Epidemiology of Aneurysmal Subarachnoid Hemorrhage in Australia and New Zealand
Incidence and Case Fatality From the Australasian Cooperative Research on Subarachnoid Hemorrhage Study (ACROSS)
Background and Purpose—More data on the epidemiology of subarachnoid hemorrhage (SAH) are required to increase our understanding of etiology and prevention. This study sought to determine the incidence and case fatality of SAH from 4 prospective, population-based registers in Australia and New Zealand.
Methods—We identified all cases of “aneurysmal” SAH from November 1995 to June 1998 in Adelaide, Hobart, Perth (Australia), and Auckland (New Zealand), a total population of approximately 2.8 million, using standard diagnostic criteria and uniform community-wide surveillance and data extraction procedures.
Results—A total of 436 cases of SAH were registered, including 432 first-ever events and 4 recurrent events. The mean age of cases was 57 years (range, 16 to 94 years), and 62% were female. From the 400 first-ever events registered over whole years, the crude annual incidence for the total population was 8.1 per 100 000 (95% CI, 7.4, 9.0), with rates higher for females (9.7; 95% CI, 8.6, 11.0) than for males (6.5; 95% CI, 5.5, 7.6). Age-specific rates showed a continuous upward trend with age, although the shape and strength of this association differed between the sexes. Standardized annual incidence of SAH varied across centers, being highest in Auckland largely because of the high rate in Maori and Pacific people. The 28-day case fatality rate for the total population was 39% (95% CI, 34%, 44%), with little variation in ratios across centers.
Conclusions—There is variation in the incidence of SAH in Australia and New Zealand, but the rates are consistently higher for females. A monotonic increase in incidence with age suggests that exposures with cumulative effects and long induction times may be less relevant in the etiology of SAH.
Subarachnoid hemorrhage (SAH), due to rupture of intracerebral aneurysms, accounts for approximately 4% of all strokes, but its impact is relatively greater because it tends to affect younger adults in good health, often with devastating consequences.1 2 Despite improvements in surgery and medical care for the condition, case fatality and morbidity remain high,2 3 4 with a significant proportion of deaths occurring before any specific treatment can be contemplated. Although there has been a decline in mortality rates from SAH since the 1970s,5 6 it is uncertain how much of this reflects declines in incidence and case fatality during this period.1 6 A better understanding of the incidence of and important causal risk factors for SAH could lead to improvements in prevention and treatment of the condition.
Many epidemiological studies of SAH have used hospital-based surveillance or prospective case ascertainment in tertiary referral centers.7 8 These studies, however, are confounded by referral bias, in particular the exclusion of patients who die early or who are unsuitable for surgical intervention. A population-based registry can overcome this deficiency provided that a comprehensive surveillance system is used to ensure complete ascertainment of cases in a defined population.9 Although a number of well-designed population-based studies of SAH have been undertaken, the small numbers of patients registered have prevented firm conclusions regarding age- and sex-specific rates and secular trends in incidence.9 Moreover, they do not provide reliable evidence about important causal factors other than a strong association with cigarette smoking.10
The Australasian Cooperative Research on Subarachnoid Hemorrhage Study (ACROSS) was initiated as a large prospective, multicenter, population-based, case- control study in Australia and New Zealand to determine incidence, risk factors, and prognosis of SAH; factors of importance in triggering the event; and management and long-term outcome of SAH. We present here the incidence and early case fatality for SAH.
Subjects and Methods
ACROSS used population-based registers in 4 major centers of Australia and New Zealand: Adelaide (South Australia), Hobart (Tasmania), Perth (Western Australia), and Auckland (New Zealand). These cities are ideally suited for epidemiological studies of SAH because the populations are well defined and have similar healthcare systems, which include regional neurosurgical services for SAH located at tertiary care public teaching hospitals. In addition, Perth and Auckland have well-established systems for assessing trends in the incidence of stroke.11 12 Population figures for those residents aged 15 years and older are available from a 1996 census for each city. With the exception of Hobart, the cities are of comparable size: Adelaide (inner metropolitan region, 870 965), Hobart (153 397), Perth (1 021 770), and Auckland (823 890), for a total study population of approximately 2.8 million. The surveillance time periods varied across the centers but included whole years: November 1, 1995, to March 31, 1998, for Adelaide (29 months); December 1, 1996, to January 31, 1998, for Hobart (26 months); December 1, 1996, to February 28, 1998, for Perth (27 months); and June 23, 1997, to June 22, 1998, for Auckland (12 months).
