Estimating Levels of Consumption and Patterns of Drinking

General population surveys can provide a wide range of information on where drinking takes place, patterns of consumption, sociodemographic correlates, and alcohol- related harms. At the simplest level of analysis, they can provide information regarding, for example, the number of abstainers in the population in the last 12 months, lifetime abstainers, and consumption of alcohol by gender. There is variation worldwide on the rates of abstinence for men and women.

There are countries in which the rate of abstinence is very high for both genders (e.g., where drinking is prohibited for religious reasons), and countries in which most of the drinking is done by males, with females drinking relatively less. Some European countries present no differences between genders. When rates of abstinence are considered, the average amount of alcohol consumption per drinker can be much greater than expected.

Population-level data cannot, however, identify different drinking patterns in a population, who does the drinking (e.g., which age groups), how patterns relate to socioeconomic characteristics of the population or to gender, where the drinking takes place, when and how it is consumed, if it is concentrated into special occasions of high consumption, such as in festivals or holidays, or is more evenly distributed throughout the year.

Surveys can also indicate the prevalence of high-, moderate-, and low-risk drinkers, according to clearly defined criteria. Patterns of drinking are not uniform across different studies thus, for the global burden of disease study, patterns of drinking have to be defined and estimated. A key informant questionnaire was sent to countries and, after two surveys, data were obtained from 63 countries in all regions but the Eastern Mediterranean.

Four different aspects of drinking were covered by the survey: heavy drinking occasions (e.g., festive drinking at fiestas or community celebrations, the proportion of drinking occasions in which drinkers get drunk, the proportion of drinkers who drink daily or nearly daily, drinking with meals, and drinking in public places.

The results of this and other such surveys were combined with available national or regional survey results (unpublished information or publications in peer-review journals). They were then analyzed using optimal scaling analysis, which is similar to factor analysis but permits the simultaneous inclusion of ordinal and categorical data, to determine the number of underlying dimensions and the relations of items to each dimension. The analysis identified one dimension called a detrimental impact that leads to increased mortality and morbidity.

The countries were then classified into four categories and assigned values from 1 (least risky drinking pattern) to 4 (most risky drinking pattern). Rates of abstention were taken into account separately in the final comparative risk analysis (for the final algorithms for calculating pattern scores, see Rehm et al., 2003b).

Rates of a variety of alcohol-related problems can be explored by surveys and linked to patterns of drinking and amount or frequency of drinking. Information on alcohol use disorders, using validated instruments such as AUDIT (Babor et al., 2001) and CAGE (Ewing, 1984), and diagnostic criteria such as those found in the ICD-10 or DSM- IV, can be of value. This information is important when assessing coverage of treatment services in a particular country and then planning the organization of health treatment systems to respond to the range of alcohol-related problems.

On surveys, the most used and recommended questions regarding alcohol consumption are quantity- frequency, graduated frequency, and recent recall (WHO, in press). Even though there is still little agreement in the literature on which questions to include in surveys, and how to ask them, international collaborative studies have tried to increase comparability of data by agreeing on common indicators on a core number of areas, thus some progress has been made.

Two studies are the GEN- ACIS study (Gender, Alcohol, Culture: An International Study), which included 35 countries from most regions of the world and assessed alcohol consumption and related problems from the adult general population, and the ESPAD (European School Survey Project on Alcohol and Drugs), which collected information from school students 13-15 years of age in European countries in 1995, 1999, and 2003, using the same basic questionnaire.

There are numerous methodological issues to be considered when planning, undertaking, and interpreting data from surveys. These include reference period, beverage- specific or overall questions on consumption, quantity per drinking day versus quantity per drinking occasion, drink size and alcohol content of alcoholic beverages (and mixed drinks), criteria for defining a drinker or nondrinker, validity and reliability of the survey instrument in the absence of international standardization, criteria for risk drinking (on a single occasion and average daily consumption, for acute or chronic problems), sample selection, and sample size, among others. The WHO guide (2000) is a good source of information on these methodological issues and how to address them.