In June 2010, England launched one of the most ambitious patient-safety programmes in its recent history. Every adult admitted to an NHS hospital would receive a documented assessment of venous thromboembolism (VTE) risk. Thromboprophylaxis would follow for those who needed it. Trusts would report compliance quarterly. Financial penalties would bite if performance fell short. NICE had already published its guideline — CG92 in January 2010 — setting out what good prevention looked like. The aim was clear: reduce deep vein thrombosis (DVT), pulmonary embolism (PE), and the deaths that follow from hospital-associated thrombosis.
Fifteen years on, the compliance charts look like a quality-improvement textbook. National assessment rates climbed from roughly 53% in mid-2010 to above 90% within two years. From April 2013, when the target was raised to 95%, the NHS met it and kept meeting it — quarter after quarter, trust after trust. By 2019, the figure sat at 95–96% and barely moved. On paper, the programme worked.
The NHS did not make thromboprophylaxis mandatory in the abstract. It made documented risk assessment mandatory — and then treated assessment completion as the primary measure of national success. CQUIN payments, standard-contract clauses, and board-level dashboards all converged on one question: was the form filled in?
That is a process metric — not whether the right patient received prophylaxis, whether doses were administered, or whether fewer patients developed clots. Root-cause analysis was part of the policy bundle, but compliance with assessment drove behaviour and reputation.
When a system optimises for what it measures, it should surprise no one when the measurement diverges from the outcome.
Hospital Episode Statistics (HES) do not record "hospital-acquired DVT" as a discrete field. What they do record — reliably, at national scale, year on year — is whether a DVT code appeared as a secondary diagnosis on an admission episode whose primary reason for hospital contact was something else. That is an imperfect proxy. It may include some pre-existing clots. It will miss silent events never coded. It cannot distinguish community DVT from thrombosis provoked by the index admission.
It is, nonetheless, the closest thing we have to a ten-year national signal of DVT arising in the context of hospital care — and that signal is not comforting.
| Period | Per 100,000 episodes |
| 2012/13 | ~166 |
| 2019/20 | ~210 |
| 2021/22 | ~225 |
| 2020/21 (COVID) | ~253 |
That is a red flag. It is proof that the assessment-based process the NHS built around the guideline — mandatory forms, compliance targets, central returns — did not produce the epidemiological pattern you would expect if that process were reliably reducing hospital-context DVT at scale. The failure lies in what was industrialised, not in the existence of thromboprophylaxis as a clinical intervention.
Those two lines were never supposed to run together.
If the assessment process were reliably triggering effective prevention among the patients the programme was designed to protect, we should see that in hospital-linked VTE mortality — deaths in hospital or within 90 days of discharge among people with a recent admission, with VTE on the death certificate. That is NHS Outcomes Framework indicator 5.1 (I00675), the official national outcome measure aligned with NICE's hospital-associated thrombosis framing.
Here the official narrative and the epidemiology part company.
|
The published story (rate)
72.8 → 62
Fatal VTE per 100,000 admissions
2007/08 to 2019/20 (61.2 in 2023/24) ✓ Widely cited as success
|
The epidemiology (count)
8,106 → 9,087
Absolute hospital-linked VTE deaths
2007/08 to 2019/20 (+12% pre-COVID) ↑ Significant upward SPC trend from 2013
|
The rate fell because the denominator grew faster than the numerator — adult hospital admissions increased by roughly a third over the same period — not because England was clearly putting fewer people in the ground from hospital-associated thrombosis.
Why does a falling rate mislead so convincingly? Because NHS OF 5.1 spreads fatal events across a denominator far broader than the population where those deaths actually occur.
National linkage work by Catterick et al. (BMJ Open, 2024), using the same case definition as the outcomes-framework indicator, found that 86% of hospital-linked fatal VTE followed emergency inpatient admission. Planned inpatient and day-case pathways together accounted for only about 12% of deaths — yet they form a large share of all admitted activity.
A rate can fall while the count of people dying does not — and even rises.
Day cases are included in both numerator and denominator of OF 5.1, so a day-case surge could artefactually lower the rate. HES data rule that out: day cases rose in number but held at about one-third of activity for over a decade. Fatal VTE stayed anchored in emergency inpatient care. The problem is a rate-based metric on a heterogeneous admission base, sold on the promise of fewer clots and fewer deaths.
NICE CG92 — later updated as NG89 — describes a pathway: identify risk, prescribe appropriate thromboprophylaxis for those who need it, consider post-discharge extension where indicated, investigate incidents, learn from harm. That is more than a tick-box.
What England actually mandated and measured was the first step — documented VTE risk assessment — and treated completion of that step as if it stood for the whole pathway. Form completion, central reporting, and a 95% compliance target became the national product. Thromboprophylaxis remained in the guideline text; it did not become the national scoreboard.
That substitution is why I regard the guideline as well meaning but failed in practice: not because prevention is futile, but because the assessment-based process given statutory force was the wrong proxy for it.
| Metric | What happened | Signal |
| VTE assessment compliance | Target met, sustained, celebrated | ✓ |
| HES secondary DVT | Upward trend through compliance era | ↑ |
| Fatal VTE (absolute deaths) | Flat to rising from 2013; SPC significant | ↑ |
| Fatal VTE (published rate) | Modest decline; cited as success | ↘ |
Assessment completion should be a gateway metric, not the finish line. Outcomes would be tracked in absolute terms and pathway-specific strata — emergency medical admissions first — with prophylaxis administration audited alongside forms. Secondary DVT in HES would be monitored as a sentinel, coding caveats acknowledged but not deployed to dismiss the signal.
The lesson generalises beyond thrombosis. When a guideline becomes mandatory NHS activity with financial teeth, ask which part of the pathway was actually enforced. If the answer is a single process step — here, risk assessment — you may get excellent compliance on that step and no superior clinical outcome on the harm the guideline was written to prevent.
The clinical intention was humane. The assessment-based architecture was not equal to it.
- NICE. Venous thromboembolism in over 16s (NG89). nice.org.uk/guidance/ng89
- NICE. VTE: reducing the risk for patients in hospital (CG92). nice.org.uk/guidance/cg92
- Department of Health. Report of the Independent Expert Working Group on Prevention of VTE. 2007.
- NHS England / NHS Digital. VTE risk assessment quarterly data and CQUIN specifications, 2010–2020.
- Catterick MD, Hunt BJ. Impact of the national VTE risk assessment tool in secondary care in England. Blood Coagul Fibrinolysis. 2014;25(6):631–635.
- Hughes F, et al. HES DVT/PE trends. BMJ Open. 2025. doi:10.1136/bmjopen-2024-090301
- Catterick MD, et al. Who dies from VTE after hospitalisation in England? BMJ Open. 2024. doi:10.1136/bmjopen-2023-078898
- NHS England Digital. NHS Outcomes Framework 5.1 (I00675). Feb 2025 release
- Nuffield Trust. Blood clots following hospital care. nuffieldtrust.org.uk
- NHS England Digital. Hospital Admitted Patient Care Activity (HES). digital.nhs.uk
- Hunt BJ, et al. VTE prevention: UK experience. Res Pract Thromb Haemost. 2023. PMC9903667
