Predictors of In-Hospital Mortality Among Stroke Patients at a Tertiary Care Hospital in Nepal: A Prospective Cohort Study

Abstract

Stroke is a leading cause of morbidity and disability, with limited data on in-hospital mortality from low-resource settings. This study aimed to identify predictors of in-hospital mortality among stroke patients at a tertiary care hospital in Nepal. A prospective cohort study was conducted among 120 stroke patients aged ≥ 18 years, enrolled between November 2023 and April 2024. The primary outcome was in-hospital mortality following admission. Data was analysed using SAS version 9.4. Kaplan–Meier survival analysis and Cox proportional hazards regression were employed to identify predictors of in-hospital mortality. A p-value < .05 was considered statistically significant. The cohort comprised 68.3% ischemic and 31.7% haemorrhagic strokes, with an overall in-hospital mortality rate of 9.0%. Multivariate analysis revealed that a Glasgow Coma (GCS) score < 8 (AHR: 12.36; 95% CI: 2.73-56.00), National Institutes of Health Stroke Scale (NIHSS) ≥12 (AHR: 14.75; 95% CI: 3.01-72.28), moderate to severe disability (mRS ≥ 3; AHR: 9.92; 95% CI: 1.10-89.24), hemiplegia (AHR: 6.70; 95% CI: 1.835-53.748), territorial infarcts (AHR: 26.33; 95% CI: 2.093-331.203), capsuloganglionic infarcts (AHR: 14.6; 95% CI: 1.819-160.877), presence of chronic obstructive pulmonary disease (COPD) (AHR: 2.48; 95% CI: 1.317-45.091), and alcohol use (AHR: 3.87; 95% CI: 1.014-18.478) were significant predictors of in-hospital mortality. Neurological impairment at admission, specific infarct locations, hemiplegia, COPD, and alcohol use are significant predictors of in-hospital mortality among stroke patients. These findings underscore the importance of early neurological assessment, systematic risk stratification, and targeted interventions to improve stroke outcomes in resource-constrained settings.

Keywords

in-hospital mortality predictors stroke tertiary care hospital

Introduction

Stroke is an escalating global public health concern, placing a substantial burden on communities worldwide. According to the Global Burden of Disease (GBD) 2021 estimates, approximately 11.9 million new stroke cases occurred globally, resulting in 7.3 million deaths and 16.5 million disability-adjusted life years (DALYs). Stroke is currently the second leading cause of mortality and the third leading cause of DALYs worldwide.¹ Notably, over 80% of stroke-related deaths occur in low- and middle-income countries (LMICs), underscoring the disproportionate burden faced by resource-constrained settings.^2,3

The increased stroke burden in LMICs is driven by systemic challenges, including inadequate healthcare infrastructure, shortages of trained professionals, inadequate risk-reduction strategies, and poor access to timely interventions.⁴ These issues, compounded by demographic transitions including aging population and rapid urbanization, are projected to further exacerbate stroke incidence and mortality in these regions.

Stroke is broadly categorized into ischemic stroke, caused by cerebral arterial occlusion, and haemorrhagic stroke, resulting from intracranial vessel rupture. Of these, ischemic stroke accounts for 65.3% of cases globally.⁵ Importantly, ischemic stroke is a heterogeneous condition that can be subdivided into 5 major subtypes according to the TOAST classification: large-artery atherosclerosis (atherothrombotic infarct), cardioembolic stroke, small-vessel occlusion (lacunar infarct), infarct of unusual aetiology, and cryptogenic (essential cerebral) infarction.^5,6 Differentiating these subtypes is of substantial clinical relevance, as they vary significantly in their underlying risk factors, stroke severity, recurrence patterns, treatment responses, and prognostic trajectories. Recognition of this heterogeneity is therefore essential for effective prevention strategies, individualized therapeutic decision-making, and accurate prognostication. Recent evidence further underscores that failure to adequately differentiate ischemic stroke subtypes may obscure important variations in clinical outcomes.⁶

In-hospital outcomes among stroke patients are further shaped by sociodemographic and clinical factors, acute medical and neurological complications, and the quality of hospital-based stroke care.⁷ In high-income countries, in-hospital mortality rates typically range from 3% to 11%, while rates in LMICs are considerably higher, ranging from 7% to 15%.⁸

Predictors of in-hospital mortality include stroke subtype, baseline severity at admission, comorbidities (e.g. hypertension, diabetes mellitus, dyslipidaemia), complications such as pneumonia or cerebral oedema, symptom onset-to-admission time, and length of hospital stay. Neurological impairment quantified using standardized scales such as the Modified Rankin Scale (mRS), Glasgow Coma Scale (GCS), and National Institutes of Health Stroke Scale (NIHSS), is strongly associated with in-hospital mortality, as these scores reflect functional status and brain injury severity.^4,9
-11

Standardized tools such as mRS, GCS and NIHSS also allow meaningful comparisons across populations and studies, thereby supporting evidence-based guidelines and improving resources allocation.^4,12 A comprehensive understanding of the predictors of in-hospital mortality is therefore essential to inform clinical practice, optimize patient management, and design targeted interventions to improve outcomes.

