Junkfood Science: Was this really proof that bariatric surgeries save lives?

August 25, 2007

Was this really proof that bariatric surgeries save lives?

Not a single medical professional, scientist or journalist has dared take a critical eye to this study. That fact alone is the best evidence yet of the power of financial interests and bias. What is most disturbing, and should be for everyone who cares about research being used to learn the truth rather than sell us on something, is that this is about life and death.

Media outlets — professional and consumer — all came out with nearly identical stories and interpretations of this study on the same day — having, of course, received the same embargoed press releases providing the script. Great marketing, but it would have benefitted their readers, viewers and patients if one of them had done their jobs. Starting with reading the study.

Instead, we heard: “‘Weight loss’ surgery extends lifespan.”

According to the Pennington Biomedical Research Center press release:

Lars Sjöström, M.D., Ph.D., and Claude Bouchard, Ph.D., of the Pennington Biomedical Research Center, were members of an international research team that tracked more than 4,000 obese individuals for upwards of ten years. About half of the individuals underwent bariatric surgery (the general term for various forms of weight loss surgery), the other half received conventional treatments....Those results are published in the August 23rd issue of The New England Journal of Medicine and, according to the researchers, is the first major report to confirm that weight loss of this magnitude decreases mortality. To the contrary, according to the researchers, some studies had reported in the past that weight loss was associated with increased mortality....

Led by Lars Sjöström, professor at the University of Gothenburg Medical Center and an adjunct and visiting professor at the Pennington Biomedical Research Center, the researchers examined 4,047 obese subjects in a long-term study called the Swedish Obese Subjects (SOS) study. 2,010 underwent bariatric surgery in one of three forms: gastric bypass, vertical-banded gastroplasty or banding while 2,037 subjects received conventional weight loss treatments. The researchers then tracked the weight, behavior and health indicators of the subjects for an average of 11 years.

The average 11-year weight loss of the non-surgery group was less than 2-percent, while the average weight loss among the three surgical groups ranged from 14 to 25-percent. During the study, researchers recorded 129 deaths in the non-surgery group and 101 in the surgery group, most commonly caused by myocardial infarction and cancer. This result indicates a 23-percent reduction in total mortality within the surgical groups.

“We count these results as a milestone in our understanding of the benefits of bariatric surgery for obesity,” Bouchard said. “We are confident in the results and believe this will lead to an acceptance that bariatric surgery is a viable, life-saving option for severely obese patients.”

The glowing, one-sided coverage by the Washington Post was typical of the news around the world. It quoted:

· An editorial by Dr. George Bray, also of the Pennington Center, saying: “The question as to whether intentional weight loss improves life span has been answered. The answer appears to be a resounding yes.”

· Dr. David Flum, of the Department of Surgery, University of Washington, Seattle, and lead author of the largest report on actual mortality rates of bariatric surgeries done in the United States, saying: “This is very exciting...This is huge.”

· Dr. Louis Aronne, director of the Comprehensive Weight Control Center in New York City and co-chair of GlaxoSmithKline’s Reality Council, adding: “Now we have proof. These papers represent a milestone.”

· Kelvin Higa, president of the American Society for Metabolic and Bariatric Surgery, calling it another landmark study.

Missing throughout the media were viewpoints and analyses from professionals without financial interests.

Even this study entitled, “Effects of Bariatric Surgery on Mortality in Swedish Obese Subjects,” was funded by grants from Hoffmann–La Roche, AstraZeneca, Cederroth, and the Swedish Medical Research Council. [See below for the financial disclosures of the authors.]

Medical professionals who’ve followed the inordinate complication rates, the long-term health and nutritional problems, and excessive death rates documented by all objective sources, as well as those who’ve personally watched as fat friends and family die, have good reason to examine this study more closely.

