Heart-Rate-to-Blood-Pressure Ratios Correlate with Malignant Brain Edema and One-Month Death in Large Hemispheric Infarction: A Cohort Study

Introduction: Large hemispheric infarction (LHI) can lead to fatal complications such as malignant brain edema (MBE). We aimed to investigate the correlation between heart-rate-to-blood-pressure ratios and MBE or one-month death after LHI. Methods: We prospectively included LHI patients from a registered cohort. Hourly heart-rate-to-blood-pressure ratios were recorded as a variation of the traditional shock index (SI), SIs and SId (systolic and diastolic pressures, respectively), and calculated for mean and variability (standard deviation) in 24 h and two 12 h epochs (1–12 h and 13–24 h) after onset of symptoms. MBE was defined as neurological deterioration symptoms with imaging evidence of brain swelling. We employed a generalized estimating equation to compare the trend in longitudinal collected SIs and SId between patients with and without MBE. We used multivariate logistic regression to investigate the correlation between SIs, SId and outcomes. Results: Of the included 162 LHI patients, 28.4% (46/162) developed MBE and 25.3% (40/158) died within one month. SIs and SId increased over baseline in all patients, with a similar ascending profile during the first 12 h epoch and a more intensive increase in the MBE group during the second 12 h epoch (p < 0.05). During the overall 24 h, patients with greater SId variability had a significantly increased MBE risk after adjustment (OR 3.72, 95%CI 1.38–10.04). Additionally, during the second 12 h epoch (13–24 h after symptom onset), patients developing MBE had a significantly higher SId level (OR 1.18, 95%CI 1.00–1.39) and greater SId variability (OR 3.16, 95%CI 1.35–7.40). Higher SId and greater SId variability within 24 h independently correlated with one-month death (all p < 0.05). Within the second 12 h epoch, higher SIs, higher SId and greater SId variability independently correlated with one-month death (all p < 0.05). No significant correlation was observed in the first 12 h epoch. Conclusions: Higher and more fluctuated heart-rate-to-blood-pressure ratios independently correlated with MBE development and one-month death in LHI patients, especially during the second 12 h (13–24 h) epoch after onset.


Introduction
Stroke is the leading cause of death and disability worldwide [1,2], and the most common type of stroke is acute ischemic stroke [2]. Large hemispheric infarction (LHI) is the very ischemic stroke subtype with the poorest prognosis [3]. LHI patients may develop rapidly progressing malignant brain edema (MBE), likely resulting in an impaired state of consciousness, herniation, severe independence or death [4]. The fatality of MBE under conservative intensive care is around 80% [5,6], with medication or surgery having limited effectiveness [3,5]. Therefore, it is necessary to prevent the occurrence of MBE to reduce fatality and improve the prognosis of LHI.
The autoregulation of cerebral blood flow is injured after stroke, increasing the dependence of cerebral blood supply on systemic blood pressure [7,8]. The formation of MBE mainly results from disruption of the blood-brain barrier, which physiologically correlates with blood pressure [4,5,9]. Some previous studies explored the correlation between blood pressure and prognosis of general stroke [10,11], but investigation into LHI patients is limited. Moreover, few studies have examined the impact of hemodynamic indexes that physiologically link to blood pressure, for example, the heart rate.
The shock index (SI), calculated as the heart rate divided by systolic blood pressure, is a commonly used index in intensive care [12]. Recent studies showed that admission SI at extremities appeared to predict poor outcomes in general stroke [13,14]. Here, we hypothesize that SI may correlate with MBE or related fatality in LHI patients. No study has explored this correlation. Moreover, alteration of HR affects diastolic blood pressure more significantly than systolic pressure, and diastolic pressure takes up a greater proportion of the mean artery pressure, which determines cerebral blood flow. It is reasonable to suspect that the heart-rate-to-diastolic-pressure rate may correlate with, if not more significantly than the traditional shock index, MBE or related fatality after LHI.
The natural history of brain edema can be categorized into cytotoxic and vasogenic edema. Cytotoxic edema is defined as cell swelling caused by intracellular fluid accumulation and is initially observed within hours after stroke onset and then declined within 1 day. Vasogenic edema is caused by extracellular fluid accumulation resulting through disruption of the blood-brain barrier and is usually developed within two to three days after onset and maintained for several days [9]. Therefore, the influence of heart-rate-to-blood-pressure ratios may vary at different recording epochs.
In this research, we aimed to explore the correlation between heart-rate-to-bloodpressure ratios and MBE, as well as one-month death after LHI.

