Specialist alcohol inpatient treatment admissions and non-specialist hospital admissions for alcohol withdrawal in England: an inverse relationship

ABSTRACT Aims We assessed the relationship between specialist and non-specialist admissions for alcohol withdrawal since the introduction of the UK government Health and Social Care Act in 2012. Methods Using publicly available national data sets from 2009 to 2019, we compared the number of alcohol withdrawal admissions and estimated costs in specialist and non-specialist treatment settings. Results A significant negative correlation providing strong evidence of an association was observed between the fall in specialist and rise in non-specialist admissions. Significant cost reductions within specialist services were displaced to non-specialist settings. Conclusions The shift in demand from specialist to non-specialist alcohol admissions due to policy changes in England should be reversed by specialist workforce investment to improve outcomes. In the meantime, non-specialist services and staff must be resourced and equipped to meet the complex needs of these service users.


BACKGROUND
In England, alcohol-related hospital admissions have risen from 493,760 in 2003/04 to over 1.26m in 2018/19 (Public Health England, 2020. A recent study estimated 1 in 5 patients admitted to hospital experience harmful drinking, with 1 in 10 experiencing alcohol dependence (Roberts et al., 2019). A disproportionate impact of alcohol on the National Health Service (NHS) is exerted by those with chronic alcohol disorders accessing care via emergency departments (Phillips et al., 2019). Alcohol-related disorders have been estimated to cost the NHS £3.5bn per year (Department of Health, 2013).
Effective treatment services should respond to the full spectrum of risks and acute, chronic and complex needs (Babor et al., 2008), with inpatient care treating those at greatest risk of severe presentations by providing medically assisted alcohol withdrawal (National Institute for Health and Care Excellence (NICE), 2010, 2011). However, reductions in public health funding in England to commission specialist alcohol treatment since the introduction of the Health and Social Care Act (2012) (HSCA) have been associated with the closure of numerous specialist inpatient units and increasing pressures on acute hospital services (Robertson et al., 2017;HC Deb 2019;Drummond, 2017).
We examined publicly available data to estimate the potential impact of the HSCA policy changes before and after 2013 in relation to access to specialist and non-specialist inpatient admissions for alcohol withdrawal care.

METHODS AND DESIGN
We conducted secondary analysis of publicly available official national data from 2009 to 2019 to examine the associations between specialist and non-specialist admissions following the introduction of the HSCA.
All adult services commissioned to provide specialist alcohol treatment in England report the number of inpatient treatment episodes to the National Drug Treatment Monitoring System (NDTMS). From 2009/10 to 2012/13 data are available on the number of service users accessing 'planned inpatient alcohol detoxification' to manage alcohol withdrawal. This definition was changed in 2013/14 to identify those service users accessing 'inpatient prescribing interventions'. We have treated data on 'inpatient alcohol detoxification' and 'inpatient prescribing interventions' with equivalence and defined these as 'specialist admissions'.
Hospital Episode Statistics (HES) are provided by NHS Digital, and include the annual number of non-specialist hospital admissions for a primary or secondary diagnosis of alcohol withdrawal . The number of admissions for alcohol withdrawal is based on methodology originally developed by the North West Public Health Observatory (Jones et al., 2008), which avoids double counting of specific alcohol diagnoses. We used diagnostic codes defined by the International Classification of Diseases, tenth version (ICD-10; WHO, 1992), to identify the number of alcohol withdrawal admissions (F10.3) from 2009 to 2019, which we have defined as 'non-specialist admissions'. Admissions recorded as 'withdrawal state with delirium' (F10.4) were excluded from this analysis as these conditions represent a medical emergency requiring acute hospital settings only.
Our study also examined the cost of admissions in the two settings. The estimated daily cost for specialist admissions set at £341 per bed day was taken from research commissioned by UK Department of Health (Brennan et al., 2019). The average length of specialist admissions varies (e.g. 7-14 days), however, we defined the length of specialist admissions as 10 days according to previous research conducted in the UK (Parrott et al, 2006).
Non-specialist admission costs for alcohol withdrawal in NHS hospitals were extracted from our previous study examining the burden of alcohol disorders on non-specialist settings (Phillips et al., 2019). This study identified 231,237 individuals with alcohol disorders with a mean annual total cost related to hospital admissions of £6749 per individual accounting for a mean total of 15.14 days, which equates to £446 per bed day. Those admitted with alcohol withdrawal experienced on average 3.36 admissions/year and 17.23 total bed days. Therefore 5 days (i.e. 17.23/3.36) was estimated to be the mean length of non-specialist admissions. All costs were adjusted using UK Gross Domestic Product deflator calculations (HM Treasury, 2019).
The annual number of people accessing specialist and nonspecialist admissions for alcohol withdrawal from 2009/10 to 2018/19 was presented as a time series. The mean number of admissions and costs before the implementation of HSCA (Time point 1 (T1) = 2009-2014) was compared with data since the policy change (Time point 2 (T2) = 2014-2019) using the Mann-Whitney U test. Correlations were reported using the Spearman's rank correlation coefficient. We utilized R language 3.5.0 and Stata 15 for these analyses.

