Randomized controlled trials of aprotinin in cardiac surgery: Could clinical equipoise have stopped the bleeding?

Article (PDF Available)inClinical Trials 2(3):218-29; discussion 229-32 · February 2005with78 Reads
DOI: 10.1191/1740774505cn085oa · Source: PubMed
Abstract
Aprotinin is a serine protease inhibitor used to limit perioperative bleeding and reduce the need for donated blood transfusions during cardiac surgery. Randomized controlled trials of aprotinin evaluating its effect on the outcome of perioperative transfusion have been published since 1987, and systematic reviews were conducted in 1992 and 1997. A systematic search was conducted for all RCTs of aprotinin that used placebo controls or were open-label with no active control treatment. Data collected included the primary outcome, objective of each study, whether a systematic review was cited or conducted as part of the background and/or rationale for the study and the number of previously published RCTs cited. Cumulative meta-analyses were performed. Sixty-four randomized, controlled trials of aprotinin were found, conducted between 1987 and 2002, reporting an endpoint of perioperative transfusion. Median trial size was 64 subjects, with a range of 20 to 1784. A cumulative meta-analysis indicated that aprotinin greatly decreased the need for perioperative transfusion, stabilizing at an odds ratio of 0.25 (p < 10 - 6) by the 12th study, published in June of 1992. The upper limit of the confidence interval never exceeded 0.65 and results were similar in all subgroups. Citation of previous RCTs was extremely low, with a median of 20% of prior trials cited. Only 7 of 44 (15%) of subsequent reports referenced the largest trial (N = 1784), which was 28 times larger than the median trial size. This study demonstrates that investigators evaluating aprotinin were not adequately citing previous research, resulting in a large number of RCTs being conducted to address efficacy questions that prior trials had already definitively answered. Institutional review boards and journals could reduce the number of redundant trials by requiring investigators to conduct adequate searches for prior evidence and conducting systematic reviews.
Randomized controlled trials of aprotinin in
cardiac surgery: could clinical equipoise have
stopped the bleeding?
Dean Fergusson
a,b
, Kathleen Cranley Glass
b,c
, Brian Hutton
a
and Stan Shapiro
b,c,d
Background Aprotinin is a serine protease inhibitor used to limit perioperative
bleeding and reduce the need for donated blood transfusions during cardiac surgery.
Randomized controlled trials of aprotinin evaluating its effect on the outcome of
perioperative transfusion have been published since 1987, and systematic reviews
were conducted in 1992 and 1997.
Methods A systematic search was conducted for all RCTs of aprotinin that used
placebo controls or were open-label with no active control treatment. Data collected
included the primary outcome, objective of each study, whether a systematic review
was cited or conducted as part of the background and/or rationale for the study and
the number of previously published RCTs cited. Cumulative meta-analyses were
performed.
Results Sixty-four randomized, controlled trials of aprotinin were found, conducted
between 1987 and 2002, reporting an endpoint of perioperative transfusion.
Median trial size was 64 subjects, with a range of 20 to 1784. A cumulative meta-
analysis indicated that aprotinin greatly decreased the need for perioperative
transfusion, stabilizing at an odds ratio of 0.25 (p , 10 2 6) by the 12th study,
published in June of 1992. The upper limit of the confidence interval never exceeded
0.65 and results were similar in all subgroups. Citation of previous RCTs was
extremely low, with a median of 20% of prior trials cited. Only 7 of 44 (15%) of
subsequent reports referenced the largest trial (N ¼ 1784), which was 28 times
larger than the median trial size.
Conclusions This study demonstrates that investigators evaluating aprotinin were
not adequately citing previous research, resulting in a large number of RCTs being
conducted to address efficacy questions that prior trials had already definitively
answered. Institutional review boards and journals could reduce the number of
redundant trials by requiring investigators to conduct adequate searches for prior
evidence and conducting systematic reviews. Clinical Trials 2005; 2: 218232.
www.SCTjournal.com
Introduction
The ostensible purpose of randomized controlled
trials (RCTs) is to answer an unsettled question
about an intervention’s efficacy. This paper uses the
example of RCTs of aprotinin treatment to address
the issue of how much experimentation is enough,
and if better procedures are needed to prevent RCTs
a
Ottawa Health Research Institute, Clinical Epidemiology Program, Ottawa, Ontario, Canada,
b
Clinical Trials Research
Group, McGill University, Montreal, Quebec, Canada,
c
Departments of Human Genetics and Pediatrics and Biomedical
Ethics Unit, McGill University, Montreal, Quebec, Canada,
d
Department of Epidemiology and Biostatistics, McGill
University, Montreal, Quebec, Canada
Author for correspondence: Dean Fergusson, Clinical Epidemiology Program, Ottawa Health Research Institute,
c/o Ottawa Hospital, 501 Smyth Road, Box 201, Ottawa, Ontario KIH 8L6, Canada. E-mail: dafergusson@ohri.ca
ETHICS Clinical Trials 2005; 2: 218232
# Society for Clinical Trials 2005 10.1191/1740774505cn085oa
from being initiated or conducted when the
question they are addressing has already been
answered.
Aprotinin is a serine protease inhibitor used to
limit perioperative bleeding and thus reduce the
need for allogeneic (donated) red blood cell transfu-
sions [1]. A meta-analysis of 16 trials of aprotinin in
cardiac surgery published in 1994 [2] concluded that
it was highly effective in reducing the proportion of
patients requiring a transfusion (odds ratio 0.23;
95% confidence interval, 0.16 to 0.33). Another
meta-analysis was published in 1997 by one of the
authors (DF) [3]. This review of 45 randomized
clinical trials further affirmed the effectiveness of
aprotinin (odds ratio 0.31; 95% confidence interval,
0.25 to 0.39). While conducting the meta-analysis
it became obvious that randomized controlled
trials (RCTs) had continued to be proposed, funded,
conducted, and published well after the effective-
ness of aprotinin had been established.
An RCT is permissible when there is no scientific
consensus about the relative efficacy of two
competing interventions. Freedman referred to
this state of uncertainty in the expert community
as clinical equipoise [4]. Clinical equipoise provides
a justification for randomising individuals to
competing therapies or to placebo. Clinical equi-
poise must be based on awareness of the existing
medicalevidence,whichinturnrequiresthe
organized study of that evidence. This should be
presented to colleagues, Institutional Review Boards
(IRBs), funding and regulatory agencies, journal
editors and peer reviewers, and prospective partici-
pants as part of the justification for a given RCT. In
light of the obligation to justify a claim of clinical
equipoise between treatment options before pro-
ceeding with a trial, one of the first questions to be
asked is whether trialists systematically reviewed the
prior literature.
Methods
An unrestricted Medline and EMBASE literature
search was conducted for the dates January 1966 to
March 1997 with the text word aprotinin to identify
randomized controlled trials in cardiac surgery. An
updated systematic literature search with an RCT
filter [5] was conducted to identify all cardiac
surgery RCTs indexed in Medline after 1996. Only
randomized trials that described the proportion of
patients receiving at least one unit of allogeneic red
blood cells were eligible. Studies were included
regardless of whether they were full publications,
abstracts or letters to the editor; or were published in
a language other than English. All RCTs had to be
either placebo controlled or open-label with no
active control. Duplicate publications, publications
without data on the proportion of patients trans-
fused, and single-centre publications that were
part of a multi-centre publication were excluded.
Pseudo-randomized trials (e.g., randomized by
birthdate), noncontrolled trials, review articles,
and observational studies were excluded. RCTs
with an active comparator and no open-label or
placebo control were excluded.