Experienced trained nurses scrutinized daily the medical records of all persons with any of specific clinical diagnoses—stroke, intracerebral hemorrhage, SAH, and headache—who presented to the accident and emergency departments or were admitted to any of the acute public teaching hospitals in each of the study centers. In addition, these nurses made twice weekly visits of neurosurgical or medical wards in the hospitals, checked hospital discharge records, both public and private, and reviewed all death certificates and coroners’ reports for a diagnosis of SAH as either the underlying or a contributing cause of death. Final checks for completeness of ascertainment were made by reviewing computerized hospital separation data for hospitals within and surrounding the study areas with the use of the International Classification of Diseases, Ninth Revision code 430 for SAH as either a primary or secondary diagnosis and by searching official mortality statistics with key words for SAH.
SAH was defined, according to standard criteria,13 as an abrupt onset of severe headache and/or loss of consciousness, with or without focal neurological signs, with CT, necropsy, or lumbar puncture evidence of focal or generalized blood in the subarachnoid space. We excluded patients in whom the hemorrhage was found definitely to originate from sources other than an intracranial aneurysm, including primary intracerebral hemorrhage, arteriovenous malformations, trauma, infections, bleeding diathesis, and neoplasms. Those patients in whom an aneurysm could not be identified by cerebral angiography, necropsy, or the presence of a localized collection of blood in a fissure on CT were included but analyzed separately. Attention was given to diagnosing cases with a perimesencephalic pattern of hemorrhage on CT and normal 4-vessel angiography.14 Patients with CT alone and no specific pattern of hemorrhage were classified as “uncertain” aneurysmal SAH, while those with acute severe headache followed by death within hours were classified as “probable” SAH. Each event during the study period was further classified as being the patient’s first-ever or a recurrent SAH. For patients with multiple events, the index event was defined as that event which occurred nearest to the time when the patient was first registered. A case “managed in hospital” was one in which admission involving an overnight stay occurred within 28 days of the onset of the event. Twenty-eight-day case fatality was defined as the proportion of all events resulting in death within 28 days of onset. Efforts were made to maintain uniform diagnostic standards, and study nurses discussed difficult cases with the study neurologist in each of the centers.
As soon as possible after notification, the study nurses undertook face-to-face interviews with patients or, when the patient was deceased or disabled, the partner or next of kin. A structured questionnaire was used to obtain information regarding demographics, clinical features, investigations and management, medical and family history, health behavior, health status, and risk factors. General practitioners and hospital medical records were reviewed to obtain more information about each event and about previous illnesses. To ensure standard procedures, all study nurses attended an initial start-up meeting and had their first 10 interviews checked by tape recorder. Information about the study was provided at regular meetings with collaborating hospital staff by the study team. The protocol for ACROSS was approved by all relevant institutional ethics committees in each of the study centers, as outlined in the Appendix. Consent from next of kin was obtained for patients who were severely ill, unconscious, or deceased.
Crude incidence, together with 95% CI, was calculated for each age, sex, and city category by the exact approach15 and with 1996 census population data for Australia and New Zealand. In view of seasonal variability in the incidence of SAH,9 16 rates were derived only for whole years of surveillance (1996 and 1997 for Adelaide, Hobart, and Perth; mid-1997 to mid-1998 for Auckland). Given the high incidence of SAH among Maori and Pacific people in Auckland,17 the rates were recalculated separately for these groups, with appropriate adjustment of the population denominator. Standardized rates were derived by the direct method and 10-year age groupings (≥15 years) of Segi’s “world”18 and estimated Australasian populations as the external reference. Rates are presented as 10-year age- and sex-specific rates per 100 000 person-years. The effects of age and sex on incidence were estimated with a Poisson regression model. Data were analyzed for heterogeneity across centers by 1-way ANOVA for continuous variables that were approximately normally distributed and the Kruskal-Wallis test for continuous variables with evidence of nonnormal distribution. Categorical variables were compared with the χ2 test. All calculations were performed with the use of SAS19 and SPSS for Windows software.20
Overall, 1609 possible cases of SAH were registered over the full 29-month study period. After review of clinical data and other information, 1055 of these patients were excluded either because they were not residents of one of the study centers or the diagnosis was not SAH. A final diagnosis of SAH was confirmed in 554 cases, but this included 39 (7%) due to arteriovenous malformations and 79 (14%) secondary to head trauma. In addition, 3 patients had a past history of idiopathic SAH, and 1 patient experienced a recurrent event during the study period.
