Stroke is a leading cause of death and disability worldwide [1,2]. Stroke also imposes an important economic burden. In Western countries, approximately 2-5 per cent of total health-care costs are due to cerebrovascular diseases [3]. Within the first year following a stroke, acute in-patient care accounts for the majority of direct health-care costs [3-5]. The management of acute stroke in hospital has a variable duration, lasting from 1 week to 2 weeks on average [6,7]. One of the main predictors of acute in-patient care cost is length of stay (LOS) [6]. Most of the reports concerning association of stroke and LOS concern three major stroke categories (ischemic stroke, 87%; intracerebral hemorrhagic stroke, 10%; and subarachnoid hemorrhage stroke, 3%) [6,8-13]. Previous studies have identified a number of these key factors including stroke severity, stroke type and others [12,14,15]. LOS of stroke is known to associate with many demographic and clinical factors, and observed associations are sometimes inherent within health care systems. Prolonged length of stay following stroke inflates health care cost, increases risk for hospital-acquired complications, in-hospital death and has been associated with worse outcome [16,17]. The majority of stroke was managed in stroke units, neurology departments and rehabilitation departments. The human, social and economic consequences of stroke are particularly devastating in African countries, particularly because of the weakness of the health system in terms of early detection and adequate care [18]. According to a study carried out in Benin, the direct hospital cost of stroke was significantly related to the length of patient hospitalization [19]. However, there are few studies in Africa on the factors associated with the lengthening of hospital stays in patients with stroke. The only one study in Sub Saharan African was carried out in Nigeria [20]. In Burkina Faso, the care of strokes is devalued at a few tertiary hospitals, remains dependent on the insufficiency of the technical platform and of qualified personnel (neurologists, physiotherapists). Moreover, there are no studies on the hospital cost of strokes and its predicting factors such as hospital stay. The aims of our study are to determine the predicting factors of longer hospital stay for acute stroke and to contribute to the improvement of the hospital care of patients suffering from stroke.

Study profile

This cross-sectional study was carried out in the neurology department of University Hospital Yalgado Ouedraogo in Ouagadougou, the capital city of Burkina Faso.

Study population

The study included stroke patients admitted in the neurology during the period from January 1, 2019 to December 31, 2019. Stroke were confirmed by computed tomography. Patients discharged against medical advices, patients placed under observation and those with incomplete records were not included in the study.

Data collection technique and tools

The data were collected by OF. We used the patient's clinical record as a collection tool. A collection sheet comprising a pre-written questionnaire was completed using the information contained in the patient's medical file.

Study variables: The study questionnaire included socio-demographic variables (age, sex, place of residence, occupation, marital status, educational level, socioeconomic level); clinical variable (vascular risk factors, neurological symptoms, stroke investigations, treatments, stroke subtypes, length of stay).

Data analysis and processing

The data were analyzed on a computer using Excel and Epi info software in version 7.2. Quantitative variables were expressed as mean and standard deviation. The qualitative variables were expressed in number and in percentage. For the analysis, we used Student's test and Karl Pearson’s chi² test for observations with sample sizes greater than or equal to 5; the YATES corrected chi-square test was used for observations with a sample size of less than 5. The threshold of significance was 5%. We considered that there was a statistical difference when the “p” value found was less than 5%.

Assessments

LOS was defined as the difference between admission to discharge, death or other residential institution. Numerically, it was calculated by discharge date minus admission date. If admission date and discharge date were the same, LOS was set to 1 day [11]. Delayed assessment: for biological and radiological examinations (blood count, lipid assessment, urea, serum creatinine, blood monogram and glycemia, cerebral CT) carried out after 24 hours of the request. Longer LOS is defined as hospital stay period greater than or equal to 14 days.

Frequency

Three hundred and four stroke patients were admitted in the neurology department during study period. Of them, 200 (65.79%) patients had met the inclusion criteria.

Sociodemographic characteristics and vascular risk factors

The mean age of patients was 57.45 ± 14.32 years, ranging from 21 and 99 years. Most patients were male gender (108; 54%), married (67; 83.5%) and residing in urban areas (121; 60%). The majority of patients had no formal education (102; 51%). Hypertension and alcohol consumption were present respectively in 61.5% and 28.5% of patients. Of these 123 (74.1%) were known uncontrolled hypertensive and 25.90% new diagnosed. Table 1 summarizes the demographic and clinical characteristics of the 200 patients.