Two reviewers, independently, assessed each
citation for eligibility and a total of 62 publications
representing 64 trials met full inclusion criteria
[667]. In addition to the 45 trials from a
previous meta-analysis [3,645,47,48,50,51], the
updated literature search identified 19 further
trials published between 1996 and 2004
[46,49,5267].
Data collected from each trial included the
objective, patient characteristics, whether a sys-
tematic review was conducted as part of the
background and/or rationale for the study, number
of subjects, publication date, dates of study enrol-
ment, and all previously published randomized
trials and systematic reviews cited. In addition, trial
quality was assessed using a published, validated
quality scale [68]. If the study stated more than one
objective without stating which one was the
primary objective, all were considered as primary.
A cumulative meta-analysis that produces an
updated measure of effect by statistically pooling
studies after each new study is completed was
performed with all 64 trials. Due to the lack of
reporting of dates of randomization, we used the
publication date as the study completion date.
To elucidate possible reasons for the continued
use of placebo or open-label control arms, subgroup
cumulative analyses were performed stratified by
methodological quality and patient characteristics.
To evaluate the effect of trial quality upon the results
of these meta-analyses, the quality scale was used
[68] along with a subgroup analysis of open-label
versus placebo controls. The trial quality scale
assesses quality based on randomization, blinding,
and the description of withdrawals. The highest
possible score is five; the lowest is zero. A score equal
to or greater than three was considered good and
less than three was considered poor quality. This
judgment is consistent with the original publication
[68]. Each trial was evaluated independently by two
individuals, with differences resolved by either
consensus or independent evaluation of a third
party. Outcome data from each trial (proportion of
subjects requiring at least one unit of allogeneic red
blood cell transfusion) was analysed using software
(Meta-Analyst
.977
, J Lau and T Chalmers) with a
random-effects model. Effect sizes are presented as
odds ratios (OR) with 95% confidence intervals. An
OR of 1 suggests no difference between intervention
and control; an OR , 1 suggests that fewer subjects
Randomized controlled trials of aprotinin in cardiac surgery 219
www.SCTjournal.com Clinical Trials 2005; 2: 218232
in the intervention group received at least one
allogeneic red blood cell transfusion while an
OR . 1 suggests the reverse.
Results
Figure 1 presents the date of publication, sample size
and odds ratio for each of the 64 trials. Overall, there
were 8040 subjects entered in the 64 trials. Median
trial size was 64 subjects (range 201784) with 46 of
64 (72%) enrolling fewer than 100 patients. The
largest trial consisted of 1784 patients and was
published in 1992. Figure 2 illustrates the published
randomized trials cited in each of the 64 trials. Dates
of study enrolment were reported in 15 of the
64 trials.
The cumulative meta-analysis presented in
Figure 3 indicates that a clinically significant result
was achieved in the very first trial of 22 patients (OR
0.03, 95% CI: 0.000.56). After the 12th study, the
cumulative effect estimate stabilizes in the range of
0.250.35. Variability around this estimate narrows
as the cumulative sample size increases. Throughout
the cumulative meta-analysis, the upper limit of the
confidence interval never crosses 0.65.
Results of the subgroup analysis for good quality
and placebo-controlled trials are presented in
Figure 4a and b. By including only the studies
assessed as good quality, a highly clinically and
statistically significant association is evident by
1990 after the third such trial (OR 0.09, 95% CI
0.020.54). A further 31 good quality randomized
controlled clinical trials were published between
1990 and 2001. Results are similar for sub group
analyses of placebo-controlled studies.
The objectives of each of the trials were examined
to identify possible justifications for conducting
further trials. All 64 publications stated an objective
with blood loss or transfusion requirements men-
tioned as a primary objective or outcome in 49
(77%) of the trials. Of the 64 trials, 53 stated
whether the patient population included primary
(38 trials), repeat (five trials), or a combination of
primary and repeat surgery (10 trials) cases. Separate
cumulative meta-analyses indicate that effective-
ness in each of the three surgical groups was
established in the early 1990s (Figure 5). Of the 64
trials, 51 provided information on aspirin use
(Figure 5). For the 24 trials that enrolled a
proportion of patients who were taking aspirin at
the time of surgery, the upper limit of the
confidence interval did not cross 0.62 after the first
trial. For trials enrolling patients exclusively taking
aspirin at the time of surgery, the cumulative effect
size became nonsignificant after the fourth of five
trials. After the fifth trial, the overall effect was an
odds ratio of 0.39 (95%CI: 0.170.89).
Overall, assuming a one year lag in publication,
the median number of prior trials cited was four and
the median percentage of prior cited trials per
publication was 20%. For published trials 110,
1140, and 4164, the respective median percen-
tage of cited trials were 33%, 31% and 10%.
Figure 6 illustrates the cumulative total number
of trials, assuming a one-year lag in publication, and
the corresponding number of published studies
cited in each publication. The largest trial, pub-
lished in 1992, was referenced by only seven of the
subsequent 44 trials published more than one year
later (Figure 2), even though it was almost 20 times
larger than the median trial size. The measure of
effect and confidence intervals for this trial were
almost identical to the pooled estimate of the 45
trials (OR 0.32, 95%CI: 0.260.40 versus OR 0.34,
95%CI: 0.290.41).
Discussion
To be ethical, clinical research must be valuable [69].
To be of value, a trial must add to current
knowledge. An integral step in evaluating the
evidence is to conduct a systematic review of the
literature. A systematic review or meta-analysis
refers to an overview of the literature conducted in
a well-defined, systematic and thorough manner, be
it quantitative or qualitative. Beyond justifying a
research question, a systematic review serves three
purposes: 1) trialists become aware of the full extent
of the literature related to their research question;
2) it provides a transparent, traceable path of due
diligence for research review boards, journal editors,
readership and, ultimately, prospective patients;
and 3) it provides trialists a comprehensive list of
clinical, design, and statistical issues that may be
relevant for the design or interpretation of their
proposed trial.
A multi-country survey conducted between 1995
and 1997 found that drugs to minimize peri-
operative bleeding and transfusion requirements
were used in a large proportion of hospitals [70]. Of
these drugs, aprotinin was used in the greatest
proportion of hospitals. While it is arguable whether
or not aprotinin was or should be a standard of care
at cardiac surgery centres due to issues of costs,
safety and the availability of other agents, the
effectiveness of aprotinin at reducing transfusion
requirements and blood loss, the ostensible focus of
the vast majority of RCTs of aprotinin, has been well
established since the mid-1990s [2,3]. More impor-
tantly, the use of antifibrinolytic pharmacological
agents in cardiac surgery, especially when substan-
tial blood loss or transfusion requirement is
expected, was also well established.
220 D Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
Figure 1 Proportion of patients transfused.
Randomized controlled trials of aprotinin in cardiac surgery 221
www.SCTjournal.com Clinical Trials 2005; 2: 218232
Our study illustrates that trialists evaluating
aprotinin were not adequately citing previous trials
nor conducting systematic reviews to support the
need for additional trials. It seems reasonable to
conclude that there remains a barrier between
published evidence-based medicine (clinical trials
and systematic reviews) and the review of this
evidence in considering whether further trials are
warranted. None of the 64 aprotinin trials reported
that a systematic review or meta-analysis had been
conducted and only two (3%) trials referenced
published systematic reviews. Instead, selective
early trials, especially the initial trial, were cited
to support the objective of the study. Thirty-six of
63 subsequent trials (57%) referenced the first
published trial (Figure 2). Overall, assuming a one-
year lag in publication, the median number of prior
trials cited was four and the median percentage of
prior cited trials per publication was 20%. For
published trials 110, 1140 and 4164, the
respective median percentage of cited trials were
33%, 31% and 10%.
