Effect of oral nimodipine on cerebral infarction and outcome after

subarachnoid haemorrhage: British aneurysm nimodipine trial

J D Pickard, G D Murray, R Illingworth, M D M Shaw, G M Teasdale, P M Foy, P R D Humphrey,
D A Lang, R Nelson, P Richards, J Sinar, S Bailey, A Skene

Abstract Conclusions-Oral nimodipine 60 mg four hourly

Objective-To determine the efficacy of oral is well tolerated and reduces cerebral infarction and
nimodipine in reducing cerebral infarction and improves outcome after subarachnoid haemorrhage.
poor outcomes (death and severe disability) after
Wessex Neurological subarachnoid haemorrhage.
Design-Double blind, placebo controlled, Introduction
Centre, Southampton
General Hospital, randomised trial with three months of follow up and Despite advances that have reduced considerably
Southampton S09 4XY intention to treat analysis. To have an 80% chance the risks of operation to secure a ruptured cerebral
J D Pickard, MCHIR, with a significance level of 0-05 of detecting a 50% aneurysm the overall mortality and morbidity during
professor of clinical reduction in an incidence of cerebral infarction of the management of patients with subarachnoid
neurological sciences 15% a minimum of 540 patients was required. haemorrhage who have survived and not been
R Nelson, FRCS, senior Setting-Four regional neurosurgical units in the devastated by the initial ictus has not fallen dramatic-
registrar in neurosurgery ally. This is mainly because such patients rebleed and
S Bailey, RGN, Medical United Kingdom.
Research Council research Patients-In all 554 patients were recruited have delayed cerebral ischaemia. The management
sister between June 1985 and September 1987 out of dilemma remains the need to weigh the risk of
a population of 1115 patients admitted with sub- precipitating cerebral ischaemia by operation against
Medical Statistics Unit, arachnoid haemorrhage proved by the results of that of rebleeding while awaiting surgery.
Department of Surgery, lumbar puncture or computed tomography, or both. Subarachnoid haemorrhage provides the clearest
University of Glasgow, The main exclusion criterion was admission to clinical opportunity for pretreating ischaemia, but,
Glasgow GIl 6NT the neurosurgical units more than 96 hours after although the calcium antagonist nimodipine has been
G D Murray, PHD, senior subarachnoid haemorrhage. There were four reported to reduce the incidence of delayed cerebral
lecturer in medical statistics ischaemia after subarachnoid haemorrhage,' 2
breaks of code and no exclusions after entry. One
Regional Neurosciences patient was withdrawn and in 130 treatment was definitive evidence is lacking. Previous trials were
Centre, Charing Cross discontinued early. All patients were followed up for small and relied on selecting a subgroup of patients
Hospital, London W6 8RF three months and were included in the analysis, with cerebral ischaemia presumed to be due to cerebral
R Illingworth, FRCS, except the patient who had been withdrawn. vasospasm. Ischaemia may be the result of several
consultant neurosurgeon Interventions-Placebo or nimodipine 60 mg was factors35 and cannot be attributed simply to an
P Richards, FRCS, consultant given orally every four hours for 21 days to 276 and exaggerated contraction of cerebral arteries (vaso-
neurosurgeon 278 patients, respectively. Treatment was started spasm); thus defining such a subgroup is difficult and
within 96 hours after subarachnoid haemorrhage. ignores any potential benefit of treatment in other
Mersey Regional causes of ischaemia.
Department of Medical and End points -Incidence of cerebral infarction and
Surgical Neurology, ischaemic neurological deficits and outcome three Nimodipine has various effects on the cerebral
Walton Hospital, Liverpool months after entry. circulation, not all of which may be beneficial.
L9 1AE Measurements-Demographic and clinical data, Cerebrovascular smooth muscle is more sensitive
M D M Shaw, FRCS, including age, sex, history of hypertension and than systemic arterial smooth muscle to changes in
consultant neurosurgeon subarachnoid haemorrhage, severity of haemorrhage extracellular calcium concentration and to calcium
P M Foy, FRCS, consultant according to an adaptation of the Glasgow coma antagonists such as nimodipine.6-8 In addition to being
neurosurgeon scale, number and site of aneurysms on angio- a moderate cerebral vasodilator, however, nimodipine
P R D Humphrey, MRCP, graphy, and initial findings on computed tomography impairs cerebrovascular reactivity.9-" Nimodipine
consultant neurologist does not reduce the incidence or severity of cerebral
were measured at entry. Deterioration, defined as
Institute of Neurological development of a focal sign or fall of more than one vasospasm detected by angiography in humans or
Sciences, Southern point on the Glasgow coma scale for more than six primates,'2-'7 but it reduces the size of cerebral infarcts
General Hospital, Glasgow hours, was investigated by using clinical criteria and in rats when given before but not after occlusion of a
G514TF by computed tomography, by lumbar puncture, or cerebral artery. 8
G M Teasdale, FRCS, at necropsy when appropriate. All episodes of We therefore conducted a prospective trial to
professor of neurosurgery deterioration and all patients with a three month establish the effect of nimodipine on the incidence of
D A Lang, FRCS, registrar in outcome other than a good recovery were assessed both ischaemic events and proved cerebral infarction
neurosurgery by a review committee. and on outcome after subarachnoid haemorrhage.
J Sinar, FRCS, registrar in
neurosurgery Main results-Demographic and clinical data at
entry were similar in the two groups. In patients
Department of given nimodipine the incidence of cerebral infarction Patients and methods
Mathematics, University of was 22% (61/278) compared with 33% (92/276) in SELECTION OF PATIENTS AND DIAGNOSIS
Nottingham, Nottingham those given placebo, a significant reduction of 34% Eligible patients were admitted to one of four centres
NG7 IRD (95% confidence interval 13 to 50%). Poor outcomes (in London, Southampton, Liverpool, and Glasgow)
A Skene, PHD, senior lecturer were also significantly reduced by 40% (95% within 96 hours after the onset of symptoms and signs
in medical statistics confidence interval 20 to 55%) with nimodipine (20% of subarachnoid haemorrhage. This time limit was
Correspondence to: (55/278) in patients given nimodipine v 33% (91/278) specified because our hypothesis was that nimodipine
Professor Pickard. in those given placebo). Adverse reactions were should be given soon after the bleed to have a
reported in 14 patients given nimodipine and 10 given prophylactic effect upon subsequent ischaemia. The
Br.rlcj7 19'9;298:636-42 placebo. diagnosis was confirmed by lumbar puncture or

