Public health engagement in alcohol licensing in England and Scotland: the ExILEnS mixedmethod, natural experiment evaluation

1.1.1 Alcohol availability and harm

Systematic reviews, and reviews of reviews, have concluded that controlling the ease with which alcohol can be obtained can be effective in reducing alcohol-related harms. 25–27 As few territories in the world have a total prohibition on alcohol, most of the evidence on alcohol control relates to interventions that regulate and modify alcohol availability. These ‘alcohol availability’ interventions take a variety of approaches: for example, the number and proximity of alcohol outlets (physical availability), their hours and days of sale (temporal availability), restricting specific population subgroups from purchasing alcohol (e.g. age restrictions). 28 While governments commonly control alcohol availability around the world, the system, type, degree, and location of control (i.e. local, regional, national, supranational) varies by jurisdiction. There is consistent evidence of an association between increased physical29,30 and temporal availability31–33 of alcohol and higher rates of consumption and associated alcohol-related harms, including several UK studies. 34–37 Alcohol outlet density has been shown to be higher in deprived areas in both Scotland and England. 36,38 However, the extent to which increased availability causes alcohol harms in different contexts, and the mechanisms by which effects are exerted, remains unclear, since much of the research is cross-sectional and the validity of measures of the availability of alcohol premises is variable. 37,39–43 A review of 160 studies found that a causal relationship between public health activities, specific local licensing controls, indicators and types of availability and alcohol-related harms is not clear or consistently demonstrated in the literature. 39 The same review noted the difficulty of translating research findings into practice, due both to these limitations and to the lack of clear theories of change. Research into local licensing practitioners’ use of evidence suggests a preference for locally relevant evidence over evidence relating to different (international) contexts and interventions. 44 This presents challenges for alcohol availability decisions in the UK, as most of the applicable evidence is from the USA and Australia. 39,45

1.1.2 Regulation of availability in the context of England and Scotland

In the UK, age restrictions, physical availability and temporal availability controls are largely administered through licensing systems at the local government (LG) level, although mandatory and discretionary powers of LGs are specified by national legislation. 46,47 Although the context of alcohol consumption is broadly similar across the UK, three different legislative and administrative systems exist for controlling alcohol availability in the UK’s constituent nations: one in England/Wales, one in Scotland, and one in Northern Ireland. 48,49 In all three, as in many other jurisdictions worldwide, the sale of alcohol requires a premises licence (a permit). In Northern Ireland, such licences are issued by local courts, whereas both in England/Wales and in Scotland, they are issued by LG bodies known as Licensing Committees or Licensing Boards, respectively. 50,51

Major changes to the licensing systems were introduced in legislation in England and Wales in 2003,50 and in Scotland in 2005. 51 The powers and processes defined through these Licensing Acts continue to be developed through guidance and legislation – for examples, see Fitzgerald et al.,49 Home Office,52 Scottish Parliament53 and UK Parliament. 54 In this new era of alcohol licensing, decisions to grant, amend or refuse licence applications are guided by statutory ‘licensing objectives’: preventing crime and disorder, promoting public safety, preventing public nuisance, protecting children (and young people) from harm and, in Scotland only, protecting and improving public health. Licensing Committees/Boards, made up of elected councillors, are responsible for licensing decisions in their local area. The form and powers of these committees differ somewhat between England and Scotland. 49 However, it is worth noting that, once granted, licences cannot be rescinded solely to reduce availability and are rarely revoked (normally only in the case of premises that are found to be in serious breach of the law or licensing objectives). Applications for alcohol licences must indicate whether they are seeking a licence for an outlet to sell alcohol for consumption on site (on-trade premises or ‘on-licences’) or away from the premises (off-trade premises or ‘off-licences’), and must also provide other information on their proposed hours of operation, type of business and so on. In normal circumstances, the default assumption is that a licence application will be accepted unless a convincing case is made that it undermines one or more of the licensing objectives. Much licensing activity focuses on establishing, and ensuring adherence to, specific conditions placed on individual licences.

Under the current legislation, specified local bodies must receive notice of alcohol premise licence applications so that they can choose to review the application in the context of the above objectives. They may then seek to have applications amended (through a representation) or declined (through an objection). These ‘responsible authorities’ (England/Wales) or ‘statutory consultees’ (SCs; Scotland) include police, health and other public sector bodies. English health authorities became classed as responsible authorities in April 2012. 55,56 In 2013, many public health functions in England were transferred from the NHS to LGs. 57 As a result, LG-based public health practitioners became involved in alcohol licensing with a statutory role as responsible authorities. In Scotland, local NHS Boards (responsible for administering the NHS in the area) became SCs in October 2011. 58 In practice, health input to licensing in Scotland involves a diverse group of health practitioners including staff within NHS Board public health teams/departments, NHS health improvement departments, Health and Social Care Partnerships (which are joint NHS–LG bodies), or Alcohol and Drug Partnerships (which can be NHS or LG-based). For ease of writing, we use the term ‘public health team’ (PHT) to describe any combination of these groups in Scotland, as well as public health teams in English LG. 1

Besides considering individual applications, local licensing authorities must also produce a ‘Statement of Licensing Policy’ (SLP) every 5 years and are required to consult publicly on their proposed policy. 59 The SLP should outline the authority’s strategic approach to promoting the licensing objectives. In Scotland, SLPs must include a statement on ‘overprovision’; that is, whether there are areas where the number or density of outlets, or of a particular type of outlet, is deemed excessive for any reason. 51,53 A licence application may be refused in Scotland on grounds of overprovision alone. This overprovision legislation does not apply to England. However, English licensing authorities possess a discretionary power to create ‘Cumulative Impact Zones’54 (CIZs) [originally introduced in guidance as ‘cumulative impact policies’ (CIPs)/‘cumulative impact areas’]. In both CIZs and overprovision areas, the burden of proof regarding the licensing objectives is reversed: instead of assuming applicants will be granted a licence unless it is successfully argued that to do so would contravene one or more of the licensing objectives, licence applications are expected to be refused unless the applicant can demonstrate that granting the licence would not undermine the objectives. Hence, overprovision areas and CIZs provide some means of containing alcohol area-level outlet density, at least with regard to new licence applications.

Licensing authorities also have some control over temporal availability. The hours of sale for a given premises are determined by the conditions of the licence granted by the local licensing body in both nations. In Scotland, additional national statutory restrictions mean that alcohol cannot be sold for off-premises consumption (i.e. to take away) outside the hours of 10.00 and 22.00, and include a statutory presumption against 24-hour licences for on-trade premises. 51 In England, there are no statutory restrictions on hours of sale. 52 Since 2013, discretionary powers have been available to English licensing authorities to enforce area-wide prohibition of alcohol sales between midnight and 06.00, but these ‘Early Morning Alcohol Restriction Orders’ have never been implemented. 60 Stipulations relating to hours of sale have sometimes been included in what are now known as CIZs, and trading hours can also be reviewed in overprovision assessments in Scotland. English licensing authorities also have discretionary powers under the Police Reform and Social Responsibility Act 2011 to introduce a scheme known as the ‘Late Night Levy (LNL)’. LNL schemes allow licensing authorities to collect funds from alcohol outlets operating in the period from midnight to 06.00, for the specific purpose of contributing to the costs of policing the night-time economy. 61 Such schemes may be hypothesised to deter some outlets from late-night alcohol sales – to avoid paying the levy – but no robust impact evaluation of the LNL has been published. Only 10 licensing authorities in England currently operate an LNL, while two others previously operated LNLs that subsequently ceased. 62

