How efficiency varies across NHS acute trusts

Over the past few years, the NHS has expanded its workforce and received substantial funding increases. Yet activity (for example, the number of appointments or procedures) has not risen proportionately, reflecting a sustained decline in efficiency – that is, how effectively ‘inputs’ such as staff, beds and equipment are being turned into ‘outputs’ (admissions, appointments, procedures). 

This means less care is being delivered than would be expected given the resources available, contributing to longer waits for treatment. Recent work has documented this ‘productivity puzzle’ and explored the reasons behind it. NHS England has also developed more granular productivity measures that track activity and input levels over time to better understand performance across the system.

The decline in NHS efficiency is a relatively recent phenomenon. Between 1995/96 and 2018/19, the Office for National Statistics (ONS) estimated that NHS efficiency had grown by an average of 0.6% per year, rising to 1% when adjusted for the quality of care. The pandemic triggered a sharp reversal, with efficiency falling by around a quarter. While measures from the ONS and NHS England saw some recovery by 2022/23, efficiency still remained below levels seen a decade earlier. 

Restoring efficiency growth has become a central policy objective. The government has set a target to achieve 2% annual efficiency growth, partly by narrowing variation in performance between trusts. Understanding the extent to which the national efficiency level could be increased by reducing such variation is critical for policymakers.

This analysis examines how NHS efficiency has changed since the pandemic, the extent of variation across trusts and whether reducing that variation could provide a feasible route to improving efficiency nationally. Specifically, we ask whether the national decline in efficiency is being driven by a downward shift in efficiency levels across all trusts or a widening of variation between them. 

Health care productivity describes the relationship between the resources used and the care delivered. When productivity rises, the NHS can provide more or better care without additional investment in staff, technology or capital. When it falls, even large funding increases may not translate into higher levels of care.

Productivity is hard to measure and can be defined in different ways. The most commonly used measure is technical efficiency: how effectively inputs such as staff time and equipment are being converted into outputs such as consultations, procedures and admissions. However, there are legitimate concerns about an overly narrow focus on technical efficiency. Many argue that productivity should be defined more broadly to reflect how care translates into health outcomes (effectiveness) and whether the right care is being delivered in the first place (allocative efficiency). These concepts are being explored by our NHS Productivity Commission.

Nevertheless, improving technical efficiency is important. When resources remain unchanged, higher efficiency indicates more care is being provided with the same inputs. Improving technical efficiency and reducing variation across the NHS are central policy objectives reflected in the NHS Productivity Plan and reiterated in the 10 Year Health Plan. This includes aims to ‘close the productivity gap back to 2019/20 and level up worst performers with operational and clinical productivity improvements’, and the publication of a new trust-level efficiency dataset.

Our analysis therefore focuses on technical efficiency, using an approach close to the ONS non-quality-adjusted measure (see Box 1). We weighted outputs by cost to reflect differences in resource requirements – for example, knee surgery carries a greater weight than an outpatient appointment. Inputs were measured using staff volumes weighted by salaries, alongside spending on other inputs (adjusted for inflation, to account for changes in cost over time).

We analysed efficiency across 110 acute NHS trusts between 2018/19 and 2024/25 and found a clear decline in their ability to convert resources (inputs) into activity (outputs). Historically, NHS inputs and outputs have followed a broadly linear relationship: trusts with greater resources generally deliver higher levels of activity. That relationship still holds but has weakened since the pandemic. In practical terms, the trusts in our sample now deliver less activity for a given level of input than they did before the pandemic. This reflects trends seen in other measures of efficiency, which also show NHS efficiency being below pre-pandemic levels.

The pandemic created a major operational shock, requiring trusts to reduce routine activity to manage surges in complex COVID-19 admissions and implement new ways of working to prevention and control infection. At the peak of disruption in 2020/21, trusts were around 28% less efficient than in 2018/19. Although cost-weighted activity is now higher than in 2018/19, as of 2024/25, efficiency remains around 14% below pre-pandemic levels. Our measured efficiency level remains below those reported by NHS England and the ONS, likely due to differing methodologies. For example, our approach does not equally weight outpatient activity and elective day case activity, nor include advice and guidance. There are also differences in the way we measured inputs (see Box 1).

Looking beneath the national trend, there is meaningful variation in efficiency across trusts. In 2024/25, the trust at the 90th percentile of efficiency produced around 56% more outputs per unit of measured input than the 10th percentile trust (Figure 2), equivalent to being 1.6 times as efficient. This gap narrowed slightly after the pandemic but remains modest compared with the wider economy. Using a comparable measure for large companies (over 250 employees), the 90^th percentile firm is around 3 times as efficient as the 10^th percentile. This suggests there is less variation in efficiency between trusts than between organisations in the wider economy.

