Intra-hospital mortality of stroke and its predictive factors in a reference hospital in Ouagadougou, Burkina Faso

Introduction: The aim of this study was to identify predictive factors for intra-hospital stroke mortality in a cohort of patients hospitalized for stroke within 72 hours of the onset of signs at the Tingandogo University Hospital in Ouagadougou, (Burkina Faso). Patients and methods: It was a prospective, analytical study of consecutive patients hospitalized for stroke from March 2015 to February 2016. Sociodemographic characteristics, vascular risk factors, comorbidities, clinical, neuroradiological and biological data at admission, as well as Intra-hospital mortality rates have been described. An univariate, then multivariate analysis with logistic regression allowed to identify the independent predictors of intra-hospital mortality of stroke. Results: During the study period, 157 patients were consecutively hospitalized for stroke, 113 cases of cerebral infarction (72%) and 44 cases of intracerebral hemorrhage (28%). The male patients accounted for 61.1% of the workforce, the average age was 61.8 years (range 28-92 years). At admission, the National Institute of Health Stroke Score (NIHSS) averaged 15 (range 0-38), neurological deficit was severe (NIHSS ≥ 17) in 46.7%, Glasgow admission score averaged 13.7 (extremes 3 and 15) and 12 patients (7.6%) were in a coma (Glasgow ≤ 8). The mean hospital stay was 13.2 days (range 3 and 40 days). The intra-hospital mortality rate was 28.7% with an average intra-hospital death rate of 12.1 days (range 3 and 40 days). Independent predictive factors for intra-hospital mortality were NIHSS at admission ≥ 17 (OR 2.909, p = 0.036), admission hyperglycemia (OR 6.752, p = 0.000), renal failure at admission (OR 3.903, p = 0.031), hemorrhagic stroke (OR 5.580, p = 0.003) and cardiac comorbidities (OR 6.393, p = 0.009). Intra-hospital mortality is high in Tingandogo University Hospital in Ouagadougou, Burkina Faso. Reasons for the increased mortality rate have been discussed. Conclusion: Intra-hospital mortality of stroke remains high in Tingandogo University Hospital. Reducing early mortality by stroke in sub-Saharan hospitals requires early access to the highest number of patients to quality care in high quality specialized facilities such as stroke units. Correspondence to: Lompo Djingri Labodi, University Hospital of Tingandogo, Unit of Formation and Research of the Sciences of the Health, University Ouaga I-Pr Joseph Ki-Zerbo, Burkina Faso; Tel : +226 70 23 98 34; E-mail: labodilompo@yahoo.fr


Introduction
Stroke appears today as a major cause of preventable death and disability in the world [1,2]. Today, early intra-hospital or monthly mortality per stroke varies from less than 15% to 22.9% in Western countries [3][4][5]. Early mortality of cerebral infarcts has declined sharply, dropping from 10. 2-12. 7% in 2003 to 8. 4-10. 1% in 2013, as a result of improved access to quality care, including timely patient transportation, evidence-based medical interventions, and specialized high-quality facilities such as stroke units (SU) [6]. About Intracerebral haemorrhage (ICH), their intra-hospital mortality has remained quasistationary [7] due to the absence of specific therapeutics effective in reducing mortality [8].
This high stroke mortality rate in sub-Saharan Africa may be due to the higher proportion of ICHs in blacks [18] and their high lethality around the world, the high incidence of severe or catastrophic stroke [19,20], the shortage and prohibitive costs of medico-technical equipment, inadequate health human resources [21,22]. Many predictive characteristics of early stroke death have been identified [14,23,24]. Hospital mortality of stroke is an excellent tool for measuring performance and quality of hospital management of stroke; its evaluation and a better understanding of its predictive factors are useful for the implementation of specific therapies and effective strategies for the management of high-risk patients [4,25], in a resource-constrained environment. To our knowledge, no studies have yet been made in Burkina Faso on the intra-hospital mortality of stroke and its predictive factors of occurrence.
The purpose of this study was to evaluate intra-hospital mortality and to identify independent predictive factors for intra-hospital deaths of patients hospitalized with stroke less than 72 hours.

Patients and methods
This is a prospective, transverse, observational, descriptive and analytical study carried out at the University Teaching Hospital (UTH) square test to compare percentages; the value of p <0.05 was considered as a threshold of statistical significance. Univariate analyzes between the different characteristics of the patients and intra-hospital mortality made it possible to select the variables significantly associated with intra-hospital mortality. Finally, multiple logistic regression analysis identified independent factors influencing intra-hospital mortality. Only variables with a p <0.20 value in bivariate analysis were taken into account for multivariate analysis.