1.1.3 Studies of alcohol licensing in the UK

Recent studies have found that local authorities in England with a more active licensing regime (defined as those with a CIP in place, and which had declined at least one licence application) experienced a 5% reduction in alcohol-related hospital admissions rates from 2009 to 2015 (or 2% annually),63 as well as a 4–6% reduction in public nuisance and alcohol-related crime rates,64 over and above what would have been expected had there been a less active licensing regime (as defined above) in place. The English literature also includes research into specific aspects of local licensing policy. The implementation of CIPs, for example, has been found to differ greatly from one area to another, but all focused on influencing the types of premises that would be awarded new licences (typically favouring restaurants, cafes and arts venues, while at times discouraging traditional pubs or nightclubs). 65,66 A comparative case study found some evidence of PHTs advocating the rejection of all new licence applications in CIP areas, but even licensing authorities that appeared to take a hard line enforcing CIPs still accepted applications relating to premises perceived by the Licensing Committee to be compatible with licensing objectives. 65 Thus, CIPs are unlikely to contain overall outlet density but may impact on density of outlets of specific types. A quantitative evaluation of one local authority’s CIP in London found moderate reductions in crime and no impact on ambulance call-outs. 67 A study using novel quantitative methods to assess the impact of specific licensing decisions or activities found some evidence that closure of a nightclub (after a series of problems prompted a Licensing Committee review) and, in a separate case, local policy guidance led to small improvements in social outcomes. 68 LNLs remain under-researched, although a qualitative evaluation is currently going through peer review. A House of Lords Select Committee postlegislative assessment of the 2003 Act argued for a range of reforms, including the merger of the licensing and planning functions within local authorities. 69 A review of the English Licensing Act by the Institute of Alcohol Studies identified some potential for health and social benefits but put forward the view that the Act had been interpreted to the advantage of the licensed trade and reform was necessary. 47

In Scotland, research by Alcohol Focus Scotland (AFS) tracks the development of SLPs in Scotland from 2013 to 2018 to identify emerging trends in licensing approaches, and the extent to which policies meet legal requirements and incorporate identified good practice. 70 The most recent review found that many of the policies showed noticeable positive progress in the consideration and interpretation of evidence and in approaches to promoting the licensing objectives. Notably, it found that Licensing Boards were seeking to respond to the trend towards purchasing alcohol from off-sales premises for consumption at home, and an increased awareness of alcohol-related harms occurring in the private as well as the public sphere. The review highlights a new approach by the Licensing Board in Glasgow, which explicitly links the availability of alcohol from off-sales premises with high levels of health harm, and states that the health objective can be used as grounds for refusal even in the absence of an overprovision policy. The review suggests that this may be a more effective way of addressing concerns about the impact of availability on health harm because it does not require data on alcohol harm to be linked directly to licensed premises at the application stage, potentially avoiding the need to demonstrate causality, which remains a challenge in the use of overprovision policies in Scotland. This is important, because the review also finds that differing approaches to the assessment of overprovision persist due to varying interpretations of what is required by law in this regard, leading to uncertainty in practice.

1.1.4 Public health team engagement in alcohol premises licensing

Following the enhancement of their statutory roles, many professionals with an interest in reducing alcohol-related harms increased their engagement with premises licensing, often supported by national bodies such as Public Health England (PHE) and AFS, among others. 71–74 However, much of the evidence relating to their role within local licensing is qualitative, is focused on a small number of areas, and does not seek to systematically identify public health activity nor to robustly quantify health or social impacts. 27 No single study has collected data in both Scotland and England.

At the time when English public health functions transferred to LG, Martineau et al. 75 hypothesised that the absence of a public health objective for alcohol licensing in English legislation could undermine PHTs’ efforts to influence local licensing systems. Subsequent mixed-methods research into licensing systems in London supported this view, finding that public health professionals felt a lack of status within their local licensing system. Some felt confident enough to make representations on licence applications independently, but others sought to strengthen their position by working with other responsible authorities, providing data to support decision-making and/or seeking to influence SLPs. 76,77 Some PHTs attempted to work around the lack of a public health objective by focusing on social harms that have public health relevance. 76,77 A separate qualitative interview and observation study focused on public health involvement in licensing in six London boroughs also found varying perceptions of public health success and reported practices, and noted that some public health actors felt that working in the political environment of LG required a change in their professional identity away from a medical focus. 78 A comparative case study of two urban local authorities in the north-east of England found that economic, organisational and personal factors influenced their licensing authorities’ different approaches to public health and alcohol harm reduction. 79

Early research in Scotland found that licensing stakeholders struggled to understand the public health objective and how to operationalise it. 80 Early efforts by public health actors were felt to have achieved mixed results, with some Licensing Boards introducing large-scale overprovision policies and others strongly resisting public health engagement. 72,81–83 While research in England (above) found the lack of a public health objective to be a barrier to public health engagement, Scottish studies have considered its limitations. Fitzgerald et al. 81 concluded that, notwithstanding the introduction of the public health objective, there remain significant political and other challenges in orienting local Licensing Boards towards decisions more likely to contain the availability of alcohol in Scotland.

Research into public health involvement in English and Scottish licensing has extended its focus beyond the licensing objectives to look more generally at how PHTs engage with licensing systems. PHTs may, for instance, make representations directly to licensing authorities, provide data in support of a representation by the local police or Trading Standards, respond to consultations on CIPs or, as has been more common in Scotland, take the lead in developing the case for the establishment of overprovision areas. 65,72,82,84 These and other studies found that some PHTs have developed processes for reviewing and responding to licence applications, collated local data sets on outlet density and alcohol-related harms, supported the development of licensing policies, involved local communities, or directly engaged with licence holders. 65,72,82,84,85 In 2018, Fitzgerald et al. 85 considered the Scottish licensing system through the lens of democracy and power, concluding that many stakeholders involved – including public health representatives and the wider public – were relatively disempowered in decision-making processes. Research from England similarly found that opportunities for public involvement in local alcohol decision-making were limited, although the study identified some evidence of PHTs engaging with local residents, as well as business owners, to gather evidence to support proposals for CIPs. 86

1.1.5 The impact of public health engagement in licensing on alcohol-related harms

Despite this activity from public health actors, no studies have yet measured the health and social impacts of public health involvement in the English or Scottish licensing systems. PHTs’ decisions to commit resources to involvement in local licensing systems have substantial implications, particularly given the fiscal restrictions experienced by LG over the last decade in both England and Scotland and the continued financial uncertainty affecting public health and other LG budgets. 87,88 Hence it is important to assess whether PHT involvement in alcohol licensing is likely to be a worthwhile use of resources. Somerville noted that PHTs felt that their potential to achieve improvements in population health through licensing was probably small at best, and recommended further research to ascertain the actual impact of public health involvement on population-level alcohol-related health harms. 78

Attempts to influence alcohol harms through controls on availability, including PHTs’ engagement in licensing, can be viewed as interventions taking place within a wider, interactional system. 89 Conceptualisations of alcohol systems vary but in relation to licensing will involve responsible authorities/SCs (including PHTs), Licensing Committees/Boards, the alcohol trade and the general public. In engaging with the licensing system, and with support from national organisations such as the Office for Health Improvement and Disparities (formerly PHE) and AFS, PHTs have developed a range of approaches. 84,90 These approaches are used to varying degrees of intensity and in varying combinations in local areas across England and Scotland, creating a natural experiment, which has yet to be evaluated.

2.2.1 Recruitment and sampling

We recruited PHTs working in 40 local authority areas (28 in England, 12 in Scotland) by soliciting expressions of interest and making direct approaches. PHTs were deliberately sampled to maximise variation in terms of levels of activity in engaging with alcohol premises licensing, as well as being purposively varied by region and rurality. Twenty higher-activity PHT areas were identified through intelligence gathering and scoping calls and recruited. The identity of recruited areas was passed to the WP2 team, who matched them to candidate lower-activity PHT areas for the WP1 team to assess and approach. Twenty lower-activity PHT areas were successfully recruited in this way. One lower-activity area subsequently dropped out, leaving 39 PHTs in the study. A full profile of participating PHTs is included in Appendix 2, reproduced from Fitzgerald et al. 2

2.2.2 Development of the PHIAL measure

The development of the PHIAL measure is described in further detail in Fitzgerald et al. 22 The purpose of the measure was to categorise PHT activity and to enable quantification of the level of activity by each PHT over time. Preliminary categories of PHT activity were developed based on prior literature, and guided data collection with PHTs. Relevant activity from April 2012 to March 2019 was identified through structured interviews (n = 66), documentation analysis and follow-up checks, and expert consultation. Extensive internal discussion led to the development of a system of grading that would allow simple categorical assessment of the extent and nature of each activity for each 6-month period, given the nature and depth of typical PHT data. Following several iterations and testing, this included binary ratings (e.g. yes/no), or up to a four-point rating scale (e.g. higher/medium/lower/none) depending on the activity. The resulting grading system was applied to the data from all 39 areas, to generate preliminary grades for each time period, noting any instances where grading was not straightforward. These informed further team discussions and amendments to finalise grading scales for all activities, along with accompanying guidance notes. Grades were converted to fractional scores out of 1 for each activity type for each time period. With further expert consultation and team discussion, a consensus was reached to categorise each activity as ‘high’, ‘medium’ or ‘low’ according to their likely relative impact on alcohol-related health harms and crime. Weights were applied as a simple multiplier (low = × 1; medium = × 2; high = × 3) to the score for each activity type in each 6-month period. The final measure (abbreviated version in Table 2, full version in Table S4 in Fitzgerald et al. 2) therefore enables calculation of a weighted overall intensity score for total relevant activity for each 6-month period.

Relevant PHT activity for the 39 participating areas was assessed for each 6-month period using the final PHIAL measure. Nineteen activity types in six categories were mapped out and described in detail in the measure, which was used to describe the patterns and range of activities used in different areas including comparing timing and intensity of activity for PHTs in Scotland and England. The six categories were as follows and are described in the findings below:

1. staffing for PHT activity to influence local alcohol licensing

2. reviewing alcohol licensing applications

3. influencing and responding to individual licence applications

4. use of routine or bespoke data on alcohol licensing and alcohol-related harms

5. influencing local stakeholders or licensing policy

6. engagement or involvement of the public.

We successfully recruited PHTs that varied demographically, and that varied in the ‘timing, breadth, components and intensity of their efforts to engage in alcohol premises licensing’. 2

2.2.3 Deviation from protocol

We originally sought to gather and quantify the strength of the licensing regime in place in each area (policies and application decisions), and of potential confounding activities (such as PHT-led training for health staff on alcohol brief interventions, or drink-drive initiatives); however, the development of the PHIAL measure proved more onerous than anticipated (see Appendix 4, ExILEnS lessons learned). While we gathered qualitative data on these aspects, we were unable to quantify them in a way that would have enabled their use in WP2. The only way to gather this kind of licensing data currently is through highly onerous manual collation, comparison and checking, because the data do not exist in any accurate, accessible national data set. This gap in national data was a further barrier to the development of a licensing intensity measure.

2.3.1 Quantitative data collection of selected alcohol harm and crime outcomes

We collected 2009–18 time-series data, aggregated at a local authority level to 6-month periods to match the PHIAL exposure metric temporal resolution, for a set of key outcome measures determined at the start of the project. These included measures of harm: alcohol-related hospital admissions (narrow measure), which was the primary outcome on which statistical power analyses were based; hospital admissions for acute conditions related to alcohol; alcohol-related mortality; alcohol-specific mortality; and ambulance call-outs. They also included crime indicators: sexual and violent crimes and public order offences. In England hospital admissions were obtained from UKHSA Local Alcohol Profiles, and data for other outcomes from the Office for National Statistics. In Scotland, health harms data were obtained from the Information Services Division Scotland and crime data from the Scottish Government. Outcomes data were linked to the PHIAL measure at the same temporal and spatial resolutions and additionally to a set of area-level temporally variable covariates: nation (England or Scotland), season, area-level socioeconomic status [Index of Multiple Deprivation (IMD)], population density, average age in area, and, for England only, whether an area was a local alcohol action area (LAAA). 93,94

2.3.2 Statistical modelling

Time series were modelled using negative binomial mixed-effects models incorporating a random intercept for each area. An offset was specified as the mid-year population number for each area. Associations between each outcome and ‘18-month average PHIAL’ (the primary exposure metric), ‘cumulative PHIAL score’ and ‘change in PHIAL score’ were modelled such that the PHIAL score at each time point impacted on the outcome in the same period (a ‘snare’ model). In addition, we modelled 6-month lagged correlations in which PHIAL scores were hypothesised to affect the subsequent 6-month period. Baseline models included PHIAL score and time, while random and fully adjusted models also included all covariates. Final models were the most parsimonious models, with the lowest Akaike information criterion values and fewest number of predictors, but with the predictors for the correlation between exposure and outcome being similar to the fully adjusted model. Where statistically significant associations (p < 0.05) were observed, 1-holdout resampling was conducted in which areas were sequentially removed and models recalculated to assess whether significant findings were stable or resulted from inclusion of single areas.

2.3.3 Deviations from protocol

We originally aimed to also include A&E attendances for alcohol-related conditions. However, we decided not to use this outcome as we were unable to link attendances to all local authorities because of overlap of A&E catchment areas and study areas, while we were also unconvinced that the number of A&E attendances in a particular hospital is representative of the burden in a particular area. We further analysed the total composite PHIAL score as the exposure measure and have not conducted PHIAL component-specific analyses; we aim to do this in future.

2.4.1 Model development

We developed new versions of the Sheffield Alcohol Policy Model95 for each of the recruited LG areas, to allow both short-term modelling of any inequality impacts of changes in PHT involvement in licensing and longer-term modelling of the overall impact of PHT involvement. The following data were collated to prepare the models:

• population estimates for each age, sex and quintile of the English and Scottish IMDs96,97

• hospital admissions rates for each age, sex and IMD quintile for each of 45 different alcohol-related conditions98

• mortality rates for each age, sex and IMD quintile for each alcohol-related condition

• healthcare costs per admission for each alcohol-related condition

• weekly alcohol consumption in units

• alcohol prices per beverage category, per age, sex and IMD quintile subgroup

• criminal offence count estimates by age and sex for a range of different offence categories

• costs associated with each criminal offence.

We identified and obtained the most appropriate data for each of these categories. There were challenges in ensuring maximum comparability across both crime and health data between English and Scottish areas, due to differences in the criminal justice systems and the way in which diagnoses are recorded for hospital admissions. We consulted with stakeholders and the wider project team to make sure all data were as similar as possible across all models.

Where possible, local area-level data were used; however, if such data were unavailable, work was undertaken to find the best possible alternative solution. For example, neither the Health Survey for England nor the Scottish Health Survey have large enough sample sizes to provide robust estimates of alcohol consumption at LG area level. We therefore revised (for England) and adapted (for Scotland) an approach we had previously developed99 using English data which reweights the national survey data using local data to produce a synthesised local consumption survey data set for each of our areas of interest.

