"The Effect of Age on the Association between  Body-Mass Index and Mortality"
J. Stevens et al

New England Journal of Medicine, Volume 338, Number  1, January 1, 1998. , pp 1-7.

"Losing  Weight-An Ill-Fated New Year's Resolution"
J.P. Kassirer and M. Angell

New England Journal of Medicine, Volume 338, Number  1, January 1, 1998. , pp 52

The study was based on subjects from The American Cancer Society's Cancer   Prevention Study I (1959 through 1972). The 62,116 men and 262,019 women  all filled out questionnaires at the start of the study. These white men  and women never smoked and had no known cardiovascular disease and their  vital statistics (If they died, what was the cause of death?) were followed  for 12 years.  

Body-Mass Index (weight in kilograms divided by height in square meters) was   adjusted for age, physical activity, education level and alcohol  consumption. The relative risk of death from all causes or from  cardiovascular disease was found to be higher with increasing BMI in age  groups from 30 years to 75 years for both men and women. This higher  relative risk with greater BMI did not persist after age 75. 

A good model was generated from the data that can express this lower  relative risk in the older subjects. For a 35 year-old, 5' 10" man to  increase his relative risk of dying for any reason by 50%, he would have to  weigh 166 lbs (BMI=23.8). For the same fictional man at age 80 years to  increase his relative risk of dying for any reason by 50%, he would have to  weigh 294 lbs (BMI=42.1)! 

Does this mean that we can tell our older patients to get as fat as they  like? Well, no we shouldn't. We can continue to tell our older patients  that the weight they carry around is not as dangerous to them at 80 as it  was at 35. Weight loss is more clearly associated with adverse events in  the elderly, including death.   

Criticisms of the study included the fact that it is based on data from  almost 30 years ago, the questionnaires showed self-reported height and  weight, the BMI in older people can be incorrect due to loss of normal  height from osteoporosis, the subjects were unrealistically healthy and did  not smoke, the subjects were not followed over time to assess the risk of  change in BMI and mortality, and death certificates are routinely filled out  with limited information about the actual cause of death and cardiovascular  diseases is often cited as the "default" cause of death.  

The editorial goes further in pointing out that there is still not enough  good data to link mild or modest obesity with increased mortality even  though Americans spend billions of dollars annually to try, but fail, to  lose weight. For physicians to hop on the bandwagon in stigmatizing these  people can have harmful psychological effects as well as potentially harmful  physical effects if drastic fluctuations in weight are the result.

Submitted by Joe Trumble, M.D.

Both articles can also be found on NLM's PubMed Site