Mortality rate and predictors of Stroke: A Meta-Analysis and Systematic Review

Background Global data on stroke mortality remained to be sparse. In light of this, we aimed to conduct a Meta-analysis and systematic review of observational studies to estimate the mortality of stroke and to identify risk factors that predispose patients for stroke-related death.Methods This study was conducted based on the Meta-Analyses of Observational Studies in Epidemiology (MOOSE) guidelines. Overall mortality, in-hospital and 30- day mortality due to stroke were the primary outcome measures of the study. The meta-analysis was performed using Stata (Version 14, Stata Corp, College Station, Texas). Random-effect models were used for estimating pooled effects.Findings Overall, thirty two studies assessed overall mortality due to stroke. A total number of 2,885, 126 patients were recruited for the study. Pooled estimate indicated that the overall mortality of stroke was reported to be 20% (19%-22%). Whereas, the 30-day and In-hospital mortality were found to be 18% (16%-20%) and 16% (16%-19%), respectively. A subgroup analysis revealed that Africa registered the highest stroke-related death 29% (23%-36%). Hypertension was found to be an important risk factor for mortality secondary to stroke 61.9% (52.8%-71.1%).Conclusion Overall mortality of stroke was estimated to be twenty percent. The burden of stroke mortality was prominent in Africa region. Hypertension remained to be an independent risk factor for stroke mortality. Mortality of stroke can be minimized by establishing stroke centers that promptly deliver emergency management of stroke event. (ASA)

Overall mortality due to stroke, in-hospital and 30-day mortality were the primary outcome measures of the study. Risk factors for stroke which constitute factors that predict the incidence of stroke and mortality were also identi ed.

Data Collection Process
A preliminary reading of titles and abstracts was carried out to include all relevant papers. Final inclusion of articles was made upon a thorough reading of full texts. Three investigators namely (TMA, BMG, MAS) undertook the independent review of all articles. In cases of disagreement, a fourth author (EAG) was involved and any discrepancy was resolved in consensus.

Data extraction
All records that were found from searches of the electronic databases were exported into the ENDNOTE software version X5 (Thomson Reuters, USA) so as to avoid duplication. Speci c datasheet was prepared using Microsoft Excel software version 2013. Information on studies including year of publication, objectives of the study, sociodemographic and clinical characteristics of participants, complications of stroke, risk factors and types of stroke was taken from each study. The mortality rate of stroke reported as overall mortality, in-hospital mortality and 30-days mortality was fetched and stored in an independent sheet stated/leveled as "main analysis".

Data analysis
The meta-analysis was performed using Stata (Version 14, Stata Corp, College Station, Texas) [15]. Heterogeneity and publication bias were assessed using Comprehensive Meta-analysis version-3 (Biostat, Englewood, NewJersey, USA).Results were summarized using conventional forest plots. Random-effects models for estimating pooled effects were considered preferable rather than xed-effect models because high variability across the included studies was expected. The heterogeneity in pooled estimation was determined by the DerSimonian-Laird (DL) approach and was assessed using I 2 . Sensitivity analysis was conducted to determine the robustness of the results and sources of variation in pooled estimation, respectively. Subgroup analysis was performed based on study design and geography of study subjects. Moreover, publication bias for the primary outcome was assessed by Egger test and inspection of funnel plots. The quality of the studies was evaluated using STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) scale [16]. Accordingly, we arbitrarily classi ed included studies into high quality (≥75% of the STROBE checklist) and low quality (<75% of the STROBE checklist).

Study characteristics and the selection process
Overall, 8408 records were generated from four electronic databases including PubMed (1654), Scopus (2857), Google scholar (3670) and Embase databases (227), respectively. No grey literature was included. The preliminary screening based on titles and abstracts identi ed forty-one papers as potentially relevant for the full-text review and thirty two articles were nally deemed eligible for the systematic review and meta-analysis ( g 1).

Patient Characteristics
A total number of 2,885, 126 patients were recruited for the study. The mean age of participants was found to be 69.8± 13.8 years. There were 1172581(40.64%) males and 1712545(59.36%) females, respectively. Overall stroke mortality was documented for 302,592 (10.49%) patients. The rate of 30day mortality was reported in 19057(0.66%) subjects while in-hospital mortality was recorded in 108, 368 (3.75%) participants.

