Chapter 7: Healthcare

Healthcare is the largest public service by spend, representing around 41.5% of total public service expenditure in 2021. It has consistently been prioritised for measurement development since the Atkinson Review in 2005.

By the time of the Bean Review (2016), methods were sufficiently well advanced for standalone National Statistics to be published, on the usual calendar year basis. This was further enhanced in 2018 with the addition of England specific financial year statistics to meet the needs of the Department of Health and Social Care (DHSC) and National Health Service (NHS) England. These statistics have always made significant use of measures produced by the University of York. Detailed information on the methodology can be found in Public service productivity estimates: healthcare quality and methodology information (QMI). 

7.1 Core methodological and data challenges identified

The coronavirus pandemic (COVID-19) introduced significant challenges for measurement of healthcare output and hence productivity, not least the emergence of new activities established to manage and mitigate its impact. Furthermore, since the start of the pandemic, notable changes have been made to the delivery of many services, such as ‘virtual wards’ where patients receive care at home that traditionally would have been provided in hospital. Such changes present substantial challenges to the measurement of productivity as little data are available to assess the relative value of services provided by traditional and new modes of delivery. From 2020, the Office for National Statistics (ONS) captured the volume of coronavirus related testing, tracing and vaccination services provided, both in the UK National Accounts and the public service productivity estimates. These represented a sizeable contribution to public service healthcare output between 2020 and 2022. 

Despite already being at a high level of quality and coverage, and improvements made during the coronavirus pandemic (COVID-19), there remain measurement challenges for healthcare output around three main issues:

1. data quality and coverage,
2. the suitability of cost weights for assigning relative values to different services and
3. the treatment of preventative services. 

7.2 Data quality and coverage

While the data used for measuring much of healthcare output draws on the most detailed data used for any public service, the availability and detail varies by sub-sector. There are currently large gaps in activity data for NHS-funded services contracted from the independent sector, public health services and substantial portions of healthcare output in the devolved governments. 

In addition, even more detailed data are available. NHS England’s Hospital Episode Statistics (HES) provide person-level data, which are used by the University of York in their healthcare output measure. These data enable output to be measured using Continuous Inpatient Spells (CIPS). This is a broader activity measure than the Finished Consultant Episodes (FCEs) used by the ONS. The different data used mean differences are expected between ONS’s output and quality adjustment estimates and those produced by the University of York. 

For FCEs, additional activity is recorded when patients move between consultants during a hospital stay. In contrast, with CIPS, one hospital stay accounts for one activity regardless of any moves between consultants, thus preventing increases in the transfers of patients between consultants from affecting output growth. While the challenges of acquiring and processing HES data would require more time than enabled by the Review, the ONS will consider the use of these data in future.  

The ONS relies on data produced by the University of York, funded by the National Institute of Health and Care Research, to produce the quality adjustment for healthcare services. If quality adjustment is to be included in the UK National Accounts to comply with the System of National Accounts 2025 (SNA25) (see Chapter 6), the ONS needs to ensure these priority statistics can be published with minimal risks beyond its control.

The suitability of cost weights 

Cost weights are standardly used in public service output measures to assign relative values to different activities. While costs can serve as a proxy for value, using cost weights within a productivity measure can result in that measure missing a key source of productivity growth – the replacement of higher-cost service designs with lower-cost service designs which deliver the same outcome but at lower cost.  

A hospital moving to treating non-complex knee replacements as day surgery, whilst its neighbour retains an in-patient delivery model for cases of the same complexity would, if cost weighted at the lower price, appear to have reduced its output, even if it delivered the same volume of knee replacements. This is obviously erroneous. Within healthcare, there are a range of services where the NHS has sought to deliver equivalent outcomes for patients by employing lower-cost forms of healthcare provision. Such service changes are seen as an important contributor to improvements in productivity.  

Treatment of preventative services 

Preventative services merit special consideration as the value of the output they generate is not in the number or cost of patients treated, but in the number or cost of patients who avoid the need for treatment in future. A naive application of cost weights to preventative services, which typically have relatively low costs and hence low weights, therefore fails to reflect the relative value being generated by such services and changes in this over time, as discussed in Chapter 3.  

7.3 Improvements to Inputs Estimates

A review of healthcare inputs, specifically the value of goods and services used in the provision of healthcare activities (intermediate consumption), was prioritised in Year One of the Review. This found that the ONS was largely using the best available existing data, and therefore only small improvements were required, such as the inclusion of legal and audit services. These were incorporated into published statistics from Spring 2024. The weights used to calculate labour inputs were also updated to better align with our measure of full-time equivalent (FTE) staff. The overall impact of these changes was minimal and there appears little scope to deliver further advances, at least in England, at this time. 

