Rapid evaluation of the 2022 to 2023 discharge funds

Question 1

Executive summary

Accepted Answer

During 2022, there were a rising number of patients who were still in hospital when they did not clinically meet the criteria to reside. To improve flow through hospitals and free up hospital beds over the winter, the government announced 2 additional funding streams:

the £500 million Adult Social Care Discharge Fund, in September 2022, for 2022 to 2023 (‘£500 million discharge fund’)
the £200 million step down fund for integrated care boards in January 2023 (‘£200 million discharge fund’)

A rapid evaluation of this discharge funding was conducted exploring:

how much of the funds were spent
changes in discharge metrics over this period until June 2023
an analysis of the impact of the funding

Local areas have successfully spent:

£486 million of the £500 million discharge fund on a range of services including workforce, home care, intermediate care and residential care
£138 million of the £200 million discharge fund on step down care including on beds, wraparound care and therapeutic packages

Acute discharge delays data shows that the number of patients with no criteria to reside (NCTR) and not discharged from hospital increased between October 2022 and January 2023 to a peak of 14,385 before falling to an average of 13,452 in the week ending 31 March 2023. NCTR continued to fall in the subsequent 3 months to 12,200 in the week ending 30 June 2023 (the scope of metrics in this evaluation). There is evidence that this fall is associated with the discharge fund.

Local areas reported having been able to use the fund to begin improving key elements of the system that contribute to discharge delays and introduce schemes to support flow through the system. The results highlight progress across important areas including:

workforce and recruitment
securing care capacity
patient support

However, there are some barriers and challenges that may have affected the use and impact of the funding, including:

tight time frames
increasing patient demand
workforce shortages

How local areas spent the £500 million discharge fund 2022 to 2023

Local areas spent a total of £486 million (97%) of this winter’s £500 million discharge fund, and the majority of local authorities and integrated care boards (ICBs) spent the entirety of their allocated funds.

The fund was used across a range of service types with the largest area of spend being on workforce (£107 million), followed by home or domiciliary care (£97 million) and intermediate bed-based care (£95 million).

The fund supported a range of activities to secure additional care capacity, including block-booking beds, procuring specialist beds, and working with the community to provide interim care and accommodation.

Local areas also used the fund to improve workforce capacity and recruitment by expanding and reallocating workforce provision, creating new roles, launching recruitment initiatives, providing existing staff with benefits to improve retention, working with volunteers and utilising technology.

Many local areas invested in patient support and preventive interventions to help patients leaving hospital and prevent future admissions. This included mental health services, personal health grants, additional home care visits, health and safety advice, and housing support

Most local areas identified key barriers and challenges that could limit the impact of the fund. These include persistent national and local workforce shortages, difficulties securing bed capacity (particularly in smaller areas or for complex cases), increasing demand for patient services, tight timings for mobilisation, and uncertainty around future funding.

Some local areas identified challenges around monitoring data collection, including uncertainty around whether impact can be observed or directly attributed to the fund.

How local areas spent the £200 million discharge fund 2022 to 2023

Local areas split their spending of the £200 million discharge fund across step down beds and packages of care for those who occupied the beds.

Of the total 7,134 step down beds purchased nationally, the regions that purchased the most were the North West with 26%, and North East and Yorkshire with 23%. Almost 10,500 people were moved to step down beds using the fund, of which the Midlands moved 21%, and London moved 18%.

Nationally, 19,441 packages of care were purchased with the £200 million discharge fund. Of this total, the South East purchased 24%, and London purchased 22%.

The impact of the funding on key outcome metrics

Overall, there is some evidence that the funding is associated with a positive impact for patients with NCTR. There was a significant break in the growing trend of NCTR patients that coincided with the implementation of the funding.

It was not possible to consistently link greater improvement in NCTR to areas with higher spend. Challenges to evaluation as well as data quality might have affected these findings.

The number of discharges with further support continued to grow nationally after the funding, but there was no break in the previous trend. It was also not possible to conclusively attribute greater increases in discharges to greater spend in some areas. This estimation was further complicated by data quality.

There was no observed effect of the funding on hospital admissions and discharges with no support.

Conclusions and considerations for future funding

The findings suggest that, in most cases, the funding was used to increase the number of discharges and support the reduction of discharge delays. Areas spent the £500 million discharge fund to improve workforce capacity, care provision and patient support.

While most areas spent all their £500 million discharge fund allocation, some areas were unable to spend as planned. Workforce spending was limited in some areas due to challenges around recruitment. Local areas also reported that planning for the funds was made challenging due to the short-term nature of the funding. Spending from the £200 million discharge fund was more varied across areas.

There is evidence that the funding was associated with reduced NCTR levels. However, local areas reported that, due to barriers such as workforce challenges and increasing demand on services, the impact of the funding could have been limited or delayed. The ability to estimate any impact is limited by factors related to data quality and policy characteristics.

Considerations for future funding include the following:

future funding should be distributed over a longer period, with as much advance notice as possible to allow local areas to implement interventions with longer lead-in times
future funding should provide as much certainty as possible regarding continuity and exit planning. Additional funding should be seen as one part of the government’s response, alongside policy development to address longer-term or systemic issues
any grant conditions for future funding should be as clear as possible. As the thematic analysis does not provide much information on which aspects of the conditions were unclear, this should be explored further via other primary research and engagement with local areas
monitoring data requested from local areas should be meaningful, necessary and proportionate. The Department for Health and Social Care (DHSC) should consider how data from the 2022 to 2023 fund was used to identify what is required in future. This should include consideration of the frequency of data collection, and how we can ensure consistency and quality across local areas. The clarity of qualitative questions should also be reviewed, particularly in relation to innovation and impact. DHSC should engage with local areas to understand how best to collect the necessary data while minimising burden, and test plans to ensure that data collected will be meaningful
evaluating impact of discharge interventions is challenging. These challenges can be reduced by improved data quality, longer time series and considering evaluation in the policy design phase
data collection should also be considered in the context of other funding, including the Better Care Fund where spend on discharge may overlap, to help understand how different allocations are spent

Question 2

1.  Introduction

Accepted Answer

There had been substantial pressure on hospital beds in autumn 2022, with people experiencing lengthy waits to be admitted and ambulances facing difficulties handing patients to emergency departments. Contributing to this was the increasing number of people remaining in hospital despite being clinically ready to leave.

In September 2022, the government announced Our plan for patients to improve care for patients in winter 2022 to 2023 and beyond. This plan included launching a £500 million discharge Fund to provide additional funding to help speed up safe discharge of patients from hospital and into social care support.

The intention was to free up hospital beds as well as to help to retain and recruit more care workers. This was in the context of increasing numbers of patients being in hospital each day who no longer met the criteria to reside over 2022. Achieving this was expected to improve the flow in emergency departments and help reduce ambulance delays.

The details of the £500 million discharge fund were announced on 17 November 2022. £300 million was given to ICBs, and £200 million to local authorities to free up the maximum number of hospital beds and bolster the social care workforce. Local authorities and ICBs – organisations that bring the NHS together locally to improve health in the community – worked together to agree on spending across their regions, introducing tailored solutions to speed up discharge and benefit patients in their areas.

In addition to the £500 million discharge fund, a further £200 million discharge fund was announced on 9 January 2023. This fund supported discharging patients from hospital beds into step down beds to improve patient care and system flow. It has been considered alongside our assessment of impact.

The January announcement also included nearly £50 million of capital funding for ambulance hubs and discharge lounges to free up hospital beds. This capital funding is out of scope of this report.

Question 3

2.  Background and policy context

Accepted Answer

Enabling people to be discharged from hospital more quickly with the right care and support in place contributes to speedier recovery and better outcomes for patients, including reducing the risk of:

medical complications such as deep-vein thrombosis
hospital-acquired infections
loss of independence

This in turn improves patient flow and reduces delays elsewhere in the system.

The health and care system has faced significant challenges from prior to the COVID-19 pandemic with admissions rising and delayed discharges being a barrier to patient flow. Accident and emergency (A&E) attendances were around 1,949,762 in April 2017, rising to 2,112,165 in April 2019. During the pandemic, attendances dropped substantially. They returned to previous levels of 2,028,796 in April 2022 and have been relatively similar since. Acute discharge delay data shows the number of patients with NCTR and not discharged from hospital increased between October 2022 and January 2023 to a peak of 14,385.

