950
Acute Stroke Therapy Trials:
Problems in Patient Accrual
Linda J. LaRue, PhD, Milton Alter, MD, PhD, Neal D. Traven, PhD,
Arnold B. Sterman, MD, Eugene Sobel, PhD, and Jude Kleiner, RN
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
Temple University Hospital participated in a multicenter acute stroke trial but enrolled only one
patient out of 192 screened over 2 years; other centers had similar difficulty in patient
recruitment. We analyzed our screening data to determine which enrollment criteria created
difficulties in recruitment and whether the problem was attributable to any single criterion or
to combinations of criteria. Six individual criteria were frequent causes for exclusion; however,
>80% of the patients were excluded for multiple reasons. Consequently, modifying or
eliminating any single criterion did not appreciably increase patient accrual. Only 17 of 210
possible pairs of criteria occurred with statistically significant frequency (/><0.05), and these
were most likely random associations. Therefore, only by minimizing the number and
stringency of enrollment criteria will patient accrual be at a level that allows the study to be
completed in a timely manner with a fiscally reasonable number of centers. (Stroke 1988;19:
950-954)
T
he efficacy of new therapies for acute stroke
can be assessed formally only by controlled
clinical trials. However, one difficulty in
conducting such trials has been patient accrual. For
example, the recent Scandinavian multicenter hemodilution trial enrolled only 32% of 604 patients
screened.1 Other stroke studies have experienced
even greater difficulties. After 980 patients were
screened for a study of naftidrofuryl in acute cerebral infarction, only 100 (10.2%) met the enrollment
criteria and were actually included in the study.2
Another recent 4-year trial of intravenous heparin
for acute partial stable thrombotic stroke enrolled
only 7.4% of the screened patients.3
Patient accrual was difficult in these studies in
part because restrictive enrollment criteria have
been devised to provide as homogeneous a population of stroke patients as possible. This is a reasonable approach since a given therapy may be targeted toward modifying only a specific aspect of
stroke pathophysiology. While such homogeneity
simplifies interpretation of results and increases the
chance of detecting any beneficial effect of the new
From the Neuroepidemiology Section, Temple University
Hospital, Philadelphia, Pennsylvania (L.J.L., M.A., N.D.T.,
J.K.), Sandoz Research Institute, East Hanover, New Jersey
(A.B.S.), and the Division of Biostatistics, Department of Preventive Medicine, University of Southern California School of
Medicine, Los Angeles, California (E.S.).
Address for correspondence: Linda J. LaRue, PhD, Neurology Department, Temple University Hospital, Broad and Tioga
Streets, Philadelphia, PA 19140.
Received January 12, 1988; accepted April 6, 1988.
therapy on specific stroke types, these restrictive
criteria can lead to difficulty in enrolling enough
patients within a reasonable period and in generalizing the study results to the broader population of
stroke patients.
Temple University Hospital (TUH) was involved
recently in an acute stroke multicenter clinical trial
of a calcium channel blocker that used restrictive
enrollment criteria formulated to maximize homogeneity among patients. However, application of
these criteria produced an unacceptably low patient
enrollment. We describe the enrollment criteria and
their successive revisions. The frequency with which
each criterion alone and in combination prevented
accrual of patients was analyzed. Based on this
experience, we gained insights that should help in
the design of future clinical trials.
Subjects and Methods
A multicenter clinical trial of a new calcium
channel blocker in acute ischemic stroke was initiated in 1984. TUH was one of several study centers.
