Back in January 2013, the Journal of the American Medical Association published a seriously flawed study in which lead author Dr. Flegal and her associates at the Centers for Disease Control and Prevention (CDC) claimed that overweight people were actually less likely to die than normal weight people. They even claimed that those with stage 1 obesity (BMI of 30 to <35) were no more likely to die than normal weight people (1). Despite the serious flaws in the design and interpretation of this study's purported findings, for the most part the news media proclaimed that the results actually showed that being overweight appeared to reduce mortality and even stage 1 obesity was not associated with premature death compared to normal weight people. No doubt this study and the news media hype of the results were comforting for the approximately 70% of Americans who are overweight or obese. Certainly the association between BMI (and especially excess adiposity) and life expectancy can be helpful for estimating its public health impact, but this is true only if the impact of BMI is not distorted by the impact of smoking and other causes of ill health that cause weight loss, or even faster aging. It is also important that any statistical association between increased BMI and morbidity and mortality not be distorted by adjusting one's data for other causes of illness and death such as dyslipidemia, elevated blood pressure, and/or insulin resistance and type 2 diabetes. Why? Because it is known that those risk factors and many others are actually caused in part by weight gain and increasing BMI.
Critics of Dr. Flegal's study focused on some of the questionable methods she used in compiling statistics that distorted her results. For example, Dr. Flegal included people too thin to fit what most consider to be normal weight range (BMI 20 to 24.9). It is known that many people with a BMI under 20 were likely already somewhat emaciated by cancer, failing hearts, emphysema, senility, or another serious disease at the time their BMI was measured. She also included smokers, who tend to be thinner but who have a much greater risk of dying from heart disease, emphysema, and cancer than nonsmokers. The statistical associations created by this study's inclusion of smokers and people with pre-existing illnesses was likely largely responsible for producing questionable association between being overweight and being at reduced risk of dying.
New Much Larger Meta-Analysis Refutes Flegal's Analysis
The Global BMI Mortality Collaboration published data in The Lancet on July 13, 2016 (2). This group's meta-analysis of 239 prospective studies on four continents examined the association between BMI and mortality over an average of 13.7 years. They focused mainly on the results of 189 studies with nearly 4 million nonsmoking subjects without chronic diseases at recruitment who survived 5 years, of whom 385,879 died. Unlike Dr. Flegal's study, this far larger and better designed meta-analysis reported all-cause-mortality was actually the lowest in normal weight subjects with a BMI of 20 to <25. The risk of death was significantly greater for people whose initial BMIs put them in the overweight (25 to 29.9) and obese (30 or greater) categories, compared with the normal weight subjects. They also found that the risk of being overweight was greater for men than women. This new study's data did show the risk of dying for those who were very thin (BMI<20) and especially underweight (BMI <18.5) was higher than for the normal weight nonsmokers.
Flegal defended her work by claiming she used standard categories for weight classes and made some statistical adjustments for smokers, but she included data in her meta-analysis from studies that defined "normal weight" of a BMI of less than 20. Dr. Flegal adjusted her data by eliminating people who were clearly ill at the time their BMI was measured by eliminating those in hospitals and under hospice care. However, not all people who lose weight due to chronic ills are in hospitals or hospices as Dr. Flegal assumed. Far more are in nursing homes and free living and those “thin due to illness” subjects were included in her study. A far better way to statistically correct for illness that is causing poor appetite and weight loss and increasing mortality is to wait several years after BMI is measured to start looking for deaths. The Global BMI Mortality study waited 5 years but Dr. Flegal included all deaths in her analysis, including those that occurred shortly after the subjects’ BMIs were recorded. While waiting 5 years after the BMI measurement to start looking at mortality probably eliminates much of the illness-caused weight loss due to ill health that likely lead to earlier deaths, , it certainly does not eliminate all illness-related weight loss. Failure to correct for all illness-caused weight loss increases the correlation between lower BMI and greater mortality. This results in what statisticians call reverse causality. The failure to correct for this reverse causality creates a non-causal statistical correlation between lower BMI and increased mortality that no doubt grossly distorted Dr. Flegal's results and contributed to her erroneous conclusions. It is likely that even the 5-year waiting period used in The Global BMI Mortality Collaboration's study did not completely eliminate all of this reverse causality. For example, unintentional weight loss often begins 10 or more years before Alzheimer's disease is diagnosed. The aging process itself results in unintentional weight loss so there is reason to believe that people who are aging more quickly may see greater drops in their BMIs than those who are aging more slowly. So if declining BMI predicts faster aging and faster aging leads to increased risk of dying, then this too would create a correlation between a lower BMI, especially in older people and an increased risk of dying. But again this correlation would not mean that older overweight or obese people who lose weight intentionally by adopting a healthier diet and exercise program are going to age faster and make themselves more likely to die.
