| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the University of Maryland Medical System, Baltimore (M.A.S.); the
Cleveland Clinic Foundation, Cleveland, Ohio (C.A.S., E.J.T.); Duke Clinical
Research Institute, Durham, North Carolina (K.W.M., C.B.G., C.L.G., R.M.C.);
University of Washington, Seattle (W.T.L., W.D.W.); Erasmus Universiteit,
Rotterdam, the Netherlands (P.K., M.L.S.); Green Lane Hospital, Auckland, New
Zealand (H.D.W.); and the University of Massachusetts Medical Center,
Worcester (J.M.G.).
Correspondence to Michael A. Sloan, MD, Department of Neurosciences, Harbin Clinic, 1825 Martha Berry Blvd, Rome, GA 30165-1698.
Methods and ResultsWe performed an observational
analysis within a randomized trial of 4
thrombolytic therapies, conducted in 1081 hospitals in
15 countries. Patients presented with ST-segment elevation
within 6 hours of symptom onset. Our population was composed of the 268
patients who had primary intracranial hemorrhage after
thrombolysis. With univariable and
multivariable analyses, we identified clinical and brain
imaging characteristics that would predict 30-day mortality among these
patients. CT or MRI were available for 240 patients (90%). The 30-day
mortality rate was 59.7%. Glasgow Coma Scale score, age, time from
thrombolysis to symptoms of intracranial
hemorrhage, hydrocephalus, herniation, mass effect,
intraventricular extension, and volume and location
of intracranial hemorrhage were significant univariable
predictors. Multivariable analysis of 170 patients with
complete data, 98 of whom died, identified the following independent,
significant predictors: Glasgow Coma Scale score (
ConclusionsThis model provides excellent discrimination between
patients who are likely to live and those who are likely to die after
thrombolytic-related intracranial
hemorrhage; this may aid in making decisions about the
appropriate level of care for such patients.
In the Global Utilization of Streptokinase and t-PA (alteplase) for
Occluded Coronary Arteries (GUSTO-I)
trial,14 268 patients had intracranial
hemorrhage.15 The baseline clinical
predictors of 30-day mortality in the entire population have been
reported.16 We wished to develop a model that
also incorporated neurological and brain imaging variables to
better predict 30-day mortality among patients who experience
intracranial hemorrhage after thrombolysis for
AMI.
Classification of Intracranial Hemorrhage
Volume and Extent of Lesions
Grading of intraventricular hemorrhage and
hydrocephalus has been described.17 The presence
of subarachnoid hemorrhage was graded
semiquantitatively according to the method of Hijdra et
al.23 Brain edema was defined as the low-density
region contiguous to the hematoma. The volume of edema was measured by
the 3-dimensional volumetric methoddrawing a pencil line around the
border of the edema, measuring the volumes of the hematoma and edema,
and subtracting hematoma volume from the total. Mass effect was graded
according to the degree of ventricular compression and
cisternal effacement (ambient cistern, quadrigeminal cistern). The
presence and direction of any herniation were recorded: subfalcial,
downward, or upward.
Clinical Features
Statistical Analysis
A multivariable logistic regression model24
examined individual and joint relations between baseline clinical
predictors of mortality in the GUSTO-I
population,16 neurological and neuroimaging
features, and death within 30 days of randomization for the 268
patients with intracranial hemorrhage during hospitalization.
These baseline variables were examined graphically and
statistically to assess the assumption that they related linearly to
the log-odds of the outcome event (30-day mortality). Adjustments were
made for nonlinear relations. We also tested for interactions among the
significant variables. Complete data existed for 170 patients
(63.4%); these data were used to construct the multivariable
outcome prediction model.
