Design: Retrospective cohort with linked primary care, hospital and death data
Population: 476,620 knee, hip and shoulder arthroplasties in the UK, 2007 to 2019
Primary outcome: Venous thromboembolism within 6 months
Effect: 1.5% overall (knee 1.6%, hip 1.4%, shoulder 1.0%); over 30% found only in linked data

A retrospective cohort in The Bone & Joint Journal (1 September 2026) used the UK's CPRD Aurum primary care database linked to hospital and death records to follow 476,620 knee, hip and shoulder arthroplasties performed from 2007 to 2019. The outcome was venous thromboembolism within six months.

The six-month rate was 1.5%: 1.6% after knee, 1.4% after hip and 1.0% after shoulder replacement. More than 30% of events were picked up only through the linked primary care or death data. Rates fell over time after hip and knee replacement, but the successive NICE guidelines did not produce a measurable step change; after shoulder replacement, rates fell by 9% following the 2018 guideline. Risk also varied by season.

The main lesson is about measurement. A unit counting clots only from its own readmissions will undercount. The six-month window also captures clots well after the usual prophylaxis course has ended.

Tell patients that clot risk lasts weeks after discharge, and what symptoms to report.
Use a follow-up method that captures clots treated outside your hospital if you audit VTE rates.
Complete the full prescribed course of thromboprophylaxis after discharge.
Include shoulder arthroplasty in your thromboprophylaxis risk assessment.

Why it matters

Unit audits that rely on readmission data are likely to underestimate clot rates after arthroplasty.

The statistics, in plain English

A rate of 1.5% means about 15 clots per 1,000 operations. Because the study was observational, it can show that rates fell over time but cannot separate the effect of guidelines from other changes in practice.