Despite the growing global body of evidence on stroke, evidence from Nepal remains limited, particularly regarding in-hospital mortality and its determinants. Hence, this study aims to address this gap by identifying predictors of in-hospital mortality among stroke patients at a tertiary care hospital in Nepal.

Methods

Study Design, Setting and Population

A prospective cohort study was conducted among stroke patients at Annapurna Neurological Institute and Allied Sciences (ANIAS), Maitighar, Kathmandu, Nepal, from November 2023 to April 2024. ANIAS is a 100-bed multidisciplinary hospital established in 2009. It offers advanced care in neurology, neurosurgery, and allied specialties and serves as a neurosurgery residency and international fellowship training centre. ANIAS integrates clinical services, medical education, and research to provide comprehensive, patient-centred healthcare. All patients of either sex aged ≥ 18 years with a confirmed diagnosis of stroke (radiologically confirmed by CT scan and/or MRI) and an identified primary caregiver aged above 18 years were included during their inpatient stay. Patients who were unable to communicate directly (e.g., those with dementia or aphasia), unwilling to participate, uncooperative, or diagnosed with terminal illnesses such as end-stage cancer or renal failure were excluded from the study (Figure 1).

Figure 1.

Sample enrolment.

Sample Size and Sampling Procedure

According to the hospital administration, an average of 50 stroke patients are admitted monthly. With a data collection period of 3 months, a total of 150 stroke patients comprised the study population. Using Yamane’s formula for sample size calculation for a finite population, the sample size was determined as follows: n = N/(1 + N * e²), where N represents the total population (150) and e is the margin of error (0.05). The calculation yielded n = 150/(1 + 150 * (0.05)²) = 150/(1 + 150 * 0.0025) = 150/1.375 ≈ 109.09. Therefore, the calculated sample size was approximately 109. To account for a 10% non-response rate, the final required sample size was adjusted to 120 patients. A convenience sampling technique was used to select the participants for the study.

Data Collection

Data was collected using a semi-structured questionnaire adapted from previous comparable studies and it contained the following 3 sections: sociodemographic information, clinical details and pharmacological management (Supplemental File 1). The pretesting of the questionnaire was carried out among 10% of the respondents, that is, 12 participants, in similar settings. The internal consistency was tested using Cronbach’s alpha coefficient, and the alpha value obtained was .81.

Measures

Outcome Variables

The primary outcome variable was in-hospital mortality following patient admission. All patients were prospectively followed from the time of hospital admission until one of the following endpoints: death during hospitalization, discharge, referral to another facility, or leaving against medical advice (LAMA). In-hospital deaths were identified through daily monitoring of each patient’s clinical status and were confirmed by reviewing medical records, discharge summaries, and official death certificates issued by the attending physician.

Predictor Variables

The degree of disability was assessed using the Modified Rankin Scale (mRS). The mRS was developed and validated by John van Swieten, with scores ranging from 0 to 6. These scores are categorized as follows: 0 (no symptoms), 1 (no significant disability), 2 (slight disability), 3 (moderate disability), 4 (moderately severe disability), 5 (severe disability), and 6 (death), with higher scores indicating greater neurological deficit or more severe physical disability.¹³ The Glasgow Coma Scale (GCS) was used to assess the level of consciousness, with scores ranging from 3 to 15.¹⁴ Patients were categorized as follows: good GCS (13-15) indicating mild brain injury (alert), moderate GCS (9-12) indicating moderate injury (drowsy), and poor GCS (≤8) indicating severe injury (unconscious).

Stroke severity was assessed using the National Institutes of Health Stroke Scale (NIHSS), a standardized tool that quantifies neurological deficits in stroke patients. The scale evaluates consciousness, visual function, motor strength, sensory loss, language, and other neurological functions. Based on established intervals,¹⁵ patients were categorized as follows: NIHSS 0 to 6 as mild, 7 to 12 as moderate, 13 to 20 as severe, and ≥21 as very severe.

The classification of stroke into haemorrhagic and ischaemic types was determined based on computed tomography (CT) scan results. The length of hospital stay was calculated from the time of hospital admission to discharge or death.

Covariates

Age, sex, marital status, occupation, ethnicity, religion, education, residence, smoking status, alcohol consumption, family history, chief complaints infract location and types, and comorbidities were measured as covariates. Age was categorized into three groups: “≤ 40,” “41-60” and “> 60.” Sex was categorized as “male” and “female”, and religion was dichotomized as “Hindu” and “Other than Hindu.” Ethnicity was recoded as “Brahmin/Chhetri” and “other than Brahmin/Chhetri”. Marital status was classified as “ever married”, and “never married”. Occupation was regrouped as “professional worker,” “unskilled/manual labor/daily worker,” “retired”, “homemaker”, and “agriculture/farmer. Education was categorized as “unable to read and write”, “primary (1-5)”, “secondary (6-10)”, “higher secondary (+2 levels)”, and “college/university or above.” Residence was dichotomized as “urban”, and “rural.” Smoking status was categorized as “ever smoked,” and “never smoked,” and alcohol consumption as “ever consumed,” and “never consumed.” Family history was classified as “positive” and “negative.”