636 BMJ VOLUME 298 11 MARCH 1989

 computed tomography, or both-angiography was not was deemed to be non-compliant. The protocol
required before admission to the trial. The reasons allowed for separate analysis of this group, but this was
for exclusion were pregnancy; major renal, hepatic, or found not to be necessary. Plasma samples were taken
pulmonary disease; pre-existing cardiac decompensa- from 35 patients (eight taking placebo, 27 nimodipine)
tion or a recent (within six months) myocardial and nimodipine concentrations measured-there was
infarction; age below 18 years; a subarachnoid no evidence of patients given placebo receiving active
haemorrhage that produced a coma in the week treatment, or vice versa.
preceding the most recent subarachnoid haemorrhage;
and the patient or relative being unwilling to give CLINICAL MONITORING
consent. Eligibility for the trial was checked in The severity of subarachnoid haemorrhage was
specially designed record forms with a check list of graded according to an adaptation of the Glasgow coma
entry and exclusion criteria. A separate record was scale,' which is similar to the scale recently adopted by
kept of patients who did not fulfil the entry criteria. the World Federation of Neurosurgical Societies.24
Patients' eligibility was rechecked during outcome Patients whose clinical grade was I had a score on the
review meetings of all the centres. There were no Glasgow coma scale of 15 and were neurologically
exclusions. Written informed consent was obtained intact, apart from having cranial nerve palsy. Those in
from the patient, care having been taken to avoid grade II had a score of 15 and were neurologically
undue stress, or from a relative. The ethical committee intact, apart from having cranial nerve palsy, with neck
of each centre approved the study and the Department stiffness or headache, or both. Those in grade III had a
of Health and Social Security supplied a clinical trial score of 13-14 and were with or without neurological
exemption for the drug before the start of the study. deficit and those in grade IV a score of 8-12 with or
without neurological deficit. Finally, those in grade V
DESIGN OF STUDY had a score of 3-7 and were in a coma with or without
Our aim was to establish with a prospective, abnormal posturing.
multicentre, randomised, double blind, placebo Computed tomography was performed within
controlled trial whether oral nimodipine (60 mg every 24 hours after admission to the trial and repeated
four hours) reduced the incidence of episodes of whenever necessary if the patient's clinical condition
cerebral ischaemia and cerebral infarction arising anew deteriorated at all. Angiography, operation, and other
after spontaneous intracranial haemorrhage from aspects of management remained the responsibility of
a cerebral aneurysm. The outcome events to be the consultant in charge of the patient. The cause
specifically examined prospectively were the inci- of subarachnoid haemorrhage was determined by
dences of cerebral infarction and ischaemic deficit and angiography or at necropsy or, in two cases, from the
the outcome three months after the bleed, with the typical.appearance of a giant aneurysm on a computed
principle distinction being between a moderate or tomogram.
good recovery and a bad outcome (death or severe Patients were monitored clinically throughout their
disability). Each of the centres had experience with time in hospital: blood pressure, pulse rate, and
similar large studies of dipyridamole,'9 antifibrinolytic neurological observations were recorded. Particular
treatment,20 or computed tomography2l in patients attention was paid to the patient's degree of conscious-
with subarachnoid haemorrhage. From these results ness, focal neurological features, and symptoms and
we expected that at least 15% of control patients would signs suggestive of rebleeding, ischaemia, or another
have a poor outcome attributable to cerebral infarction. adverse event.
A minimum of 540 patients was required for the trial to The minimum criterion to define deterioration was
have an 80% chance with a significance level of 0 05 of either development of a focal sign or a decline by more
detecting a 50% reduction in an incidence of cerebral than one point on the Glasgow coma scale for more
infarction of 15%. than six hours. The aetiology of deterioration was
investigated by using clinical criteria with results from
RANDOMISATION computed tomography whenever possible; when
Patients were randomised to treatment or placebo by appropriate it was investigated by lumbar puncture or
using separate randomisation lists for each centre at necropsy. All episodes of deterioration and all
balanced in blocks of four. Bayer UK supplied the patients whose outcome at three months was not a good
pharmacy in each centre with bottles containing recovery were assessed at regular meetings of a review
tablets of either nimodipine or placebo. Bottles were committee comprising representatives from each of the
numbered consecutively and assigned to eligible four centres. The review committee assessed outcome
patients in that order. Each centre was supplied with blind to allocated treatment. Causes of deterioration
cards that could break the code for each patient were classified as rebleed, cerebral ischaemia or
individually and be used if the doctor in charge of the infarction, or other, when the specific complication
patient decided that the code needed to be broken. was recorded. The diagnosis of rebleeding or of
ischaemia or infarction was classified as definitive or
TREATMENT probable according to whether confirmatory evidence
Treatment was started within 96 hours after ictus was available from computed tomography, from an
and routinely continued for 21 days in survivors, operation, or at necropsy.
unless there were clinical indications for stopping. A patient with two infarcts was counted once for the
Nimodipine (or matching placebo) was given as fast statistical analysis of cerebral infarction. If a patient
release tablets containing 30 mg active compound (two had both a rebleed and an infarct he (she) appeared in
tablets given orally every four hours). The dose was both analyses. Deterioration in a patient known to have
based on the results of previous studies.' 1322 Tablets an intracerebral haematoma was deemed to be due to
were swallowed with water, or if the patient was unable ischaemia if as the new signs developed there was a
to swallow the tablets were crushed and washed down a growing area of low density surrounding a clot that
nasogastric tube with normal saline. The protocol had not itself increased on computed tomography.
allowed for adjustment of the dose if hypotension Outcome was assessed at least three months after entry
occurred, but this was not found to be necessary. to the trial according to the five point Glasgow outcome
Compliance was checked while the patient was in the scale2' by a doctor who was not responsible for the
neurosurgical unit by comparing the number of tablets patient's early management or by postal or telephone
prescribed with the number taken. Any patient who inquiry when a patient had moved out of the region. In
missed more than one third of the daily dose on any day patients who were dead or disabled at three months we