During the pandemic, hospital discharge guidance was changed and adapted to ensure that hospitals, community providers and local authorities were able to adapt to the changing context of the impact of the pandemic. Guidance released on 21 August 2020 instructed the health and care system to continue with a ‘home first’ discharge-to-assess model, and resume NHS continuing healthcare (CHC) assessments in community settings from 1 September 2020.

From 1 September 2020 to 31 March 2021, government funding of £588 million was in place to provide funded support packages of care and rehabilitation for individuals with new care requirements on discharge from hospital. The funding model provided up to 6 weeks of centrally funded care for new or additional needs on discharge from hospital.

A hospital discharge fund of £476 million was allocated to integrated care systems to support discharges with up to 6 weeks of funded care for the period between 1 April and 30 June 2021, and up to 4 weeks of funded care for the period between 1 July and 30 September 2021. A further £490 million was allocated for the period between October 2021 and March 2022 to cover up to 4 weeks of funded care until 31 March 2022.

The hospital discharge fund continued between October 2021 and 31 March 2022 to cover 4 weeks of funded care. Following this, it was expected that local systems continued to make best use of existing resources and local budgets to support safe and effective discharges, within local priorities, based on existing joint arrangements and best practices agreed locally.

The £500 million discharge fund for 2022 to 2023 aimed to support timely and safe discharge from hospital into the community by reducing the number of people delayed in hospital awaiting social care. The funding intended to help local authorities and ICBs to:

deliver additional care packages and beds
provide equipment to support people to return home
boost the social care workforce

The funding was to be used flexibly by local health and care systems to enable more people to be discharged from hospital, including mental health inpatient settings, to the most appropriate setting, with health and social care support as required. 

The £500 million discharge fund was pooled into the Better Care Fund and shared locally – with 40% (£200 million) distributed to local authorities and the remaining 60% (£300 million) to ICBs – to facilitate joint planning and decision-making across health and care systems. The £500 million discharge fund was provided in 2 tranches:

the first 40% in December 2022
the remaining 60% by the end of January 2023, which was conditional upon reporting requirements being met

This funding provided additional support to systems beyond existing spend from other budgets such as the Better Care Fund.

The government provided a further £200 million to the NHS in January 2023 to fund step down capacity, and £50 million in capital spending to expand hospital discharge lounges and ambulance hubs. The £50 million capital spending is not in scope of this evaluation report.

The Better Care Fund also committed over £7 billion in 2022 to 2023 to enable people to stay well, safe and independent at home for longer by funding things like adaptations to homes for disabled people and rehabilitating people back into their communities after a spell in hospital. Although this funding supports a wider range of services, it includes services that enable discharges from hospital. The Better Care Fund is not in scope of this evaluation report – however, this has an impact on the services provided and discharge performance.

Hospital discharge process for delayed discharges

The hospital discharge process for patients is a complex process.

Figure 1 is a simplified view of the hospital process for discharge using a stock and flow diagram. The stock and flow diagram illustrates the ‘stock’ of patients with criteria to reside or NCTR as those patients in hospital at a given point during the day. The ‘flow’ of patients are those that enter hospitals, then leave either with a normal discharge or via one of the NCTR pathways.

Figure 1: simplified stock and flow diagram of the hospital process for discharge

The diagram above (Figure 1) shows a simplified version of the flow of patients through the hospital system. An arrow of admissions rate flows into the total stock of patients meeting the criteria to reside in hospital.

This stock has arrows coming out that represent the flows of mortality rate, discharge with no delay and discharge approvals.

Discharge approvals flow into the stock of patients who do not meet the criteria to reside. Within this stock, patients are split into those discharged on the same day as being recorded as NCTR and those discharged at least a day later. Discharges that happen at least a day later than being first recorded as NCTR are delayed discharges. When patients recorded as NCTR (delayed and not delayed) are discharged, these discharges are categorised according to pathways 0, 1, 2, or 3.

Every hospitalised patient’s criteria to reside are reviewed daily to determine the rationale for their continued acute hospitalisation. A hospital patient can be considered fit for discharge to a less acute setting if they do not meet any of the criteria to reside (NCTR status). If a patient is still hospitalised a day or more after being assessed as no longer meeting the criteria to reside, they are considered to have a delayed discharge. Criteria to reside are standardised across all NHS trusts and include 11 conditions warranting hospitalisation in an acute setting.

Discharge data is grouped according to pathways. Pathway 0 indicates discharge with no additional support, and pathways 1, 2 and 3 indicate various levels of support provided. More information on the discharge pathway definitions is set out below.

Pathway 0

Simple discharge home or to usual place of residence (or to temporary accommodation) co-ordinated by the ward without involvement of the care transfer hub. Pathway 0 involves:

no new or additional health and/or social care and support
self-management with signposting to services in the community
voluntary sector support
restart of pre-existing home care package at the same level that remained active and on pause during the person’s hospital stay
returning to original care home placement with care at the same level as prior to the person’s hospital stay

Pathway 1

Discharge home or to usual place of residence (or to temporary accommodation) with health and/or social care and support co-ordinated by the care transfer hub. Pathway 1 involves:

home-based intermediate care on a time-limited, short-term basis for rehabilitation, reablement and recovery at home
restart of home care package at the same level as a pre-existing package that lapsed
returning to original care home placement with time-limited, short-term intermediate care
long-term care and support at home following a period of intermediate care in the community

Pathway 2

Discharge co-ordinated via the care transfer hub to a community bedded setting with dedicated health and/or social care and support. Pathway 2 involves:

bed-based intermediate care on a time-limited, short-term basis for rehabilitation, reablement and recovery in a community bedded setting (bed in care home, community hospital or other bed-based rehabilitation facility)

Pathway 3

In rare circumstances, for those with the highest level of complex needs, discharge to a care home placement co-ordinated via the care transfer hub. Pathway 3 involves:

care home placement for assessment of long-term or ongoing needs and facilitation of patient choice in relation to the permanent placement
long-term care and support in a care home following a period of intermediate care in the community

Question 4

3.  Methodology

Accepted Answer

This rapid evaluation was conducted using a multi-method approach that involved:

analysis of the key metrics to assess discharge delays and the wider context
analysis of £500 million discharge fund fortnightly returns from ICBs and local authorities, including monitoring data
descriptive statistics on the £200 million spend
impact analysis of the funding
in addition, primary qualitative work with 6 local areas has been conducted by the King’s Fund. This report, Hospital discharge funds: experiences in winter 2022–23, has been published separately

Analysis of key metrics to assess discharge delays and the wider context

The primary focus of the discharge funding was to:

reduce delayed discharges
improve access to urgent and emergency care through improved flow of patients in hospital
improve patient outcomes

The key discharge metrics used to measure this were:

the number of patients in acute hospitals with NCTR and not discharged by the end of the day
the number of discharges from NCTR into each pathway
the total number of discharges

These are useful metrics to understand the performance of hospitals in relation to discharge. However, they have limitations as they can be affected by a range of factors such as admissions.

For example, consider a situation where the number of patients admitted into a hospital increases by 10%. This could have an effect of increasing the number of patients that are declared NCTR, which, in turn, is likely to increase the number of patients who have no criteria to reside at the end of the day and are not discharged.

It may be the case, in this scenario, that both admissions and discharges increase, but discharges do not increase at a level to match the increased admissions. The number of patients who have no criteria to reside and are not discharged by the end of the day could also increase. The increase in NCTR could mask that the hospital has actually improved its flow rate.

Although there are more patients in hospital declared NCTR and delayed, the hospital has also increased the number of admissions in this scenario. Therefore, the flow in the hospital has increased. The total bed occupancy would also have increased in this scenario which, when too high, can be a detriment to flow through hospital.

Discharge metrics are also likely to be affected by seasonality. The NCTR data has only been recorded since the COVID-19 pandemic, so there is limited historical data to provide comparisons.

It is also difficult to make consistent comparisons as the pressures seen in 2020 and 2021 are not comparable to a normal winter. Hospitals did not operate in the manner that would be observed outside the pandemic, with restrictions on admissions, elective care and bed usage affecting NCTR data.

Due to such wider factors, it is important to also consider other metrics that can help explain the trend in discharge performance. The key metrics considered are:

number of admissions
bed occupancy

There are interactions between these metrics and discharge metrics over time. For example, increased admissions are likely to lead to higher bed occupancy and higher discharges.

