read the data, not the abstract
I know I do this myself, I read the abstract and think I know what the study says, but sometimes the abstract puts a spin on the data. So it is the case with the mortality figures of one of the more thorough studies into obesity and mortality.
The study entitled "Excess Deaths Associated With Underweight, Overweight, and Obesity" gives the impression that there is proof here that being overweight or underweight is a risk factor for early death. (http://jama.jamanetwork.com/article.aspx?articleid=200731) The abstract even says:
It says that having a BMI over 30 is associated with a greater chance of dying, and that is not untrue of their data, yet that is also misleading.
Look then at the graph of the results. these results show the different results of 3 different population studies:
Suddenly the perspective changes. Being overweight with a BMI between 25-30 is consistently on the line with being within supposedly normal BMI range of 20-25, if not slightly better. But what differs from the abstract is the fact that being obese with a BMI 30-35 is barely a risk at all. In fact within all age groups, the relative risk of an obese person (BMI 30-35) dying compared to the risk associated with having a 'normal' BMI is not statistically significant (as demonstrated by the error bars on the data points). In one study (NHANES II), this group fared the best in the 25-59 year age group, while the other 2 studies suggested there was an elevated risk in this age group. The large error bars here tell us this age range had few deaths so the studies are not powered enough to give us great certainty. So even if one were to argue that the obesity carries an elevated risk for people aged less than 60, then it needs to be taken in the context that this absolute risk is still very small!
What comes out is that it is only the morbidly obese category that suffers a statistically significant increased risk, but this elevated risk was not found in the over 70 year old groups - where more dying occurs!
What we can take from this is that the supposed risk of being overweight and obese is almost certainly not as certain as most health professionals and health writers wish to make out. Part of this problem stems from the fact that abstracts and titles of these studies try to put a spin on the results to create a false impression that their study is consistent with the general health message.
It is also a problem that not only do these studies have poor correlations, but that health professionals continue to use these weak associations to assume causality. In essence, the elevated risk of dying associated with being morbidly obese does not tell us whether 'excess' weight causes disease or whether there is a confounding variable at play. Given the many factors outside of a persons control that play a huge role in obesity, it is not unfair to assume that any small correlation between 'morbid obesity' and disease is not a causal relationship, but a feature of both sharing a common cause beyond current lifestyle.
The study entitled "Excess Deaths Associated With Underweight, Overweight, and Obesity" gives the impression that there is proof here that being overweight or underweight is a risk factor for early death. (http://jama.jamanetwork.com/article.aspx?articleid=200731) The abstract even says:
"Relative to the normal weight category (BMI 18.5 to <25), obesity (BMI ≥30) was associated with 111 909 excess deaths (95% confidence interval [CI], 53 754-170 064) and underweight with 33 746 excess deaths (95% CI, 15 726-51 766)."
It says that having a BMI over 30 is associated with a greater chance of dying, and that is not untrue of their data, yet that is also misleading.
Look then at the graph of the results. these results show the different results of 3 different population studies:
Suddenly the perspective changes. Being overweight with a BMI between 25-30 is consistently on the line with being within supposedly normal BMI range of 20-25, if not slightly better. But what differs from the abstract is the fact that being obese with a BMI 30-35 is barely a risk at all. In fact within all age groups, the relative risk of an obese person (BMI 30-35) dying compared to the risk associated with having a 'normal' BMI is not statistically significant (as demonstrated by the error bars on the data points). In one study (NHANES II), this group fared the best in the 25-59 year age group, while the other 2 studies suggested there was an elevated risk in this age group. The large error bars here tell us this age range had few deaths so the studies are not powered enough to give us great certainty. So even if one were to argue that the obesity carries an elevated risk for people aged less than 60, then it needs to be taken in the context that this absolute risk is still very small!
What comes out is that it is only the morbidly obese category that suffers a statistically significant increased risk, but this elevated risk was not found in the over 70 year old groups - where more dying occurs!
What we can take from this is that the supposed risk of being overweight and obese is almost certainly not as certain as most health professionals and health writers wish to make out. Part of this problem stems from the fact that abstracts and titles of these studies try to put a spin on the results to create a false impression that their study is consistent with the general health message.
It is also a problem that not only do these studies have poor correlations, but that health professionals continue to use these weak associations to assume causality. In essence, the elevated risk of dying associated with being morbidly obese does not tell us whether 'excess' weight causes disease or whether there is a confounding variable at play. Given the many factors outside of a persons control that play a huge role in obesity, it is not unfair to assume that any small correlation between 'morbid obesity' and disease is not a causal relationship, but a feature of both sharing a common cause beyond current lifestyle.