Odds ratios and 95% CI for each variable in the full model were
computed. For this analysis, total volume was truncated at a
lower limit of 50 mL (the 25th percentile) to meet linearity
assumptions. The predictive performance of the model was
validated with bootstrapping techniques. The model was refitted for
each of 50 bootstrap samples (which included samples of the same size
as the original population, drawn randomly with replacement from the
original sample), then tested on the original sample. This estimates
the predictive accuracy of the model if applied to an independent
sample of patients.25 26 27 28 The measure of the
model's predictive discrimination was the area under the
receiver-operator characteristic curve (the C-index). This curve shows
all possible pairwise sensitivity and specificity values for the
regression model. The C-index measures the concordance of predictions
with actual outcomes29 30 ; an area of 1.00
represents a model with perfect discrimination. Calibration of
the model was assessed by comparing the average model prediction
with the observed mortality rate across deciles of
risk.16 A nomogram was constructed using the
coefficients of a simplified version of the regression model to predict
the individual risk of mortality at 30 days in patients with
intracranial hemorrhage after
thrombolysis.31
Patients with intracranial hemorrhage were more likely to die
within 30 days if they had a lower Glasgow Coma Scale score, a
combination parenchymal-subdural hematoma, multiple parenchymal
intracerebral hemorrhages, or a deep
parenchymal intracerebral hemorrhage
(especially within the putamen/internal capsule and brain stem; Table 2
Univariable and Multivariable Predictors of
Mortality
The most significant independent predictors of 30-day mortality (in
decreasing order of importance) were time from
thrombolysis to hemorrhage, Glasgow Coma Scale
score, total hemorrhagic volume, and the GUSTO-I clinical predictors
(Table 4
The coefficients from the simplified model were used to create a
nomogram to predict 30-day mortality in individual patients (Figure 1
Figure 2
Mortality associated with spontaneous parenchymal
intracerebral hemorrhage has been found to vary
from 83% for pontine to 16% for subcortical lobar
sites.32 33 34 Independent predictors of mortality
from spontaneous supratentorial
intracerebral hemorrhage have included
hemorrhage volume,19 20 35 36 37 38 low
Glasgow Coma Scale score,19 20 35 36 37 38 pulse
pressure,35 37 38
intraventricular
hemorrhage,35 and an interaction term
between intraventricular hemorrhage and
Glasgow Coma Scale score.35 The reported
sensitivity, specificity, and accuracy of these models is 62% to 97%,
87% to 97%, and 92%,
respectively.19 20 35 37 38 These models allow
development of highly accurate algorithms to predict 30-day mortality
in patients with supratentorial parenchymal
hemorrhage, which can aid in the selection of appropriate
patients for aggressive medical or surgical
treatment.19 20 35 37 38
Predictors of increased mortality 30 days after intracranial
hemorrhage in GUSTO-I are similar to those of spontaneous
supratentorial parenchymal hemorrhage and
oral anticoagulant therapyrelated intracerebral
hemorrhage. Univariable analyses show significant
relations between 30-day mortality and GUSTO-I clinical
predictors,16 Glasgow Coma Scale score, hematoma
location, hematoma volume, time from thrombolysis to
intracranial hemorrhage onset, hydrocephalus, herniation, mass
effect, and intraventricular hemorrhage. In
fact, the combined parenchymal-subdural lesion was the most ominous,
being fatal in 28 of 36 patients (78%). Our results thus confirm and
extend the results from the first 3 Thrombolysis and
Angioplasty in Myocardial Infarction (TAMI) trials and the Duke
Databank for Cardiovascular
Disease.10
In multivariable analysis, Glasgow Coma Scale score
and total hemorrhage volume were independent predictors of
30-day mortality, similar to results reported for spontaneous
supratentorial parenchymal
hemorrhage.19 20 35 37 38 In addition, a
shorter time from thrombolysis to intracranial
hemorrhage onset was a significant, independent predictor of
30-day mortality. Our data suggest some correlation between shorter
time from thrombolysis to intracranial
hemorrhage onset and intracranial hemorrhage volume.
This implies a temporal link between the biological effects of
thrombolysis, clinical concomitants of intracranial
hemorrhage, and mortality. Alternatively, mild intracranial
hemorrhages may not have been recognized as quickly as were
severe ones. Although intracranial hemorrhage volumes were
significantly larger in patients treated with combined
thrombolytic therapy,17 this
factor was not an independent predictor of mortality. We also found
that the baseline clinical predictors of 30-day mortality in the
overall GUSTO-I population, especially age, were independently
associated with 30-day mortality from intracranial hemorrhage.