Time period under review

The details of the £500 million discharge fund were published 17 November 2022. The baseline period chosen is the month of October 2022. This is due to October being close to the winter months and being prior to the announcement of funding detail for each local authority or ICB.

The period considered to look at the spend and activity of the £500 million discharge fund is January 2023 to March 2023. This aligns with the fortnightly reporting period for local authority data on spend and activity for the discharge funding. However, we continue to monitor the metrics until the end of June as the funding is likely to have had continued impact beyond the period of spend.

Analysis of fortnightly returns from ICBs and local authorities, including monitoring data

As part of the monitoring of the £500 million discharge fund, a reporting template for spending plans was made available to local areas to set out how the funding would be used within the context of funding conditions.

The template included an expenditure sheet and scheme types that focused on spend that was relevant to the funding, and additional expenditure categories related specifically to recruitment and retention of the social care workforce. The spending plans were due to be submitted by 16 December 2022.

Thereafter, fortnightly activity reports were submitted, for each local authority footprint, detailing what activities had been delivered in line with commitments in the spending plan. An end of year report was also to be submitted (alongside the wider Better Care Fund end of year report) detailing total spend of the £500 million discharge fund by 2 May 2023.

As part of the data returns for monitoring of the fund, local authorities were asked a series of open text questions about their progress using the £500 million:

Question 1

“Please use the space below each theme to describe progress made in this period to use the additional funding to improve discharge outcomes. Where possible, please also give an indication of realised or expected impact on reducing delays.

“Where you have identified a shortfall in capacity, indicate the main causal factors. This might include:

a) progress in securing additional workforce, or increasing hours worked by the existing workforce
b) progress in commissioning additional domiciliary care and intermediate care capacity
c) other activity funded through this additional funding
d) new or innovative initiatives
e) any other themes.”

Question 2

“Please use this section to briefly describe:

a) any barriers or challenges you have faced in spending the discharge fund
b) level of confidence in your ability to spend the funding to impact on discharge delays.”

To support the interpretation of the quantitative data, and understand the uses and impact of the fund in more detail, responses to the qualitative questions have been analysed using thematic analysis. The findings capture a national snapshot of progress using the fund and any barriers or challenges faced, rather than at an individual local authority or ICB level.

Further qualitative work conducted by the King’s Fund focuses on local level issues in case study areas. The work involved interviews with ICBs, local authorities, acute trusts and stakeholders across 6 different areas.

The research captured information at a local level about how the funding had been used in the area, how decisions across different parts of the system were agreed and what barriers were faced.

The findings from this work, Hospital discharge funds: experiences in winter 2022–23, have been published separately.

Thematic analysis is a widely used analytical technique to describe and understand large volumes of qualitative data. It involves systematically reading and reviewing the data to identify key themes and subthemes by coding the findings according to the different topics covered.

In total, we received 4,456 written responses across 7 waves of fortnightly returns. Data cleaning removed 1,587 duplicates, invalid responses and ‘no change’ answers, leaving us with 2,869 to include in the analysis. Due to the high volume of responses, repetition across submissions, changes to the form template and most detailed responses occurring in the first 2 waves, we took a random sample instead of analysing them all. As part of the quality assurance process, we analysed another set of responses to ensure nothing had been missed.

In total, 1,125 responses were analysed, using the following iterative approach:

reviewing responses, and producing a list of themes and subthemes (code frame)
drawing a random sample of responses for analysis
coding – using the code frame to tally where themes occurred
consolidating and organising findings, and highlighting any new themes or subthemes
quality assurance, checking accuracy of coding and purposive coding of 83 additional responses
testing the results by re-applying themes and subthemes to the data

For more details on the approach to sampling and analysis, see ‘Thematic analysis methodology’ in the annexes.

Descriptive statistics on £200 million spend

Trusts were asked to record what the £200 million discharge fund was spent on with daily returns.

Impact analysis of the funding

The aim of this impact analysis was to better understand the impact of the discharge funding and whether it achieved its overall aims.

The research question we are aiming to answer is:

Did 2022 to 2023 discharge funding have a causal effect on the outcomes of interest at a local level?

Data used for impact analysis

This impact evaluation considers 3 outcome metrics. They are:

the number of patients with NCTR and not discharged by the end of the day
the number of discharges (either supported or not supported with additional care)
hospital admissions

The main data source to measure impact on discharge outcomes was the acute daily discharge situation report, which provided data on all 3 outcome metrics. This data is reported daily by NHS trusts and then aggregated at 3 footprints: England, ICBs and NHS trusts. Discharge data in its current format began in 2020, although the data is less reliable in earlier years due to being a new data collection, and trusts improving their understanding of the guidance and definitions over time.

Discharge funding at a local area was measured using the fortnightly reporting returns conducted by ICBs and local authorities for the £500 million discharge fund, as well as a separate data return from NHS England for the £200 million discharge fund. The spend data for £500 million closely matches allocations, with almost 90% of areas spending more than 95% of allocated funding. Other data from the same monitoring returns, such as number of care packages purchased, was considered but not used for the impact analysis because of concerns over data quality and comparability between areas.

For impact analysis, discharges were only considered as either unsupported (pathway 0) or supported (pathways 1, 2 and 3).

As part of the analysis, the relationship between the funding and hospitalisations was explored, though there was not an expectation that there would be an impact or direct affect. Hospital admissions were considered as a sum of elective and emergency hospitalisations.

More detail can be found in the ‘Further details on impact analysis data’ section of the annexes.

Methodology used for impact analysis

Due to evaluation challenges related to the policy design of discharge funding, this impact analysis estimates effects with multiple methods for causal inference.

The 3 approaches to estimation are:

interrupted time series (or autoregressive integrated moving average (ARIMA)), which uses time series data to test whether there is a change in the trend of outcomes following the introduction of an intervention
synthetic control, which uses historical data to construct a ‘synthetic clone’ of an area receiving a particular intervention. Divergence between the treatment and its synthetic clone provide the impact estimate
difference-in-differences (two-way fixed effects), which studies the outcome of interest before and after the intervention, and makes comparisons between multiple areas

More detail can be found in the ‘Further details on impact analysis data’ section of the annexes.

Question 5

4.  Results

Accepted Answer

The key research questions we sought to address using the methodology outlined in the section above are:

How did local areas spend the £500 million discharge fund?
Were local areas able to spend the £500 million discharge fund as planned?
How did trusts spend the £200 million discharge fund?
What was the impact of the funding on hospital discharge outcomes?
Did the funding have a causal effect on the outcomes of interest at a local level?

How local areas spent the £500 million discharge fund

In summary:

the largest area of spend across local areas was on workforce, followed by securing home or domiciliary care and bed-based intermediate care
spend on workforce aimed to increase capacity and increase staff retention through incentive payments, recruitment initiatives and training
other spending included patient support and preventive interventions to help patients leaving hospital and prevent future admissions, and technology

Nationally, local authorities and ICBs reported spending a total of £486 million (97%) of the 2022 to 2023 £500 million winter discharge fund in the end of year reports submitted to end of May 2023.

The breakdown of the spending is provided in Figure 2 below.

Figure 2: total spend from discharge fund by service type

Workforce was the largest area of spend (£107 million) across local areas, followed by home or domiciliary care (£97 million) and intermediate bed-based care (£95 million). Almost half (£48 million) of the spend on workforce focused on workforce retention through incentive payments and retention payments.

There was significant variation in the use of funding at a local area level. Most areas did not spend the funding on all categories with some focusing more on one service type than others (for example, residential placement being over half the spend for one area vs another area focusing on bed-based intermediate care services).

All health and wellbeing boards were asked to report spend on the total allocation (combining their ICB allocation and local authority allocation). Some areas did not report the breakdown of their spend into what proportion was from different allocations, which meant it was difficult to interpret the reason for unspent funding.

Some areas also reported spending over their planned local authority and ICB allocation. Some areas said this was due to funding being deemed insufficient. It may also have been due to changes in the planned spend from what was initially reported.

Figure 3: cumulative spend of discharge fund for all local areas

Figure 3^{[footnote 1]} above shows that the spend from the fund was spread over the duration of 2022 to 2023, increasing with each return period.

Local areas were asked to describe their progress spending the fund through the fortnightly monitoring returns. Thematic analysis of these responses provides additional insights into how local areas spent the fund.

Workforce spending helped with providing additional hours for existing staff, reallocating resource, carrying out recruitment initiatives and encouraging retention through bonuses, implementing the national living wage and childcare vouchers, providing staff training, and providing necessary equipment.