BMJ VOLUME 298 11 MARCH 1989 637

 assessed whether the cause was a direct effect of Health and Social Security (clinical trial exemption)
the initial bleed, cerebral ischaemia or infarction, regulations, the pink copy to GDM, and the white
rebleeding, hydrocephalus, intracranial haematoma, copies were kept in each centre. GDM and AS (for
operative complication, another complication of Bayer UK) analysed the data independently.
treatment, or other. In five cases there was sudden
death with no clinical evidence or evidence from
computed tomography, from lumbar puncture, or at Results
necropsy of the cause; these were classified as sudden Between June 1985 and September 1987, 554
death of unknown cause. patients were recruited into the trial out of a total
population of 1115 patients admitted to the four
STATISTICAL ANALYSIS centres with a diagnosis of subarachnoid haemorrhage:
The two treatment groups were compared for rates 151 in Southampton, 101 in London, 199 in Glasgow,
of infarction (primary outcome event), poor outcome and 103 in Liverpool. A total of 108 patients did not
(secondary outcome event), and rebleeding by using fulfil the entry criteria for the trial in Southampton,
x2 tests. These analyses were supported by a detailed 207 did not in Liverpool, 214 in Glasgow, and 32 in
analysis that used stepwise logistic regression to London. The most- common reason for not fulfilling
adjust the comparisons for the effects of any chance the entry criteria was a delay of more than 96 hours
imbalance in the prognostic factors identified from an between bleeding and admission (317 patients). All but
earlier study20 with the addition of data from computed two patients were entered within the time limit of
tomography. Interaction terms between the prognostic 96 hours; these two patients remained in the analysis as
factors and treatment were included in the models to they had violated the criterion by less than 24 hours.
test whether any treatment effects were confined to The mean time from ictus to entry was 1 9 days.
particular subgroups of patients. The prognostic One patient withdrew his verbal consent after
factors considered in this analysis were age; sex; loss of randomisation (placebo group) but before he had
consciousness at ictus; time from haemorrhage to signed the consent form or received any treatment. His
entry; score on the Glasgow coma scale on entry; limb case was fully documented and followed up to three
weakness, neck stiffness, and headache (all on en'try); months (he made a good recovery) but was not
hypertension (defined by a history of hypertension, included in the statistical analysis. There were four
receiving hypertensive treatment on entry, or a breaks of code in two centres: sudden hypotension
diastolic blood pressure greater than 100 mmHg on with induction of anaesthesia (placebo), jaundice of
entry); angiographic findings (positive, negative, or unexplained cause (placebo), sudden intraoperative
not done); and computed tomographic findings hypotension (placebo), and jaundice shown at
(normal, abnormal, or not done). necropsy to have been due to rupture of an aortic
All clinical data generated during the study were aneurysm (nimodipine). Treatment was discontinued
recorded in the patients' record forms by us - the within 21 days in 130 patients (70 given nimodipine,
blue National Cash Register copy was sent monthly to 60 placebo), the main reason being a negative result on
Bayer UK (Dr L Porto) to comply with Department of angiography with no evidence of an imbalance between
treatments (87 patients).
TABLE I-Demographic data on and indices of severity of initial subarachnoid haemorrhage in patients The cases of 272 patients were reviewed by the
treated with nimodipine or placebo. Values are numbers of patients unless stated otherwise committee. No patients were lost to follow up. All
patients recorded as survivors were known to be alive
Patients taking nimodipine Patients taking placebo 90 days after entry.
(n= 278) (n= 276)
Mean (SD) age (years) 46 (13) 48(12) DEMOGRAPHIC AND CLINICAL DATA
Men 114 107
Women 164 169 The demographic data and various indices of
Alcohol abuse 12
severity of subarachnoid haemorrhage (table I) show
Diabetes mellitus 4 3
Chronic airways disease 5 7 the overall comparability of the treatment groups. The
Peripheral vascular disease 2 group given nimodipine had a higher prevalence
Smoker (> 10 cigarettes/day) 120 143
Other cardiovascular disease 2 11 of hypertension, neck stiffness, and non-reacting
Other diseases 28 33 pupils, factors generally considered prognostic of
Hypertension:
History 33 23 poor outcome. The group given placebo, however,
Being treated at entry 20 16 contained more patients with a history of cardio-
Present at entry 40 35 vascular disease and a higher number of smokers.
Any feature 71 52
Abnormal electrocardiogram on admission 28 33
Previous subarachnoid haemorrhage 37 43
Loss of consciousness at ictus 126 129 INTERCENTRE VARIABILITY
Clinical grade:
8 12 The overall comparability of the two groups was
IV 168 159 maintained within centres but there were differences
III 76 72 among the four centres, reflecting different patterns of
IV 19 25
V 7 8 referral, systemic disease, and resources.
Initial computed tomography 273 276 In London older patients (mean age 50) were entered
Time from ictus to scan (days) 1 5 I5
Normal results 35 45 earlier (mean 1 4 days, with fewer admissions too
Cisternal blood (>2 mm thick) 177 168 late for entry (14% (19/133)). The patients were
Intracerebral haematoma 58 61
slightly more hypertensive, with a longer duration of
Intraventricular blood 83 83
Hydrocephalus 34 36 unconsciousness at ictus (in 11% (11/101) it was over
Angiography 243 229 24 hours), a lower clinical grade (55% (55/101) had a
Time from ictus to angiography (days) 5 5 5-1
Aneurysm: good grade), and less neck stiffness (7% (7/101));
Proved 187 181 angiography and operations were performed much
Multiple 40 29
Carotid 68 later (mean of 11 and 19 days, respectively).
Anterior cerebral 81 83 In Southampton patients were entered latest
Middle cerebral 54
Posterior circulation 20 15 (2 3 days) and there was a higher prevalence of
Spasm present 54 46 previous subarachnoid haemorrhage (23% (35/151))
Operations:
No 165 154 and loss of consciousness at ictus (54% (82/151)) and
Time from ictus to operation (days) range) 10-8 (2-60) 113 (2-116) less headache (25% (35/142)). Computed tomography
yielded more normal results (25% (37/147)), with

