Implementing Acute Stroke Services in Sub-Saharan Africa: Steps, Progress, and Perspectives from the Tanzania Stroke Project Open Access

Globally, stroke is the third leading cause of death and the fourth leading cause of disability-adjusted life years (DALYs), accounting for 7.3 million deaths and 160.5 million DALYs in 2021 [1]. Mortality and DALYs from stroke are projected to increase by 50% and 31% by 2050, with more than 90% of the deaths occurring in low- and middle-income countries, particularly in sub-Saharan Africa (SSA) [2].

Tanzania, a country in SSA with a population of 69 million, ranks stroke among its top ten leading causes of death [3]. Between 2004 and 2006, the age-adjusted incidence of stroke in urban areas of Tanzania was 315.9 per 100,000 person-years [4], compared to 118.7 per 100,000 in high-income countries (HICs) [5]. Stroke in Tanzania primarily affects individuals in their fourth to sixth decades of life and is associated with substantial morbidity and mortality [6].

Despite this high burden, Tanzania faces significant shortages in stroke care infrastructure [7], with only one active stroke registry, lack of dedicated stroke units, and most adults treated for stroke do not receive care that meets standard quality recommendations [7, 8]. To address these unmet needs, we established the Tanzania Stroke Project (TSP) in cooperation with the Tanzanian Ministry of Health, the World Stroke Organization (WSO), WSO-Future Stroke Leaders Program (WSO-FSLP), and Hospital Directorates. TSP highlights the baseline results from the stroke registries, details the training of healthcare workers (HCWs), and describes the establishment of stroke units.

The TSP was an implementation study that ran from August 2023 to August 2024 at two of the largest academic referral hospitals in Tanzania: the Muhimbili National Hospital-Mloganzila (MNH-M) in Dar es Salaam and Bugando Medical Center (BMC) in Mwanza. These study sites were chosen because they received official endorsement from the Tanzanian Ministry of Health, highlighting the urgent need to improve stroke care in the region, and the consent of the Hospital Directorates.

MNH-M is a branch of the main national hospital serving the entire population, while BMC is a large tertiary hospital that serves approximately 15 million people. MNH-M has a bed capacity of 608, while BMC has 1,080. Both hospitals are equipped with 24-h emergency medicine departments, intensive care units, laboratories, computed tomography scans, magnetic resonance imaging, and rehabilitation departments. They serve as the main referral hospitals for stroke patients. Staffing in the neurology unit at MNH-M includes 7 neurologists, 5 neurology residents, 2 internal medicine physicians, and 2 general medical doctors. BMC on the other hand is staffed by 1 internal medicine physician and 1 general medical doctor. Both hospitals lacked dedicated stroke units, trained multidisciplinary stroke teams, and access to acute stroke therapies (thrombolysis and mechanical thrombectomy), limiting stroke treatment to secondary prevention.

We prospectively collected de-identified data from consecutively admitted adults (≥18 years) to the medical wards presenting with the World Health Organization (WHO) clinical definition of stroke [9] at MNH-M and BMC between October 2023 and August 2024. HCWs were selected to participate in the training by the department heads, through the Directors of Medical Services, and included neurologists, neurosurgeons, neuroradiologists, radiologists, physicians, cardiologists, general medical doctors, emergency medicine physicians, intensive care unit intensivists, nurses, physiotherapists, speech therapists, occupational therapists, psychologists, psychiatrists, and nutritionists. The training was also made available to HCWs from other departments.

The intervention followed a 3-tier approach: (1) establishment of a stroke registry; (2) training of HCWs; and (3) the development of stroke protocols and establishment of stroke units.

Stroke experts from the WSO, in collaboration with neurologists from Tanzania, standardized the variables for the stroke registry. The registry design followed recommendations from the global register of stroke care quality (RES-Q, https://eso-stroke.org/projects/eso-east/registry-of-stroke-care-quality-res-q/), the Swedish Stroke Register, and the WHO stepwise stroke surveillance in low- and middle-income countries [10, 11] (online suppl. File 1; for all online suppl. material, see https://doi.org/10.1159/000545954).