Figure 6 demonstrates a profound and troubling
gap in available versus cited publications. The
largest trial (by an order of magnitude) was not
cited by 37 of 44 trials published more than a year
later. All 62 publications representing 64 trials were
easily identifiable through Internet literature search
portals and all trials, save one, were indexed on
Medline within weeks of publication. A simple
literature search using the National Library of
Medicine’s universally accessible and free of charge
PubMed world wide web portal with the term
“aprotinin” restricted to the publication type
“randomized controlled trials” identified 58 of the
62 publications (94%) and a PubMed search with the
terms “aprotinin” and “random
identified 51
(82%) publications. Moreover, the vast majority of
articles were published in top-tier cardiothoracic
specialty journals.
Clarke and Gotzsche have addressed the import-
ant issue of citing previous research [71,72]. They
concluded that researchers do not satisfactorily
address nor do journals adequately reflect the
Figure 2 Citations of previous studies.
222 D Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
Figure 3 Cumulative meta-analysis of all RCTs.
Randomized controlled trials of aprotinin in cardiac surgery 223
www.SCTjournal.com Clinical Trials 2005; 2: 218232
totality of prior relevant evidence. Gotzsche notes
that identifying previous studies by simply scanning
bibliographies of a convenient sample of publi-
cations can produce a biased sample of articles [72].
This is referred to as reference bias. The patterns in
Figure 2 provide a strong suggestion of this: trials
referenced in one paper were often picked up in
subsequent papers, but those that were missed early
remained largely uncited. This is evidence that
researchers were not conducting their own inde-
pendent reviews of the literature, but were instead
depending on previous incomplete searches by
others.
Unfortunately, too few trial publications (23%)
provided dates of patient accrual that would have
given a more accurate depiction of publications
available to investigators at time of study com-
mencement. Of those reporting accrual dates, the
vast majority accrued over a one-year period with
the trial publication one year after patient enrol-
ment ended. Even accounting for substantial lag
time, Figures 2 and 6 clearly demonstrates that
available evidence was not being evaluated in a
systematic manner prior to the start of these trials.
This makes it inevitable that trials were conducted
that did not need to be initiated.
Our present study focused on identifying all
randomized trials that reported the proportion of
patients transfused, as this outcome reflects the
1993 Food and Drug Administration’s approval
indication. With respect to their objectives and
outcome measures, the trials were quite hom-
ogenous. Of the 64 trials, 15 provided a primary
outcome or objective other than blood transfusion
or blood loss. Nine trials evaluated graft patency or
myocardial infarction as a primary objective, but no
trial had the primary objective of assessing allergic
reactions, mortality, or other serious thrombotic
events. As for homogeneity with respect to patient
populations, separate cumulative analyses were
carried out for primary surgery patients only, repeat
surgery patients only, combination of primary and
repeat surgery patients, patients all on aspirin, some
patients on aspirin, no patients on aspirin, and a
Figure 4 Cumulative meta-analysis of placebo-controlled trials and cumulative meta-analysis of good quality trials.
224 D Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
range of aprotinin doses from ultra low to high.
The effectiveness of aprotinin was established early
and remained consistent in all the above categories
of patients except for trials conducted exclusively
in patients taking aspirin at the time of surgery.
Given the estimates provided in Figure 1, it is not
surprising that the effectiveness of aprotinin
remains consistent across different clinical
subgroups.
The 64 RCTs examined here were not the only
studies conducted and published on this topic.
Numerous randomized trials not examining the
outcomes studied here were conducted, as were
numerous pseudo- or nonrandomized trials,
active-controlled and uncontrolled trials of aproti-
nin. Indeed, between 1997 and 2004, 61 random-
ized controlled trials of aprotinin in cardiac surgery
were identified in our systematic literature search,
only 19 of which met eligibility for the present
study.
Despite the substantial efficacy evidence that has
accrued on this treatment, trials of aprotinin in
cardiac surgery continue. One could argue that not
all relevant outcomes have been sufficiently studied,
such as serious adverse events (e.g., thrombotic
events, mortality, myocardial infraction). However,
to examine these endpoints trials of much greater
sample size would have to be conducted and trialists
would have had to provide evidence that the risk of
harm from administering aprotinin outweighed the
risks of not receiving aprotinin in the placebo group.
Other unsettled questions could be that certain
patient subsets may respond differently to different
administration schedules and/or doses of aprotinin.
But those questions do not require the use of a
placebo or open-label arm. If dosing is the primary
objective the control arm should receive the already
established effective dose(s) as outlined in the
product monograph.
Another possible motivation for conducting a
trial is to gain local experience with an intervention
before it is adopted. Local factors such as surgical
expertise, transfusion thresholds, and various
co-interventions can alter the effectiveness of
Figure 5 Subgroup cumulative meta-analysis.
Randomized controlled trials of aprotinin in cardiac surgery 225
www.SCTjournal.com Clinical Trials 2005; 2: 218232
aprotinin in a particular setting. However, employ-
ing a randomized controlled trial for the purpose of
gaining local experience is less than ideal from an
ethical or pragmatic standpoint.
IRBs are established to ensure that research
involving humans meets the established ethical
requirements, which are summarized by Emanuel:
value, scientific validity, fair subject selection,
favourable risk-benefit ratio, independent review,
informed consent and respect for potential and
enrolled subjects [73]. To this end, IRBs must
critically evaluate the background and rationales
provided by the investigators. Some have questioned
whether IRBs have duly fulfilled their role, citing
recent trials of established effective treatments using
placebo control. Others have suggested that the
performance and accountability of IRBs would be
improved by requiring trialists to submit systematic
reviews in support of their application [74]. This
requirement affirms the responsibility of both IRB
and investigator to adequately consider the extant
evidence base when assessing the need for a study,
and it allows the IRB to consider the evidence in an
unbiased and transparent manner. The results of our
study show the potential consequences of not
having such a requirement.
As the single largest disseminator of research
results, journals have a responsibility for ensuring
they publish only scientifically and ethically valid
and valuable research. Assessing whether clinical
equipoise was present at the start of the trial must be
part of the editorial calculus. One suggestion would
be that the CONSORT statement [75] be amended
to require authors to explicitly state whether a
systematic review was conducted to support a state
of clinical equipoise. The systematic review can
either be an original undertaking or an update of a
previous systematic review.
Over the past century and a half we have made
considerable progress in transforming the art of
medicine into the science of medicine [76].
However, systematizing the review of available
evidence remains an area where we are moving
more slowly than desirable. How do we prevent
unnecessary trials? Requiring investigators to con-
duct more organized reviews of the evidence using
Figure 6 Citations of prior publications.
226 D Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
well-established standards and scientific methods
of systematic reviewing, with the diligence of this
effort assessed by IRBs and journals, would be a
constructive step in the right direction.
Acknowledgements
The authors thank Dr Paul He
´
bert and Dr Leon Glass
for the helpful comments and suggestions in the
preparation of this manuscript.
References
1. Smith PK, Shah AS. The role of aprotinin in a blood-
conservation program. J Cardiothorac Vasc Anesth 2004;
18(4 Suppl.): S2428.
2. Fremes SE, Wong BI, Lee E et al. Metaanalysis of
prophylactic drug treatment in the prevention of
postoperative bleeding. Ann Thorac Surg 1994; 58:
158088.
