What Came First, Statistics or the Egg?
Maintaining or improving healthy eating habits is on the mind of many people as we enter the holiday season. But what to eat can be confusing because of mixed messages. Nutritional epidemiology is often used as a starting place to test relationships between nutrients, foods or a variety of foods (diet patterns) and health outcomes. The caveat to epidemiological studies is that unlike an intervention, nutrients or foods are not controlled and individuals often have very different characteristics. As Nicklas and colleagues point out in a recent publication, this can result in inconsistent findings if the statistical approach, covariates used and assumptions are not appropriate for the study.
Nicklas and colleagues examined the relationship between egg consumption, adiposity and waist circumference in 18,987 adults using two statistical approaches. The first approach categorized people as egg consumers or nonconsumers. Consumers had significantly higher body mass index and larger waist circumference than nonconsumers. The second approach identified patterns of egg consumption using cluster analysis which groups foods consumed together in a pattern. Of the 8 egg patterns identified, only 2 were associated with differences in BMI and waist circumference and less than 2% of the study population had these dietary patterns. The authors then controlled for other food groups consumed and medication use and the association between egg patterns, BMI and waist circumference were no longer significant. This is a clear illustration of reaching two very different conclusions regarding “risks” of egg consumption.
The message that statistical methods and interpretation of results needs careful consideration to avoid misinformation is important but unfortunately not new as can be seen here. Earlier this year a paper from Annals of Internal Medicine reported that relationships between cardiovascular disease, polyunsaturated fatty acids and saturated fats are similar. However, that conclusion was born from several errors and a corrected version showed beneficial associations for polyunsaturated fatty acids (omega-3’s, DHA, EPA and DHA+EPA) and cardiovascular disease. Similarly, a meta-analysis by Jiao and colleagues reported no benefit of DHA on cognition in children and adults but the methods used introduced bias and in some instances only two studies were used to draw conclusions.
Care clearly needs to be taken when examining diet and health relationships to minimize confusion. Studies like Nicklas that advocate responsible practices are an important step forward.
Nicklas TA, O’Neil CE, Fulgoni VL. Differing statistical approaches affect the relation between egg consumption, adiposity, and cardiovascular risk factors in adults. 2014 J Nutr doi:10.3945/jn.114.194068
Chowdhury R, Warnakula S, Kunutsor S, Crowe F, Ward HA, Johnson L, Franco OH, Butterworth AS, Forouhi NG, Thompson SG, Khaw K-T, Mozaffarian D, Danesh J, Di Angelantonio E. Association of dietary, circulating, and supplemental fatty acids with coronary risk: A systematic review and meta-analysis. Ann Intern Med 2014;160(6):398-406 (corrected paper)
Jiao J, Li Q, Chu J, Zeng W, Yang M, Zhu S. Effect of n-3 PUFA supplementation on cognitive function throughout the life span from infancy to old age: a systematic review and meta-analysis of randomized controlled trials. Am J Clin Nutr 2014;100:1422-1436.