W Liang1, T Chikritzhs. 1. National Drug Research Institute, Curtin University, Perth, Western Australia, Australia. w.liang@curtin.edu.au
Abstract
BACKGROUND: Over several decades, many cohort studies from the medical epidemiology literature have observed that compared with abstainers, moderate drinkers experience lower risk for a range of diseases. It was very difficult to separate the hypothesised physiological protective effect of moderate drinking from its well-correlated confounders in observational study settings. AIMS: To investigate the association between current alcohol consumption levels of randomly selected family members and the current self-reported health status of other members living within the same family, using a large-scale representative general population survey. METHOD: Poisson regression models of the association between randomly selected key respondent alcohol consumption and health status of cohabiting family members using data from the 2008, 2009 and 2010 National Health Interview Surveys. Self-reported alcohol consumption of randomly selected key participants and self-reported health status of adult and child (parent reported) family members living in the same household were measured and compared. RESULTS: After controlling for a large range of commonly reported confounders, inverse associations were evident between light and moderate alcohol consumption of key participants and the prevalence of adverse health status among their family members, including children. CONCLUSIONS: The superior health status attributed to family members of light and moderate drinkers is highly likely to be spurious and due to residual confounding rather than physiologically protective effects of alcohol. Unaccounted for confounding is likely to underpin apparent 'protective effects' due to moderate drinking commonly reported from observational studies of all-cause mortality, heart disease, stroke and other chronic diseases.
RCT Entities:
BACKGROUND: Over several decades, many cohort studies from the medical epidemiology literature have observed that compared with abstainers, moderate drinkers experience lower risk for a range of diseases. It was very difficult to separate the hypothesised physiological protective effect of moderate drinking from its well-correlated confounders in observational study settings. AIMS: To investigate the association between current alcohol consumption levels of randomly selected family members and the current self-reported health status of other members living within the same family, using a large-scale representative general population survey. METHOD: Poisson regression models of the association between randomly selected key respondent alcohol consumption and health status of cohabiting family members using data from the 2008, 2009 and 2010 National Health Interview Surveys. Self-reported alcohol consumption of randomly selected key participants and self-reported health status of adult and child (parent reported) family members living in the same household were measured and compared. RESULTS: After controlling for a large range of commonly reported confounders, inverse associations were evident between light and moderate alcohol consumption of key participants and the prevalence of adverse health status among their family members, including children. CONCLUSIONS: The superior health status attributed to family members of light and moderate drinkers is highly likely to be spurious and due to residual confounding rather than physiologically protective effects of alcohol. Unaccounted for confounding is likely to underpin apparent 'protective effects' due to moderate drinking commonly reported from observational studies of all-cause mortality, heart disease, stroke and other chronic diseases.
Authors: Annemieke Ruitenberg; John C van Swieten; Jacqueline C M Witteman; Kala M Mehta; Cornelia M van Duijn; Albert Hofman; Monique M B Breteler Journal: Lancet Date: 2002-01-26 Impact factor: 79.321
Authors: Kristi Reynolds; Brian Lewis; John David L Nolen; Gregory L Kinney; Bhavani Sathya; Jiang He; Brian L Lewis Journal: JAMA Date: 2003-02-05 Impact factor: 56.272