Risks associated with alcohol consumption may be exaggerated
The controversy over whether any amount of alcohol consumption is safe continues, as authors of a research letter published in JAMA Internal Medicine suggest that a previous study may have exaggerated the risks associated with alcohol consumption by not considering underreporting.
“Systematic measurement error in risk factor assessment is a common problem in medical studies,” Mary C. Vance, MD, MSc, an assistant professor in the department of psychiatry at Uniformed Services University, told Healio Primary Care. “Survey studies are particularly prone to this problem, because self-reports often contain measurement error due to recall, nonresponse and social desirability biases.”
More recently, Vance and colleagues conducted a quality improvement study, using a sensitivity analysis to assess the degree of misclassification that would lead to bias in estimating the risk associated with low levels of alcohol consumption.
To that end, they calculated the “true relative risk [RR]” for cardiac or all-cause mortality using the Lancet study and another major report.
Vance and colleagues focused on two types of drinking: at-risk drinking, which was defined as having three to five drinks per day, and heavy drinking, defined as having six or more drinks each day. The researchers assumed two types of underreporting occurred in the previous studies — one in which at-risk drinkers underreported by one to two drinks per day, and another in which heavy drinkers underreported their actual consumption dramatically to hide their drinking problems.
Vance and colleagues found that, if 30% of those who consumed at least three drinks each day had downplayed their alcohol consumption, then the observed RRs of those who reported one to two drinks per day were about the same as the risks observed in the Lancet study. If 40% of at-risk and heavy drinkers underreported their alcohol consumption, the observed RRs began to exceed those observed in the study.
“If 30% of at-risk drinkers underreport their drinking by one to two drinks per day, and if 30% of heavy drinkers try to hide their drinking by saying they only drink one to two drinks per day, this underreporting could lead to biased study findings that misrepresent a reality in which one drink per day reduces the risk of adverse health outcomes by 5%, and two drinks per day result in no net harm,” the researchers wrote.
Vance said that although it is generally accepted that systematic biases can result in overestimations or underestimations of population averages — in this case, how many drinks a person has each day — “it can be easy to overlook their impact on associations between the risk factor and outcomes of interest — eg, overestimation of the association between a low amount of alcohol use and risk of adverse outcomes, especially in the likely case that many heavy drinkers significantly underreport their true level of alcohol use.”
She added that “when researchers have a good reason to suspect systematic information bias in risk factor assessment, we recommend, at the minimum, a sensitivity analysis to assess the impact of that misspecification on the observed associations between risk factors and outcomes.”
“Applied to the study of risks associated with alcohol use, future studies would benefit from taking into account the potential impact of inaccurate reporting of alcohol use in survey data in their statistical analysis and interpretation of results,” she continued. – by Erin Michael
Disclosures: Vance reports no relevant financial disclosures. Please see study for all other authors’ relevant financial disclosures.