7.4 Improvements to Outputs Estimates

The ONS has, following the advice of the Review, made improvements to the coverage of preventative healthcare activities by reviewing the extent to which local authority commissioned public health activities are captured in its output measure. Many local authority funded services were already included, as a wide range of services are commissioned to NHS providers. However, some services are also commissioned to a wide range of alternative providers. Treatments for alcohol misuse, drug misuse, and smoking cessation services commissioned to non-NHS providers were added to the ONS output measure from Spring 2024. However, the ONS has not yet been able to apply the alternative weighting methods proposed in the Treatment of preventive services section of Chapter 3. Given the growing importance of preventative services the Review considers this should remain a priority for future work. 

The ONS remains in conversation with the devolved governments around improving coverage of their data (as per Recommendation 14). There also remains a lack of data on NHS-funded services contracted from the independent sector and this limitation is acknowledged within the healthcare sector. 

From the Spring 2025 annual release the ONS will enhance the healthcare output measure to incorporate certain screening services, improved measures for primary care, equivalised unit costs for equivalent treatments across different modes of provision, and removal of excess bed days activity. 

Screening services 

Historically, abdominal aortic aneurysm, bowel cancer, breast, and cervical screening services have been excluded because of data limitations. These can now be included because of improved expenditure data availability from the financial year ending 2023 National Cost Collection (NCC) and long-term activity data from NHS England and DHSC. 

The ONS will use expenditure proportions from the financial year ending 2023 NCC as a baseline and activity growth data from NHS England and DHSC to establish a reliable back series from financial year ending 2015. This method accounts for both the relative cost differences between screening programmes and the annual shifts in activity. This provides a more accurate reflection of change over time than other methods considered (applying the financial year ending 2023 unit costs to each year in the back series, or assuming the cost of one screening programme relative to the others in financial year ending 2023 has remained consistent over time). 

There are currently quality issues with the NCC activity data, therefore the ONS is only using this source for expenditure and using other sources for activity. When the quality of the NCC activity data improves, the ONS intends to make this the sole source to measure screening services, streamlining the process, and reducing reliance on other datasets.

These changes are being applied to England only initially, as regular activity and expenditure data for these services for Wales, Scotland and Northern Ireland are not currently available to the ONS. The ONS will continue to work with the devolved health administrations to assess the feasibility of incorporating these adjustments for the other UK nations in the future. However, the inclusion of these services for England increases coverage of healthcare services, ensuring a more comprehensive and accurate reflection of healthcare activity and expenditure in the UK productivity measures. 

Primary care 

The Review considers that the consistency of the dental activity time series can be enhanced by using an alternative source of data to estimate activity growth from 1996 until 2006, and an improved method of linking different activity data sources. The Review also found that the cost weighting for ophthalmic services and NHS telephone and website services can be improved, which means these services are weighted more appropriately relative to other Healthcare outputs. For ophthalmic services this can be achieved by ensuring the overall weight of the component is consistent with expenditure as reported in the DHSC annual accounts. For NHS telephone and website services this can be achieved by uprating the historical unit costs to account for general inflation in NHS costs.  

To better represent activity undertaken over the pandemic period, for NHS dental services, the ONS will reduce the size of the expenditure weight used to calculate the contribution of dental output in financial year ending 2022. Rather than the standard approach of taking net dental expenditure as the dentistry weight, the ONS will uprate the financial year ending 2020 unit costs by a growth factor equivalent to growth in other healthcare services between financial year ending 2020 and financial year ending 2021. This approach ensures that the resulting implied expenditure weight for dentistry is more reflective of the cost of the activities performed by dentists over the year, rather than the full contract value. This approach is more consistent with the cost weighting applied to other healthcare services and ensures the weight given to the fall in output from dentistry in financial year ending 2021 and recovery in financial year ending 2022 do not result in an unrepresentative contribution to growth from dentistry. 

Equivalisation of weights for (i) acute care 

As discussed in Chapter 3, in the absence of prices, cost weights are typically used as a next best alternative to approximate the value of different services. The granular data available for healthcare activity and costs enables a high degree of differentiation in value between different services. However, where process improvements lead to lower-cost service delivery methods which are recorded as separate activity types, they are assigned a lower weight in output, meaning efficiency gains from moving to lower cost treatment are not represented in the productivity measure. 