The presence of comorbidities was assessed using yes/no questions regarding specific health conditions or diseases, including “hypertension,” “diabetes,” “dyslipidemia,” “chronic obstructive pulmonary disease (COPD),” and “thyroid.” Infarct locations and types were regrouped as “Middle cerebral artery,” “Posterior cerebral artery,” “Internal capsule,” “Lacunar infarcts,” “Capsuloganglionic,” and “No visible infarcts on imaging.”

Statistical Analysis

Data was entered into Epi-Data 3.1 and exported to SAS version 9.4 for analysis. Descriptive statistics, along with bivariate tests of association (Chi-square, Fisher’s exact test, and Likelihood ratio), were performed to explore the distribution of the variables. Kaplan–Meier survival analysis was used to estimate in-hospital survival probabilities. The assumptions of the Cox proportional hazards model were assessed using graphical methods and global goodness-of- fit test or Schoenfeld residual tests. Multicollinearity among predictor variables was evaluated using the variance inflation factor (VIF) and pairwise correlation analysis; all variables in the model exhibited VIF values less than 2, indicating that multicollinearity was not a concern. The Cox proportional hazards regression model was employed to examine the association between predictor variables and in-hospital mortality. Initially, bivariate Cox regression was conducted to assess the relationship between each predictor variable and the outcome. Variables with a p-value ≤ .25 in the bivariate analysis were included in the multivariate Cox regression model. Adjusted hazard ratios (AHRs) with 95% confidence intervals (CIs) were reported, and variables with a p-value < .05 were considered statistically significant predictors of in-hospital mortality.

Results

Among the 120 patients included in the study, two-thirds (68.3%) had experienced an ischemic stroke, while the remaining (31.7%) had a haemorrhagic stroke. Most patients were over 60 years of age (50.0%), male (65.0%), ever married (98.3%), and professional workers (30.0%). In terms of ethnicity, Dalit and Janajati groups constituted a slightly higher proportion (55.0%) compared to Brahmin/Chhetri (45.0%). A considerable number of patients were unable to read and write (44.2%). The majority were Hindu (83.3%) and resided in urban areas (63.3%). Regarding social history, nearly half of the patients had never smoked (48.3%) or consumed alcohol (45.0%), while the remainder were former smokers (51.7%) and alcohol consumers (55.0%). Only a small number reported a family history of stroke (3.3%). Notably, marital status, occupation, educational level, and place of residence were significantly associated with stroke type (p < .05), as shown in Table 1.

Table 1.

Sociodemographic Characteristics and Bivariate Analysis by Types of Strokes (n = 120).

Characteristics	Total, n (%)	Ischaemic stroke, n (%)	Haemorrhagic stroke, n (%)	p-value
Overall	120 (100)	82 (68.3)	38 (31.7)
Age (years)				.119
≤40	15 (12.5)	10 (66.7)	5 (33.3)
41-60	45 (37.5)	26 (57.8)	19 (42.2)
>60	60 (50.0)	46 (76.7)	14 (23.3)
Sex				.90
Male	78 (65.0)	53 (67.9)	25 (32.1)
Female	42 (35.0)	29 (69.0)	13 (31.0)
Marital status				.036*,^a
Ever married	118 (98.3)	82 (69.5)	36 (30.5)
Never married	2 (1.7)	0 (0.0)	2 (100.0)
Occupation (over the last year)				.051*,^b
Professional worker	36 (30.0)	21 (58.3)	15 (41.7)
Unskilled/manual labor/daily worker	31 (25.8)	20 (64.5)	11 (35.5)
Retired	10 (8.3)	8 (80.0)	2 (20.0)
Homemaker	26 (21.7)	17 (65.4)	9 (34.6)
Agriculture/farmer	17 (14.2)	16 (94.1)	1 (5.9)
Ethnicity				.722
Brahmin/Chhetri	54 (45.0)	35 (64.8)	19 (35.2)
Other than Brahmin/Chhetri	66 (55.0)	47 (71.2)	19 (28.8)
Religion				.296
Hindu	100 (83.3)	66 (66.0)	34 (34.0)
Other than Hindu	20 (16.7)	16 (80.0)	4 (20.0)
Education				.045*,^b
Unable to read and write	53 (44.2)	38 (71.7)	15 (28.3)
Primary (1-5)	17 (14.2)	15 (88.2)	2 (11.8)
Secondary (6-10)	19 (15.8)	10 (52.6)	9 (47.4)
Higher secondary (+2 level)	14 (11.7)	7 (50.0)	7 (50.0)
College/University or above	17 (10.8)	12 (61.5)	5 (38.5)
Residence				.031*,^c
Urban	76 (63.3)	50 (65.8)	26 (34.2)
Rural	44 (36.6)	31 (70.5)	13 (29.5)
Smoking status				.276
Never smoked	58 (48.3)	40 (69.0)	18 (31.0)
Ever smoked	62 (51.7)	42 (67.7)	20 (32.3)
Alcohol consumption				.844
Never consumed	54 (45.0)	36 (66.7)	18 (33.3)
Ever consumed	66 (55.0)	46 (69.7)	20 (30.3)
Family history				.777
Positive	4 (3.3)	3 (75.0)	1 (25.0)
Negative	116 (96.7)	79 (68.1)	37 (31.9)

Note. Other than Hindu: Buddhism, Christianity, Muslim; other than Brahmin/Chhetri: Dalit, and Janajati; Ever smoked: occasional or active or former smoker; Ever consumed; occasional or light or heavy or former.

n = frequency.