Types of stroke
About 121 3064 (42%) patients experienced ischemic stroke. hemorrhagic stroke was identi ed in 190,906(6.6%) of patients. The remaining stroke events were undifferentiated or unclassi ed.

Risk factors
The following risk factors were reported hypertension 1159461(40.2%) with mean systolic blood pressure of 154.29±21. 45

Determinants of mortality
Pooled odds ratio of precipitating factors revealed that hypertension was found to be an important risk factor for mortality secondary to stroke 61.9% (52.8%-71.1%). The probability of death has increased in a quarter among stroke patients who were having hypercholesterolemia 20.2% (10.0%-30.4%) and alcohol consumption 21.5% (10.4%-32.6%), respectively (table 2).

Sensitivity and subgroup analysis
The sensitivity analysis showed that omission of anyone of the included studies did not signi cantly affect the pooled mortality of stroke (all P < 0.05) (additional le 2). We performed subgroup analysis based on geographic area of the studies and the type of study design. Accordingly, overall mortality was found to be higher in Africa 29% (23%-36%) than any other continents ( g 5).In addition, mortality was higher among prospective studies 26% (22%-31%) versus retrospective 17% (15%-20%) (Additional le 3).

Discussion
Stroke is a sudden cerebrovascular accident that poses signi cant mortality and morbidity. Stroke mortality can be summarized as overall mortality, 30-day mortality and in-hospital mortality. Post-stroke complications of stroke denotes a constellation of disabilities including: neurologic sequelae, infection, psychiatric disorder, pain, movement disorders, cognitive and speech problems [46]. The determinants of mortality secondary to stroke represent an array of modi able and non-modi able risk factors including diabetes [47], hypertension [48], hypercholesterolemia [49], alcohol consumption [50], cigarette smoking [51] and atrial brillation [52].Despite the hassle burden of stroke, data regarding stroke mortality and its determinants were found to be scarce. To this end, a systematic review and meta-analysis of observational studies has been designed so as to explore the mortality rate of stroke on the basis of gross, thirty-day and in-hospital mortalities. In addition, the current review highlighted risk factors that implicate with stroke mortality.
A rigorous review of 32 articles provided a total of 2,885, 126 stroke patients. According to the pooled analysis, the overall mortality of stroke was reported to be 20% (CI: 19%-22%).This gure falls between ranges of mortality values reported by different studies. 7-8 Subgroup analysis con rmed greater incidence of overall stroke mortality in Africa29% (23%-36%). In Africa, the burden of stroke appeared to be in unprecedented rise which cause international inequality in relative to developed nations. Contemporary reports indicated 316 per 100 000incidence rate of stroke per year. Socio-demographic and lifestyle changes in the general population including population growth and longevity were supposed to attribute for this substantial burden [53][54].Ironically, there is limited intervention to reduce the impact of stroke on the lives of individuals. For instance, African counties lack national strategies to stop smoking, alcohol consumption and to promote physical activity [54][55].To date; there is no "state of the art" stroke care in African countries. Rather, the management of stroke is directed to supportive therapies. Further, diagnosis is speculative and complex due to lack of imaging modalities. Patients' pre hospital delay precludes the timely initiation of thrombolytic therapy [56].In contrary, low rate of crude mortality was observed in Europe (22%), America (16%), and Asia (16%). Recent advances in system and processes enable developed countries to e ciently use reperfusion therapy. Reperfusion improves outcomes by reducing the volume of brain tissue injury. Tele-stroke and mobile stroke units in the pre hospital setting have increased thrombolytic utilization and reduced delays to treatment. However, these technologies are hardly implemented in low and middle income countries despite their effectiveness [57]. 30-day mortality was found to be 18%(16%-20%) in the present review. Different explanations have been forwarded for the escalated burden of mortality in stroke patients within thirty days of the stroke event. Among ischemic stroke patients, the mortality of stroke depends on the e cacy of anticoagulation and the mortality associated with intracranial hemorrhage due to the supra-therapeutic anticoagulation [58].We can extrapolate that the 30-day mortality is not independently attributed from stroke, rather it could be secondary to the bleeding related to anticoagulation. In addition, the thirty-day mortality could be in uenced by integrated system of stroke care. For instance, retrospective analysis of a large data base of Canadian stoke registry reported a signi cant decline of 30-day mortality to 12.7% in well-organized stroke center. However, the magnitude of 30-day mortality of stroke is expected to be signi cant in developing countries where advanced stroke service and nursing care is inadequate [59].A prospective evaluation of death among Nigerian stroke patients with thirty days of ictus reported a 19.69% of mortality which is nearly a similar gure with the pooled estimate [17]. A multi-center analysis of large data of acute ischemic stroke patients in United States revealed 15.32% 30-day mortality. 39 In United Kingdom, time-trend analysis of the general practice research stroke database indicated similar rate of mortality (15.32%) [32].
