Long-term Survival After Sleeve Gastrectomy Versus Gastric Bypass in a Binational Cohort Study

Obesity, defined as BMI >30 kg/m², is an increasing global health problem that currently affects 30–40% of adults in the Western world (1). Obesity is associated with comorbidities, mainly diabetes, hypertension, cardiovascular diseases, cancer, and psychiatric disorders, as well as reduced overall survival (2,3). In severely obese individuals (BMI ≥35 kg/m²), bariatric surgery, and specifically gastric bypass, may induce substantial and long-lasting weight loss, resolve obesity-related comorbidities, and prolong life expectancy (4–11). In gastric bypass, the distal stomach and duodenum are bypassed, resulting in malabsorption of nutrients, and a fashioned gastric pouch restricts food intake. This procedure has been the predominant bariatric surgical method during the last few decades but carries a risk of serious adverse events (12). Sleeve gastrectomy, where the greater curvature of the stomach is resected, is a less extensive bariatric procedure that has gained increased popularity during recent years due to its simplicity and lower risk of adverse events (12). However, the weight loss is less than after gastric bypass (13,14), which implies that obesity-related comorbidities and all-cause mortality might not decrease to the same extent as after gastric bypass.

This study aimed to clarify potential differences in long-term overall survival after sleeve gastrectomy compared with gastric bypass in a cohort study from two Nordic countries (Sweden and Finland) with similar health care systems and data resources and to also assess the risk of mortality in the most lethal obesity-associated comorbidities (i.e., cardiovascular diseases and cancer).

This was a population-based and binationwide cohort study in Sweden and Finland from 2007 to 2020. The study retrospectively analyzed prospectively collected data. Eligible for the cohort were all adult residents (≥18 years) who underwent primary laparoscopic sleeve gastrectomy or gastric bypass for obesity any time between 1 January 2007 and 31 December 2019 in Sweden, and between 1 January 2007 and 31 December 2018 in Finland. Follow-up ended on 31 December 2020 in Sweden and on 31 December 2018 in Finland. The study population was retrieved from an updated version of the Swedish and Finnish part of the Nordic Obesity Surgery Cohort (NordOSCo), which has been described in detail previously (4,15). In short, NordOSCo contains data from the national patient registries, prescribed drug registries, and causes of death registries in Sweden and Finland. These registries are continuously updated with prospectively collected data from all health care institutions and pharmacies in Sweden and Finland as part of routine health care and are mandated by law. Because of the nationwide design, all individuals in Sweden and Finland were potentially eligible for the study, and no sampling was performed. Thus, no sample size calculations were performed. The Swedish and Finnish national health care registries have been validated extensively, with excellent results for research purposes (16–20). Personal identity numbers are used by all residents and recorded in the registries, and these numbers enabled accurate linkages and merging of information for each individual. The personal identity system also allowed complete follow-up of all participants in population registries until death (also abroad), thus avoiding any losses to follow-up. The study was approved by the Regional Ethical Review Board in Stockholm, Sweden (registration number 2019-05828) and by the Finnish National Institute for Health and Welfare, Statistics Finland, and the Population Registry Center (15). Informed consent was waived given the registry-based design, and all data were analyzed and presented anonymously.

Swedish and Finnish guidelines recommend bariatric surgery only for patients with a BMI >35 kg/m² with failed serious attempts at conventional weight loss. The study compared primary laparoscopic sleeve gastrectomy (defined by the operation codes JDF97 or JDF01, according to the Nordic Medico-Statistical Committee Classification of Surgical Procedures) with primary laparoscopic gastric bypass (JDF11). Data regarding these procedures came from the national patient registries available in both Sweden and Finland. These registries have been validated for high completeness and accuracy compared with medical records data (16,20) and have been specifically validated for bariatric surgery, showing 97% concordance to medical records in 938 patients (21).