Study Population
We enrolled patients with acute ischemic stroke consecutively admitted between January 2020 and December 2021 from our registry cohort database [15,16]. Acute ischemic stroke was diagnosed according to established guidelines [17]. LHI was defined as infarction with visible hypodensity over a third of the middle cerebral artery (MCA) territory in computerized tomography and/or magnetic resonance imaging within 6 h of onset, or over half the territory of MCA within 48 h of onset [18]. All patients completed computerized tomography on admission. Follow-up brain imaging was performed within 7 days of admission or when clinical worsening occurred. LHI adult patients admitted within 24 h of onset were included. Patients were excluded if they: (1) had bilateral ischemia, (2) had recurrent stroke, (3) had parenchymal hemorrhage type 2 [19] occurring before or at follow-up imaging, and/or (4) had no blood pressure or heart rate measurements.

Data Collection
We collected demographic and clinical data including age, gender, body temperature, National Institute of Health stroke scale (NIHSS) and Glasgow Coma Scale (GCS) on admission, hypertension, diabetes mellitus, hyperlipidemia, atrial fibrillation and stenosis/occlusion of intra/extracerebral arteries; ischemic area (whether invisible hypodensity ≥ 1/2 territory of MCA), Trial of ORG 10,172 in Acute Stroke Treatment (TOAST) classification [20]. Treatments including thrombolysis, thrombectomy, and dehydration treatment were documented. Any in-hospital infection, including pneumonia, urinary infection or others, was also recorded.

Heart-Rate-to-Blood-Pressure Ratios Measurement
We collected repeated time-stamped blood pressure and heart rate of all included patients within the first 24 h post-onset. Blood pressure and heart rate were measured simultaneously by noninvasive monitoring devices and were recorded by trained nurses hourly (±30 min) after admission. The relationship between heart rate and diastolic blood pressure can be represented as a variation of the traditional shock index, wherein heart-rate-to-blood-pressure ratios are assigned as SI s and SI d for systolic and diastolic pressures, respectively. The equations for SI s and SI d are derived by dividing heart rate by systolic or diastolic blood pressure. Hourly SI s and SI d were recorded and calculated for mean, maximum, minimum, range (difference between maximum and minimum), standard deviation (SD) and coefficient of variation (CV) for the entire recording epoch (24 h) and during two 12 h epochs (1-12 h and 13-24 h after onset). For patients who developed MBE within 24 h of onset, the heart-rate-to-blood-pressure ratios collected after MBE development were excluded for analysis in order to ensure the appropriate sequential order of studied parameters and outcome.

Outcomes
The primary outcome was MBE, defined as midline shift ≥ 5 mm with a deteriorated status of consciousness or anisocoria, or indications for decompressive craniectomy. All images were independently viewed by two researchers (X.S and W.G) who were blinded to clinical data. A final decision was made by a senior neurologist (M.L) if two researchers could not reach a consensus. Presence of mild degree of hemorrhagic transformation was permitted, since it was not considered to result in mass effect [21]. Secondary outcomes were one-month death and three-month poor functional outcome (modified Rankin scale (mRS) ≥ 3). Patients or, if not possible, their relatives were followed up by telephone interview at one and three months after stroke. Failure to contact after three different attempts was recorded as lost to follow-up.