Thus, a total of 432 cases were registered with a final diagnosis of first-ever SAH (62% female; mean±SD age, 57±17 years). Table 1⇓ outlines the patterns of notification and first sources of information regarding events across the centers. Although there were significant differences in the “onset-to-notification” and “onset-to-assessment” times across the centers, mainly because of delayed notifications in Hobart and Perth, overall these were short, with median times of 2 (interquartile range, 1 to 25) and 4 (interquartile range, 1 to 14) days for these intervals, respectively. There were also significant differences in the first source of notification across centers, probably reflecting local “hot-pursuit” strategies, but the profile still reflects the importance of using multiple sources of case ascertainment in such studies.
Overall, SAH was verified by CT in 394 (90.4%) or by necropsy alone in 42 (9.6%), as shown in Table 2⇓. There were 330 patients (76%) who had the aneurysmal origin of the SAH diagnosed by angiography, surgery, or at autopsy. In addition, 3 patients were included who met the clinical criteria alone because they died before investigations could be done (they were classified as probable SAH), and there were 13 cases (3%) of perimesencephalic hemorrhage. There was little variation in the proportional frequencies of relevant investigations across the centers (Table 2⇓). The proportion of patients who underwent angiography varied from 63% (Adelaide) to 71% (Hobart), while lumbar puncture and necropsy were undertaken in 54 (12%) and 42 (10%) cases overall, respectively. The great majority of patients (92%) were managed in hospital during the acute phase, and 255 (59%) had neurosurgical intervention, usually within 48 hours of onset (n=175, 69%).
Table 3⇓⇓ shows the age- and sex-specific incidence of SAH, by center and for the total population, estimated from the subgroup of 400 cases registered over whole years. The crude annual incidence of first-ever SAH in the total population (aged ≥15 years) in 1996–1998 was 8.1 (95% CI, 7.4, 9.0) per 100 000 (6.5 per 100 000 in males and 9.7 per 100 000 in females). When age- and sex-adjusted by the direct method to the 1996 Australasian population, the annual incidence was 6.7 per 100 000 (5.6 per 100 000 in males and 7.7 per 100 000 in females). Standardized by the direct method to the world population of Segi, the annual rate was 6.5 (95% CI, 5.8, 7.2) per 100 000.
Across the centers, the rates were highest in Auckland (9.1 per 100 000 for males and 11.3 per 100 000 for females), intermediate for Hobart, and lowest in Perth and Adelaide (for males in Perth, 3.9 per 100 000; for females in Adelaide, 6.1 per 100 000). Recalculation of rates for the major ethnic groups with the use of revised population denominators in Auckland showed that the higher rates in this center could be accounted for in part by the high rate of disease in Maori and Pacific people (Table 4⇑).
The Figure⇓ shows the age-specific crude incidence for each sex. Females showed a continuously rising trend of incidence with age, while for males, the trend appears bimodal, with peaks in younger adults (groups aged 34 to 44 and 45 to 54 years) and in the oldest old (group aged ≥85 years). These age-specific incidence curves for males and females were generally consistent across centers, as was the higher rate of disease in females. Table 4⇑ shows the truncated age-standardized incidence rates and female:male crude incidence rate ratios obtained from Poisson regression. The incidence in females was 60% greater than that for males (rate ratio, 1.6; 95% CI, 1.3, 2.0) in the total population (P<0.001), although there was variability in the rate ratios across centers. Modeling the interaction age×sex indicates that the higher rate of disease applied only to older females (≥55 years) (P<0.001), whereas the sex-specific rates were comparable in the younger age groups (15 to 54 years). Among Maori and Pacific people, however, the reverse applied, with the highest relative rate of disease among younger females (Table 4⇑).
The overall 28-day case fatality was 39% (170/436) for all cases and 38% (152/400) for first-ever SAH recorded over whole years. The corresponding ratios for deaths within 7 days of onset were 29% and 28%. There was no significant difference in case fatality ratios across the centers (Table 2⇑).
