Scientific evidence for diets: don't believe everything you read13 years, 8 months ago
Posted on Mar 01, 2005, 4 a.m.
By Bill Freeman
In a society increasingly fixated with body image, we are bombarded with so-called scientific evidence promoting the use of a myriad of diets. An article published today in the Open Access journal BMC Medical Research Methodology suggests that we shouldn't take everything we read at face value, as most research articles reporting weight loss studies fail to indicate crucial patient characteristics that may bias the results.
Cheryl Gibson, from the University of Kansas School of Medicine, and colleagues found that over 90% of diet studies did not adequately and comprehensively describe their subjects, making them almost impossible to interpret accurately.
In the USA, 97 million adults are overweight or obese. Extensive studies on the role of diet, exercise and genes are being carried out in a general effort to better understand and prevent obesity.
The researchers analysed 231 articles reporting studies of the effects of diet restriction, diet restriction and exercise, or exercise only, on weight loss; as well as research studies on body composition, fat distribution, metabolism and aerobic fitness. The analysis included articles from 1966 to 2003 that reported studies of obese adult participants.
The team focused on how the articles reported subject characteristics, using the Consolidation of the Standard of Reporting Trials Characteristics (CONSORT) as a guide. The CONSORT statement is a list of 21 elements, recommended as essential for a study to be valid, by a panel of clinical investigators, epidemiologists, biostatisticians and journal editors. The characteristics the researchers chose to concentrate on were age, gender, general health, medication use (other than drugs taken to control obesity), ethnicity and postmenopausal status. It has been shown that all of these characteristics affect weight and can alter the effect of diet and exercise on weight loss.
The researchers also examined whether the sample size at the beginning and at the end of the study was indicated, by gender. "Without knowledge of the number of subjects who were lost to follow-up, readers are unable to judge the effectiveness of a clinical treatment or ascertain whether or not a research finding has practical significance," explain the authors.
Their results show that medication use was the least indicated of the characteristics, as 92% of the articles missed it out. The subject's health status was ignored by 34% of the studies. Ethnicity was not indicated in 86% of the studies, and the subject's age was not reported in 11% of the studies. 8% of the studies failed to report the postmenopausal status of their participants, and 4% did not indicate gender.
"We found major shortcomings in the reporting of subject characteristics," conclude the authors. "Many studies did not report variables that may explain some of the variance in outcomes [...] and reveal poor adherence to published standards of reporting".
Inadequate reporting can lead to biased results being accepted as valid. The quality of reporting for weight loss studies has to improve if we are ever to understand obesity.
### This press release is based on the following article:
Reporting Quality of Randomized Trials in the Diet and Exercise Literature for Weight Loss Cheryl A Gibson, Erik P Kirk, James D LeCheminant, Bruce W Bailey Jr, Guoyuan Huang and Joseph E Donnelly BMC Medical Research Methodology 2005, 5:9 (23 February 2005)