Statistical Analysis
All data were reported as mean ± standard deviation (SD), median (interquartile range, IQR) or count (percentage). Inter-group differences were compared by χ2 test or Fisher's exact test for categorical variables, and Student's t-test or Mann-Whitney U test for continuous variables as appropriate. We employed the generalized estimating equation (GEE) to compare the trend in longitudinal collected SI s and SI d measurement values between patients with and without MBE, as well as between patients who survived and those who died within one month. This method allowed us to account for the correlation between the repeated measurements over time, improving the accuracy of our parameter estimates and providing scientific rigor to our analysis. We used multivariable logistic regression to analyze independent correlation between SI s or SI d parameters and outcomes. SD and CV values of SI s and SI d were entered into the regression model as log-transformed values due to the skewness distribution of data, while other SI s and SI d parameters were entered as ten-fold values (per 0.1-unit increase). We preselected the adjusted covariates with clinical significance for MBE (age, body temperature, NIHSS, ischemic area, hypertension, atrial fibrillation, dehydration treatment, and in-hospital infection), and one-month death/three-month poor outcome (age, body temperature, NIHSS, ischemic area, hypertension, atrial fibrillation, and in-hospital infection). Cumulative risk of one-month death was estimated with Kaplan-Meier curves and compared by log-rank test. We further conducted a restricted cubic spline regression model to assess the non-linear dose effect of SI s or SI d parameters on outcomes with four knots (at the 5th, 35th, 65th, and 95th percentiles) [22]. We determined the cutoff of SI s or SI d that best predicted MBE and one-month death based on receiver operating characteristic curves and the Youden index. All statistical analyses were performed by SPSS (version 26.0; IBM, Chicago, IL, USA), STATA 16.0 (Corporation, College Station, TX, USA), and GraphPad Prism (version 8.01; GraphPad Software, San Diego, CA, USA). A two-sided p < 0.05 was considered statistically significant.

Temporal Evolution of Heart-Rate-to-Blood-Pressure Ratios
We included 2138 sets of time-stamped blood pressure and heart rate measurements in total. For up to 24 h after onset, 14 ± 6 sets of blood pressure and heart rate were recorded per patient, 6 ± 3 and 8 ± 3 for the first and second 12 h epoch, respectively. Missing data were mainly due to pre-hospital referral and transferring for thrombectomy or follow-up imaging.
Temporal evolution of SI s and SI d in LHI patients with and without MBE are shown in Figure 1. SI s and SI d increased in the early phase (first 12 h epoch, 1-12 h after symptom onset) for both the MBE group and non-MBE group, then reached a plateau for the remainder of the monitoring epoch in the non-MBE group. In contrast, SI s and SI d continued to rise at a lower rate in the MBE group, beginning at approximately 13 h after onset. During the second 12 h epoch (13-24 h after symptom onset), in patients with MBE, SI s and SI d were higher than that in patients without MBE (p < 0.001 for SI s , p = 0.036 for SI d by GEE method for MBE). We did not find differences in SI s and SI d during the entire 24 h recording period or during the first 12 h epoch between patients with and without MBE (all p > 0.05 for SI s and SI d by GEE method for MBE). The temporal evolutions of SI s an SI d were similar in patients who died or had survived at one month ( Figure S2).

Correlation of Heart-Rate-to-Blood-Pressure Ratios with Malignant Brain Edema
The clinical features and heart-rate-to-blood-pressure ratio parameters in patients with and without MBE were compared in Tables 1 and 2 to rise at a lower rate in the MBE group, beginning at approximately 13 h after onset. During the second 12 h epoch (13-24 h after symptom onset), in patients with MBE, SIs and SId were higher than that in patients without MBE (p < 0.001 for SIs, p = 0.036 for SId by GEE method for MBE). We did not find differences in SIs and SId during the entire 24 h recording period or during the first 12 h epoch between patients with and without MBE (all p > 0.05 for SIs and SId by GEE method for MBE). The temporal evolutions of SIs an SId were similar in patients who died or had survived at one month ( Figure S2).
Diagnostics 2023, 13, x FOR PEER REVIEW 5 of 14 Figure 1. The 24 h temporal evolution of heart-rate-to-blood-pressure ratios in LHI patients with and without MBE. (a) Temporal evolution of SIs. (b) Temporal evolution of SId. During 13-24 h after onset, in patients with MBE, SIs and SId were higher than that in patients without MBE (p < 0.001 for SIs, p = 0.036 for SId by GEE method for MBE). LHI, large hemispheric infarction; MBE, malignant brain edema; SI, shock index; GEE, generalized estimating equation; heart-rate-to-blood-pressure ratios are assigned as SIs and SId for systolic and diastolic pressures, respectively. * p < 0.05; ** p < 0.001.