This is particularly notable in the case of elective surgery, where procedures may be carried out either as an inpatient procedure, a day case procedure or an outpatient procedure. Historically, separate unit costs have been used for each procedure type. Therefore, where more procedures transition from overnight hospital stays to same-day treatments and costs fall, this results in more lower-weighted activity and so appears as a reduction in output, even though the same care is being provided more efficiently. 

The ONS has developed equivalised unit costs for equivalent treatments across different modes of provision. These are applied by combining activity and expenditure across different services categories within each Healthcare Resource Group (HRG). HRGs are clinically meaningful groupings of patient activity derived from NHS patient records, primarily using procedure and diagnosis codes. They provide a means of determining fair and equitable reimbursement for healthcare services by providing consistent ‘units of currency’, based on expected resource use. This approach generates a new unit cost, calculated as a weighted average of the previously separate unit costs, reflecting both higher and lower-cost modes of care. 

For inpatient and day case procedures, there is no restriction on inclusion in the equivalisation. If a HRG exists in more than one of those components, an equivalised weight will be applied. For outpatient procedures, unit costs are only equivalised where the HRGs tariff (the price paid by commissioners under the NHS Payment Scheme) is equal to that of elective inpatient and day cases. This follows the approach used by NHS England in their new productivity measure. 

Equivalisation of weights for (ii) ambulance services  

There is a similar issue with the reduction of avoidable ambulance service activity through increasing provision of medical advice by telephone and care in situ by ambulance crews without conveying patients to hospital. This approach improves patient outcomes and reduces unnecessary hospital admissions, easing pressure on emergency care services. Over time, this should lead to a decline in the more expensive ‘See, treat and convey’ service and an increase in the lower cost modes of ambulance service delivery. Further details on reducing avoidable conveyance can be found in Planning to Safely Reduce Avoidable Conveyance. This aligns with the NHS Long Term Plan which introduces new standards to ensure sickest patients receive the fastest possible response, while also ensuring that all patients receive appropriate care the first time, within a clinically suitable timeframe.  

The ONS has addressed this issue for ambulance services by calculating an average unit cost across the following components of treatment: 

• Hear and treat or refer – when a person does not require an ambulance, but a clinician is able to provide treatment and advice over the phone.
• See and treat or refer – when a person does not require hospital care but instead a paramedic or another clinician provides treatment at the scene. 
• See, treat and convey – when a clinician or paramedic assesses and provides treatment at the scene and determines that the patient requires further care, resulting in the patient being transported to a hospital or other healthcare facility for additional treatment. 

As a result, a shift in the composition of ambulance activity towards more calls being responded to with advice or care in-place without conveyance will not result in a reduction in output. A shift is likely to lead to a reduction in inputs, hence productivity would improve. Emergency 999 calls and other ambulance services, such as those referred from NHS 111, are not included in the equivalisation because the ONS intends to retain the difference in the weight of these services as they are not necessarily alternative services to ambulance conveyances. 

Equivalisation of weights for (iii) the devolved governments 

These changes have currently only been applied to England’s healthcare output. In Wales and Northern Ireland, inpatient and day case activity are already aggregated and equivalised in an equivalent manner to methodological developments applied to England in the source data provided by these administrations. For Scotland, inpatient and day case equivalisation has not yet been implemented, but the feasibility of implementation can be evaluated in future.  

The equivalisation of outpatient procedures and ambulance services are not currently applicable to the devolved governments because of differences in data reporting and availability. The ONS will continue to investigate opportunities to develop methods for Wales, Scotland, and Northern Ireland. 

Excess bed days 

A further improvement developed is the removal of excess bed day activity from historic volume calculations. Excess bed days are additional hospital stays extending beyond the expected treatment period for clinical reasons e.g. complications or other medical needs. Prior to financial year ending 2019, excess bed days were treated as additional activity categories within NHS England’s National Cost Collection. From financial year ending 2019 onwards, the excess bed day activity data were removed and the costs were captured within the elective and non-elective care components.  

This improvement integrates expenditure from excess bed days into the respective elective and non-elective care categories and removes excess bed days from the activity measure entirely. This ensures that growth in the number of excess bed days no longer contributes to output growth during the financial year ending 2015 to financial year ending 2019, aligning with the NHS’ financial year ending 2019 update.  

Excess bed days are not identified as a distinct component in the data from the devolved governments. Therefore, the change the ONS is implementing for England does not need to be applied for the devolved governments.

Both the removal of excess bed days and the equalisation of weights across admitted patient care and ambulance services have been applied to the growth rate from financial year 2015 onwards, as changes in how activity was recorded in financial year ending 2014 present challenges for reprocessing earlier years. The impact is expected to be largest on earlier years, particularly financial year ending 2015, when significant shifts in care delivery methods began. The ONS has also applied minor consistency adjustments to data from financial year 2015 onwards.

