The CHAT Classification of Stroke
EUGENE F. BERNSTEIN, M.D., PH.D., and NORMAN L. BROWSE, M.D., F.R.C.S.
Current terminology for clinical episodes relating to stroke is
inconsistent and unclear, does not permit inclusion of data re-
garding the location and magnitude of extraanial and intra-
cerebral arterial disease, does not coincide with existing clas-
sifications in Europe, and characteizes a hemispheric entity only,
as opposed to a global description including prior symptoms in
both hemispheres. A new classification system (CHAT) has been
designed to deal with these problems, including the current clin-
ical presentation, historical clinical episodes, the site and patho-
logic type of arterial disease, and information regarding abnor-
malities of the brain. Using this system, a retrospective review
of 480 consecutive carotid endarterectomies is presented, dem-
onstrating the advantages of the CHAT classification. Data in-
clude a significant difference in the probability of survival after
carotid endarterectomy for asymptomatic stenosis in patients
with prior symptoms on the opposite side, as well as a significant
difference in the probability of stroke-free survival between pa-
tients with amaurosis fugax and those with prior carotid cortical
symptoms (TIAs) as the presentng dinical condition. The CHAT
classification is suggested as a sinificnt advance in the reporti
of all surgical cerebrovascular disease experience, and has par-
ticular implications for the current randomized trials between
medical and surgical therapy for carotid artery disease.
T n HE MODERN CLASSIFICATION of stroke was in-
troduced by Whisnant, Siekert, and Millikan in
1960' and includes most of the terms and con-
cepts that are generally accepted in Europe and North
America as describing the clinical presentations of this
disease (Fig. 1). Their definition of clinical categories in-
cluded the term incipient stroke, or transient focal cerebral
ischemia (which has evolved to the term transient ischemic
attack for clinical events that completely reverse within
24 hours or less), as well as the terms advancing stroke or
progressing stroke for focal symptoms that progress over
a period of hours or days. During the ensuing 25 years,
other attempts to subdivide stroke problems in different
Correspondence: Eugene F. Bernstein, M.D., Division of Vascular
and Thoracic Surgery, Scripps Clinic and Research Foundation, La Jolla,
CA 92037. (Reprints will not be available from the author.)
Submitted for publication: June 27, 1988.
242
From the Departments of Surgery, Scripps Clinic and
Research Foundation, La Jolla, California, and St. Thomas'
Hospital Medical School, London, England
ways have resulted in significant incompatibility between
the systems currently in use. In the Whisnant description,
there was no provision for asymptomatic patients with
cerebrovascular pathology. In 1973, Natali and Thevenet2
introduced a system in which asymptomatic patients were
classified as Stage 0 (Fig. 1). Subsequently, Vollmar3 pro-
posed another classification in which asymptomatic pa-
tients were identified as Stage I; other abbreviations and
acronyms were also introduced (Table 1). In addition, a
detailed "Classification and Outline of Cerebrovascular
Diseases" was initially developed in 1958 by the Com-
mittee on Cerebrovascular Disease of the National Insti-
tute of Neurological Diseases and Blindness chaired by
Millikan.4 This was revised in 1975, but now has so many
subdivisions that it has proven difficult to use in reporting
or comparing clinical experiences.5
Thus, the continuing problems in the classification of
stroke include: 1) confficts between the existing classifi-
cations of Whisnant, Natali, and Vollmar, and the re-
sulting difficulties in comparing data published from dif-
ferent countries; 2) controversy as to whether the patient
should be considered as having a clinical presentation that
is a hemispheric entity (in which case only the current
presentation is described) as opposed to a global entity
(in which case a global description includes prior symp-
toms in both hemispheres); 3) failure to include the site(s)
and degree(s) ofarterial disease and its relationship to the
clinical presentation; 4) lack of information regarding
brain parenchymal lesions as discovered by computed to-
mography (CT) and magnetic resonance imaging scanning
(MRI); 5) failure to define the degree of recovery; 6) sub-
division by the duration of symptoms; 7) confusing ab-
breviations and acronyms; and 8) an inability to classify
Vol. 209 No. 2 THE CHAT CLASSIFICATION OF STROKE 243
Incipient Progressing Completed TABLE 1. Acronyms Used to Classify or Describe