Stroke characteristics

The delay of admission of patients was 3.81 ± 3.009 days, ranging from 1 and 28 days. Twenty-two (11%) of patients were admitted in the first 24 hours. The main symptoms of stroke were motor deficit (n=186; 93%) and language disorders (n=90; 45%). The mean NIHSS score was 10.2 points ranging from 1 to 22. Sixty (30%) patients had severe stroke (≥16). Hemorrhagic strokes predominated accounted for 56.5% (n=113) and ischemic stroke for 43.5% (n=87). Six patients with IS had cardiovascular stroke. Biological investigations revealed hyperglycemia in 44/187 (22%) patients, anemia in 45/200 (21%) and hyperleukocytosis with elevated CRP in 61 (30.5%) patients. Turnaround time for cardiovascular investigations was respectively 67 hours 41 minutes for echocardiography, 76 hours 19 minutes for Echo TSA and 56 hours 24 minutes for electrocardiography. Turnaround time for laboratory tests was 26 hours 10 min (lipid profile, blood ionogram, creatininemia, fasting blood sugar, blood count, Plasma urea). All the patients had performed the head CT scan in the first 24 hours of stroke onset. The frequency and delay in stroke investigations were summarized in the Table 2. Six patients were treated with antivitamin K (3%) and 7 (3.5%) with oxygen therapy. All the patients need physical rehabilitation. All the sessions were attended by physical therapist assistants. Of them, 51 (25%) had performed no therapy sessions during hospitalization. The mean length of stay was 10.81 ± 5.07 days, ranging from 2 to 30 days. Fifty-seven (28.5%) patients a longer stay (>14 days), 50 (20%) under 7 days and 93 (76%) between 7 and 14 days. Mean LOS was respectively 10.6 days in ischemic stroke and 8.9 days in hemorrhagic stroke. Pneumonia (14%), neuropathic pain (18%) and (Urinary tract infection (UTI) (10%) were the 3 most common complications during hospitalization. The mortality rate was 5.5% (n=11). Table 3 gives the clinical characteristics of stroke. All the survived patients were transferred from the rehabilitation hospitals to neurology department for further management of their medical problems.

Factors associated with lengthening hospital stays

The Table 3 highlights A the results of bivariate analysis. Among clinical characteristics, patients with high fever, speech disorders, swallowing disorder, and consciousness disorder had a significantly longer LOS (p-value<0.05). Among therapeutic characteristics, low physical sessions, vitamin K antagonist treatment and urinary catheterization were significantly associated to longer LOS (p-value<0.05). Among clinical outcome, patients with respiratory infections, bedsores and urinary tract infections, had a significantly longer LOS (p-value<0.05). In the multiple regression model, there was no factor had significant influence on longer LOS. The Table 4 summarizes the results of multiple regression analysis.

This study investigated the association between LOS and demographic, clinical and therapeutic data of patients admitted in the neurology department of Yalgado Ouedraogo during the year 2019 for acute stroke (ischemic and hemorrhagic).

The analysis presented above showed that the mean LOS of stroke was 10.81 ± 5.07 days and that this LOS was similar than reported in Brazil for patients managed on a general/neurology ward (10.8 days) [21], in France (10 ± 12 days) [22], in Taiwan (11days) [12], in Denmark (13 days) [23] and in Nigeria (13.7 ± 8.9 days) [20]. In contrast, this LOS was significantly higher than that those reported in Ghana (6.2 days) [24], in Madagascar and Turkey (7 days) [25,26], in India (5.6 ±4.3 days) [27], and in Iran (5.6 ± 2.1 days) [28]. Higher mean LOS was observed in Korea (115.6 ± 219.0 days) [29], in Togo (19.9 ± 13.5 days) [30], in China (17.4 and 19 days) [10,31] and in Australia (21 days) [16]. This difference could be explained by the socio-demographic characteristics of the patients, the severity of the stroke, the type of service (stroke unit, neurology ward, rehabilitation wards) and the delay in carrying out additional examinations. The proportion of overall patients with a longer LOS was 28.5%, higher than those reported in Taiwan (10.4%) [32] but lower than those reported in India (41.8%) [27]. This difference between the studies would be linked to the method of calculating the long stay. Unlike studies in Taiwan and India, we used 14 days instead of 7 days as the cut-off value. In our context, we do not have a neurovascular care unit that would not shorten patient length of stay. Although researchers have established that management of patients in an SU is associated with a reduction in length of hospital stay compared with other wards [4,10]. Average length of stay was longer in patients managed on general neurology/medical wards compared with those managed on the stroke unit [10]. Among the socio-demographic factors studied, only the socioeconomic level of the patients are significant predictors of increased LOS, in line with several studies [33-35]. In our contexts, these patients (housewife, farmer, itinerant merchant) do not manage to honor the additional examinations on time. The association between age and LOS was significant in our series, consistent with the results of previous studies [27,28,32] in contrast with the study of Lee and Spratt [16,36]. In these latest studies, older age is a strong predictor of prolonged hospital stay in acute stroke patients. The relationship between gender and LOS for stroke patients remains unclear, and various results have been reported in various studies [12,25]. In our study, gender was not associated with long hospital stay, in agreement with that of Hsuei-Chen et al and Majidi-Shad concerning the gender of patients [28,32]. To our knowledge, there was not studies evaluating the impact of residence and educational status on Stroke length of stay. In our study, residence and educational status were not associated to LLOS. Rural populations and less educational status were associated to severity of stroke in Burkina Faso.

