Exploring the characteristics of newly defined at-risk drinkers following the change to the UK low risk drinking guidelines: a retrospective analysis using Health Survey for England data

In England, alcohol guidelines form part of a suite of population level interventions to address problematic alcohol consumption [1, 2]. They are used to aid individuals and healthcare professionals to assess the health risks associated with alcohol consumption [3, 4]. The UK Low Risk Drinking Guidelines were changed in 2016, reducing the maximum weekly alcohol consumption guideline down from 21 to 14 units per week for men, due to new evidence of the health risks associated with drinking at lower levels [3]. To assist with targeted screening for alcohol risk in healthcare services and targeted alcohol education programmes, this study used annual, nationally representative survey data to retrospectively explore how many men would have been redefined from low risk to at-risk drinkers according to the new guidelines, which demographic characteristics were associated with being an at-risk drinker before and after the change in guidelines and whether the characteristics associated with being an at-risk drinker changed under the new guidelines.

Globally, alcohol causes approximately 5.9% of deaths each year and is linked to over 200 disease and injury conditions, accounting for 5.1% of the global burden of disease [5] and presenting a significant public health risk. In England in 2016, over 23,000 deaths were alcohol-related [6] and despite a gradual decrease in the reported alcohol consumption of men in England [7], the number of hospital admissions linked to alcohol consumption in England continues to rise [8]. In addition to health consequences for the drinker, alcohol is linked to numerous other consequences such as violence and crime [9, 10], child abuse and neglect [11, 12], reduced productivity and unemployment [13, 14] and many more intangible consequences such as reduced quality of life, pain and suffering, both for the drinker and those around them [15].

In response to concerns over alcohol-related deaths and illnesses, the first UK guidelines on individual drinking were produced in 1984 in That’s the Limit, with ‘safe limits’ described as 18 standard drinks (equivalent to 18 units) for men and 9 standard drinks for women [16]. Alcohol units (where 1 unit = 8 g pure alcohol) were introduced three years later along with a ‘sensible limits’ guideline of 21 units per week for men and 14 units per week for women, with ‘too much’ defined as 36 and 22 units respectively [16]. These guidelines continued until 1995, when, despite the British Medical Association [17] and the Royal Colleges of Physicians, Psychiatrists and General Practitioners [18] endorsing the continuation of the 21/14 units per week guideline, the Sensible Drinking guidelines changed to not regularly drinking more than 3 to 4 units per day for men and 2 to 3 units per day for women [19]. Whilst the move to daily limits caused concern amongst critics who viewed the change as, at best, being irrelevant and therefore ignored by the majority of drinkers who drink up to twice per week [20, 21] and at worst, encouraging daily drinking [22], all whilst significantly increasing the weekly limit [23]; in reality, the 21/14 units per week message has persisted alongside the daily guideline. This was demonstrated following the introduction of the latest guidelines in 2016, recommending that men and women drink no more than 14 units per week [3], when numerous headlines, including one from the Department of Health [24] reported a reduction in the weekly guidelines for men from 21 to 14 units [25, 26].

The latest guideline review was prompted by new evidence (since the 1995 guidelines) regarding the health risks associated with alcohol consumption, particularly related to cancer risks, where the risk of developing certain cancers (including, but not limited to cancers of the throat, mouth and breast) is now known to increase even at low levels of alcohol consumption [20]. The new guidelines were developed following a consultation which reviewed the recent evidence [27] and used the Sheffield Alcohol Policy Model to review morbidity and mortality risk estimates related to different alcohol consumption levels [28]. The resulting Low Risk Drinking Guidelines state that, for both men and women: “To keep health risks from alcohol to a low level it is safest not to drink more than 14 units per week on a regular basis” (3, p4) with additional guidance around spreading these units across three or more days. The decision to align the male and female weekly guideline was taken following evidence that at lower levels of alcohol consumption, the risks for men and women are similar, with men being more vulnerable to shorter-term risks and women more vulnerable to longer-term risks in general [3].

The aim of drinking guidelines is to inform the public in order for people to make informed decisions about their alcohol consumption [3, 29]. Whilst drinking guidelines have been criticised for their ineffectiveness in reducing alcohol consumption or changing drinking behaviour [30, 31], the value of drinking guidelines as part of a suite of education and information interventions, which also include health warnings on alcohol containers, counter-advertising and school interventions, has been recognised [1, 30]. Research into public awareness of changes to alcohol guidelines in the UK has repeatedly found that awareness peaks around the time of the change but is not sustained and long-term promotional campaigns have been recommended [32‐34].