(Intermittent insufficiency; '(advancing) * (stable) Cerebral Arterial Disease
transient ischaemic attacks) x ,,,t
,1 00 q.. Acronym Definition
SFR Stroke with full recovery
RIA Reversible ischemic attack
Whisnant ; TIA Transient ischemic attack
RIND Reversible ischemic neurologic deficit
PRIND Partially reversible ischemic neurologic deficit
PRIND Prolonged reversible ischemic neurologic deficit
PRINS Partially reversible ischemic neurologic symptoms
TRINS Totally reversible ischemic neurologic symptoms
IRINS Irreversible ischemic neurologic symptoms
Absent Sx month 68hor|Weso
onh

1973l |stage 0 Stage _Stage II |Stage I I The CHAT Classification

Voilmar Stage Stage II (<24 hrs.) PRINS=mild
1980 Stage 111(>24 hrs.) Stage IV=severe The CHAT system has been designed for application
to all varieties of stroke from any cerebro-vascular term-
F;IG. 1. Schematic comparison ofthe classification systems of Whisnant, tory. The term CHA T is an acronym where
Natali, and Vollmar.'~3 Note the differing terminology and stage num-
bering for similar clinical situations. C designates current clinical presentation
H is for historical clinical episodes
A describes the site(s) and pathologic type of arterial
diffiuse symptoms not easily localized to a particular area disease
of the brain. T adds information regarding abnormalities in the tar-
In an attempt to deal with these questions, a Sympo- get organ, which in the case of stroke is the brain. Each
sium was organized by Robert Courbier in Marseilles in of these major classification letters is followed by one or
more subscripts.
September 1984, with participation by international rep-
resentatives from neurology and vascular surgery.6'7 Those
present at the Marseilles Symposium agreed that the most Clinical Subscripts (0-8)
important basis for the classification of cerebral arterial The first subscript to "C" would deal with the type of
disease must continue to be the patient's symptoms, and current clinical presentation, which would generally be
they developed a code of carefully described symptoms
(Table 2). In addition, an effort was made to make the
terminology consistent, using the word stroke as part of TABLE 2. Marseille Classification ofCerebral Arterial Diseases*
the description of each ofthe clinical presentations. Thus
Classification of the Patient and Description of Disease Grade
the term transient and reversible stroke replaced that of
transient ischemic attack in referring to clinical events Classification of the Patient
having a duration of less than 24 hours, and prolonged Asymptomatic (on both sides) 0
and reversible stroke was used to describe episodes lasting Symptoms of stroke in relation to a clinically defined
vascular territory (carotid artery, vertebrobasilar
more than 24 hours but less than 3 weeks. Details of this artery)
ckassification system have been published,6'7 but no clinical Transient and reversible stroke
evaluation of this proposal has been presented as yet. Short TIA (<24 hours) IA
Prolonged and reversible stroke (>24 hours, <3
Stimulated by the Marseilles meeting, we have at- weeks) IB
tempted to describe a clinically usefull system that would Changing stroke (<3 weeks) II
be easy to remember and universally acceptable, analo- Improving
Worsening
gous to the TNM (tumor, nodes, metastases) system for Established stroke (>3 weeks) IIIA
tumor classification. It aims to combine information re- With mild deficit
garding the current illness, historical background, arterial With severe deficit IIIB
Symptoms in relation to a clinically uncertain
disease, and underlying pathology in a manner permitting vascular territory
an easy classification of patients, based on the information Symptoms not of vascular origin
available at any time during the course of their illness. Description of Disease
Pathogenesis
This report describes our proposed classification and pre- Vascular lesion
sents information concerning its initial evaluation when Cerebral hemispheric lesion
applied to a consecutive series of patients undergoing ca- Risk factors
rotid endarterectomy. * References 6 and 7.
 BERNSTEIN AND BROWSE Ann. SUrg * Februr 1989
244
TABLE 3. CHAT Classification of Stroke: Brief stroke is what has previously been called a tran-
("C')-Current Clinical (<1 Year) sient ischemic attack (an episode lasting less than 24 hours)
Subscripts (Cl, Table 3).
to "C" Symptoms Subscripts Vascular Territory Temporary stroke with full recovery (C2) is used to de-
0 Asymptomatic a Carotid ocular scribe those episodes lasting more than 24 hours but with
(amaurosis fugax) return to normal neurologic function within 3 weeks.