Further, the present review included 108,368 subjects from 22 articles to determine the in-hospital mortality of stroke. Accordingly, the pooled in-hospital mortality was found to be 16% (14%-19%). The frequency of in-hospital mortality was reported among different countries such Iran 20.52% [34] Sweden(15.47%) [19] and Cameroon 20.62% [30]. Even though hospital admission facilitates the recovery of stoke patients, in the meantime, it predisposes patients for hospital acquired infections which complicate the survival of stroke patients. More importantly, the likelihood of in-hospital mortality is substantiallyaffected by the prompt and vigilant care. 60 In addition, the coincidence of comorbidities such as diabetes, AF and other cardiovascular disorders deteriorates the prognosis of patients admitted with stroke. 61 Increased risk of in-hospital mortalitywas also noted in patients who were presented withsevere form of stroke manly middle cerebral artery occlusion with herniation, basilar artery occlusion, and hemorrhagic transformation [62].Therefore, the contribution of in-hospital mortality for the burden of overall mortality of stroke could be minimized through delivery of pragmatic care for hospitalized stroke patients.
Our review highlighted that, hypertension was the most common precipitating factor for the onset of stroke and subsequent death. HTN accounted to a 61.9% (52.8-71.1) increase in the incidence of mortality. The occurrence of stroke on the top of elevated blood pressure is viewed as one of the hypertensive crisis encountered in emergency department. The prompt management of hypertensive emergency determines the prognosis of patients. Hemorrhagic transformation of ischemic stroke is a clinical deterioration and the most fatal scenario observed in patients who are not properly managed in critical setting. Poor outcome was noticed in hemorrhagic stroke patients due to the development of hematoma. Therefore, adequate blood pressure should be a priority so as to reduce the mortality of stroke [48,63].We also investigated other risk factors such as diabetes, AF and dyslipidemia as independent determinants that in uence stroke outcomes. Additional studies are warranted to explore strategies to minimize the consequence of these risk factors on stroke mortality.
In general, the current review revealed that the mortality of stroke found to be more signi cant worldwide to the extent that it becomes a huge hassle for the global health. But, the study was not without limitations. Firstly, some of the studies included in the review showed higher level of heterogeneity and we could not detect the source of variation with the available data. However, we conducted a sensitivity analysis to avoid the 'drowning effect' from large sample size studies. For example, one-on-one exclusionof Margaret C et al 2014 [41] and Agarwal S et al 2015 [42]found that the overall mortality did not differ from the original overall estimates.Secondly, it should be noticed that data regarding mortality might not be comprehensive since information from grey literatures were not searched.

Conclusions
Overall mortality of stroke was estimated to be twenty percent. The burden of stroke mortality was prominent in Africa region. Hypertension remained to be an independent risk factor for stroke mortality. On the basis of the ndings of the present review, it is recommended that the overall mortality of stroke can be minimized by establishing stroke centers that promptly deliver emergency management of stroke event. Particularly, these services should be extended in African region which represent the highest rate of mortality due to stroke. Further, blood pressure control should be given a due attention to reduce the risk of mortality secondary to cerebrovascular incidence. TMA conceived the study, prepared the study protocol, involve in acquisition of data, performed review, and analyze the data and write-up of the nal part of the manuscript. MAS, EAG, AGM, ASB, MBA and BMG prepared the study protocol, interpret and analyzed the data and wrote the initial draft of manuscript. All the authors read, approved the nal manuscript and agreed to be accountable for all aspects of the work.

Figure 2
The overall mortality of stroke patients