Improving stroke outcomes through progressive implementation of stroke unit care in Brazil: a longitudinal observational study

Introduction Stroke remains a leading cause of death and disability worldwide, with a disproportionate burden in low- and middle-income countries (LMICs). Although stroke unit (SU) care improves outcomes, evidence from LMIC settings is still limited. We evaluated the impact of three sequential stroke-care models implemented in a Brazilian public university hospital. Patients and methods This longitudinal observational study included 1,889 patients with ischemic or hemorrhagic stroke across three care models: before stroke unit (BSU), acute stroke unit (ASU), and comprehensive stroke unit (CSU). Demographic characteristics, stroke subtype, baseline severity, imaging metrics, and outcomes were collected using retrospective and prospective approaches. Primary outcomes were 90-day case fatality and functional status assessed by the modified Rankin scale (mRS). Secondary outcomes included door-to-CT time, thrombolysis rates, pneumonia, access to rehabilitation, and length of stay. Multivariable logistic regression was used to identify predictors of mortality and excellent functional outcome (mRS 0–1). Results The implementation of structured stroke-care models was associated with significant improvements in outcomes. Functional independence at 90 days (mRS 0–2) increased from 45.6% (BSU) to 60.3% (ASU) and 56.3% (CSU) ( p < 0.001), while case fatality declined from 24.3 to 10.2% and 7.7%, respectively ( p < 0.001). Key quality indicators improved substantially: mean door-to-CT time decreased from 527 to 170 and 107 min ( p < 0.001), thrombolysis rates increased from 0 to 14.1% and 22.2% ( p < 0.001), and post-stroke pneumonia rates declined from 29.4 to 16.3% and 12.5% ( p < 0.001). In multivariable analyses, older age and higher baseline NIHSS were independently associated with increased mortality, whereas intravenous thrombolysis and SU care were associated with reduced odds of death. SU care was associated with more than a threefold increase in the likelihood of excellent functional outcome, while thrombolysis remained the strongest predictor (OR 6.19). Conclusion Stepwise implementation of structured stroke-care models significantly reduced mortality and improved functional outcomes in this LMIC setting. These findings support the effectiveness and scalability of SU—particularly comprehensive models—as key strategies to strengthen stroke systems of care.