Over 2 years, 192 patients admitted to TUH with
diagnosed focal neurologic deficits consistent with
acute stroke were screened using the enrollment
criteria shown in Table 1. The screening of patients
was not continuous; there were interruptions during
the 2-year interval while enrollment criteria were
reviewed and revised to address the difficulty of
enrolling patients. The rationale for selecting the
various enrollment criteria have been reviewed
elsewhere.4
LaRueetal
Acute Stroke Therapy Trials
951
TABLE 1. Enrollment Criteria for Acute Stroke Multicenter Clinical Trial
Revision 1 (n = 78)
Revision 2 (n = 20)
Criterion
% excluded
Criterion
% excluded
12
12
18
>35
lschemic
MCA/ACA
3
17
26
>35
lschemic
MCA/ACA
0
0
15
>35
lschemic
MCA/ACA
0
0
20
24
9
11
23
8
14
>48hr
Unstable course
Decreased level of
consciousness
Lacune/TIA
Hemiplegia
Ipsilateral,
contralateral with
residual
Significant hepatic or
renal disease,
alcohol or drug
abuse
Unstable atrial
fibrillation, valvular
heart disease,
cardiac surgery <6
months, and other*
25
0
5
27
0
4
0
>48hr
Unstable course
Decreased level of
consciousness
Lacune/TIA
Hemiplegia
Ipsilateral,
contralateral with
residual
Significant hepatic or
renal disease,
alcohol or drug
abuse
*
4
Active peptic ulcer
5
Active peptic ulcer
0
Seizures
nonexclusionary
Severe movement
disorder
Not specified
0
Initial (n = 94)
Criterion
% excluded
Inclusion criteria
Age (yr)
40-80
Stroke type
lschemic
Territory
MCA
Exclusion criteria
Time of screening
>24 hr
Stroke
Unstable course
Decreased level of
consciousness
Lacune/TIA
Severity
Plegia of one limb
Previous stroke
None allowed
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
Cardiac disease
28
Congestive heart
failure, atrial
fibrillation, valvular
heart disease,
clinically significant
arrhythmias, cardiac
surgery, and other*
Gastrointestinal
disease
Neurologic
disease
Active peptic ulcer
4
>24hr
Unstable course
Decreased level of
consciousness
Lacune/TIA
Hemiplegia
Ipsilateral,
contralateral with
residual
Significant hepatic or
renal disease,
alcohol or drug
abuse
Decompensated
congestive heart
failure, atrial
fibrillation, valvular
heart disease,
clinically significant
arrhythmias, cardiac
surgery, and other*
Active peptic ulcer
Seizures
13
Seizures
14
Seizures
10
All other neurologic
diseases
Significant
9
21
All other neurologic
diseases
Significant
25
2
All other neurologic
diseases
Significant
t
30
t
15
Medical
conditions
Psychiatric
illness
Concomitant
medications
Laboratory
tests
Other
5
14
31
Hepatic or renal
34
disease, alcohol or
drug abuse
5
12
18
32
13
4
8
Clinically significant 19 Clinically significant
deviations from
deviations from
normal
normal
17
10
Untestable
Untestable
3 Intracranial surgery
Intracranial surgery
3
Questionable diagnosis 24 Questionable diagnosis 17
Pregnant or lactating
Nonneurologic deficit!
0 Pregnant or lactating
9 Nonneurologic deficit§
0
1
0
20
15
25
0
Final (n = 71)
Criterion % excluded
Nitroglycerine,
0
t
/3-blockers,
anticoagulants,
antiplatelet agents,
and hypnotics
allowed
Clinically significant
5 Clinically significant
deviations from
deviations from
normal
normal
Untestable
15
Untestable
0 Intracranial surgery
Intracranial surgery
Questionable
Questionable diagnosis 25
diagnosis
Pregnant or lactating 0 Pregnant or lactating
Nonneurologic
Nonneurologic deficit! 0
deficit!
1
11
14
18
13
0
4
0
0
0
0
0
0
MCA, middle cerebral artery; ACA, anterior cerebral artery; TIA, transient ischemic attack.
•Cardiac exclusions were myocardial infarction <4 months, acute or noncompensated heart failure, complete heart block, ventricular
arrhythmia, unstable angina, cardiac or carotid surgery s i month, blood pressure of > 190/110 or <90/60 mm Hg, or pulse of > 110/min
or <50/min.
tExclusionary concomitant medications were vasodilators, barbiturates, naloxone, /3-blockers, other calcium channel antagonists,
dextrans including mannitol, anticoagulants, antiplatelet agents, corticosteroids, neuroleptics, antidepressants, and hypnotics.
tDisallowed concomitant medications were vasodilators, naloxone, other calcium channel antagonists, dextrans including mannitol,
corticosteroids and barbiturates other than short-acting; other medication allowed; 325 mg aspirin/day required in nonanticougulated
patients.
§Examples include amputations and other defects making assessment impossible.
952
Stroke
Vol 19, No 8, August 1988
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
Screening for acceptability was done by a member of the Neurology Department of TUH within a
few hours after the patient was admitted to the
hospital. The patient's clinical history was obtained,
a neurologic examination was performed, the computed tomogram (CT scan) was reviewed, and laboratory studies were done. As the study progressed
and accrual remained nil, individual enrollment
criteria were modified, but often a given patient was
excluded by multiple criteria. Therefore, modification of one or several criteria had disappointingly
little effect on patient accrual.
The frequency with which sets of enrollment
criteria were exclusionary was noted. In the sets
containing two criteria, a x2 test with 1 degree of
freedom was used to determine which combinations
occurred frequently enough to be statistically significant. Expected frequencies were calculated by
assuming independence of the criteria and multiplying the proportion excluded by each criterion in the
pair by the total number of patients.
