Burden, Clinical Outcomes and Predictors of Stroke in Hospital Mortality among Adult Patients Admitted to Stroke Unit of Jimma University Medical Center: Prospective Cohort Study

Background: Global burden of stroke epidemiology is changing rapidly. Over the 1990–2013 period, there was a significant increase in the absolute number of deaths and incident events of stroke. The burden of ischemic and hemorrhagic stroke varies between regions and over time in Ethiopia. The paucity of data has limited research output and consequently the response to this burden in our country. Methods: Prospective cohort study was carried at stroke unit of Jimma University Medical Center (JUMC) from March 10- July 10, 2017. The outcome of interest was mortality and time to death. Data was analyzed using SPSS version 20. Multivariable Cox regression was used to identify the predictors of in hospital mortality and time to death from hospital arrival. Predictor variables with P< 0.05 were considered statistically significant. Results: A total of 116 eligible stroke patients were followed with the mean age of 55.1+14.0 years and males comprised of 73 (62.9%) with male: female ratio of 1.70:1. Stroke accounted for 16.5 % of total medical admissions and 23.6 % of the total cases of in hospital mortality. A total of 91 (78.4%) of patients were discharged being alive making in hospital mortality of rate of 25 (21.6%). The median time of in hospital mortality after admission and length of hospital stay of the patients was 4.38 days and 9.21 days, respectively. The prominent suspected immediate cause for in hospital mortality was increased intracranial pressure 17 (68.0%) followed by respiratory failure secondary to aspiration pneumonia 11 (44.0%). Brain edema (AHR: 6.27, 95% CI: 2.50-15.76), urine incontinence (AHR: 3.48, 95% CI: 1.48-8.17), National Institute of Health Stroke Scale (NIHSS) >13 during hospital arrival (AHR: 22.58, 95% CI: 2.95-172.56) and diagnosis of stroke clinically alone (AHR: 4.96, 95% CI: 1.96-12.54) were the independent predictors of in hospital mortality. Conclusions: The mortality of stroke in this set up was similar to other low-and middle-resource countries.


Background
Global burden of stroke epidemiology is changing rapidly [ 1]. As heart disease and stroke statistics of 2016 report from American Heart Association 3 (AHA), stroke was the second-leading cause of death behind heart disease in 2013, accounting for 11.8% of total deaths globally [ 2]. First-time incidence of stroke occurs almost 17 million times a year worldwide; one every two seconds [ 3]. It is an important disease worldwide, constituting a big burden on the public health purse as well as on patients and their relative [ 4,5]. Stroke is a devastating and disabling cerebrovascular disease with significant amount of residual deficit leading to economic loss [ 6,7]. The burden of stroke is high and is not only attributable to its high mortality but also its consequent high morbidity [7][8][9]. One in six people worldwide will have a stroke in their lifetime [ 5]. Patients with stroke of under the age of 50 years, account for 5-10% of all stroke worldwide [ 10].
The global burden of disease (GBD) study also indicated that, 80% of stroke deaths occur in low and middle income countries (LMICs) [ 4], showing that the developing world carries the highest burden of stroke mortality and stroke-related disability [ 11]. The poor are increasingly affected by stroke, because of both the changing population exposures to risk factors and inability to afford the high cost of stroke care [ 5]. Moreover it remains uncertain if increased urbanization and life expectancy in some parts of sub-Saharan African nations will shift the region to higher cardiovascular diseases burden in future years [ 12].
The burden of ischemic and hemorrhagic stroke varies in Ethiopia between regions and over time [ 4 13]. Most deaths occurred early after admission due to stroke related acute complications with respiratory failure. As patients usually present late and the standard of care is poor compared to hospitals in developed countries, the in hospital mortality is expected to be higher [ 9]. Thus patients with stroke are often poorly managed and discharged from hospital without receiving adequate rehabilitation services. This has a series implication in terms of saving the life of patients especially with hemorrhagic stokes which are characterized by sever neurologic complications [ 4]. Additionally, in our country in hospital stroke mortality is higher (14.7%) and more than half of the patients were discharged with severe disability [ 14].
Despite the high burden of strokes globally, there are few available data and there is insufficient information on the current epidemiology, prevention and management of stroke in African countries and other LMICs [ 9, [15][16][17]. This is due to lack of adequately trained manpower and other resources to combat the epidemic [ 16,17]. The paucity of data has limited research output and consequently the response to this burden [ 18]. Hence this study was aimed to assess burden, clinical outcomes and predictors of stroke in hospital mortality among adult patients admitted to stroke unit of JUMC.