The development of new roles such as ‘brokerage assistants’ and expansion of social work provision also supported the co-ordination of discharge. Some local areas also increased workforce capacity through collaborative working – for example, joining up workforce partnerships, working across systems in multidisciplinary hubs and teams or increased use of charity and voluntary workers.

A local authority-ICB return from Central England reported:

There is ongoing progress towards securing additional workforce to support facilitating timely hospital discharges. A social work post has been recruited to support with D2A [discharge to assess].

A local authority-ICB return from the North East reported:

Workforce recruitment and retention funding is supporting direct recruitment into social care (50% of funding is a direct recruitment service and 50% of funding is to support social care career promotion and recruitment).

Local areas reported that utilising the fund allowed them to secure additional care capacity – for example, by purchasing additional care beds (residential and/or intermediate) and care hours, block-booking beds to ensure spaces are available for discharge, and procuring specialist beds and accommodation for those with more complex needs – such as for those with dementia, those who are homeless, those with COVID-19 and/or those with learning difficulties.

Local areas also reported using the funding to secure additional care capacity through community services and reablement schemes. In some cases, local areas also utilised spend to uplift funding to care providers.

Some areas took forward initiatives aimed at improving patient outcomes, providing support to patients and increasing care quality. This included mental health support, support for carers (unpaid and paid), transport, short-stay wards, additional domiciliary care or visits, grants for patients and/or carers, and working with patient support charities and volunteers.

Some also provided additional discharge support into community settings, hotels, hospices or end of life care.

In addition, some areas invested in patient-related preventive interventions to reduce re-admissions following discharge. This often involved working with charities, education for patients to avoid illnesses, fall prevention interventions, housing advice for the homeless, mental health and emotional support for patients, and the use of personal health grants.

Some local areas deployed technology to support flow through the system, staff capacity and reduce discharge delays – for example through virtual homecare monitoring, virtual wards, online recruitment portals, assistive technology systems and e-bikes for home care staff.

A local authority-ICB return from the North West reported:

An ‘assistive technology first’ approach to hospital discharge ensures that more complex support arrangements are reserved for those who truly need it.

A local authority-ICB return from the East of England reported:

“Digital home care system has been procured; 48 robots have been set up. Training for staff on the use of the portal and other associated systems has taken place.”

Finally, some local areas said they used the fund to support business-as-usual activity and did not necessarily take forward new initiatives. The fund instead enabled them to support service recovery considering increased expenses in previous years.

Spending of the £500 million discharge fund according to spending plans

In summary:

most areas were able to spend all of their allocated funding, but with some areas spending less than planned on workforce
many areas identified barriers that made it challenging to spend the fund on reducing delays. This included the short-term nature of the fund limiting spend on recruitment, delays in mobilising schemes and external pressures, such as national workforce shortages and COVID-19

The majority of local areas spent most of the £500 million discharge fund.

Figure 4: distribution of spend of combined allocation for each local area

Figure 4 above shows the breakdown of total spend as a percentage of total allocation for each local area. It shows the original total spend from plans submitted in December 2022 with a black line and compares it with the total from final end of year reports for each service type.

Fourteen areas spent less than 90% of what they had included in their spending plans and 7 spent less than 80%. Five areas reported a final spend of at least 2% higher than their planned spend, with some stating that the funding was insufficient and they therefore felt they needed to add their own contribution to the funding.

Figure 4 also shows that workforce spend was 15% lower than planned when aggregating at a national level.

The end of year reports highlighted that local areas encountered recruitment challenges, with actual spend on local recruitment initiatives being over 20% lower than planned, as can be seen in Table 1 below.

Table 1: breakdown of the £107 million spend on workforce from the £500 million discharge fund

Category	Percentage of planned spend that was spent
Workforce retention	88%
Additional or redeployed capacity	83%
Local recruitment	79%
Overtime for existing workforce	91%

This underspend has gone into other areas such as residential placements, bed-based intermediate care services and ‘other’ where spend is higher than in the plans.

In the fortnightly monitoring reports, local areas were also asked about their progress spending the fund. Thematic analysis found that, throughout the monitoring period, most areas expressed medium to high levels of confidence in their ability to spend the money.

Many local areas recorded and monitored the use of the spend from the fund and impact using internal systems. This included the use of current and future spending forecasts, updates on spend monitoring, and investment in financial recording systems.

Having systems to monitor progress and spending allowed some local areas to evaluate their most successful schemes, anticipate underspend and reallocate funding where required.

Some local areas said they were able to use the fund to support pilot schemes and monitor impact. Some areas had alternative initiatives that they could implement if other schemes did not show impact.

As part of the progress questions, local areas were asked about any barriers or challenges to spend. In some cases, these barriers impacted what activities local areas were able to pursue.

One of the biggest challenges identified was the impact of tight timings.

Local areas reported that the timescales of the fund were too short for some schemes to be implemented fully and for them to have a meaningful impact on discharge delays. This was, firstly, because some interventions (for example, recruitment) have long lead-in times and, secondly, because uncertainty over future funding meant necessary longer-term commitments could not be made. As a result, some local areas also said that they were aiming for ‘quick wins’ by allocating funding to initiatives they perhaps already had or were quicker to implement. They also said that the timing of the fund was not well aligned with winter planning and would have been more beneficial in the summer.

Some areas said that the funding amount was insufficient, citing increasing costs in securing capacity within the market and additional costs being faced to provide interim patient support on discharge. There was some confusion around grant conditions, and disagreement around the purpose of the funding and policy. A small group of local areas said that the scope of the funding was too narrow, but did not elaborate on why or how else the funding should be used.

Most local areas expressed difficulty filling staffing positions due to competition across the sector. Finding specialist staff able to manage patients with complex needs was particularly challenging. Funding timelines meant that, since areas were unable to offer full-time positions, filling short-term vacancies was difficult. Timelines also limited the process for securing international recruitment licences.

A local authority-ICB return from London reported:

Recruitment in general is extremely competitive at present… the key barrier to spending grants will be securing the additional workforce for a time limited period and/or balancing any deficits.

A local authority-ICB return from the South West reported:

Recruitment of workforce for both short-term contracts and full-time roles continues to be difficult, especially social workers and specialist nursing.

Local areas reported that, due to high workloads, they had limited capacity for staff to take on additional work.

Frontline staff and those responsible for mobilising implementation on the ground were particularly stretched, limiting the ability to introduce new activities. Some local areas also said that, due to challenging work and living circumstances, staff were often prioritising time off and rest over taking on overtime or additional hours. Many local areas also reported contending with staff absence due to annual leave, sickness and strikes, as well as poor staff retention, which limits capacity. Local areas also noted that the resource required to onboard new staff members and upskill the existing workforce to treat increasingly complex patients makes it challenging to increase workforce capacity. Some areas reported national supply shortages of equipment and technology.

Alongside workforce capacity, local areas reported difficulty in securing bed capacity due to insufficient availability in the market as well as competition for beds, particularly in smaller areas. Some local areas experienced unwillingness from providers to accept short-term admissions. They also noted long waits for complex cases and specialist care, most notably for young people, mental health and substance abuse cases.

Some areas reflected on how increasing demand on patient services and admissions, and an increase in complex cases, paired with limited capacity and existing delays, meant that flow through services was slower and workload increased. Local areas felt that this could limit or delay the impact of funding.

A local authority-ICB return from the East of England reported:

The acute settings are continuing to see an increase in complexity and acuity of patients, which is of more of an issue than the numbers.

A local authority-ICB return from the South West reported:

We are also seeing increased demand at the hospital front door, which in turn is impacting on flow – while this continues and resource remains primarily focused on discharge, there will also remain a risk that significant improvements in flow will not be achieved.

How local areas spent the £200 million discharge fund

In summary:

almost all regions spent over half of their maximum funding allocation, with North East and Yorkshire spending 24%
almost 10,500 people moved into step down beds, with 21% people moving into step down beds in the Midlands and 18% in London
London and the South East are the regions that purchased the most packages of care

Of the total £200 million discharge fund, 69% was spent by local areas.

Figure 5: proportion spent of maximum allocation

Figure 5 above shows the region that spent the greatest proportion of their allocation was the North West, spending 97%, and the region that spent the least was the North East and Yorkshire at 24%.