Research in numerous animals shows that a reduced calorie intake reduces their weight relative to animals fed ad libitum, but the lower calorie intake actually slows down the aging process and prolongs survival. This animal data suggests that eating more calories and gaining weight may actually cause people to age a bit faster. How does Dr. Flegal explain how weight gain in middle or older age would improve health or slow aging? On the other hand, being very lean may not always improve survival. For example, older people with more body fat may be more likely to survive a serious infection simply because their initially higher energy reserves when they become sick make it less likely for them to reach a critically low BMI level that may impair their ability to fight off the infection. However, being overweight or obese promotes many types of cancer, raises blood pressure, promotes type 2 diabetes, increases gallstones, osteoarthritis, senility, dyslipidemia, heart attacks, & heart failure, and numerous other ills that can increase both morbidity and mortality. A recent controlled clinical trial of 220 normal weight to slightly overweight subjects who were randomly assigned to either continue their current diet or adopt a high satiety reduced calorie diet for 2 years found that weight loss resulted in a significant reduction in inflammatory markers including CRP and TNF-alpha (3).
The absurdity of any claim that it is healthier to be overweight or obese especially in older people could easily be seen if a thin or normal weight person gained a significant amount of weight. What would happen? Their blood lipids generally become more atherogenic (decreased HDL-C, increased triglycerides, and nonHDL-C), and many would also see increased fasting insulin levels and more insulin resistance, increased blood pressure, increased CRP and other inflammatory markers. By contrast, if that person lost 10 or 20 pounds, all these risk factors for CVD and type 2 diabetes would likely improve. And there are now at least 10 to perhaps 12 types of cancer including breast (in postmenopausal women), prostate, colo-rectal, stomach, and esophageal that are significantly more common in overweight and especially obese individuals. So how do researchers like Dr. Flegal suspect being overweight improves health and increases longevity?
We all know people who were overweight or obese most of their lives, but unless they die suddenly from an accident, heart attack or stroke, they typically lose a lot of weight before dying from some degenerative diseases and/or the aging process itself. Since illness- and aging-caused weight loss is much greater in older people than in young to middle aged people, this probably explains why the association between increased BMI and mortality tends to be reduced in older subjects. This is exactly what The Global BMI Mortality Collaboration meta-analysis showed.
Bottom Line: There is a big difference between losing weight unintentionally due to faster aging and/or illness and intentionally losing excess adiposity by adopting a healthier diet and exercise program. People with Alzheimer's, congestive heart failure, emphysema, renal failure, and numerous other ills generally lose weight unintentionally for years before succumbing to their disease. Unintentional weight loss is often associated with serious disease that markedly increases the risk of dying. However, this disease-promoted weight loss is very different than people intentionally losing weight by adopting a healthful diet and exercise program. Research has shown marked improvement in numerous disease risk factors associated with an increased risk of dying from many serious degenerative diseases if weight is lost through said diet and exercise program.
By James J. Kenney, PhD, FACN
- Flegal KM, et al. JAMA 2013;309: 71-82 or http://jama.jamanetwork. com/article.aspx?articleid=1555137.
- Meydani SN, Das S, Pieper CF, et.al. Long-term calorie restriction inhibits inflammation without impairing cell mediated immunity: A randomized controlled trial in non-obese humans. Aging. 8(7). Published online July 13, 2016.
Stephanie Ronco has been editing in a professional capacity for the past 10 years. In addition to her work as an editor, Ronco has also served as a ghostwriter and writing tutor. A voracious reader, Ronco loves watching language evolve and change. When she’s not delving into her latest project, Ronco can be found teaching acting classes, performing in community theater, or sailing with her husband.