638 BMJ VOLUME 298 11 MARCH 1989

 fewer proved aneurysms (56% (84/151)). Computed the good recovery end of the scale compared with
tomography in the patients treated with nimodipine placebo. One hundred and three patients died within
yielded fewer normal results (17% (12/72)) than in 90 days after starting treatment, 43 in the group given
those given placebo (33% (25/75)). nimodipine and 60 in that given placebo (table III).
In Liverpool the greatest number of patients
were admitted too late for entry (39% (122/310)), the TABLE iII-Effect of nimodipine on score on Glasgow outcome scale
prevalence of loss of consciousness at ictus was lowest three months after subarachnoid haemorrhage. Values are numbers of
(30% (31/103)), lumbar puncture studies were used the patients
most for diagnosis (91% (94/103); range 46-55% in the
other centres), and patients had the best overall grade. Patients taking nimodipine Patients taking placebo
Outcome (n=278) (n=276)
In Glasgow the prevalence of aneurysms was greatest
(77% (154/199)), including multiple aneurysms (27% Death 43 60
(40/150 angiograms)), and the prevalence of reported Vegetative state 1 2
Disability:
alcohol abuse was lowest (2% (3/199)). Severe 11 29
Overall, the earlier computed tomography was Moderate 24 16
Good recovery 199 169
performed the greater the incidence of significant
cisternal blood and hydrocephalus. The incidence of
proved aneurysms was lowest in Southampton and Mortality in the two groups was not significantly
London (56 and 60%, respectively). different at the 5% level but had an observed sig-
INFARCTS
nificance level of 6% (relative reduction 29% (95%
confidence interval -i to 50); X2=3 60, df= 1).
Nimodipine reduced significantly the number of The principal reason for the improvement in
primary outcome events - that is, the number of all outcome was a reduction in the number of deaths
cerebral infarcts-by 34% (table II). The number of and patients with disability considered to be due to
definite infarcts was reduced by 37% (61 in placebo episodes of delayed ischaemia. Other causes of death or
group, 39 in nimodipine; X2 =6-10, df=1; p=0 014). disability (initial bleed, rebleeding, intracranial
There was no evidence that the time from ictus or entry haematoma, operative complication, hydrocephalus,
to first infarct was affected by treatment (ictus to complications of treatment, other causes, and
infarct: nimodipine 9-0 days, range 1-38; placebo unknown) were similar in the two groups. In particular,
8 4 days, range 1-40). Late infarcts seemed to be recovery in patients who had sustained a severe initial
associated with later operations, although the numbers bleed did not seem to be influenced (table IV).
were small. The effect of nimodipine was seen both
before and after operation and in patients considered to
be too ill for angiography or surgery. TABLE iv-Causes of disability and death in patients taking
The following factors were individually (rather than nimodipine or placebo. Values are numbers (percentages) of patients*
independently) associated with the rate of infarction:
sex (marginal), loss of consciousness, angiographic Disability Death
findings, centre (marginal), hypertension, findings on Patients Patients Patients Patients
computed tomography, and treatment. Adjusting taking taking taking taking
nimodipine placebo nimodipine placebo
simultaneously for findings on angiography and Cause (n=278) (n=276) (n=278) (n=276)
computed tomography and hypertension removed the Cerebral ischaemia 20 (7) 34(12) 17 (6) 25 (9)
effects of sex, loss of consciousness, and centre, but Initial bleed 19 (7) 17 (6) 7(3) 10(4)
treatment remained highly significant (p<0001). Rebleeding 5 (2) 7 (3) 14 (5) 24 (9)
There was no evidence of any interactions between Intracranial haematoma 1 (<1) 1 (<1) 3 (1)
Hydrocephalus 1 (<1) 6 (2) 1 (<1) 2 (1)
prognostic factors and treatment. Operative complicationt 8 (3) 6(2) 4 (1) 3 (1)
In summary, there was clear evidence of a reduced Other 5 (2) 2 (1) I (<1) 7 (3)
Unknown 5 (2)
incidence of infarction in patients given nimodipine.
The difference could not be explained by an imbalance *Patients had more than one cause of disability or death.
in prognostic factors (indeed, adjusting for the tComplications of treatment with either nimodipine or placebo did not
prognostic factors if anything pointed to an enhanced cause any disability or death.
benefit), and there was no evidence that the benefit of
treatment was confined to a particular subgroup of the The following factors were individually related to
patients. outcome: age, sex (marginal), loss of consciousness,
OUTCOME
score on Glasgow coma scale, angiographic findings,
limb weakness, neck stiffness (marginal), hyper-
Nimodipine reduced significantly the secondary tension, computed tomographic findings, and
outcome event-namely, the incidence of poor treatment. Adjusting simultaneously for age, loss of
outcome (dead, vegetative state, and severe disability) consciousness, score on Glasgow coma scale, and
-from 33% (91 patients) to 20% (55)-that is, by 40% findings on angiography and computed tomography
(table II). Given the distribution of outcomes with the removed the effects of sex, limb weakness, neck
five category Glasgow outcome scale, there was stiffness, and hypertension, but treatment remained
clear evidence (significant at the 1% level) that the highly significant (p= 0001). There was no evidence of
distribution of outcome changed with treatment: any interaction between any prognostic factor and
nimodipine shifted the pattern of outcomes towards treatment.
In summary, there was clear evidence that treatment
with nimodipine reduced the incidence of a poor
TABLE II-Effect of nimodipine on incidence of cerebral infarction and outcome after subarachnoid outcome. This could not be explained by an imbalance
haemnorrhage. Values are numbers (percentages) of patients unless stated otherwise in prognostic factors between the treatment groups.
There was no evidence that the benefit of treatment
Patients taking Patients taking Relative 95% was confined to any particular subgroup of the
nimodipine placebo reduction Confidence Significance patients.
(n=278) (n=276) (%) interval (p value)
Cerebral infarct 61 (22) 92 (33) 34 13 to 50 0-003 (, 899; df= 1) REBLEEDS
Poor outcome 55 (20) 91 (33) 40 20 to 55 <0 001 (y = 12-41; df= 1)
Rebleed 25 (9) 38(14) 35 -5to59 0077(7 3 13;df=1) The incidence of rebleeding, either probable or
definite, in patients treated with nimodipine was 9%

