Reader Comments

Post a new comment on this article

Must demostrate uncorrelated statistics

Posted by plosmedicine on 31 Mar 2009 at 00:20 GMT

Author: David Strip
Position: Modeler
Institution: Sandia National Laboratories
E-mail: drstrip@sandia.gov
Submitted Date: February 18, 2008
Published Date: February 19, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Much in line with the response by Mittendorf, the validity of the results lies very strongly on key assumptions that are not demonstrated. The analysis assumes that the cost of an incidence of the 22 key diseases is independent of the risk factors being tested. Likewise, remaining health care costs, which account for 85% of health-care spending in the Netherlands , are assumed to be uncorrelated to risk factors. Given that this latter class of spending dwarfs the former, the importance of demonstrating the lack of correlation is particularly important. The incidence of numerous co-morbidities with obesity argues, in fact, that one might reasonably expect to find that the annual health costs are higher in the obese and that the cost of treatment in the last months preceding death may be quite different from the non-obese.

No competing interests declared.