Results
Ninety-four patients were screened using the
initial enrollment criteria. Of these, all were excluded.
Twenty-three patients (24%) were excluded because
their stroke occurred >24 hours before our screening,
but in none was time of screening the sole reason for
disqualification. Evidence of a previous stroke by
history, CT, or other prior neurologic deficit excluded
29 (31%) of the 94 screened, but previous stroke was
the sole disqualifier in only four (4%). Medical conditions (e.g., blood pressure consistently >190 mm Hg
systolic or >110 mm Hg diastolic, a disease affecting
drug absorption or excretion) excluded 32 (34%);
however, 31 of these 32 patients were also excluded
on one or more additional criteria. Cardiac disease
(congestive heart failure, atrial fibrillation, valvular
heart disease, cardiac arrhythmias, or recent cardiac
surgery) disqualified 26 of the 94 patients (28%);
cardiac disease was the sole reason for exclusion of
only two patients (2%). Concomitant medications
excluded 28 (30%) of the 94 screened, but no patient
was disqualified solely on the basis of concomitant
medications. Among the 94 patients initially screened,
only 14 (15%) were disqualified on a single enrollment
criterion; thus, most patients had more than one
reason for exclusion. In 27 (29%) there were two
disqualifying factors; 19 (20%) had three and 11 (12%)
had four disqualifying factors. In 23 patients (24%)
there were five or more reasons for exclusion.
After 5 months of screening patients with no
enrollments, six criteria were modified slightly to
make them less restrictive (Table 1, Revision 1).
The lower age boundary was dropped from 40 to 35
years with no restriction placed on the upper boundary, the vascular territory affected by the stroke
was extended to include the anterior cerebral artery
(ACA), severity of the stroke was changed to exclude
those with hemiplegia, patients with evidence of a
previous contralateral stroke without residual defi-
TABI.E 2. Number of Patients Excluded by Given Number of
Enrollment Criteria in Initial, Revision 1, and Revision 2 Sets
Patients
Number of enrollment
criteria excluding
No.
%
0
1
2
3
4
5
6
7
8
9
1
35
64
35
28
18
3
4
3
1
0.5
18.3
33.3
18.3
14.6
9.4
1.6
2.1
1.6
0.5
Only one (0.5%) of 192 patients screened was eligible for
enrollment.
cits were included, of medical conditions only significant hepatic or renal disease were exclusionary,
and the criterion for presence of heart disease was
modified to allow enrollment of those with compensated congestive heart failure. Using the Revision 1
enrollment criteria, another 78 patients were
screened, but again, none were enrolled. As with
the initial enrollment criteria, multiple reasons for
exclusion disqualified the majority of potential
patients. The percent of these patients excluded by
each Revision 1 criterion is shown in Table 1. Thus,
none of the six modifications in enrollment criteria
of Revision 1 improved patient accrual. In retrospect, this might have been anticipated. For example, dropping the lower age boundary from 40 to 35
years would not be likely to improve patient recruitment since very few strokes occur in the 3540-year-old age range. However, removing the upper
limit did reduce the number of patients excluded by
this criterion. Broadening the vascular territory
affected to include the ACA territory did not permit
us to enroll more patients as none of the initial 94
patients screened had been excluded on the basis of
ACA territory strokes, and none of the 78 subsequent patients screened had ischemia confined to
the ACA territory.
Since no patients were enrolled with Revision 1
criteria, additional changes in four more criteria
were made (Table 1, Revision 2). Using Revision 2
criteria, another 20 patients were screened and one
was enrolled. The enrollment of one patient after
screening 192 with almost 2 years' work is obviously inefficient. Other centers participating in the
trial had similar difficulty in enrolling patients. Therefore, a careful reconsideration of the enrollment
criteria was clearly necessary.
In formulating reasonable rules for enrollment,
the criteria must be considered in combination.
Only 35 (18.3%) of the 192 patients in this study
were excluded on a single criterion; most were
excluded on multiple criteria (Table 2). Table 3
gives the percentages excluded for each criterion
LaRue et al
TABLE 3. Number of Patients Excluded by Each Enrollment
Criterion
Enrollment criterion
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
Medical conditions
Time of screening
Previous stroke
Other (questionable
diagnosis)
Territory
Concomitant medications
Cardiac disease
Neurologic disease (all
other)
Severe deficit
Neurologic disease
(seizure s/phenytoin)
Laboratory tests
Other (untestable)
Stroke type
Stroke (decreased level of
consciousness)
Stroke (unstable course)
Age
Stroke (lacune/transient
ischemic attack)
Other (nonneurologic
deficit)
Gastrointestinal disease
Psychiatric illness
Other (intracranial surgery)
All patients
%
No.