2.4.2 Deviation from protocol and potential future value of this work

The original plan was to incorporate the findings from WP2 with the local models we had built. However, the findings from WP2 were non-significant and close to zero, with little or no evidence to support an association between PHT involvement in the licensing process and the outcomes used in our models (i.e. alcohol-related hospital admissions and mortality). As a result, we were unable to undertake the planned modelling, although the models have been built and will be available for use in future projects which look at the longer-term impacts of local or national policies that might influence alcohol consumption. Such policies might include action on pricing, alcohol availability or other public health interventions. In addition to these models, we have also developed new approaches to estimating local area-level alcohol consumption in Scotland and generated more comparable data on alcohol-related health and crime between England and Scotland, both of which are likely to be of value in future work. We collated data collected through the WP1 analysis and interviews relating to the staff commitments associated with PHT activity. Our aim was to combine this information with standard pay scales as well as any other information collected on other related costs, to estimate the overall cost of PHT involvement in the licensing process. Unfortunately, in practice these data were inconsistently recorded or unclear, with either estimates of staff full-time equivalent commitments to PHT activity, or information on the pay band of those staff being difficult to obtain. As a result, we were unable to produce any robust estimates of the staff costs associated with PHT implementation actions.

2.5.1 Recruitment, sampling, data collection and analysis

Within the 20 higher-activity areas, potential stakeholders for interview were identified through direct contact, initial site visits undertaken as part of WP1, and snowball sampling. Purposive participant selection aimed to optimise diversity in terms of public health and licensing stakeholder remit and location. Informed consent was obtained prior to each interview. Interviews were conducted by telephone, between November 2018 and October 2020. Fifty-five individuals agreed to participate, and two were uncontactable to arrange interviews. Separate topic guides were developed for the five stakeholder groups [public health, licensing managers/officers, elected members, police (licensing) officers and licensing lawyers/clerks] based on existing alcohol licensing literature and research team discussion. The topic guides (summarised in O’Donnell et al. 5) included a focus on interviewee roles, responsibilities and purpose in the licensing system, public health approaches to engaging in licensing, and interviewee views on such approaches and the ways in which temporal and spatial availability impact on alcohol harms. Interviews lasted between 32 and 156 minutes (median = 72 minutes) and were audio-recorded. Interview transcripts were analysed thematically with NVivo 12 (Lumivero, Denver, CO, USA) using inductive and deductive approaches.

Secondly, two small online discussion groups were conducted in January 2021 with a different sample of stakeholders working in England (n = 6) and Scotland (n = 4) with expertise in public health and licensing. These groups included representation from local and national government, police, third sector and academia identified via relevant networks and recommendations. Both discussions were audio-recorded, with participant consent obtained in advance. Groups discussed views on the ways in which temporal and spatial availability interventions might impact on alcohol-related harms, and the possible mechanisms of change involved, including relevant preliminary findings from in-depth interviews (questions are summarised in O’Donnell et al. 5). 99 Transcripts were analysed to identify extracts providing confirmatory and contradictory insights compared with in-depth interview findings. These extracts also informed further analysis of the in-depth interview data, alongside team meeting discussions and with input from advisory group members with legal and policy backgrounds.

2.5.2 Developing theories of change

Prior to the study’s commencement, we developed simple linear theories of change and an overarching logic model suggesting that diverse local PHT activities may contribute to changes to local licensing policy, effect change in the environment (places), and thus influence health and crime outcomes (for people). Over the course of the study, we discussed at multiple team meetings and with expert advisors what the pathways to change might be from different public health activities to licensing change, and from changes in licensing to alcohol-related harm. On the former, we considered how to group PHT activity types (from the 19 identified) to reflect differing PHT approaches, and whether different approaches might have differing degrees of success. We developed preliminary logic models and directed acyclic graphs to help develop our theories; however, given the limitations in prior evidence, we had no strong basis on which to refine or finalise these theories. On the latter, we examined evidence and theory on links between temporal and spatial availability and harms, and developed systems maps, based on systematic inspection of the interview data and prior expertise. We used these systems maps (see Appendix 3) as an engagement tool in our online discussion groups (above) to gain stakeholder insights. These discussions introduced further complexity, especially as discussion group members and interviewees were uncertain about the local validity of the mechanisms presented and any evidence base supporting them. Ultimately, we felt that attempting to present the nuances of overlapping PHT approaches and thinking in a diagram would be unhelpfully reductionist, and that the availability systems maps need to be systematically stress-tested with theory and evidence-driven pathways (see Section 3.7). We concluded that our data would be better represented in qualitative reports rather than in the form of a diagram. Our qualitative findings add to current theorisations, but any theory of change is underdeveloped in practitioners’ thinking as well as in the international literature.

4.1.1 Diversity of types and level of activity on licensing

By including a relatively large sample of PHTs (n = 39) and assessing their engagement in licensing over a long period (7 years), we were able to identify and assess for the first time, in detail and depth, the intensity of PHT activity to engage in alcohol licensing. Our finding that there are PHTs in England which do not engage at all in local licensing concurs with the findings of a prior study of eight London areas. 77 Our finding that many PHTs across England, and some in Scotland, engage only to a very low extent (e.g. only by logging applications or occasionally supporting representations made by others) accords with other research previously carried out in London. 78 This was predicted by Nicholls48 when highlighting a ‘risk of disillusionment, frustration, or simple lack of capacity’ as threats to continued public health engagement in alcohol licensing. In our structured interviews with PHTs in lower-activity areas, it was apparent that lack of capacity, and prioritisation of other public health activities were factors in decisions not to engage during our data collection period (2012–9). The latter is likely to have been a more widespread issue during the COVID-19 pandemic.

In contrast, we also found many areas that had developed an impressive array of activities and systems to support a high level of engagement in licensing. Many of these activities have been identified previously in smaller studies as above and in Scotland, but the depth and breadth of our analysis has allowed us to distinguish cases where PHTs have gone furthest in their commitment to this work. We found examples of very detailed databases logging, tracking and evaluating decisions on licence applications which in some instances included notes on what worked well (or did not) in the PHT response to inform future practice. There is likely to be a rich data set in such systems that may be valuable to collate and analyse nationally. Many PHTs put significant effort into regularly curating data into summaries analysed by data zone or locality, presented in accessible formats for licensing colleagues and committee members. Others even established new data collection processes specifically to inform licensing. Some PHTs in England were highly active in engaging with applicants to influence premises operating plans and extract commitments on safety (e.g. CCTV), though this was generally considered a role for licensing colleagues rather than public health in Scotland. Scottish PHTs, in particular, made ample use of the public health objective to lead and submit independent representations or objections in response to licence applications. Some PHTs either led or significantly contributed to the actual drafting of SLPs, and others supported the active involvement of local communities in licensing. Differences in approach and innovative practices appear to have developed iteratively in areas and nations, but with some important influences specific to England or Scotland (discussed further below). No single PHT was highly active across all the activity types assessed, and no areas scored the maximum available. While some differences reflected deliberate decisions not to engage in certain activities, and PHT beliefs about the most effective or appropriate approaches (see below), the variation in activity points to opportunities for learning between PHTs that is not specific to Scotland or England. It also highlights the potential for further analysis of the impact of different activity types or groups. Further support from national bodies, perhaps working together, could enable such peer-to-peer sharing of practice, intelligence and experience going forward, as well as providing a route by which the practices and innovation identified in this study could be shared in greater detail.

4.1.2 Differences in activity between Scotland and England and in the approaches taken by public health teams: the influence of the public health objective

As this was the first comparative study across two licensing systems in the UK, we were able to compare in detail the extent to which being based in Scotland or England was a key factor influencing PHT practice. We found that variations in activity between PHTs in the two nations were influenced by legal, structural and philosophical differences including, but not limited to, the presence of a public health objective in Scotland. 49

Our findings suggest that the presence of a public health objective in Scotland has been helpful to PHT engagement by conferring legitimacy on public health stakeholders and data, as suggested in earlier English studies. 76,78 The objective appears to partly explain why PHTs in Scotland were more active in making representations or objections to licence applications independently of other stakeholders, and were more comfortable with adopting a challenging approach to the licensing system. The lack of a public health objective made some English PHTs less confident about doing so and was most likely a factor in some teams being more passive or choosing not to engage with licensing at all. Somerville found that some licensing and police stakeholders in England felt that a passive, supportive role for public health was most appropriate, though others felt that PHTs should have an equal role to other responsible authorities,78 more akin to the situation in Scotland.

