INTRODUCTION

Intracerebral hemorrhage (ICH) is less studied than other subtypes of cerebrovascular disease, despite higher mortality and worst short- and long-term prognosis compared with other stroke types¹. In Mexico, ICH accounts for about 30% of all types of stroke, with an acute phase case fatality rate (CFR) ranging from 30% to 50%²-⁵. Despite these alarming data, information regarding pre-hospital characteristics and determinants of ICH is scarce, which contrasts with the knowledge on acute ischemic stroke, a condition where the pre-hospital determinants that favor or delay hospital arrival are well established⁶-⁸. There is evidence that an early arrival may be beneficial for most of the acute neurovascular syndromes⁴-⁸, and it appears that implementing a stroke code has contributed to prompt hospital admission for all stroke types. This collateral beneficial effect has favored early diagnostic evaluations and prompt therapeutic interventions, such as acute blood pressure control and surgical evacuation, when needed⁹,¹⁰.

Our aim was to identify the variables determining hospital arrival in the first 6 h after ICH symptoms onset and to analyze their impact on short-term outcome. Our hypothesis was that an early arrival is associated with better prognoses.

METHODS

We analyzed the database from the multicenter RENAMEVASC registry that included all consecutive adult patients with any type of acute stroke confirmed by neuroimaging studies⁵. Briefly, this registry was a hospital-based observational study designed and conducted by the Mexican Association of Cerebral Vascular Disease (AMEVASC) to improve stroke knowledge and awareness in the community, where investigators from 25 referral hospitals of 14 Mexican federal states recruited consecutive patients with an acute stroke diagnosis. All the participants or their legal proxies signed informed consent. An Local Institutional Review Boards approved the study at each participating center.

Definitions of the analyzed sociodemographic variables, risk factors, and clinical conditions at admission were implemented according to international standards at the moment of the study. Hypertension was recorded in patients with primary or secondary hypertension history with either regular or irregular treatment, or determination of arterial blood pressure values ≥ 140/90 mmHg for more than 2 weeks after hospitalization. Alcoholism was defined as a patient with a history of chronic alcohol consumption. Smoking history was determined in patients with current or past tobacco use confirmed by patient or family member at admission. Obesity was registered in patients with body mass index ≥ 30 at hospital admission.

The ICH clinical features at onset were queried in the emergency room, which included headache, vomit, impaired consciousness, motor or sensorial deficits, seizures, and other features. Neurological status was evaluated using the Glasgow coma scale (GCS), and severity of the neurological status was classified as mild (13-15 points), moderate (9-12), or severe (< 8 points). Furthermore, ICH location (i.e., lobar, deep ganglionic, cerebellar, brainstem or intraventricular), etiology (i.e., hypertensive, arteriovenous malformation, amyloid, coagulation disorders, drug use, undetermined), as well as any ventricular system apertures, were recorded. Trauma, ruptured aneurysm, and cerebral venous thrombosis leading to ICH were excluded from the present analysis. The patients with unknown symptom onset, inaccurate registry data at hospital admission, or ICH development while hospitalized for reasons other than the initial stroke, were also excluded from the study.

Hematoma volume was calculated according to the AxBxC/2 formula, where A = largest diameter, B = perpendicular diameter to A, C = number of sections (in cm) in which the hemorrhage appears in the CT scan¹¹,¹². Established treatment in the acute phase was classified as conservative (i.e., medical) and surgical (i.e., hematoma drainage, ventriculostomy, or both). Two prognostic models were analyzed to estimate 30-day mortality; the ICH score¹³ and the ICH-grading scale (GS)¹⁴. Arrival times were analyzed during the 1st h, 3, and 6 h after ICH symptoms onset. The functional outcome was evaluated using the modified Rankin scale (mRS) at 30 and 90 days. The CFR was also assessed at discharge, 30 and 90 days after ICH onset.

RESULTS

From 500 patients with ICH included in this multicenter registry, 473 (94.6%) patients met the selection criteria for the present analysis (51% women, with a median age of 63 years, range: 18-98 years); 63 (13%) were patients under 40 years of age and 176 (37%) over 70 years. The reason for excluding the remaining 27 (5.4%) patients was the lack of data on symptoms onset. The main identified risk factors for ICH were hypertension, obesity, smoking, and alcoholism (Table 1). In all, 244 (51.6%) patients had an indirect arrival (i.e., were referred from other institutions) and received first medical assessment in a first-aid post or first-level medical care before definite reference to the final second- or third-level institution. The median delay from symptoms onset to the first medical evaluation was 1.0 h (IQR: 0.5-3 h). On the other hand, the median time from symptoms onset to final emergency-room admission was 6.25 h (IQR: 2.5-24 h): 37 (7.8%) arrived in < 1 h, 124 (26.3%) in < 3 h, 214 (45.3%) in < 6 h, and 294 (62.3%) in < 12 h. As a consequence, the final arrival time was shorter in patients who were not referred from another institution (i.e., direct arrival), as compared with those patients who received first assessment in a different hospital before final referral (i.e., indirect arrival) (median time: 3.0 vs. 12.0 h, respectively; p < 0.001).