The main outcome was all-cause mortality, defined as death from any cause occurring any time after the bariatric surgery. There were also two secondary outcomes: cardiovascular-specific mortality and cancer-specific mortality. All outcomes were identified in the national cause of death registries, which prospectively collect information on the date and causes of death as determined by a physician. Cardiovascular-specific mortality was defined by any of the diagnosis codes I00-I99, thus including all deaths from cardiovascular and cerebrovascular diseases in ICD-10 causing the death (full details are given in Supplementary Table 1). The corresponding diagnosis codes for cancer-specific mortality were C00–C97, which included death from all malignant tumors (Supplementary Table 1). The cause of death registries have been validated with 100% completeness for all-cause mortality and >96% completeness for disease-specific mortality in both countries (19,20).

Seven variables were considered potential confounders or effect modifiers: age, sex, diabetes, hypertension, Charlson comorbidity index score, country, and calendar year. Data on age, sex, Charlson comorbidity index score, country, and calendar year were identified from the national patient registries. The Charlson comorbidity index score did not include diabetes in this study, as diabetes was considered separately in the analyses, but was otherwise defined based on the most updated and validated version of this index (Supplementary Table 2) (22). In order to capture diabetes and hypertension well, also in patients managed by general practitioners only (not covered by the patient registries), these conditions were defined by the prescription and dispensation of any antihyperglycemic and antihypertensive drugs, respectively, within 1 year before the bariatric surgery as identified in the national prescribed drug registries. Antihyperglycemic drugs included insulin, insulin analogs, and other antihyperglycemics according to the corresponding Anatomical Therapeutical Chemical (ATC) codes (A10A and A10B). Antihypertensive drugs included thiazides, β-blockers, calcium channel blockers, ACE inhibitors, and angiotensin II receptor blockers (ATC codes: C03A, C03EA01, C07AB, C07AG, C08, and C09A-D). All covariates were indexed upon study entry (i.e., the date of bariatric surgery).

The patients were followed up from the date of sleeve gastrectomy or gastric bypass and ended at date of death, surgical conversion from sleeve gastrectomy to gastric bypass or vice versa, or end of the study period, whichever occurred first. Cox proportional hazards regression provided crude and adjusted hazard ratios (HRs) with 95% CIs. Multivariable modeling was used to adjust the HRs for age (continuous), sex (male or female), diabetes (no or yes), hypertension (no or yes), Charlson comorbidity index score (0 or ≥1), country (Sweden or Finland), and calendar year (continuous). The proportionality assumption of the hazards was verified by calculating the Schoenfeld residuals and was met for all models. Analysis of the main outcome all-cause mortality was also performed by introducing an interaction term between each of the seven covariates and the exposure; for each model, the interaction term was tested for statistical significance, and the effect in each strata of the covariates was also computed. A senior biostatistician (G.S.) conducted the statistical analyses according to a detailed study protocol drafted after data collection, but before data management, using Stata/MP 15.1 statistical software.

Of 61,503 included patients, 9,612 (15.6%) underwent sleeve gastrectomy and 51,891 (84.4%) received gastric bypass (Table 1). A total of 1,571 deaths (2.6%) occurred during follow-up of 415,712 person-years (mean 6.8 person-years). The majority of patients (75.4%) were women and lived in Sweden (85.8%), and the median age at the time for surgery was 42 years. A substantial proportion of the patients had diabetes (17.2%) or hypertension (33.1%), while most patients (70.3%) had a Charlson comorbidity index score of 0 (not counting diabetes and hypertension) (Supplementary Table 3). Sleeve gastrectomy patients and gastric bypass patients had comparable smoking frequency and mean BMI according to data for the Swedish part of the cohort. Most patient characteristics were similar comparing the sleeve gastrectomy and gastric bypass groups, but sleeve gastrectomy was more often performed during more recent calendar years (Fig. 1). During follow-up, 203 sleeve gastrectomy patients (2.1%) and 420 gastric bypass patients (0.8%) underwent bariatric surgery revision. The 90-day all-cause mortality was 0.04% (n = 4) after sleeve gastrectomy and 0.06% (n = 29) after gastric bypass.