Results
During the study period, we recorded 157 cases of stroke, 113 cases of cerebral infarction (72%) and 44 (28.7 %). The initial volume of HIC was ≤ 60 cc in 31 patients (70.5%) and> 60 cc in 13 patients (29.5%). General and neurological complications present at admission or appearing during hospitalization were dominated by fever with 85 cases (54.5%), pneumonia with 75 cases (48.7%), neurological deterioration with 60 cases (38.9%) and epileptic seizures with 27 cases (17.8%). Table 1 shows the distribution of patients according to their main characteristics.
The mean hospital stay was 13.2 days (range 3 and 40 days). At the end of the hospitalization, 45 patients died, ie an intra-hospital stroke mortality rate of 28.7% for an average life expectancy after stroke in the 12.1 days (extremes 3 and 40 days) and an average hospital stay of 13.7 days (extremes 4 and 33 days) in survivors (p = 0.019). Depending on the nature of the stroke, there were 26 cases of cerebral infarction deaths and 19 cases of HIC deaths, ie intra-hospital fatality rates, 23% for infarcts and 43.2% for the ICHs (p = 0.012). The intra-hospital mortality rate was 12.7% at day 7, 21% on day 14 and 28.7% at the end of hospitalization.
After univariate analysis, cardiac comorbidities, an admission delay of> 24 hours, a time to perform CT ≤ 4 hours, initial clinical severity of stroke (NIHSS ≥ 17), intake swallowing disorders, coma on admission (GCS ≤ 8), intake fever, intake hyperglycemia, intake dyskalaemia, intake dysnatraemia , renal failure at admission, intake leukocytosis, hemorrhagic nature of stroke, were the variables significantly associated with intra-hospital mortality.
After multivariate analysis using the ascending logistic regression method, the following independent intra-hospital mortality predictive of Tingandogo in Ouagadougou. It is one of the tertiary hospitals (3rd reference level) of the city of Ouagadougou, political capital of Burkina Faso. The hospital has 600 beds in 15 pavilions, of which only 200 are operational at the present stage. Our study was conducted in the department of neurology, housed in the department of medicine and medical specialties, which has 34 beds distributed in 12 rooms. Our study lasted 18 months, from March 1, 2015 to August 30, 2016. It examined adult patients (> 16 years of age) of all sexes, consecutively hospitalized in the Department of Neurology, for neuroimaging (CT or Brain MRI) confirmed stroke, up to 72 hours after its onset, excluding subarachnoid haemorrhages. For each patient, the following tests were performed at admission: blood pressure (BP), temperature, capillary blood glucose; National Institutes of Health Stroke Score (NIHSS), Glasgow coma score (GCS); Electrocardiogram (ECG) standard; brain scan; creatininemia & blood urea, numeration blood formula-platelet levels, blood crest, blood ionogram. In case of cerebral infarction, the lipid balance (total cholesterol, Low Density Lipoprotein (LDL) cholesterol, High Density Lipoprotein (HDL) cholesterol, triglycerides), transthoracic heart ultrasound and holter ECG were also performed, if necessary. The radiologists performed the interpretation of the brain scan. Additional tests were performed on a case-by-case basis: standard chest x-ray in case of suspicion of bronchopneumopathy, biological check-ups if necessary; tick drop in case of suspicion of malaria, cytobacteriological examination of the urine if suspicion of urinary infection, blood culture, in case of suspicion of septicemia; brain scanner if needed, ... The evolution of the patients during hospitalization was monitored daily by a clinical evaluation possibly aided by complementary examinations on a case by case basis and the complications regularly noted in the medical file, until the end of the hospitalization. The complications sought included those observed at admission or which appeared during hospitalization. Upon discharge from hospital, patients were subdivided according to discharge status, surviving patients and deceased patients. The variables studied included socio-demographic characteristics, vascular risk factors (VRFs), modified Rankin score (mRS) before stroke, comorbidities, care pathway, (temperature, BP, GCS, NIHSS), radiographic data on initial CT scan [nature of stroke, ICH volume, other neuroradiological abnormalities), qualitative biological data at admission (hemoglobin, leukocytes, blood sugar, serum sodium, serum potassium, serum creatinine); medical complications, present at admission or occurring during hospitalization (thromboembolic venous complications, haemorrhagic complications, metabolic complications, infectious complications, cardiac complications, respiratory complications, neurological complications present on admission or appearing during hospitalization: neurological deterioration during hospitalization (increased neurological deficit, including alertness), epileptic seizures, relapse or extension of stroke, life-threatening prognosis at the end of hospitalization ( survivors / deceased).
The consent of the patients or that of their legal representatives was ensured before the recruitment. The study protocol was approved by the ethics committee of the University of Ouaga I-Pr Joseph KI-ZERBO and by the national ethics committee of Burkina Faso.
Statistical analyzes were carried out using the SPSS12 software. Student's t-test was used to compare the averages and the Pearson Chi-= 0.000), renal insufficiency at admission (OR 3.903, 95% CI 1.131-13.475); p = 0.031). The results of the univariate and multivariate analysis are summarized in the following Table 2.
Intra-hospital mortality in our study is, however, significantly higher than that reported in developed countries, where it varies from less than 15% to 22.9% [3][4][5]17,18,27]. The lowest rates are for cerebral infarcts, whose mortality dropped from 10.2-12.7% to 8.4-10.1% between 2003 and 2013. The early mortality rate for ICHs has remained virtually unchanged [16]: 34% in France [27], 40% in the USA [8], because no specific therapies have yet proved effective in reducing mortality after ICH [8]. Reduction of early mortality in ischemic stroke patients in developed countries is due to improved access to quality care, including timely patient transportation, evidence-based medical interventions and specialized facilities of high quality such as SU [6]. Moreover, better prevention of vascular risk factors is more effective, which has reduced the incidence of catastrophic stroke still too frequent in Africa [1,19,20]. Conversely, the high mortality rate of stroke in sub-Saharan Africa may be due to a number of factors: the highest proportion of ICHs in blacks [18], the highest ICH lethality reported worldwide [18,27]; inadequate primary prevention of stroke [19,20], the shortage and prohibitive costs of medical-technical equipment, and the inadequacy of qualified human resources for investigations, emergency care and rehabilitation of stroke patients in sub-Saharan Africa [28].