At hospital admission 39% of patients had systolic blood pressure (SBP) > 180 mmHg and 33% had glucose levels > 150 mg/dL. The median GCS was 12 (IQR: 8-14), without sex or age differences. In all, 469 (99%) had head CT scan and 4 (1%) underwent MRI. Neuroimaging was obtained at a median of 9.45 h (IQR: 4.5-24 h) since symptoms onset, or 4.0 h (IQR: 3-6 h) on hospital admission (33% underwent neuroimaging in < 6 h since emergency room presentation). However, the time from symptoms onset to undergoing neuroimaging was shorter in patients who had a direct arrival, as compared with those with indirect arrival (median time: 7.3 vs. 11.0 h, respectively; p = 0.034). After etiological investigations during the hospital stay, the main cause of ICH was considered arterial hypertension in 70% of patients (64% with a history of hypertension and 6% with diagnosis after ICH), undetermined 9%, cerebral vascular malformations 7%, and amyloid angiopathy in 4% of cases. In all, 9% were receiving antiplatelet therapy, and 2% received anticoagulant before admission. Main ICH locations were supratentorial (91%) and deep (52%). Hematoma volume had a median of 20 mL (IQR: 9-45 mL), with 35% with a volume > 30 mL. Intraventricular hemorrhage occurred in 45% of the cases. A total of 85 (18%) patients received surgical evacuation of hematoma with or without ventriculostomy. The median in-hospital stay length was 9.0 days (IQR: 5-17 days). A total of 112 (23.7%) patients required admission to the intensive care unit, with a median UCI stay length of 5.0 days (IQR: 2.75-9.0 days). In all, 179 (38%) had in-hospital complications such as pneumonia (28%), urinary tract infections (12%), deep-vein thrombosis (6.2%), and other complications.

In bivariate analyses, school education, hypertension, seizures at ICH onset (11% of patients), admission blood glucose, admission GCS and the ICH score were factors associated with hospital arrival at different cut-off points (Table 2). Seizures (either focal or generalized) were more common in patients with lobar hemorrhages than in the deep location (45% vs. 26%, p = 0.001) Multivariate models for prediction of hospital arrival were reliable and consistent only for the prediction of arrival in < 6 h (Table 3). Thence, previous hypertension drug therapy, ≥ 3 years of schooling, and seizures at ICH onset were the independent predictors of a hospital arrival in < 6 h.

CFR at discharge, 30 days and 90 days after ICH onset was 28.8%, 30.0%, and 32.6%, respectively. The causes of death were neurological in 56.9%, systemic (non-neurological, such as sepsis) in 27.5%, and pertaining to both categories of causes in 15.7%. Correspondingly, mRS > 3 at discharge, 30 days and 90 days was 71.7%, 61.4%, and 56.2%, respectively (21.6% 90-day improvement). The median mRS at discharge, 30-day and 90-day follow-up was essentially unchanged, with 4 (IQR: 3-6), 4 (IQR: 3-6), and 4 (IQR: 2-6), respectively. In a Cox proportional-hazards model for prediction of in-hospital death, SBP > 180 mmHg at admission, ICH score ≥3, and GCS <8 at admission were the independent predictors (Table 4). Notably, arrival time, indirect arrival (after referral from another institution), ICH-GS, clinical features at onset or comorbidities were not independent predictors, but they were retained in the final multivariate model for adjustment.

DISCUSSION

Conditions that delay or facilitate hospital arrival have been frequently studied in ischemic stroke, even before the approval of intravenous thrombolysis with tissue plasminogen activator¹⁵,¹⁶. The diversity of diagnostic studies and time-dependent treatment options in cerebral infarction has kept the interest in this topic constant¹⁷-²⁰. In contrast, specific determinants for early or late hospital arrival for ICH have been scarcely studied⁷,²¹-²³. In our study, more than 50% of patients with spontaneous ICH had a medical assessment in a first-aid post or first-level institution that finally referred the patient to a different medical care center that took the responsibility to admit and treat the patient, losing a median time of 5.25 h, which might be essential to improve the outcome in some cases. Moreover, a quarter of patients had an early hospital arrival within the first 3 h, and nearly 50% had an early arrival in the first 6 h. In a similar contemporary study from Brazil, 64% of patients had a hospital arrival within the first 3 h²⁴. In the Spanish multicenter stroke registry, EPICES, which analyzed the hospital arrival times and delay before acute stroke patients received neurological care, 35% of patients (among 5,454 of all stroke types) arrived in < 1 h, 64% < 3 h, and 78% in < 6 h from the onset of acute cerebrovascular disease²⁵. In the same report, 771 patients with ICH had an average time to hospital arrival of 3.8 h, being the hypertensive etiology with the shortest arrival time with an average of 3.17 h, compared to other ICH causes. In a recent report from Korea analyzing the type of emergency medical service use on the time interval from ICH onset to hospital admission they found that, among 6,564 patients, there was an emergency arrival within the first hour in 23.7% of patients, particularly in those that used the emergency system with direct referral²³. Extensive data analysis between 1990 and 2008 from 75 studies and 183,009 stroke patients has reported a 6% decrease in time to hospital arrival per year, with variations depending on the period and study design, geographical or cultural barriers, and pre-hospital organization for stroke care²⁶.