In addition to informing the public, drinking guidelines enable healthcare practitioners to assess and advise patients on their alcohol consumption. Screening for at-risk drinkers (usually either defined as drinking above national guidelines or identified via alcohol risk screening tools such as the Alcohol Use Disorders Identification Toolkit (AUDIT) [35]) and offering brief interventions for alcohol use, such as brief structured advice or brief motivational interviewing [36] have demonstrated positive results [37]. Whilst statistical modelling to predict population-level benefits of screening and brief interventions support universal screening [36], National Institute for Health and Care Excellence (NICE) guidance recommends universal screening if feasible and targeted screening of groups at greater risk of alcohol-related harm (e.g. people with relevant physical or mental health problems) if universal screening is not feasible [4]. Widespread implementation of screening has proved problematic both in trials of screening and brief intervention [38] and in clinical practice [39], and policy in England has favoured targeted screening in general practice settings, including the incentivisation of primary care services in England between 2008 and 2015 to deliver ‘Identification and Brief Advice’ for new patients only [36].

With an increasing focus within healthcare on targeting at-risk drinkers who consume above the low risk drinking guidelines [4, 37‐39], the recent reduction in the low risk drinking threshold for men from 21 to 14 units per week could have a significant impact on healthcare providers as the number of at-risk drinkers increases. Furthermore, new awareness of the health risks associated with consuming above 14 units of alcohol per week means that supporting people to reduce their consumption to within these levels is a public health priority. Establishing the number of additional men who would be classified as at-risk following the guideline change may assist policy and commissioning decisions relating to screening for at-risk drinkers and provide a further incentive for improving education and awareness around drinking guidelines. Furthermore, within the English policy model of targeted screening for at-risk drinkers, it would be of benefit to know which demographic characteristics are associated with being an at-risk drinker and to establish whether the change in guidelines affected any specific demographic groups. It is feasible that men who were previously adhering to the guidelines may a) be more amenable to adhering to recommended guidelines having previously consumed within the low risk guidelines and b) not be aware that they are no longer considered to be drinking at low risk levels; therefore this group may benefit from specific targeting. Consequently, the objectives of this study were:

Whilst it could be argued that, in line with the 1995 Sensible Drinking guidelines of no more than 3 to 4 units per day for men [19], the change to guidelines constituted a 50% decrease from (a potential) 28 units per week to 14; both the media (e.g. 22–24), and the medical community [17, 18, 23, 40], including those defining increasing risk drinking in order to facilitate brief interventions in healthcare settings [4] viewed this as a reduction of one third in light of the pre-existing weekly limit of 21 units for men. Therefore, this paper will focus on a change in the Low Risk Drinking Guidelines from 21 units per week to 14 units per week for men.

The Health Survey for England (HSE) is an annual, nationally representative, cross-sectional survey using a clustered, stratified multi-stage sampling design to select a random sample of private households in England. The survey is administered face to face within the participant’s household by a trained interviewer. Full details of the HSE methodology can be found online [41]. Ethical approval was sought prior to data collection from the relevant research ethics committee and further ethical approval was not required for this study.

Unweighted and weighted sample sizes are recorded in Table 1. Sample sizes were different in each year, ranging from 3487 to 3790 (unweighted). Whilst exact percentages varied from year to year, the newly defined at-risk group (men drinking > 14 to ≤21 units per week) was consistently the smallest group of men, followed by non-drinkers and previously defined at-risk drinkers (men drinking > 21 units per week), with low risk drinkers constituting the majority of the sample each year.

Table 3 shows the annual population prevalence estimates of adult men in England (2011–2015) drinking at newly defined at-risk levels (> 14 to ≤21 units per week) by age group. The youngest (16–25 years) and oldest (75+ years) age groups consistently had amongst the lowest proportions of men drinking at newly defined at-risk levels, whilst the proportion of men drinking at newly defined at-risk levels in other age groups varied annually. The highest proportion in any age group categorised as newly defined at-risk drinkers was amongst 55 to 64-year-old men in 2015 (15.3%).

The percentage of men aged 16 and over, living in private residences in England who were drinking at newly defined at-risk levels ranged from 11.2% (2,322,896, 5.5% total population) in 2011 to 10.2% (2,182,401 men, 5% of the total population) in 2014.

2. To establish which demographic characteristics in men were associated with being at-risk (> 21 units/week) vs low risk (≤ 21 units/week) drinkers according to the previous guidelines; which demographic characteristics were associated with being at-risk (> 14 units/week) vs low risk (≤ 14 units/week) drinkers according to the new (2016) guidelines, and to explore whether the characteristics associated with being an at-risk drinker were different following the change to guidelines.

Sample characteristics for low risk vs at-risk male drinkers under the previous guidelines (men drinking ≤21 vs > 21 units/week) and low risk vs at-risk male drinkers under the new guidelines (men drinking ≤14 vs > 14 units/week) taken from Health Survey for England 2015 data are shown in Table 2. In all age groups, a higher proportion of men were classified as at-risk drinkers under the new guidelines, in line with the lower weekly threshold. The difference in percentage of at-risk drinkers under the previous vs new guidelines was greatest for 55–74 year olds (29% at-risk under the previous guidelines and 44% at-risk under the new guidelines) followed by 16–34 year olds (19% under the previous guidelines and 32% under the new guidelines).