As Junkfood Science regulars know, study designs can be manipulated in countless ways to help lead to whatever conclusions the authors or sponsors set out to "prove." Ideally, such problems are less widespread in clinical trials since people have risked their lives for science, but, sadly, that is not always the case. The tactics, however, can be hard to see and take an especially discerning eye. This study offers a valuable example of things to consider whenever you come across a study that claims to have found something that no other careful study has been able to find. As even the study authors admitted, “our observations are at variance with most other observational studies regarding weight loss.”

Since a comprehensive examination is beyond what most readers have time to read, we’ll look at just a few of the more obvious flaws in this paper. Even this simplified look will show that the actual facts presented in this study are quite different from what we’ve been hearing in the media. So, let’s dig in.

Years of follow-up

We’ll begin with one of the most common misconceptions: years of follow-up. Since this study was examining the long-term effects of bariatric surgery on mortality, it’s understandable you might think that the average 10.9 years of follow-up reported meant the patients were followed for 10.9 years after having surgery. Here is the first example of not reading the study. The follow-up period — which actually ranged from 4 to 18 years — began from the “matching date.” That was the date that the study recruitment campaign ended and the surgical candidates were chosen and matched with controls. This was about 1 1/2 years before the surgeries were actually performed. So, we’re down to just over an average 9 years of post-surgical follow-up.

While the authors spent pages describing weight loss year-by-year through 15 years (which we’ll see in a moment is useless data), when it came to reporting mortality, they lumped all of the causes of deaths together. There were holes in the data reported. They provided no detailed breakdown information on how the people died over the years. The greatest number of deaths occurred after ten years and the number of deaths rose most significantly the further out from surgery the patients were. Yet most of the available mortality data and patients in this study were newer post-ops still in the first 5-8 year honeymoon period, ameliorating the overall mortality rates and the worsening long-term outlooks. There were fewer older surgical patients in this study — only 760 patients were 12 years post-op, only 422 were 14 years out, and only 169 were 16 years out.

Types of surgeries

Not all bariatric surgeries are the same or have the same rates of complications and deaths. The SOS study began in 1987 and the patients underwent three types of surgeries. Most (68%) had vertical banded gastroplasty (VBG). Another 18% had banding and only 13% had gastric bypass. This is significant for several reasons.

This study involved a mix of procedures that's not remotely reflective of the surgeries performed today, especially in the United States. Among 16,155 bariatric procedures performed in the United States from 1997 to 2002, for example, 81% were gastric bypasses, 8.9% were VBGs, and 2.1% were bandings.

The shorter followup periods (4+ years) in the SOS study are more likely to reflect contemporary procedures, but not provide long-term data. The SOS study gives no breakdown of mortalities based on types of surgery, however, to enable anyone to find figures that are relevant for patients today.

As Dr. Michel M. Murr, MD, FACS, Director of Bariatric Surgery at Tampa General Hospital in Florida, and colleagues reported in the January issue of Bariatric Times, VBG was the operation of choice during the 1980s. It has fallen out of favor, however, because of its ineffectiveness for long-term weight loss and its significantly higher incidences of long-term complications, leading up to 36% of patients seeking revisional surgery. According to all surgical research to date, secondary operations have higher mortalities and morbidities compared to initial surgeries.

As the clinical trials for FDA approval of lap bands reported, banding patients also suffer high rates of complications (89%). Long-term complications are especially severe, and one of the U.S. centers that have been performing bands the longest recently reported 41% of patients have second surgeries to remove them because of intolerable side effects and another third were currently wanting second surgeries to have their bands removed or get gastric bypass because the bands had resulted in inadequate weight loss. One of the first long-term studies on bandings reported that 7 years post-op, 58% of patients had been reoperated on to have their bands removed or converted to bypasses.

The SOS researchers reported that 31% of their banding patients, 21% of the VBG and 17% of their gastric bypass patients had had reoperations or conversion surgeries because of poor results. (These don’t include those due to post-op complications, they said. Yet, the only additional information on reoperations provided was that 26 patients had immediate post-op complications serious enough to necessitate reoperations.) While this study was reporting on bariatric surgery-related mortalities, there were no surgical-related deaths listed among the “Causes of death.” A footnote reveals there were five deaths within the first 90 days after the study initiation among the surgical patients. “Three of these fatalities were due to gastrointestinal leakage detected too late. One death was caused by a technical mistake during a laparoscopic operation and one by postoperative myocardial infarction,” according to Anna Ryden, Ph.D. and Jarl S. Torgerson, M.D., Ph.D. in an SOS report published in Surgery for Obesity and Related Diseases last fall.