These 4 factors accounted for 93.1% of the full model's
discriminative power. In the simplified model, the larger overall
sample size of GUSTO-I and the effects of Glasgow Coma Scale score and
hemorrhage volume may have diminished the strong general effect
of age on mortality after infarction.39 40
Our findings have implications for the management of intracranial
hemorrhage after thrombolysis. It is not known
whether patients with severe intracranial hemorrhage soon after
thrombolytic therapy should receive only supportive
medical care41 or should be aggressively managed
(treatment of increased intracranial pressure, ventriculostomy,
neurosurgical evacuation). For now, the use of Figure 1
This study has several limitations. First, the models were
derived from a retrospective analysis. Second, the use of
best-estimated Glasgow Coma Scale scores may have introduced some
misclassification bias. The arbitrary assignment of a Glasgow Coma
Scale score of 7 to patients with "coma" or "unresponsiveness"
may be considered too conservative, however, thus underestimating its
contribution to the model. In addition, the excellent agreement between
a cardiologist and a neurologist in independently assigning Glasgow
Coma Scale scores suggests that misclassification bias may have been
minimized. However, prospective determination of Glasgow Coma Scale
scores would have been better. Third, missing data, particularly for
Glasgow Coma Scale scores, may have limited the power of the
analysis and introduced bias, although the sample size of our
multivariable model (n=170) compares favorably with those of recent
studies of mortality prediction for spontaneous
supratentorial parenchymal
hemorrhage.19 35 37 38 Finally, the
treating clinician's perception of poor outcome from intracranial
hemorrhage may have influenced the decision to provide
aggressive support for these patients. As a result, the model reported
here may partly reflect the clinician's behavior in treating these
patients. This model should be validated in a larger, unselected
population.
Received August 26, 1997;
revision received June 10, 1998;
accepted June 16, 1998.
2.
Gruppo Italiano per lo Studio della Streptochinasi
nell'Infarto Miocardico (GISSI). GISSI-2: a factorial randomised trial
of alteplase versus streptokinase and heparin versus no heparin among
12 490 patients with acute myocardial infarction. Lancet. 1990;336:6571.[Medline]
[Order article via Infotrieve]
3.
The International Study Group. In hospital mortality
and clinical course of 20 891 patients with suspected acute myocardial
infarction randomised between alteplase and streptokinase with or
without heparin. Lancet. 1990;336:7175.[Medline]
[Order article via Infotrieve]
4.
ISIS-3 (Third International Study of Infarct Survival)
Collaborative Group. ISIS-3: a randomised comparison of streptokinase
vs tissue plasminogen activator vs anistreplase
and of aspirin plus heparin vs aspirin alone among 41 299 cases of
suspected acute myocardial infarction. Lancet. 1992;339:753770.[Medline]
[Order article via Infotrieve]
5.
Kase CS, Pessin MS, Zivin JA, del Zoppo GJ, Furlan AJ,
Buckley JW, Snipes RG, LittleJohn JK. Intracranial hemorrhage
after coronary thrombolysis with tissue
plasminogen activator. Am J
Med. 1992;92:384390.[Medline]
[Order article via Infotrieve]
6.
Sloan MA. Thrombolysis and intracranial
hemorrhage. In: Feldmann E, ed.
Intracerebral Hemorrhage. Armonk,
NY: Futura; 1994:99149.
7.
Maggioni AP, Franzosi MG, Farina ML, Santoro E, Celani
MG, Ricci S, Tognoni G. Cerebrovascular events after myocardial
infarction: analysis of the GISSI trial: Gruppo Italiano per lo
Studio della Streptochinasi nell'Infarto Miocardico (GISSI).
BMJ. 1991;302:14281431.
8.
AIMS Trial Study Group. Effect of
intravenous APSAC on mortality after acute myocardial
infarction: preliminary report of a placebo-controlled clinical trial.
Lancet. 1988;1:545549.[Medline]
[Order article via Infotrieve]
9.