BMJ VOLUME 298 11 MARCH 1989 639

 compared with 14% in patients given placebo (table of infarction as soon as the first day and as late as the
II). This difference had an observed significance 40th day aftter the initial bleed-only 37% of
level of 8%. The following factors were individually episodes of infarction occurred from the seventh to the
associated with the risk of rebleeding: age, loss 10th day.
of consciousness, score on Glasgow coma scale Deterioration due to cerebral infarction was chosen
(marginal), angiographic findings, hypertension as the primary outcome event for several reasons.
ymarginal), and computed tomographic findings. Firstly, given the known actions of nimodipine,
Adjusting simultaneously for age and angiographic we considered that any beneficial effect would be
findings removed the effects of the other variables, and shown through a prophylactic effect" and a reduced
there was little evidence of an effect of treatment occurrence of such episodes. Secondly, although
(p=021). This analysis should be given little weight nimodipine might help to preserve neuronal integrity
because there were too few rebleeds to be confident of in ischaemia, many other factors, including specific
having identified relevant factors. treatments such as induced hypertension, contribute
to the natural recovery of function that occurs in some
BLOOD PRESSURE patients with delayed ischaemic deterioration.4 As a
Nimodipine reduced progressively over 21 days result the proportion of patients who are permanently
both systolic (by a mean (SD) of 7 1 (18 5) mmHg; disabled owing to ischaemia may provide a less
p=005) and diastolic blood pressure (by 3 6 sensitive assessment of a prophylactic measure.
(11-0) mmHg; p=0-094). This effect was possibly Furthermore, infarction can be diagnosed by clinical
slightly more obvious in hypertensive patients, but it observation by sequential computed tomography,
was not significant. Treatment had no consistent effect and at necropsy. We intentionally did not interpret
on the natural tendency of the pulse rate to increase particular episodes as being due to cerebral vasospasm.
progressively for 21 days after a subarachnoid Arterial narrowing detected by angiography is only one
haemorrhage (from a mean of 76 (12) to 85 (15) beats/ of the many factors that lead to cerebral ischaemia after
minute) or on fluid intake or urine output. subarachnoid haemorrhage. Others include raised
intracranial pressure (haematoma, hydrocephalus);
ADVERSE REACTIONS episodes of hypotension, hypoxia, hypovolaemia, or
Twenty seven adverse reactions (17 for nimodipine, hyponatraemia; predisposing factors (history of
10 for placebo) were reported in 24 patients (14 given hypertension, advancing age, cigarette smoking); and
nimodipine, 10 given placebo). The highest numbers surgery to clip the aneurysm.45 Indeed, cerebral
of reported events were for cardiovascular effects and vasospasm alone seldom precipitates critical cerebral
effects on the liver and bi iary system. Six adverse ischaemia unless the cerebral circulation can no longer
cardiovascular events were reported in the group given compensate for proximal arterial constriction by
nimodipine (three of headache, one of flushing, one of dilatation of distal cerebral arterioles. So many factors
hypertension, and one of hypotension) and three in the contribute to delayed cerebral ischaemia that an
group given placebo (two of hypotension and one of appropriate subgroup would be difficult to define.
headache). Anyway, beneficial effect of nimodipine was seen in all
Four adverse events concerned with the liver the patients.
and biliary system were reported in patients given The assessment of outcome by the Glasgow outcome
nimodipine (two of jaundice, two of liver function scale is widely accepted. To avoid unduly relying on
abnormalities) and five in those given placebo (one of the distinction between good recovery and moderate
jaundice, four of liver function abnormalities). Apart disability because such a distinction is particularly
from two events of rash with nimodipine and one of subject to variability in observers we distinguished
rash with placebo all other adverse reactions were poor outcomes (death, persistent vegetative state, and
reported on only one occasion in both groups. The severe disability) from good recovery and moderate
adverse reaction was considered to be sufficiently disability combined. The effect on outcome was
severe in eight patients taking nimodipine and three greater perhaps than we would have expected from the
taking placebo for the treatment to be withdrawn. reduction in the number of cerebral infarcts detected
Routine haematological and biochemical laboratory by computed tomography, given that there are other
data were comparable between the two groups. causes of bad outcome. Studies at necropsy have shown
that cerebral infarcts may be small and diffusely
distributed' and hence beyond the resolution of
Discussion computed tomography. Nimodipine might protect
Our results show that oral nimodipine 60 mg every partially against such minor diffuse infarction.
four hours is well tolerated and reduces cerebral We found no evidence that the beneficial effect of
infarction and improves outcome after subarachnoid nimodipine was restricted to men, contrary to a recent
haemorrhage. This effect of treatment is seen in all study of patients with stroke.26
clinical grades of patients both before and after Results compatible with our findings were obtained
operation. Although it is generally accepted that the in smaller controlled trials (table V) and in uncontrolled
peak incidence of delayed ischaemia is around seven to studies.' 322 The pioneering study of Allen et al showed
10 days after the initial bleed, the need for early and an effect of treatment only when a subgroup of patients
also prolonged treatment was shown by the occurrence with ischaemic neurological deficits attributed to

IABIE v-Comparison of results of double blind placebo controlled tnrals of oral timodipine treatment in patients with subarachnoid haemorrhage

Dose No No of patients Grade No of patients with bad outcomes Relative

(everv four hours of on reduction
Studs Year Place for 21 days) centres Total Exclusions Nimodipine Placebo entry Time Nimodipine Placebo (%)
Allen etal 1983 United States 0 35 mg/kg 5 125 9 56 60 1-2 21 days 1 8 86*
Philippon eta!l" 1986 France 60 mg 2 81 11 31 39 1-3 21 days 3 13 71*
Neil-D"wer et al 1987 United Kingdom 60 mg 1 75 25 25 25 1-5 3 months 6 12 50
Current trial 1989 United Kingdom 6( mg 4 554 0 278 276 1-5 3 months 55 91 40
Petruketal' 1988 Canada 90mg 17 188 34 72 82 3-5 121 days
*3 months
53
44
67
54 10
7

*Related to spasm alone.