Quality performance indicators were assessed through chart review, and elements included time from symptom onset to admission and imaging, medical complications, assessment of swallowing function, availability of physiotherapy, and administration of deep venous thrombosis prophylaxis. Compliance with recent intracerebral hemorrhage (ICH) bundled care was reviewed; this included blood pressure (BP) <140/90 mm Hg, temperature <37.5°C, blood glucose levels of 6.1–7.8 mmol/L for non-diabetics and 7.8–10 mmol/L for diabetics on arrival [12], and secondary prevention medication. The registries were hosted on the web-based software (REDCap, Vanderbilt University, Nashville, TN), and designated general medical doctors in the neurology unit were responsible for daily patient registration under the supervision of a senior consultant.

Collaborators from WSO developed training modules based on the ESO Stroke Guidelines: https://eso-stroke.org/guidelines/eso-guideline-directory/, AHA/ASA [13, 14], and the Angel’s initiative: https://www.angels-initiative.com/). The modules covered 31 topics on core competencies in stroke, focusing on hyperacute, acute, and post-acute care, as well as specialized nursing tasks and rehabilitation (online suppl. File 2). Training was delivered in English language and conducted virtually. At MNH-M, sessions were held twice a week for 1 h from January to March 2024, while at BMC, daily 1-h sessions were held from Monday to Friday over a 3-week period in June 2024. This was followed by an extensive 1-day onsite mentorship training in August 2024, reviewing key aspects of stroke unit care and incorporating case-based simulations.

Three questionnaires were developed for pre- and post-training assessments: (1) hyperacute, acute, and post-acute care module (20 multiple-choice questions, total score: 20 points), (2) nursing module (10 multiple-choice questions, total score: 10 points), (3) rehabilitation module (10 questions with single and multiple-choice answers, total score: 20 points) (online suppl. File 3). Participants were required to complete both the pre- and post-assessments and attend at least 70% of the module topics to successfully complete the training.

Stroke unit protocols were developed by WSO collaborators and site neurologists, using the WSO Roadmap for essential stroke services and other international guidelines [13, 14], adapted to the local context based on experiences from stroke unit initiatives in Nepal, Zambia, and Ethiopia [15‒17] (online suppl. File 4). Stroke units were established by designating 8 beds within the general medical ward for each facility. The units were equipped with cardiac monitors, digital BP machines, glucometers, and thermometers. The stroke units started their services after completing the training in August 2024.

Continuous variables were summarized and presented as means and standard deviations for normally distributed data and medians with interquartile ranges for nonparametric data with comparisons made using the Wilcoxon rank-sum test by stroke subtype. Clinical characteristics and outcomes were summarized as proportions, and differences by stroke subtype were assessed using chi-square tests or Fisher’s exact tests, with significance level set as p value <0.05.

There were 526 patients registered between October 2023 and August 2024, with 290 patients from MNH-M and 236 from BMC. Sixty-six patients were excluded due to missing brain imaging findings, leaving 460 patients in the final analysis, of whom 66% (303/460) had complete patient information.

Among all patients, 58% (269/460) had hemorrhagic stroke, and of these, 94% (253/269) had ICH, 4% (12/269) subarachnoid hemorrhage, and 2% (4/269) isolated intraventricular hemorrhage. The mean National Institutes of Health Stroke Scale (NIHSS) at presentation was 18 ± 8. Compared to ischemic stroke patients (n = 191), patients with hemorrhagic stroke were younger (57 ± 14 years vs. 65 ± 15 years, p < 0.001), more likely to have been referred from another healthcare facility (86% [229/269] vs. 63% [120/191], p < 0.001), and had higher prevalence of premorbid hypertension (86% [231/269] vs. 74% [142/191], p = 0.002) compared to those with ischemic stroke, Table 1. Overall, in-hospital mortality was 27% (93/340) and was significantly higher in adults with hemorrhagic stroke compared to ischemic stroke (33% vs. 21%).

