|
Affiliation
|
Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada. |
|
Abstract
|
BACKGROUND AND AIMS: Evidence suggests that adult per capita alcohol consumption, as estimated from self-reports of nationally representative surveys, underestimates 'true' consumption, as measured as the sum of recorded and unrecorded consumption. The proportion of total adult alcohol per capita reported in representative surveys is usually labelled coverage. The aim of the present paper was to estimate coverage for South Africa under different scenarios of alcohol use assessment and 'true' consumption.
DESIGN: Five nationally representative surveys from South Africa were used to estimate the prevalence of drinking and the grams per day among current drinkers. All surveys used a complex multistage sampling frame that was accounted for by using survey weights. The total (recorded and unrecorded), the recorded, and the adjusted total adult per capita alcohol consumption in South Africa served as different estimates of the 'true' consumption. SETTING: South Africa. PARTICIPANTS: Alcohol use information was assessed on a total of 8,155, 16,398 and 13,181 adults (15 years or older) in surveys from the years 2003, 2005, and 2008, respectively. Two surveys in 2012 included 27,070 and 18,688 adults. MEASUREMENTS: Coverage rates of the alcohol use reported in the surveys, compared with the 'true' adult per capita alcohol were calculated.
FINDINGS: The survey data covered between 11.8% (2005; 95% confidence interval (CI) 9.3-16.2%) to 19.4% (2003; 95% CI 14.9-24.2%) of the total alcohol used per capita. The highest coverage of 27.9% (95% CI 22.4-36.8%) was observed when looking only at recorded alcohol in 2003.
CONCLUSIONS: Evidence from five nationally representative surveys assessing alcohol use suggests that less than 20% of the total adult per capita alcohol consumption in South Africa get reported in surveys.
This article is protected by copyright. All rights reserved.
Language: en |