Patients
excluded by
1 criterion
No.
%
62
46
46
32
24
24
2
1
7
6
3
20
41
40
40
36
21
21
21
19
7
5
0
2
20
14
0
6
28
26
15
14
2
3
6
9
25
25
25
24
13
13
13
13
0
0
1
2
0
0
3
6
22
14
13
11
7
7
0
0
0
0
0
0
11
6
1
3
9
8
5
5
5
4
3
3
1
0
0
1
3
0
0
3
individually. Of the 210 possible pairs of the 21
enrollment criteria (calculated using the binomial
coefficient (V), 17 occurred frequently enough to
suggest possible associations. For example, exclusion for a medical condition was associated with
exclusion for decreased level of consciousness
(/? = 0.023) and significant deviation of laboratory
tests from normal values (p = 0.013). Decreased
level of consciousness was linked to an unstable
course or evolving stroke (p<0.001); unstable course
was linked to concomitant medications (p<0.02).
Nonneurologic deficit was associated with significant gastrointestinal disease (p<0.01). However, in
individually testing 210 possible combinations each
at a nominal significance level of 0.05, one would
anticipate up to 17 significant associations by chance
alone (95% confidence interval 4-17) even if no
significant association actually existed, if the combinations are independent; in fact, they are not
independent. Since only 17 significant associations
were found, we conclude that many combinations
of enrollment criteria probably were simply random
associations. Therefore, discarding or making a few
criteria less restrictive would be unlikely to result in
an appreciably increased enrollment.
Acute Stroke Therapy Trials
953
After screening 192 acute stroke patients, recruitment for the clinical trial was temporarily suspended and new enrollment criteria were formulated in an effort to obtain criteria that would permit
enrollment of enough patients to complete the trial
within a reasonable time. The final criteria listed in
Table 1 were formulated, and 71 more patients were
screened; of these, 16 (23%) were accepted. This
was clearly an improvement. Had these 16 patients
been screened according to the previous three sets
of criteria, only one would have been accepted
under either the initial or Revision 1 criteria; all 16
would have been accepted under Revision 2 criteria. Therefore, with modification, the enrollment
criteria more closely approximated the characteristics of ischemic stroke patients actually encountered in an acute-care hospital, and the efficiency of
patient accrual increased from near zero to an
acceptable level.
Discussion
Given the substantial loss of potential subjects
resulting from the failure to meet enrollment criteria, a clinical trials investigator might be tempted to
ask why any restrictions should be imposed on
patient recruitment; however, on reflection it is
clear that some restrictions are required. For example, one must match the expected action of the drug
being tested with the pathophysiology of stroke.
Since in this study calcium channel blockers are
postulated to affect only acute stroke events, enrollment of only acute stroke patients was appropriate.
However, the interval considered acute may vary.
Since it takes time for a patient to reach the hospital,
to undergo diagnostic tests, and to give informed
consent, the interval of acceptability should not be too
short. Almost one fourth of those we screened presented >24 hours after onset of the stroke. In the
National Survey of Stroke,5 35% of patients with
ischemic strokes were hospitalized >24 hours and
approximately 20% >48 hours after onset. On the
other hand, if the interval of acceptability is too long,
the beneficial effect of a new therapy may be obscured.
For example, in a study of theophylline treatment for
acute stroke, failure to observe a beneficial effect was
attributed to a delay in enrollment of up to 46 hours
after onset.67
Another factor that restricts enrollment is the
extent of the stroke. If the area of brain damaged is
very small (e.g., a lacune), the patient may recover
spontaneously. A recent hospital-based study found
that 7.5% of stroke admissions were lacunar.8 Large
lesions may produce coma, which precludes evaluation by standardized scales usually included in a
stroke clinical trial. Moreover, prognosis for survival in comatose patients is poor. When standard
efficacy measures such as the Toronto Stroke Scale9
are used, individuals with posterior circulation
strokes may be excluded because deficits such as
vertigo, ataxia, and nystagmus are difficult to quantify on those scales. In the naftidrofuryl study, 19%
954
Stroke
Vol 19, No 8, August 1988
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
of the patients screened were excluded because of
posterior circulation deficits, a percentage similar to
that we excluded for failure to meet the territory
inclusion criterion.