640 BMJ VOLUME 298 11 MARCH 1989

 vasospasm was defined: a significant benefit was not reduce costs. Exploratory analysis suggests, however,
seen when the whole population was considered.1 that there is still a trend in favour of nimodipine.
Similar results have been reported for outcome at three The mechanism of the beneficial effect ofnimodipine
months in patients of all clinical grades (table V). 6 27 is unclear. The original rationale for using calcium
Data from two controlled studies of intravenous antagonists after subarachnoid haemorrhage was to
nimodipine (1-2 mg/h for seven to 10 days)2829 and from reduce cerebral arterial spasm as seen on angiography.
open trials (preoperative, topical, and intravenous Our observation that spasm was not seen less often in
nimodipine for seven to 14 days followed by oral drug treated patients accords with previous studies that used
for up to 21 days) with historical controls'1322 suggest angiography or transcranial Doppler ultrasonography
that outcome in patients with ischaemia associated in patients and primates.'2'7 The data thus suggest that
with vasospasm was improved. nimodipine neither prevents nor reverses cerebral
Is intravenous nimodipine more effective than oral vasospasm. Calcium channel blockers such as
nimodipine? Although it is difficult to compare the nimodipine may have an effect on cerebral arterioles
patient populations, the reduction in the incidence of that are below the limits of resolution of angiography
bad outcomes with intravenous nimodipine (23% in and transcranial Doppler ultrasonography,6 37 but
the study of Ohman and Heiskanen28 and 36% in that cerebral blood flow studies do not support such a
of Jan et al29) is similar to that in our study (40%). mechanism.27 The brain uses calcium for many other
During intravenous infusion (2 mg/h) the mean plasma processes in addition to the control of cerebrovascular
concentration was reported to be 26-6 (1 8) ng/ml, smooth muscle, and the diverse effects of calcium
while during oral treatment (45 mg every four hours) antagonists are becoming known.38"4"
the peak plasma concentration an hour after each dose How should the fact that oral nimodipine 60 mg
ranged from 7 to 96 ng/ml and was 13 2 ng/ml (range every four hours reduces the incidence of cerebral
<3-38-8 ng/ml) immediately before each tablet was infarction and improves outcome affect overall
taken.30 The area under the curve was similar for the management of subarachnoid haemorrhage? Early
two methods of administration."' Because protein surgery clearly will reduce the incidence of rebleeding,
binds 98% of the drug the ratio of cerebrospinal fluid to but for most neurosurgical units in the United
plasma concentration is only 0 005-the concentration Kingdom more resources are needed to provide such
in cerebrospinal fluid was 0 3 (0 2) ng/ml compared a service. The proved efficacy of nimodipine and
with 77 (34) ng/ml in plasma.3' Whether the difference doctors' ability to assess recovery of normal cerebro-
in plasma concentration with oral and intravenous vascular reactivity in individual patients4' will
administration is reflected in the cerebrospinal fluid encourage selective early surgery and hence earlier
and whether such a difference affects nimodipine's transfer to neurosurgical units. Antifibrinolytic drugs
efficacy in preventing cerebral infarction remains to be reduce rebleeding in the short term but increase
determined. Certainly, the available clinical evidence cerebral infarction.20 Nimodipine might reduce
does not yet suggest any great difference. Oral ischaemic complications produced by antifibrinolytic
administration permits prolonged treatment, which is agents,'7 but, as with the present trial, such a combina-
useful if there is a long delay between the bleed and tion should not be used without the benefit of a
operation. substantial randomised double blind trial. Finally,
Oral nimodipine is well tolerated and in practice has the introduction of nimodipine for subarachnoid
not led to problems with interaction with anaesthetic haemorrhage is not displacing another treatment. The
agents.'2 3 We found no suggestion of episodes of apparent additional expense must, however, be viewed
severe hypotension with nimodipine when compared in the context of the overall cost effectiveness of
with placebo. The mild reduction in blood pressure managing a patient with subarachnoid haemorrhage
noted, particularly in the hypertensive patients,"4 may from ictus to final outcome and consideration should
have contributed to the improved outcome.35 not be restricted simply to the immediate burden on
Would a higher dose have any greater effect? A neurosurgical units' drug budget.
Canadian study used 90 mg every four hours and,
although it was a smaller study of patients with a poor We thank Dr L Porto and Mrs Julia Duffy of Bayer UK for
grade, showed no significant hypotensive effects,'7 but their collaboration and punctilious attention to detail; our
the effects on overall outcome were much less clear cut colleagues for permission to enter their patients into the trial
and for their enthusiastic support (Mr J Brice and Mr J
than in our trial. An unpublished study (Bayer trial No Garfield in Southampton; Mr M Rice-Edwards in London;
3247/3293) comparing 30, 60, and 90 mg orally every Mr J Miles, Mr R V Jeffreys, and Mr G Findlay in Liverpool;
four hours did not show additional benefit with the and Mr T A H Hide, Mr S Galbraith, Mr J Turner, Mr
90 mg dose when compared with the 60 mg dose. In R Johnston, Mr P Stanworth, and Mr D Mendelow in
theory, a patient who metabolises nimodipine quickly Glasgow); and Mrs Jane Roberts for her help with managing
might develop ischaemia and would benefit from a the trial in Southampton. Bayer UK provided the funding
higher dose. required to support the study in the four centres; the conduct
For how long should nimodipine be used? There was of the trial and statistical assessment were our independent
no evidence after nimodipine was discontinued of a responsibilities.
late incidence of cerebral infarction compared with
placebo. Any late infarcts seemed to occur after late I Allen GS, Ahn HS, Preziosi TJ, ei al. Cerebral arterial spasm-a controlled
operations. Prophylactic treatment should start trial of nimodipine in patients with subarachnoid hemorrhage. N EnglJMed
1983;308:619-24.
therefore before any operation, whatever its timing. 2 Anonymous. Calcium antagonists attd aneurysmal subarachnoid haemorrhage
The effect of treatment was independent of when [Editorial]. Lancet 1983;ii: 141-3.
3 Crompton MR. Cerebral infarction following the rupture of cerebral berry
patients were given nimodipine within the 96 hours aneurysms. Brain 1964;87:263-79.
after ictus. The experimental data show that pre- 4 Symon L. Disordered cerebrosascular physiology in aneurysmal subarachnoid
treatment within 20-30 minutes after occlusion haemorrhage. Acta Neurochir (Wien) 1978;41:7-22.
5 Pickard JD, Matheson M, Patterson J, Wyper D. Prediction of late ischemic
of the middle cerebral artery in rats is protective.' complications after cerebral aneurysm surgery by the intraoperative
Cerebral ischaemia after subarachnoid haemorrhage is measurement of cerebral blood flow. 7 Neurosurg 1980;53:305-8.
6 Simeone FA, Vinall P. Mechanisms of contractile response of cerebral artery to
progressive so it is not surprising that nimodipine may externally-applied fresh blood. I Neurosurg 1975;43:37-47.
be helpful even after clinical deterioration starts,'9 7 Brandt L, Andersson KE, Edvinsson L, Ljunggren B. Effects of extracellular
calcium and of calcium antagonists on the contractile responses of isolated
but prophylaxis is to be preferred. Patients with human pial and mesenteric arteries. J Cereb Blood Flow Metabol 1981;1:
negative results on high quality cerebral angiograms are 339-47.
at low risk of rebleeding or cerebral ischaemia and 8 Kazda S, Towart R. Nimodipine: a new calcium antagonistic drug with a
preferential cerebrosvascular action. Acta Neurochir (Wien) 1982;63:259-65.
stopping nimodipine treatment might be considered to 9 Harper AM\, Craigen L, Kazda S. Effect of the calcium antagonist, nimodipine,