3. Laupacis A, Fergusson D. Drugs to minimize perio-
perative blood loss in cardiac surgery: meta-analyses
using perioperative blood transfusion as the outcome.
The International Studyof Peri-operative Transfusion
(ISPOT) Investigators. Anesth Analg 1997; 85: 125867.
4. Freedman B. Equipoise and the ethics of clinical
research. N Engl J Med 1987; 317: 141 45.
5. Dickersin K, Scherer R, Lefebvre C. Identifying
relevant studies for systematic reviews. BMJ 1994; 309:
128691.
6. Royston D, Bidstrup BP, Taylor KM, Sapsford RN.
Effect of aprotinin on need for blood transfusion after
repeat open-heart surgery. Lancet 1987; 2: 128991.
7. Bidstrup BP, Royston D, Sapsford RN, Taylor KM.
Reduction in blood loss and blood use after cardiopul-
monary bypass with high dose aprotinin (trasylol).
J Thorac Cardiovasc Surg 1989; 97: 36472.
8. Fraedrich G, Weber C, Bernard C, Hettwer A,
Schlosser V. Reduction of blood transfusion require-
ment in open heart surgery by administration of high
doses of aprotinin preliminary results. Thorac Cardio-
vasc Surg 1989; 37: 8991.
9. Bidstrup BP, Royston D, McGuiness C,
Sapsford RN. Aprotinin in aspirin-pretreated patients.
Perfusion 1990; 5(Suppl.): 7781.
10. Locatelli A, Bertollo D, Bianchi T et al. Aprotinin in
cardiosurgery: a randomized prospective study with
different protocols for use. Miner Anestesiol 1990; 56:
97375.
11. Dietrich W, Spannagl M, Jochum M et al. Influence
of high-dose aprotinin treatment on blood loss and
coagulation patterns in patients undergoing myo-
cardial revascularization. Anesthesiology 1990; 73:
111926.
12. Harder MP, Eijsman L, Roozendaal KJ, v an
Oeveren W, Wildevuur CR. Aprotinin reduces
intraoperative and postoperative blood loss in mem-
brane oxygenator cardiopulmonary bypass. Ann Thorac
Surg 1991; 51: 93641.
13. Boldt J, Zickmann B, Czeke A et al. Blood conserva-
tion techniques and platelet function in cardiac surgery.
Anesthesiology 1991; 75: 42632.
14. Deleuze P, Loisance DY, Feliz A et al. Reduction of
per- and postoperative blood loss with aprotinin
(Trasylol) during extracorporeal circulation. Arch Mal
Coeur Vaiss 1991; 84: 1797 802.
15. Gherli T, Porcu A, Padua G et al. Reducing bleeding
during extracorporeal circulation interventions by
high doses of aprotinin. Minerva Cardioangiol 1992; 40:
12126.
16. Baele PL, Ruiz-Gomez J, Londot C et al. Systematic
use of aprotinin in cardiac surgery: influence on
total homologous exposure and hospital cost. Acta
Anaesthesiol Belg 1992; 43: 10312.
17. Dietrich W, Barankay A, Hahnel C, Richter JA.
High-dose aprotinin in cardiac surgery: three years’
experience in 1,784 patients. J Cardiothorac Vasc Anesth
1992; 6: 32427.
18. Vedrinne C, Girard C, Jegaden O et al. Reduction in
blood loss and blood use after cardiopulmonary bypass
with high-dose aprotinin versus autologous fresh whole
blood transfusion. J Cardiothorac Vasc Anesth 1992; 6:
31923.
19. Mohr R, Goor DA, Lusky A, Lavee J. Aprotinin
prevents cardiopulmonary bypass-induced platelet
dysfunction. A scanning electron microscope study.
Circulation 1992; 86(5 Suppl.): II405II409.
20. Cosgrove DM, Heric B, Lytle BW et al. Aprotinin
therapy for reoperative myocardial revascularization: a
placebo-controlled study. Ann Thorac Surg 1992; 54:
103138.
21. BidstrupBP,UnderwoodSR,SapsfordRN,
Streets EM. Effect of aprotinin (Trasylol) on aorta-
coronary bypass graft patency. J Thorac Cardiovasc Surg
1993; 105: 14752.
22. Hardy JF, Desroches J, Belisle S et al. Low-dose
aprotinin infusion is not clinically useful to reduce
bleeding and transfusion of homologous blood products
in high- risk cardiac surgical patients. Can J Anesth 1993;
40: 62531.
23. Isetta C, Gunness TK, Samat C et al. Antifibrinolytic
treatment and homologous transfusion in cardiac
surgery. Eur Heart J 1993; 14: 424.
24. Liu B, Belboul A, Radberg G et al. Effect of reduced
aprotinin dosage on blood loss and use of blood products
in patients undergoing cardiopulmonary bypass.
Scand J Thorac Cardiovasc Surg 1993; 27: 14955.
25. Carrera A, Martinez MV, Garcia-Guiral M et al. Use
of high doses of aprotinin in cardiac surgery. Rev Esp
Anestesiol Reanim 1994; 41: 13 19.
26. Lemmer JHJ, Stanford W, Bonney SL et al. Aproti-
nin for coronary bypass operations: efficacy, safety, and
influence on early saphenous vein graft patency. A
multicenter, randomized, double-blind, placebo-
controlled study. J Thorac Cardiovasc Surg 1994; 107:
54353.
27. Murkin JM, Lux J, Shannon NA et al. Aprotinin
significantly decreases bleeding and transfusion require-
ments in patients maintained on aspirin undergoing
cardiac surgery. J Thorac Cardiovasc Surg 1994; 107:
55461.
28. Bailey CR, Wielogorski AK. Randomized placebo
controlled double blind study of two low dose
aprotinin regimens in cardiac surgery. Br Heart J 1994;
71: 34953.
29. Bailey CR, Kelleher AA, Wielogorski AK. Random-
ized placebo-controlled double-blind study of three
aprotinin regimens in primary cardiac surgery. Br J Surg
1994; 81: 96973.
30. Maccario M, Fumag alli C, Dean geli s R et al.
Comparison between low and high doses of aprotinin
in heart surgery. Minerva Anestesiol 1994; 60: 315 20.
Randomized controlled trials of aprotinin in cardiac surgery 227
www.SCTjournal.com Clinical Trials 2005; 2: 218232
31. Rocha E, Hidalgo F, Llorens R et al. Randomized
study of aprotinin and DDAVP to reduce postoperative
bleeding after cardiopulmonary bypass surgery. Circula-
tion 1994; 90: 92127.
32. Swart MJ, Gordon PC, Hayse-Gregson PB et al.
High-dose aprotinin in cardiac surgerya prospective,
randomized study. Anaesth Intensive Care 1994; 22:
52933.
33. Tabuchi N, Huet RC, Sturk A, Eijsman L, Wild-
evuur CR. Aprotinin preserves hemostasis in aspirin-
treated patients undergoing cardiopulmonary bypass.
Ann Thorac Surg 1994; 58: 103639.
34. Blauhut B, Harringer W, Bettelheim P et al.
Comparison of the effects of aprotinin and tranexamic
acid on blood loss and related variables after cardio-
pulmonary bypass. J Thorac Cardiovasc Surg 1994; 108:
108391.
35. Kalangos A, Tayyareci G, Pretre R, Di Dio P,
Sezerman O. Inuenceofaprotininonearly
graft thrombosis in patients undergoing myocardial
revascularization. Eur J Cardiothorac Surg 1994; 8:
65166.
36. Alvarez JM, Quiney NF, McMillan D et al. The use
of ultra-low-dose aprotinin to reduce blood loss
in cardiac surgery. J Cardiothorac Vasc Anesth 1995; 9:
2933.
