The possible importance of income and education as covariates in cohort studies that investigate the relationship between diet and disease [version 2; referees: 2 approved]

Background: Many cohort studies have been carried out that have provided information on the relationship between diet and health-related outcomes. Omission of important covariates during multivariate analysis may give rise to error due to residual confounding. A possibly important covariate is socioeconomic status (SES) as this is related to both diet and health. Objective: To determine the frequency with which different measures of SES are included as covariates during multivariate analysis of cohort studies that investigated the relationship between diet and health. Methodology: An analysis was carried out of 76 randomly selected papers from 66 cohort studies. The papers covered many dietary variables and a wide variety of diseases/health-related outcomes. The cohort studies were carried out in many different locations and the subjects varied widely in age. Results: Approximately two-thirds of the papers (65.8%) used at least one measure of SES as a covariate. Education was used most often (60.5% of papers), followed by income (14.4%) and social class (2.6%). More than one measure of SES was used in 11.8% of papers. Conclusions: Failure to include income (or another measure of present SES, such as occupation) may be a common source of error in cohort studies. Over-reliance on education may be particularly important as it is likely to be a weaker measure of present SES than is income. There is a need for more research on this question. SES in childhood is almost never included in multivariate analysis in cohort studies carried out on adults. This could also play a significant role in disease risk in middle age or later. Very little is known regarding whether this is also a source of residual confounding.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.