I Brief stroke (TIA, b Carotid cortical Permanent stroke (C3 or C4) is used to describe those
<24 hours) conditions lasting more than 3 weeks. Permanent strokes
2 Temporary stroke c Vertebrobasilar
with full with minor residua permitting an independent existence
recovery (24 are classified as C3, and are separated from those with
hours to 3 major residual neurologic dysfunction (patients who are
weeks)
3 Permanent stroke, d Other focal permanently dependent and require continuing assis-
minor (<3 tance, C4).
weeks) The term Nonspecific dysfunction (C5) has been in-
4 Permanent stroke, e Diffuse
major (>3 cluded for those patients in whom the presenting symp-
weeks) toms do not clearly fit a precise vascular territory of origin
5 Nonspecific or focal symptom complex.
dysfunction
6 Improving stroke The previous Whisnant classification of progressing or
7 Fluctuating stroke advancing stroke has been retained and subdivided to dis-
8 Deteriorating tinguish between those patients in whom the condition is
stroke
worsening and those in whom it is improving. Thus, C6
represents improving stroke, C7 represents fluctuating
hemispheric (Tables 3 and 7). Each major clinical pre- stroke, and C8 identifies deteriorating stroke. These three
sentation has been described as some variety of stroke. descriptive terms apply during the first 3 weeks of symp-
toms only. After the first 3 weeks, the patient should be
The Oxford English Dictionary states that the term stroke
implies "stroke of God's hand" and cites a reference from classified into one of the first five stable categories.
1599.8 The Greek verb plesso, which means to "stroke, Localization of the symptom complex to a major vas-
hit, or beat" is the other common origin of the term, and cular territory is noted by the use of small letters a-e, as
is the basis for plegmos, which means apoplectic stroke, indicated in Table 3. Carotid ocular symptoms (a) are
and the postfix plegia, as in paraplegia and hemiplegia. distinguished from those related to the carotid motor cor-
Because these terms have the strongest historical back- tical circulation territory (b), and from vertebrobasilar
symptoms (c). In addition, a subdivision has been created
ground for the description of the disease in each of its
presentations, it seemed appropriate to eliminate the term for focal symptoms that cannot be localized into any of
transient ischemic attack, and substitute it with phrases the primary three areas (d). Finally, symptoms that are
not focal, but are diffuse or vague are classified as (e).
that use the word stroke.
-Thus, an asymptomatic patient would be classified as CO,
and a patient presenting with a brief episode ofamaurosis
TABLE 4. CHAT Classification of Stroke: ("H'7-History (>I Year) fugax would be classified Cla.
Subscripts
to "H" Symptoms Subscripts Vascular Territory History
0 Asymptomatic a Carotid ocular
(amaurosis fugax) The same symptom classification used in "C" (with
I Brief stroke (TIA, b Carotid cortical exception of the subgroups C6.8, which define an acutely
<24 hours) changing presentation) is used to describe previous epi-
2 Temporary stroke c Vertebrobasilar
with full sodes of cerebrovascular symptoms, with a similar as-
recovery (24 signment to the responsible vascular territory (Table 4).
hours to 3 Therefore, a patient with a current brief motor stroke and
weeks)
3 Permanent stroke, d Other focal no previous history of symptoms would be classified as
minor (<3 CibHo, whereas a similar patient with a prior history of
weeks) major permanent motor stroke would be classified as
4 Permanent stroke, e Diffiuse ClbH4b. Because the magnitude of cerebrovasular disease
major (>3
weeks) in the second patient might well be far more severe, sig-
5 Nonspecific nificantly increasing the operative risk and clouding the
dysfunction long-term outlook, it seems useful to provide for a sepa-
Vol. 209 - No. 2 THE CHAT CLASSIFICATION OF STROKE 245
ration of such patients, who have been lumped together TABLE 5. CHAT Classification of Stroke: ("A ')-Artery
in prior classification systems and in most, if not all, re- Subscripts
ports of clinical experience in cerebrovascular disease. The to "A" Site Subscripts Pathology
most important role of the historical section of the clas-
sification is that it permits considering the clinical pre- 0 No lesion a Arteriosclerosis
1 Appropriate lesion c Cardiogenic
sentation of the patient as a global, rather than a hemi- 2 Lesion only in another embolus
spheric, entity. Thus, prior symptoms in either hemisphere vascular pathway d Dissection
will be recorded. 3 Combined (appropriate e Aneurysm
lesion + lesion in f Fibromuscular
Artery another vascular dysplasia
pathway) r Arteritis
The arterial classification may be based on any available t Trauma
information, including ultrasonic echography, digital Significant = <50% stenosis or pathology believed to be the source of
subtraction angiography, or formal contrast angiography symptoms.