The median door to imaging time was 0 days interquartile range (0, 1). Swallow screening was documented in only 11% (n = 50/452) of patients. The most common medical complication was aspiration pneumonia 20% (90/460). Few patients achieved ICH bundle of care targets for BP and glycemic control 11% (28/253) and 47% (99/213) respectively, Table 2.

For the main training module, 82 HCWs registered and 43.9% (36/82) completed the modules for both study sites. The nursing module was attended by 27 HCWs with a completion rate of 59.3% (16/27), and the rehabilitation module was attended by 58 HCWs with a completion rate of 44.8% (26/58). The number of participants was higher in BMC in the majority of the modules compared to MNH-M (main module: 51 vs. 31; nursing: 13 vs. 14; rehabilitation 27 vs. 31, respectively), Table 3. Notably, the completion rate was significantly higher in BMC compared to MNH-M: main module – 57% (29/51) at BMC versus 23% (7/31) at MNH-M; nursing – 85% (11/13) at BMC versus 36% (5/14) at MNH-M; rehabilitation module – 71% (22/31) at BMC versus 15% (4/27) at MNH-M.

HCWs at BMC showed significant improvement in their post-assessment quiz over all modules, with no improvement at MNH-M. In the main module, HCWs at BMC demonstrated improved performance in the post-assessment, with mean scores of 18 ± 3 compared to 12 ± 4 in the pre-assessment (p < 0.0001), with no improvement at MNH-M (post-assessment, mean score 15 ± 4 vs. 12 ± 4 pre-assessment mean score, p = 0.1842).

In the nursing module, HCWs at BMC demonstrated improved performance in the post-assessment, with mean scores of 9 ± 1 compared to 6 ± 2 in the pre-assessment (p < 0.0001), with no improvement at MNH-M (post-assessment mean score 8 ± 1 vs. 7 ± 6 pre-assessment mean score, p = 0.7227). The same was observed for the rehabilitation module, where BMC HCWs showed improved post-assessment mean scores of 18 ± 4 versus 8 ± 6 mean pre-assessment score, p = 0.0001, with no improvement at MNH-M (post-assessment mean score 13 ± 4 vs. 10 ± 5 pre-assessment mean score, p = 0.7227).

The TSP was the first initiative to improve acute stroke services in a resource-limited healthcare setting in Tanzania. Within 1 year, two large academic referral hospitals successfully established their first stroke units using a three-tier implementation approach. The successful establishment of a stroke registry and the effective training of HCWs in our study provide a scalable framework that can be adapted to similar resource-limited healthcare settings. These experiences highlight the potential for replicable strategies to improve stroke care delivery and foster the development of standardized protocols and capacity building in underserved regions.

We successfully established comprehensive stroke registries at the two sites, which were crucial for creating baseline measurements for the acute stroke units. Challenges included missing data with only two-thirds of registered patients having complete information. These challenges are consistent with those reported in the Nepal Stroke Project [16]. Our data collection sheet comprised of 100 items, acknowledging that not all fields may be equally relevant for outcome improvement and system development. In response, the project team is refining the registry variables to prioritize the most critical fields, thereby reducing the rate of missing data. Additionally, systems are being developed to identify missing fields early, generate weekly quality assurance reports, and engage hospital directors to allocate dedicated time for registry management. Importantly, this registry has laid the groundwork for the development of a National Stroke Registry in Tanzania.

We observed a high proportion of hemorrhagic strokes (58% of all included stroke patients), particularly ICH (94%), occurring in younger age groups (mean age 57 ± 14 years) and high prevalence of preexisting and undertreated hypertension (86% and 90%, respectively), similar to other studies conducted in SSA [18, 19]. In contrast, stroke in HICs has been reported to occur predominantly in people over 70 years of age, with only 10–15% of patients suffering hemorrhagic stroke [20]. Hypertension remains a leading global risk factor for stroke and is a key driver for stroke in SSA [21]. More than 45% of adults in Africa over the age of 25 years are estimated to have hypertension, with less than half of those with hypertension aware of their BP status, and at least half of those diagnosed do not receiving treatment [22]. This highlights the urgent need to promote primary preventive strategies for early detection, treatment, and control of hypertension. Additionally, further investigations are warranted to better understand the etiologies of hypertension in this younger adult population. Other risk factors for stroke identified in our cohort included a history of previous stroke (13%) and diabetes mellitus (10%). A small proportion of patients (less than 5%) had HIV infection, cardiac diseases (atrial fibrillation and heart failure), and history of smoking and alcohol consumption.