Nor do the SOS researchers offer any explanations for why their mortality figures about 9 years after surgery (5%) are similar to real-life ones seen among American bariatric patients after just the first year. The study by Dr. Flum published in the October 2005 issue of JAMA, for example, reported that for bariatric surgery patients of the same average age as the SOS study patients (47 years), the one-year mortality rates were 4.1% overall (7.7% for men and 3.1% for women).

Lost in follow-up

As we’ve talked about this before, it warrants only a brief reminder. Extraordinarily high “lost in followup” (attrition) rates are common in bariatric and diet studies, meaning that most studies typically have data on fewer than half the original patients. This makes it much easier to report favorable outcomes, as a higher proportion of those who drop out of weight loss studies are those not having good results. But using data on only a small fraction of the original population makes any claims that an intervention is responsible for changes in health indices or weight loss uncredible.

At year ten, the SOS researchers had data on only 31.8% (641) of the original surgical group!

Even with these enormous gaps in the data, the SOS study reported a uniform trajectory of weight regain among all of the remaining surgical patients from years 1 through 8, which continued steadily upward for the banding patients through year 15. They also found no significant correlation between the degree of weight loss and later mortality. “The low cumulative mortality in our study hindered the detection of modest differences.”

But when reporting mortality figures, the authors used social security numbers to cross-check their database against the Swedish Population and Address Register to locate those who had died. They said they found vital statistics on all but three of the original study participants as of November 1, 2005. Does this make their mortality claims any more relevant? Of course not.

Did you catch the leap of logic? The researchers have no idea what’s been happening for nearly a decade to more than two-thirds of the patients they’d operated on ten years earlier. Yet people are ready to believe that the surgery is responsible for purported improved mortalities.

That would be like attributing your beautiful smile to a cleaning a dentist had done ten years ago, ignoring the countless other interventions (dental procedures, sealants, fluoride treatments, antibiotics, etc.) done in the meantime. [Or, perhaps you have rotting teeth ten years later. Lots of other things might explain those, too.]


The biggest problem with this study not only sets it apart from gold standard clinical trials, but calls into question every one of its conclusions. Incredibly, even the authors pointed out this problem, yet the media, medical and scientific communities have completely overlooked it.

Randomization, where study participants are randomly selected and grouped by chance, is the most important way to help ensure that intervention and control groups are comparable (even in ways researchers may not have considered) and that the participants haven’t been hand-chosen to manipulate the findings towards a predetermined conclusion. Randomization prevents stacking the deck, so to speak. While randomization isn’t always ethically possible in human studies, when there has been no randomization, a study needs to be examined with an especially critical eye. Without randomization, findings are less likely to be defensible.

The selection of intervention and control groups can be maneuvered in unimaginable and subtle ways, but the cumulative results can virtually guarantee researchers their findings will satisfy their sponsor.

The SOS study began with a massive recruitment campaign that ran 13.4 years (September 1987 to January 2001), in which application forms were sent out to 11,453 people. It was initially a registry and longitudinal intervention study, with a population study added later. The obese people for the registry study had extensive health examinations performed at 480 collaborating primary healthcare centers in 19 Swedish counties. “Obese subjects were characterized in great detail with respect to medical history, metabolic aberrations, adipose tissue distribution, food habits, and psychological, socioeconomic, and genetic variables,” explained Drs. Ryden and Torgerson. Finally, 2,010 individuals had been carefully selected for the surgical arm and invited to participate based on “detailed inclusion and exclusion criteria.” And 2,037 people were matched using a computer algorithm that took 18 variables into account...