Table 2: national breakdown of £200 million discharge fund spending

Total number of step down beds purchased	Total number of packages of care purchased	Total number of people moving from an acute hospital bed into a step down bed	Total number of people moving from a mental health bed into a step down bed	Total number of people moving from a community bed into a step down bed
7,134	19,441	8,504	1,273	698

Table 2 above shows that, nationally, 7,134 additional step down beds and 19,441 packages of care for those who occupied beds were purchased using the £200 million discharge fund.

Almost 10,500 people were enabled to move into a step down bed.

The North West, and North East and Yorkshire were the regions that purchased the most step down beds using the £200 million discharge fund, and the South East and London purchased the most packages of care, as shown in Table 3 and Figures 6 to 8 below.

Table 3: regional breakdown of £200 million discharge fund spending as proportions of national totals

Region	Total number of step down beds purchased	Total number of packages of care purchased	Total number of people moving from an acute hospital bed into a step down bed	Total number of people moving from a mental health bed into a step down bed	Total number of people moving from a community bed into a step down bed
East of England	8%	2%	12%	3%	4%
London	10%	22%	8%	61%	60%
Midlands	17%	10%	25%	1%	6%
North East and Yorkshire	23%	5%	21%	2%	15%
North West	26%	18%	14%	29%	8%
South East	11%	24%	13%	4%	6%
South West	4%	19%	6%	0%	1%

Figure 6: percentage of step down beds purchased as proportions of national total

Figure 6 above shows the amount of step down beds purchased by each region as a proportion of the total step down beds purchased by all areas using the £200 million discharge funding.

Figure 7: percentage of packages of care purchased as proportions of national total

Figure 7 shows the amount of packages of care purchased by each region as a proportion of the total step down beds purchased by all areas using the £200 million discharge funding.

Table 4: number of people moved into a step down bed by region

Region	Total people moving into a step down bed
Midlands	2,178
London	1,907
North East and Yorkshire	1,889
North West	1,645
South East	1,203
East of England	1,090
South West	562

Figure 8: number of people moved into a step down bed by region

Table 4 and Figure 8 above show that the Midlands was the region that enabled the most people to move into a step down bed using the £200 million discharge fund, with around 2,200 people enabled to move.

The impact of the funding on discharge outcomes

To assess the impact of the discharge funding, 2 approaches were taken.

Firstly, this report presents a descriptive analysis of key outcome metrics and their trends after implementation.

Subsequently, we used methods for impact analysis to estimate a counterfactual scenario and help attribute impact on the observed outcomes to discharge funding specifically.

Descriptive statistics

This section presents the data available from October 2021 until June 2023, where available, to provide useful historical context.

October 2022 is defined as our baseline period and a month average is used to mitigate the volatility with the metrics.

Data has been provided beyond the end of the funding period (March 2023) to enable us to see whether trends continued.

Figure 9 below illustrates the key metric for understanding the number of patients with NCTR who have not been discharged by the end of the day (also referred to as the stock of patients with NCTR).

Figure 9: number of patients with NCTR and who have not been discharged

While there are some limitations to this metric, the findings in Figure 9 above are encouraging. The figures show that 7-day rolling average NCTR was increasing from October 2021 to October 2022. NCTR then stabilised, before a dip between Christmas and new year, before increasing to a peak on 11 January 2023. It has since been on a downward trend through to the end of June 2023.

Figure 10: number of patients with NCTR and who have not been discharged (years overlaid)

Figure 10 above shows that, although the number of patients with NCTR was higher in December 2022 than in December 2021, we saw much less growth in NCTR figures than in winter 2021 to 2022.

NCTR figures increased from 10,377 between 1 December 2021 and the peak of 12,843 in winter 2021 to 2022 (a 24% increase). In comparison, NCTR only increased from 13,545 between 1 December 2022 and the peak of 14,385 in winter 2022 to 2023 (a 6% increase). This suggests that funding this winter may have been successful at preventing further growth in NCTR.

Since the peak observed in January 2023 of 14,385 patients with NCTR (11 January 2023, 7-day rolling average), NCTR patients reduced to 13,452 at the end of March 2-2023 (31 March 2023, 7-day rolling average) and this downward trend continued approaching the summer to 12,200 (30 June 2023, 7-day rolling average).

Table 5: change in NCTR across 2021 to 2022 and 2022 to 2023

Time period	Date of winter peak	Increase in absolute NCTR, October average to winter peak	Increase in absolute NCTR, 1 December to winter peak
2021 to 2022	24 January	3,156	2,466
2022 to 2023	11 January	755	840

The data for a more detailed breakdown of discharges from NCTR is limited to patients who have a length of stay of at least 7 days. Below we split this into groups:

supported discharges (discharges into pathways 1, 2 and 3)
unsupported discharges (discharges into pathway 0, known also as simple discharges)

Figure 11: number of discharges from NCTR – unsupported (pathway 0) and supported (pathways 1, 2 or 3)

Figure 11 above shows the number of supported and unsupported discharges over time. The number of supported discharges (pathways 1 to 3) in winter 2022 to 2023 (January to March 2023) was 17% higher compared to October 2022 and 15% higher compared to winter 2021 to 2022 (January to March 2022). The number of unsupported (pathway 0) discharges increased by 5% and 6% respectively with regards to the same periods.

The increase in supported discharges suggests at least one of the following:

increases in admissions (but the data below suggests this is not large enough to account fully for the increase in discharges)
improvements in discharges (outflow is greater than inflow)
higher medical needs or acuity of patients (they require more complex discharges)
improvements in recording of data relating to NCTR

The reduction in NCTR and increase in supported discharges is in the context of increased emergency admissions and extremely high bed occupancy over the winter period (over 94%), as well as considerable winter pressures in December 2022 associated with COVID-19 and flu. Since January 2023, wider urgent and emergency care metrics (such as A&E 4-hour performance) have also stabilised or improved.

Figure 12: average daily emergency admissions per month into acute hospitals

Source: NHS England, A&E Attendances and Emergency Admissions 2023-24.

Figure 12 above shows that emergency admissions increased slightly by 4% (16,819) in January to March 2023 compared to January to March 2022 (16,213). There was a 3% increase in January to March 2023 compared to October 2022 (16,350). In June 2023, emergency admissions were 17,441, which is slightly lower than the levels seen pre-COVID-19 (17,627 in June 2019).

This covers patients of all ages and more trusts (whereas the acute daily discharge situation report is for those aged 18 and above), but it does not cover elective patients.

Figure 13: adult G&A beds occupancy rate

Source: NHS England, Critical care and General & Acute Beds – Urgent and Emergency Care Daily Situation Reports.

Figure 13 above shows that the number of general and acute (G&A) beds available in all acute trusts increased over the winter of 2022 and was slightly higher in April 2023 compared to a year previously (around 92,711 compared to 90,991).

However, adult G&A bed occupancy remained high during the winter and peaked at 96% before reducing to around 94% in April 2023 and, as of August 2023, remains at that level.

Missing data

The acute daily discharge situation report used for the analysis collected data from 121 different acute trusts for the period between October 2022 and March 2023. As of June 2023, there are 119 acute trusts and the analysis is based on the list of trusts at this point with changes in trusts, due to mergers, from the original period reflected by aggregating data.

There are known issues in reporting and the quality of data from some trusts given the frequency of the data collection and the number of organisations involved. Below, we review the quality of the data submitted and explore the impact this could have on the analysis produced.

The data obtained at trust level could have different types of issues, which include:

organisations not submitting for a period
organisations incorrectly submitting for a period

The former is straightforward to observe and assess. However, when an area submits data that is incorrect, they can only update the submission the week before the acute discharge monthly publication is made available. We are also aware that some trusts have reviewed their approach to measuring NCTR to be more in line with the guidance over time, and this can impact comparisons.

This analysis does not aim to identify instances of incorrect submissions as this would require a full audit. Instead, we aim to estimate the impact of non-submission on national-level figures.

The number of patients remaining in hospital who no longer meet the criteria to reside is the difference between 2 metrics:

the number of patients who no longer meet the criteria to reside at the start of the day
the number of those patients who are discharged during the day

Analysis of those 2 metrics shows that 107 out of 119 trusts in scope have submitted data for each day between October 2022 and March 2023 (90%). The remaining trusts have missed submissions to varying levels in that period.

Figure 14: number of trusts with missing NCTR data or reporting zero for key metrics to calculate stock

Figure 14 above shows the number of trusts who had missing information or reported zeros for those metrics. The quality generally improved over the period observed. Notably, there is data for 2 to 3 trusts missing in October 2022, compared to 0 to 2 trusts for the rest of period, which suggests that October could be slightly under-represented.