(Table 5). If no lesion is present on an adequate angio-
graphic study, the patient is classified as Ao. If a lesion is Initial Evaluation of the CHAT Classification
believed to be responsible for the patient's symptoms and
is in an anatomically appropriate artery for causing A retrospective review of the clinical records of 407
symptoms, it is classified as A1. If arterial lesions are seen patients undergoing 480 consecutive carotid endarterec-
only in vessels other than those anatomically relevant to tomies by two surgeons at a single institution (Scripps
the symptoms, they are classed as A2. A3 includes patients Clinic, 1978-1985) was performed by an individual not
in whom both appropriate lesions and lesions in other involved with the development ofthis proposal. Data were
vascular pathways are demonstrable. collected regarding the early and late follow-up of these
A "significant" lesion is defined as one that results in patients (97% of whom were followed for an average of
greater than 50% stenosis, or is otherwise believed to be 30 months). Analyses of the results in terms ofimmediate
the cause of the presenting symptoms. Others have quan- and late strokes and death were then carried out for var-
titated the number of vessels involved in more detail,4 ious subsets within the classification in an effort to deter-
but when more subdivisions are added, the number of mine whether meaningful but previously unknown rela-
patients in each group becomes quite small and the groups tionships or distinctions could be identified.
become statistically insignificant. Thus, this approach, All data were entered on a VAX model 11/750 com-
with four major subgroups of arterial disease ranging from puter (Digital Equipment Corp., Maynard, MA) using
none to extensive, was designed to be a reasonable com- CLINFO (BBN Software Products Corp., Cambridge,
promise. The artery classification scheme also includes a MA) software for clinical research management analysis.
second subscript for defining the major kinds of recog- Life-table analyses were performed according to the pro-
nizable arterial pathology, which may produce similar cedures of Kaplan and Meier9 and Gehan.'0
symptoms, but have significantly different implications In Table 8, the distribution of the 407 patients is sum-
and prognoses. marized according to the CHAT scheme, with an indi-
cation of how CHAT differs from the current clinical
Target scheme employed in the U.S. and the U.K. The largest
The presence of brain pathology should be based on subgroup, TIA, represented 49.6% of the indications for
CT scanning, MRI, or any other method that demon- the first carotid operation in this series. However, in the
strates abnormality; but whatever method is used, it CHAT classification, these indications are further sub-
should be cited in the report. The suggested classification
system for the site of an abnormality is analogous to the TABLE 6. CHAT Classification ofStroke: ("T') Target
one proposed for arterial pathology, and includes a second Subscripts
suffix for defining any known information regarding the to "T" Site Subscripts Pathology
type of pathology in the central nervous system lesion
0 No lesion h Hemorrhage
(Table 6). I Appropriate lesion i Infarct
The definition of the word significant in the target organ 2 Lesion in another 1 Lacunar
pathology classification is any lesion that is clearly ab- vascular territory m Atriovenous -
normal, regardless of whether it is believed to be the source only malformation
3 Combined (appropriate n Neoplasm
of the symptoms. Thus, a lesion in an appropriate cerebral lesion + lesion in q Other
territory is classified as T,, in an inappropriate territory another vascular r Retinal embolus
it is called T2, and it is classified as T3 iflesions are present territory)
in both appropriate and inappropriate areas. Significant = pathology believed to be the source of symptoms.
 BERNSTEIN AND BROWSE Ann. Surg. * February 1989
246
TABLE 7. Proposed Classifcation of Stroke
"H"
"C" History (>l year) "A" "'rs
Current Clinical (<1 year) Artery Target
Symptoms/Vascular
Symptoms Vascular Territory Territory Site Pathology Site Pathology
(0) Asymptomatic (a) Carotid ocular (0) No lesion (a) Arteriosclerosis (0) No lesion (h) Hemorrhage
(1) Brief stroke TIA (amaurosis ditto (1) Appropriate lesion (c) Cardiogenic (1) Appropriate lesion (i) Infarct
(<24 hours) fugax) Current Clinical (2) Lesion in another embolus (2) Lesion in another (I) Lacunar
(2) Temporary stroke (b) Carotid cortical 1-5 a-e vascular (d) Dissection vascular (m) AVM
with full (c) Vertebrobasilar pathway only (e) Aneurysm territory only (n) Neoplasm
recovery (24 (d) Other focal (3) Combined (f) Fibromuscular (3) Combined (q) Other
hours to 3 (e) Diffuse (appropriate (r) Arteritis (appropriate (r) Retinal emb
weeks) lesion + lesion (t) Trauma lesion + lesion
(3) Permanent stroke, in another in another
minor (< 3 vascular vascular
weeks) pathway) territory)
(4) Permanent stroke,
major (> 3
weeks)
(5) Nonspecific
dysfunction
(6) Improving stroke
(7) Fluctuating stroke
(8) Deteriorating
stroke

Significant = >50% stenosis or pathology believed to be the source of symptoms.