Unnecessary accident and emergency (A&E) admissions 

The Review also considered a further improvement to output to account for reducing unnecessary A&E admissions by treating more patients in primary care settings being another potential source of productivity gains. A&E attendances carry a higher unit cost, giving them a higher weight in output calculations, relative to General Practice (GP) consultations. The Review developed a method to adjust the output weight for unnecessary A&E admissions to represent that of a typical GP consultation. This approach highlights that equivalent care could be delivered through a lower cost GP consultation, while acknowledging that A&E inputs remain higher, thereby indicating lower productivity when A&E resources are used for these cases.  

However, early estimation of the effect of this new method suggests the impact on output is very small, although this should be monitored. Given the additional processing burden for a minimal impact on overall output, the ONS has not implemented the new method at this time, although the ONS should be ready to return to this if it becomes more significant.  

7.5 Improvements to Quality Adjustments

Indicators of quality are conceptually easier to identify for healthcare than for most other public services and a number of quality adjustments that account for patient experience and effects on health-related quality of life are already applied to the productivity estimates. However, the Review identified opportunities to improve some of the existing quality adjustments. For GP patient outcomes, the ONS measures the change in the proportion of patients managing different health conditions through a selection of quality indicators from the Quality and Outcomes Framework (QOF), a tool that helps assess the performance of general practice. 

Through consultation with primary care experts at Imperial College London additional outcome metrics were identified, concentrating on indicators with a sufficient time series, or without substantial changes in definition over time. The Review also assessed the breadth of conditions covered, identifying two additional health conditions with suitable coverage to include in our measure – asthma and diabetes mellitus.  

The means of weighting the significance of the indicators used to measure outcome improvements were then reviewed. Previously the change in each indicator was weighted equally; since Spring 2024 the ONS has used QOF points to weight indicators relative to one another. This represents the value given by the NHS to each indicator, meaning that indicators for activities perceived as more valuable are given a higher relative weight than lower value indicators.  

The ONS’ existing quality adjustment for patient experience prior to the Review included patient satisfaction measures for admitted patient care, emergency services and mental health services taken from surveys conducted by the Care Quality Commission. Patient satisfaction in primary care had been a gap in coverage since financial year ending 2008, when the National Patient Survey Programme was discontinued. The Review identified that the GP Patient Survey, which has been in operation since financial year ending 2009, can be used as an alternative data source for measuring patient experience in general practice and dentistry services.

A range of other data sources investigated to further expand patient satisfaction measures include the ‘Family & Friends Test’, which is used within a wide range of NHS services. However, NHS England advised against its use because of the voluntary nature of the test, likely providing biased results. The ONS will continue to monitor progress of the Care Quality Commission in rolling out patient satisfaction surveys for other healthcare services.  

In 2024 the ONS was commissioned by NHS England to introduce the Health Insight Survey, with results on satisfaction and access with various NHS services published quarterly in Experiences of NHS healthcare services in England. While this data source does not yet have the length of time series needed to be included in the patient satisfaction quality adjustment, the ONS will evaluate its suitability for this purpose in future. 

A quality adjustment is also applied that uses the Quality-Adjusted Life Years conceptual framework to estimate the health gain from treatment over remaining life expectancy. This has 2 dimensions: 1) health gain, with an adjustment accounting for survival rates and average health gain from the procedure, and 2) time, where the health gain is applied over discounted remaining life expectancy after accounting for waiting times to treatment. Until now, the ONS has applied this adjustment in a way which means no adjustment for the time component is applied to non-elective treatments because waiting times are not  relevant to these treatments. The Review has identified that the ONS can improve the quality adjustment by assigning a time quality adjustment factor for non-elective treatments based on just the life expectancy. This is consistent with the approach taken by the University of York and allows for any changes in life expectancy in non-elective treatments to be accounted for in the quality adjustment measure. This improvement has been implemented back to financial year ending 2015 for consistency with other developments implemented as part of the Review.  

7.6 Recommendations for further work

Substitutions of higher-cost services for lower-cost services 

As already described, the Review has identified a method for ensuring that efficiency gains from process improvements that lead to lower-cost service delivery methods are represented in the productivity measure. While the Review has implemented methods to account for these productivity gains in acute care and ambulance services, new ways of delivering equivalent outcomes for patients by employing lower-cost forms of healthcare provision continue to emerge. In particular, the planned moves from hospital to community services and greater use of digital services could have large implications for productivity. Identifying these substitutions in the future and addressing them in the productivity measure would help to ensure that important contributors to improvements in productivity are captured.  