In this investigation, by far the largest population-based study to date, we have shown that aneurysmal SAH has a unique incidence profile. In contrast to the other subtypes of stroke (and cardiovascular disease, in general) in which rates increase exponentially with age, the incidence of SAH shows a monotonic increase with age, although the shape and strength of these curves differ between the sexes. For males, age-specific rates tend to be bimodal in distribution, whereas among females, the trend of increasing rates with age appears attenuated after the menopause (≥55 years). These data contradict many previous epidemiological studies of SAH that show a flattening, or even a decline, in age-specific rates in older people.1 21 22 Small numbers and logistic difficulties in verifying the diagnosis, particularly among older people, are important limitations in these earlier studies.
Saying that advancing knowledge on the etiology and pathogenesis of a particular disease requires coordinated contributions from epidemiology and the basic sciences is usually a statement of the obvious, but it is certainly true for SAH. Research findings from epidemiology, genetics, molecular biology, and other basic sciences have been mutually reinforcing in suggesting risk (and protective) factors for SAH. Epidemiological observations, in particular, are consistent with the hypothesis that cerebral aneurysms are acquired abnormalities that are likely to form and rupture as a result of the effects of cardiovascular risk factors, especially cigarette smoking and probably hypertension.23 Despite recent progress, however, it is uncertain whether the decline in the incidence of cardiovascular disease over recent decades, brought about in part by the reduction in these cardiovascular risk factors, has translated into a reduction in the incidence of SAH. Apart from 1 large population-based series over 10 years,24 a meta-analysis of well-defined studies did not confirm a decline in the incidence of SAH over the last 30 years.25
The paucity of population-based data on the incidence, risk factors, and outcome of SAH within various countries has been a major impediment to identifying etiologic clues.24 A fundamental problem is that most incidence studies have been based on small numbers of patients and therefore lacked statistical power for detecting trends in rates and associations related to all but the most common of exposures. A systematic overview may overcome the problem of imprecise results from small studies, but the approach requires careful attention to the sources, quality, and timing of the data collected in each investigation. In a recent reanalysis of pooled existing data from 15 non-Finnish and 3 Finnish prospective population-based studies, the incidence of SAH was 10.5 (95% CI, 9.9, 11.2) per 100 000 person-years (7.8; 95% CI, 7.2, 8.4 for non-Finnish studies) and higher in females (7.1; 95% CI, 5.4, 8.7) than males (4.5; 95% CI, 3.1, 5.8).24 However, there was considerable regional variation in incidence, particularly in Finland, and the precision of the estimates is unreliable because only 3 of these 18 studies involved >100 patients, and in the only study of >400 patients, less than half had CT confirmation of the diagnosis.
Our study, on the other hand, used prospective, multicenter, population-based disease registries to enlarge the study size, thus enabling more precise estimates of the incidence of SAH and associations with demographics, lifestyle, and other factors. To satisfy a set of well-established “ideal” criteria for population-based incidence studies of stroke,9 it was crucial that the case ascertainment procedures were standardized to ensure that any variation across centers could not be attributed to registration artifacts. Unlike many other studies, we were able to achieve a uniformly high proportion of CT-confirmed (and/or necropsy-confirmed) cases of SAH across centers, since it is well known that it is not possible to differentiate subtypes of stroke reliably on clinical grounds alone.25 Moreover, radiological confirmation of the aneurysmal origin of the SAH in a large proportion of cases will allow future analyses of risk factors specifically for aneurysmal SAH, excluding possible or other (including perimesencephalic) forms of the condition.14 23
While incomplete identification of nonhospitalized cases may lead to a lower apparent incidence and higher observed case fatality for other subtypes of stroke, our data indicate that routine sources of data (death certificates and hospital separation lists) are effective for the surveillance of SAH in these populations. Since our study was also concerned with the assessment of exposures by direct interview, the protocol required continuous prospective surveillance of hospitals and early assessment of cases. The “hot-pursuit” methods were tailored to systems and resources in each population, as reflected in regional variation in the timing and sources used for data collection. If there were major differences in ascertainment of mild cases across the centers, however, this would have been reflected in the figures for case fatality. Yet, this parameter was similar across the centers.
Although the crude annual incidence of SAH (8.1 per 100 000) was quite high, our results do not differ from those of other studies (except in Scandinavian countries, where the rates are very high1 9 24 ) when adjustments were made for the age distribution of our population. The high rate in Auckland could be largely attributed to the high rate of disease among Maori and Pacific people, as documented elsewhere.17 The role of modifiable and nonmodifiable risk factors (ie, familial or genetic factors) as explanations for these differing rates is not completely understood. Future analyses of these data are planned.