Methods
The study was conducted at stroke unit of Jimma university medical center (JUMC), a tertiary hospital found in Jimma city, south-west Ethiopia. The prospective cohort study was carried out from March 10-July 10, 2017.

Eligibility Criteria
All adult stroke patients diagnosed clinically or confirmed by imaging as per world health organization 5 (WHO) criteria and admitted to stroke unit of JUMC during the study period were included. Exclusion criteria -Patient or guardian not willing to give an informed consent -Those who died before evaluation by stroke neurologist/ other physician -If initial diagnosis of stroke was later changed to other (ruled out stroke).
-Patients with diagnosis of transient ischemic attack and hematomas.
-Brain ischemia with hemorrhagic transformation and vice versa.
-Stroke with undetermined type clinically where imaging was unaffordable -Stroke readmission cases were excluded.

Outcome and validating methods
In hospital mortality after hospital admission of the patient was considered as the clinical outcome of the study. Patients were followed from hospital arrival until died in hospital/ discharged. Death ascertainment was based on physician duty note along with suspected immediate causes of death.
Length of hospital stay/ admission was measured as the time gap from the patient admission to stroke unit until patient discharged or died in the hospital. In addition there were different validating methods that measures factors that were important predictors of outcome of interest. Stroke severity was obtained as per by the national institute of health stroke scale (NIHSS) [ 14] and level of consciousness was obtained by Glasgow coma scale (GCS) [ 19]. Collection of clinical endpoints and other needed parameters were performed daily from the time of patient admission until patient died in the hospital or discharged. Initial neurological assessment was performed within 24 hours of hospital arrival. The decision to perform different ancillary tests, laboratories, imaging and clinical history taking was left to the treating clinicians [ 20].

Statistical analysis
The data was analyzed using SPSS version 20. Descriptive statistics such as proportions, means, standard deviations, medians and interquartile ranges were calculated to describe the independent variables. Chi-square (χ2) test was used to test the significance of associations between categorical 6 variables. In hospital mortality rate was calculated by the Kaplan-Meier method and compared with the log rank test. Predictors of stroke mortality at hospital were investigated with the use of Cox regression to estimate the hazard ratio of explored predictors. Because of adequate significant variables were obtained at P<0.05, it was considered as cut off point for candidate selection and those identified variables at p<0.05 on binary Cox regression were subjected to multivariable Cox regression. Multivariable Cox regression with backward stepwise approach was used to identify the independent predictors of stroke mortality. Interaction between covariates and types of strokes were tested. Confidence interval which doesn't contain 1 and predictor variables with probability value less than 0.05 were considered statistically significant.

Results
During the study period there were a total of 756 medical admissions, of which 125 were stroke related admission. From these total admission, 110 of them experience in hospital mortality and stroke account for 26 of the in hospital mortality. Overall, stroke accounted for 16 Outcome and discharge condition of the patients A total of 91 (78.4%) patients were discharged alive making in hospital mortality 25 (21.6%). From those discharged patients, 67 (57.8%) discharged with improvement and 16(13.8%) left against medical advice (LAMA) on self and family request. During discharge 81(89.0%) patients were discharged to home, but the remaining 10(11.0%) were transferred/referred to other hospital/ ward /health facility.
The mean national institute of health stroke scale (NIHSS) of the patients during discharge was 10.32+5.8, which was higher in hemorrhagic stroke patients compared to ischemic stroke patient (11.10+6.4 Vs 9.75+5.3) without statistically significant difference (P=0.275). Majority of the patients had moderate NIHSS 43 (47.3%) and only one patient had severe brain injury (GCS ≤8) during discharge.
The mean modified Rankin score (mRS) at discharge was 3.97+1.5 for all stroke patients [IS=.63+1.38 and HS=4.34+1.55] which was statistically different (P=0.013). At discharge majority of patients 44(37.9%) had severe physical disability (mRS 4-5) and all patients "discharged to die" were classified as having mRS = 5 (severe disability) at the time of discharge.
The median length of hospital stay for all patients was 9.21days (ranged: 0.29-39.01days). The median length of hospital stay for ischemic stroke patients was 9.88 days, while that of hemorrhagic stroke was 8.49days. Seventeen patients (14.7%) discharged within 3 days and 22 patients (19.0%) stayed for greater than 2 weeks after hospital admission ( Table 2).