divided as ocular (18.9%), carotid cortical (24.8%) and
vertebrobasilar (5.9%). Information regarding the prob-
ability of successfully improving the stroke-free survival
of each of these groups will be derived from future anal-
yses.
Figure 2 demonstrates the difference in probability of
survival after carotid endarterectomy for asymptomatic
stenosis when the patient population is divided on the
basis of prior symptoms in the opposite hemisphere. Pre-
vious contralateral cerebrovascular symptoms signifi-

TABLE 8. Distribution of Patients' Clinical Presentations According to CHAT Scheme: Scripps Clinic (1978-1985)
Current Clinical (<I year) CHAT Classification Current
Clinical
CHAT Class Symptoms and Vascular Termtory No. % aassification
Co Asymptomatic 117 28.5 28.5%
C, Brief stroke (<24 hours)
CIa-carotid ocular 77 18.9
Clb-carotid cortical 101 24.8 49.6% = TIA
Clc-vertebrobasilar 24 5.9
C2 Temporary stroke (24 hours to 3 weeks)
C2-carotid ocular - 1 0.2
C2b-carotid cortical 7 2.9
C2C-vertebrobasilar 0 0
C3 Permanent stroke, minor
C3a-carotid ocular 14 3.4 18.5% = Prior
C3b-carotid cortical 40 9.8
Ck-vertebrobasilar 4 1.0 Stroke
CM-other focal 1 0.2
C4 Permanent stroke, major
C4a-carotid ocular 2 0.5
C4b-carotid cortical 2 0.5
C4c-vertebrobasilar 0 0
C5 Nonspecific dysfunction 12 2.9 2.9%
Total 407 100.0 100%
 Vol. 209 * No. 2 THE CHAT CLASSIFICATION OF STROKE 247
cantly decreased the probability of survival (p = 0.02), PROBABILITY OF SURVIVAL AFTER CAROTID
ENDARTERECTOMY FOR ASYMPTOMATIC STENOSIS
even though the number of patients in the prior symptom
group was very small (n = 8). In Figure 3, a similar analysis 1.0
of the probability of stroke-free survival in patients with
brief ocular stroke or CIa ( < 24 hours) is compared with o.s
I
~~~~~~~(CoHo)
those patients with brief cortical motor stroke, or CIb. ,IIo rorS (n=94)
*2 0.8
The difference is significant at p = 0.0 17. These prelim-
inary data suggest the importance of using the kinds of % 0.7 ', (COH1,3
information contained in the CHAT classification for the 1iPrior Sx, opp. side (n=8)
reporting of all cerebrovascular disease experience, but 0.6

particularly for the trials of medical versus surgical therapy 0.5

now underway for both asymptomatic and symptomatic p-0.02
patients. nfA
0 12 24 36 48 60 72 84 96
Time (months)
Discussion FIG. 2. Life-table analysis ofthe probability ofsurvival after carotid end-
arterectomy for asymptomatic stenosis (C0), demonstrating the signifi-
The introduction and acceptance of a new classification cantly better survival in those patients without any prior contralateral
of stroke will be successful if the general perception is that symptoms (COHO) as compared to those patients with previous contra-
lateral transient or permanent symptoms (CDHI, COH2, and C0H3).
