients undergoing RYGB (OR = 1.88, 95% CI: 1.73–2.03) than patients undergoing SG, and even BPD/DS had better odds of GERD remission compared with SG (OR = 1.57, 95% CI: 1.29–1.90). Finally, BPD/DS was best for remission of type 2 DM (OR = 2.53, 95% CI: 2.13–3.0) and for RYGB (OR = 1.63, 95% CI: 1.51–1.75).
P < .0001.
BPD/DS, biliopancreatic diversion with duodenal switch; CI, confidence interval; DM, diabetes mellitus; GERD, gastroesophageal reflux disease; MSD, musculoskeletal disease; OR, odds ratio; OSAS, obstructive sleep apnea syndrome; RYGB, Roux-en-Y gastric bypass; SG, sleeve gastrectomy.
Unmatched AEs and SAEs at 30 days and 1 year
AEs and SAEs were recorded at both 30 days and 1 year. At 30 days, unmatched AEs were lowest for SG at 8.04% and increased progressively to 11.77% for RYGB and 20.26% for BPD/DS. At 1 year, the AE rates were 10.03% for SG, 17.99% for RYGB, and 27.52% for BPD/DS. SAE rates for SG at 30 days were 0.81% for SG, 1.35% for RYGB, and 3.63% for BPD/DS. At 1 year, SAE rates increased slightly to 0.93% for SG, 1.58% for RYGB, and 4.60% for BPD/DS.
Rates of bleeding at 30 days were 0.63% for SG, 1.38% for RYGB, and 0.99% for BPD/DS. Leaks were 0.14% for SG, 0.36% for RYGB, and 0.89% for BPD/DS. Pulmonary embolism (PE) was 0.11% for SG, 0.13 for RYGB, and 0.54% for BPD/DS. At 1 year, bleeding was slightly higher at 0.67% for SG, 1.46% for RYGB, and 1.00% for BPD/DS. Similarly, leak rates were slightly higher at 0.24% for SG, 0.43% for RYGB, and 1.18% for BPD/DS. PE rate remained unchanged for all operations at 1 year and were 0.11% for SG, 0.14% for RYGB, and 0.74% for BPD/DS (Tables 4 and 5).
AE includes any unexpected effect; SAE includes death, cardiac arrest, heart failure, and/or pulmonary edema, myocardial infarction, multisystem organ failure, sepsis from anastomotic leak, sepsis from other abdominal source, systemic inflammatory response syndrome, stroke/cerebrovascular accident, pneumothorax, PE, respiratory failure, renal failure, anastomotic leakage, evisceration, intra-abdominal bleeding/hemorrhage, and deep venous thrombosis.
AE, adverse event; BPD/DS, biliopancreatic diversion with duodenal switch; PE, pulmonary embolism; RYGB, Roux-en-Y gastric bypass; SAE, serious adverse event; SG, sleeve gastrectomy.
AE includes any unexpected effect; SAE includes death, cardiac arrest, heart failure, and/or pulmonary edema, myocardial infarction, multisystem organ failure, sepsis from anastomotic leak, sepsis from other abdominal source, systemic inflammatory response syndrome, stroke/cerebrovascular accident, pneumothorax, PE, respiratory failure, renal failure, anastomotic leakage, evisceration, intra-abdominal bleeding/hemorrhage, and deep venous thrombosis.
P < .0001.
AE, adverse event; BPD/DS, biliopancreatic diversion with duodenal switch; CI, confidence interval; OR, odds ratio; PE, pulmonary embolism; RYGB, Roux-en-Y gastric bypass; SAE, serious adverse event; SG, sleeve gastrectomy.
