Three decades of first-stroke records from south London show hypertension and diabetes prevalence rising fastest among Black African and Black Caribbean patients, with Black African patients having strokes 10 to 12 years earlier than white patients in the same register. [1][2]
The finding comes from the South London Stroke Register, a population-based cohort of 8,515 first-stroke patients followed from 1995 to 2024 by researchers at King's College London, published as a medRxiv preprint in January and presented at the European Stroke Organisation Conference. [2][3] The Guardian's exclusive frames an inequality alarm, and the university's release echoes the widening-inequalities language. [1][2] The frame is supported. The discipline around it comes from the study itself.
Three labels never come off this finding. It is a preprint — posted January 15, not peer reviewed, and not yet a journal article. It is a register cohort — one city's population-based surveillance, not national incidence, and nothing in it licenses an extrapolation to Britain as a whole. And it is observational — it establishes that the gap widened inside the register, not why. [3]
The preprint's own counterweight deserves the same paragraph as the alarm. Treatment rates for hypertension and diabetes were also higher among Black participants across the study period. [3] Prevalence, detection, and treatment are separate stages: a group can show faster-rising recorded risk factors because incidence is rising, because detection improved, because register capture improved — or because all three moved at once. The register cannot yet decompose the widening gap into those components, and neither can the coverage. [1][3] The higher treatment rates complicate any simple care-failure reading in both directions: more people were treated, and the risk factors kept climbing anyway.
That tension is where the science actually sits. Earlier onset — 10 to 12 years among Black African patients — is the register's most striking figure and its least explained. [2] It could reflect genuine biological and social exposure differences, differential survival into the register, migration patterns shifting the denominator, or some combination the authors have not yet separated. Converting it into a causation claim, in either direction, would outrun a January preprint that has not faced peer review. [3]
What the register unquestionably offers is duration. Thirty years of person-level records from one defined population is a rare asset in stroke epidemiology, and a dismissal that waved off the cohort as "only a preprint" would discard real evidence. The alarm frame's risk is the opposite: over-reading one city's surveillance into a national verdict. [1][3] The data discipline both impulses.
The unanswered questions have owners. Has the manuscript gone to a journal since January, and what does review change? How much of the widening gap is incidence versus detection? What do treatment rates, control rates, and outcomes show by group across the same three decades — because treatment without control is its own finding? [3] UK health-equity voices on X would have frames for all three, but both documented searches were rate-limited, and no community reading is recovered here. [1]
A register that watched one city for 30 years has earned careful reading. Careful reading means the alarm and the treatment ledger travel together, under a preprint label, until peer review says which parts hold.
-- KENJI NAKAMURA, Tokyo