AIMS Trial Study Group. Long term effects of
intravenous anistreplase in acute myocardial infarction:
final report of the AIMS study. Lancet. 1990;335:427431.[Medline]
[Order article via Infotrieve]
10.
Uglietta JP, O'Connor CM, Boyko CB, Aldrich H,
Massey EW, Heinz ER. CT patterns of intracranial hemorrhage
complicating thrombolytic therapy for acute myocardial
infarction. Radiology. 1991;181:555559.
11.
de Jaegere PP, Arnold AA, Balk AH, Simoons ML.
Intracranial hemorrhage in association with
thrombolytic therapy: incidence and clinical predictive
factors. J Am Coll Cardiol. 1992;20:289294.
12.
Gore JM, Sloan M, Price TR, Randall AM, Bovill E,
Collen D, Forman S, Knatterud GL, Sopko G, Terrin ML, and the TIMI
investigators. Intracerebral hemorrhage,
cerebral infarction, and subdural hematoma after acute myocardial
infarction and thrombolytic therapy in the
Thrombolysis In Myocardial Infarction Study: TIMI Phase II,
Pilot and Clinical Trial. Circulation. 1991;83:448459.
13.
Sloan MA, Price TR, Petito CK, Randall AMY, Solomon RE,
Terrin ML, Gore J, Collen D, Kleiman N, Feit F, Babb J, Herman M,
Roberts WC, Sopko G, Bovill E, Forman S, Knatterud GL, for the TIMI
Investigators. Clinical features and pathogenesis of
intracerebral hemorrhage following rt-PA and
heparin therapy for acute myocardial infarction: the
Thrombolysis in Myocardial Infarction (TIMI) II Pilot and
Randomized Clinical Trial combined experience. Neurology. 1995;45:649658.
14.
The GUSTO Investigators. An international randomized
trial comparing four thrombolytic strategies for acute
myocardial infarction. N Engl J Med. 1993;329:673682.
15.
Gore JM, Granger CB, Simoons ML, Sloan MA, Weaver WD,
White HD, Barbash GI, Van de Werf F, Aylward PE, Topol EJ, Califf RM,
for the GUSTO-I Investigators. Stroke after
thrombolysis: mortality and functional outcomes in the
GUSTO-I trial. Circulation. 1995;92:28112818.
16.
Lee KL, Woodlief LH, Topol EJ, Weaver WD, Betriu A, Col
J, Simoons M, Aylward P, Van de Werf F, Califf RM, for the GUSTO-I
Investigators. Predictors of 30-day mortality in the era of reperfusion
for acute myocardial infarction. Results from an international trial of
41 021 patients. Circulation. 1995;91:16591668.
17.
Gebel JM, Sila CA, Sloan MA, Granger CB, Mahaffey
KW, Weisenberger J, Green CL, White HD, Gore JM, Weaver WD, Califf RM,
Topol EJ, for the GUSTO-I Investigators.
Thrombolysis-related intracranial hemorrhage: a
radiographic analysis of 244 cases from the GUSTO-1
trial with clinical correlation. Stroke. 1998;29:563569.
18.
Kwak R, Kadoya S, Suzuki T. Factors affecting the
prognosis in thalamic hemorrhage. Stroke. 1983;14:493500.
19.
Broderick JP, Brott TG, Duldner JE, Tomsick T, Huster
G. Volume of intracerebral hemorrhage: a
powerful and easy-to-use predictor of 30 day mortality.
Stroke. 1993;24:987993.
20.
Lisk DR, Pasteur W, Rhoades H, Putnam RD, Grotta JC.
Early presentation of hemispheric
intracerebral hemorrhage: prediction of outcome
and guidelines for treatment allocation. Neurology. 1994;44:133139.
21.
Hier DB, Davis KR, Richardson EP, Mohr JP. Hypertensive
putaminal hemorrhage. Ann Neurol. 1977;1:152159.[Medline]
[Order article via Infotrieve]
22.
Kase CS, Williams JP, Wyatt DA, Mohr JP. Lobar
intracerebral hematomas: clinical and CT
analysis of 22 cases. Neurology. 1982;32:11461150.