Fisher exact test.

Likelihood ratio.

Chi-square test.

Statistically significant at p < .05.

Table 2 details the clinical presentation of stroke patients distinguishing between ischaemic and haemorrhagic stroke types. Among the chief complaints reported by patients, headache was significantly more prevalent in those with haemorrhagic stroke (56.3%) compared to ischaemic stroke (43.8%; p = .01). Conversely, weakness in the right limb (82.9% vs 17.1%; p = .028) and urine/bladder incontinence (64.2% vs 35.8%; p = .001) were more frequently observed among ischaemic stroke patients. Other complaints did not show statistically significant differences between the 2 stroke types. Significant associations were also observed in infarct locations and types and comorbidities (p < .01) between ischaemic and haemorrhagic stroke patients. Infarcts in the middle cerebral artery (96.2%), posterior cerebral artery (90.9%), internal capsule (80.0%), and lacular regions (100.0%) were more predominantly linked to ischaemic stroke. Notably, no visible infarct on imaging was observed in 48.3% of patients, including 39.7% of ischaemic stroke and 31.9% of haemorrhagic stroke patients. Hypertension emerged as the most prevalent comorbidity (78.3; p < .01), along with which dyslipidemia (p = .007) and COPD (p = .02) also demonstrated significant associations with stroke type.

Table 2.

Clinical Characteristics and Bivariate Analyses by Type of Stroke (n = 120).

Characteristics	Total, n (%)	Ischaemic stroke, n (%)	Haemorrhagic stroke, n (%)	p-value
Overall	120 (100)	82 (68.3)	38 (31.7)
Chief complaints
Swelling in leg	3 (2.5)	3 (100.0)	0 (0.0)
Vomiting	17 (14.2)	10 (58.8)	7 (41.2)	.363
Headache	32 (26.7)	14 (43.8)	18 (56.3)	.01*,^b
Nausea	7 (5.8)	3 (42.9)	4 (57.1)	.135
Weakness in left limb	45 (37.5)	31 (68.9)	14 (31.1)	.919
Disorientation	23 (19.2)	14 (60.9)	9 (39.1)	.392
Weakness in right limb	35 (29.2)	29 (82.9)	6 (17.1)	.028*,^b
Slurring of speech	52 (43.3)	39 (75.0)	13 (25.0)	.170
Facial deviation	17 (14.2)	14 (82.4)	3 (17.6)	.180
Loss of consciousness	13 (10.8)	11 (84.6)	2 (15.4)	.181
Frothing from mouth	7 (5.8)	6 (85.7)	1 (14.3)	.308
Urine/bladder incontinence	14 (11.6)	9 (64.2)	5 (35.8)	.001*,^b
Hemiplegia	61 (50.8)	45 (73.8)	16 (26.2)	.193
Infarct locations and types				.01*,^a
Middle cerebral artery	27 (22.5)	25 (96.2)	1 (3.8)
Posterior cerebral artery	11 (9.2)	10 (90.9)	1 (9.1)
Internal capsule	5 (4.2)	4 (80.0)	1 (20.0)
Lacunar infarcts	10 (8.3)	10 (100.0)	0 (0.0)
Capsuloganglionic	3 (2.5)	3 (100.0)	35 (60.3)
No visible infarct on imaging	58 (48.3)	23 (39.7)	38 (31.9)
Comorbidities
Hypertension	94 (78.3)	64 (68.1)	30 (31.9)	<.01^b
Diabetes	28 (23.3)	21 (75.0)	7 (25.0)	.162
Dyslipidemia	6 (5.0)	4 (66.7)	2 (33.3)	.007^a
COPD	7 (5.8)	6 (85.7)	1 (14.3)	.02^a
Thyroid	8 (6.7)	7 (87.5)	1 (12.5)	.069
mRS at the time of admission
Mild (mRS ≤ 2)	12 (10.0)	8 (66.7)	4 (33.3)	.045*,^a
Moderate to severe (mRS ≥ 3)	108 (90.0)	74 (68.5)	34 (31.5)
GCS at the time of admission
Mild	78 (65.0)	53 (68.0)	25 (32.0)	.332
Moderate	30 (25.0)	22 (73.3)	8 (26.7)
Severe	12 (10.0)	7 (58.3)	5 (41.7)
NIHSS at the time of admission
Mean ± SD	12.33 ± 9.07	11.57 ± 5.06	13.99 ± 7.07
Mild	23 (19.2)	18 (78.3)	5 (21.7)	.155
Moderate	50 (41.7)	35 (70.0)	15 (30.0)
Severe	25 (20.8)	15 (60.0)	10 (40.0)
Very severe stroke	15 (12.5)	9 (60.0)	6 (40.0)
Length of hospital stay
1-10	82 (68.4)	54 (65.9)	28 (34.1)	0.137
11-20	32 (26.6)	22 (68.8)	10 (31.3)
>20	6 (5.0)	6 (100.0)	0 (0.00)