37. Corbeau JJ, Monrigal JP, Jacob JP et al. Comparaison
des effets de l’aprotinine et de l’acide tranexamique sur le
saignement en chirurgie cardiaque. Ann Fr Anesth Re
´
anim
1995; 14: 15461.
38. Speekenbrink RGH, Vonk ABA, Wildevuur CRH,
Eijsman L. Hemostatic efficacy of dipyridamole,
tranexamic acid, and aprotinin in coronary bypass
grafting. Ann Thorac Surg 1995; 59: 438 42.
39. Lass M, Welz A, Kochs M et al. Aprotinin in elective
primary bypass surgery. Graft patency and clinical
efficacy. Eur J Cardiothorac Surg 1995; 9: 206 10.
40. Pugh SC, Wielogorski AK. A comparison of the effects
of tranexamic acid and low-dose aprotinin on blood loss
and homologous blood usage in patients undergoing
cardiac surgery. J Cardiothorac Vasc Anesth 1995; 9:
24044.
41. Wendel HP, Heller W, Michel J et al. Lower
cardiac troponin T levels in patients undergoing
cardiopulmonary bypass and receiving high-dose
aprotinin therapy indicate reduction of perioperative
myocardial damage. J Thorac Cardiovasc Surg 1995; 109:
116472.
42. Penta de Peppo A, Pierri MD, Scafuri A et al.
Intraoperative antifibrinolysis and blood-saving
techniques in Cardiac Surgery Prospective trial of
3 antifibrinolytic drugs. Tex Heart Inst J 1995; 22:
23136.
43. Casas JI, Zuazu-Jausoro I, Mateo J et al. Aprotinin
versus desmopressin for patients undergoing operations
with cardiopulmonary bypass. A double-blind placebo-
controlled study. J Thorac Cardiovasc Surg 1995; 110:
110717.
44. Dietrich W, Dilthey G, Spannagl M et al. Influence
of high-dose aprotinin on anticoagulation, heparin
requirement, and celite- and kaolin-activated clotting
time in heparin-pretreated patients undergoing open-
heart surgery. A double-blind, placebo-controlled study.
Anesthesiology 1995; 83: 67989.
45. Levy JH, Pifarre R, Schaff HV. A multicenter,
double-blind, placebo-controlled trial of aprotinin for
reducing blood loss and the requirement for donor-
blood transfusion in patients undergoing repeat
coronary artery bypass grafting. Circulation 1995; 92:
223644.
46. Cicek S, Demirkilic U, Kuralay E, Ozal E, Tatar H.
Postoperative aprotinin: effect on blood loss and
transfusion requirements in cardiac operations. Ann
Thorac Surg 1996; 61: 137276.
47. Rodrigus IE, Vermeyen KM, De Hert SG, Amsel BJ,
Walter PJ. Efficacy and safety of aprotinin in
aortocoronary bypass and valve replacement operations:
a placebo-controlled randomized double-blind study.
Perfusion 1996; 11: 31318.
48. Menichetti A, Tritapepe L, Ruvolo G et al. Changes
in coagulation patterns, blood loss and blood use after
cardiopulmonary bypass: aprotinin vs tranexamic acid vs
epsilon aminocaproic acid. J Cardiovasc Surg 1996; 37:
401407.
49. Speekenbrink RG, Wildevuur CR, Sturk A,
Eijsman L. Low-dose and high-dose aprotinin improve
hemostasis in coronary operations. J Thorac Cardiovasc
Surg 1996; 112: 523 30.
50. D’Ambra MN, Akins CW, Blackstone EH.
Aprotinin in primary valve replacement and recon-
struction: a multicenter, double-blind, placebo-
controlled trial. J Thorac Cardiovasc Surg 1996; 112:
108189.
51. Lemmer JH Jr, Dilling EW, Morton JR et al.
Aprotinin for primary coronary artery bypass grafting:
a multicenter trial of three dose regimens. Annals of
Thoracic Surgery 1996; 62: 1659 67.
52. Ashraf S, Tian Y, Cowan D et al. “Low-dose”
aprotinin modifies hemostasis but not proinflam-
matory cytokine release. Ann Thorac Surg 1997; 63:
6873.
53. Hardy JF, Belisle S, Couturier A, Robitaille D.
Randomized, placebo-controlled, double-blind study of
an ultra-low-dose aprotinin regimen in reoperative
and/or complex cardiac operations. J Card Surg 1997; 12:
1522.
54. Hayashida N, Isomura T, Sato T, Maruyama H,
Kosuga K, Aoyagi S. Effects of minimal-dose aprotinin
on coronary artery bypass grafting. J Thorac Cardiovasc
Surg 1997; 114: 261 69.
55. Ray MJ, Marsh NA. Aprotinin reduces blood loss after
cardiopulmonary bypass by direct inhibition of plasmin.
Thromb Haemost 1997; 78: 102126.
56. Rossi M, Storti S, Martinelli L et al. A pump-prime
aprotinin dose in cardiac surgery: appraisal of its effects
on the hemostatic system. J Cardiothorac Vasc Anesth
1997; 11: 83539.
57. Klein M, Keith PR, Dauben HP et al. Aprotinin
counterbalances an increased risk of peri-operative
hemorrhage in CABG patients pre-treated with Aspirin.
Eur J Cardiothorac Surg 1998; 14: 360 66.
58. Alderman EL, Levy JH, Rich JB et al. Analyses of
coronary graft patency after aprotinin use: results from
the International Multicenter Aprotinin Graft Patency
Experience (IMAGE) trial. J Thorac Cardiovasc Surg 1998;
116: 71630.
59. Basora M, Gomar C, Escolar G et al. Platelet function
during cardiac surgery and cardiopulmonary bypass with
low-dose aprotinin. J Cardiothorac Vasc Anesth 1999; 13:
38287.
60. Ray MJ, Brown KF, Burrows CA, O’Brien MF.
Economic evaluation of high-dose and low-dose aproti-
nin therapy during cardiopulmonary bypass. Ann Thorac
Surg 1999; 68: 94045.
61. Nuttall GA, Oliver WC, Ereth MH et al. Comparison
of blood-conservation strategies in cardiac surgery
patients at high risk for bleeding. Anesthesiology 2000;
92: 67482.
62. Santamaria A, Mateo J, Oliver A et al. The effect of
two different doses of aprotinin on hemostasis in
228 D Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
cardiopulmonary bypass surgery: similar transfusion
requirements and blood loss. Haematologica 2000; 85:
127784.
63. Tassani P, Augustin N, Barankay A, Braun SL,
ZaccariaF,RichterJA.High-dose aprotinin
modulates the balance between proinflammatory and
anti-inflammatory responses during coronary artery
bypass graft surgery. J Cardiothorac Vasc Anesth 2000;
14: 682 86.
64. Dignan RJ, Law DW, Seah PW et al. Ultra-low dose
aprotinin decreases transfusion requirements and is cost
effective in coronary operations. Ann Thorac Surg 2001;
71: 15863; discussion 16364.
65. Alvarez JM, Jackson LR, Chatwin C, Smolich JJ.
Low-dose postoperative aprotinin reduces mediastinal
drainage and blood product use in patients undergoing
primary coronary artery bypass grafting who are taking
aspirin: a prospective, randomized, double-blind, placebo-
controlled trial. J Thorac Cardiovasc Surg 2001; 122: 45763.
