LAUSR

TO THE EDITOR: We believe that there are 2 potential threats to the validity of Gunter and colleagues' (1) and Park and associates' (2) studies. Daily coffee intake usually begins at approximately 20 years of age in the context of adult student life or incorporation into the job market. Nevertheless, the median age at baseline in both studies was older than 50 years. Persons from the source populations may thus have been exposed to coffee for a median of 30 years before study entry. This prevalent user design may have allowed for a decrease in the number of susceptible persons (who may have had events before qualifying for study entry) from the coffee intake groups, resulting in an exposed study population of healthy survivors. Incident user designsthat is, those that are exposure-naiveare considered the gold standard in observational pharmacoepidemiology, because the selection bias derived from inclusion of prevalent users is minimized and the associations are more consistent with those from randomized trials (3). Of interest, in an analysis of the Johns Hopkins Precursors Study by Klag and colleagues (4) in which the baseline age was younger than 30 years (that is, closer to an incident user design), there was a strong association between coffee use and cardiovascular events in multivariable analyses. The risk for residual confounding is also worth mentioning. Socially conditioned behaviors, such as coffee intake, are particularly challenging and subject to confounding by socioeconomic status (SES). As with studies of moderate alcohol consumption, these behaviors are strongly prevalent in groups with higher SES, and higher SES in itself is a protective factor for health outcomes (5). Although the authors did adjust for education, SES is a complex, latent construct for which proxies (such as education, income, or occupation) are often measured erroneously when measured at all and may not fully capture the effect of SES. Of note, Gunter and colleagues' (1) sensitivity analysis assessing robustness to residual confounding found that a risk factor with a hazard ratio of 1.50 (within the realm of the feasible effect sizes of poverty [5]) needs only a 20% difference in the confounder between groups. Because of these limitations, we believe that the reports of these studies in the lay press may have exaggerated their causal significance. Future research on coffee intake and health should aim to include younger persons and use more comprehensive measurements of SES.