Early studies in Scotland show that the introduction of the public health objective was far from instantly transformative, and that many challenges emerged as stakeholders grappled with engaging in a licensing system that carried a traditional focus on local and individual concerns. 48,80 Public health stakeholders were new to licensing, and their expectations and approach have evolved over time as they have learned from experience and licensing colleagues. 5,82 For example, they no longer object to all licence applications and are more focused on the importance of the SLP, as well as building relationships with licensing colleagues and raising awareness of alcohol harms. Overall, there is a sense in Scotland that the public health objective, and the efforts of PHTs in conjunction with national advocacy work, may be slowly reorienting local licensing systems and stakeholders to be more comfortable with a broader, population-based lens. A comparison of local SLPs published in 2013 and 2018 in Scotland suggests that a more strategic approach is evolving on the promotion of the licensing objectives, as the statements ‘attempt to place licensing policy in context, provide explanations for policy positions, refer to supporting evidence, and acknowledge and reflect the views of consultees in the formulation of policy’. 70 Other tweaks to the Scottish system may aid its evolution: a requirement for Licensing Boards to publish annual reports is seen as a positive move towards greater accountability on all the licensing objectives. 53,105

Early struggles with the public health objective partly arose from a clash of perspective between public health and licensing teams, the former favouring international academic evidence, and the latter being more likely to value local data and practical knowledge in making decisions. 48,81,82 Our new data suggest that these struggles have lessened in some areas in Scotland (and to some extent in England), with both parties apparently developing a constructive understanding and appreciation of the value of different kinds of evidence. Some licensing stakeholders described how useful PHT approaches to, and contributions of, evidence have been. At the same time, public health stakeholders appear to have lowered prior expectations that the licensing system could simply reject licence applications across the board, or make a substantial difference to availability (see O’Donnell et al. 5 and below).

In summary, our data provide extensive evidence of successful PHT engagement in the licensing system in Scotland and England, but also show that such activity is generally higher in Scotland. 49 There is a widespread sense that the public health objective strengthens the role and legitimacy of public health involvement in Scotland and could helpfully do the same in England. It is important not to view the public health objective as a panacea, however, and to consider its introduction alongside other changes that may enable the licensing system to better protect against alcohol-related harms and within a broader evidence-based alcohol policy framework (see recommendations for decision-makers below).

4.1.3 Differences in activity between Scotland and England and in the approaches taken by public health teams: other important influences

By comparing the legal systems and structures in Scotland and England in detail alongside our in-depth data on PHT activities, we were able to identify influences on PHT activity other than the public health objective that have not previously been highlighted in the literature. Seemingly small differences in the set-up of the licensing system appear to have an important influence on how PHTs engage in this arena. 49 The system in England allows a licence application to be granted without being considered at a committee meeting (‘hearing’) if there are no objections, creating an incentive for parties to agree. Thus, PHTs in England sought to negotiate with applicants or their solicitors to reach agreement on operating conditions or other changes to licence applications to avoid needing a hearing. This practice has been noted previously – for example, by Reynolds et al. 77 – but our analysis highlights that the same incentive does not exist in Scotland, where all new licence applications are considered at Licensing Board meetings. This also partly explains why PHTs in Scotland were less likely to negotiate on licence applications: they tended to be more focused on containing availability by preventing new premises from opening than on shaping how premises operated,5,106 and felt that negotiation would not have made the same sense anyway.

It is also apparent that support from national bodies (AFS and PHE, as was) alongside regional groups in England is likely to have been an important influence. In England, the work of PHE on this agenda has recently waned, and some regional bodies no longer exist (e.g. DrinkWise North West). This may have influenced a falling-off of PHT attention to licensing in our study, which is quite likely to have been exacerbated by the COVID-19 pandemic. We also identified a surprising (mistaken) belief among some experienced PHT colleagues in Scotland that working with other SCs to share data and coordinate a joint response to a licence application would constitute illegal ‘collusion’. There is no basis for this belief in civil licensing law, and it suggests a need for corrective communication with SCs, perhaps by licensing teams or AFS. Finally, Scottish PHTs did not generally feel it was appropriate for them to liaise directly with licence applicants; they seemed to be more sensitive to likely conflicting interests between public health and licence applicants. This may also reflect differing national contexts, with the UK government reported to be more comfortable with industry influence in alcohol policy. 107,108

4.1.4 How public health team activity, and the licensing system more generally, may impact on alcohol-related harms

We found that the extent to which PHTs engaged in diverse activities to influence alcohol premises licensing policies and decisions was not associated with measurable reductions in health harms or crimes linked to alcohol, nor in overall ambulance call-outs, over the 7-year period of this study. This null finding raises important questions relating to alcohol availability, licensing and public health.

We were unable to quantitatively assess direct impact on the licensing system, in part because data on licensing decisions and licensed premises are not consistently or easily available, reliable or accurate. This means that we cannot be sure whether the lack of impact of public health involvement was because their activities failed to make a material change to the licensing system in their area (e.g. in terms of changing the number of licence applications submitted or the proportion that were approved) or because changes in the licensing system as a result of their involvement had no impact on alcohol-related harms. If the former, it would mean that PHTs did not materially influence licensing, perhaps merely reinforcing what would have happened anyway in the system. However, our in-depth interviews with public health and other licensing stakeholders suggested that PHTs provided valued input into alcohol licensing, contributing to or influencing decisions and policies in identifiable ways5 as in earlier studies,77 and the more likely explanation therefore is that such changes had little or no impact on alcohol-related harms.

However, PHT impact is situated within, and necessarily limited by, the relatively permissive nature of the licensing system in the two nations. Alcohol outlet density in the UK is high by international standards,109 and even the strongest local licensing policies could not legally reduce outlet numbers, but merely contain them at current levels. Existing licences cannot be revoked on overprovision/cumulative impact grounds and licence applications can continue to be granted even in such areas. 65 In practice, even containment of physical availability is unlikely in most areas; while some Scottish Licensing Boards have put in place overprovision policies covering their entire jurisdictions, they usually apply to a particular premises type. Furthermore, the systems for assessing CIZs/overprovision are not equipped to take account of the rise of remote sales, online and via apps, which involve direct delivery to consumers. Licensing stakeholders, PHTs and clinicians are united in their concern about the rise in such sales, particularly since the COVID-19 pandemic, and the inability of the licensing system to contain this kind of availability in any way. 70,104 National action to address online alcohol sales before they increase even further would likely help to prevent further rises in availability, levels of drinking at home, and related harms.