Within the 2015 HSE sample, the demographic groups with the greatest difference in proportion of at-risk drinkers under the previous vs new guidelines were men working in managerial or professional occupations (26% of men under the previous guidelines vs 41% under the new guidelines), married or cohabiting men (24% vs 37%), men living in the North East (30% vs 46%) and South West (22% vs 38%) of England, white men (25% vs 38%), men who were current (30% vs 44%) or ex-regular (29% vs 43%) smokers and men with non-limiting long-lasting illness (24 vs 39%). For all demographic variables, the groups with the highest percentage of at-risk drinkers remained similar under the new guidelines.

Table 4 shows the results of two logistic regression models: column 1 shows the odds ratios for at-risk vs low risk (reference category) drinkers under the previous guidelines (> 21 vs ≤21 units/week) and column 2 shows the odds ratios for at-risk vs low risk (reference category) drinkers under the new guidelines (> 14 vs ≤14 units/week). All results relate to data from the HSE 2015.

Most demographic characteristics which were significantly associated with at-risk drinking under the previous guidelines remained so under the new guidelines. Under both the previous (OR 1.71, 95%CI 1.25–2.34) and new (OR 1.63, 95%CI 1.25–2.12) guidelines, 55–74 year old men had significantly greater odds of being classified as at-risk (vs low risk) drinkers when compared to the youngest age group (16–34 year olds); the association was slightly attenuated by the lowered threshold under the new guidelines. Under both the previous (OR 0.61, 95% CI 0.40–0.92) and new (OR 0.69, 95% CI 0.48–0.98) guidelines, men over 75 years old had significantly lower odds of being at-risk (vs low risk) drinkers when compared to the youngest age group.

Under the previous guidelines, 35–54 year olds had significantly greater odds of being at-risk drinkers (vs low risk drinkers) compared to the youngest age group (OR 1.42, 95% CI 1.08–1.98); however, under the new guidelines, 35–54 year old men no longer had significantly greater odds of being categorised as at-risk drinkers than the younger age group (OR 1.18, 95% CI 0.92–1.51). This was also the case in the unadjusted regression model (OR 1.22, 95% CI 0.97–1.54).

Men working in managerial or professional occupations had significantly greater odds of being classified as at-risk (vs low risk) drinkers compared to men working in manual or routine occupations under both the previous (OR 1.46, 95% CI 1.15–1.86) and the new (OR 1.64, 95% CI 1.34–2.00) guidelines, with the magnitude of association increasing following the change to the guidelines.

Under the previous guidelines, single men had significantly greater odds of being at-risk (vs low risk) drinkers compared to men who were married/cohabiting (OR 1.41, 95% CI 1.06–1.88); however, under the new guidelines this association was no longer significant (OR 1.28, 95% CI 0.99–1.67). It was also the case within the unadjusted regression model that single men did not have greater odds of being at-risk drinkers than married/cohabiting men under the new guidelines (OR 0.90, 95% CI 0.71–1.13).

Under both the previous and new guidelines, men from the North East (previous: OR 1.75, 95% CI 1.09–2.79; new: OR 2.08, 95% CI 1.38–3.13) and North West (previous: OR 1.78, 95% CI 1.26–2.51; new: OR 1.91, 95% CI 1.41–2.60) of England had greater odds of being classified as at-risk (vs low risk) drinkers than men from the South East of England. Under the new (but not the previous) guidelines, men from the West Midlands (OR 1.68, 95% CI 1.17–2.42) and London (OR 1.53, 95% CI 1.03–2.28) had greater odds of being classified as at-risk drinkers than those in the South East.

For men in the West Midlands, this association was also significant in the unadjusted model (OR 1.46, 95% CI 1.02–2.08). For men living in London, the association was not significant in the unadjusted model (OR 1.29, 95% CI 0.87-1.91), and when all variables were entered in a stepped process, the association was significant only when ethnicity was also included in the model.

Under both the previous (OR 0.44, 95% CI 0.25–0.77) and new (OR 0.53, 95% CI 0.34–0.80) guidelines, non-white men had significantly lower odds of being at-risk drinkers than white men.

Current smokers (previous: OR 2.30, 95% CI 1.72–3.08; new: OR 2.26, 95% CI 1.73–2.94), ex-regular smokers (previous: OR 2.11, 95% CI 1.67–2.67; new: OR 2.01, 95% CI 1.63–2.47) and ex-occasional smokers (previous: OR 2.10, 95%CI 1.36–3.25; new: OR 1.85, 95% CI 1.25–2.74) all had significantly greater odds of being at-risk drinkers under both the previous and the new guidelines. The association was slightly attenuated under the new, lower guidelines.

