Methods and definitions

There are currently two sets of questions that can be used to estimate alcohol consumption based on conversion factors in the HSE. One relates to drinking in a typical week and the other to the maximum amount of alcohol consumed on any day in the previous week (see Appendix A for further information). In both modules participants who drink alcohol are asked about a range of alcohol products and drink sizes which are then converted into units of alcohol using a standard conversion table, as detailed below:

• the frequency of drinking different types of drink and the amount of each drunk on a typical day in the last 12 months (providing average usual weekly consumption)
• the amounts of different types of alcohol drunk on the day they drank most (providing maximum alcohol consumption on any day in the last week)

To examine usual weekly consumption, adults who drank alcohol are divided into three groups:

• lower risk (up to 14 units for men and women)
• increasing risk (above 14 units and up to 50 units for men, above 14 and up to 35 units for women)
• higher risk (above 50 units a week for men, above 35 units for women)

To examine the maximum amount drunk on any day in the last week, adults are divided into four groups:

• none
• up to and including 3 units (for women) / 4 units (for men)
• more than 3 units (for women) / 4 units (for men), up to and including 6 units (for women) / 8 units (for men)
• more than 6 units (for women) / 8 units (for men)

The usual weekly consumption classification is based on the current guidelines to low risk drinking (UK Chief Medical Officers’ Low Risk Drinking). The maximum daily consumption classification is based on previous guidance (Department of Health. Sensible drinking) since the current guidelines no longer include specific daily limit recommendations. Both these classifications were used in HSE 2019 and 2021 reports.

One unit of alcohol is 10ml by volume of pure alcohol. Table A below shows the conversion factors used between 2006 and 2021 (hereon in referred to as ‘original’) and revised conversion factors for converting what and how much of it was drunk into units of alcohol. Based on the HSE Alcohol Working Group recommendations, increases in the unit conversion factors were applied to the following beverages:

• normal strength beer, lager, stout, cider, shandy (pints and large cans)
• strong beer, lager, stout, cider (large cans)
• wine (bottles, small, medium, and large glasses)

Conversion factors for other beverage sizes remained unchanged. Further details on the rationales for the updated conversion factors are outlined in Appendix A.

Data from HSE 2019 has been used as the basis for this evaluation ahead of the HSE 2022 report, since the data collection methods were similar in both survey years with interviews occurring face-to-face in participants’ homes. Owing to the Covid-19 pandemic, the fieldwork for the 2020 survey ceased in March. In 2021, the data collection approach differed from previous years. This included a change in mode, from face-to-face interviewer visits to remote telephone and video interviews, to limit contact between participants and interviewers. Therefore, HSE 2021 estimates are considered not directly comparable with previous years (HSE 2021 Methods report). Comparative analyses were carried out with HSE 2021 to investigate whether the mode of data collection affected the impact of the revised conversion factors, and whether discrepancies differed by survey year.

Usual weekly alcohol consumption and the maximum alcohol consumed on any day of the last week were estimated using the revised conversion factors and compared with published estimates in HSE 2019 and HSE 2021, which used original conversion factors. This was initially done for only adults that drank alcohol, as the alcohol unit conversion factors apply only to those who reported drinking alcoholic beverages. The numbers of non-drinkers remain unchanged.  The same analyses were carried out among all adults, to establish the prevalence of alcohol consumption in the general population after applying the revised conversion factors. The data tables show estimates from the original and revised conversion factors for men, women and age group separately, to illustrate the difference between the conversion factors. As in each survey year, estimates are subject to a margin of error and standard errors and 95% confidence intervals around the new estimates are also presented in the tables.

The data is considered paired since estimates using both original and revised conversion factors occur for each participant. Therefore, appropriate statistical tests were used for paired data (paired t-test for continuous measures, or a McNemar test for proportions) to examine whether differences between the estimates using the original and revised conversion factors were statistically significantly different among adults, men and women.

In addition, to examine whether the impact of using the revised conversion factors differed by sex and age group, regression model analyses were conducted. Linear regression was used to estimate the differences in mean consumption by sex, age, and an age and sex interaction. Logistic regression models were used to estimate the odds of adults exceeding thresholds when previously they had not, compared with no change in classification following the revision, after adjusting for sex, age, and an age and sex interaction. A Wald test was carried out to test the overall significance of age group rather than assessing statistical significance of specific sub-groups. Complex survey design and non-response weighting were applied to analyses.