How does stroke risk change after quitting smoking—and does the type matter?

01

A blockage and a bleed are not two names for the same event

An ischaemic stroke involves an obstruction of blood flow to the brain. A haemorrhagic stroke involves a ruptured vessel; studies may separate intracerebral haemorrhage, within brain tissue, from subarachnoid haemorrhage, around the brain. Smoking-related vascular injury and clotting help explain why ending smoke exposure matters, but a clot-related explanation is not the whole explanation for every bleeding outcome.

This matters when a headline says ‘stroke risk falls’. It may count all types together or only ischaemic events. A combined result does not tell us the amount of change for each subtype. A study of new strokes also does not measure how much an existing brain injury has healed.

[1][3]

02

What three different results looked like in one real cohort

Moon and colleagues' 2023 analysis followed 719,040 Korean men selected from insured health-examination participants. People with a reported history of stroke or cancer were excluded. Compared with sustained smokers, men classified as long-term quitters had lower adjusted hazards of ischaemic stroke, intracerebral haemorrhage and subarachnoid haemorrhage. The table shows three estimates from the same analysis: they were favourable, but not identical.

‘Long-term’ meant reporting former smoking at both assessed visits, roughly four years apart; it was not daily verification of an exact quit date. Smoking status was self-reported, the analysis lacked detailed smoking amount and duration, and the population excluded women. These are associations after statistical adjustment, not random assignment to stop smoking.

There is also an instructive difference between crude and adjusted results. The long-term quitter group's unadjusted total-stroke rate was 229 per 100,000 person-years, versus 221 among sustained smokers; yet its adjusted hazard ratio was 0.62. The groups differed in age and other characteristics. Those two numbers do not contradict each other: the crude rate describes the observed group, while the model accounts for specified differences. Neither number is a personal risk estimate or proof that quitting alone caused the whole difference.

[2]

03

Why ‘not significantly different’ does not establish a reset date

Luo and colleagues' review, published in 2022 with searches through July 2021, combined 25 prospective cohorts. Compared with never smokers, former smokers did not show a clear pooled difference for total stroke, intracerebral haemorrhage or stroke mortality; small higher estimates remained for ischaemic stroke and subarachnoid haemorrhage. Thus even a broad ‘former smoker’ result did not mean every subtype was identical to never smoking.

The cohorts used different definitions of smoking status and follow-up periods, with substantial heterogeneity. The review converted several relative measures to odds ratios. Its duration curves were statistical models, not repeated scans showing an individual's vessels repaired. A confidence interval including no difference is uncertainty, not proof of exact equality. The review also lacked enough data to model cessation duration against mortality from each separate stroke subtype.

[1]

04

Risk reduction supports stopping; it does not replace stroke care

Use the evidence as a reason to end continued smoke exposure, not to stop prescribed care, skip follow-up or assume a previous stroke has healed. Questions about your vascular risk or an existing diagnosis belong with a qualified clinician. Suspected stroke requires immediate local emergency help, even if the signs have gone away; do not wait for a quit-smoking milestone or a counselling appointment.

For the separate task of stopping smoking, NHS routes can help people in England contact local support. They are not emergency stroke services. This page does not collect health histories, calculate personal stroke probabilities or choose medicines, doses or treatment.

[4][5][6]

05

What to keep in mind

• Stopping smoking has value without a personal normalization date.
• Stroke subtype, new events and stroke deaths are different endpoints.
• Neither a favourable model nor a quit anniversary replaces medical care.