In hospital mortality of stroke patients
The in-hospital stroke mortality was higher for hemorrhagic stroke compared to ischemic stroke (32.1% Vs 11.7%) (P=0.01). The median time of in hospital mortality after admission was 4.38 8 (ranged: 0.29-13.75 days). The median time of in hospital mortality after admission for ischemic stroke was 4.30 days and hemorrhagic stroke was 4.41days. From total 25 patients died in hospital, ten patients (8.6%) were died within 3 days, 10 (8.6%) between 3 and 7 days and 5 (4.3%) died after one week of hospital admission.

Immediate causes and predictors of in hospital mortality
The prominent suspected immediate cause for in hospital mortality forwarded by clinicians was increased intracranial pressure 17 (68.0%) followed by respiratory failure secondary to aspiration pneumonia 11 (44.0%) ( Survival probability curves derived from Log rank Kaplan Meier in hospital mortality with different factors was shown ( Figure 1).

Discussion
In this study stroke accounted 16.5 % of total medical admissions, and 23.6 % of the total medical cases of in hospital mortality. This admission rate was higher than findings from Gambia in which the stroke patients made up 5% [ 23] and in southwestern Nigeria made up of 4.5% medical admission [ 24]. But the finding was in agreement with previous study conducted in Hawasa Ethiopia in which stroke accounted for 13.7% of all medical admissions [ 14]. The elevated number of stroke admission in Ethiopia may be due to lack of awareness, poor risk factor control and being hospital based study with referral bias.
A total of (78.4%) patients were discharged being alive from the hospital with in hospital mortality rate of (21.6%). From those discharged being alive more than half (57.8%) were discharged with improvement which was lower as compared to study by Masood  . For the shorter length of hospital stay in our set up, multiple reasons could be explained. Some patients were rapidly improved and discharged due to the stroke unit had proper possible care as compared to other wards in the hospital. Secondly some patients were died rapidly, some discharged LAMA and others discharged with medical advice without improvement due to small bed occupy of the stroke unit of the hospital. In this regard if the patient stayed longer than other patients and any improvement to the condition was seen, the bed would be left for new stroke patients.
In some patients there was shorter length of stay because of stroke unit provides better quality of care during the early phase. Additionally in some patient's delays in complimentary evaluations is one of the most feasible explanations for the prolonged admission time, which not only significantly increases the costs for stroke care, but also increases the risks for infection, other complications, and recurrence in patients with suboptimal treatment and evaluation.  In general the in hospital case fatality rate of stroke in our study was higher than reports from western studies, but was quite similar to SSA studies. This difference could reflect the limited access to hospital care, limited staffing, including availability of physiotherapy and occupational therapy similar to other developing countries as well as insufficient number of hospital beds for longer period care. In addition to this some caregivers/patients belief that people should die at home, where they spent most of their lives, with family members around and caring for them. Absence of treatment with thrombolytic, the low frequency of treatment with antiplatelets for patients with ischemic stroke and lack of evaluation with neuroimaging suggest that suboptimal care be the most likely explanation.  14]. It has been stated that the high mortality rate in this study during the first one-week (17.2%) may be due to raised intracranial pressure and aspiration pneumonia.
Brain edema, urine incontinence, NIHSS>13 during hospital arrival and diagnosis of stroke clinically alone were the independent predictors of in hospital mortality up on multivariable cox regression.
Except stroke severity other factors were not reported study by Atadzhanov et al in Zambia [ 15]. In this study increased NIHSS was associated with stroke severity constituting decreased level of consciousness. High NIHSS score as a predictor of mortality, was consistent with previous study done by Deresse et al [ 14].
In this study brain edema as complication was one predictor of in hospital mortality unlike study by ]. The overall difference in independent predictors of in hospital mortality could be due to sample size, study design, significance value considered and inclusion criteria of the patient.