RESULTS
Prevalence estimates of the number of adults with alcohol dependence in potential need of specialist treatment remained static over the assessed period. However, the percentage of adults accessing specialist admissions fell from 11.3% in 2009/10 to 5.8% in 2018/19 (see Table 1).
Before 2013/14 the mean annual number of specialist admissions remained relatively static at 10,012 admissions. However, the mean annual number of specialist admissions decreased to 5453 admissions between 2013/14 and 2018/19, a reduction of 45.5%. Overall, annual specialist admissions decreased by a mean of 4559 comparing T1 to T2, which was statistically significant (P = 0.01).
By contrast, there was a 43.8% increase in non-specialist admissions between 2009/10 and 2018/19, increasing by a mean of 4736 per annum comparing T1 to T2, which was also statistically significant (P = 0.01). Figure 1 presents changes in the number of specialist admissions and non-specialist admissions relative to a baseline in 2009/10. When considered together, the overall number of adults accessing either specialist or non-specialist admissions for alcohol withdrawal appears relatively stable, suggesting that the reductions in specialist admissions has been displaced to non-specialist admissions since 2013/14.
The annual total specialist admissions were compared with non-specialist admissions for the same years using the Spearman's rank correlation. A negative correlation was observed (r = −0.93, P < 0.01), providing strong evidence of an association between the annual reduction in specialist admissions and the annual increase in non-specialist admissions.
The annual estimated total costs of specialist admissions reduced from £31.7m to £15.9m from 2009/10 to 2018/19. Comparing T1 and T2, these costs decreased by a mean of £13.6m per annum: a statistically significant reduction of 41.3% (P = 0.01). In contrast the estimated costs for non-specialist admissions increased by 67.3% between 2009 and 2019: a mean increase of £16.3m per annum comparing T1 to T2, which was statistically significant (P = 0.01). Despite the decrease in specialist admission costs and increases in non-specialist admission costs, the overall total cost estimates remained relative static over the reporting period with no significant difference in mean cost between T1 to T2.
The average cost per admission varied over time and between settings with a specialist admission at £3180 in 2009/10 increasing to £3700 in 2018/19, compared to non-specialist admission £2,230 and £2595 respectively. Despite the cost per day being lower for specialist admissions, the estimated length of stay was twice that of non-specialist admissions (i.e. 10 days versus 5 days). Therefore, the estimated cost of each non-specialist admission was 29.8% less than for specialist admissions. The overall cost per admission remained relatively unchanged from 2009/10 to 2018/19 and was not found to be statistically significant between T1 and T2.

