What’s the skinny on the “Obesity Paradox”?

If you have been reading medical news headlines recently, you’ve probably heard about the “obesity paradox”, and the relationship between body mass index (BMI), a measure of body composition, and risk of death, i.e., mortality. For decades, doctors and researchers have been puzzled by research results that have suggested a protective effect of obesity on risk of mortality. The suspected mechanism behind the protective effect of being slightly overweight has often been explained as the benefits of having a “metabolic reserve”, or extra stores to draw upon if you get severely sick or hospitalized. Obese people have an advantage over an underweight frail person when facing critical illness. This apparent paradox received lots of attention about a year ago, in the Journal of the American Medical Association (JAMA)1, researchers reported a meta-analysis of 141 studies that examined the relationship between BMI and all-cause mortality. The findings revealed people who were severely obese had a significantly higher risk of all-cause mortality, relative to people of normal weight. However, people who were overweight or mildly obese had a significantly lower risk of all-cause mortality.

When this research was reported, it generated many comments and critiques from epidemiologists, doctors, and other scientists. Ultimately, several of the leading epidemiologists conducted new analyses in different datasets with improved methods to see if this apparent “paradox” was reproducible. These new studies, reported in a January 2014 editions of the New England Journal of Medicine2 and the Journal of General Internal Medicine3 , provide perspectives on the “obesity paradox” specific to people with diabetes. Their findings found a more nuanced relationship between BMI and mortality risk, specifically related to people with diabetes.

BMI: Body Mush Index or Barely Meaningful Index?

First, let's review some of the definitions used in these studies, starting with BMI. BMI is a calculation using height and weight that is used to classify individuals into normal weight, overweight, and obese; the standard definitions follow below in Table 1. Despite being frequently used to define obesity, as an indicator of risk for diabetes, there are limitations to the use of BMI. Specifically the BMI calculation does not differentiate body weight that is due to fat, or body weight that is due to lean muscle mass. This differentiation is important because lean muscle mass is highly metabolically active tissue resulting from physical activity and strength training, which is much, much different from fat stores that are the consequence of over-eating and inadequate physical activity! Additionally, fat in the body affects risk differently depending on where in the body it is, i.e., fat under the skin (i.e., subcutaneous fat) confers less risk than fat stored in the organs (i.e., visceral fat) or fat around the abdomen - and again this differentiation is not considered by the BMI calculation.

Table 1: Definitions of Overweight and Obesity by BMI


 BMI range



 Normal weight


- Lower end 


- Upper end 




- Lower end 


- Upper end 




- Class 1 


   - Class 2 


- Class 3 



Now that we better understand the application and limitations of “BMI”, let’s review some of the findings of the newer studies exploring the “obesity paradox” and consider some explanations for why lower weight people may be at greater risk for death. One important factor is that lower body weight people aren’t all lean athletes, this group often also includes people who are frail or elderly, heavy smokers, those who have lost weight due to serious diseases, and undernourished people. The new studies controlled for smoking (specifically never smokers vs. people who quit), looked at younger populations, and controlled for serious diseases like cancer and heart disease. The news studies also looked specifically at people with diabetes.


Recent Studies Evaluating BMI and Risk of Death

The recent NEJM study looked at over 11,000 people with diabetes who were free from cancer or heart disease. They followed this cohort and tracked BMI and deaths. They found a more common-sense pattern, a linear relationship of increasing risk of death with increasing BMI among never smokers and people who were diagnosed with diabetes under age 65. A “J-shaped curve” was identified for people with diabetes with other health risk factors (e.g., smoking), which means the risk of death was higher among those at both the very low end and at the high end of the BMI scale. Interestingly, the “obesity paradox” was still supported in overweight & Class 1 obese people without diabetes!

The subsequent report in the J of General Internal Medicine took a different approach (in terms of statistical analysis) and attempted to determine if diabetes influenced the relationship between obesity and mortality. This study looked at over 74,000 people, some of whom had diabetes and some who did not. According to their results, people with diabetes had more total deaths at all BMI categories, compared to those without diabetes, and specifically increased waist circumference or belly fat predicted higher mortality. However, as BMI increased, the rate of death within each BMI category was slower for people with diabetes, suggesting a potentially protective effect, or a “paradox”, of higher BMI for people with diabetes.

However, the second study was limited methodologically in several important ways. First, it measured BMI at the time of diabetes diagnosis, and did not track weight change over time, and thus was unable to account for people who may have lost weight or reversed diabetes. Also, the categorization of participants as “diabetic” also occurred only at the beginning of the study, and therefore the investigators were unable to account for those developing diabetes over the course of the study, and thus some people may have been misclassified as not having diabetes for the duration of the study. Additionally, although investigators considered the importance of balancing many risk factors other than BMI in their analysis, they did not adjust for differences in specific cardiovascular risk factors like concentration of blood cholesterol or degree of high blood pressure. Although the diagnosis of diabetes is based on blood glucose, factors other than blood glucose impact risk of death even when blood sugar is high, including measures of inflammation. Another consideration for the apparent protective effect of BMI in those with diabetes, is that people with very advanced diabetes would be more likely to be very lean and frail, and thus die earlier in the study period. It is also feasible that the protection may have nothing to do with BMI or diabetes, but may be because people with diabetes are more likely to have closer contact with their physicians and thus increased medical surveillance for preventable diseases, as well as, more aggressive treatment and management of risk factors and acute conditions as they occur.

What have these studies taught us?

While the “obesity paradox” hasn’t been proven or disproven by epidemiologists, these latest studies provide people with diabetes some useful information about how to reduce their risk:

  • Smoking is a bigger risk factor for death than obesity. If you smoke, ask for help to quit.

  • Being very obese, i.e., a BMI over 35, leads to increased death.

  • Weight loss is still important for people with and without diabetes. Even a 10% weight loss has substantial health benefits

  • Abdominal and visceral obesity matter most.

  • Metabolic “fitness”, represented by normal blood sugar and lots of lean muscle mass, is healthy at any size!

Remember, adipose tissue isn’t just inert fat- it releases unique hormones and contributes to inflammation.4 Inflammatory cytokines, e.g., TNF-alpha and IL-6, come from adipose tissue and cause low-grade inflammation, contributing to the development of diabetes, heart disease, cancer, and other diseases which have a more direct affect on mortality risk. Interestingly, my recent research in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort demonstrates the activity of a liver enzyme called “GGT” is strongly associated with obesity, but also with IL-6 and other measures of inflammation5, and even small increases in GGT are known to interact with BMI, greatly increasing risk for developing diabetes6. As discussed in previous articles, small elevations in GGT activity also increase risk for high blood pressure, cancer, and death.

Next directions

Despite the results of these recent studies, which have only partially clarified the “obesity paradox”, most people, with our without diabetes, are actively looking for strategies to lose weight. Whether or not the extra weight increases your risk for death, carrying unwanted body weight creates a burden, and increases risk for arthritis and depression, which reduce active living. Have there been any advances in weight loss treatments, whether natural, pharmaceutical or surgical? What’s the update on gastric bypass surgery for weight loss and diabetes prevention? Stay tuned for next month’s article!

In health, Ryan Bradley, ND, MPH

February, 2014



  1. Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA : the journal of the American Medical Association. Jan 2 2013;309(1):71-82.

  2. Tobias DK, Pan A, Jackson CL, et al. Body-mass index and mortality among adults with incident type 2 diabetes. The New England journal of medicine. Jan 16 2014;370(3):233-244.

  3. Jackson CL, Yeh HC, Szklo M, et al. Body-Mass Index and All-Cause Mortality in US Adults With and Without Diabetes. Journal of general internal medicine. Jan 2014;29(1):25-33.

  4. Gregor MF, Hotamisligil GS. Inflammatory mechanisms in obesity. Annual review of immunology. 2011;29:415-445.

  5. Bradley R FA, Lee DH, Swords-Jenny N, Jacobs DR, Herrington D. . Associations between γ -glutamyltransferase (GGT) Activity and Atherosclerosis: The Multi-Ethnic Study of Atherosclerosis (MESA). . Atherosclerosis. 2014.

  6. Lim JS, Lee DH, Park JY, Jin SH, Jacobs DR, Jr. A strong interaction between serum gamma-glutamyltransferase and obesity on the risk of prevalent type 2 diabetes: results from the Third National Health and Nutrition Examination Survey. Clin Chem. Jun 2007;53(6):1092-1098.