BMJ VOLUME 298 1 1 MARCH 1989 641

 on cerebral blood flow and mctabolism in the primate. ] Cereh Blood Flo-w 26 Gielmers HJ, Gorter K, de Weerdt Cj, Wiezer HJA. A controlled trial of
Mletabol 1981;1:349-56. nimodipine in acute ischemic stroke. N EnglJ Med 1988;318:203-7.
10 Harris RJ, Branston NMi, Svmon L, Bayhan M, Watson A. The effects of a 27 Neil-Dwver G, Mee E, Dorrance D, Lowe D. Early intervention with
calciurn antagonist, nimodipine, upon physiological responses of the nimodipine in subarachnoid haemorrhage. Eur Heart J 1987;8(suppl K):
cerebral sasculature and its possible influence upon focal cerebral ischemia. 41-7.
Stroke 1982;13:759-66. 28 Ohman J, Heiskanen 0. Effect of nimodipine on the outcome of patients after
11 Haws CH, Heistad DD. Effects of nimodipine on cerebral sasoconstrictor aneurvsmal subarachnoid hemorrhage and surgery. J Neurosurg 1988;69:
responses. Am ] Physiol 1984;247:H 170-6. 683-6.
12 Grotenhuis JA, Bettag WXI, Fiebach BJO, Dabir K. Intracarotid slow bolus 29 Jan M,, Buchheit F, Tremoulet M. Therapeutic trial of intravenous nimodipine
injection of' nimodipine during angiography for treatment of cerebral in patients with established cerebral sasospasm after rupture of intracranial
vasospasm after subarachnoid hemorrhage. ] Neurosurg 1984;61:231-40. aneurvsms. Neurosurger-e 1988;23:154-7.
13 Ljunggren B, Brandt L, Saseland H, et al. Outcome in 60 consecutive patients 30 Vinge E, Andersson KE, Brandt L, Ljunggren B, Nilsson LG, Rosendal-
treated with early aneurysm operation and intravenous nimodipine. Helgesen S. Pharmacokinetics of nimodipine in patients with aneurysmal
7 Neurosurg 1984;61:864-73. subarachnoid haemorrhage. Eurj Clin Pharmacol 1986;30:421-5.
14 Espinosa F, Weir B, Overton r, e al. A randomized placebo-controlled 31 Ramsch KD, AhrG, Tettenborn D, Auer LM. Ov erview on pharmacokinetics
double-blind trial of nimodipine after subarachnoid hemorrhage in of nimodipine in healthy volunteers and in patients with subarachnoid
monkeys. 1. Clinical and radiological findings. ] Neurosurg 1984;61:231-40. haemorrhage. Neurochirurgia (Stutig) 1985;28(suppl 1):74-8.
15 Nosko M, Weir B, Krueger C, et al. Nimodipine and chronic vasospasm in 32 Stulleken EH, Balestrieri FJ, Prough DS, McWhorter JM. The hemodvnamic
monkeys. I. Clinical and radiological findings. Neurosurgerv 1985;16:129-36. effects of nimodipine in patients anesthetised for cerebral aneurysm
16 Philippon J, Grob R, Dagreou F, Guggiari Ni, Rivierez M, Viars P. Prevention clipping. Anesthestology 1985;62:346-8.
of vasospasm in subarachnoid haemorrhage. A controlled study with 33 Merin RG. Calcium channel blocking drugs and anesthetics-is the drug
nimodipine. -cta N'eurochir (Wien) 1986;82: 110-4. interaction beneficial or detrimental? Anesthesiology 1987;66:111-3.
17 Petruk KC, West NI, Mohr G, et al. Nimodipine treatment in poor 34 Tettenborn D, Dycka J, Volberg E, Dudden P. Blood pressure and heart
grade aneurysm patients. Results of a multicentre, double-blind, placebo rate during treatment with nimodipine in patients with subarachnoid
controlled trial. ] Nreurosurg 1988;68:505-17. haemorrhage. Neurochirurgia (Stuttg) 1985;28(suppl 1):84-6.
18 Gotoh 0, Mohamed AA, McCulloch I, Graham Dl, Harper AM, 35 O'Neill P, West CR, Chadwick DW, et al. Post-ictal blood pressure in
Teasdale GM. Nimodipine and the hemodvnamic and histopathological aneurysmal subarachnoid haemorrhage. BrJ Neurosurg 1988;2:153-60.
consequences of middle cerebral artery occlusion in the rat.]J Cereb Blood 36 Brandt L, Ljunggren B, Andersson KE, et al. Effects of topical application of a
FlowMeiab 1986;6:321-31. calcium antagonist (nifedipine) on feline cortical pial microvasculature
19 Shaw MDNi, Foy PM, Conway M, e al. Dipyridamole and post-operative under normal conditions and in focal ischemia. J Cereb Blood Flowv Metabol
ischemic deficits in aneurysmal subarachnoid hemorrhage. ] Neurosurg 1983;3:44-50.
1985 ;63:699-703. 37 Takavasu M, Bassett JE, Dacey RG. Effects of calcium antagonists on
20 V'ermeulen NI, Lindsay KW, Cheah MF, ei al. Antifibrinolytic treatment in intracerebral penetrating arterioles in rats. J Neurosurg 1988;69: 104-9.
subarachnoid hemorrhage. N EnglJ]fed 1984;311:432-7. 38 Middlemiss DN, Spedding M. A functional correlate for the dihydropyridine
21 Mohsen F, Pomonis S, Illingworth R. Prediction of delayed cerebral ischaemia binding site in rat brain. Nature 1985;314:94-6.
after subarachnoid haemorrhage by computed tomography. ] Neure' 39 Pickard JD, Walker V, Vile J, Perry S, Smythe PJ, Hunt R. Oral nimodipine
Neurosurg Psychialry 1984;47:1197-202. reduces prostaglandin and thromboxane production by arteries chronically
22 Auer LM. Acute operation anid preventive nimodipine improve outcome in exposed to a periarterial haematoma and the antifibrinolvtic agent
patients with ruptured aneurysms. Neurosurgerv 1984;15:57-66. tranexamic acid. J Neurol Neurosurg Psychiatry 1987;50:727-3 1.
23 Teasdale GM, Lindsay KW, Allardyce G, Dharker S, Ward P. Standardized 40 Phillis JW, O'Regan MH, Walter GA. Effects of nifedipine and felodipine on
clinical grading of patients with subarachnoid haemorrhage: a uniform adenosine and inosine release from the hypoxemic rat cerebral cortex.
international system? In: Auer LNI, ed. Timing of aneurysm surgery. Berlin: J Cereb Blood Flouw Metab 1988;8:179-85.
Walter de Gruvter, 1985:9-14. 41 Pickard JI), Read DH, Lovick AHJ. Preoperative assessment of cercbro-
24 Teasdale GNI. A universal subarachnoid haemorrhage scale: report of a sascular reactivity following subarachnoid haemorrhage-clinical
committee of the World Federation of Neurosurgical Societies. J Neurol correlations. In: Auer LM, ed. Ttmtng of aneurvsm surger,. Berlin: Walter de
NeurosurgPsychiatry 1988;51:1457. Gruyter, 1985:47-51.
25 Jennett B, Bond M. Assessment of outcome after severe brain damage. A
practical scale. Lancet 1975;i:480-4. (Accepted 20 December 1988)