What does an increase in incidence with age indicate in relation to etiology of SAH? For the vast majority of chronic diseases, both vascular and neoplastic, incidence increases exponentially with age because of the compounding effects of antecedent exposure (ie, exposure in one’s 20’s interacts with exposure in one’s 30’s, which interacts with exposure in one’s 40’s, etc). However, a monotonically increasing relationship between incidence and age may suggest that SAH is triggered in susceptible individuals (genetically or otherwise) and that the increase in incidence with age exactly reflects the increase in the risk of exposure with age. A plausible hypothesis is that most aneurysms form over a relatively short time (hours, days, or weeks),26 and the trigger might be an acute increase (or rapid fluctuations) in blood pressure.3 For example, compared with younger, normotensive, active individuals, it may be more likely for a transient increase in systolic blood pressure to occur among sedentary older people with high resting blood pressure. Perhaps the transient decline in the risk of SAH among males with respect to females after middle age could relate to the higher competing risk of cardiovascular disease from shared chronic and acute exposures.27 28
In summary, our data, compiled with the use of a unique multicenter registry, show regional differences in the incidence of SAH between Perth (low), Hobart, Adelaide, and Auckland (high) that can be explained in part by ethnic differences in disease risk. For both sexes, attack rates increase with age, but for males the association appears bimodal, with peak rates among younger adults and the oldest old. For females, rates increase continuously with age, although the trend is attenuated after the menopause. These data suggest that exposures with cumulative effects and long induction times may be less relevant in the etiology of SAH. Further analyses of the case-control component of ACROSS may help to unravel the puzzle.
The following are the committee members, principal investigators, and study coordinators of ACROSS: Steering Committee: C. Anderson (study chair); N. Anderson, R. Bonita, D. Dunbabin, G. Hankey, K. Jamrozik. Writing Committee: C. Anderson, G. Hankey, K. Jamrozik, D. Dunbabin. Data Management and Statistics: D. Bennett, R. Broadhurst, J. Duncan, C. Ni Mhurchu, S. Rubenach. Study Coordinators: J. Bennett (Study Manager), D. Healy, S. Rubenach (Adelaide); J. Sansom and J. Flecker (Hobart); J. Harvey, J. Linto, G. Mann, K. White (Perth); and S. Hawkins and C. Mulholland (Auckland). Neurosurgical Investigators: B. Brophy (Flinders Medical Center), J. Liddell (Royal Hobart Hospital), E. Mee (Auckland Hospital), G. McCulloch (Queen Elizabeth Hospital), N. Knuckey (Sir Charles Gairdner Hospital), and P. Reilly (Royal Adelaide Hospital). Clinical Centers: Ashord Hospital, Flinders Medical Center, Memorial Hospital, Repatriation General Hospital, Queen Elizabeth Hospital, and the Royal Adelaide Hospital (Adelaide, South Australia); the Royal Hobart Hospital, Calvary Hospital, and St Helen’s Private Hospital (Hobart, Tasmania); Fremantle Hospital, Royal Perth Hospital, St John of God Hospital Subiaco, St John of God Hospital Murdoch, and Sir Charles Gairdner Hospital (Perth, Western Australia); Auckland Hospital, North Shore Hospital, Middlemore Hospital, and Waitakare Hospital (Auckland, New Zealand).
This study was supported by grants from the National Health and Medical Research Council of Australia, the Health Research Council of New Zealand, and the Sylvia and Charles Viertel Charitable Foundation of Queensland, Australia. We thank Anthony Rodgers, Gary Whitlock, and Valery Feigin for reviewing earlier drafts of this article. We are indebted to the study investigators and coordinators for their dedication and performance; Janet Bennett for her efforts; the support of the coroner’s department in each center; the assistance of the Australian Bureau of Statistics and Statistics New Zealand; and the help provided from nursing, administration, and medical records staff of the clinical centers.
A list of all ACROSS participants is given in the Appendix.
- Received March 6, 2000.
- Revision received May 3, 2000.
- Accepted May 3, 2000.