23.
Hijdra A, Brouwers PHAM, Vermeulen M, van Gijn J.
Grading the amount of blood on computed tomograms after
subarachnoid hemorrhage. Stroke. 1990;21:11561161.
24.
Hosmer DW, Lemeshow S. Applied Logistic
Regression. New York, NY: John Wiley & Sons, Inc; 1989.
25.
Harrell FE Jr, Pollock BG, Lee KL. Graphical methods
for the analysis of survival data. In: Proceedings of the
12th Annual SAS Users Group International Conference. Cary, NC:
SAS Institute Inc; 1987:11071115.
26.
Harrell FE Jr, Lee KL, Pollock BG. Regression models in
clinical studies: determining relationships between predictors and
response. J Natl Cancer Inst. 1988;80:11981202.
27.
Chambers JM, Hastie JT. Statistical Models in
S. Pacific Grove, Calif: Wadsworth & Brooks/Cole; 1992.
28.
Harrell FR Jr. Transcan. S function for
transformation and imputation using canonical variates. Available from
http://www.statlib@lib.stat.cmu.edu; 1993. Request "send transcan
from S."
29.
Lusted LB. Decision-making studies in patient
management. N Engl J Med. 1971;284:416424.
30.
Hanley JA, McNeil BJ. The meaning and use of the area
under a receiver operating characteristic (ROC) curve.
Radiology. 1982;143:2936.
31.
Harrell FE Jr. Design: S plus functions for
biostatistical/epidemiologic modelling, testing, estimation,
validation, graphics, prediction, and typesetting by storing enhanced
model design attributes in the fit. UNIX version available from
http://www.statlib@lib.stat.cmu.edu; 1996.
32.
Kase CS, Mohr JP. General features of
intracerebral hemorrhage. In: Barnett HJM, Mohr
JP, Stein BM, Yatsu FM, eds. Stroke: Pathophysiology, Diagnosis
and Management. New York, NY: Churchill-Livingstone;
1986:497523.
33.
Caplan LR. Clinical features of spontaneous
intracerebral hemorrhage. In: Kaufman HH, ed.
Intracerebral Hematomas. New York, NY: Raven
Press, Ltd.; 1992:3147.
34.
Feldmann E. Intracerebral
hemorrhage. Stroke. 1991;22:684691.
35.
Tuhrim S, Dambrosia JM, Price TR, Mohr JP, Wolf PA,
Hier DB, Kase CS. Intracerebral hemorrhage:
external validation and extension of a model for prediction of 30 day
survival. Ann Neurol. 1991;29:658663.[Medline]
[Order article via Infotrieve]
36.
Portenoy RK, Lipton RB, Berger AR, Lesser ML, Lantos G.
Intracerebral hemorrhage: a model for the
prediction of outcome. J Neurol Neurosurg Psychiatry. 1987;50:976979.
37.
Tuhrim S, Dambrosia JM, Price TR, Mohr JP, Wolf PA,
Heyman A, Kase CS. Prediction of intracerebral
hemorrhage survival. Ann Neurol. 1988;24:258263.[Medline]
[Order article via Infotrieve]
38.
Tuhrim S, Horowitz DR, Sachen M, Godbold JH. Validation
and comparison of models predicting survival following
intracerebral hemorrhage. Crit Care
Med. 1995;23:950954.[Medline]
[Order article via Infotrieve]
39.
White HD, Barbash GI, Califf RM, Simes RJ, Granger CB,
Weaver WD, Kleiman NS, Aylward PE, Gore JM, Vahanian A, Lee KL, Ross
AM, Topol EJ. Age and outcome with contemporary
thrombolytic therapy: results from the GUSTO-I trial.
Circulation. 1996;94:18261833.
40.
Maggioni AP, Maseri A, Fresco C, Franzosi MG, Mauri F,
Santoro E, Tognoni G, on behalf of the Investigators of the Gruppo
Italiano per lo Studio della Sopravvivenza nell'Infarto Miocardico
(GISSI-2). Age-related increase in mortality among patients with first
myocardial infarctions treated with thrombolysis.
