Defining Global Benchmarks in Bariatric Surgery

A Retrospective Multicenter Analysis of Minimally Invasive Roux-en-Y Gastric Bypass and Sleeve Gastrectomy

Daniel Gero, MD; Dimitri A. Raptis, MD, MSc, PhD; Wouter Vleeschouwers, MD; Sophie L. van Veldhuisen, MD; Andres San Martin, MD; Yao Xiao, MD; Manoela Galvao, MD; Marcoandrea Giorgi, MD; Marine Benois, MD; Felipe Espinoza, MD; Marianne Hollyman, MD, PhD; Aaron Lloyd, MPH; Hanna Hosa, MD; Henner Schmidt, MD; José Luis Garcia-Galocha, MD; Simon van de Vrande, MD; Sonja Chiappetta, MD; Emanuele Lo Menzo, MD; Cristina Mamédio Aboud, RN, MSc; Sandra Gagliardo Lüthy; Philippa Orchard, MD; Steffi Rothe, MBA; Gerhard Prager, MD; Dimitri J. Pournaras, MD, PhD; Ricardo Cohen, MD; Raul Rosenthal, MD; Rudolf Weiner, MD; Jacques Himpens, MD, PhD; Antonio Torres, MD, PhD; Kelvin Higa, MD; Richard Welbourn, MD; Marcos Berry, MD; Camilo Boza, MD; Antonio Iannelli, MD; Sivamainthan Vithiananthan, MD; Almino Ramos, MD; Torsten Olbers, MD, PhD; Matias Sepúlveda, MD; Eric J. Hazebroek, MD, PhD; Bruno Dillemans, MD; Roxane D. Staiger, MD; Milo A. Puhan, MD, PhD; Ralph Peterli, MD; Marco Bueter, MD, PhD


Annals of Surgery. 2019;270(5):859-867. 

In This Article


This multicenter study established outcome benchmarks for the 2 most frequently performed bariatric procedures by applying a recently developed standardized methodology.[2] In the current report of low-risk BS patients operated in 1 of 19 high-volume referral centers located on 3 continents, main findings were a zero 90-day mortality rate, a low early postoperative morbidity rate with the majority of reinterventions occurring after the first 90-days. The cohort's 1-year percentage weight loss was comparable to the mean procedure-specific pooled outcomes published in the 2018 IFSO global registry report.[43]

Identified outcome benchmarks may serve as a reference for bariatric centers to compare their own outcomes in similarly low risk or even all bariatric patients and to take action when eventual performance gaps are identified. So far, these attempts were impeded by great variability in case mix between centers.[17,43,44] The approach is reminiscent to the concept of propensity scoring study participants, where randomization at baseline is mimicked by cohorts that are comparable on main measured covariates.[45] In practice, random allocation of patients to different bariatric centers is not feasible; therefore, the concept of establishing global benchmarks based on multicentric data of low-risk patients seems to be an appealing alternative to allow comparison of outcomes and thus, of surgical quality.

This study aimed to represent the "real world" by including European, Northern and Southern American centers led by recognized experts in the field of BS. All centers had sufficient caseload, a prospective bariatric database and previous publication(s) on surgical outcomes. The selection of benchmark patients was performed by a strict and stepwise risk stratification aiming to identify the "healthiest" BS candidates with the least expected complications. Each submitted case was read by one of the principal investigators in Zurich, and clarification was requested from the coinvestigators in case of incompletely submitted case report forms. Our protocol focused on CD >II events, given that medically treated complications of BS are frequently managed by nonsurgeon healthcare providers outside of bariatric centers and thus, do not obviously appear in institutional databases.[46]

All included patients were operated in academic centers with teaching assignments, and consequently, with potentially increased operation duration.[47] This may explain why established outcome benchmarks for operation times in our study were not shorter, but in a similar range as those reported for all laparoscopic RYGB and SG (including high-risk patients and redo surgery) performed in the USA in 2015 to 2016.[45]

Length of stay is not a pure quality indicator, since it highly depends on health care systems and at some extent, on the attitude of patients. We are witnessing a trend toward shorter inpatient stays following BS. The safety of early discharge on the first postoperative day in a selective group of bariatric patients without significant comorbidities is supported by the 2015 dataset of the MBSAQIP.[48] Furthermore, SG in selected low-risk patients has been recently found to be safely feasible even as an outpatient procedure.[49] Therefore, it is of no surprise that the 2018 IFSO global registry report showed shorter lengths of stay for RYGB (2 vs 4 d) and a similar one for SG (3 d) in comparison with the benchmarks identified in the current study, which were derived from a cohort operated between 2012–2017.[43]