To assess post-pandemic recovery among the most efficient trusts (the ‘frontier’), we used the 90^th percentile as a proxy. Efficiency among this frontier fell by 28% during the pandemic, and despite some recovery, remains around 15% below its pre-pandemic level. This suggests that even the highest performing trusts have yet to recover. 

By this measure, relative positions across trusts have been broadly stable. Comparing the 90th percentile with the median shows that the typical trust has remained at a similar distance from the frontier since 2018/19. By 2024/25, the frontier remained 20% more efficient than the median, broadly unchanged from its pre-pandemic level. This suggests it is not a shifting of the frontier and a ‘catch-up’, but a systemic shift downwards across the board. 

Overall, this evidence suggests that while the variation between trusts has stayed broadly similar, efficiency has fallen across the board. If poorer-performing trusts were driving this trend, we would expect the gap to have widened as these fell further behind. Equally, if stronger trusts were driving recovery, the gap would also have widened as they pulled ahead. As neither pattern is observed, it seems there is a common constraint affecting all trusts, including the highest performers. This points away from a simple ‘leaders and laggards’ explanation and instead suggests a system-wide decline in efficiency. 

With variation and relative trust positions remaining broadly stable over time, we examined how efficiency changed for trusts at the top and bottom of the distribution post-pandemic. This helps identify whether differences in trust characteristics, inputs or outputs may explain variation in recovery.

We grouped trusts from the top and bottom quartiles of efficiency in 2018/19 and tracked them over time. While all trusts experienced a sharp deterioration in efficiency during the pandemic, recovery has been uneven. Between 2018/19 and 2024/25, trusts in the lowest quartile recovered more quickly on average than those in the highest (Figure 3), slightly narrowing the gap between the two groups.

Improving efficiency is a central policy objective, with recent NHS England plans setting out a target of 2% sustained annual efficiency growth and a programme of measures focused on variation reduction, operational improvement and better use of existing capacity.

To understand how NHS efficiency has evolved since the pandemic, we examined the extent of variation across trusts and the relative impact of changes in average performance versus the dispersion of efficiency levels across trusts.

Our findings suggest that the post-pandemic efficiency shortfall is best understood as a system-wide decline rather than a widening gap between higher and lower-performing trusts. This has important implications for current and future policy approaches.

First, the fall in efficiency across the entire distribution, including at the frontier, indicates that constraints are system wide rather than concentrated in underperforming trusts. This aligns with NHS England’s emphasis on pressures such as bed occupancy, delayed discharge, diagnostics and workforce capacity. It also suggests that policies focused primarily on identifying and correcting ‘low performers’ or reducing variation are unlikely to be sufficient on their own. Such policies may provide a one-off boost to system-level efficiency, but they will not address the shared structural constraints affecting all providers.

Second, the distance between median and frontier performance has changed little since before the pandemic. This stability implies that closing gaps between trusts can contribute to improvement but is unlikely to drive system-wide recovery over the long term. These findings also highlight the limitations of technical efficiency as a measure. Not all cross-trust differences should be interpreted as inefficiencies, particularly given variations in their case mixes and organisational functions, such as teaching and research activity. Current efficiency measures do not go far enough to distinguish between unavoidable costs and low efficiency amenable to intervention. Refining efficiency measures to appropriately capture such activities would be an improvement in this area.

Third, weak efficiency growth appears more attributable to rising non-workforce costs than to labour-related factors. While efficiency as measured by outputs per worker has shown some recovery, spending on servicing capital and operating expenditure – including on estates, supplies, drugs and outsourced services – has risen more rapidly. Some of this improvement in outputs per worker may have been driven by post-COVID-19 efforts to reduce spending on temporary staff. The reasons for increases in other inputs are less clear, but increases in drugs costs and overheads such as energy costs are contributing factors. We did not find evidence of input costs being driven by investments in new buildings and equipment (referred to as ‘capital deepening’), but rather by an increase in the running costs associated with capital.

Overall, reducing variation between trusts can support a one-off boost to system-wide efficiency, but on its own is unlikely to close the post-pandemic efficiency gap or to lead to a sustained higher level of efficiency. Because the shortfall is largely common across providers, restoring efficiency growth also will require system-wide interventions that address shared constraints on capacity and input costs.