Under-reporting of alcohol consumption in household surveys: a comparison of quantity-frequency, graduated-frequency...

To compare alternative survey methods for estimating (a) levels of at risk alcohol consumption and (b) total volume of alcohol consumed per capita in comparison with estimates from sales data and to investigate reasons for under-reporting.

The homes of respondents who were eligible and willing to participate.

A total of 21,674 Australians aged 14 years and older.

A 2001 national household survey of drug use, experiences and attitudes with weights applied for age, sex, geographic location and day of week of interview.

Self-completion questionnaire using quantity-frequency (QF) and graduated-frequency (GF) methods plus two questions about consumption 'yesterday': one in standard drinks, another with empirically based estimates of drink size and strength.

The highest estimate of age 14 + per capita consumption of 7.00 l of alcohol derived from recall of consumption 'yesterday' or 76.8% of the official estimate. The lowest was QF with 49.8%. When amount consumed 'yesterday' was recalled in standard drinks this estimate was 5.27 l. GF questions yielded higher estimates than did QF questions both for total volume (5.25 versus 4.54 l) and also for the proportion of the population at risk of long-term alcohol-related harm (10.6%versus 8.1%). With the detailed 'yesterday' method 61% of all consumption was on high risk drinking days.

Questions about typical quantities of alcohol consumed can lead to underestimates, as do questions about drinking 'standard drinks' of alcohol. Recent recall methods encourage fuller reporting of volumes plus more accurate estimates of unrecorded consumption and the proportion of total alcohol consumption that places drinkers at risk of harm. However, they do not capture longer-term drinking patterns. It is recommended that both recent recall and measures of longer-term drinking patterns are included in national surveys.