The study revealed an interesting pattern in postoperative morbidity, as most reoperations and reinterventions occurred beyond 90 days. This is in line with postbariatric readmission rates observed in The Danish National Health Surveys,[50] but it is in contrast to previous reports on the temporal occurrence of reinterventions after other types of major abdominal surgeries (ie, hepatectomy, esophagectomy), where the vast majority of CD grade >II events were recorded within 30 days.[6,7] Several explanations to this finding may be possible. First, both BS patients and bariatric centers are committed to perform a lifelong follow-up, which may not be the case for other types of surgeries. Second, severe obesity is a chronic disease, which is often characterized by cyclic episodes of weight loss and weight regain, as well as a higher risk for the development of a series of associated conditions that may require surgical care (ie, gallstone disease, GERD).[51] In this study, symptomatic cholelithiasis, GERD, and weight regain were among the most frequent reasons of long-term postbariatric readmissions, although the prevalence of these pathologies may not entirely depend on surgical performance at the index procedure, and increases with length of follow-up. Abdominal pain of unknown origin was the most common reason of readmission after RYGB and the second most frequent one after SG. This is somewhat surprising and may be in part related to the retrospective design of the study. Surgical databases often record the chief complaint of the patients' at presentation and are not always updated by the definitive diagnosis retained at the end of the often time-consuming work-up.

The concept of establishing benchmarks in BS was validated by complementary approaches. First, we found that compared with the zero 90-day mortality of benchmark patients, the same centers recorded a 0.06% mortality rate following RYGB and SG in nonbenchmark cases. This small difference emphasizes the need for quality indicators focusing on postoperative morbidity. Second, we found that in 1 participating center the overall morbidity in the nonbenchmark cases was above the global benchmark cutoff, mainly due to the higher frequency of CD ≤IIIb complications. Third, the previous outcome reports on consecutive or secondary RYGB published by the participating centers showed a higher 30-day major complication rate than the benchmark cutoff (10.75% vs 5%), highlighting the additional burden of postoperative morbidity observed in "high-risk" cases.

This study has some inherent limitations. First, the quality of current institutional surgical databases seems to be suboptimal for capturing the full spectrum of postoperative morbidity, especially beyond 90 days. This could be improved in the future by external auditing of surgical databases and by replacing self-reporting with automatized data input methods. Ibrahim et al[17] also found a wide variation in postoperative severe complications rates among accredited BS centers in the USA. In the current study, it was not possible to judge whether between-center differences reflected variability in surgical performance or were due to missing data regarding postoperative events. Although benchmark cutoffs were established until 90-days postoperatively in patients with 100% follow-up, we cannot exclude that some complications may have been underreported. Second, to minimize the confounding effect of a potentially hostile intra-abdominal status and of associated procedures on postoperative morbidity, we excluded cases with previous abdominal surgery and with additional nonbariatric procedures performed in combination with the index procedure, including cholecystectomies. This may at least partly explain why symptomatic cholelithiasis ranked among the most frequent postbariatric reasons for readmission.[51]Third, the current methodology of global surgical benchmark establishment is bounded by logistic obstacles, leading to a considerable burden for its future reproductions. The rapid evolution of surgical and endoscopic bariatric procedures, as well as the increasing caseload and experience of referral centers will mandate the regular update of bariatric benchmarks. Ultimately, this process needs to be automatized by development and adaptation of BS registries. Fourth, the case mix in the presented study does not reflect current practices: SG, the dominant operation worldwide,[11] represented only 26% of benchmark cases, thus future studies should aim to update benchmarks for SG with the inclusion of centers with a higher experience with this technique.

In conclusion, we consider this project as an inaugural study introducing the concept of benchmarking in BS. The surgical community's genuine desire to improve patients' postoperative outcomes has a crucial role in increasing penetrance of BS and in decreasing complication-related patient discomfort and healthcare expenditures. The concept of benchmarking is expected to be embedded in surgical quality improvement cycles, and to stimulate the need for comprehensive large databases allowing precise and timely identification of both global benchmarks and institutional outcomes.