Correlation of Heart-Rate-to-Blood-Pressure Ratios with Malignant Brain Edema
The clinical features and heart-rate-to-blood-pressure ratio parameters in patients with and without MBE were compared in Tables 1 and 2  The 24 h temporal evolution of heart-rate-to-blood-pressure ratios in LHI patients with and without MBE. (a) Temporal evolution of SI s . (b) Temporal evolution of SI d . During 13-24 h after onset, in patients with MBE, SI s and SI d were higher than that in patients without MBE (p < 0.001 for SI s , p = 0.036 for SI d by GEE method for MBE). LHI, large hemispheric infarction; MBE, malignant brain edema; SI, shock index; GEE, generalized estimating equation; heart-rate-to-bloodpressure ratios are assigned as SI s and SI d for systolic and diastolic pressures, respectively. * p < 0.05; ** p < 0.001.    According to restricted cubic spline regression, during the second 12 h epoch after onset, the correlation between the mean SI d level and MBE were in a non-linear dose-effect manner ( Figure 2). The Youden index identified a mean SI d level of 1.11 bpm/mmHg during the second 12 h epoch after onset (13-24 h after onset) as able to discriminate between patients with and without MBE. The area under the receiver operating characteristic curve was 0.61 bpm/mmHg, sensitivity was 78%, and specificity was 47%. According to restricted cubic spline regression, during the second 12 h epoch after onset, the correlation between the mean SId level and MBE were in a non-linear dose-effect manner ( Figure 2). The Youden index identified a mean SId level of 1.11 bpm/mmHg during the second 12 h epoch after onset (13-24 h after onset) as able to discriminate between patients with and without MBE. The area under the receiver operating characteristic curve was 0.61 bpm/mmHg, sensitivity was 78%, and specificity was 47%.  Restricted cubic spline regression analysis to explore the dose effect of mean level of SI d 13-24 h after onset on MBE; solid black line stands for odds ratios, and the area between dotted black lines for 95% confidence intervals is based on standard errors. Red horizontal lines reflect grid lines when the odds ratio equals 1. Odds ratios were adjusted for age, body temperature, National Institute of Health stroke scale, ischemic area, hypertension, atrial fibrillation, dehydration treatment, and in-hospital infection. MBE, malignant brain edema; OR, odds ratio; SI, shock index; SI d was calculated as heart rate divided by diastolic blood pressure.

Correlation of Heart-Rate-to-Blood-Pressure Ratios with Secondary Outcomes
Table S1 shows the clinical features for patients who died or survived within one month. Patients who died within one month had greater variability of heart-rate-to-bloodpressure ratios than those who survived within one month during the overall 24 h recording epoch (all p < 0.05, Table S2). During the 13-24 h epoch, higher SI s (0.71 bpm/mmHg vs. 0.64 bpm/mmHg, p = 0.049) and SI d (1.26 bpm/mmHg vs. 1.13 bpm/mmHg, p = 0.013) levels were observed in patients who died within one-month (Table S2).
No SI s or SI d parameters in the first 12 h epoch presented a significant correlation with one-month death. During the second 12 h epoch, a higher SI s level (OR 1.28, 95%CI 1.00-1.65, p = 0.049) independently correlated with one-month death. Meanwhile, a higher SI d level (OR 1.25, 95%CI 1.06-1.48, p = 0.009), a higher maximum SI d value (OR 1.16, 95%CI 1.05-1.28, p = 0.004), a wider SI d range (OR 1.18, 95%CI 1.05-1.32, p = 0.004) and greater SI d variability (OR for log-SD 2.38, 95%CI 1.04-5.43, p = 0.039) during this epoch independently correlated with one-month death. The cumulative incidences of one-month death were significantly higher among patients with higher SI s or SI d levels during the 13-24 h epoch ( Figure S3). The Youden index identified a mean SI s of 0.65 bpm/mmHg and SI d level of 1.15 bpm/mmHg 13-24 h after onset as able to discriminate between patients who died or survived within one month. The area under the receiver operating characteristic curve for SI s and SI d was 0.63 bpm/mmHg (sensitivity 72%, specificity 53%) and 0.64 bpm/mmHg (sensitivity 72%, specificity 56%), respectively.
No difference in SI s or SI d parameters was observed between patients with and without a three-month poor functional outcome (all p > 0.05).