Figure 15: NCTR stock comparison between published figures and rate with imputations for missing data

Figure 15 above compares a 7-day rolling average of actual NCTR at England level against an approximation of what NCTR would be without the missing submissions.

The approach taken is to consider the metrics at the trust level and, for those with missing or zero data, to use an imputed value, which is the last non-zero value that exists for that trust prior to the missing information. This is aggregated together to provide an imputed NCTR.

The analysis suggests that, after the corrections, the gap for NCTR in October 2022 is larger than for other months and is relatively small for the rest of the period observed.

The other metric of interest is discharges from NCTR.

The completion of this metric is the same as seen in Figure 14 for NCTR stock because the 2 metrics are related, therefore up to 3 trusts may not have submitted for a given day in the time period considered. However, there are more likely to be zero values submitted for this field. These could be accurate zero values (smaller trusts may not have pathway 1, 2 or 3 discharges from NCTR on a day-by-day basis) or inaccurate reporting from areas of discharges.

Figure 16: number of trusts reporting zero discharges or missing values from NCTR by pathway each day

Figure 16 above shows the number of trusts who have reported zeros or have missing data for delayed discharges by pathway.

The figures for pathway 0 are similar to the total figure. However, many trusts have zeros reported for pathways 1, 2 and 3. The fluctuation seen also presents a weekend effect of reporting zeros. The number of discharges varies across the week, with many hospitals managing weekend capacity by discharging more patients on a Friday. According to NHS England Discharge planning (PDF, 61KB), discharges then are lower until Monday morning (or afternoon in some cases), which means Saturday, Sunday and Monday have fewer discharges.

While the number of missing or zero reports is quite high for pathways 1, 2 and 3, the pattern is consistent with total discharges from NCTR (when discounting for Christmas and the new year). We would expect any impact of any missing data to be similar across the baseline and impact periods, and therefore have minimal impact on the findings.

Similarly to the imputed calculations of stock NCTR to account for missing trusts, we have also considered the impact of this on NCTR discharges. It should be noted that data on patients discharged to a hospice is omitted at trust level due to disclosure risks, which can have a small impact on imputed values and difference with national totals.

Figure 17: NCTR discharges comparison between published figures and rate with imputations for missing data

Figure 17 above presents the difference between the actual discharges from NCTR and the imputed values. The differences were largest in the month of October 2022 (average of 166 extra per day), indicating that discharges from NCTR were higher than the data suggested and, therefore, the increase in NCTR discharges from October 2022 to the January to March 2023 period was potentially smaller than estimated.

Table 6: actual vs imputed numbers of NCTR discharges, accounting for missing data, between October 2022 and January to March 2023

Type of discharge	Time period	Actual number	Imputed number	Difference
NCTR	October 2022	13,630	13,927	297
NCTR discharges	October 2022	9,044	9,201	157
NCTR	January to March 2023	13,788	13,835	47
NCTR discharges	January to March 2023	9,669	9,695	26

Table 7: percentage change in NCTR discharges from NCTR between October 2022 and January 2023 to March 2023

Type of discharge	Actual percentage change	Imputed percentage change
NCTR	1%	−1%
NCTR discharges	7%	5%

The methodology used indicates that the difference between reported and actual NCTR patients could be up to around 300 on a given day. This is higher in October 2022 than in January to March 2023 where we have much less missing data. Similarly, it also shows the difference between reported and actual number of NCTR discharges could be around 160 on a given day.

While the number of missing or zeros reported are quite high for NCTR discharges into pathways 1, 2 and 3, it is relatively consistent over the period considered. We have slightly more trusts missing in October 2022 than in January to March 2023. We would expect any impact of any missing data to be similar across the baseline and impact periods, and therefore missing data is likely to have a minimal impact on the findings.

Results of impact analysis

In summary:

there was a significant break in the trend for NCTR values at England level coinciding with implementation of discharge funding. This suggests that the funding was associated with reduced NCTR levels
it was not possible to consistently link greater improvement in NCTR to areas with higher spend. Challenges to evaluation as well as data quality might have affected these findings
for supported discharges, a growing trend continued, but there were no significant changes to the trend after the implementation of the fund. There was no observed effect on hospital admissions

The main challenges to the impact analysis were connected to the following considerations:

funding was implemented in all areas in England at the same time
the areas with more significant problems with delayed discharges were targeted with more funding. This means it is possible to confuse observed improvements in those areas for actual effect of the funding
there were other policies and shocks affecting health and social care (for example, the Market Sustainability and Improvement Fund, Better Care Fund and general local authority funding)
there were quite significant data quality issues, particularly with metrics at localised footprints before 2022 – for example, missing values and outliers in the data
there was a significant risk of spill-over effects between local areas

Due to these challenges, the approach for the impact analysis relied on implementing multiple estimators with varying assumptions.

There is evidence that the discharge funding is associated with a decrease in NCTR numbers. Discharges continued a growing trend, but there was no break in trend associated with discharge funding, nor was it possible to estimate a difference compared to a modelled counterfactual.

The number of admissions was stable and unrelated to levels of discharge funding spend.

Overall, the results highlight the challenges to robust evaluation of discharge interventions.

Figure 18: ARIMA estimation for rates of NCTR, England footprint

Interrupted time series estimation for NCTR using ARIMA (a statistical method for forecasting based on past time series values and trends), as shown in Figure 18 above, suggests a statistically significant break in the previously rising trend for NCTR. Assuming the previous trend from 2020 to 2022 had continued, NCTR numbers would be higher by around 3,000 than observed in June 2023.

This difference could be explained by the implementation of discharge funding. Alternatively, it might have been caused by other changes that coincided with the funds.

It was not possible to conclusively identify differences in impact between areas with different levels of spending. Synthetic control estimations (ICB footprint) provided inconsistent results depending on the specification and whether the £200 million discharge fund was included in the analysis. Difference-in-differences effects estimations had no significant coefficients for funding levels at ICB and NHS trust footprints.

The relatively low statistical power of the study with 42 and 115 areas for respective footprints may have contributed to the findings. Similarly, they could be affected by data quality problems (missing data, outlier values and breaks in trends).

Overall, this does not suggest that the funding had no overall impact, but may suggest that small differences in spend between ICBs were not sufficient for detecting significant differences in impact between localities.

Figure 19: ARIMA estimation for numbers of supported discharges, England footprint

As Figure 19 above shows, supported discharge (pathways 1, 2 and 3) continued an upward trend after the implementation of discharge funding.

However, there was no statistically significant break in this trend that would coincide with the implementation of funding.

Synthetic control estimations (ICB footprint) suggest that a typical high-spend area might have experienced a relative increase in supported discharge compared to the lower-spend counterfactual. However, this estimation was not statistically significant, possibly in part due to problems with data quality for discharge.

Figure 20: ARIMA estimation for numbers of unsupported discharges, England footprint

Figure 20 above shows that unsupported discharge (pathways 0) also continued increasing following the implementation of discharge funding. However, there was no statistically significant break in this trend that coincided with the implementation of funding.

Synthetic control estimations (ICB footprint) suggest that a typical high-spend area might have experienced around the same levels of unsupported discharge as compared to the lower-spend counterfactual. This estimation was not statistically significant, possibly in part due to problems with data quality for discharge.

Figure 21: ARIMA estimation for hospital admissions, England footprint

Interrupted time series estimation found hospitals admissions were generally stable at England footprint, as shown in Figure 21 above.

No effects of discharge funding on admissions were observed in both synthetic control and fixed effects estimations, although this is to be expected as the funding was not provided for admissions avoidance.

Combined, these findings provide evidence that there has been a positive impact on NCTR that is associated with discharge funding, despite the challenges to this evaluation. However, the findings are not sufficiently robust for precise estimates of total effect or comprehensive value-for-money evaluation.

The details on specifications and estimates are provided in the ‘Further details on impact analysis data’ and ‘Impact analysis methodology’ sections of the annexes below.

Question 6

5.  Conclusions and recommendations

Accepted Answer

The findings from the rapid evaluation of the 2022 to 2023 discharge funds suggest that in most cases, funding was used to increase the number of discharges and to support the reduction of discharge delays.

There is evidence that the funding was associated with reduced NCTR levels. However, local areas reported that, due to barriers related to spend, the impact of the funding could be limited or delayed.