PHTs described two approaches to their licensing work, with some PHTs applying both. The first approach sought to influence local licensing policies to take a more assertive approach to containing alcohol availability in order to effect longer-term, potentially more radical, culture change. The second approach attempted to achieve more immediate impacts through ‘fitting in’ with local licensing system goals. For some teams engaging in the latter approach, this meant deprioritising a focus on availability. While the positive effects of reducing availability on alcohol-related harm is well supported by systematic review evidence,29,32,110,111 the relevance of such evidence locally in the UK context is uncertain. 39 The evidence for effectiveness of interventions to manage the drinking environment (e.g. server training) is low, however. 111,112 No stakeholders raised the potential of other licensing systems that take a stronger approach to managing availability, such as through a ‘surrender principle’ as exists in Northern Ireland (whereby a new premises can only open by obtaining a licence surrendered by another premises)113 or through restricting the sale of alcohol above a certain strength to be sold only in shops owned by the state, as in several Nordic countries. 114

An important observation of our data is that PHTs and licensing stakeholders paid relatively little attention to opening hours for licensed premises, and that when talking about addressing availability they tended to think in terms of spatial (numbers of premises) rather than temporal availability. Recent studies in diverse high-income cities suggest that restrictions on temporal availability, particularly after midnight, may be very effective in reducing short-term alcohol harms from on-trade premises. 115,116 This may work even where physical availability is high, but was rarely a focus of public health stakeholders in this study. One reason for this may be uncertainty about the mechanism by which changes in opening hours may lead to any harms (see below). Importantly, participants cited the 24-hour availability of alcohol in England, and the resulting ease of access to online sources of alcohol at any time of day, as one moderating factor that might impede any benefits from early closing of on-trade premises. This raises the question of whether 24-hour availability of off-trade alcohol remains appropriate in light of new rapid delivery services and easy online access, and what its impact might be on the hospitality sector (see Section 3.5). Overall trends towards lower consumption since 2005 may undermine confidence in any link between greater temporal availability and increased consumption, though diverse (and sometimes contrasting) trends in different subgroups may be an important part of the discussion.

Closing premises earlier may be more compatible with economic development than preventing licences being granted. 117–119 This highlights the importance of documenting the impacts of licensing beyond health harms and crime to include impacts on nightlife and trade in future studies. In Scotland, recent analysis of alcohol policy statements found that more Licensing Boards across Scotland had extended premises’ opening hours than reduced them, and there is a risk that there will occur a ‘creeping’ extension to hours. 70 Given robust evidence from other high-income countries associating later opening hours with increased assaults and ambulance call-outs, PHTs and licensing stakeholders may wish to be cautious. A separate study examining extensions in late-night hours in Glasgow and Aberdeen cities is under way. It is worth noting that the policy statement analysis for Scotland was conducted by AFS (without specific funding to do so) and is very helpful for understanding trends and developments in local licensing. Currently, there is no national monitoring or comparison of local alcohol policy statements in England.

The alcohol licensing system is a centuries-old, highly legalistic, local ‘centre’ of policy-making, with cultures of evidence and practice that are very different from those of public health. It has been argued that long-term engagement may be necessary to see substantial impact. 48,82,120 It may therefore be unrealistic to expect PHT engagement in the licensing system to impact in a measurable way on health and crime outcomes, given the above constraints, in a 7-year period. However, such engagement may be starting to reshape thinking and practice in licensing to be more strategic and/or population-focused in some local areas, especially in Scotland. Radical changes to completely reshape the system (e.g. to that of Northern Ireland or to government retail monopolies)113,114 are not under discussion; however, where national policy-makers are supportive, this may lead to small changes to licensing laws and guidance that will gradually have a positive impact. For example, Licensing Boards in Scotland are now required to publish ‘annual functions reports’ to facilitate transparency and some accountability. We also identified possible mechanisms by which changes in availability may gradually result in longer-term benefits (see below). It may well be that the impact of public health actors on the licensing system takes a long time to emerge, while, at the same time, unsuccessful PHT efforts lead to changes in the licensing system. This seems more likely where a public health objective has been set for licensing but is not clearly being achieved. Strong action on availability alone will not halt the normalisation of alcohol in everyday life, but it likely has a role to play in reducing harms as part of a package of wider policy measures, most of which are not under local control. 81

Previous studies have shown a small positive effect of stronger licensing policies across a larger sample of local authorities in England63,121,122 and in one London authority. 67 Our methods may have meant that we were unable to detect effects either because our area of analysis was too large (at LG level) or because our linear exposure–response analyses were aimed at obtaining average effects despite the complexity of the system under study. The effects of licensing (though not PHT activity) may be measurable in small areas with good, hyperlocal information on interventions, exposure and outcomes. 123 Better data on licences in use, premises operations, and so on would greatly assist with improving knowledge.

4.1.5 Mechanisms of impact of temporal and spatial availability on alcohol harms

Five overarching mechanisms by which alcohol availability was perceived to impact on harms emerged from in-depth interviews with licensing stakeholders. 99 These were: access/convenience, visibility (including pathways via drinking cues and normalisation), premises- and area-level norms of consumption and behaviour linked to characteristics of different premises, affordability (including that arising from availability of high-strength/low-cost alcohol), and management of the night-time economy (e.g. through staggered closing times). These mechanisms are plausible and not typically discussed or scrutinised in previous availability literature, in which theories largely focus on explaining why clusters of outlets or specific individual outlets are associated with elevated rates of harm, especially crimes. 124–126 Further research could identify and assess the evidential basis for the mechanisms suggested (see below).

Current evidence underpinning the links between availability and harms is mixed and also merits further critical scrutiny, especially for specific outlet types, specific subgroups (including individuals in recovery, heavy drinkers and women), areas of differing geographical, cultural and socioeconomic profiles, and over a longer time period, as well as on outcomes such as types and levels of drinking and harms, and harms to others from alcohol. 39,127 Further limitations in the evidence around the role of online availability,128,129 availability relative to a person’s place of work or daily travel, and interactions between availability and price39 may also give rise to expanded areas of focus for developments in the licensing system and public health practice.

Of the mechanisms suggested in our data, the importance of visibility is supported by evidence on the pro-consumption influence of visibility as a ‘drinking cue’ for both drinkers and those in recovery,130 triggering a desire for alcohol131,132 outwith the control of the individual. 133 The idea that visibility of premises might also be a normalising influence on children’s attitudes towards alcohol (like a form of marketing) is emerging from studies of children’s exposure to alcohol in different settings, and children’s own expressed desire for reduced exposure. 134,135

Stakeholders also suggested that attempts to remove low-cost, high-strength alcohol from sale in off-licences were motivated by a desire to ‘control the drinking of those unable to control it themselves’, such as people with alcohol dependence. While there was an implied link in stakeholder reports between earlier alcohol consumption and greater consumption later on a night out, they did not mention the direct influence of intoxication as being a similar barrier to control for late-night drinkers. This may be an important omission in discussion of restrictions on sales late at night – and is easy to understand. The risks suggested by stakeholders arising from the norms of specific types of premises, including ‘vertical drinking’ establishments, can be explained by Gruenewald’s ‘niche theory’: alcohol premises engage in ‘niche marketing’ to specific social groups, drinkers return to establishments at which they find people like themselves, and consequent social stratification of the marketplace increases the levels of related problems in some outlets. 128 Efforts to ‘place-shape’ through licensing policies welcoming applications from certain types of licensed premises while discouraging vertical drinking establishments (e.g. in cumulative impact areas) may have limited effectiveness.

Licensing stakeholders, including public health, found it challenging to articulate specific pathways by which changes in alcohol availability may impact on alcohol harms, and were largely unfamiliar with thinking about mechanisms of change. This may be due in part to the absence of discussion and scrutiny of some of these mechanisms in the availability literature, as well as to weaknesses and limitations in the studies that have considered links between availability and harms. The inability of stakeholders to clearly articulate stories of these mechanisms is likely to hinder their ability to effectively advocate for changes in licensing at local or national level, as evidence suggests that such stories may be at least equally important as, if not more important than, academic evidence in influencing local policy-makers. 48,136

4.2.1 Implications for policy and policy-makers

1. A strategic national approach in England and in Scotland to managing the availability of alcohol is needed to monitor, evaluate and propose responses to trends and progress in availability and licensing in terms of the statutory objectives, with joint leadership from government departments responsible for justice, health and child protection. This national approach should include:

   1. an open online licensing and availability database (to address several weaknesses in current data) comprising an accurate and live register of licensed premises currently trading, including: type of premises, ownership, conditions of licence, size, indoor and outdoor capacity and trading hours, date opened, and sales information for alcohol delivered (see Section 2.2)

   2. routine comparative analysis of local licensing policy statements and licensing data for content, development and trends in efforts to deliver on licensing objectives in the public interest (see Section 4.1.4)

   3. adequately resourced national support for local PHTs involved in licensing to inform them about relevant legal and national developments and to enable peer-to-peer sharing of practice, intelligence and experience, including identification of common challenges and solutions (see Section 4.1.1)

   4. regular review of whether the licensing systems are likely achieving the objectives set for them, including consideration of whether reform or replacement of the system would be in the public interest and potential benefits or risks of alternatives in terms of achieving the objectives (Section 4.1.4 & 4.1.5).

