Bariatric surgery and death rates: the Swedish Obese Subjects study

Why this study matters

Weight loss improves many risk factors, such as blood pressure and blood sugar. But before this study, no prospective study had tested whether an intervention that causes weight loss also lowers the overall risk of death. Some observational studies had even suggested that losing weight was linked to a higher risk of dying. This report from the Swedish Obese Subjects (SOS) study was the first to look at overall deaths after bariatric surgery, which is an operation on the stomach or gut to help with weight loss. Canadian readers weighing long-term options may find it useful to know what this landmark study did and did not show.

Who was in the study

The study enrolled 4047 adults living with severe obesity in Sweden. Of these, 2010 had bariatric surgery and 2037 received conventional care. Follow-up for deaths averaged 10.9 years. When the analysis was done (November 1, 2005), the researchers knew whether all but three people were alive or dead, a follow-up rate of 99.9%. The published summary does not give the age range.

What the researchers did

This was a prospective, controlled cohort study. "Prospective" means people were enrolled first and then followed forward in time. "Controlled" means there was a comparison group. It was not randomized: a computer did not assign people to surgery or conventional care by chance. Instead, each person who chose surgery was matched to a similar person who received conventional treatment. Surgery types were gastric bypass, vertical-banded gastroplasty, or banding. The summary does not describe the conventional care in detail. Weights were recorded for up to 15 years.

What they found

The main result was the number of deaths. There were 129 deaths in the conventional care group and 101 in the surgery group. The unadjusted hazard ratio was 0.76 (P=0.04). After adjusting for sex, age, and risk factors, the hazard ratio was 0.71 (P=0.01) (R32-C94). A hazard ratio compares the rate of an event in two groups over time; a value below 1 means the event happened less often in the surgery group. The "P" value tells you how likely a result this size would be by chance alone; values below 0.05 are usually called statistically significant. The summary we reviewed does not report a confidence interval for the hazard ratio.

The most common causes of death were heart attack (25 in the control group, 13 in the surgery group) and cancer (47 in the control group, 29 in the surgery group).

Weight results differed by type of surgery. Weight loss peaked after 1 to 2 years and then partly returned before settling:

In the conventional care group, average weight changed by less than 2% in either direction over the up to 15 years that weights were recorded.

Side effects and people who stopped

The published summary we reviewed does not report surgical complications, side effects, or how many people left the study for reasons other than death. It does state that vital status was known for 99.9% of participants.

What this study does not tell you

• It shows an association, not proof of cause. Because people were not randomized, those who chose surgery may have been healthier or different in ways that the matching and adjustment did not capture.
• The operations used (gastric bypass, vertical-banded gastroplasty and banding) were done decades ago and differ from the procedures most common today. The result may not apply in the same way to current surgery.
• The summary does not describe conventional care, and the study predates modern weight-management medicines, so it does not compare surgery with those medicines.
• No confidence interval was given in the summary, so the range of uncertainty around the 0.71 hazard ratio is not shown here.
• The summary we reviewed does not state who funded the study.
• Later SOS reports on heart events and life expectancy use the same people. They are not independent confirmation.

What it means in Canada

This study is often cited when Canadian clinicians and guidelines discuss the long-term effects of bariatric surgery. It does not tell you whether surgery is right for any one person, and it says nothing about how surgery compares with the medicines used today. If you are reading about medicines, see semaglutide in Canada and tirzepatide in Canada. For questions about what provincial plans pay for, see public drug coverage by province.

Related reading

Other studies in this library that bear on the same question.

Source