66. Greilich PE, Okada K, Latham P, Kumar RR,
Jessen ME. Aprotinin but not epsilon-aminocaproic
acid decreases interleukin-10 after cardiac surgery with
extracorporeal circulation: randomized, double-blind,
placebo-controlled study in patients receiving aprotinin
and epsilon-aminocaproic acid. Circulation 2001;
104(Suppl. 1): I26569.
67. Englberger L, Kipfer B, Berdat PA, Nydegger UE,
Carrel TP. Aprotinin in coronary operation with
cardiopulmonary bypass: does “low-dose” aprotinin
inhibit the inflammatory response? Ann Thorac Surg
2002; 73: 1897904.
68. Jadad AR, Moore RA, Carroll D et al. Assessing
the quality of reports of randomized clinical trials:
is blinding necessary? Control. Clin. Trials 1996; 17:112.
69. Freedman B. Scientific value and validity as ethical
requirements for research. IRB: A Review of Human
Subjects Research 1987; 9: 710.
70. Fergusson D, Blair A, Henry D et al. Technologies to
minimize blood transfusion in cardiac and orthopedic
surgery. Results of a practice variation survey in nine
countries. International Study of Peri-operative Transfu-
sion (ISPOT) Investigators. Int J Technol Assess Health Care
1999; 15: 71728.
71. Clarke M, Chalmers I. Discussion sections in reports
of controlled trials published in general medical
journals: islands in search of continents? JAMA 1998;
280: 28082.
72. Gotzsche PC. Reference bias in reports of drug trials. Br
Med J (Clin Res Ed) 1987; 295: 65466.
73. Emanuel EJ, Wendler D, Grady C. What makes
clinical research ethical? JAMA 2000; 283: 270111.
74. Savulescu J, Chalmers I, Blunt J. Are research ethics
committees behaving unethically? Some suggestions for
improving performance and accountability. BMJ 1996;
313: 139093.
75. Moher D, Schulz KF, Altman DG. The CONSORT
statement: revised recommendations for improving the
quality of reports of parallel-group randomised trials.
Lancet 2001; 357: 119194.
76. Matthews JR. Quantification and the quest for medical
uncertainty. Princeton, New Jersey: Princeton University
Press; 1995.
Discussion
Comment
The scandalous failure of science to cumulate evidence
scientifically
Iain Chalmers
a
The article by Dean Fergusson and his colleagues [1]
in this issue of the journal [p] is the most recent
evidence of an ongoing scandal in which research
funders, academia, researchers, research ethics
committees and scientific journals are all complicit.
New research should not be designed or
implemented without first assessing systematically
what is known from existing research [2,3]. The
failure to conduct that assessment represents a lack
of scientific self-discipline that results in an
inexcusable waste of public resources. In applied
fields like health care, failure to prepare scientifi-
cally defensible reviews of relevant animal and
human data results not only in wasted resources but
also in unnecessary suffering and premature death
[411].
Fergusson and his colleagues [1] have used the
technique of cumulative meta-analyses of random-
ized trials, pioneered by Tom Chalmers and his
colleagues more than a decade ago [4,5], to analyse
64 trials assessing the effect of aprotinin on
perioperative blood loss, as judged by the use of
blood transfusion They show, as they did eight
years ago [12], that placebo controlled trials of
aprotinin have continued to be done long after
strong evidence has accumulated showing that
the drug substantially reduces the use of blood
transfusion.
In addition to this litany of unnecessary, and
therefore unethical research, Fergusson and his
colleagues present an analysis of the extent to which
authors of the reports of the 64 trials cited relevant
earlier trials. Their shocking findings are summar-
ized in Figures 2 and 6: between 1987 and 2002 the
proportion of relevant previous reports cited in
successive reports fell from a high of 33% to only
10% among the most recent reports. Furthermore,
only seven of 44 subsequent reports referenced
the report of largest trial (which was 28 times larger
than the median trial size); and only two of the
reports referenced systematic reviews of these trials
published in 1994 and 1997.
a
Iain Chalmers is editor of The James Lind Library (www.jameslindlibrary.org), a web-based resource containing material
about the evolution of fair tests of medical treatments, and co-ordinator of the secretariat for the James Lind Alliance
(www.lindalliance.org), a coalition of patients and clinicians collaborating to confront important uncertainties about the
effects of treatments. He was director of the UK Cochrane Centre between 1992 and 2002, and director of the National
Perinatal Epidemiology Unit between 1978 and 1992.
Comments on Fergusson et al. 229
www.SCTjournal.com Clinical Trials 2005; 2: 218232
This is simply the latest example of the
consequences of a lack of scientific and ethical
self-policing among researchers and those who fund
their activities. But what of the research ethics
committees that approved these studies? Ten years
have passed since ethics committees were chal-
lenged publicly to recognise that they were behav-
ing unethically by not taking steps to assure that
they were approving only necessary research [6], yet
there is very little evidence that they have taken this
challenge seriously [13,14].
And what were the editors of journals doing
accepting reports of redundant research for publi-
cation? Had any of them taken seriously the pro-
posal that systematic reviews should be used by
editors and peer reviewers to judge submitted
manuscripts in the context of related, previous
studies [15]?
I propose that the research ethics committees and
journals who approved and published studies of
aprotinin after 1990 should be invited to send
Clinical Trials their comments on the paper by
Fergusson and his colleagues. Not only would this
help to show ethics committees and editors how
they are failing patients and the public in this
domain, but publication of these comments should
help to uncover some of the academic, commercial
and practical pressures that are leading to this
indefensible situation.
As the paper emphasizes, science is meant to be
cumulative, but many scientists are not cumulating
scientifically and those who can call researchers
and academia to account are failing to do so. Not
only are most new studies not designed in the light
of systematic reviews of existing evidence, new
evidence is only very rarely reported in the context
of updates of those reviews [16,17], even though it
was pointed out years ago that the potential for
doing this has been transformed by electronic
publishing [18].
The idea that new research results should be set
in context has existed for well over a century [3].
In 1884, in his Presidential Address to the meeting
of the British Association for the Advancement
of Science in Montreal, Lord Rayleigh, Professor of
Physics at the University of Cambridge, noted
that “the work which deserves, but I am afraid
does not always receive, the most credit is that
in which discovery and explanation go hand
in hand, in which not only are new facts
presented, but their relation to old ones is pointed
out” [19].
The scientific and ethical consequences of
academia’s failure to take research synthesis suffi-
ciently seriously in biomedical and clinical research
[20] are that patients (and the public more
generally) suffer directly and indirectly; policy-
makers, practitioners, and patients have inadequate
information to guide their choices among alterna-
tives; and limited resources for health care and new
research are used inefficiently. Those who wield
power within academia should either publicly
defend their failure to take effective action [21], or
act more forcefully to change this unacceptable
state of affairs.
This paper by Fergusson and his colleagues
compellingly demonstrates why all new research
whether basic or applied should be designed in the
light of scientifically defensible syntheses of existing
research evidence, and reported setting the new
research “in the light of the totality of the available
evidence” [22], thus making clearer to readers what
contribution if any new studies have made to
knowledge.
References
1. Fergusson D, Glass KC, Hutton B, Shapiro S
Randomized controlled trials of aprotinin in cardiac
surgery: could clinical equipoise have stopped the
bleeding? Clinical Trials 2005; 2: 218232.
2. Horder TJ. The organizer concept and modern
embryology: Anglo-American perspectives. Inter-
national Journal of Developmental Biology 2001; 45:
97132.
3. Chalmers I, Hedges LV, Cooper H. A brief history of
research synthesis. Evaluation and the Health Professions
2002; 25: 1237.