Matched AEs and severe AEs at 30 days and 1 year
AE and SAE rates for SG, RYGB, and BPD/DS after matching for baseline characteristics are shown in Table 5. Using SG as the reference odds of any AE at 30 days were 1.51 higher after RYGB (95% CI: 1.43–1.59), and 2.97 higher after BPD/DS (95% CI: 2.60–3.4). Matched odds of SAEs at 30 days were 1.7 higher after RYGB (95% CI: 1.45–2.00) than after SG, and 4.31 higher after BPD/DS (95% CI: 3.06–6.07). Matched odds of bleeds were 2.17 higher after RYGB (95% CI: 1.79–2.61) and 2.06 higher after BPD/DS (95% CI: 1.24–3.44). Matched odds of PE at 30 days was not significantly higher after RYGB (OR 1.18, 95% CI: 0.74–1.89) than after SG, but was 3.51 higher after BPD/DS (95% CI: 1.21–10.17).
At 1 year (Table 5), the matched odds of having an AE compared with SG were 1.93 higher for RYGB (95% CI: 1.84–2.02) and 3.51 higher for BPD/DS (95% CI: 3.11–3.96). The odds of SAE were 1.73 for RYGB (95% CI: 1.49–2.01) and 4.79 for BPD/DS (95% CI: 3.51–6.55).
Discussion
This study examines the outcomes of RYGB and DS using the SG as a reference operation from a large national cohort of patients. We found that at baseline, BPD/DS was associated with higher mean BMI and comorbidity burden compared with both the RYGB and the SG. BPD/DS had an average BMI reduction of 5.3 U, and RYGB 2.2 U compared with SG in matched patients. If degree of weight loss is the primary determinant for obtaining a bariatric operation, this information is helpful to surgeons and patients by quantifying expected average weight loss after each of the three bariatric operations.
With regard to GERD, RYGB had the best resolution, with odds of 1.88 times higher compared with SG. The odds of GERD resolution were 1.57 times higher even after BPD/DS and may be related to a larger pouch and diversion of acid and bile further downstream. These findings are consistent with single-institution studies in the literature that show a postoperative rate of GERD of up to 22% after SG.4–8
We also found that type 2 DM was best resolved by BPD/DS, and RYGB was superior to SG. 3 This has been previously demonstrated and could be explained on the basis of stronger incretin effect from more distal delivery of nutrients in the gut for RYGB and BPD/DS. Other authors have compared RYGB to SG and shown similar results.10,11 The chances of resolving sleep apnea were equivalent after either RYGB or BPD/DS, and were 1.46 and 1.76 times higher compared with SG. Comparing BPD/DS to RYGB difference in rates of resolution was not significant (OR = 1.13, 95% CI: 0.9–1.34).
Although patient weight and comorbid conditions at baseline are important data, there are several other considerations that effect the choice of an operation. These include facility, surgeon, and patient factors. Each operation has its unique set of benefits and management issues. SG is technically less demanding and can be performed expeditiously. Recovery is rapid and therefore it is more suited for facilities with limited resources. BPD/DS on the other hand is more challenging technically and requires surgeon expertise with intracorporeal suturing. Also, due to its potential for malabsorption of vitamins, BPD/DS requires a compliant patient and additional expenditure on special vitamins. Serum levels of these vitamins also need to be periodically monitored. While the BPD/DS is usually associated with more frequent bowel movements (about 2–3 a day is average) and more foul flatus, 12 the RYGB on the other hand is associated with more dumping and marginal ulceration. These side effects of an operation also need to be considered when selecting a particular operation. Hence, in patients who are on chronic nonsteroidal anti-inflammatory agents for severe arthritis, the RYGB, which is prone to developing marginal ulcers, would not be a good choice and SG may be more suitable. In the SG, no bowel is bypassed and therefore absorption of nutrients and medications is more predictable. These factors are important considerations for patients with end-organ failure, inflammatory bowel disease, or other conditions where bowel bypass may not be desirable, such as significant adhesions from previous operations. In summary, empiric data as well as common sense and practical limitations need to be considered in selecting a particular operation for a patient.
Conclusions
Our data suggest that RYGB is best for GERD and BPD/DS is best for weight loss, DM, hypertension, and OSAS. SG has the lowest AE rates. In addition, there are common sense and practical limitations that need to be considered in selecting a particular bariatric operation for a patient.
Footnotes
Disclosure Statement
No competing financial interests exist.