MEASURING ALCOHOL CONSUMPTION IN THE 2007-08 NATIONAL HEALTH SURVEY

In order to estimate levels of alcohol consumption, respondents in the 2007-08 National Health Survey (NHS) were asked about their alcohol consumption in the last week and over the last 12 months. See the National Health Survey: Data Reference Package, 2007–08 (cat. no. 4363.0.55.002) for a copy of the questionnaire used, as well as the National Health Survey: Users' Guide, 2007-08 (cat. no. 4363.0.55.001) for more information.


CALCULATING LONGER TERM RISK ACCORDING TO THE 2001 GUIDELINES

To measure risk in the longer term according to the 2001 National Health and Medical Research Council (NHMRC) guidelines, the NHS asks questions about how recently a person had consumed alcohol, and how frequently in the last 12 months. If they had consumed alcohol in the last week, they were then asked on which days of the week they had consumed alcohol. These respondents were asked further, detailed questions about the alcoholic beverages they had consumed on each of the 3 most recent days on which they had consumed alcohol. This information was used to calculate their average consumption over those 1–3 days, from which their average daily alcohol consumption over the last week (in standard drinks) was estimated. It should be noted that in this series of questions, respondents were not asked about standard drinks; rather, they were asked about the type, brand, number and serving sizes of the beverages they consumed, from which the number of standard drinks was derived.

Using the estimated average daily alcohol consumption over the last week, respondents were grouped into three categories of relative risk level: low, risky or high risk, based on the 2001 NHMRC guidelines for minimising risk in the longer term. Individuals whose consumption placed them in the risky and high risk categories had exceeded the recommended guidelines.

Diagram 1 illustrates how respondents were grouped into categories of longer term risk based on the 2001 guidelines.




According to the 2001 guidelines, in 2007-08 there were 2.1 million persons (13.4% of persons aged 18 years or over) who consumed alcohol at risky or high risk levels in the longer term, based on consumption of alcohol in the last week. A further 492,900 people (3.1%) who did not consume alcohol in the last week, normally consumed alcohol on a weekly basis. It is possible that a proportion of these also consumed alcohol at risky or high risk levels. Likewise there were 2.7 million people (17.2%) who, while they did not consume alcohol in the last week, normally consumed alcohol but on a less than weekly basis during the last year. It is also possible that a proportion of these people consumed alcohol at risky or high risk levels.

At the same time, there were 1.6 million people (10.0%) who had never consumed alcohol and a further 938,700 people (6.0%) who last consumed alcohol 12 or more months ago.


CALCULATING LIFETIME RISK ACCORDING TO THE 2009 GUIDELINES

The questions asked in the 2007-08 NHS can also be used to estimate the number of people who exceeded or did not exceed the 2009 NHMRC lifetime guidelines. As for the 2001 guidelines, this is based on their average daily alcohol consumption over the most recent 3 days on which they consumed alcohol.

Diagram 2 illustrates how respondents were grouped according to the 2009 guidelines.


According to the 2009 guidelines, in 2007-08 there were 3.3 million persons (20.9% of persons aged 18 years or over) who exceeded the recommended guidelines for minimising risk in the longer term. This is greater than the number exceeding the 2001 guidelines for longer term risk (2.1 million persons, or 13.4%). The difference is due to the lower threshold of standard drinks at which a male exceeds the 2009 guidelines compared with the 2001 guidelines.

As for the 2001 guidelines, it is possible that a proportion of the people who normally drank alcohol weekly but did not do so in the previous week also exceeded the 2009 guidelines.

	Persons exceeding guidelines '000	Proportion %
2001 longer term guidelines	2 109.9	13.4
2009 lifetime guidelines	3 294.0	20.9
(a) Persons aged 18 years or over.

Issues potentially affecting data quality

The collection of accurate data on consumption of alcohol is complex, particularly where recall is concerned. In general, under-reporting of consumption, both in terms of persons identifying as having consumed alcohol in a specific period, and in the quantities reported, is a common occurrence in population surveys such as the National Health Survey (NHS).

The overall extent to which under-reporting occurs and its effect on the accuracy of survey estimates is not known, although investigations in relation to previous National Health Surveys have indicated possible under-reporting to be as high as 50% for some beverages. For example, when compared to 'apparent' consumption of alcohol (essentially, statistics on the production and supply of alcoholic beverages, which may overestimate actual consumption) the NHS appears to underestimate consumption of alcohol. Any under-reporting however does not invalidate survey results as indicators of relative consumption levels (current and over time), or the relative health risks of the consumption levels identified.