4. Antman EM, Lau J, Kupelnick B, Mosteller F,
Chalmers TC. A comparison of results of meta-analyses
of randomized control trials and recommendations of
clinical experts. JAMA 1992; 268: 240 48.
5. Lau J, Schmid CH, Chalmers TC. Cumulative meta-
analysis of clinical trials builds evidence for exemplary
medical care. Journal of Clinical Epidemiology 1995; 48:
4557.
6. Savulescu J, Chalmers I, Blunt J. Are research ethics
committees behaving unethically? Some suggestions for
improving performance and accountability. BMJ 1996;
313: 139093.
7. Chalmers I. Using systematic reviews and registers of
ongoing trials for scientific and ethical trial design,
monitoring, and reporting. In Egger M, Davey Smith G,
Altman D eds. Systematic reviews in health care: meta-
analysis in context, 2nd ed of Systematic Reviews London:
BMJ Books, 2001: 42943.
8. Horn J, de Haan RJ, Vermeulen M, Luiten PGM,
Limburg M. Nimodipine in animal model experiments
of focal cerebral ischaemia: a systematic review. Stroke
2001; 32: 243338.
9. Sandercock P, Roberts I. Systematic reviews of animal
experiments. Lancet 2002; 2: 586.
10. Pound P, Ebrahim S, Sandercock P, Bracken M,
Roberts I. Where is the evidence that animal research
benefits humans? BMJ 2004; 328: 514 517.
11. Gilbert R, Salanti G, Harden M, See S. Infant
sleeping position and the sudden infant death syn-
drome: systematic review of observational studies and
historical review of clinicians’ recommendations from
19402000. International Journal of Epidemiology (in
press).
12. Laupacis A, Fergusson D. Drugs to minimize
perioperative blood loss in cardiac surgery:
230 Comments on Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
meta-analyses using periopearetive blood transfusion as
the outcome. The International Study of Peri-operative
Transfusion (ISPOT) Investigators. Anesthesia and Analge-
sia 1997; 85: 1258 67.
13. Chalmers I. Lessons for research ethics committees.
Lancet 2002; 359: 174.
14. Mann H, Djulbegovic B. Why comparisons must
address genuine uncertainties. James Lind Library,
www.jameslindlibrary.org (accessed 21 April 2005).
15. Jefferson T, Deeks J. The use of systematic reviews for
editorial peer reviewing: a population approach. In
Godlee F, Jefferson T eds. Peer review in health sciences.
London: BMJ Books, 1999: 22434.
16. Clarke M, Chalmers I. Discussion sections in reports
of controlled trials published in general medical
journals: islands in search of continents? JAMA 1998;
280: 28082.
17. Clark e M, Alderson P, Chalmers I. Discussion
sections in reports of controlled trials published
in general medical journals. JAMA 2002; 287: 2799801.
18. Chalmers I. Electronic publications for updating
controlled trial reviews. Lancet 1986; 2: 287.
19. Rayleigh, The Right Hon Lord. Presidential address
at the 54th meeting of the British Association for the
Advancement of Science, Montreal, August/September
1884. London: John Murray, 1885: 323.
20. Alderson P, Gliddon L, Chalmers I. Academic
recognition of critical appraisal and systematic reviews
in British postgraduate medical education. Medical
Education 2003; 37: 386 87.
21. Chalmers I. Academia’s failure to support systematic
reviews. Lancet 2005; 365: 469.
22. Begg C, Cho M, Eastwood S et al. Improving the
quality of reporting of randomized controlled trials. The
CONSORT statement. JAMA 1996; 276: 637 39.
Comment
John G. Augoustides MD, FASE
a
and Lee A. Fleisher
MD, FACC
b
How many randomized controlled clinical trials
(RCTs) are required to document that aprotinin
improves hemostasis after cardiac surgery? Clinical
equipoise, real doubt about drug efficacy, is violated
when RCTs are conducted even after drug efficacy
has been proven. When clinical equipoise is violated
in an RCT, patients may be unethically randomized
to placebo and denied proven benefit from the
“study drug”.
Fergusson et al. present a comprehensive meta-
analysis (64 RCTs: n ¼ 8040) that details the extent
of redundant aprotinin RCTs conducted after the
hemostatic efficacy of aprotinin in cardiac
surgery was established. An RCT to evaluate
drug efficacy is redundant if efficacy has already
been proven. In 1997 Fergusson co-authored an
extensive meta-analysis (45 RCTs: n ¼ 5808) that
quantified the hemostatic efficacy of aprotinin in
cardiac surgery showing a significantly decreased
allogeneic blood exposure (odds ratio 0.31, 95%
confidence interval 0.25 0.39; P , 0.0001) and
significantly decreased reoperation rate for bleeding
(odds ratio 0.44, 95% confidence interval 0.27
0.73; P ¼ 0.001) [1].
In this follow-up meta-analysis, focused more on
aprotinin RCT design rather than efficacy, Fergusson
et al. demonstrate and quantify the degree of RCT
redundancy. Substantial aprotinin efficacy was
observed in the very first of 64 RCTs; even when
the analysis was restricted to those trials deemed of
highest quality, 31 aprotinin RCTs were published
after 1990, when aprotinin efficacy should have
been concluded from the first three trials. Redun-
dancy persisted in the subgroup analyses stratified
by aspirin exposure and primary and repeat cardiac
surgery. As stressed by these investigators and by
many who have preceded them [25], a systematic
literature review preceding a trial should prevent
RCT redundancy.
What are the possible explanations for
this apparent tremendous redundancy in aprotinin
RCTs? The first possible may be that the most
cost-effective hemostatic aprotinin dosage regimen
remained undefined, an issue because high-
dose aprotinin is expensive. Perhaps ongoing RCTs
were required to delineate the lowest dose of apro-
tinin that produced adequate hemostasis [69].
A second possible explanation for the seeming
excess RCTs is that they were mounted to better
quantify the balance between the benefit on the
bleeding outcome and the risk of vascular graft
thrombosis, which was a major controversy [10,11].
Although the authors claim this would require
larger RCTs than were conducted, this may merit
further inquiry.
A third possible explanation is that aprotinin’s
biologic effects beyond hemostasis justified evi-
dence-based evaluation, such as in vitro antic-
oagulant properties [12], platelet-sparing effects
[13], anti-inflammatory properties [14,15], and
organ protection properties [16]. It is doubtful
whether such endpoints would justify RCTs in the
face of proven clinical efficacy, but it would help us
better understand the phenomenon observed here
to know how many of the protocols cited such issues
as part of the trial justification.
a
John G. Augoustides is an Assistant Professor of Anesthesia, University of Pennsylvania School of Medicine. He is a Fellow
of the American Society of Echocardiography. He practices as a cardiovascular anesthesiologist, and has a special interest
in thoracic aortic procedures, including aprotinin.
b
Lee A. Fleisher is the Robert D. Dripps Professor and Chair of Anesthesia as well as Professor of Medicine at the University
of Pennsylvania School of Medicine. He is an associate scholar of the Center for Clinical Epidemiology and Biostatistics at
the University.
Comments on Fergusson et al. 231
www.SCTjournal.com Clinical Trials 2005; 2: 218232
A fourth possible explanation is that redundant
aprotinin RCTs were conducted to improve market
share and diffuse information about the effectiveness
of aprotinin into regions that did not use it. These
motives for RCTs are less than ideal, and represent
suboptimal clinical practice, but the ethical argu-
ment becomes more complex if the routine care in
the study center did not include the drug.