The ability to estimate any impact is limited by factors related to data quality and policy characteristics.

The fortnightly returns and end of year reports show that service provision in areas such as domiciliary care, residential care, reablement and bed-based intermediate care were key areas of spend identified by local areas for the £500 million discharge fund. Recruitment and retention were also a key priority for local areas with workforce being the largest spend.

Some areas observed improvements in flow through the system, and a reduction in discharge delays and wait times. Many areas were able to spend the funding to secure additional care capacity, and invest in schemes to provide patient support and interventions to prevent re-admissions. Some areas reflected on new partnerships with community services, charities and providers to increase care provision and provide intermediate support.

The fortnightly returns have shown that most areas were able to spend their £500 million discharge fund allocation, although this may not have been in the same services as initially planned. Workforce was particularly an area where local areas were not able to spend as planned, with some highlighting recruitment challenges over the period, and reflecting on the short-term nature of the funding and conditions also having an impact on this.

Additional challenges to spending included systemic or structural issues that made it difficult to reduce discharge delays – examples included increasing demand and complexity of need. Additional funding alone, particularly short-term funding, was not sufficient for local areas to completely overcome these issues. The lack of certainty around future funding also meant that some local areas had to fund existing schemes instead of developing new ones. Some local areas also felt that the scope for the funding was too narrow or unclear.

Many local areas also reported that the monitoring required by DHSC was burdensome and did not align with their own data systems. Some areas said they were unable to attribute impact to specific funding pots.

Impact analysis shows that there was a significant break in the rising NCTR trend at national level, with NCTR decreases coinciding with the implementation of the fund. This suggests that the funding could have contributed to the reduction in NCTR. However, there are other possible reasons for the break, such as change in admission and/or discharge criteria occurring at the same time. NCTR discharges continued to grow with no breaks in trend coinciding with the implementation of funding while hospitalisation remained stable.

Looking at the £500 million discharge fund in isolation, the results from the impact analysis indicate that higher discharge funding could contribute to lower NCTR rates. When including the additional £200 million discharge fund, the analysis found the opposite effect. These inconsistent results do not mean the policy had no overall impact, but may suggest that small differences in spend between ICBs were not sufficient for detecting significant differences in impact between localities.

Key considerations for future funding, based on the findings from this research, include the following:

future funding should be distributed over a longer period, with as much advance notice as possible to allow local areas to implement interventions with longer lead-in times
future funding should provide as much certainty as possible regarding continuity and exit planning. Additional funding should be seen as one part of the government’s response, alongside policy development to address longer-term or systemic issues
any grant conditions for future funding should be as clear as possible. As the thematic analysis does not provide much information on which aspects of the conditions were unclear, this should be explored further via other primary research and engagement with local areas
monitoring data requested from local areas should be meaningful, necessary and proportionate. DHSC should consider how data from the 2022 to 2023 fund was used to identify what is required in future. This should include consideration of the frequency of data collection, and how we can ensure consistency and quality across local areas. The clarity of qualitative questions should also be reviewed, particularly in relation to innovation and impact. DHSC should engage with local areas to understand how best to collect the necessary data while minimising burden, and test plans to ensure that data collected will be meaningful
evaluating impact of discharge interventions is challenging. These challenges can be reduced by improved data quality, longer time series and considering evaluation in the policy design phase
data collection should also be considered in the context of other funding, including the Better Care Fund where spend on discharge may overlap, to help understand how different allocations are spent

Question 7

Annexes with detailed methodology

Accepted Answer

Data sources

Data on acute hospital discharge activity is collected through a data collection called the Acute daily discharge situation report. This is a daily collection of several metrics including:

number of patients in hospital with NCTR
number of patients discharged each day by pathways and type of destination
primary reason attributed for delay
total bed days lost since NCTR status declared for patients currently in hospital

Data is collected from local areas as part of fortnightly reporting and end of year reporting on spend from discharge funding in different service types, including:

Details on acute daily discharge situation report data

The acute daily discharge situation report data collection is available online from April 2021 and replaces the previous data collection on delayed transfers of care (DTOC). DTOC collected information on the number of patients experiencing delays and reasons for delay from August 2010 to February 2020. DTOC data is not comparable with acute discharge delays data.

Acute daily discharge situation report data is available at a national, regional, ICB and trust level for the following metrics:

number of NCTR patients at beginning of day (data available daily)
number of NCTR patients discharged (data available daily)
number of NCTR patients at end of day (11:59pm) (data available daily)
count of NCTR patients broken down by attributed primary reason for delay and for stays that are 14 or more days in length (data available is the weekly snapshot average for the month)
number of discharges of patients with NCTR broken down by their intended discharge destination (by pathway) (data available as monthly total)

Thematic analysis methodology

Responses to the open-text fortnightly monitoring questions were analysed using thematic analysis. Thematic analysis is a widely used technique to systematically identify key themes and subthemes in the data.

The data contained responses from 7 waves of fortnightly returns. Out of the 4,456 written responses, 1,587 were removed during data cleaning. This was due to a high number of duplicates and invalid responses, such as ‘N/A’, or misplaced data from the quantitative questions.

After the second wave of data collection, an option was added into the template for local areas to select ‘no change’ instead of providing a new qualitative response. For the purposes of this analysis, the ‘no change’ responses were also removed. After data cleaning, 2,869 responses remained.

During the data familiarisation stage (see Table 8 below), repetition was observed in the weekly responses. This is likely due to the frequency of monitoring submissions or an overlap in question topics. It was found that the responses with the most detail occurred in the first 2 waves and, after this, non-response and repetition increased.

Given the high volume of data, as well as the declining detail and responses after waves 1 and 2, the possibility of reaching saturation was assessed. Saturation in qualitative research occurs when, through the process of analysing data, the same themes emerge. Beyond data saturation, no additional insights are found.

Assessing when saturation will occur can be estimated based on experiences from related studies,^{[footnote 2]} using statistical analysis of expected or desired theme prevalence,^{[footnote 3]} or by:

‘stretching data’ – taking a purposive approach to assessing different groups within your data set to ensure that all findings have been captured^{[footnote 4]}
using an iterative approach – taking a random sample for coding and then testing the code frame by re-applying it to a selection of the un-coded responses^{[footnote 4]}

To ensure validity of findings, avoid analysing beyond saturation and maximise efficiency, the 2 above approaches were combined. Additional purposive coding of outlier local areas conducted as part of quality assurance checked for uncovered themes missing from the random sample.

More information on the sampling and each stage of the analysis can be found in Table 8 below.

Table 8: stages of thematic analysis

Activity	Steps
1. Familiarisation and producing a code frame	Reviewed an initial sample of responses to identify key themes to be coded during the analysis. Cleaned the data to ensure any blanks or invalid responses have been removed.
2. Iterative sampling	Initial random sample of 1,042 responses drawn, containing half of responses (50%) from waves 1 and 2, and one third (33%) from waves 3, 4, 5, 6 and 7. Oversampling in waves 1 and 2 ensured the most detailed responses were captured.
3. Coding	Using the code frame, all sampled responses reviewed to record which codes applied and identify any new themes or subthemes.
4. Consolidation	Consolidation of findings to interpret data, organise results and identify areas in which further analysis was required.
5. Quality assurance	Development and interpretation of the data checked independently against quality assurance log. Independent re-coding of previously coded work to check for consistency, subjectivity and errors in interpretation. Additional round of purposive coding on a selection of 33 outlier local authority-ICBs (83 written responses) with varying characteristics (geography, discharge delays and funding levels) to check for any missing themes. Assessment by the coding team and quality assessor about whether saturation had been reached.
6. Testing	The final lists of themes and subthemes were tested independently, and re-applied to the data to check accuracy.

Local issues, change throughout reporting and the final spending reports have not been included in this phase of analysis.

The questions were not designed or tested by social researchers, and there was a wide interpretation of questions and variation in the quality of responses.

Numerical prevalence of findings cannot be drawn from this work due to the chosen question format and analytical technique.

Further details on impact analysis data

For analysis at the level of local areas, 2 variables are used for scaling to ensure comparability of local areas. Without scaling, trusts with higher NCTR may be deemed to have a bigger delayed discharge problem when, in fact, it is because the trust is larger in size.

For outcome metrics, a metric for available beds (acute daily discharge situation report) was considered as the main scaling variable to account for hospital size.