The precise knowledge of hospital arrival times is relevant for the application of directed interventions in the current management of ICH. Hemostatic therapy, temperature, blood pressure, and glucose control require evaluation in an emergency setting within the first hours after symptoms onset according to current recommendations⁹,²⁷-³⁰. A recent article by Colton et al., emphasizes that early symptom recognition and activation of a stroke code favors an earlier hospital arrival of patients with ICH and shortens other strategies such as door-to-CT scan time frame and other essential interventions¹⁰. In our report, the median time elapsed from the onset of ICH symptoms to the performance of neuroimaging was 9.45 h (4 h after hospital presentation).

In the present study, we found that prehospital characteristics such as medical treatment, clinical features, and patient's education level are associated with an early arrival. Similarly, arrival characteristics such as prognostic scores and blood pressure were found associated with the short-term outcome. These findings suggest that very early characteristics of ICH patients can predict a prompt hospital presentation and short-term outcome, since it is possible that patients with a severe clinical presentation arrive earlier than those with little neurological impairment. The clinical severity of ICH had been previously found associated with an early hospital arrival after ICH⁵,²⁴,³¹.

We found an independent association between the presence of seizures and early hospital arrival. Seizures were more common in patients with lobar hemorrhages than in deep location, which is in accordance with the literature³². It is possible that witnessing a seizure in our environment translates into a cardinal event for seeking medical care. Perceptions, myths and orally transmitted knowledge about epilepsy could be also considered as potential influence³³.

In the bivariate analyses, we found that GCS and ICH score were associated with early arrival at some time points. This finding can be associated with a worse clinical state being more likely to receive a prompt medical attention, reference, and diagnostic evaluation. However, to our surprise the time to hospital arrival was not independently associated with outcome in the multivariate analysis, which could have been analytically overcome by the clinical state itself, reflected by the prognostic scores.

Although ICH usually implies a poor prognosis, determining factors of the outcome are characteristics and severity of the hematoma, such as volume, location and intraventricular irruption⁹,¹⁴. However, small hemorrhages can present with minor symptoms as a lacunar stroke syndrome, which are associated with a good outcome and CFR fewer than 10%³⁴,³⁵. Moreover, the site of bleeding is the second most important hematoma variable that determines the outcome, especially for infratentorial and deep lobar structures³⁶, since, for example, even small hematomas located in the thalamus are associated with a prognosis comparable to a brainstem ICH³⁷. Therefore, it is possible that the more severe hemorrhages are naturally selected for hospitalization and that the worse-case scenarios are over represented in hospital series. In our study, early hospital arrival was not statistically associated with the functional outcome or death in the short term. Our results coincide with those of Naganuma et al., who analyzed 1226 patients with acute ICH and demonstrated that the time elapsed to hospital arrival had no effect on functional prognosis or death in the short term³⁸. However, the fact that early arrival might not impact ICH short-term prognosis should not be misinterpreted as evidence justifying delays in the rapid recognition of an acute neurovascular syndrome and transfer to the hospital, especially since at the very beginning it is difficult to distinguish ischemic from hemorrhagic stroke based only on clinical features⁴. Moreover, we do not know the impact of early hospital arrival on the mid- and long-term or for specific outcomes, such as motor, cognitive or non-neurological complications.

In the present cohort only about one-quarter of the patients with ICH were admitted to the ICU, in spite that this condition is considered critical. The limited resources and scarce spaces in ICU may explain this phenomenon, but it may also be explained by cognitive errors such as fatalism and self-fulfilling prophecy influencing medical decisions, and thus, limiting the opportunities to survive to these patients.

Limitations of this study include that we did not consider certain characteristics associated with hospital arrival, such as previous knowledge about stroke symptoms, educational level of bystanders, reasons for arriving indirectly after a hospital referral, and medications used before final hospital admission. Registering of the bystander's education level would also offer more insight into the determinants for a prompt hospital arrival. Furthermore, even when the time to hospital arrival is more likely to be associated with short- instead of long-term outcome, we did not analyze possible impairments and mortality in the long run. Such analysis would provide with more data on the consequences of expedited medical care.

In conclusion, in this multicenter registry from Mexico, nearly half of the patients had an arrival within 6 h. Seizures at ICH onset, patient's basic level of education, and hypertension therapy were the main clinical predictors of early arrival. More severe clinical conditions were associated with short-term mortality. Further studies are necessary to elucidate whether time elapsed from symptoms onset to hospital arrival is determinant for short- and long-term outcome when effective and specific treatments are finally available.