Conclusions
During discharge majority of the patients were alive and discharged from the hospital with improvement. The mean NIHSS of the patients during discharge was moderate, but majority of patients had severe physical disability. Similar to other most of LMICs, Stroke mortality was very high in this setup. The in-hospital stroke mortality was higher for hemorrhagic stroke and the prominent immediate cause for in hospital mortality was increased intracranial pressure and respiratory failure secondary aspiration. Development of brain edema, urine incontinence NIHSS>13 during hospital arrival and diagnosis of stroke clinically alone were the independent predictors of in hospital mortality up on multivariable cox regression.
The following points were forwarded as strategies to improve the stroke outcome.
The non-governmental organizations (NGOs) and other non-profit organizations that work in areas of non-communicable diseases should focus towards the current debilitating conditions of stroke in SSA 13 including Ethiopia through better funding of the health care system to improve the quality of care.
There should be burning need to establish and strengthen the available stroke units which are wellequipped and staffed with intensive health care teams in different hospitals across the country.
Organized inpatient stroke unit care probably benefits a wide range of stroke patients in a variety of different ways, i.e., reducing death from secondary complications of stroke and reducing the need for institutional care through a reduction in disability.
There should be aggressive propaganda from every social, media, and political level of the country in the purpose of increasing the awareness of risk factors and needed interventions to overcome the current challenges of stroke by making the people to understand the devastating effect of the stroke.
Timely evaluation and initiation of treatment for stroke patient is critical to optimal patient outcomes. Since the cost of t-PA is a limiting factor in developing countries including Ethiopia, government subsidy or free provision of t-PA in government hospitals can improve thrombolysis rate.
Long-term comprehensive care to address risk factor control, mortality and the impact of ongoing disabilities resulting from stroke should be addressed through coordinated work by different sectors.
Access to stroke experts, neurologists, rehabilitation therapists and well-coordinated stroke team, as well as well as availability of well-equipped diagnostic instruments are necessary to improve therapeutic strategies as well as reducing mortality and morbidity.
Finally, future work must be designed to identify the barriers to improve stroke outcomes and recovery. With this a prospective community based longitudinal studies are required to identify burden, risk factors and outcomes of stroke. We thank Jimma University for supporting the study. We are grateful to staff members of stroke unit of JUMC, data collectors and study participants for their cooperation in the success of this study. Our thanks extends for Mr. Hunduma Wakassa for his especial contribution in the success of the study.
Permission was obtained from all authors for using previously published (table 1). Thus we acknowledge the authors of the article [ 20,22].

Funding
The study was funded by Jimma University. The funder had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Competing Interests: No competing interests exist Authors' Contributions GF contributes in the design the study, analysis and write up the manuscript. AK made the data analysis and interpretation of the data. LC contributed to the design of the study, monitoring the study and edition of the manuscripts. All authors critically revised the manuscript and have approved the final manuscript.
Ethics approval and consent to participate Ethical clearance was obtained from the Institutional Review Board (IRB) of Jimma University, Institute of health. At hospital written informed consent was obtained from the participants. All patients got the right to opt out of the research. This was done by explaining the objective and importance of the study as it would be beneficial for patient's quality service delivery for future encounters. The data from the case records and interview was handled with strong confidentiality. Neither the case records nor the data extracted was used for any other purpose. The confidentiality and privacy of patients was assured throughout by removing identifiers from data collection tools using different codes.