Research in Practice · Source reviewed: The complete 12-page NEJM article, published on 7 November 2026. The separate supplementary appendix and protocol were not supplied; details available only in those documents have not been independently checked.
What question did the trial address?
OPTIMAL-STROKE examined whether a lower systolic blood-pressure target could improve cardiovascular outcomes in people with elevated blood pressure and a previous ischaemic stroke or transient ischaemic attack (TIA). The comparison was a treatment strategy targeting below 120 mmHg versus one targeting below 140 mmHg.
The distinction matters: the study compared target-based strategies, rather than a particular antihypertensive drug. Its findings concern participants with previous cerebrovascular events. Participants had to be clinically stable for at least 48 hours before enrolment. The median interval since the qualifying event was 1.1 years, and 71% were enrolled at least 120 days afterwards. These findings must not be applied as an acute-stroke or immediate post-thrombectomy BP-lowering protocol. Methods and Discussion, pp. 2, 4 and 9
How was the study conducted?
The multicentre, open-label randomised trial, with blinded outcome assessment, recruited 4,368 participants at 30 sites in Brazil. Of these, 2,172 were assigned to intensive treatment and 2,196 to standard treatment. Mean age was 63.9 years; 47.9% were women. Median follow-up was 3.9 years. Original paper
| Study feature | Verified detail |
|---|---|
| Population | Elevated BP and previous ischaemic stroke or TIA |
| Design | Multicentre, open-label randomised trial; blinded outcome assessment |
| Intensive strategy | Systolic BP target <120 mmHg |
| Standard strategy | Systolic BP target <140 mmHg |
| Setting and sample | 30 Brazilian sites; 4,368 participants |
| Follow-up | Median 3.9 years |
Adults were eligible with systolic BP 130–180 mmHg. Key exclusions included modified Rankin score ≥4, known secondary hypertension and eGFR <20 mL/min/1.73 m². Randomisation was central and web-based, stratified by site with variable block sizes. An independent committee unaware of assignment adjudicated outcomes. Treatment was adjusted using prespecified algorithms; extra monthly visits were arranged as needed in the intensive group. Methods, p. 2
The primary outcome combined death from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke, and hospitalisation for unstable angina or heart failure. This is a broader cardiovascular endpoint than recurrent stroke alone. A favourable composite result does not mean that every individual component necessarily improved.
What were the main results?
A primary-outcome event occurred in 236 participants (10.9%) assigned to intensive treatment and 297 (13.5%) assigned to standard treatment. The hazard ratio was 0.79 (95% CI 0.67–0.94; P=0.007). Event rates were 3.25 versus 4.11 per 100 person-years. Original paper
| Measure | Intensive versus standard treatment |
|---|---|
| Participants with a primary event | 236/2,172 versus 297/2,196 |
| Observed event proportions | 10.9% versus 13.5% |
| Primary-outcome hazard ratio | 0.79; 95% CI 0.67–0.94 |
| Mean achieved systolic BP* | 122.0 versus 135.2 mmHg |
| Composite renal outcome | 3.9% versus 2.2%; HR 1.76 (95% CI 1.24–2.50)* |
| Serious adverse events overall | 15.1% versus 15.3% |
*Renal outcome estimates are descriptive: formal hierarchical testing stopped before this endpoint; its confidence interval was not adjusted for multiplicity.
Achieved BP figures exclude the first three months of follow-up. The mean between-group difference was 13.1 mmHg (95% CI 12.5–13.8).
The hazard ratio corresponds to a 21% lower estimated hazard of the composite outcome during follow-up. It is a relative time-to-event measure, not a 21-percentage-point absolute reduction. The observed proportions differed by 2.6 percentage points, calculated from the rounded percentages; that arithmetic difference should not be mistaken for a formally estimated risk difference at a fixed time point.
Which secondary findings are confirmatory?
The key secondary composite of cardiovascular death, nonfatal MI or nonfatal stroke occurred in 9.8% versus 11.8% (HR 0.81; 95% CI 0.68–0.98; P=0.03). The next composite, which also included all-cause death, was favourable. Cardiovascular death itself was not significantly different (4.6% versus 4.7%; HR 0.96; 95% CI 0.73–1.26), so formal testing stopped there. Subsequent outcomes cannot be called statistically significant on the basis of their unadjusted intervals. Table 2, p. 7
Stroke occurred in 4.7% versus 6.4% (HR 0.73; 95% CI 0.56–0.94), but this is supportive rather than confirmatory evidence. The much smaller TIA subgroup had an imprecise estimate (HR 1.34; 95% CI 0.68–2.64); the overall result does not establish benefit specifically for TIA alone. Subgroup intervals were not multiplicity-adjusted. Results, p. 6; Figure 3, p. 10
Did participants actually achieve the assigned targets?