Independent predictors of intra-hospital mortality
Our study identified several independent predictors of intrahospital mortality after stroke.
The initial clinical severity of stroke (initial NIHSS ≥ 17) recognized as an independent predictor of early mortality by our study and several other studies [13][14][15], directly reflect the extent and severity of neurological damage secondary to stroke [3,13,14,27]. We have shown, like other studies [13,14,[29][30][31][32], that admission hyperglycemia was a predictor of early post-stroke mortality. Indeed, hyperglycemia often complicates severe acute strokes, in response to the major stress triggered by lesions of extensive cerebral necrosis, which explains the pejorative life prognosis in these patients. Hyperglycaemia, in turn, contributes to an aggravation of the initial infarction, via its toxicity in the ischemic penumbra zone and potentiation of early reperfusion lesions; the ultimate consequence is an increase in the final volume of the cerebral infarction, a greater risk of hemorrhagic transformation and a pejorative functional and vital prognosis [29][30][31][32]. Some studies have reported that hyperglycemia is associated with a poor prognosis after ICH but the underlying mechanisms have not yet been identified [33,34]. Onset, vigilance, severity of neurological deficit and blood glucose levels identify subgroups of patients at increased risk of early stroke death to ensure appropriate monitoring and care ideally in an intensive care stroke unit. We have found as others [13,22,27,35] that ICHs were an independent predictive factor of early post-stroke intra-hospital mortality. It is universal that ICHs, compared to cerebral infarcts, have a higher mortality rate [18,27]. Indeed, they are more often accompanied by an early alteration of vigilance, or even coma, due to a more frequent and earlier intracranial hypertension.
We have shown, like others [25,[36][37][38], that cardiac comorbidities were also an independent predictive factor of intra-hospital mortality, due to the often advanced age of these patients, the usual clinical extent and severity of cardiac embolism, of a high rate of early cardiac mortality [36,39,40]. Specialized cardiac assessment and management is required in patients with acute stroke associated with cardiac comorbidities [36].
Finally, we identified renal insufficiency at admission as an independent predictive factor of intra-hospital mortality per stroke, in agreement with some publications [41][42][43][44]. This could be explained by an increase in the susceptibility of these patients to infections and a greater frequency of fatal cardiac complications secondary to metabolic disorders [45].
Other independent predictors of intra-hospital stroke mortality, identified by some authors, such as sex and especially age [27], severe initial hypertension [22], history of stroke [10,22], were not found in our study.
However, our study suffered from a number of shortcomings: the fact that patients were hospitalized more than 72 hours after the onset of stroke significantly reduced the number of patients, the late admission of some patients (16.6%) did not make it possible to obtain the early data of the beginning of the stroke and finally we did not have the neuroradiological data for the intra-hospital follow-up of our patients because of the non-accessible financial possibility of the cerebral scanner. However, these shortcomings have not hindered the relevance of our study.

Conclusion
Intra-hospital mortality of stroke remains high in Sub-Saharan Africa, ranging from 25% to 45%, and the hospital mortality rate of 28.7% in our study falls within this range. The initial clinical severity of the stroke and initial hyperglycemia, indicative of the extent of neurological damage caused by stroke, cardiac comorbidities due to the usual severity of cardio-embolic infarction and resulting cardiac mortality, the haemorrhagic nature of stroke due to high coma propensity due to intracranial reaction hypertension and renal failure by associated cardiac and infectious complications are the independent predictive factors of early post-stroke deaths. Strategies to reduce early stroke mortality should include screening, monitoring and management of subgroups of patients most at risk of early death, specifically those with hyperglycemia and / or severe neurological deficit and / or coma at the outset, as well as early detection and specialized cardiac and / or nephrological management of patients with cardiac or renal comorbidities. These strategies should also include the prevention and early treatment of intracranial hypertension, ideally in SUs that have proven to be highly effective in reducing mortality and stroke disability around the world.