Improving quality adjustment of waiting times 

The Review started to explore whether the ONS’ existing quality adjustment for waiting times could be a) expanded, so that an adjustment for waiting times is applied to a wider range of healthcare services, and b) improved, by taking account of longer waiting times leading to worse health outcomes. 

The current waiting times adjustment is applied to elective inpatients and day cases only and concentrates on the delay in health gain. Any change in waiting times is spread over remaining life expectancy, meaning the effect is trivial. Accounting for changes in waiting times for a wider range of services and adding a further adjustment to account for longer waiting times leading to worse health outcomes for patients would mean that a change in waiting times would carry more weight in the productivity estimates.

A literature review identified a link between delays and increased risk of death for some treatments, but a lack of evidence exists to suggest this approach is generalisable. Therefore, further consideration of this approach is needed before any changes can be implemented. Some data sources have been identified that could potentially be used to create a waiting times quality adjustment where the ONS does not currently have one, however, further research is needed to consider how the adjustment would be weighted and applied for different service types.  

Preventative services – a case study 

As discussed in Chapter 3, preventative services, such as public health interventions aimed at reducing the incidence of disease, merit special consideration in the measurement of public service output as the public value they generate is typically disconnected from the cost weighted bundle of activities delivered. Preventative services instead reduce the volume of cost weighted activities demanded by the public in future, such as reducing the need for surgery, while also leading to improved public welfare.

As part of the Review the ONS commissioned Weale (2024) to consider potential methodologies for measuring preventative services in public service output and productivity, using the Diabetes Prevention Programme as a case study.  

Further practical challenges to implementation include the lack of time-variant data on the outcomes of preventative programmes and the wide variety of such schemes, with each having only a marginal impact relative to healthcare overall. The ONS intends to further investigate the feasibility of the approach proposed by Weale (2024) to scope other preventive programmes where the methodology could be applied, and determine if the challenges can be sufficiently addressed to merit the implementation of such a bespoke approach to the output of preventative services. 

NHS Payment Scheme 

NHS trusts are funded through payment for activity based on the cost of delivering different activities. However, for some activity, NHS England modifies the costs used in the payment scheme to incentivise trusts to adopt improvements in clinical practice. This may be through improving outcomes or adopting new practices or technology to deliver similar services at lower cost.

The ONS has started to explore the feasibility of adjusting the cost weighting of certain HRGs within the output measure to incorporate these incentive-based modifications for acute care. The ONS found that while it may be possible to use the NHS Payment Scheme weights for acute care, there remain significant difficulties with implementing this approach into healthcare output.  

In particular, the matching of the NHS Payment Scheme data with the NCC data is challenging, particularly prior to financial year ending 2018, because of many HRGs in the NCC data not having equivalent categories in the NHS Payment Scheme data. Another issue is that acute trusts do not only receive funding for acute care through the NHS Payment Scheme, which means they have additional costs which would still need to be accounted for when weighting their care against other services which are cost weighted. 

Coherence of the ONS and NHS England healthcare productivity estimates 

In developing this report, the Review has had the opportunity to observe the reaction of stakeholders to the data which has been published during the life of the Review, especially where new data have been made available. In February 2025 the ONS published new quarterly estimates of healthcare productivity, which differed to those published concurrently by NHS England. These differences are explained by a) the ONS measuring the UK whilst NHS England measures England, b) the ONS measuring acute, GP, and community services, including COVID-19 vaccination services and NHS England only measuring acute services, c) the ONS comparing quarter with the same quarter the previous calendar year whilst NHS England compares the first seven months of the financial year, d) the ONS applying seasonal adjustment while NHS England do not, and e) minor methodological and data differences.

Nevertheless, the Review noted that this debate allowed the relative productivity performance of the hospital sector versus other parts of the healthcare system to be exposed as a potentially important area of investigation. It suggests that where healthcare accounts for 41.5% of public services, to report this as one service whilst services such as tax and social security administration (which account for less than 2% each), are given equal treatment, may hinder user understanding of the key drivers of productivity growth. Taking this example, where NHS England are able to produce statistics coherent with the ONS which allow greater granularity to be shared with users, the ONS should look to develop a more granular breakdown of the healthcare service to be included in its statistics. 

In addition, during the completion of the Review, the potential to explore working with NHS Wales to explore its datasets to augment the measures of Wales for Healthcare services presented itself. This occurred too late in the process to be taken forward as part of the Review itself, but is an opportunity the ONS should take forward.