dear andrew perrin: i was wrong and you were right on the obesity and mortality correlation

A while back, Andrew and I got into an online discussion about the obesity/mortality correlation. He said it was true, I was a skeptic because I had read a number of studies that said otherwise. Also, the negative consequences of obesity can be mitigated via medical intervention. E.g., you may develop diabetes, but you can get treatment so you won’t die.

The other day, I wanted to follow up on this issue and it turns out that the biomedical community has come up with a more definitive answer. Using standard definitions of obesity (BMI) and mortality, Katherine Flegal, Broan Kit, Heather Orpana, and Barry I. Graubard conducted a meta-analysis of 97 articles that used similar measures of obesity and mortality. Roughly speaking, many studies report a positive effect, many report no effect, and some even report a negative effect. When you add them all together, you get a correlation between high obesity and mortality, but it is not true at ranges closer to non-overweight BMI. From the abstract of Association of All-Cause Mortality With Overweight and Obesity Using Standard Body Mass Index Categories: A Systematic Review and Meta-analysis, published in the 2013 Journal of the American Medical Association:

Conclusions and Relevance Relative to normal weight, both obesity (all grades) and grades 2 and 3 obesity were associated with significantly higher all-cause mortality. Grade 1 obesity overall was not associated with higher mortality, and overweight was associated with significantly lower all-cause mortality. The use of predefined standard BMI groupings can facilitate between-study comparisons.

In other words, high obesity is definitely correlated with mortality (Andrew’s claim). Mild obesity and “overweight” are correlated with less mortality (a weaker version of my claim). The article does not settle the issue of causation. It can be very likely that less healthy people gain weight. E.g., people with low mobility may not exercise or take up bad diets. Or people who are very skinny may be ill as well. Still, I am changing my mind on the basic facts – high levels of obesity increase mortality.

50+ chapters of grad skool advice goodness: Grad Skool Rulz ($2!!!!)/From Black Power/Party in the Street

Written by fabiorojas

June 4, 2015 at 12:01 am

7 Responses

Subscribe to comments with RSS.

  1. It is significant, however, if I am reading this right, that being overweight is associated with LOWER mortality than being a normal weight. I think that this is important for two reasons: (1) a lot of discussions of the obesity epidemic lump everyone over “normal” weight into one category, as if this is terrible, when apparently it might actually be good to be overweight, and (2) no one is freaking out that apparently being normal is (more) dangerous (than being overweight).
    Also re: causation, there is also evidence that very obese people get worse treatment (i.e., are not as likely to be screened as thoroughly), which would also have an impact on mortality.

    Liked by 1 person


    June 4, 2015 at 12:40 am

  2. @Sarah: Isn’t there also a question as to the causality of the association between being overweight and lower mortality, though? Essentially the argument being that many terminal illnesses cause weight loss.

    Liked by 1 person


    June 4, 2015 at 2:51 am

  3. This article make me hungry


    Robert Eberhart

    June 4, 2015 at 4:12 am

  4. The life course matters. High overweight/mild obesity in early adulthood is bad because it raises the risk of obesity later on, it triggers some early complications with high blood pressure, joint/tissue damage, etc. Being overweight relative to ‘normal’ becomes a very good thing once you’re older.



    June 4, 2015 at 1:48 pm

  5. All these comments show that life course is a crucial component of understanding health data.



    June 4, 2015 at 4:59 pm

  6. Thanks for posting this, I hadn’t noticed it but will read it soon. I agree with the points on life course and causality for “overweight” (non-obese) categories. Causality is very thorny in all of this, particularly if you add in causal pathways that include fat stigmatization. The lower-mortality finding should probably be tempered by considering not just mortality but morbidity as well, both in terms of quality-of-life and health care utilization.



    June 4, 2015 at 8:28 pm

  7. Overweight people have a higher HR when you remove eg people with a history of cancer that puts them at a lower weight:

    “For the overweight category only, we also repeated analyses including the results from a study that pooled data from 19 cohorts. After excluding ever smokers and those with a history of cancer or heart disease, Berrington de Gonzalez et al8 found a HR of 1.11 (95% CI, 1.07-1.16) for men and 1.13 (95% CI, 1.09-1.16) for women with a BMI of 25 to 29.9 relative to those with a BMI of 20 to less than 25 (Amy Berrington de Gonzalez, DPhil, written communication, June 16, 2011).”

    Please read the article..


    Kim Josefsen

    June 9, 2015 at 8:33 pm

Comments are closed.

%d bloggers like this: