A figure circulates on surgical wards: the risk of pulmonary embolism in hospital inpatients is about one per cent. The risk of catastrophic bleeding from pharmacological thromboprophylaxis is also about one per cent. Equal numbers. A stand-off. Many surgeons — particularly in general surgery — treat that parity as a reason to hesitate, to individualise, to let the VTE assessment form decide.
The arithmetic is seductive. The conclusion is wrong — not because surgeons are careless, but because the two percentages measure different things and cause different harms.
In pooled randomised data for moderate-risk general surgery, low molecular weight heparin prevents clinical venous thromboembolism in roughly one patient per 150 (number needed to treat ~147). Major haemorrhage attributable to prophylaxis occurs in roughly one per 67 (number needed to harm ~67). On a pure event-count basis, more patients may suffer prophylaxis-related bleeding than avoid thrombosis — in that selected trial population.
That is the evidence behind the worry. It is why some analyses argue chemoprophylaxis should be reserved for groups where clinical VTE risk exceeds roughly three per cent. It is why a surgeon who sees a wound haematoma or a return to theatre blames the heparin.
The ward conversation — one per cent versus one per cent — compresses that literature into a single intuitive dread. Rate parity feels like equipoise. It is not.
Major bleeding in prophylaxis trials means transfusion, drop in haemoglobin, sometimes reoperation. It is serious. It is not the same as catastrophic bleeding — intracranial haemorrhage, exsanguination, permanent neurological injury. Those outcomes from prophylactic-dose heparin in fit surgical patients are rare. Fatal bleeding at prophylactic dose is a fraction of the major-bleed numerator.
Pulmonary embolism is different. Symptomatic PE carries mortality often quoted around ten to fifteen per cent in hospitalised patients; massive PE kills quickly. The prevented event is frequently life-ending. The caused event, in most cases, is treatable morbidity.
| Prophylaxis harm | Untreated PE | |
| Typical event | Transfusion, haematoma, delay | Hypoxia, collapse, death |
| Catastrophic tail | Uncommon at prophylactic dose | Common enough to fear |
| Reversibility | Often | Sometimes, if caught |
Equal rates do not mean equal stakes. Harm reduction and life-saving logic favour prevention when thrombotic risk is genuine — even if the numerators look similar on a spreadsheet.
The escape hatch in guidelines is the low-risk patient: fully mobile, short stay, medical admission without immobility. In a contemporary UK acute NHS hospital, that patient is rare.
Medical beds are not filled by people walking the corridor awaiting a taxi. They are filled by hypoxia, sepsis, heart failure, delirium, and frailty collapse. Surgical beds hold patients who have had general anaesthesia — venous stasis, dehydration, tissue injury, inflammatory hypercoagulability — in a population that is older, heavier, and less mobile than a decade ago. Day-case volume has grown, but the biology of operation does not shrink because discharge is earlier.
When a risk assessment tool labels someone "low risk," it often reflects which boxes were ticked, not physiology. The same patient can score differently on different models; recommendations for prophylaxis have ranged from one in ten to nine in ten depending on the tool. Low risk on paper is a weak exception in practice.
For major orthopaedic surgery — hip fracture, arthroplasty, major trauma — the debate barely applies. VTE risk without prophylaxis is among the highest in surgery. The surgeon's question there is timing after neuraxial block and duration post-discharge, not whether clot risk exists.
Day-case surgery does not mean low thrombotic risk. Much hospital-associated VTE is diagnosed after discharge. Procedural catastrophic bleeding in standard day cases is uncommon. Day-case growth widens the surgical pool at risk; it does not create a legion of mobile low-risk patients.
Orthopaedic exceptions are specific: active bleeding, severe thrombocytopenia, neuraxial anaesthesia where prophylaxis must be delayed until safe — not withheld because VTE risk is doubted.
Across medical, surgical, and orthopaedic inpatient wards, the withhold reasons that survive scrutiny are a short list:
- active haemorrhage or high bleeding risk by NICE criteria
- neuraxial timing (start when safe)
- existing therapeutic anticoagulation
- defined contraindication (e.g. severe thrombocytopenia)
- palliative intent or informed refusal
Procedure-specific high bleeding risk in moderate general surgery — the category that fuels the one-per-cent debate — is uncommon. What trials count as extra bleed from LMWH is not the same as the operation being inherently catastrophic-bleed territory.
That is not a rejection of NICE. It is acute illness and reduced mobility logic applied directly. The impact weighting — fatal PE prevented versus major bleed caused — supports default even when rates look close.
What does not follow is that the mandatory VTE assessment form delivers this. England measured form completion, not correct prescribing, not administration, not post-discharge extension. Inter-hospital variation in appropriate prophylaxis ran from forty to one hundred per cent while assessment compliance sat near ninety-five per cent.
NICE published NG89 acknowledging the national risk assessment tool had not been validated for identifying who will develop VTE. The programme was industrialised nationally — CQUIN, contracts, dashboards — without operational efficacy data. Excellence is ward execution: protocol plus contemporaneous checking, not another tick-box.
Treat every inpatient as at risk until proven otherwise.
Default pharmacological prophylaxis for acute inpatients — unless a narrow withhold applies — is rational harm reduction. The VTE form was meant to help; it was never validated to prove it does. National policy counted the form. Patients needed the prescription.