Discussion
To our knowledge, this study is the first to examine the temporal evolution of heartrate-to-blood-pressure ratios (SI s and SI d ), as well as their impact on MBE and one-month death. We are also the first to investigate this correlation among specific, fatal stroke subtypes as LHI patients, using a registry cohort database. In this study, SI s and SI d increased over baseline in all patients, with similar profiles during the first 12 h epoch and a more intensive increase in patients with MBE or one-month death during the second 12 h epoch. Higher and more fluctuated heart-rate-to-blood-pressure ratios independently correlated with increased risk of MBE and one-month death in LHI patients, especially during the second 12 h (13-24 h) epoch after onset.
The median interval from onset to MBE development in our study was 36.3 h, which corresponded to previous studies [9], and 85% (39/46) of MBE was developed with an interval over 24 h. For the seven patients who developed MBE within 24 h of onset, we did not apply the heart-rate-to-blood-pressure ratios collected after MBE development for analysis, in order to ensure the appropriate sequential order of studied parameters and outcomes. For the remaining patients, we included all collected heart-rate-to-bloodpressure ratios for analysis. Therefore, it is much more likely that the MBE development was influenced by ratio changes and not the other way around, since for each individual patient, the studied ratios were collected strictly before MBE occurred. We failed to conduct sensitivity analysis by excluding those who developed MBE within 24 h of onset due to limited study samples, which needs to be validated in a further enlarged cohort.
Our study revealed a significant correlation between the variability of continuous measurement of SI d levels within 24 h after onset and the occurrence of MBE. Additionally, we found a significant correlation between an increase in the mean and variability of SI d levels during the 13-24 h epoch after onset and the occurrence of MBE. The relationship between heart-rate-to-blood-pressure ratios and one-month mortality showed similar significance, that higher SI d with greater variability is associated with one-month mortality, particularly during the 13-24 h epoch after onset. Higher SI s during the 13-24 h epoch after onset was also independently correlated with one-month mortality. Our result is in agreement with previous studies. Two studies have investigated the correlation between SI s and stroke outcome. McCall suggested that admission SI s at both high and low extremities may predict short-term fatality among general stroke [13]. According to Myint, higher SI s at presentation to emergency predicts patient-related clinical outcomes in general stroke [14]. Of note, previous studies only recorded the first admission SI s of unselected patients, including both ischemic and hemorrhagic stroke populations. Our result specifically focused on LHI patients, who are more likely to develop MBE, thus needing more intensive monitoring. Meanwhile, we investigated the impact of both the absolute level and variability of the heart-rate-to-blood-pressure ratio by serially collecting them for 24 h after onset. Moreover, compared to previous studies, we extended the traditional shock index by adding a variation of it, SI d . Diastolic pressure has a greater impact than systolic pressure on the mean arterial pressure, which is the most commonly used target regarding cerebral blood flow. In our research, the SI d parameter showed a significant correlation with MBE, while the SI s parameter did not, suggesting that in LHI patients, diastolic blood pressure may play a more critical role than systolic blood pressure in brain edema development. This finding suggests that more attention should be paid to diastolic blood pressure control when preventing MBE development in clinical practice.
The mechanism underpinning differences in SI s and SI d remains unknown. No difference in systolic and diastolic blood pressure was observed between groups. Previous research reported an increased incidence of arrhythmias, or changes in the electrocardiogram in stroke [23][24][25][26][27]. Ischemic or hemorrhagic cerebral damage may impair the function of the autonomic nervous system and involves a higher sympathetic activity and a reduction in vagal tone, leading to an autonomic imbalance in heart rate regulation, most of which is tachyarrhycardia [27][28][29][30]. The risk of post-stroke tachyarrhycardia increases proportionally with stroke severity [31,32]. Therefore, LHI patients may feature a higher risk of post-stroke tachyarrhycardia, which correlate with poor clinically related outcomes [29,32,33]. HR was higher in patients who died within one month, although the difference was not statistically significant after adjustment for covariables (data now shown). This finding partially explained the impact of SI s and SI d on poorer outcomes, while also suggesting that heart rate alone may not independently influence the outcome.