There was no statistically significant relationship between duration of hospitalization and past medical history (diabetes milletus, hypertension, alcohol, tobacco, obesity) as in the study of Majidi-Shad in Iran [28]. However, just as Saxena et al. also reported, the presence of hypertension was not statistically associated with LOS among stroke victims in this study [27]. Our study was in contrast with some studies in which alcohol and smoking could increase the risk of hospitalization and prolong LOS [37,38]. DM and increased number of co-morbidities was found to be related to prolonged hospital stay in previous studies done by Spratt et al. and Lee et al. [16,36]. Among clinical variables, patients with speech disorders had significantly longer LOS in line with the study of Hsuei-Chen et al. [32]. Patients with aphasia do not complain and often have swallowing complications. Neither the Stroke subtypes nor the stroke severity were not associated to longer LOS in contrast with the study of Hsuei-Chen et al. [32] in which these two factors were associated to stroke severity and stroke subtype. Among predictors of prolonged hospitalization, Majidi-Shad et al. identified stroke subtype (hemorrhagic), and stroke volume (moderate, large) [28]. The delay in blood investigations (lipid profile, blood ionogram, creatinine, plasma urea, fasting blood sugar, hemogram) and cardiovascular investigations (cardiac ultrasound, ASD ultrasound, electrocardiogram) were not associated to longer LOS. In our series, the majority of patients had their biological and cardiovascular workup performed within 24 hours of the request. This is because most of the tests were available at the hospital lab. According to treatment, urinary catheter placement was associated to longer LOS, as in the study of Stott et al. [39]. In our situation, bladder catheterization often causes urological complications such as trauma to the urethra, urinary tract infections. Acenocumarol use was associated to longer LOS. This situation could be explained by the late initiation of treatment with acenocoumarol, mainly linked to the equilibration phase of acenocoumarol. Patients with lower number of physical sessions had longer LOS. In our context, patients pay the rehabilitation costs in advance. In addition, rehabilitation sessions for hospitalized patients are done twice a week due to lack of staff. There is no assigned physiotherapist in the neurology department and often the visit schedule is not followed. All the sessions were attended by physical therapist assistants. Patients treated with oxygen therapy had longer LOS. Hsuei-Chen et al. [32] found that oxygen therapy was associated with a long hospital stay. Urinary tract infections, bedsores and post stroke pneumonia were associated with longer LOS. Lee et al. in 2016 found that the presence of pressure ulcers was predictive of an extended length of stay [36]. The finding that urinary catheterization and nasogastric tube insertion were associated with increased LOS is also supported by a study by Roth et al. [40]. Stenzelius et al. [41] in a prospective study in 2016 found that urinary tract infection prolonged the length of stay by 3.5 days. Like many studies, our study did not find independent factors influencing longer LOS.

This study aimed to investigate the factors that influence the LOS in stroke patients. This study has several limitations due to the study profile and study location. First, it was a retrospective study and some data may be missing. Second, the LOS examined in this study was the LOS of acute care hospitalization (neurology) while most studies were performed in Stroke Units and physiotherapy departments. Finally, factors related to the health system could not be analyzed, such as the impact of the stay in Emergency Department (ED), the shortage of staff, the treatment protocols put in place.

Most stroke patients had longer LOS in our study. The factors associated to longer LOS were infections, complications of decubitus, anticoagulant treatment, waiting for rehabilitation, conscience disorders and socioeconomic status.

Ethical approval and consent to participate

This study was approved by the Ethical board of medical school of Joseph Ki Zerbo University and the local ethical Committee of Yalgado Ouedraogo University Teaching hospital. The confidentiality of patients’ data was taken into account.

Availability of data and materials

All the data and materials were available with the corresponding authors.

Competing interests

The authors do not report any conflict of interest.

Funding

no funding.

Author’s contributions

AAD, AD, OLF, JMAK had contributed to the data collection or processing, analysis or interpretation, literature research and writing. CN, AM and JK had contributed to concept and design of the study. All the authors read and approved the final manuscript.

Acknowledgement

Authors would like to thank Mrs. Zongo L Carine Patricia for translation.