SES. What factor plays one of the most important roles in health outcomes and healthcare that is always controlled for in the most careful medical research? Socioeconomic class/status. The researchers had this information on all of the applicants, yet SES was the most obviously missing variable in their study and one they specifically chose not to control for. That does not mean it wasn’t part of their selection process.

They also chose to not control for other factors, such as genetics/family history, dieting behaviors or physical activity. The SOS study also made no mention of race or ethnicity. All are well-known factors that can affect health and mortality outcomes.

AGE. What is the single strongest predictor of mortality? Age, of course. Even the authors revealed that in both of their computer models of hazard ratios for overall mortality, age was the strongest predictor.

Yet, the people selected to be in the control group for this study were older than the surgical group, by about 1.5 years. As the authors said, this significant difference “would benefit survival in the surgery group.” As expected, mortality among the older participants (over age 47.8) in the control group was higher.

An interesting note regarding age was found in the supplemental data. A greater percentage of younger patients (under 47.8 years) died in the surgical group than the control group. The younger adults had a 50% greater chance of dying having surgery than if they hadn’t had it. The study authors provided no explanation.

PRE-EXISTING CONDITIONS. The authors stated: “Overall mortality was higher in subjects who had had cardiovascular events (myocardial infarctions [heart attacks] or stroke) before baseline [surgery] than in subjects without such events.”

Yet, the people selected for the control group had 25% higher rates of previous heart attacks and strokes, compared to the surgical group. They also had 6.5% higher rates of diabetes at the time of surgery. Not surprisingly, the mortality rates reported for the controls were higher than other general population studies for people of similar size and age.

Are you starting to see why randomization is so important?

With all of this, we might expect dramatically better mortality rates among their bariatric surgical patients as compared to their controls. The fact there wasn’t lends further question as to the value of bariatric surgery. The difference in actual mortality was a mere 1.3% between their surgical and control groups. In statistical reality, no difference. No better than chance!


There is another finding that wasn’t reported in the news but is valuable to note, especially given that bariatric interests are using this study to support bariatric surgeries for the “severely obese.” Those with BMIs ≥40.8 (where bariatric surgery is recommended) who had bariatric surgery had higher risks of dying than those of the same size who didn’t have surgery. Those with BMIs ≥40.8 who had bariatric surgery were 43% more likely to die over the follow-up years than those with the same BMIs in the control group. This is also what other studies to date have reported.

And also confirming other studies, more men died after bariatric surgeries than women. Men actually increased their risks of dying by having the surgery than if they didn’t.

I hope this was a valuable exercise in the importance of looking closely at studies and media claims. The study authors concluded that “bariatric surgery for severe obesity is associated with...decreased overall mortality.”

That is true as far as it goes, but as we’ve seen, it doesn’t begin to tell the full story.

© 2007 Sandy Szwarc

Financial disclosures of the authors of “Effects of Bariatric Surgery on Mortality in Swedish Obese Subjects:”

Dr. L. Sjöström reports receiving grants from Sanofi-Aventis and Ethicon and lecture and consulting fees from AstraZeneca, Biovitrum, Bristol-Myers Squibb, GlaxoSmithKline, Johnson & Johnson, Lenimen, Merck, Novo Nordisk, Hoffmann–La Roche, Pfizer, Sanofi-Aventis, and Servier, having an equity interest in Progenit, NMCT, and Lenimen, and serving on the board of directors for NMCT, PMCT, and Lenimen; Dr. C.D. Sjöström, receiving consulting and lecture fees from Sanofi-Aventis.

Dr. Lystig, being employed by AstraZeneca and having an equity interest in the same company and Amgen.

Dr. B. Carlsson, being employed by AstraZeneca and having an equity interest in the same company.

Dr. Jacobson, receiving research grants from Hoffmann–La Roche; Dr. Karlsson, receiving consulting fees from Pfizer.

Dr. L.M.S. Carlsson, receiving consulting fees from AstraZeneca and having an equity interest in Progenit and Sahltech.

Bookmark and Share