It is notable that only a small proportion of patients with ICH achieved targets for BP and glycemic control as recently recommended by the INTERACT-3, ICH bundle of care [12]. This likely contributed to the high in-hospital mortality rate observed (27%), which was significantly higher in adults with hemorrhagic stroke compared to ischemic stroke (33% vs. 21%, respectively), emphasizing the need for immediate intervention. Our study population was characterized by severe clinical presentations, with a median NIHSS score of 18. We hypothesize that the high incidence of ICH and stroke severity observed may be attributed to the fact that only the most critical cases are referred to the specialized hospitals.

Over two-thirds (75%) of adults in this cohort were referred from lower level healthcare facilities, with over 50% of adults arriving after 24 h after the onset of stroke symptom. This delay underscores the importance of community and HCW awareness and the need to strengthen prehospital referral systems to ensure rapid transfer of stroke cases to stroke-ready centers in Tanzania. Based on these data, a stroke awareness campaign using “UPESI,” the Swahili translation of the FAST acronym for community sensitization [23], is being implemented. This initiative aims to improve early recognition of stroke symptoms and promote timely medical intervention, ultimately reducing delays in seeking and receiving appropriate care.

Beyond delayed presentation, the high in-hospital mortality rate also reflects gaps in the quality of stroke care. Limited adherence to essential care practices, such as swallowing screening, was evident, with more than two-thirds of patients (86%) lacking a documented swallow screen assessment. This likely contributed to the high incidence of aspiration pneumonia (20%). Similarly, over one-third of patients did not receive speech and language therapy due to the unavailability of a speech therapist, while the majority received physiotherapy only after 48 h of arrival. These delays in poststroke rehabilitation can lead to prolonged functional recovery and negatively impact long-term patient outcomes. Given that strokes primarily affect young adults in this setting, the long-term socioeconomic consequences have implications beyond individual patients, potentially influencing both the family and nation’s economy.

Robust training modules were provided by WSO collaborators to our local teams focusing on core stroke management skills. While the training aimed to provide comprehensive stroke knowledge, several limitations were noted: there were notable differences in training completion rates between the two hospitals. Similarly, HCWs at the BMC had better scores on the post-assessment quizzes compared to MNH-M. This is likely because MNH-M was the first study site to initiate the trainings, which were open to all HCWs without the identification of focal persons to form the multidisciplinary teams. Similarly, the trainings in MNH-M lasted longer (about 3 months), which increased the likelihood of dropout rates and nonadherence. Conversely, BMC identified focal HCWs to attend the training that was shorter in duration (3 weeks), which likely increased attendance rates and therefore knowledge acquisition, which was reflected in the post-evaluation quizzes. Another plausible explanation for the higher benefit of the latter BMC cohort is that the WSO-FSLP trainers adapted their lecture content to align more closely with the specific needs of Tanzanian HCWs following the initial training session at MNH-M. Consequently, the subsequent training sessions conducted at BMC were more effectively tailored to address the participants’ requirements, enhancing their relevance and impact. Our findings support the need for continued training of HCWs at least twice a year, with shorter intense duration. However, it has to be assessed if the shorter training duration will still lead to long-lasting gain in knowledge.

Our study represents a significant milestone in advancing the understanding of stroke care system development within a resource-constrained healthcare setting. It has made a substantial contribution to enhancing quality monitoring, expanding knowledge, and promoting the standardization of stroke care in Tanzania. The involvement of international stroke experts from diverse healthcare environments ensured the development of high-quality training modules and protocols. Nonetheless, our study is not without its limitations: a limitation of the TSP was that it involved only two sites. Similarly, the high rates of high hemorrhagic strokes observed are contrary to the previous population-level data [24] and likely influenced by selective admission, where patients with more severe disease are likely to be referred and seek hospital care, hence limiting generalizability of the results.