Increased risk of atherosclerosis in women after the menopause

Jacqueline C M Witteman, Diederick E Grobbee, Frans J Kok, Albert Hofman, Hans A Valkenburg
Abstract may help to clarify this issue. Information on the extent
An increase in the incidence of cardiovascular of atherosclerosis in premenopausal and postmeno-
disease has generally been observed in postmeno- pausal women has been gained mainly from necropsy
pausal women, but there have been few studies of studies. Compared with women with intact ovaries
the association between menopausal state and women who had had bilateral oophorectomy had an
atherosclerosis. In this study 294 premenopausal excess of coronary atherosclerosis, which approached
and 319 postmenopausal women aged 45 to 55 were that in men. We investigated the association between
examined radiographically for calcified deposits in menopausal state and the presence of calcified deposits
the abdominal aorta, which have been shown to in the abdominal aorta as seen on lateral radiographs of
represent intimal atherosclerosis. Aortic athero- the lumbar spine. The presence of such deposits
sclerosis was present in eight (3%) of the premeno- represents true intimal atherosclerosis5 and is a strong
pausal women and in 38 (12%) of the postmenopausal predictor of death from cardiovascular causes.6
women. After adjustments for age and other
Department of indicators of cardiovascular risk women with a
Epidemiology, Erasmus natural menopause had a 3*4 times greater risk of Subjects and methods
University Medical School, atherosclerosis than premenopausal women (95% A population study on chronic diseases was
Rotterdam, confidence interval 1-2 to 9-7; p<O0O5); women who conducted in Zoetermeer, a suburb of The Hague
The Netherlands had had a bilateral oophorectomy had a 5*5 times in The Netherlands, between 1975 and 1978. All
Jacqueline C M Witteman, greater risk (1-9 to 15-8; p<O0OO5). No excess risk of inhabitants of one rural and one urban district who
MSC, resident in epidemiology atherosclerosis was observed among women who were aged 5 years and over were invited for a medical
Diederick E Grobbee, MD, had had a hysterectomy without removal of both examination. The response rate among middle aged
senior lecturer in epidemiology women was 77%. Details of the initial study have been
Frans J Kok, PHD, senior ovaries.
lecturer in epidemiology These results suggest that when oestrogen reported.7 All women aged 45 to 55 without a history of
Albert Hofman, MD, production stops, either naturally or after surgery, cardiovascular disease (n=676) were eligible for our
professor of epidemiology the risk of atherosclerosis is increased. study. The lower age limit was 45 because radiography
Hans A Valkenburg, MD, of the lumbar spine was performed from this age
emeritus professor of onwards. The upper age limit of 55 was chosen to
epidemiology Introduction include sufficient premenopausal and postmenopausal
An increase in the incidence of cardiovascular women of similar ages.
Correspondence to: disease after the menopause has generally been We diagnosed atherosclerosis by radiographic
Miss Witteman. detection of calcified deposits in the abdominal aorta.6
observed, though not all studies agree with this
BrMed7 1989;298:642-4 finding.' 2 Non-invasive assessment of atherosclerosis Lateral radiographs of the lumbar spine were available

642 BMJ VOLUME 298 1i MARCH 1989