Making sense of alcohol consumption data in Australia

Farhat Yusuf and Stephen R Leeder
Med J Aust 2015; 203 (3): 128-130. || doi: 10.5694/mja15.00151

Has alcohol consumption in Australia increased in recent years?

Estimating the consumption of alcohol by individuals and societies is notoriously difficult, especially as it frequently relies on self-reported data. Much depends on the quality of the data-gathering instrument, the questions asked, and the veracity of the survey sample.

Please login with your free MJA account to view this article in full

  • Farhat Yusuf
  • Stephen R Leeder

  • 1 University of Sydney, Sydney, NSW
  • 2 Macquarie University, Sydney, NSW

Competing interests:

No relevant disclosures.


remove_circle_outline Delete Author
add_circle_outline Add Author

Do you have any competing interests to declare? *

I/we agree to assign copyright to the Medical Journal of Australia and agree to the Conditions of publication *
I/we agree to the Terms of use of the Medical Journal of Australia *
Email me when people comment on this article

access_time 11:39, 3 August 2015
Michael Kemp

What exactly are the "weighted daily averages" of drinking? Is it average over the three days the respondents were asked in detail about, or the extrapolated average over 7 days? There is a difference since as the ABS site says ( more interviews were done earlier in the week so weekends (when alcohol consumption is highest) were over-represented. This inflates these values a bit however measured - but more so if averaged over 3 days rather than 7. The figures are useful for comparisons (as the article uses), but I wonder about the magnitude. As the ABS puts it "ABS analysis indicated that the 3 day methodology has a small impact on the overall level of health risk at the population level, however, as the effect is considered to be stable over time, analysis of relative risk levels over time should not be affected. " (4363.0.55.001 - Australian Health Survey: Users' Guide, 2011-13 -> Health Risk Factors -> Alcohol).

Also, are the 40% of respondents who did not consume alcohol in the last week included in these daily averages? If not, then again these averages are inflated.

Competing Interests: No relevant disclosures

Dr Michael Kemp
Charles Sturt University

access_time 05:14, 11 August 2015
Michael Livingston

Yusuf and Leeder (1) showed alcohol consumption increased in Australia between 2001 and 2011 when examining ABS National Health Surveys (NHS). However, this is a selective and flawed picture of Australian drinking trends.

Between 2001 and 2011, per-capita alcohol consumption figures produced by the ABS declined (from 10.22l/person to 10.04 (2)). Further, these data show an increase from 2001 to 2007, followed by a sharp decline of 8.9% between 2007 and 2013 -- a decline matched closely in data from the National Drug Strategy Household survey (10.5%, unpublished analyses).

By analysing just two waves of the NHS data, the authors miss a significant turning point in consumption trends, led by declines among young drinkers (3). Further analyses using 2004/05 and 2007/08 NHS would provide a clearer picture of consumption trends (e.g. ABS reports on the NHS show a recent decline in risky drinking (4)), but these surveys were not included.

Alcohol consumption is measured idiosyncratically in the NHS using flawed measures involving the last three drinking occasions (within the past week), meaning that estimates are biased by the day of the week that the survey is conducted (5). Further, Yusuf and Leeder appear to report consumption per person who drank in the week before the survey (rather than per capita), making comparisons between teenagers and adult drinkers problematic, given their markedly different drinking frequencies.

Relying solely on NHS data to estimate Australian alcohol consumption trends is problematic; a more complete picture than that presented in Yusuf and Leeder is required.

1. Yusuf, F. & Leeder, S. R. (2015) Making sense of alcohol consumption data in Australia, Medical Journal of Australia, 203, 128-130.

2. Australian Bureau of Statistics (2015) Apparent Consumption of Alcohol, Australia, 2013-14 (Canberra, Australian Bureau of Statistics).

3. Livingston, M. (2014) Trends in non-drinking among Australian adolescents, Addiction, 109, 922-929.

4. Australian Bureau of Statistics (2012) Australian Health Survey: First Results, 2011-12 (Canberra, ABS).

5. Donath, S. (1999) Estimated alcohol consumption in the 1995 National Health Survey: Some methodological issues, Australian and New Zealand Journal of Public Health, 23, 131-134.

Competing Interests: No relevant disclosures

Dr Michael Livingston
National Drug and Alcohol Research Centre, University of New South Wales

access_time 06:19, 13 August 2015
Farhat Yusuf

“Average alcohol consumption converted into standard drinks - over days recorded” was the only suitable variable that was available in the confidentialised unit record files for the 2001 and 2011-12 National Health Surveys.
The weighted daily averages were based on responses to questions about alcohol consumption (in standard drinks) over the three most recent days when alcohol was consumed during the week prior to the interview (see the titles and Y-axes of Boxes 1 and 2). Persons not reporting alcohol consumption during the past week were not included in the denominators of these weighted averages.
Apart from collecting the consumption data for each day of the week - which ABS did not do –converting a 3-day into a 7-day average by extrapolation will not necessarily provide an accurate estimate of the weekly average. In any case, such a conversion would not have changed our findings regarding the increased alcohol consumption over the 10-year period.

Competing Interests: No relevant disclosures

Prof Farhat Yusuf
The University of Sydney and Macquarie University

access_time 06:22, 13 August 2015
Farhat Yusuf

Livingston and Dietze’s second reference contains the following two statements from the ABS:
1. “Estimates of 'apparent consumption' are obtained from information related to supply (for example, excise data on alcohol produced for domestic consumption, and data on imports) and do not represent actual consumption.” (See: Media Release section).

2. “This estimation differs from personal consumption estimates from a survey from which individual and group patterns may be derived.” (See: About This Publication section).
Since both NHS were each conducted over two financial years (2001 NHS from February to November 2001 and the 2011-12 NHS from March 2011 to March 2012), the apparent consumption figures corresponding to them were 10.12 and 10.17 litres respectively (derived from data in Table 7 of the second reference). These figures corroborate our finding of an increase in alcohol consumption over the 10-year period rather than the claims made by Livingston and Dietze.
We used only the 2001 and 2011-12 NHS and not the two intervening surveys conducted in 2004-05 and 2007-08 as Livingston (reference 3) did in his paper by ignoring the 2004 and 2007 National Drug Strategy Household surveys.
The criticisms made by Donath (reference 5) of AHS methods referred to the 1995 NHS and do not necessarily apply to the surveys used in our paper. Finally, the consumption measures used in the NHS were quite similar to measures used in surveys in UK, USA and elsewhere.

Competing Interests: No relevant disclosures

Prof Farhat Yusuf
The University of Sydney and Macquarie University

Responses are now closed for this article.