N Engl J Med. 1993;329:14421448.
41.
Sane DC, Califf RM, Topol EJ, Stump DC, Mark DB,
Greenberg CS. Bleeding during thrombolytic therapy for
acute myocardial infarction: mechanisms and management. Ann
Intern Med. 1989;111:10101022.
© 1998 American Heart Association, Inc.
Clinical Investigation and Reports
Prediction of 30-Day Mortality Among Patients With Thrombolysis-Related Intracranial Hemorrhage
![]()
Abstract
Top
Abstract
Introduction
Methods
Results
Discussion
References
BackgroundLimited information
exists on risk factors for mortality after
thrombolysis-related intracranial hemorrhage.
We wished to determine the characteristics associated with 30-day
mortality after thrombolysis-related intracranial
hemorrhage.
2,
19.3; P<0.001), time from thrombolysis
to intracranial hemorrhage (
2, 15.8;
P<0.001), volume of intracranial hemorrhage
(
2, 11.6; P<0.001), and baseline
clinical predictors of mortality in the overall GUSTO-I trial
(
2, 10.3; P=0.001). The final model had a
C-index of 0.931.
Key Words: thrombolysis hemorrhage mortality prognosis myocardial infarction
![]()
Introduction
Top
Abstract
Introduction
Methods
Results
Discussion
References
Intracranial hemorrhage is the most feared
complication of thrombolytic therapy for acute
myocardial infarction (AMI). Its frequency in large trials varies from
0.22% to 0.70%, depending in part on the thrombolytic
agent used.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Several of these trials have
shown mortality rates from this complication of 48% to
74%.1 2 3 4 5 7 8 9 10 11 12 13 14 15 Higher doses of the
thrombolytic agent,12 13 the type
of thrombolytic agent used,4 a
larger volume of intracerebral hemorrhage,
significant mass effect with midline shift, and multiple bleeding
sites10 have been associated with increased
mortality after intracranial hemorrhage in past
thrombolytic trials, but such analyses have
included relatively few patients with this complication.
![]()
Methods
Top
Abstract
Introduction
Methods
Results
Discussion
References
The GUSTO-I study included 41 021 patients AMI who
presented with ST-segment elevation within 6 hours of symptom
onset from 1081 hospitals in 15 countries. Descriptions of the entire
population, thrombolytic and adjunctive treatments, end
points, data management and quality assurance, reporting, and
classification of cerebrovascular events have been
reported.14 15
Criteria for parenchymal intracerebral
hemorrhage, subdural hematoma,
intraventricular hemorrhage, and
subarachnoid hemorrhage have been
published.6 12 13 15 17 If multiple lesions or
lesion types were present, the most important lesion, clinically
speaking, was the largest or the one that best explained the
patient's neurological state. The features of each hemorrhagic lesion
on brain images were classified centrally by 2 investigators (C.A.S.,
M.A.S.), as described.17
Each lesion was identified by brain imaging and measured by
investigators blinded to patient treatment and
outcome.17 A pencil line was drawn around the
margin of the lesion on all available brain image slices. For
parenchymal intracerebral hemorrhages, 2
methods of volume measurement were used: the modified (greatest
vertical height x greatest anterior-posterior extent x
greatest medial-lateral extent)/2 method13 18 19 20 21 22
or the 3-dimensional stereotactic/planimetric method
(proprietary software run on a Sun Microsystems SparcStation),
performed by the Center for Computer Assisted Neurosurgery, Department
of Neurosurgery, Cleveland Clinic Foundation. For subdural hematomas,
volumes were calculated on the basis of a further modification of the
first method shown above.18 19 20 21 22 The greatest
vertical height was calculated as for the intracerebral
hemorrhages. The maximum anterior-posterior lesion length was
estimated by drawing a straight line between the anterior and posterior
borders of the lesion. The medial-lateral extent of the lesion was
estimated from the slice portion that showed the greatest distance
between the cortical surface and the inner table of the
skull.23 We used the volumes calculated by the
3-dimensional method for all analyses reported here.