- Copyright © 2000 by American Heart Association
Ingall TI, Wiebers DO. Natural history of subarachnoid hemorrhage. In: Whisnant JP, ed. Stroke: Populations, Cohorts, and Clinical Trials. Boston, Mass: Butterworth-Heinemann Ltd; 1993:174–186.
Bonita R, Thomson S. Subarachnoid hemorrhage: epidemiology, diagnosis, management, and outcome. Stroke. 1985;16:591–594.
Longstreth WT, Koepsell TD, Yerby MS, van Belle G. Risk factors for subarachnoid hemorrhage. Stroke. 1985;16:377–385.
Longstreth WT Jr, Nelson LM, Koepsell TD, van Belle G. Clinical course of spontaneous subarachnoid hemorrhage: a population-based study in King County, Washington. Neurology. 1993;43:712–718.
Ingall TJ, Whisnant JP, Wiebers DO, O’Fallon WM. Has there been a decline in subarachnoid hemorrhage mortality? Stroke. 1989;20:1150–1155.
Truelsen T, Bonita R, Duncan J, Anderson N, Mee E. Changes in subarachnoid hemorrhage mortality, incidence, and case fatality in New Zealand between 1981–1983 and 1991–1993. Stroke. 1998;29:2298–2303.
Østbye T, Levy AR, Mayo NE. Hospitalization and case-fatality rates for subarachnoid hemorrhage in Canada from 1982 through 1991: the Canadian Collaborative Study Group of Stroke Hospitalizations. Stroke. 1997;28:793–798.
Juvela S, Hillbom M, Numminen H, Koskinen P. Cigarette smoking and alcohol consumption as risk factors for aneurysmal subarachnoid hemorrhage. Stroke. 1993;24:639–646.
Sudlow CLM, Warlow CP. Comparing stroke incidence worldwide: what makes studies comparable. Stroke. 1996;27:550–558.
Longstreth WT Jr, Nelson LM, Koepsell TD, van Belle G. Cigarette smoking, alcohol use, and subarachnoid hemorrhage. Stroke. 1992;23:1242–1249.
Bonita R, Anderson C, Broad J, Jamrozik K, Stewart-Wynne E, Anderson N. The incidence and case fatality of stroke in Australasia: comparison of the Perth and Auckland population-based stroke registers. Stroke. 1994;25:552–557.
Vermeulen M, van Gijn J. The diagnosis of subarachnoid hemorrhage. J Neurol Neurosurg Psychiatry. 1996;243:496–501.
Clayton D, Hills M. Statistical Models in Epidemiology. Oxford, England: Oxford University Press; 1993.
Elliott WJ. Circadian variation in the timing of stroke onset: a meta-analysis. Stroke. 1998;29:992–996.
Bonita R, Broad JB, Beaglehole R. Ethnic variations in stroke incidence and case fatality: the Auckland Stroke Study. Stroke. 1997;28:758–761.
Waterhouse J, Muir CS, Correa P, Powell J. Cancer Incidence in Five Continents. Vol 3. Lyon, France: International Agency for Research on Cancer; 1976:456.
Statistical Analysis System Institute Inc. SAS Version 6.0. Cary, NC: SAS Inc; 1991.
SPSS for Windows, Version 7.5.1. Chicago, Ill: SPSS Inc; 1996.
Bamford J, Sandercock P, Dennis M, Burn J, Warlow C. A prospective study of acute cerebrovascular disease in the community: the Oxfordshire Community Stroke Project 1981–86, II: incidence, case fatality rates and overall outcome at one year of cerebral infarction, primary intracerebral hemorrhage and subarachnoid hemorrhage. J Neurol Neurosurg Psychiatry. 1990;53:16–22.
Teunissen LL, Rinkel GJE, Algra A, van Gijn J. Risk factors for subarachnoid hemorrhage: a systematic review. Stroke. 1996;27:544–549.
Linn FHH, Rinkel GJE, Algra A, van Gijn J. Incidence of subarachnoid hemorrhage: role of region, year, and rate of computerized tomography: a meta-analysis. Stroke. 1996;27:625–629.
Sandercock PAG, Allen CMC, Corston RN, Harrison MJG, Warlow CP. Clinical diagnosis of intracranial hemorrhage using Guys Hospital Score. BMJ. 1985;291:1675–1677.
Wiebers DO, Feigin VL, Brown RD. Cerebrovascular Disease in Clinical Practice. New York, NY: Little, Brown & Co; 1997.