2. A public health objective for alcohol licensing should be introduced in England without further delay, as there is convincing evidence that it would be valuable to, and valued by, public health and other licensing stakeholders. This would help to legitimise the role of health and health data in the licensing system in England and has been implemented in constructive ways in the system in Scotland. The new objective should be planned alongside other changes in law or guidance to better enable the licensing system to achieve the licensing objectives (Section 4.1.2).

3. Policy-makers should act as a matter of urgency to review, trial if necessary, and implement policy options for the regulation and restriction of remote (online/app-based/telephone/delivery-based) sales of alcohol nationally, given the reported rise in such sales and in harms arising for some groups following the COVID-19 pandemic. Such a review should include concerns about age verification but go beyond this to consider the contribution of these sales to overall availability and harms, and how they undermine local efforts to address the cumulative impact of alcohol outlets in a given area (Section 4.1.5).

4. Given the constraints on the licensing system in terms of reducing or even containing availability, the Scottish Government should clarify how current licensing (or other) legislation is intended and expected to address the availability of alcohol and its intended contribution to reducing alcohol-related harms (overarching recommendation). The national alcohol framework for Scotland identifies the licensing system as the mechanism through which action on the availability of alcohol will be taken forward, in part because addressing availability is one of just three ‘best buys’ for alcohol policy identified by the World Health Organization. However, it is not currently clear how progress can be achieved.

5. A national alcohol strategy for England is needed, to include action on alcohol availability and the matters raised in this report, as one of the three ‘best buys’ identified by the World Health Organization for reducing alcohol-related harms (alongside regulation of marketing and pricing of alcohol). (Overarching recommendation).

6. Research funders (NIHR, CSO, Scottish Government, UK Government) should seek to address gaps in current understanding about mechanisms of impact of availability changes on harms. This should include a scoping review of relevant evidence as well as primary research to fill gaps identified; for example, studies of how children and people in recovery from alcohol problems are exposed to and impacted by alcohol’s spatial and temporal availability and how it may promote the normalisation of alcohol consumption across diverse groups and situations. Further work is needed to create accurate evidence-informed stories of the potential mechanisms for PHTs to draw on when communicating with policy-makers. In the meantime, these potential mechanisms can be borne in mind by public health and licensing stakeholders when making decisions on local licensing policy (Section 4.1.5).

4.2.2 Implications for practice (including public health teams)

1. Given the importance of availability as a key driver of alcohol harms, PHTs should continue to engage with local alcohol premises licensing systems to provide an important and necessary context for policy development and decisions. Given that the value of such engagement remains uncertain in the short to medium term, PHTs should be realistic about how their involvement could influence local licensing policy and practice over time. Their involvement and experiences should be carefully evaluated, as reporting the challenges faced by PHTs, or their failures, may be a prerequisite for improvements in national licensing legislation and is often valued by other stakeholders (Section 4.1.4).

2. Reducing or preventing increases in availability is a long-term aim, unlikely to be realised in the lifetime of a SLP, but increased use of health data and evidence in the formulation of licensing policy and decisions should be recognised as a legitimate and positive outcome of public health engagement (Section 4.1.4).

3. Greater consideration should be given by PHTs and other licensing stakeholders to current evidence suggesting that increases in temporal availability may drive increases in harms (Section 4.1.5).

4.2.3 Implications for research and recommended future research

In what we believe to be the first study of its kind, we developed and applied a novel measure to systematically assess the nature and intensity of public health engagement in local alcohol premises licensing in England and Scotland, identifying and describing 19 distinct activity types. We used this assessment to examine the impact of public health engagement in licensing on health and crime outcomes in the local licensing area over the 7-year period of this study, and found no evidence of such impact. As the first comparative study of licensing in England and Scotland, our findings add to understanding of the licensing systems in the two nations by highlighting legal, structural and philosophical reasons for differences in public health practice on licensing, expanding current knowledge by identifying important influences beyond the presence/absence of a public health objective. We found that PHTs in some Scottish areas are now working well within their local licensing systems and making some progress in establishing the legitimacy and value of their data and input, influencing policy and decisions. This advances earlier evidence by suggesting that initial challenges facing PHTs may be overcome in some areas. We report novel data from a large in-depth interview study that highlights potential mechanisms of impact of alcohol availability that are not typically discussed in current literature, including visibility and affordability. Taken all together, this study adds to knowledge about the current state of the licensing systems in Scotland and England from a public health and public interest perspective. It also highlights other limitations of the systems brought to the fore due to the COVID-19 pandemic, including an inability to address online sales and the lack of a national approach to data, monitoring and support.

We discuss areas for future research below.

4.3.1 Overall strengths

To our knowledge, this is the first study to gather primary data on public health engagement in licensing from professionals across differing licensing regimes, and to seek to quantitatively assess the impact of that engagement on health and crime outcomes. The study was ambitious in both scale and endeavour. The data set is based on 94 structured interviews with public health professionals, alongside extensive documentation analysis from 39 PHTs in diverse communities over a 7-year period. It also includes 53 in-depth interviews with licensing stakeholders, including public health, in 20 local authority areas. The natural experiment that we sought to evaluate – diversity of public health engagement in licensing by different PHTs around the UK – was not created by the presence or absence of a discrete action or policy in different locations. Instead, it was a complex intervention consisting of multiple components that evolved and differed in nature and intensity within and between PHTs over time, without a clear start or end point – and which had not previously been systematically documented or assessed. While not a systems study in the sense that we did not analyse other potential influences on the outcomes in question during the time period in each local area, the use of extensive qualitative data to help understand and explain the quantitative findings fits well with developments in natural experiment evaluations more generally. 138,139 We published a detailed protocol identifying our primary outcome measure,1 and registered our analysis plans in advance (researchregistry6162).

The new PHIAL measure that we developed, and the methods we used to develop it,2 provide a template for future research in the UK and elsewhere, both on public health involvement in licensing and on similarly complex real-world practices in other fields. Our comparison of legal, structural and philosophical reasons for differences in public health practice on licensing between Scotland and England expands current knowledge by identifying important influences beyond the presence or absence of a public health objective. This analysis provides an authoritative and thorough guide to the similarities and differences between the two licensing systems, and consequent implications for public health practice, that is likely to be of interest to those considering the design and implementation of licensing regulations in similar permit-based systems worldwide. Our in-depth interviews provide novel data by considering licensing stakeholder views on the specific mechanisms by which temporal and spatial availability interventions may impact on alcohol-related harms. These findings add considerably to the international literature by proposing potential mechanisms of impact not routinely discussed in relation to availability of alcohol, including visibility of alcohol cues and affordability. Finally, the combination of quantitative and qualitative analysis, in-depth interviews and expert input enabled a comprehensive and realistic assessment of the likely impacts of public health engagement in licensing, highlighting the limitations of the two licensing systems in addressing availability and harms, and recommendations for development of the systems in both nations from a public health perspective.