DISCUSSION
Our analysis identifies that before the introduction of the HSCA, specialist alcohol services offered approximately 10,000 specialist admissions annually. This study found from 2013/14 to 2018/19 Table 1. Annual reported data for specialist (NDTMS) and non-specialist admissions (HES) for alcohol withdrawal, including estimated costs    a decline in the clinical population accessing specialist treatment and a significant rapid reduction in specialist admissions for alcohol withdrawal. The parallel increase in non-specialist admissions and the strong statistical association between these trends suggest a net substitution of care from specialist to non-specialist admissions since the implementation of the HSCA. However, the overall total number of all admissions for alcohol withdrawal interventions remained relatively static, emphasizing a continued and increasing need for inpatient treatment for alcohol withdrawal. The transfer from specialist to non-specialist admissions was also reflected in the costs for inpatient care that are met by different governmental organizations. We found a significant shift in costs for alcohol withdrawal from local authorities responsible for specialist admissions to local NHS commissioners responsible for the provision of non-specialist admissions.
It is important to acknowledge several limitations. Firstly, the definitions used to describe alcohol withdrawal admissions vary between time points and settings. It is widely acknowledged that inpatient treatment under specialist alcohol services will involve medically managed alcohol withdrawal regardless of their coding, be it 'inpatient alcohol detoxification or inpatient prescribing interventions'. Equally, non-specialist admissions for alcohol withdrawal are based on the clinical presentation of alcohol withdrawal symptoms requiring clinical management and therefore provide a comparable data source.
Secondly, the economic analysis provides estimates based on aggregate costs and lengths of admission across both settings. We chose a more conservative period for specialist admissions based on literature that was more reflective of recent practice. The length of stay for non-specialist admissions is drawn from a national sample in 2009/10. While we feel this accurately reflects the average length of admission at this time, it is possible that variation in practice exists, skewing costs.
Thirdly, while the rapid decline in specialist admissions coincides with the service changes prompted by the HSCA, and further exacerbated by significant reductions to the public health grant to local authorities in 2015/16 (Department of Health, 2015) used to support specialist alcohol treatment, this time series analysis is unable to identify a specific causal relationship between the fall and rise in alcohol withdrawal admissions.
International literature examining alcohol treatment systems has identified that, as unmet demand for treatment decreases in one part of the system, there are increases in another part of the system (Ritter et al., 2019). Significant evidence of the effectiveness and costeffectiveness of specialist alcohol treatment exists (NICE, 2011), with improved outcomes being experienced when specialist admissions are integrated with community treatment (Eastwood et al., 2018). Previous studies in North America and Scandinavia have identified that increases in effective specialist alcohol treatment are associated with decreases in liver morbidity (Smart et al., 1996), liver and allcause mortality (Holder & Parker, 1992;Rautiainen et al., 2019) as well as hospital admissions (Smart & Mann, 2000). A recent governmental inquiry into the fall in the number of people entering specialist alcohol services between 2013 to 2017 concedes financial pressures increased barriers and reduced access to specialist treatment (Public Health England, 2018b). It is therefore plausible that the observed increase in non-specialist admissions may in part be influenced by the reduced ability of specialist services to provide the full recommended range of interventions including specialist admissions for alcohol withdrawal.
Equally, it might be argued that the clinical characteristics of those requiring treatment have changed over the last decade, meaning that specialist services are unable to fully meet the needs of severely ill service users experiencing alcohol withdrawal. Firstly, there has been an ageing cohort of harmful and dependent drinkers with comorbid conditions (Drummond et al., 2016). In addition, increasing non-specialist admissions for alcohol-related liver disease has worsened particularly in deprived areas (Williams et al., 2020), pointing towards greater clinical complexity. Our previous study identified that unplanned non-specialist admissions were greater in those with chronic alcohol disorders, including alcohol withdrawal, than those with an acute or no alcohol disorder (Phillips et al., 2019). Therefore, the shift in demand from specialist to non-specialist admissions might be best explained by reductions in funding exacerbated by the increasing complexity and clinical needs of the service users.
Despite the higher individual bed day cost, non-specialist admissions are briefer than recommended (NICE, 2011) and significantly shorter compared to specialist admissions, thereby reducing the cost of each admission. Shorter lengths of admission suggest that many will continue to experience withdrawal features on discharge, promoting potential relapse and readmission (Yedlapati and Stewart, 2018). Furthermore, socially disenfranchised groups who experience greater risk factors are less likely to engage in follow-up treatments and are more likely to be readmitted (Neighbors et al., 2018).
To realize the full public health benefits of alcohol treatment, there is a requirement to evaluate and develop a system of care based on population needs that ensures services are accessible, efficient and appropriately resourced (Babor et al., 2008;Rush & Urbanoski, 2019). Although desirable, it is unlikely that the closure of specialist inpatient units will be reversed in the short-term. We suggest the remaining specialist services should be preserved to support integrated care pathways and act as specialist training centres. The reduction in specialist admissions means that specialist care for alcohol withdrawal is being transferred to non-specialist care settings that may be less equipped to meet the current and predicted increase in alcohol-related hospital admissions (McQuire et al., 2019). The recent development of care pathways and hospitalbased Alcohol Care Teams may promote the completion of alcohol withdrawal programmes in the community that shorten nonspecialist admissions (Public Health England, 2018a; NHS England, 2019). In addition, these initiatives provide opportunities for specialist and non-specialist practitioners to develop shared competencies (Phillips et al., 2020), innovative service models, deliver comprehensive packages of care within the hospital and integrate effective community treatments (Moriarty, 2019;Drummond et al., 2019).
Previous commentators have warned of the consequences of cuts to specialist treatment services following the introduction of the HSCA (Drummond, 2017). Our analysis suggests that those in need of specialist inpatient care are likely to be disproportionately affected by changes in funding following the implementation of this policy. Service models within non-specialist care settings should evolve within an integrated model of provision to ensure the needs of service users are fully met.