Two independent investigators (M.A.S., K.W.M.) retrospectively
determined the Glasgow Coma Scale score at the initial evaluation of
202 of the 268 patients with intracranial hemorrhage (75.4%),
using all available clinical information. Patients described as being
"in coma" or "unresponsive" and who had no additional
discriminating information were assigned a score of 7. If seizures were
associated with the clinical presentation, the best
estimated Glasgow Coma Scale score before the seizure or after complete
recovery from the postictal state was used. Agreement between the 2
evaluators was excellent; for the 174 patients classified by both, the
median (25th, 75th percentile) absolute difference in scores was 0
(0,1), and the mean± SD absolute difference was 1.0±1.8.
Baseline characteristics were summarized as frequencies and
percentages for categorical data and as medians and interquartile
ranges for continuous variables.
2 and
Fisher's exact tests were used to examine relations between
categorical variables, whereas the Wilcoxon rank-sum test
was used for continuous variables.
![]()
Results
Top
Abstract
Introduction
Methods
Results
Discussion
References
In 240 patients with intracranial hemorrhage, the most
prominent lesion was a parenchymal intracerebral
hemorrhage in 197 patients (82.1%), a subdural hematoma in 7
patients (2.9%), and both in 36 patients (15%). The parenchymal
intracerebral hemorrhage group included
subarachnoid and intraventricular
hemorrhages (2 had intraventricular or
subarachnoid hemorrhages only). The overall mortality
rate from intracranial hemorrhage was 59.7%. Patients who died
were older and had intracranial hemorrhage sooner after
thrombolysis than those who survived (Tables 1
and 2
).
The median (25th, 75th percentiles) time from treatment to death for
the 160 patients with intracranial hemorrhage who died was 54.4
hours (24, 134). The median time from stroke symptom onset to death for
these patients was 40.5 hours (12, 116).
View this table:
[in a new window]
Table 1. Baseline Clinical
Characteristics
View this table:
[in a new window]
Table 2. Neurological Event-Related
Characteristics
). For other neuroimaging characteristics, the presence of mass
effect, herniation, hydrocephalus, and
intraventricular hemorrhage were all
significantly more common in patients who died.
Intraventricular hemorrhage was an
independent predictor of 30-day mortality, but there was no relation
between intraventricular hemorrhage volume
and 30-day mortality (median, 14 mL; interquartile range 4 to 36 mL for
patients who died versus 11 mL [5 to 23 mL] for survivors,
P=0.56).
In univariable analysis, a lower Glasgow Coma Scale
score was most strongly associated with 30-day mortality, followed by
shorter time from thrombolysis to hemorrhage;
larger total hemorrhagic volume; and the presence of hydrocephalus,
herniation, or mass effect (Table 3
). The
baseline clinical predictors of 30-day mortality in the overall GUSTO-I
population and location of hemorrhage also were significantly
associated with 30-day mortality.
View this table:
[in a new window]
Table 3. Univariable Predictors of Mortality After
Thrombolysis-Related Intracranial Hemorrhage
). Because these 4 variables
accounted for 93% of the total prognostic information, we developed a
reduced version of this model (Table 5
).
Because age was by far the most predictive variable in the overall
GUSTO-I model,16 a simpler model that substituted
age alone for the combined "GUSTO-I clinical predictors" yielded
similar results (Table 5
).
View this table:
[in a new window]
Table 4. Multivariable Predictors of Mortality After
Thrombolysis-Related Intracranial
Hemorrhage
View this table:
[in a new window]
Table 5. Multivariable Predictors of Mortality After
Thrombolysis-Related Intracranial
Hemorrhage
). For an 80-year-old with a Glasgow
Coma Scale score of 5 due to an 80-mL intracranial hemorrhage
occurring 20 hours after thrombolysis, the estimated
probability of 30-day mortality would be 14+16+93+5=128, or
90%.
For a 60-year-old with a Glasgow Coma Scale score of 12 due to a 30-mL
intracranial hemorrhage occurring 40 hours after
thrombolysis, the estimated probability of 30-day
mortality is 7+5+87+0=99, or 7.5%.

View larger version (39K):
[in a new window]
Figure 1. Nomogram to predict individual risk of mortality
at 30 days in patients with intracranial hemorrhage after
thrombolysis, using the simplified model given in Table 5
. The C-index for the model is 0.923.
shows the calibration
(reliability) of the full model (Table 4
). This figure shows the
observed 30-day mortality rates across deciles of predicted risk versus
the predicted mortality rates of the full multivariable model. The
validation of the full model with bootstrapping techniques showed that
the C-index was overoptimistic by 0.01, yielding a bias-corrected
C-index of 0.93.

View larger version (9K):
[in a new window]
Figure 2. Graph showing observed 30-day mortality after
thrombolysis- related intracranial hemorrhage
versus mortality predicted by the full multivariable logistic
regression model (Table 4
). Patients were grouped into deciles
according to their predicted probability of death. The actual mortality
rate for the patients within each decile (dots) is plotted against the
average model prediction. The solid line reflects perfect calibration
of the model predictors.
![]()
Discussion
Top
Abstract
Introduction
Methods
Results
Discussion
References
We anticipated that several factors would be highly predictive of
30-day mortality among patients with intracranial hemorrhage
after thrombolysis, such as Glasgow Coma Scale score,
total intracranial hemorrhage volume, and age. We also
discovered a powerful new factor: time from
thrombolytic treatment to onset of symptoms of
intracranial hemorrhage. With a model that incorporates these
and other variables, mortality among these patients can be
predicted with a high degree of accuracy.
may aid in the
clinician's empirical decision-making for such patients.
![]()
Acknowledgments
This study was funded by Genentech (South San Francisco, Calif),
Bayer (New York, NY), CIBA-Corning (Medfield, Mass), ICI
Pharmaceuticals (Wilmington, Del), and Sanofi Pharmaceuticals
(Paris, France).
![]()
References
Top
Abstract
Introduction
Methods
Results
Discussion
References
1.
Gruppo Italiano per lo Studio della Streptochinasi
nell'Infarto Miocardico (GISSI). Effectiveness of
intravenous thrombolytic treatment in acute
myocardial infarction. Lancet. 1986;1:397402.[Medline]
[Order article via Infotrieve]
This article has been cited by other articles:
![]() |
S. Schulman, R. J. Beyth, C. Kearon, and M. N. Levine Hemorrhagic Complications of Anticoagulant and Thrombolytic Treatment: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition) Chest, June 1, 2008; 133(6_suppl): 257S - 298S. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. G. Goodman, V. Menon, C. P. Cannon, G. Steg, E. M. Ohman, and R. A. Harrington Acute ST-Segment Elevation Myocardial Infarction: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition) Chest, June 1, 2008; 133(6_suppl): 708S - 775S. [Abstract] [Full Text] [PDF] |
||||
![]() |
V. Menon, R. A. Harrington, J. S. Hochman, C. P. Cannon, S. D. Goodman, R. G. Wilcox, H. J. Schunemann, and E. M. Ohman Thrombolysis and Adjunctive Therapy in Acute Myocardial Infarction: The Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy Chest, September 1, 2004; 126(3_suppl): 549S - 575S. [Abstract] [Full Text] [PDF] |
||||
![]() |
E. M. Ohman, R. A. Harrington, C. P. Cannon, G. Agnelli, J. A. Cairns, and J.W. Kennedy Intravenous Thrombolysis in Acute Myocardial Infarction Chest, January 1, 2001; 119 (2009): 253S - 277S. [Full Text] [PDF] |
||||
![]() |
T. G. Phan, M. Koh, R. A. Vierkant, and E. F.M. Wijdicks Hydrocephalus Is a Determinant of Early Mortality in Putaminal Hemorrhage Stroke, September 1, 2000; 31(9): 2157 - 2162. [Abstract] [Full Text] [PDF] |
||||
![]() |
Predicting Mortality in Patients with ICH After Thrombolysis for AMI Journal Watch Emergency Medicine, December 1, 1998; 1998(1201): 9 - 9. [Full Text] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Circulation Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 1998 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |