Four systemic failures explain why good technologies take years to be adopted correctly in the NHS — and why strong evidence alone is not enough. The Implementation Engine closes all four. And because it accumulates evidence with every device and every site that passes through it, it compounds: the adoption infrastructure built on this methodology becomes more capable with each technology it proves.
NICE identifies promising technologies. AAC endorses them. The MedTech Funding Mandate funds specific ones. None of these mechanisms address what happens between national endorsement and a Trust using technology correctly, generating the outcomes that were promised — at scale, consistently, and in a way that makes the next adoption faster.
National evidence does not automatically become a local capital committee approval. No mechanism exists to express a device's value in the specific governance language — BAF risk closure, activity income under the current payment regime, workforce exposure, CFO-recognisable financial return — that each Trust actually decides in. Valuable technologies fail at committee not because the evidence is weak, but because the translation does not exist.
High-capital devices sit on 10-year depreciation cycles. Even when multi-centre evidence clearly establishes a superior alternative, the accounting logic of an undepreciated asset defers displacement for years. The cost of not adopting — avoidable complications, foregone coding income, preventable patient harm accumulated year on year — is never modelled. The governance question is always "can we afford to adopt?" It is never "can we demonstrate we can afford not to?"
NHS Trust governance can only account rationally for impacts within its own budget boundary. Technologies with whole-pathway consequences — reduced downstream readmissions at a different provider, reduced community follow-up demand, long-term care cost reduction across the ICS — are systematically undervalued. The investment case cannot see beyond the Trust wall, and the technologies with the widest reach pay the highest price for that structural blindness.
After adoption, monitoring stops. No mechanism exists to determine whether technology was implemented correctly, whether training produced genuine protocol fidelity, or whether the promised outcomes are materialising. The evidence that would accelerate the next adoption — real-world process quality data, fidelity signals, implementation lessons — is never generated. Each adoption is, from a system learning perspective, a dead end.
The four failures above apply to technologies that have already reached national endorsement. But most valuable specialty innovations never get there. The formal adoption pathway — NICE assessment, AAC acceleration, MedTech Funding Mandate — is a needle's eye: tens of thousands of technologies in active NHS use, roughly 30–50 NICE MIBs per year, perhaps 20–30 in active AAC acceleration, as few as 8–12 with a MedTech Funding Mandate. Technologies that are genuinely effective but commercially small, specialty-specific, or evidenced only through dispersed single-centre studies never consolidate the signal needed to enter it. The Implementation Engine operates in both directions: closing the adoption failures for technologies that have been endorsed nationally, and surfacing the real-world evidence that identifies which technologies deserve to be.
A device generates value across multiple archetypes simultaneously. NHS Trust investment governance does not evaluate devices on integrated value — it evaluates them on a much smaller set of decision-currency signals: BAF risk closure, regulatory compliance, activity impact under the current pay regime, workforce exposure, external funding leverage, board attention. Under budget pressure the filter narrows further to what is mechanically affordable AND what is already a recognised problem at board level. The Implementation Engine surfaces both — the archetype profile (what value the device creates) AND the decision-currency translation (the language Trust governance actually decides in). Without this translation, even the most valuable investment fails the committee.
Every case produces comparable per-patient cost and per-patient benefit metrics across the asset's usable life. Two questions are forced into the open: is recoverable value being identified or actually realised, and where there is no cash release, what is the clinical-quality cost of the alternative — leaving harm in place?
Investment value is modelled across the full system, not the Trust budget envelope. When an HLM reduces neurological injury, the downstream beneficiary is long-term care. When it accelerates throughput, the DGH absorbs less escalation pressure. When clinical-quality benefit is real but cash release is absent, the case still needs to be visible — to the ICB, to NHS England, to the right system payer.
When clinical-quality benefit is high but Trust cash position is marginal or negative, the engine raises the case to the system payer rather than letting it fail at the local committee.
A capital committee responds to payback periods and cash release. The clinical value of preventing a stroke, eliminating a post-operative AKI episode, or avoiding post-operative cognitive dysfunction can dwarf the cost of the device that prevents it — but under current NHS payment structures that value does not flow back to the Trust that made the investment. It sits outside the financial model: real, large, and invisible to the decision. The Implementation Engine makes both visible simultaneously and keeps them deliberately separate: cash release tracked against deployment cost, clinical-quality value flagged alongside it. Not because they should be conflated — they should not — but because a governance system that can only see the smaller figure is systematically failing the patients whose long-term outcomes depend on a decision that never gets made.
Switch the valuation lens. Under conservative scoring the clinical-quality value is comparable to deployed capital. Under NICE QALY-equivalent valuation it dwarfs it. The case for an expanded CapEx envelope is the difference between what is flowing through the system and what the cash-only view is capturing.
Evidence generated at Site A is translated into Site B economics using local data — case volume, outcome baselines, DGH catchment intensity, downstream care patterns. The same finding produces a different projected return at every tier: cash release to the Trust, escalation offload to local DGHs, downstream care effect to community and long-term services, clinical-quality value across the population. Below the evidence threshold, the finding is informative. At threshold, it is actionable. Above threshold, it is a procurement signal — and a Trust whose cash position is marginal but whose system value is substantial gets escalated to the ICB rather than failed at the local committee. When evidence is stronger still — consistent signals across five or more centres, validated through calibrated implementation cycles, surviving translation across different patient populations and case mixes — the cost-of-not-adopting model generates an explicit displacement case: the financial and clinical argument that writing down an undepreciated device early is superior to continued deferral. At the highest confidence level, the accumulated evidence becomes an AAC referral candidate. The Implementation Engine is not a parallel track to the national adoption pathway — it is a feeder into it, surfacing the technologies that deserve national endorsement but would never have generated the consolidated evidence to reach it.
