The value of focus in the NHS (at least £8bn per year …)

Strasys analysis shows that NHS Trust productivity tends to decrease as the size of the Trust increases. In addition, health outcomes relative to disease prevalence within the population also decline as Trust size increases, whether Trust size is measured by activity volume or Operating Expenditure. This creates an interesting “trapped value” scenario within the NHS.

There are 13 acute trusts over £1.5bn in annual Operating Expenditure (FY23-24). Not one of those trusts can match the average productivity of the other 104 smaller acute trusts; if all 13 did manage that average productivity, the NHS would have saved over £8bn in financial year 2023-2024. Moreover, health outcomes would be significantly improved.

Given we all know that “There. Is. No. More. Money”, maths and the reasoning set out below would suggest that “There. Must. Be. More. Focus”. This is central to the Strasys thesis for better healthcare.

It often pains me when economic and business terminology is misapplied in the public sector. After 26+ years with a foot in both the private and public camps, I know that bi-directional sharing of lessons is beneficial, but I also know that context and reasoning from first principles is essential.

Take the application of “economies of scale”; whilst the idea of amortising fixed costs over the largest possible variable cost base and reducing marginal costs per unit through scale has some obvious merit, what is often missed is the tipping point at which complexity of service lines means bigger is harder, not better.

Economies of scale, and – perhaps more importantly for healthcare – economies of learning, occur when an organisation can do more of the same thing, and do that thing in a standard way. A hospital that just does hernia repairs can get really good, and really efficient, at fixing hernias by doing lots of them in a consistent fashion (e.g. Shouldice Hernia Hospital).

An acute hospital with typically 50+ medical service lines does not necessarily get more efficient or more effective by adding more volume; and it certainly compounds its problems by adding more service lines – i.e. it experiences diseconomies of scope. In healthcare, economies of focus need to be better understood.

My esteemed @Strasys colleague, Dr Nadeem Moghal, in his latest Friday Fish and Chip Paper quoted Peter Drucker saying that hospitals are “the most complex human organisation ever devised”. It stands to reason that such complex human organisations follow different rules to simpler companies addressing narrow customer segments or delivering related products (economies of scope).

This analysis started by asking the simple question of whether the 15 mergers of acute NHS Trusts (in 1 instance a merger of a Specialist trust into an acute) that occurred between 1st April 2017 and 1st April 2023 drove noticeable improvements relative to the performance of the pre-merged trusts. To cut short a lot of data cleansing and time buried in Python and spreadsheets, the answer is no.

There is of course a much longer answer, but not suitable for this medium – do reach out if you would like to discuss. What surprised me most was that it became hard to tell if relative poor performance of merged trusts was because of the merger or just because larger trusts under perform. Obviously, when trusts merge they get bigger. It is not the case that this invariably leads to poorer productivity and health outcomes – some trusts do well (kudos to Royal Devon and especially South Tyneside and Sunderland), but the results are certainly worrying enough that any Trust Boards currently planning a merger should pause and give us a call.

I cannot do justice to the analysis in a post, but a bit of necessary statistics follows. Trust size, defined as OpEx less Impairments and High Cost Area Supplement (HCAS) costs, has a Pearson correlation coefficient of -0.31 to productivity, defined as total activities delivered per £1k of adjusted OpEx. That has a p score of 0.0006, so well below the p < 0.05 statistical significance threshold. That means size only determined ~10% of productivity – i.e. lots of other factors are at play – but the impact as shown in the scatterplot above is significant. None of the 15 largest trusts (measured by OpEx) manage to get to average England Trust productivity.

A linear regression line is shown in the chart above, but the relationship is not strictly linear. The ‘big trust penalty’ gets worse for the very largest trusts. A trust two standard deviations (SDs) below the mean size loses ~0.20 SD of productivity for each +1 SD size bump. At +3 SD (very large trusts) the loss grows to ~0.42 SD per +1 SD – roughly double the penalty faced by small trusts. Even at this large scale, trust size only explains ~ 11% of variation, but the financial sums here are staggering. If the 13 trusts over £1.5bn in annual OpEx to FY ending 2024 were to get to the average productivity of the 104 smaller acute trusts, we would be talking about over £8bn per year in FY24 figures of savings.

It is not just productivity that is adversely impacted by trust size. Whether you look at OpEx or total activity volume, there is clear negative correlation with what we at Strasys call Health Alpha.

Health Alpha is a trust’s outcome score (based on a combination of mortality and emergency re-admissions) relative to the disease prevalence within the population the trust serves. A positive health alpha means better outcomes than would be expected for that disease prevalence; a negative health alpha means worse outcomes. Total volume of activity and Health Alpha have a Pearson correlation coefficient of -0.33 (p 0.00023); OpEx and Health Alpha have a Pearson correlation coefficient of -0.32 (p 0.00041) – see chart below.

Again, these scores mean size only determines approximately 10% of health outcomes – many other factors are at play – but scale that out across England and we are talking about a lot of excess deaths and readmissions.

If you are wondering why this is not well known against publicly available data, I will happily regale you with the intricacies of trying to clean and map data from multiple public data sources including TAC (keyed on non-standardised trust names – who does that??), HES, ECDS, CSDS, QOF, SHMI and others to get to a comparable view.

Absolutely not.

Years and £billions would be wasted. What the NHS does need is the implementation of new models of care, centred on meticulous understanding of patient needs, volumes, and behaviours. You reduce friction (i.e. cost) and improve outcomes by designing services around the patient.

There are four strategic focus lenses that healthcare organisations can apply and 6 high-level models of care that drive organisational effectiveness. In the interests of brevity I will leave description of those for another day. If we build around those and stop the fallacy of patient choice and tackle the idea that DGHs can still be most things to most patients and be effective, then there are huge improvements to be made within the existing financial envelope. Money that then be spent on new treatments and addressing #Gainsbury’s Law.

Please do reach out if you have questions or would like to discuss any of the above.