| Reference | Study type | Population | Intervention and comparison | Outcomes | Risk of bias |
|---|---|---|---|---|---|
| RCT=randomized controlled trial; SR=systematic review; MA=meta-analysis | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | MA |
Patients with prior stroke or transient ischemic attack from 10 RCTs across multiple countries | More intensive vs less intensive blood pressure lowering | Primary outcome: Recurrent stroke; Secondary outcomes: major cardiovascular events, ischemic stroke, hemorrhagic stroke, fatal/disabling stroke, myocardial infarction, death from cardiovascular causes, death from any cause, heart failure |
Moderate |
| Reference | Comments |
|---|---|
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | This meta-analysis investigated whether more intensive blood pressure (BP) lowering
is more effective than less intensive treatment in preventing recurrent strokes among
patients with a history of stroke or transient ischemic attack (TIA). The study synthesized
data from ten randomized controlled trials published between 1980 and 2022, encompassing
a total of 40,710 patients. Eligible studies were randomized trials that included patients with prior stroke or TIA, compared more intensive versus less intensive BP-lowering strategies—either through antihypertensive drugs versus placebo or lower versus higher BP targets—and reported recurrent stroke as an outcome. Studies were required to report the magnitude of systolic BP reduction between groups and have a minimum treatment duration of one year. Trials were excluded if they were published before 1980, used only non-pharmacological interventions, compared different antihypertensive drugs, had paradoxical BP outcomes (i.e., higher BP in the intensive group), enrolled patients within three days of stroke, or included a significant proportion of patients with end-stage kidney disease. Among the 10 included trials, 6 compared antihypertensive drug(s) to placebo or no antihypertensive therapy, and 4 trials compared a lower BP target to higher BP target. The mean baseline BP was 146 mmHg and diastolic BP was 85 mmHg across all trials. The magnitude of BP reduction varied across studies. On average, the differential SBP reduction between the more intensive and less intensive groups was 6.7 mmHg, and the differential diastolic BP reduction was 2.8 mmHg. The meta-regression analysis revealed a log-linear relationship between the magnitude of BP reduction and the risk of recurrent stroke. For example, a 5 mmHg greater systolic BP reduction was associated with an RR of 0.90, and a 10 mmHg reduction with an RR of 0.67. Similarly, a 3 mmHg greater diastolic BP reduction was associated with an RR of 0.84, and a 5 mmHg reduction with an RR of 0.60. These findings suggest that greater BP reductions are associated with proportionally greater reductions in stroke risk. For the primary outcome, recurrent stroke, number needed to treat (NNT) is 58 in 3 years. For major cardiovascular events, NNT is 61 in 3 years. The quality of the included studies was assessed using the Cochrane Risk of Bias tool. All studies were randomized. However, allocation concealment and blinding varied. Some studies were double-blinded (particularly those comparing drugs to placebo), while others—especially those comparing BP targets—were open-label, introducing potential performance bias. Blinding of outcome assessment was inconsistently reported, which may have led to detection bias, particularly in trials where stroke was not the primary outcome. Attrition bias appeared minimal, as sensitivity analyses showed that removing individual trials did not significantly alter the overall results. Selective outcome reporting was not evident, and the primary and secondary outcomes were consistently documented across studies. Publication bias was assessed using the trim-and-fill method, and no major asymmetry was found, suggesting low risk of bias from unpublished negative studies. The analysis was conducted at the trial level, not the individual patient level, which limited the ability to explore subgroups such as patients with baseline systolic BP below 140 mmHg. There was substantial heterogeneity among the included trials, likely due to differences in BP reduction magnitude and study design. Additionally, two large trials (PRoFESS and PROGRESS) contributed more than half of the total patient population, which may have disproportionately influenced the pooled results. The definition of BP reduction varied across studies, with some reporting mean differences over time and others using single time-point comparisons. Target BPs in the intervention group were also defined differently between those included studies that compared intensive vs. less intensive BP targets, e.g. systolic BP <120 mmHg; systolic BP <125 mmgHg; systolic BP <130 mmHg; systolic BP <130 mmHg or a 10 mmHg reduction if baseline systolic pressure was <140 mmHg mmHg; or BP 120/80 mmHg. Furthermore, in several trials, stroke was not the primary endpoint, which could affect the accuracy of event reporting. |
Results
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I=intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 10 studies; patients (I: 20344 / C: 20366) | Mean 2.8 years (range 1–4 years) | 1704 (8.4%) | 2061 (10.1%) | RR: 0.83 (95% CI: 0.78–0.88) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I=intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 9 studies; I: 19582 / C: 19608 | Mean 2.8 years (range 1–4 years) | 2348 (12.0%) | 2679 (13.7%) | RR: 0.88 (95% CI: 0.83–0.92) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I= intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 6 studies; I: 16134 / C: 16189 | Mean 2.8 years (range 1–4 years) | 1217 (7.5%) | 1403 (8.7%) | RR: 0.88 (95% CI: 0.83–0.92) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I= intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 6 studies; I: 16134 / C: 16189 | Mean 2.8 years (range 1–4 years) | 114 (0.7%) | 212 (1.3%) | RR: 0.54 (95% CI: 0.43–0.68) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I= intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 6 studies; I: 8537 / C: 8539 | Mean 2.8 years (range 1–4 years) | 253 (3.0%) | 334 (3.9%) | RR: 0.76 (95% CI: 0.64–0.89) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I=intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 9 studies; I: 20344 / C: 20366 | Mean 2.8 years (range 1–4 years) | 365 (1.8%) | 409 (2.0%) | RR: 0.89 (0.78–1.03) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I= intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 9 studies; I: 19711 / C: 19736 | Mean 2.8 years (range 1–4 years) | 629 (3.2%) | 728 (3.7%) | RR: 0.86 (0.78–0.96) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I=intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 10 studies; I: 20344 / C: 20366 | Mean 2.8 years (range 1–4 years) | 1511 (7.4%) | 1554 (7.6%) | RR: 0.97 (95% CI: 0.91–1.04) |
| Level of evidence: moderate | |||||
| Reference | Number of studies and number of patients (I/C) | Follow-up time | Absolute number of events (%) I | Absolute number of events (%) C | Relative effect (95% CI) |
|---|---|---|---|---|---|
| I= intervention; C=comparison; CI=confidence interval; RR=relative risk | |||||
| «Hsu CY, Saver JL, Ovbiagele B, ym. Association Bet...»1 | 2 studies; I: 10779 / C: 10816 | Mean 2.8 years (range 1–4 years) | 126 (1.2%) | 120 (1.1%) | 1.05 (0.82–1.35) |
| Level of evidence: low | |||||