Two points should be noted in regard to the use of alcohol consumption over the last week to assess longer term/lifetime health risks in the 2007-08 NHS:

• while both the 2001 and 2009 NHMRC risk levels relate to consumption on an ongoing basis, estimates of the number of people exceeding the guidelines relate to individuals' alcohol consumption during the last week only, and do not take into account whether this was more than, less than, or similar to their usual consumption;
• there is a relatively large number of people (3.2 million persons, or 20.5% of all persons aged 18 years or over) who did not consume alcohol in the last week but did in the last 12 months (see Diagram 1). While it is not possible to calculate longer term risk for these people, it is likely that a proportion did consume alcohol at levels exceeding the guidelines.

As the 2007-08 NHS was conducted over an 11-month period, while data may not necessarily reflect usual drinking behaviour of respondents at the individual level, at the population level, data are expected to be representative of the total population.

While respondents were asked about all days in the previous week on which they had consumed alcohol, details of consumption (type, brand and quantity) were collected for the 3 most recent days of consumption only. Some further issues that should be considered in relation to this method are:

• recording consumption for up to 7 days may provide different results to 3 days, noting factors such as respondent recall (whereby accuracy of recall for earlier days in the reference week may be lower than that for more recent days) and respondent burden (due to the potentially large amount of information collected for each day, asking questions for up to 7 days may be perceived by respondents as onerous, thereby creating the potential for avoidance of questions and/or changes in responses, resulting in lower quality data);
• more people were interviewed early in the week, which may have resulted in an under-representation of mid-week drinking occasions and an over-representation of weekend drinking occasions; and
• the proportion of persons reporting drinking was highest on weekends (with 64% of adults reporting drinking on Saturday, compared with 38%-40% for Monday to Thursday), while the proportion of persons providing consumption details was also highest for weekends and Monday (with 70%-78% of adults who drank on either Saturday, Sunday or Monday providing consumption details for those days, compared with 48%-64% for Tuesday to Friday). See the National Health Survey: Users' Guide, 2007-08 (cat. no. 4363.0.55.001) for more information.

Analysis has indicated that the 3-day methodology has a small effect on the overall level of health risk at the population level, however, as this is considered to be stable over time, analysis of relative risk levels over time are not affected. It is expected that further analysis will be conducted when results of the 2011-13 Australian Health Survey become available.


CALCULATING SHORT TERM RISK ACCORDING TO THE 2001 GUIDELINES

To measure short term risk according to the 2001 guidelines, respondents who had consumed alcohol in the last 12 months were asked about single occasion drinking behaviour over the last 12 months. In particular, they were asked how often (in times per week or times in the last 12 months year) their consumption exceeded the following levels:

• for males, drinking 7 or more standard drinks on a single occasion, and for females, 5 or more standard drinks - classified as 'risky'; and
• for males, drinking 11 or more standard drinks on a single occasion, and for females, 7 or more standard drinks - classified as 'high risk'.

For the calculation of short term risk, this recall method of estimating drinking patterns over the last 12 months is considered more reliable than the 3-day method used for longer term/lifetime risk.

Diagram 3 illustrates how respondents were grouped into categories of short term risk based on the 2001 guidelines.




In 2007-08 there were 5.9 million persons (37.7% of all persons aged 18 years or over) who consumed alcohol at risky or high risk levels in the short term, according to the 2001 guidelines, on at least one occasion during the last 12 months.


CALCULATING SINGLE OCCASION RISK ACCORDING TO THE 2009 GUIDELINES

Using the existing questions to measure short term risk, it is not possible to calculate from the 2007-08 NHS the total number of persons who consumed alcohol at levels that exceeded the 2009 NHMRC guidelines for single occasion risk. The number of females exceeding the guidelines can be calculated, as the threshold for females (5 or more standard drinks) has not changed (see Table 2). For males, however, the new and lower threshold of 'more than 4 standard drinks' in the 2009 guidelines cannot be assessed, as male respondents were only asked questions about occasions on which 7 or more, or 11 or more, standard drinks were consumed.

Questions in the 2011-13 Australian Health Survey have been designed to allow calculation of single occasion risk according to both the 2001 and 2009 guidelines, for both males and females.

	Persons exceeding guidelines '000	Proportion %
2001 short term guidelines
Males	3 543.2	45.6
Females	2 388.7	29.9
Persons	5 931.9	37.7
2009 single occasion guidelines
Males	na	na
Females	2 388.7	29.9
Persons	na	na
(a) Persons aged 18 years or over. na not available