Furthermore, the stroke unit protocols and algorithms were primarily derived from HIC recommendations and may not always be practical in a resource-limited setting. Missing data and the relatively short study duration also pose challenges in drawing definitive conclusions regarding the impact of KPIs on patient outcomes.

Future implementation strategies will focus on tailoring standard operating procedures to better align with the local healthcare context to enhance feasibility and effectiveness in Tanzania. Additionally, efforts will include comparing patient outcomes pre- and post-training, assessing HCWs’ acceptance and satisfaction with the model.

This was the first initiative to implement acute stroke services in two large academic centers in Tanzania. Few patients received and achieved standard of care treatment targets as recommended by the WSO Roadmap. Our findings support the need to strengthen Tanzania’s healthcare systems to improve screening, treatment, and control of hypertension to prevent complications such as stroke. Additionally, the TSP highlights the need for multilevel stakeholder involvement using a three-tier approach in countries beginning to establish stroke services. Continuous monitoring of care quality, HCW training, and community sensitization are essential for sustained improvements in stroke care delivery.

Our gratitude goes to hospital Directors, Professor Mohamed Janabi and Dr. Fabian Massaga, for supporting this initiative and thanks to Professor. Deanna Saylor and Dr. Omary Ubuguyu for their invaluable inputs to the project. RP is supported by a grant from the National Heart, Lung and Blood Institute (K24 HL118107; R01 HL160332). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

This study was performed in accordance with the Declaration of Helsinki. This human study was approved by National Institute of Medical Research, approval: NIMR/HQ/R.8a/Vol.IX/4555. Written informed consent to take part in the study was obtained from patients and from patient’s next of kin, where they were unable to provide written informed consent.

Ms. Menglu Ouyang, Dr. Emily Ramage, and Professor Craig Anderson were members of the journal’s Editorial Board at the time of submission. The remaining authors have no conflicts of interest to declare.

This study was funded by the World Stroke Organization-Future Stroke Leaders Program.

Sarah Shali Matuja: conceptualization, healthcare worker training, stroke registry and protocol development, data collection, analysis, interpretation, and writing the initial manuscript. Christine Tunkl, Tamer Roushdy, Linxin Li, Menglu Ouyang, and Faddi G. Saleh Velez: training module development, healthcare worker training, stroke registry and stroke unit protocol development, and writing the initial manuscript. Meron Gebrewold, Jatinder S Minhas, Zhe Kang Law, Aristeidis H. Katsanos, Teresa Ullberg, Maria Giulia Mosconi, Maria Khan, Matias Alet, Radhika Lotlikar, Alicia Richardson, Bogdan Ciopleias, Mirjam R Heldner, Susanna Maria Zuurbier, Emily Ramage, Selam K Kifelew, Vasileios Lioutas, Marika Demers, Marina Charalambous, Dorcas BC Gandhi, Urvashy Gopaul, Leonardo Carbonera, and Ralph Akyea: training module development, healthcare worker training, stroke registry and stroke unit protocol development, and critically reviewing and revising the manuscript. Ladius Rudovick: conceptualization, data collection, and critically reviewing and revising the manuscript. Bahati Wajanga, Semvua Kilonzo, Robert Peck, Paschal Ruggajo, Tumaini Nagu, and William Matuja: conceptualization, and critically reviewing and revising the manuscript. Mohamed Mnacho and Faraja S Chiwanga: conceptualization, training, and stroke registry and protocol development. Brighton Mushengezi, Kigocha Okeng’o, Henrika Kimambo, and Mohamed Manji: training, stroke registry variables and protocol development, and data collection. Akili Mawazo: data analysis and interpretation. Louise Johnson, Octávio Pontes-Neto, Craig Anderson, and Sheila Ouriques Martins: conceptualization, developing the training modules and stroke unit protocols, training healthcare workers, and critically reviewing and revising the manuscript. All authors read and approved the final manuscript.

The data that support the findings of this study are not publicly available to maintain patient confidentiality but are available from the corresponding author upon reasonable request.