For variables related to discharge policy, metrics are scaled by the number of disabled people by age group from the 2021 Census (aged 65 years and above in this analysis).

Outcome metrics were processed at the NHS trust footprint. The NHS trust data was aggregated for the ICB footprint and, similarly, aggregated for England. Control variables – typically reported at the level of local authorities – were aggregated at ICB level.

Outcome metrics reported by several NHS trusts had problems of missing data, implausible patterns and extreme values. Four hospital trusts were excluded from the analysis due to concerns over data quality and the scale of missing data.

For all remaining trusts, data problems were mitigated by the following adjustments and transformations:

missing values were linearly interpolated based on the nearest reported values (7-day rolling average)
if the first or last value in the time series was missing (only one hospital trust affected) and prevented interpolation, the median of the nearest 50 values (providing representative pool) was imputed instead as the procedure for reducing the influence of outliers)
to deal with random fluctuations and fluctuations driven by administrative processes, the 28-day rolling median was used. This approach was used to account for outliers in the data, and capture weekly and monthly cycles in hospitals
to improve data quality, the analysis was restricted to data from March 2022 to June 2023, which had fewer data quality issues

The data collected for this study was informed by the previous literature on the integration of health and social care. In particular, it relied upon a previous Better Care Fund evaluation from 2018, A system-level evaluation of the Better Care Fund: final report. This includes variables such as local deprivation, urban density and existing social care capacity. This study attempts to control for some of these factors indirectly using area-level data as part of the evaluation design.

The following sets out the specific data collected as part of this study and used as control variables in multivariate analysis:

social care beds per local area from 2021 to 2023 – based on quarterly Capacity Tracker data
population density – based on 2021 population data and calculated from the Office for National Statistics (ONS) Data combining multiple variables, England and Wales: Census 2021
strikes among healthcare workers from 2022 to 2023 as a binary variable – taken from NHS England Industrial action in the NHS
attendance allowance and income support recipients per capita from 2021 to 2023 – based on Department for Work and Pensions benefit statistics
estimates for unemployment rate in local authorities from 2022 to 2023 – based on ONS Labour market in the regions of the UK: April 2023 data
flu and respiratory disease hospitalisations in winter 2022 to 2023 – taken from NHS England Urgent and Emergency Care Daily Situation Reports 2022-23
Better Care Fund spend plans for 2022 to 2023 – taken from NHS England Better Care Fund: 2022-23 planning data to account for the influence of the Better Care Fund

The analysis includes key external factors that may have an impact on hospital discharge, but there will likely be other unobserved factors that may have an impact which have not been included in the analysis. The methodology used can account for a part of these unobserved influences, but it cannot account for them completely.

Impact analysis methodology

It is possible to estimate impact by comparing the observed outcomes of the funding to a constructed counterfactual in which there is no funding.

There are 2 broad approaches for estimating the counterfactual used in this analysis. One approach uses past values and trends as the basis for constructing the counterfactual, and compares it to the observed outcomes after discharge funding. The other approach compares outcomes in different sub-units with varying levels of spend. The first approach corresponds to the interrupted time series (ITS) estimation, while the other corresponds to synthetic control and difference-in-differences.

The ITS method was implemented using 2 estimators: segmented regression and ARIMA. It was implemented with an England footprint (n = 1), covering the period from January 2021 to June 2023.

The most common method for implementing ITS is the segmented regression. It is a linear regression with terms for step change after intervention and change in the post-intervention slope. It is possible to add terms to account for seasonal effects. Due to the use of daily data, this approach is susceptible to problems with autocorrelation, which motivated the use of ARIMA as a more sophisticated ITS methodology.

ARIMA is a modelling method enabling more sophisticated analysis of time series data, accounting for autocorrelation structures, which are a common problem with time series data. There are 3 key terms in ARIMA models – the:

autoregressive term (using previous values in time series as a regression term)
difference term (explaining differences over time)
moving average term (explaining dependent variable through residuals from the previous period)

These 3 elements are combined, although some can be omitted to improve model fit or when theoretically justified. It is also possible to account for seasonal effects and general trends.

ARIMA for ITS was implemented with an automatic function finding the optimal parameters for a given time series. Similarly to the segmented regression approach, regression terms for change in level and slope of outcomes were included in the estimation. Data frequency was reduced to weekly and bi-weekly (specifically 7 and 14-day rolling averages) to simplify the potential analysis of seasonality. Seasonal terms were rejected by the automatic function, but the analysis was also conducted with a fixed (rather than automatic) parameter for seasonality as a robustness check.

Synthetic control was implemented at the level of ICBs (n = 42). Because all units were exposed to discharge funding to some degree, this required collapsing the continuous spend variable into a binary variable, indicating high and low spending areas (cut off at the median scaled spend). All treated units were then combined into an ‘average high-spend ICB’, which was calculated as the weighted (by bed number) average of outcome and control variables. Synthetic control was then estimated with the low-spend areas as donor units and average high-spend ICB as the treated unit. Standard placebo testing was conducted to estimate statistical significance of the findings.

Synthetic control requires defining predictors, which will be used to allocate the appropriate weights for donor units. The predictors included outcome variables of interest in the pre-treatment period at 50-day intervals. The non-outcome control variables included in this modelling are:

population density
social care capacity (beds in residential care)
unemployment rate
attendance allowance recipients per capita
income support recipients per capita
number of strike days
the Better Care Fund spending

The units in two-way fixed effects analysis are NHS trust (n = 115) and ICBs (n = 42). The key independent variable is the level of discharge spending per disabled person over the age of 65. The model includes dummy variables for each of the analysed units and time periods. The regression specification also includes time-varying control variables, including data on strikes and flu hospitalisations. Errors are clustered to account for the clustered nature of trusts (contained within ICBs) and nature of the intervention (not occurring at the trust level).

Sensitivity analysis was conducted to test the robustness of the findings. Typically, it included:

shifting of the treatment cut-off (set at 1 December 2022)
shifting of the included timescales
changing model specification by including or excluding covariates

For ARIMA modelling, sensitivity testing also included fitting a seasonal model and changing reporting frequency of the data (at a weekly and 2-weekly interval).

For synthetic control, this also included changes to how ‘average treated unit’ was constructed (weighted mean reflecting the size of ICB populations), adjusting the predictor variables and changing the treatment cut-off from median spend level.

The difference between Figure 2’s total of £483 million reported at the end of March 2023 and £486 million quoted in Figure 3 is due to additional funding being recorded in the final year report that was not recorded in the final fortnightly report. ↩
Tran VT, Porcher R, Falissard B, and Ravaud P. ‘Point of data saturation was assessed using resampling methods in a survey with open-ended questions.’ Journal of Clinical Epidemiology 2016: volume 80, pages 88-96. ↩
Lowe A, Norris A, Farris A, and Babbage D. ‘Quantifying Thematic Saturation in Qualitative Data Analysis.’ Field Methods 2018: volume 30, issue 3. ↩
Vasileiou K, Barnett J, Thorpe S and others. ‘Characterising and justifying sample size sufficiency in interview-based studies: systematic analysis of qualitative health research over a 15-year period.’ BMC Medical Research Methodology 2016: volume 18, article 148. ↩ ↩²

Cookies on GOV.UK

Applies to England

Executive summary

How local areas spent the £500 million discharge fund 2022 to 2023

How local areas spent the £200 million discharge fund 2022 to 2023

The impact of the funding on key outcome metrics

Conclusions and considerations for future funding

1. Introduction

2. Background and policy context

Hospital discharge process for delayed discharges

Pathway 0

Pathway 1

Pathway 2

Pathway 3

3. Methodology

Analysis of key metrics to assess discharge delays and the wider context

Time period under review

Analysis of fortnightly returns from ICBs and local authorities, including monitoring data

Question 1

Question 2

Descriptive statistics on £200 million spend

Impact analysis of the funding

Data used for impact analysis

Methodology used for impact analysis

4. Results

How local areas spent the £500 million discharge fund

Spending of the £500 million discharge fund according to spending plans

How local areas spent the £200 million discharge fund

The impact of the funding on discharge outcomes

Descriptive statistics

Missing data

Results of impact analysis

5. Conclusions and recommendations

Annexes with detailed methodology

Data sources

Details on acute daily discharge situation report data

Thematic analysis methodology

Further details on impact analysis data

Impact analysis methodology

Is this page useful?

Help us improve GOV.UK

Help us improve GOV.UK