High-volume tertiary centre (1,200 cases/year)
Adjustable local parameters
The Trust sees its own cash position. The system sees the upstream effects (pre-op optimisation, fewer cancellations rippling through the elective recovery pathway) and the downstream effects (less DGH escalation, less community follow-up, less long-term care). The same investment produces different break-even curves depending on which lens is applied.
The 90-day cycle is a calibration framework, not a fixed implementation timeline. Complex technologies may take longer to stabilise — and the engine tracks the full adoption curve however long it runs, separating implementation immaturity from true device or protocol effect throughout. What 90 days marks is the first point at which a validated signal can be trusted: calibrated, attributed, and ready to inform the next decision. By embedding quality intelligence and the supplier-Trust relationship from day one, adoption difficulties become visible and workable as they arise rather than discovered at the end. And when a second site adopts the same technology, it starts with the first site's calibrated evidence — the adoption curve accelerates because the learning transfers.
—
I have known about the next generation of heart-lung machine since it was released eight years ago. The HLM fleet at Barts Heart Centre was three years old at that point — adequate, meeting every safety standard, but missing the features that move perfusion forward. The newer device offered them, and through eight years of incremental development those features have continued to accumulate: a ventilation module, a host of features built around the needs of the perfusionist — making better and safer care possible without adding to the documentation burden that already competes with clinical attention.
For most of those eight years I have had to wait. First for the natural end-of-life of a fleet that meets every minimum requirement but does not enhance care. Then, separately, to make the argument that replacement should be an upgrade — not a like-for-like substitution. The business-as-usual case (this one is broken, buy another of the same) is not the value-based case (this is the moment to introduce technology we have evidenced for nearly a decade). They are different arguments, made to different committees, against different envelopes, requiring different evidence. Both arguments had to be won, sequentially, for one fleet, in one centre.
We succeeded locally. The argument was made, the case was won, and the next generation is on track. But the success was local. Every cardiac centre in the UK runs a similar fleet. Every centre has equipment somewhere in its replacement cycle. Every centre's perfusionists will have read the same evidence I have. Whether each centre ends up with a best-in-class HLM at the highest specification — the device that improves outcomes, accelerates training, reduces complications, and runs the structured informatics this prototype is designed to make universal — remains a national lottery.
That is the gap the Implementation Engine closes. Local success becomes national signal. The evidence that took eight years to translate into one fleet replacement becomes the evidence that translates into adoption across the network — without every centre having to wait for its own equipment to reach end of life, and without every clinical lead having to win the upgrade argument from scratch against an envelope that defaults to like-for-like. The lottery becomes a pathway.
"When you start these things the data is never as good as you would like and that’s where the courage comes in… data only becomes good when you use it."
The Implementation Engine enters the Trust through finance and governance — where the immediate problem is acknowledged and the appetite for better tooling is real — and exits the initial implementation cycle having added a structured, validated node to the clinical intelligence model. It then propagates that node into a translation layer that turns local evidence into national signal, and continues running as a permanent QA mechanism that surfaces drift before it becomes harm.
This is how a provisional model becomes a validated one — through accumulated, calibrated implementation cycles, not a single speculative build. It is the first operational mechanism in NHS digital that closes the loop between local evidence, system economics, national procurement intelligence, continuous quality surveillance, national process quality registry feed, and compounding analytical maturity — without requiring new infrastructure, new data sets, or new clinical work.
At the centre of this is a structural alignment that no existing NHS procurement mechanism creates. The device supplier needs real-world outcome and process-quality data to validate its investment case and accelerate adoption at the next centre. The Trust needs implementation support and ongoing fidelity monitoring to realise the benefits its capital committee approved. Both parties want the same thing: the technology working as evidenced, embedded in the clinical environment as intended, generating outcomes that justify the decision. CauseACTION gives both the structure to embed that shared interest permanently — turning what is typically a transactional procurement relationship into a continuing evidence partnership, held together by objective data neither party controls alone.
Multi-archetype scoring including clinical quality, patient-normalised economics, 90-day cycle adding a twin node per investment.
System economics surface investments whose cash value sits outside the Trust budget, or whose primary value is clinical, and route them to the right system payer.
Local-data translation generates site-specific ROI from multi-site evidence including DGH offload and downstream care effects. The evidence threshold triggers acceleration ahead of replacement cycle.
The cycle persists as permanent QA. Protocol drift, staff change, and workload pressure are surfaced in real time, not at the next audit.
Each new device in the ecosystem strengthens every other node. Noise reduces. Signals separate. The model improves through use.
The accumulated cross-site evidence base — validated through calibration cycles, permanent QA, and national registry feeds — becomes the methodology for value-based procurement evaluation within NHS Supply Chain frameworks. Whole-lifecycle economics, not capital cost alone, become the standard by which devices are compared and frameworks are awarded.
Every device that passes through this methodology — evaluated, implemented, monitored, and calibrated — adds a permanent, validated node to a clinical intelligence model. The post-adoption quality assurance layer the engine creates is not a separate tool: it is the same data architecture that CauseACTION builds across NHS pathways. The supplier-Trust relationship it structures, held together by objective process quality data, is the beginning of the connective tissue that the NHS already has the skeleton for — in its policies, its protocols, and its standards — but has not yet been able to make visible in real time.
CauseACTION applies this same principle across clinical pathways — connecting the data that already exists, surfacing the gap between work as imagined and work as done, and turning every routine case into evidence that makes the next one safer. The goal is not another reporting system. It is the intelligence layer that makes variation attributable, improvement automatic, and quality universally measurable — without adding to the burden of the people already keeping the system alive.
Join Us Thinking Ltd · NHS Clinical Entrepreneur Programme Cohort 10