Another underlying mechanism for the association between elevated heart-rate-toblood-pressure ratios and MBE as well as one-month mortality after acute stroke might be due to abnormalities in autonomic control, such as decreased baroreflex sensitivity (BRS). BRS is an index that quantifies the extent of baroreflex modulation of the heart rate in response to changes in blood pressure [34]. Previous studies reported a reduction in BRS in stroke patients [35], which is linked to an imbalance in autonomic regulation with a decline in the function of the parasympathetic nervous system and enhanced sympathetic activity [36] Sympathetic activation following a stroke can cause an increase in blood pressure while reduced BRS leads to weakened regulation of the heart rate in response to changes in blood pressure, leading to an insufficient decline in the heart rate and higher heart-rate-to-blood-pressure ratios. Moreover, BRS is inversely correlated with blood pressure variability [37], and impaired BRS can lead to an increase in blood pressure variability, which may also increase the variability of heart-rate-to-blood-pressure ratios. Impaired BRS appears to be related to brain hypoperfusion [38], thus leading to MBE development. Additionally, impaired BRS was reported to be associated with increased long-term mortality after acute ischemic stroke [36].
Elevated SI s is reported to be of prognostic importance for infection, which may influence stroke outcomes [12]. Patients diagnosed with LHI have a relatively high incidence of in-hospital infection, with the pneumonia rate estimated to be between 39% and 54% [39,40] and urinary incontinence around 18.4% [39]. LHI patients with poor functional outcomes have a considerably higher incidence of pneumonia (67.1%) and urinary incontinence (27.9%) [39]. This phenomenon may be attributed to various factors, such as impaired consciousness post-stroke, aspiration, and prolonged hospitalization [41]. In this study, the rate of any in-hospital infection rate was 77.6%. The higher infection rate may be due to older age (70 vs. 61 years) and higher NIHSS (16 vs. 14) when compared to the previous study population [39]. It is also important to note that the COVID-19 pandemic may have contributed to an increase in in-hospital infection rates since patients in our study were included during this specific period. As a result, we included in-hospital infection as a covariate in our analysis to account for this confounding factor, and still identified a prognosis significance of elevated SI s or SI d for MBE and one-month death. However, due to a limited sample size, we regret that we were unable to perform a sensitivity analysis on patients without in-hospital infections. Therefore, further research involving larger sample sizes is needed to explore this issue further.
We investigated the association between the heart-rate-to-blood-pressure ratio parameters and the natural history of MBE progression by dividing the overall 24 h recording epoch into two 12 h epochs. The progression of MBE can be separated into two stages: cytotoxic edema and vasogenic edema. The initial stage of cytotoxic edema occurs within the first few hours and declines within one day, which aligns temporally with the first 12 h epoch. The second stage of vasogenic edema develops within one to three days and is maintained for several days [9], which could be impacted by physiological changes during the 13-24 h epoch after onset. In this study, a higher SI d with greater variability was observed during the overall 24 h recording epoch and second 12 h epoch (13-24 h after onset), but not the first 12 h epoch, which suggested that SI d may affect outcomes by mainly interfering vasogenic edema 13-24 h after onset.
Our study has some limitations. Firstly, the population is relatively small due to the low incidence of LHI. Secondly, the ratios were collected less within the first 12 h epoch after onset due to the hospitalization delay of patients. Thirdly, our study lacked more hemodynamic parameters, such as PaCO2 and beat-to-beat data, which could have provided further insights into the underlying mechanisms. In order to validate our results, further studies are required across multiple centers, with a larger sample size. However, the limitations of this study do not obscure its strengths. Firstly, we focused on LHI, which is the very subtype of ischemic stroke with the worst outcome. We provided some specific evidence for intensive care in this population. Secondly, SI s and SI d are two simple indices derived from two readily available recordings during monitoring. It can be measured in emergencies or in the ICU without training from a neurologist, with usefulness in a clinical setting. Thirdly, we conducted serial measurements of HR and BP, revealing the temporal and variability pattern of the ratios.

Conclusions
A higher and more fluctuated heart-rate-to-diastolic-blood-pressure ratio (SI d ) independently correlated with increased risk of MBE and one-month death in LHI patients, especially during the second 12 h (13-24 h) epoch after onset. A higher heart rate-to-systolic blood pressure ratio (SI s ) 13-24 h after onset was independently correlated with one-month death in LHI patients.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/diagnostics13152506/s1, Figure S1. The flow diagram of patient selection; Figure S2. The 24 h temporal evolution of heart rate/blood pressure ratios in LHI patients who died and survived within one month; Figure S3. Kaplan-Meier survival curves for cumulative incidence of one-month death among LHI patients with high or low heart rate/blood pressure ratios 13-24 h after onset; Table S1 The baseline characteristics of large hemispheric infarction patients who died and survived within one month; Table S2 Calculated parameters of overall and epoch-based heart-rate-to-blood-pressure ratios in patients who died and survived within one month.