n = frequency; SD = standard deviation; COPD = chronic obstructive pulmonary disease; GCS = Glasgow coma scale; mRS = modified rank score; NIHSS = National institute of health stroke; Ref = reference.

Fisher exact test

Chi-square test.

Statistically significant at p < .05.

The mRS at admission revealed a statistically significant difference between groups (p = .045), with 90% of patients exhibiting moderate to severe disability. In contrast, no such significant differences were observed in GCS (p = .332) and NIHSS (p = .155) scores between stroke types. Although most patients (68.4%) stayed at the hospital for 1 to 10 days, the length of hospital stay did not differ significantly by stroke type (p = .137).

Bivariate and multivariate Cox regression analysis identified several predictors of in-hospital mortality among admitted stroke patients. Patients presenting with hemiplegia had a significantly increased risk of mortality (AHR: 6.70; 95% CI: 1.835-53.748; p = .047). GCS score < 8 emerged as a strong predictor (AHR: 12.36; 95% CI: 2.73-56.00; p = .001), as like a NIHSS score ≥ 12 (AHR: 14.75; 95% CI:3.01-72.28). Similarly, patients with moderate to severe disability at admission (mRS ≥ 3) had a significantly higher mortality risk (AHR: 9.92; 95% CI: 1.10-89.24; p = .041). In terms of radiological findings, territorial infarcts (AHR: 26.33; 95% CI: 2.093-331.203; p = .011) and capsuloganglionic infarcts (AHR: 14.6; 95% CI: 1.819-160.877; p = .048) were significantly associated with increased mortality. The presence of COPD (AHR: 2.48; 95% CI: 1.317-45.091; p = .039) and a history of alcohol consumption (AHR: 3.87; 95% CI: 1.014-18.478; p = .013) were significantly associated with increased in-hospital mortality. Other variables including age, stroke type, hypertension, diabetes, thyroid disorders, smoking status, and other infarct locations, did not show statistically significant associations after adjustment, as shown in Table 3.

Table 3.

Bivariate and Multivariate Cox Regression Analysis Identifying Predictors of In-Hospital Mortality Among Stroke Patients (n = 120).

Characteristics	Deadn	Surviven	CHR [95% CI]	p-value	AHR [95% CI]	p-value
Age (years)
>50	1	31	Ref.		Ref.
≤50	10	78	0.937[0.282 to 3.104]	.0613	1.05[0.293 to 3.764]	.078
Type of Stroke
Haemorrhagic	2	36	Ref.		Ref.
Ischaemic	9	73	1.49[0.313 to 7.100]	.617	1.53[0.292 to 8.105]	.612
Hemiplegia
No	1	58	Ref.		Ref.
Yes	10	51	6.73[2.839 to 54.036]	.053	6.70[1.835 to 53.748]	.047*
Hypertension
No	1	25	Ref.		Ref.
Yes	10	84	1.86 [0.231 to 14.958]	.561	1.94[0.237 to 15.764]	.538
Diabetes
No	8	84	Ref.		Ref.
Yes	3	25	0.98[0.259 to 3.717]	.977	0.94[0.246 to 3.594]	.929
Thyroid
No	10	102	Ref.		Ref.
Yes	1	7	1.89 [0.234 to 15.334]	.550	1.78 [0.218 to 14.634]	.589
COPD
No	10	103	Ref.		Ref.
Yes	1	6	2.64[1.53 to 21.344]	.036*	2.48[1.317 to 45.091]	.039*
NIHSS
≤12	2	82	Ref.		Ref.
≥12	9	27	13.66 [2.85to 65.45]	.001*	14.75 [3.01 to 72.28]	.001*
GCS score
<8	8	21	10.25 [2.31 to 45.57]	.002*	12.36 [2.73 to 56.00]	.001*
9-12	2	11	4.82 [0.45 to 51.61]	.193	5.10 [0.50 to 52.10]	.171
13-15	1	77	Ref.		Ref.
mRS at hospital arrival
Mild (≤2)	0	12	Ref.		Ref.
Moderate to severe (≥3)	11	97	8.76 [1.03 to 74.54]	.047	9.92 [1.10 to 89.24]	.041*
Infarct Locations and Types
Territorial	7	36	14.4 [1.961 to 106.432]	.009*	26.33[2.093 to 331.203]	.011*
Capsuloganglionic	1	2	8.06 [2.708 to 81.849]	.008*	14.6[1.819 to 160.877]	.048*
Internal Capsule	1	4	4.04[0.357 to 45.981]	.251	6.21[0.446 to 86.427]	.174
No Visible Infarct on Imaging	2	56	Ref.		Ref.
Smoking status
Never Smoked	7	51	Ref.		Ref.
Ever smoked	4	58	1.17[0.325 to 4.253]	.804	1.19[0.331 to 4.313]	.786
Alcohol consumption
Never consumed	9	45	Ref.		Ref.
Ever consumed	2	64	3.79 [1.797 to 18.100]	.014*	3.87[1.014 to 18.478]	.013*