Targets and achieved pressures should be distinguished. Although the intensive strategy aimed below 120 mmHg, its mean systolic pressure after the initial three months was 122.0 mmHg; the median was 119.6 mmHg. Corresponding values with standard treatment were 135.2 and 135.0 mmHg.
The result therefore describes assignment to the intensive treatment strategy, with the pressures achieved in this trial. It does not demonstrate that every participant must maintain a reading below 120 mmHg to obtain benefit, or that an isolated clinic reading can reproduce the trial's approach.
What is the renal safety signal?
The composite renal outcome was more frequent with intensive treatment. It comprised a sustained reduction of at least 50% in estimated glomerular filtration rate or progression to end-stage renal disease. There were 84 events (3.9%) versus 49 (2.2%); HR 1.76 (95% CI 1.24–2.50). Most were sustained eGFR reductions, while long-term dialysis occurred in six versus three participants. Analyses of individual renal components were post hoc. The authors note that baseline-relative eGFR changes cannot distinguish an initial haemodynamic fall from progressive functional loss; reversibility and long-term significance remain uncertain. These findings require attention but are not proof of irreversible kidney injury. Table 2 and Discussion, pp. 7–9
Overall serious adverse events occurred in 15.1% versus 15.3%. Sodium <130 mmol/L was recorded in 3.3% versus 2.2% (HR 1.53; 95% CI 1.07–2.21); these safety intervals were not multiplicity-adjusted. Safety, p. 8 That broad comparison does not negate the specific renal finding or establish equivalence for every adverse event. Cardiovascular benefit and the renal signal need to be read together.
How should clinicians interpret this result?
The trial supports consideration of a more intensive BP-control strategy in the population studied, while retaining attention to renal safety and individual tolerability. This is editorial interpretation, not a new guideline recommendation or a universal prescription.
The authors support considering lower targets in clinically stable patients while monitoring kidney function and electrolytes as treatment is intensified. The evidence concerns a supervised treatment strategy, rather than an instruction for patients to adjust medication themselves.
What limits confidence and generalisability?
All participants were recruited in Brazil and were younger than many stroke cohorts; severe disability and very low eGFR were excluded. The open-label design could introduce bias, although independent adjudicators were blinded. About 17% of visits were remote in both groups; 3.6% of participants were lost to follow-up and 0.4% withdrew consent. The main paper reports similar results in analyses addressing incomplete follow-up.
Analyses followed intention-to-treat. The trial initially planned approximately 7,000 participants; blinded aggregate reviews revised this to 4,060 participants and 689 primary events, then lowered the event threshold to 495 after recruitment. Actual enrolment was 4,368. These changes did not use treatment assignments or treatment-effect estimates, but the authors acknowledge reduced power for modest effects. The primary Cox model included treatment as a fixed effect and site as a random effect. Statistical Analysis, pp. 3–4; Limitations, p. 9
Clinical takeaway
In OPTIMAL-STROKE, targeting systolic BP below 120 rather than below 140 mmHg reduced the primary composite cardiovascular outcome after previous ischaemic stroke or TIA over a median 3.9 years. The renal outcome signal qualifies the benefit. Use the study to inform an individualised discussion, and retain caution in populations excluded from the trial and avoid extrapolation to acute stroke care.
Original research
Sampaio Silva G, Machline-Carrion MJ, Santo K, et al., for the OPTIMAL-STROKE Trial Investigators. Intensive Blood-Pressure Control in Patients with Previous Ischemic Stroke. New England Journal of Medicine. Published online 7 November 2026. doi:10.1056/NEJMoa2612924 . Trial registration: NCT04036409. Funding: Hospital Israelita Albert Einstein with approval and support from the Brazilian Ministry of Health. Servier donated fixed-dose combinations to both groups and had no other trial role, according to the paper. Individual disclosure forms were not supplied.
Cover: AI-generated conceptual clinical still life.