there is a significant need to alter current terminology and
commonly used methods of classification, and if the sug-
gested alternatives seem more logical and useful. In ad- sification system relates to the resolution of current con-
dition, a period of trial and evaluation of the proposals troversies regarding selection of patients for carotid end-
will be necessary for seeing whether the proposed system arterectomy. For example, the patient who has had a prior
is more satisfactory in providing an intelligent separation stroke in one hemisphere with permanent neurologic def-
of different clinical symptom complexes and their meth- icit and CT evidence of one or more infarcts and who
ods of management and prognosis. For example, it would subsequently develops an asymptomatic carotid lesion in
appear intuitively logical to separate the ocular presen- the contralateral artery probably does not have the same
tations of anterior circulation transient episodes from outlook as the patient in whom the asymptomatic stenosis
those that occur within the motor cortex. Indeed, it now is the only abnormality present. Yet, in the currently ac-
appears that a number of subdivisions of the ocular pre- cepted terminology, both of these patients are referred to
sentations may also become appropriate in the future, as having "asymptomatic stenosis." The same term is ap-
because there is a wide spectrum of ocular lesions that plied to those patients who have had contralateral carotid
portend varying prognoses and require different forms of surgery. Certainly, any proposed study of the fate of pa-
therapy. tients with "asymptomatic stenosis" with or without an
The proposed system does not include a stratification
of risk factors, such as hypertension, diabetes, and smok-
PROBABIUTY OF STROKE-FREE SURVIVAL
ing. Although these factors have well-recognized influence AFTER CAROTID ENDARTERECTOMY
on the pathogenesis and prognosis of stroke, they have
not been included because suitable systems for identifying 1.0
and codifying them are already in general clinical use, W ~~Amaurosis fugax n=76
and specific new adaptations of these schemes for the sub- 0.9 (Cia)
classification of stroke do not appear necessary in the ini- .E
0.8 -
Carotid cortical n=101
tial evaluation of the proposed CHAT scheme. (C1
It would be particularly valuable for the proposed sys- 0.7 -

tem to be discussed and considered by all members of the .011

40
concerned medical specialties, particularly neurologists, 0.6 -

neurosurgeons, vascular surgeons, and ophthalmologists,

0.5 -p=0.017
and we hope this will occur. In addition, it would be ap-
propriate if this scheme were subjected to critical trial by 0.410
others than the authors, in order to provide for a wider 12 24 36 48 60 72 84 96
or broader evaluation. In addition, the method could cer- Time (months)
tainly be applied to currently ongoing or proposed clinical FIG. 3. Life-table analysis of.the probability of stroke-free survival after
trials of intervention in cerebrovascular disease. carotid endarterectomy for amaurosis fugax (Cl.) compared with carotid
cortical transient symptoms (CUb) reveals a significantly better prognosis
One of the most important justifications for a new clas- with amaurosis fugax.
248 BERNSTEIN
intervention should take cognizance of the full spectrum
of other abnormalities, as proposed by the current clas-
sification system.
Similarly, the patient who presents with an initial single
transient ischemic attack and no other evidence of disease
except an ipsilateral internal carotid artery lesion should
be differentiated from the patient with a similar presen-
tation who has a significant background of past and/or
contralateral disease and/or prior carotid surgery. Thus,
the major advantage of the currently proposed classifi-
cation is that it gives a more precise definition of the pa-
tient's presenting clinical status, as well as additional in-
formation concerning the historical background of clinical
disease, arterial pathology, and brain pathology. The use-
fulness of such a classification must be proven by clinical
studies whose aim is to determine whether the subclas-
sifications used to separate varying groups of patients will
permit a more accurate assessment of the risk of stroke
or recurrent stroke, and the relative risks and benefits of
various forms of intervention, both medical and surgical.
It is the authors' hope that this method may prove more
AND BROWSE
Ann. Surg. February 1989
useful and acceptable than the existing approaches to the
problem.
References
1. Whisnant JP, Siekert RG, Millikan CH. Appraisal of the current
trend toward surgical treatment of occlusive cerebral vascular
disease. Med Clin North Am 1960; 44:875-886.
2. Natali J, Thevenet A. Chirurgie des Arteres Carotides et Vertebrales
Dans Leur Segment Extra-Cranien. Paris: Masson & Cie, 1973.
3. Voilmar JF. Reconstructive Surgery of the Arteries. Stuttgart: G.
Thieme Verlag, 1980.
4. Millikan CH, Adams RD, Fang H, et al. A classification and outline
of cerebrovascular diseases. Neurology 1958; 8:396 464.
5. Millikan CH, Bauer RB, Goldschmidt J, et al. A classification and
outline of cerebrovascular diseases. Stroke 1975; 6:564-616.
6. Courbier R, ed. Basis for a classification of cerebral arterial diseases,
Excerpta Medica. Curr Clin Pract 1985, Series 22:1 309.AQI2
7. Courbier R. Basis for an international classification of cerebral arterial
diseases. J Vasc Surg 1986; 4:179-183.
8. Dirckx JH. Letter to the Editor. Stroke 1986; 17:559.
9. Kaplan EL, Meier P. Nonparametric estimation from incomplete
observations. Am Stat Assn J 1958; 53:457-481.
10. Gehan EA. A generalized Wilcoxon Test for comparing arbitrarily
singly-censored samples. Biometrika 1965; 52:203-223.