The cumulative probability of survival did not reveal any major differences between the sleeve gastrectomy and gastric bypass groups (Fig. 2). In multivariable analyses, the adjusted risk of all-cause mortality during the entire study period was similar among patients who underwent sleeve gastrectomy compared with gastric bypass (HR 0.98, 95% CI 0.81–1.20) (Table 2). The interaction term analysis showed diabetes was a statistically significant effect modifier for the association between the type of bariatric surgery, with higher risk of all-cause mortality after sleeve gastrectomy compared with gastric bypass (HR 1.54, 95% CI 1.06–2.24). In stratified analyses, diabetes status suggested potentially decreased mortality after sleeve gastrectomy versus gastric bypass in patients without diabetes (HR 0.82, 95% CI 0.63–1.06) and increased mortality in patients with diabetes (HR 1.26, 95% CI 0.95–1.68). The interaction term analyses showed no statistical significance of the other six covariates (data not shown). The all-cause mortality was statistically significantly decreased after sleeve gastrectomy from the year 2014 onward (HR 0.72, 95% CI 0.54–0.97) and tended to be increased during the earlier period (HR 1.25, 95% CI 0.96–1.63 before year 2014). Other stratified analyses did not show any statistically significant differences in the risk of all-cause mortality comparing sleeve gastrectomy with gastric bypass (Table 2).

Cardiovascular diseases (n = 311 [19.8%]) and malignant tumors (n = 304 [19.4%]) were the most-common causes of death among the study participants (Supplementary Table 4). Death due to poisoning by drugs or ethanol (n = 277 [17.6%]) was the third most-common cause of death, which was overrepresented after gastric bypass (n = 267 [18.4%]) compared with sleeve gastrectomy (n = 10 [8.2%]). In multivariable analysis, sleeve gastrectomy was not associated with cardiovascular-specific mortality (HR 1.16, 95% CI 0.82–1.63) or cancer-specific mortality (adjusted HR 1.30, 95% CI 0.88–1.93).

The all-cause mortality was similar after primary laparoscopic sleeve gastrectomy and gastric bypass during the full study period but was lower after sleeve gastrectomy from the year 2014 onward. Interaction term analysis showed all-cause mortality was decreased after sleeve gastrectomy in patients without diabetes but increased in those with diabetes. There were no statistically significant differences between sleeve gastrectomy and gastric bypass in relation to cardiovascular-specific and cancer-specific mortality.

Methodological strengths include the population-based design and the use of prospectively collected data from complete and well-validated national databases in two entire countries, combined with a long and complete follow-up. These strengths ensured statistically robust and internally valid results and facilitated generalizability. An inevitable source of error with an observational study is residual or unknown confounding. The results were therefore adjusted for several relevant covariates, which should counteract confounding, but other potentially relevant factors (e.g., changes in BMI, tobacco smoking, and alcohol consumption) were not adjusted for. However, the distribution of BMI and smoking were similar in the sleeve gastrectomy group and the gastric bypass group (in the Swedish data set), changes in BMI after the bariatric surgery would be a mediator rather than a confounder, and heavy drinkers were captured through the Charlson comorbidity index, which includes alcohol-related diseases. Thus, none of these three factors should much influence the findings. Sleeve gastrectomy gained popularity in more recent calendar years, which led to a shorter follow-up time in this group compared with gastric bypass. However, we adjusted all analyses for calendar year of surgery and, therefore, indirectly for follow-up time. Finally, although the large cohort size gave allowed subgroup analyses, the stratified analyses had limited power because of the low death rates in this cohort, consisting mainly of younger adults.

Few observational studies have directly compared sleeve gastrectomy and gastric bypass in relation to long-term survival, and these have generally been underpowered. An exception was a recent multicenter cohort study from the U.S. of 33,560 patients that, similar to the current study, did not find any difference in 5-year mortality comparing sleeve gastrectomy and gastric bypass (HR 0.94, 95% CI 0.73–1.23) (23). However, that study covered only parts of all bariatric procedures in the study population, included no subgroup analyses, and had a short follow-up (up to 5 years) (23). In contrast, the current study had a population-based covering of almost all bariatric surgery, included several subgroup analyses, and had a larger sample size and a longer follow-up (up to 14 years). A French nationwide cohort study with 7-year follow-up found that gastric bypass and sleeve gastrectomy were both associated with decreased mortality compared with matched nonoperated-on control subjects, but the two methods were not directly compared (24). While the main finding of the study was that sleeve gastrectomy performed equally as well as gastric bypass regarding overall all-cause mortality, stratified analyses indicated that sleeve gastrectomy was followed by superior survival in more recent calendar years. This may be partly explained by the strongly increased annual volume of sleeve gastrectomy during more recent years, while gastric bypass has become less frequent, and as of today, sleeve gastrectomy is the most commonly performed bariatric procedure in Sweden. Sleeve gastrectomy is a relatively simple procedure and should not have a significant learning curve for bariatric surgeons already proficient in gastric bypass. The improved survival after sleeve gastrectomy observed in more recent calendar years may rather be a result of more standardized surgery and postoperative care due to the increased annual volumes in many centers (25).

Previous studies with long-term follow-up have shown that the weight loss is substantial, persists for at least 10 years after gastric bypass, and is followed by high rates of recovery from comorbidities, although some of these studies had >20% of losses to follow-up (26–30). Three randomized clinical trials have compared sleeve gastrectomy with gastric bypass in relation to weight loss or comorbidities. In a Finnish trial (Laparoscopic Gastric Bypass Versus Sleeve Gastrectomy to Treat Morbid Obesity [SLEEVEPASS] study), the sleeve gastrectomy group (n = 121) displayed less weight loss and less resolution of hypertension than the gastric bypass group (n = 119) 5 years after surgery (13), in a Swiss trial (Swiss Multicentre Bypass or Sleeve Study [SM-BOSS] study), sleeve gastrectomy (n = 107) and gastric bypass (n = 110) were comparable regarding weight loss and resolution of comorbidities at 5 postoperative years (31), and a Norwegian trial found the gastric bypass group (n = 54) had better remission of diabetes compared with the sleeve gastrectomy group (n = 55) 1 year after surgery (32). Taken together, gastric bypass may provide superior weight loss and thus better resolution of diabetes than sleeve gastrectomy. It is thus possible that the statistically significant interaction between diabetes and all-cause mortality, showing better survival in patients with diabetes after gastric bypass in the current study is mediated by a more pronounced weight loss. Similarly, the possible slightly higher (although statistically nonsignificant) cardiovascular- and cancer-related mortality after sleeve gastrectomy suggested in the current study could be mediated by superior weight loss after gastric bypass.

Randomized clinical trials of the long-term survival in patients who undergo sleeve gastrectomy compared with gastric bypass are not feasible considering the required sample size and length of follow-up. The current study may therefore provide the best available guidance when considering the type of bariatric surgery. This is the first study showing that the long-term survival following sleeve gastrectomy is equal to or even superior to gastric bypass in recent years, which indicates that sleeve gastrectomy can be recommended as well as gastric bypass for patients selected for bariatric surgery. The interaction analyses suggested better outcomes with gastric bypass in patients with diabetes and with sleeve gastrectomy in patients without diabetes. Thus, sleeve gastrectomy may be the primary procedure in patients without diabetes, whereas in patients with diabetes, gastric bypass may instead maximize life expectancy.

In conclusion, this two-country population-based cohort study with a long and complete follow-up found a similar survival after sleeve gastrectomy and gastric bypass, and the survival after sleeve gastrectomy was even superior in a more recent calendar period. This suggests that sleeve gastrectomy may be considered an established bariatric procedure. The findings also indicate a role for a tailored surgical approach in relation to patients’ diabetes status, where sleeve gastrectomy may be recommended for patients without diabetes, while gastric bypass may offer better long-term survival in patients with diabetes.

Funding. The study was funded by Vetenskapsrådet (The Swedish Research Council) (2019-00209) and Svenska Läkaresällskapet (Swedish Society of Medicine).

The funding organizations had no role in the design of the study, collection, management, analysis, and interpretation of the data, preparation, review, or approval of the manuscript, and decision to submit the manuscript for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. D.H. and J.L. researched data and wrote the manuscript. G.S. and J.H.K. researched data and reviewed/edited the manuscript. S.R.M. reviewed/edited the manuscript. D.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis.