by Lindsey Smith, PhD Student in Nutrition Epidemiology at UNC
Every year on January 1st, millions of Americans toss out their sweets, join the gym, and shell out big bucks on fancy workout gear with the resolution to lose weight. This year, however, dieters got a welcome reprieve from those pesky resolutions to shed excess fat when a recent scientific report made headlines nationwide, with bold titles claiming “Overweight Americans face lower mortality risk” and “Best to be overweight?” The study, which was published in JAMA (the premier medical journal), is a systematic review and meta-analysis. The results of this meta-analysis showed that, relative to normal weight, grade 1 obesity (the mildest type of obesity) was not associated with higher all-cause mortality. What’s more, overweight was associated actually associated with lower all-cause mortality.
So does this mean we should all toss our weight-loss efforts out the window and tear into that bag of Oreos?
Not so fast.
Unfortunately, most news articles are forced by their brevity and readership to skim over key details, which caution against a strict interpretation of these results and suggest we might reconsider before abandoning obesity-prevention efforts. Here’s why:
(1) The nature of the study design: meta-analyses are a type of study which combine results from previously published articles to draw conclusions. However, publication bias presents a serious problem to the validity of this methodology. In layman’s terms, studies that present significant findings are often more likely to be published than studies that find no differences. This means that meta-analyses are often only able to take into account one piece of the whole picture. Although there are ways to test whether publication bias influences results, there is no way to truly know whether publication bias came into play.
(2) Age/period/cohort: It is very difficult to disentangle the effects of age, period, and cohort when examining these types of studies. Typically, meta-analyses pool results from studies that may span decades as long as the studies meet certain inclusion criteria. But, as US adults have become increasingly overweight in the last 50 years, the fundamental meaning of overweight and obesity has changed. In 1970, someone who was obese might have been considered very ill. In 2010, when obesity is the norm, an obese person might actually represent the normal state. Being normal weight or underweight could be the result of pre-existing or latent illness, which would explain why lower weights are linked to increased risk of death.
(3) Treatment effects: treatments for obesity and co-related morbidities (such as diabetes and hypertension) have dramatically improved in recent decades, extending the life of those who might otherwise died earlier due to these complications. In this case, it would be wrong to interpret these results as meaning that obesity isn’t associated with increased death—it simply means that we have come a long way in keeping sick people alive longer.
(4) Biologic plausibility: the authors, and most epidemiologists, use cut-points to define overweight and obesity. Yet, intuitively, one might question whether there is much biologic difference between someone with a BMI of 24.5 (“normal weight”) and a BMI of 25.5 (“overweight”). In many cases, how a variable is classified can make a difference in the outcome. While this type of categorization is often necessary in epidemiological analysis, it is important to draw a distinction between what is statistically true and biologically meaningful. It is possible that if BMI were constructed differently, the observed associations in this study might disappear or even reverse.
The key here is to read the literature with a discerning eye. Although this study used rigorous epidemiological methods, distilling the results into a single bullet or headline can be misleading. It is also dangerous to draw conclusions from one study without considering others. For example, a recent study from Columbia University found the exact opposite results: that obesity increase risk of mortality, and this effect gets stronger with age. More research is required—ideally through randomized controlled trials—to truly understand the complex relationship between overweight/obesity and mortality.
Flegal KM, Kit BK, Orpana H, Graubard BI. Association of All-Cause Mortality with Overweight and Obesity Using Standard Body Mass Index Categories. JAMA2013; 309(1):71-82.
Masters RK, Powers DA, Link BG. Obesity and US Mortality Risk over the Adult Life Course. American Journal of Epidemiology 2013;177(5):431-42.