Monday, December 6, 2010

Is a little cushioning protective?

Here is the information on the study I mentioned yesterday.  It is in the current issue of the New England Journal of Medicine.

The title of the research article is Body-Mass Index and Mortality among 1.46 Million White Adults. The title does not explain the research but indicates to whom the results can be applied.

I did read the article today and the results offer strong associations between a BMI range and all cause mortality.  The study does have some limitations. For instance, it does not start with a group of indivudals that are followed over time and tested periodically over that time, but reviews previous studies. That being said, the research is sound for what it claims and is important for all of us to consider.

The scientists chose 19 prospective studies that collected baseline data on persons and followed them for no less than five years.  Each study had its own criteria for inclusion and its own outcome of interest.  To be included in the new study, there had to be at least 1000 white person deaths during the course of the study and the study had to include information on age, marital status, alcohol consumption, physical activity, education, smoking history and disease status.  The new study researchers had a cut off of BMI as well.  They did not include persons with BMIs under 15 or over 50 and they excluded people who were less than age 19 or more that age 85 at the start of those individual studies.  The particpants that were included in this study were white.  The researchers felt that race differences could impact the outcome and separate studies should be completed to measure BMI and mortality with regard to specific races.

The researchers did look at everyone together and compared their BMI (adjusted and unadjusted) with rates of death in each 5 pt BMI category during the study periods.  They controlled for all of the above mentioned variables so that alcohol consumption or education for example, did not impact the statistics.  They also standardized age so that the increase in mortality of older persons did not confuse the data on BMI risk.

Separately they analyzed ONLY those who were healthy at the start of each study and who had never smoked.  The findings were similar in the all subject and select subject groups but the association was stronger in the second group.  As the people did not smoke and were healthy at the start, the higher hazard ratios suggest that the BMI - the weight - is the reason for the increased death.  (see below)

What are the results?  They are given in hazard ratios (the odds that one group will have the outcome vs a referent group).  For example, the odds that a smoker will have a heart attack vs the odds that a non smoker (referent) will.

In this study, the group that the overweight and obese persons were compared to - or the referent group - were those whose BMI was between 22.5 and 24.9.  They were considered to have expected or average mortality rates.

For each five point rage, hazard ratios were calculated, including the lower BMI groups.

I am starting to sound more technical than I prefer for the blog.  If you want to know more about analyses, confounders, limitations,  etc - please email or comment.

The rates of death for any cause were higher for every group as the weights (which BMI represents) went up so that the group with BMI of 35 to 40 was 1.8 times more likely to die from anything than the referent group (BMI 22.5 - 24.9).  For men, that same group (35-40) was twice as likely to die as the referent.  These are obese persons.  What the authors really wanted to determine (in response to some conflicting data) was if just being a little overweight was also associated with increase in mortality.  They have found this to be true.  Starting with BMI over 25 the HR go above one.

Also noted, there was a HR of 1.4 for BMI 15 to 18.4  and a HR of 1.1 for 18.5 to 19.9.  There were several reasons offered for the higher mortality in low BMIs, such as disease, low weight due to fitness and low weight due to wasting.  They found that if a study lasted more than 5 years, the low BMI HR was no longer strongly associated with increased death.  Thus the authors suggest a weak correlation between very low BMI and all cause mortality but do not rule it out. 

The researchers do state that a BMI between 20.0 to 24.9 is health promoting.  And to answer the question in the post title - NO< being a little overweight is not a good thing.

Some last words:  The BMI is helpful with regard to population numbers, and this research is about populations not individuals.  Previous research on weight and disease tells us that body fat is detrimental and the BMI is a poor substitute for indivual body fat status. For most people however, if the BMI chart shows them as overweight, they do have excess body fat.  The better measure is waist circumference but people don't always calculate it correctly.  The best measurement involves water and we certainly can't go putting thousands of people in those underwater pods.  So the BMI is helpful, not perfect. 

These results also tell us that people with BMIs at this range are more likely to have an outcome (mortality) but that doesn't mean that you (insert name) will die before someone else because of your number.  

I want to include one of the tables from the study because it is very important.  The reference for the study is below.  As the journal allowed the download of this graphic I am taking that as permission to use it with credut.  The study article is not available without a journal subscription but you may have access if you are in school or work for a company that pays for a journal service.

If you can see this chart, there are confidence intervals.  That means that the researchers are sure that all the people in this BMI category have a HR between those two numbers in the parenthesis.  In some cases the low BMIs actually have LOWER death rates than those with the normal range.  With a HR you are looking at anything above or below the number one. 
 If it is below one, there is less death if it is above one there is more death. One is even - no difference. 
Think about all that while you study this chart.  And if you do not know your BMI you can use the link here to calculate it.  You can also see in this chart that the obese persons rates of death stay high regardless of the length of the study.

Reference: Berrington de Gonzalez, A., Hartge, P., Cerhan, J. R., Flint, A. J., Hannan, L., MacInnis, R. J., . . . Thun, M. J. (2010). Body-Mass Index and Mortality among 1.46 Million White Adults. New England Journal of Medicine, 363(23), 2211-2219. doi: doi:10.1056/NEJMoa1000367

No comments: