Skip to content
Main menu hidden.

Image: Elin Berge

Health Outcomes and Resource Utilization for Bariatric Surgery: Real World Evidence from Sweden

Research project Obesity is associated with significant mortality, increasing risks of developing and worsening ranges of comorbidities such as diabetes and cardiovascular disease.

Bariatric surgery (BS) is more effective in reducing weight for morbid obesity, relative to non-surgical interventions. In Sweden, utilization of BS has increased over past 10 years, with gastric bypass (GBP) being the dominant procedure (90% in 2012, 64% 2016), and a recent increase in sleeve gastrectomy (SG) since 2013 (17% in 2012, 34% in 2016).

Head of project

Sun Sun
Research fellow

Project overview

Project period:

2019-01-01 2022-12-31

Participating departments and units at Umeå University

Department of Epidemiology and Global Health

Research area

Public health and health care science

External funding


Project description

However, knowledges regarding comparing health outcomes and resource utilization for BS vs non-surgical treatment, and between different surgical procedures, especial GBP vs SG, are limited. Majority of such comparisons were based on clinical trials, where both sample size and follow-up time were limited (mostly <2 years).

The unique value of Scandinavian Obesity Surgery Registry and possibility of linkage with Patient Registry, Drug registration, Longitudinal Integration Database, Sickness benefit, and Västerbotten Intervention Programme (VIP) data, enables comparisons based on a national population and with follow-ups up to 10 years.

Weight loss might increase physical health, but not necessarily mental health. Better understanding of BS' impact on different health domains is needed. Reliable prediction of surgery outcome of BS is helpful for decision makers; however, the current models were commonly based on study design with small sample size and specific patient groups, resulting the limited external generalization. The richness of Swedish data enables the possibility of construct a reliable, generalizable, robust and validated risk prediction model based on RWE.

Traditionally, regression modelling has been applied to generate risk prediction models. Machine learning algorithms have recently emerged as a viable alternative to conventional regression-based risk prediction, which might offer more objective and complete means for characterising risk.

External funding

Latest update: 2018-11-15