Ethical and safety considerations restrict recruitment, but few patients at our center were excluded
for these reasons alone. Cardiac disease and medical conditions, when severe, usually exclude a
patient. In the National Stroke Survey,5 35% of
patients with embolic stroke had valvular heart
disease, and over half of all patients with ischemic
strokes had chronic heart disease, atrial fibrillation,
and other arrhythmias. We excluded 19% of the
patients screened for cardiac disease under the
initial, Revision 1, and Revision 2 criteria, but when
the criterion was modified to exclude only those
with acute or unstable heart disease, nine patients
were excluded. Our experience compares with the
7% excluded for cardiac reasons in the Scandinavian Hemodilution Study.1
It is necessary to exclude patients who must
receive drugs the effects of which potentially confound those of the test drug, and in our study 21% of
those patients screened under the initial, Revision
1, and Revision 2 criteria were excluded because of
concomitant medications.
The above exclusions consider loss of patients
according to individual criteria, but an important
insight gained from our study was that modifying
any one or even many criteria would have a much
less salutary effect on recruitment of patients than
would be anticipated from noting the proportion
rejected when a specific criterion was considered
alone. Only a few modifications led to an appreciable decline in the percentage excluded by an individual criterion (Table 1). For example, removing
the upper age boundary decreased exclusions
because of that criterion, as did allowing previous
contralateral stroke without residual deficit. Allowing compensated congestive heart failure and
changes in disallowed concomitant medications
decreased exclusions due to those criteria. Other
changes, such as enlarging the vascular territory to
include the ACA and allowing patients with plegia
of one limb, did not appreciably change the percentage excluded by those criteria. Some criteria that
were not modified showed fluctuations in the percentage excluded and probably reflect fluctuation in
the characteristics of successive groups of patients
who were screened.
Most patients were disqualified on more than one
criterion so modification or elimination of many
criteria was needed to capture a significant number
of patients. However, elimination of exclusion criteria also reduces homogeneity of the study population, and compromises must be made. Strict enrollment criteria reduce the number of patients who can
be enrolled in any given center, but this problem
can be overcome by adding more centers. Of course,
costs and variability in the study data increase with
the addition of centers. Some indication of the costs
of including various enrollment criteria can be determined from our study, and our experience can serve
as a guideline in selecting criteria for future acute
stroke trials. The number of patients excluded by
various enrollment criteria should be calculated in
other medical centers to assure that our experience
was representative.
References
1. Scandinavian Stroke Study Group: Multicenter trial of hemodilution in ischemic stroke—Background and study protocol. Stroke 1985;16:885-890
2. Steiner TJ, Rose FC: Towards a model stroke trial: The
single-centre naftidrofuryl study. Neuroepidemiology 1986;
5:121-147
3. Duke RJ, Bloch RF, Turpie AGG, Trebilcock R, Bayer N:
Intravenous heparin for the prevention of stroke progression
in acute partial stable stroke: A randomized controlled trial.
Ann Intern Med 1986;105:825-828
4. Sterman AB, Furlan AJ, Pessin M, Kase C, Caplan L,
Williams G: Acute stroke therapy trials: An introduction to
recurring design issues. Stroke 1987;18:524-527
5. Walker AE, Robins M, Weinfeld FD: The National Survey
of Stroke: Clinical findings. Stroke 1981;12(suppl 1):I-131-44
6. Britton M, DeFaire U, Helmers C, Miah K: Lack of effect of
theophylline on the outcome of acute cerebral infarction.
Ada Neurol Scand 1980;62:116-123
7. Yatsu FM, Pettigrew LC, Grotta JC: Medical therapy of
ischemic strokes, in Barnett HJM, Mohr JP, Stein BM,
Yatsu FM (eds): Stroke: Pathophysiology, Diagnosis, and
Management. New York, Churchill Livingstone Inc, 1986,
vol 2, pp 1069-1083
8. LaRue L, Alter M, Lai SM, Friday G, Sobel E, Levitt L,
McCoy R, Isack T: Acute stroke, hematocrit, and blood
pressure. Stroke 1987;18:565-569
9. Norris JW: Comments on "Study Design of Stroke Treatments." Stroke 1982; 13:527-528
KEY WORDS • cerebrovascular disorders • clinical trials •
patient identification systems
Acute stroke therapy trials: problems in patient accrual.
L J LaRue, M Alter, N D Traven, A B Sterman, E Sobel and J Kleiner
Stroke. 1988;19:950-954
doi: 10.1161/01.STR.19.8.950
Downloaded from http://stroke.ahajournals.org/ by guest on July 28, 2017
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1988 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/19/8/950
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click Request
Permissions in the middle column of the Web page under Services. Further information about this process is
available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/