Note: AHR: Adjusted hazard ratio, CHR: Crudes hazard ratio, CI: Confidence level; GCS: Glasgow coma scale, mRS: Modified rank score: Ref: Reference; COPD: chronic pulmonary obstructive disease.

Statistically significant at p < .05.

Kaplan–Meier survival analysis showed that stroke patients had a mean survival time of 35.08 days (95% CI: 29.31-40.85). The cumulative survival probability was approximately 96% by Day 5, decreasing to about 95% by day 10. By Day 15, the survival probability dropped to nearly 90% and further declined to approximately 80% by Day 20. By Day 30, survival probability fell to about 70% and plunged below 50% after Day 40. This pattern indicates a progressive decline in survival over time, suggesting that longer hospitalization was associated with increased mortality risk among stroke patients, as shown in Figure 2.

Figure 2.

Kaplan–Meier survival curve of stroke patients by hospital stay duration.

Discussion

This prospective cohort study assessed key predictors of in-hospital mortality among stroke patients at a tertiary neurological hospital in Nepal. Several significant predictors of in-hospital mortality were identified, including an NIHSS score ≥ 12, GCS score < 8, moderate to severe disability (mRS score ≥ 3) on admission, territorial and capsuloganglionic infarcts, hemiplegia, COPD, and alcohol consumption. Details of pharmacotherapeutic management during hospitalization are provided in Supplemental File 2.

A significant finding of the study was the predominance of ischemic stroke, consistent with global epidemiological trends.^16,17 The increasing burden of stroke in low- and middle-income countries (LMICs) is likely driven by longer life expectancy and a shift in disease patterns toward non-communicable diseases (NCDs).¹⁸ Demographic factors such as marital status, education level, occupation, and place of residence were significantly associated with stroke type. Clinically, symptoms such as headache and right limb weakness, as well as infarct location and type, also showed significant associations. Moreover, comorbid conditions, particularly hypertension and dyslipidemia, complications like COPD, and mRS score at admission were significantly linked to stroke type.

The study reported a hospital survival rate of 91.0%, comparable to rates in Pakistan (91.0%) and Kenya (93.8%),^19,20 and higher than rates reported in Ethiopia (78.4% and 59.18%).^4,7 The mean survival time among patients who died during hospitalization was 35 days, substantially longer than the median survival times of 4.5 and 6 days reported in previous studies.^4,7 These differences may reflect variations in complication rates, healthcare infrastructure, the presence of specialized stroke units, and timely access to quality care. The in-hospital mortality rate of 9.0% observed in this study aligns with reports from other low-resource settings such as Pakistan (9%) and Kenya (6.2%), but is lower than rates reported in Ethiopia (21.6%), Tanzania (33.3%), and Zambia (40%).^21
-23 In contrast, studies from high-income western countries consistently report lower mortality rates. For instance, a large German stroke registry reported an in-hospital mortality of 4.5% among ischemic stroke patients, while national registries from the United States and other European countries typically report rates ranging from 3% to 7%.^6,24,25 These differences highlight the influence of geography and healthcare infrastructure on stroke outcomes. Western healthcare systems benefit from organized stroke units, rapid neuroimaging, early repurfusion therapies (IV-tPA and thrombectomy), and multidisciplinary rehabilitation services, all of which contribute to reduced mortality.

Several key predictors of in-hospital stroke mortality were identified, primarily reflecting stroke severity. Patients with GC scores ≤ 8, NIHSS scores ≥ 12, and mRS scores ≥ 3 had a significantly higher risk of mortality. These findings are consistent with existing evidence indicating that poor neurological status at admission strong predicts adverse outcomes.^4,5,26 Such impaired scores often reflect decreased consciousness and extensive brain injury, frequently involving the brainstem or large haemorrhagic regions.^4,27 Radiologically, capsuloganglionic and territorial infarcts were strongly associated with in-hospital mortality. These infarct types indicate extensive cerebral damage and poor collateral circulation, affecting critical brain regions and leading to worse functional outcomes and higher fatality rates.²⁸

Additionally, clinical features such as hemiplegia and chronic COPD were significant predictors of mortality. This is consistent with previous studies showing that functional impairments and underlying chronic systemic conditions exacerbate stroke severity and impede recovery.^4,5,7,29 Alcohol consumption was also significantly associated with increased in-hospital mortality risk, in line with literature linking alcohol use to poorer neurological outcomes, likely due to pro-inflammatory responses and elevated risk of haemorrhagic transformation.^30,31

Interestingly, age, stroke subtype (ischemic vs. haemorrhagic), and common comorbidities such as hypertension and diabetes did not independently predict mortality after adjustment. This suggests that functional and neurological status at admission may be more critical in determining immediate outcomes than the presence of chronic conditions.

This study has several notable strengths. It is one of the few prospective cohort studies from Nepal to systematically examine predictors of in-hospital mortality in stroke patients. The use of standardized clinical assessment tools such as mRS, GCS, and NIHSS, along with radiological confirmation, strengthened diagnostic accuracy and allowed robust risk stratification. Additionally, Kaplan-Meier survival analysis and Cox proportional hazards modelling provided reliable estimates of survival and predictors of mortality.

Nevertheless, several limitations should be acknowledged. Being a single-centre study with a modest sample size, the findings may not be generalizable to other healthcare settings. The use of convenience sampling may have introduced selection bias. Long-term outcomes beyond hospital discharge were not assessed, limiting insights into post-discharge survival and functional recovery. Important prognostic variables such as time-to-treatment, thrombolysis use, structural heart disease, and differentiation of ischemic subtypes (lacunar vs. non-lacunar, cardioembolic) were not systematically evaluated. Finally, while the study was adequately powered for major predictors, the small number of deaths may have limited the precision of effect estimates for less common variables.

Conclusion

This study found a 9% in-hospital mortality rate among stroke patients, with mortality strongly predicted by severe neurological impairment (GCS < 8, NIHSS ≥ 12, mRS ≥ 3), hemiplegia, territorial and capsuloganglionic infarcts, COPD, and alcohol consumption. These findings underscore the need for early neurological assessment, systematic risk stratification, and targeted interventions to reduce stroke-related death in resource-limited settings. Future research should focus on long-term post-discharge outcomes, the impact of acute interventions such as thrombolysis and stroke unit care, and the role of stroke subtypes and cardiac comorbidities in prognosis. Larger multicentre studies are warranted to validate these predictors and inform evidence-based stroke care in South Asia.

Supplemental Material

sj-docx-1-inq-10.1177_00469580251385397 – Supplemental material for Predictors of In-Hospital Mortality Among Stroke Patients at a Tertiary Care Hospital in Nepal: A Prospective Cohort Study

Supplemental material, sj-docx-1-inq-10.1177_00469580251385397 for Predictors of In-Hospital Mortality Among Stroke Patients at a Tertiary Care Hospital in Nepal: A Prospective Cohort Study by Saru Panthi, Sabina Sankhi, Bibek Bhandari, Shishir Paudel and Nirmal Raj Marasine in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Supplemental Material

sj-docx-2-inq-10.1177_00469580251385397 – Supplemental material for Predictors of In-Hospital Mortality Among Stroke Patients at a Tertiary Care Hospital in Nepal: A Prospective Cohort Study

Supplemental material, sj-docx-2-inq-10.1177_00469580251385397 for Predictors of In-Hospital Mortality Among Stroke Patients at a Tertiary Care Hospital in Nepal: A Prospective Cohort Study by Saru Panthi, Sabina Sankhi, Bibek Bhandari, Shishir Paudel and Nirmal Raj Marasine in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Footnotes

Acknowledgements

We sincerely thank the Annapurna Neurological Institute and Allied Sciences Hospital, and all the patients who took part in this study for their invaluable time and support. This study would not have been possible without their contribution

ORCID iDs

Sabina Sankhi

Nirmal Raj Marasine

Ethical Considerations

Ethical approval was taken from the Institutional Review Committee of Annapurna Neurological Institute & Allied Sciences (IRC-ANIAS, Ref No.: 115-2023/24), the study followed the Helsinki Ethical Guidelines.

Consent to Participate

Written informed consent was obtained from each patient or their caretaker prior to data collection. The collected data were used solely for research purposes. Participation was entirely voluntary, and participants were free to withdraw at any time without any consequences.

Author Contributions

SaP: Conceptualization, methodology, validation, project administration, data curation, resources, Writing – review & editing; SS: methodology, validation, formal analysis, writing – original draft, Writing – review & editing; BB: Writing – review & editing; SP: methodology, validation, Writing – review & editing; NRM: Conceptualization, methodology, validation, formal analysis, supervision, visualization, project administration, resources, writing – original draft, Writing – review & editing.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Supplemental Material

Supplemental material for this article is available online.

References

Feigin

Abate

. Global, regional, and national burden of stroke and its risk factors, 1990-2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Neurol. 2024;23:973-1003.

Johnson

Nguyen

Roth

. Global, regional, and national burden of stroke, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016 Lancet Neurol. 2019;18:439-458.

O'Donnell

Chin

Rangarajan

, et al Global and regional effects of potentially modifiable risk factors associated with acute stroke in 32 countries (INTERSTROKE): a case-control study. Lancet. 2016;388:761-775.

Fekadu

Chelkeba

Kebede

. Burden, clinical outcomes and predictors of time to in hospital mortality among adult patients admitted to stroke unit of Jimma University Medical Center: a prospective cohort study BMC Neurol. 2019;19(1):213.

Feigin

Brainin

Norrving

, et al World stroke organization: global stroke fact sheet 2025. Int J Stroke. 2025;20(2):132-144.

Gasull

Arboix

. Molecular mechanisms and pathophysiology of acute stroke: emphasis on biomarkers in the different stroke subtypes. Int J Mol Sci. 2022;23(16):9476.

Deresse

Shaweno

. Epidemiology and in-hospital outcome of stroke in South Ethiopia. J Neurol Sci. 2015;355:138-142.

Koennecke

Belz

Berfelde

, et al Factors influencing in-hospital mortality and morbidity in patients treated on a stroke unit. Neurol. 2011;77:965-972.

Greffie

Mitiku

Getahun

. Risk factors, clinical pattern and outcome of stroke in a referral hospital, Northwest Ethiopia. Clin Med Res. 2015;4:182-188.

10.

Labodi

Kadri

Valentin

, et al Intra-hospital mortality of stroke and its predictive factors in a reference hospital in Ouagadougou, Burkina Faso. Brain. 2017;1:1-6.

11.

Admas

Teshome

Petrucka

Telayneh

Alamirew

. In-hospital mortality and its predictors among adult stroke patients admitted in Debre Markos Comprehensive Specialized Hospital, Northwest Ethiopia. Sage Open Med. 2022;10:20503121221122465

12.

Siniscalchi

. Use of stroke scales in clinical practice: current concepts. Turk J Emerg Med. 2022;22:119-124.

13.

Broderick

Adeoye

Elm

. Evolution of the modified Rankin scale and its use in future stroke trials. Stroke. 2017;48:2007-2012.

14.

Teasdale

Jennett

. Assessment of coma and impaired consciousness: a practical scale. Lancet. 1974;2:81-84.

15.

Brott

Adams

Jr Olinger

, et al Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989;20:864-870.

16.

Johnson

Onuma

Owolabi

Sachdev

. Stroke: a global response is needed. Bull World Health Organ. 2016;94:634-634A.

17.

Neshuku

Kirchner-Frankle

Nangolo

, et al The burden and In-hospital mortality of stroke admissions at a tertiary level hospital in Namibia: a retrospective cohort study. Stroke Res Treat. 2023;2023:1-12.

18.

Temesgen

Teshome

Njogu

. Treatment outcomes and associated factors among hospitalized stroke patients at Shashemene Referral Hospital, Ethiopia. Stroke Res Treat. 2018;2018:8079578.

19.

Masood

Hussain

Abbasi

. Clinical presentation, risk factors and outcome of stroke at a district level teaching hospital. J Ayub Med Coll Abbottabad. 2013;25:49-51.

20.

Jowi

Mativo

. Pathological sub-types, risk factors and outcome of stroke at the Nairobi Hospital, Kenya. East Afr Med J. 2008;85(12):572-581.

21.

Alhazzani

Mahfouz

Abolyazid

, et al

In hospital stroke mortality: rates and determinants in Southwestern Saudi Arabia

Int J Environ Res Public Health. 2018;15:927.

22.

Okeng'o

Chillo

Gray

Walker

Matuja

. Early mortality and associated factors among patients with stroke admitted to a large teaching hospital in Tanzania. J Stroke Cerebrovasc Dis. 2017;26:871-878.

23.

Atadzhanov

Mukomena

Lakhi

Ross

Meschia

. Stroke characteristics and outcomes of adult patients admitted to the University Teaching Hospital, Lusaka, Zambia. Open Gen Intern Med J. 2012;5:3-8.

24.

Reeves

Fonarow

Zhao

Smith

Schwamm

. Quality of care in women with ischemic stroke in the GWTG program. Stroke. 2009;40(4):1127-1133.

25.

Heuschmann

Kolominsky-Rabas

Roether

, et al Predictors of in-hospital mortality in patients with acute ischemic stroke treated with thrombolytic therapy. JAMA. 2004;292(15):1831-1838.

26.

Jauch

Saver

Adams

Jr , et al Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2013;44:870-947.

27.

Banks

Marotta

. Outcomes validity and reliability of the modified Rankin scale: implications for stroke clinical trials: a literature review and synthesis. Stroke. 2007;38:1091-1096.

28.

Hankey

. Stroke. Lancet. 2017;389(10069):641-654.

29.

Sun

. Evaluation of Immunomodulatory Mediators in Chronic Obstructive Pulmonary Disease and Cardiovascular Disease. Doctoral dissertation. Sidney Sussex College and Division of Experimental Medicine and Immunotherapeutics; 2022.

30.

Kim

Park

Chang

Kim

Lee

. Inflammation after ischemic stroke: the role of leukocytes and glial cells. Exp Neurobiol. 2016;25:241-251.

31.

Kolonne

Mudalige

Dissanayaka

, et al Alcohol consumption and stroke mortality: global patterns, risks, and public health implications. Int J Ment Health Addict. 2025;1-4.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

0.03 MB