4.3.2 Strengths and limitations in structured data collection (and exposure measurement)

We sought to capture PHT activity levels accurately across the time period, using extensive documentary and interview data. Given the duration and breadth of data of interest, our large sample of diverse PHTs is a strength and provides examples of a comprehensive range of relevant activities. The activity categories and some of the subcategories (e.g. public involvement; influencing stakeholders) likely have widespread relevance. We took a team approach to coding and analysing the interview data, keeping reflective notes to iteratively develop the coding framework (which ultimately informed the PHIAL measure). We continually discussed the emerging findings across the team and checked and rechecked the coding against the data. We drew on extensive interview data in our analyses and involved legal experts and colleagues from AFS and PHE (as was) to strengthen our knowledge base, in particular when interpreting the potential reasons for differences in practice between Scotland and England.

In terms of limitations, we are unlikely to have captured all possible public health approaches to engagement in alcohol licensing in England and Scotland, and less so where different licensing regimes apply (in Northern Ireland as well as abroad). It is possible that the teams that agreed to be part of the study were biased in terms of being more interested in this issue generally than those which did not. A further possible selection bias could be that teams that did not participate were busier or otherwise under more pressure for time than those which did. In terms of affecting the transferability of our findings, given that we did not find any effect of PHT involvement in licensing on health and crime outcomes even in these interested teams, it is unlikely that the findings would have been different for teams that were on average less interested or too busy to engage with the study. Our data were also limited by recall bias: staff changes and poor records sometimes meant that there were gaps in interviewee knowledge of earlier practices or the reasons for such practices. In many, but not all, cases we were able to interview former staff to minimise the impact of this.

4.3.3 Public health engagement in alcohol licensing measure

Standardisation of measurement is an essential tenet of the scientific method, but applying standard measures within complex public health systems can be challenging. 140 The PHIAL measure is new, has been transparently and robustly developed, and explicitly accounts for the diversity in real-life approaches taken by PHTs as they engage in local alcohol premises licensing. The measure successfully captured differences in type and level of public health engagement in licensing within and between areas over time, with high face validity. Advisory input from a wide pool of PHT and licensing representatives as well as national experts gives us confidence that the PHIAL measure will be relevant and applicable to PHTs across England and Scotland, despite differences in organisational structures and licensing law. Our approach was informed by best practice in developing composite measures. 141 As this is the first measure of its kind in the UK or internationally to our knowledge, there is no gold standard or alternative measure against which we can assess PHIAL.

The measure and the scores generated are subject to several limitations. Through in-depth discussion among our large and varied team, with input from experts, and taking cognisance of prior literature, we made a series of judgements on the scope and the granularity of measurement. These ultimately impacted on what was scored and weighted. Our final definition of relevant PHT activity excludes efforts to change retail practices through direct engagement with existing premises or support for industry-led voluntary best practice schemes (which were in our first draft) but includes efforts to place binding operating conditions on new licences. This reflected our primary interest in licensing systems rather than business practices, and evidence suggesting that such direct working was unlikely to have significant impact. 142,143 There has also been some research into industry-led local initiatives which, though not part of the licensing system, are sometimes delivered in partnership with local authority stakeholders involved in licensing. 144–146 For example, voluntary withdrawals of cheap, high-strength beers and ciders from off-licences may increase the average price of certain types of available beverage,144 but can be circumvented by shoppers purchasing different products or shopping at non-participating stores. 145 We discussed these initiatives in our consideration of mechanisms of impact in in-depth interviews, but they would not have been captured by the PHIAL measure except in so far as they formed part of negotiations on new licence applications.

In our comparison of practice between Scotland and England, we did not statistically analyse differences, as the study was not designed to do so; thus we only explore differences here that are very clearly visible in the data reported (see Figure 3). Secondly, some of the 19 types of PHT activity cover more than one specific activity and we are unable to disaggregate these to compare practice in Scotland and England at a more granular level. 49

We assessed activity over a lengthy period (7 years), which meant that some uncertainties in historical activity could not always be resolved. Having exhausted all PHT sources, we resolved uncertainties in discussion by taking an ‘on the balance of probabilities’ approach where necessary. This limitation would not apply to a prospective study in which the PHIAL measure could be applied to contemporaneous public health activity. The grading process for individual PHTs inevitably involved some subjectivity of judgement, especially where data quality was lower. We sought to reduce variability by asking graders not to grade data if unsure, keeping reflective logs of uncertainties, having a second researcher review all grading, and resolving all issues by consensus. Grading guidance notes were added to the measure as needed. While this will have improved the reliability of grading, we were unable to conduct formal inter-rater reliability checks due to changes of staff within the study team.

4.3.4 In-depth interview data

The size and diversity of our data set – drawn from 53 interviews with public health, police and licensing stakeholders, including local authority lawyers and elected politicians from regions in both Scotland and England – is a key strength of this element of the study. The use of lengthy qualitative interviews provided each interviewee with the scope to discuss a wide range of views and issues based on their own understanding of what was important. While the number of participants was fewer than we originally envisaged (80), the interviews were longer than anticipated and it is questionable whether a larger sample size would have offered value for money or added substantially to our findings. We did not include PHTs that had little or no engagement in alcohol licensing, though we were able to recruit 19 of them for structured interviews. Their rationale for not engaging may have shed further light on challenges or limitations in this work, though those are arguably well covered in earlier work. 76–78,81,85,105

Some public health interviewees struggled to articulate their rationale for involvement in alcohol premises licensing or the mechanisms of impact of availability on harms at the time of interview; interview topics could be provided to interviewees in more detail in advance to overcome this limitation in future studies. The findings on public health approaches to licensing are necessarily limited to the experiences of these participants. We did not routinely collect data on participants’ years of relevant experience, so we could not examine whether that affected approaches or mechanisms discussed. The mechanisms identified by interviewees are necessarily grounded in an England/Scotland perspective but, given the novelty of the work, have potential to contribute to theory, practice and research internationally.

4.3.5 Quantitative analysis of impact of public health team engagement on licensing

The study was based on a matched sample of areas of higher and lower engagement that were comparable with respect to other relevant characteristics. Furthermore, inferences were strengthened by the longitudinal nature of the data. Temporal patterns of primary exposure varied between and within areas over time, avoiding erroneous inferences resulting from positive or negative correlations between exposure and outcome observed in all areas across the time period. This is similarly a strength for the ‘change in PHIAL score’ metric, but not for cumulative exposure, which increases over the study period. An important limitation of this study is that the PHIAL exposure metric itself is not intuitively interpretable. Although the semiquantitative score enables the assessment of correlations between exposures and outcomes, the practical implication of observed associations is unclear. The analyses would further have benefited from additional information on alcohol licensing policies and environmental changes that, for example, may have affected alcohol availability and related harms in the areas, and which would have enabled modelling of complete hypothesised causal pathways from PHT engagement influencing licensing policies and decisions which in turn might impact on alcohol-related harms. We had originally intended to develop a second intensity measure exploring the strength of the local licensing system from a public health perspective, but we had underestimated the work involved in developing a new composite measure. We did not have the capacity to develop a second such measure, given that this would have required extensive manual data collection because of a lack of centralised, systematic and accurate data on licensed premises. It is likely that such a measure would still be useful for future analyses. Our sample size was informed by statistical power calculations for alcohol-related hospital admissions and crimes based on relevant previous research1 but which was not an exact parallel, as this was the first study of its kind. As our focus was evaluating the indirect effect of PHT engagement on downstream harms, it is likely that, if associations exist, the effect size would be smaller than in the studies used in our statistical power analyses. This quantitative aspect of the study would have benefited from a larger sample size. However, with 39 areas and 119 interviews already included in the qualitative elements that generate the PHIAL scores, adding more PHTs to the study would quickly have become prohibitively resource-intensive.