Neither limiting, nor non-limiting long-lasting illness were significantly associated with being classified as an at-risk drinker, either under the previous or the new guidelines.

As health care services increasingly focus on screening for at-risk drinkers and offering brief interventions in order to reduce their alcohol intake [4, 37‐39], the sudden increase in the number of at-risk drinkers by an additional 2.3 million men (based on 2015 figures) caused by the change to alcohol guidelines could have a significant impact on services, all at a time when incentivisation for identification and brief advice in primary care has been withdrawn [39]. Policy within England favours targeted screening [36]; a decision which is likely to have been influenced by the significant cost and practical implications of universal screening and the decision by the UK National Screening Programme not to recommend screening for alcohol misuse nationally [57]. It is therefore unlikely under the current system that these men will be identified through routine screening, and, particularly in light of the current evidence relating to poor awareness of changes to guidelines in the UK [32‐34], these men may well continue to drink at levels which pose a risk to their health.

Knowing which groups of men have greater odds of being at-risk drinkers, and particularly those groups which would have been redefined as at-risk drinkers following the change to guidelines, could allow for 1) more targeted screening for at-risk male drinkers, and 2) targeted and sustained educational programmes around drinking guidelines.

Under both the previous and the new drinking guidelines, men aged 55–74 had greater odds of being at-risk drinkers, which correlates with current evidence around ‘middle-aged men’ being the most at-risk in terms of their drinking [58] and reinforces the need for more targeted campaigns for this age group. This could also provide an age criterion with which to screen for at-risk drinkers. Under the previous, but not the new guidelines, men aged 35–54 had greater odds of being at-risk drinkers, a change which may be explained by the greater increase in the proportion of at-risk drinkers in the reference category (16–34 year olds, 13% increase in at-risk drinkers) compared to 35–54 year olds (11% increase in at-risk drinkers). Around a third of men in the younger age group would now be considered at-risk from their drinking and the youngest age group experienced the second largest increase in the number of people considered at-risk under the new guidelines, indicating that a considerable proportion of this age group were drinking between 14 and 21 units per week. With this in mind, it is important to ensure that any sustained promotional activity does not focus on one age group to the detriment of others as has been suggested with the former focus on problematic younger drinkers [59].

Further criteria for targeted screening and education might include white men and men working in managerial and professional occupations, who had consistently greater odds of being at-risk drinkers in line with existing evidence [45, 52]. Routes to delivery here might be through screening and education in workplaces. The limited sample size only permitted analysis of a binary ethnicity category and further investigation of at-risk drinking within non-white ethnic groups is warranted. Whilst being single was associated with being an at-risk drinker under the previous guidelines, this was no longer the case under the new guidelines. This may be accounted for by the greater increase in the percentage of at-risk drinkers in the reference category (married/cohabiting men: 13% increase in at-risk drinkers) compared to single men (12% increase in at-risk drinkers). Whilst marriage is often associated with reduced drinking levels among men [60, 61], this finding indicates that drinking habits of married men could be putting them at-risk. Whether the higher number of married/cohabiting men drinking between 14 and 21 units indicates an attempt to drink within (previous) guidelines would require further investigation.

Targeted screening and the offer of brief interventions for alcohol consumption have been implemented differently across the devolved nations. In Wales, clusters of health behaviours are targeted [39] and the finding that current and ex-smokers had more than twice higher odds of being at-risk drinkers under both the previous and new drinking guidelines (consistent with the literature on the common co-occurrence of smoking and alcohol consumption [62, 63]), indicates that England could benefit from a similar system. In Scotland, there has been a stronger focus on screening and intervention in line with a more severe alcohol problem [36], providing evidence of the benefits of tailoring services according to geographical location. The change in guidelines strengthened the association between living in the North East and North West and being an at-risk drinker and under the new guidelines only, living in the West Midlands, and London (when adjusted for ethnicity), were also significantly associated with being an at-risk drinker. Therefore more tailored service provision of screening and interventions for alcohol could be put in place in the North East, North West, West Midlands and London, where men have odds between one and a half and twice higher of being at-risk drinkers.

The change to the Low Risk Drinking Guidelines would have resulted in more than 2 million additional male at-risk drinkers in England. Men aged 55–74, current or ex-smokers, men in managerial or professional occupations and white men had greater odds of being at-risk drinkers under the new and previous guidelines and these are the groups already known to be drinking the most, strengthening the case for targeted screening and education around alcohol consumption for known groups of at-risk men. In addition, this study found that men in the West Midlands and London had greater odds of being at-risk drinkers under the new guidelines, and 16–35 year olds and married men had a marked increase in the proportion of at-risk drinkers. These are all groups that have not previously been identified as at-risk drinkers, and may benefit from specific education around the new Low Risk Drinking Guidelines.