In summary, this high-quality meta-analysis
demonstrates significant redundancy in aprotinin
RCTs within the cardiac surgical population, includ-
ing subgroups exposed to aspirin, primary and
repeat surgery. A full understanding of the reasons
for this redundancy may require further investi-
gation with respect to the four justifications out-
lined here. These investigations might reduce what
wejudgetobethedegreeofaprotininRCT
redundancy, but will probably not eliminate it.
In the future we can anticipate more aprotinin
RCTs, addressing clinical concerns such as 1) safety
in specific cardiac surgical subsets (off-pump coron-
ary bypass, deep hypothermic circulatory arrest); 2)
platelet protection (patients exposed to aspirin and
clopidogrel); and 3) organ protection (the brain and
heart, in particular). Each RCT is justified as long as
clinical equipoise is respected and the hypothesis is
clearly formulated a priori, based on a thorough
literature review. This meta-analysis highlights how
essential these steps are for responsible RCT design.
We will need to follow them to ethically advance our
understanding, in an evidence-based fashion, of
ways to improve perioperative outcomes for our
patients.
References
1. Laupacis A, Fergusson D. Drugs to minimize
perioperative blood loss in cardiac surgery:
meta-analyses using perioperative blood transfusion as
the outcome. The International Study of Peri-operative
Transfusion (ISPOT) Investigators. Anesth Analg 1997; 85:
125867.
2. Browner WS. Clinical research: a simple recipe for
doing it well. Anesthesiology 1994; 80: 92328.
3. Begg CB, Cho MK, Eastwood S et al. Improving the
quality of reporting of randomized controlled trials: the
CONSORT statement. JAMA 1996; 276: 637 39.
4. Moher D, Schulz KF, Altman D. The CONSORT
statement: revised recommendations for improving the
quality of reports of parallel-group randomized trials.
Lancet 2001; 357: 119194.
5. Todd MM. Clinical research manuscripts in Anesthe-
siology. Anesthesiology 2001; 95: 1051 53.
6. Liu B, Belboul A, Radberg G et al. Effect of reduced
aprotinin dosage on blood loss and use of blood products
in patients undergoing cardiopulmonary bypass. Scand J
Thorac Cardiovasc Surg 1993; 27: 149 55.
7. Bailey CR, Kelleher AA, Wielogorski AK. Random-
ized placebo-controlled double-blind study of three
aprotinin regimens in primary cardiac surgery. Br J Surg
1994; 81: 96973.
8. Alvarez JM, Quiney NF, McMillan D et al. The use
of ultra-low-dose aprotinin to reduce blood loss in
cardiac surgery. J Cardiothorac Vasc Anesth 1995; 9:
2933.
9. Levy JH, Pifarre R, Schaff HV. Amulticenter,
double-blind, placebo-controlled trial of aprotinin
for reducing blood loss and the requirement for
donor-blood transfusion in patients undergoing repeat
coronary artery bypass grafting. Circulation 1995; 92:
223644.
10. Alderman EL, Levy JH, Rich JB et al. Analyses of
coronary graft patency after aprotinin use: results from
the International Multicenter Aprotinin Graft Patency
Experience (IMAGE) trial. J Thorac Cardiovasc Surg 1998;
116: 71630.
11. Cosgrove DM, Heric B, Lytle BW et al. Aprotinin
therapy for reoperative myocardial revascularization: a
placebo-controlled study. Ann Thorac Surg 1992; 54:
103138.
12. Dietrich W, Dilthey G, Spannagl M et al. Influence
of high-dose aprotinin on anticoagulation, heparin
requirement, and celite- and kaolin-activated clotting
time in heparin-pretreated patients undergoing open-
heart surgery. A double-blind, placebo-controlled study.
Anesthesiology 1995; 83: 67989.
13. Mohr R, Goor DA, Lusky A et al. Aprotinin prevents
cardiopulmonary bypass-induced platelet dysfunction. A
scanning electron microscope study. Circulation 1992;
86: II405409.
14. Ashraf S, Tian Y, Cowan D et al. “Low-dose”
aprotinin modifies hemostasis but not proinflam-
matory cytokine release. Ann Thorac Surg 1997; 63:
6873.
15. Tassani P, Augustin N, Barankay A et al. High-dose
aprotinin modulates the balance between proinflamma-
tory and anti-inflammatory responses during coronary
artery bypass graft surgery. J Cardiothorac Vasc Anesth
2000; 14: 68286.
16. Wendel HP, Heller W, Michel J et al. Lower cardiac
troponin T levels in patients undergoing cardio-
pulmonary bypass and receiving high-dose aprotinin
therapy indicate reduction of perioperative myo-
cardial damage. J Thorac Cardiovasc Surg 1995; 109:
116472.
232 Comments on Fergusson et al.
Clinical Trials 2005; 2: 218 232 www.SCTjournal.com
    • "Indeed, cases exist in medicine where primary research is shown to be no longer necessary, since combining studies in one synthesis shows significance where individual studies fail to find any. For example, a SR which conducted cumulative meta-analytical techniques on 64 trials investigating the effectiveness of the drug Aprotinin at controlling perioperative bleeding showed that the effectiveness was apparent after only 12 trials (Fergusson et al., 2005 ). Thus this SR identified 52 unnecessary trials that had a SR been performed after the twelfth study, the treatment effect would have been apparent, duplicate trials would have been avoided, and patients would have experienced the benefit of a useful drug ten years earlier (Freeman et al., 2006). "
    [Show abstract] [Hide abstract] ABSTRACT: The volume of scientific literature continues to expand and decision-makers are faced with increasingly unmanageable volumes of evidence to assess. Systematic reviews (SRs) are powerful tools that aim to provide comprehensive , transparent, reproducible and updateable summaries of evidence. SR methods were developed, and have been employed, in healthcare for more than two decades, and they are now widely used across a broad range of topics, including environmental management and social interventions in crime and justice, education, international development, and social welfare. Despite these successes and the increasing acceptance of SR methods as a 'gold standard' in evidence-informed policy and practice, misconceptions still remain regarding their applicability. The aim of this article is to separate fact from fiction, addressing twelve common misconceptions that can influence the decision as to whether a SR is the most appropriate method for evidence synthesis for a given topic. Through examples, we illustrate the flexibility of SR methods and demonstrate their suitability for addressing issues on environmental health and chemical risk assessment.
    Full-text · Article · Sep 2015
    • "Throughout the cumulative meta-analysis, the upper limit of the confidence interval did not go higher than 0.65, and in this study, the final meta-analyses was 0.34 (95% CI: 0.24–0.41) after the publication of a trial in June 2002, by which time a total of more than 8000 patients had been randomised [13]. "
    [Show abstract] [Hide abstract] ABSTRACT: “Cumulative meta-analysis” describes a statistical procedure to calculate, retrospectively, summary estimates from the results of similar trials every time the results of a further trial in the series had become available. In the early 1990s, comparisons of cumulative meta-analyses of treatments for myocardial infarction with advice promulgated through medical textbooks showed that research had continued long after robust estimates of treatment effects had accumulated, and that medical textbooks had overlooked strong, existing evidence from trials. Cumulative meta-analyses have subsequently been used to assess what could have been known had new studies been informed by systematic reviews of relevant existing evidence and how waste might have been reduced.
    Full-text · Article · Jul 2014
    • "We notice that there is a strikingly different behavior of the DL estimator between simulations with dichotomous andFig. 3. A double triangular test [15] was used for an SMA with α = 0.01 and power = 0.90 for the " Aprotinin in Cardiac Surgery " data [19]. The relevant effect size was equal to log(OR = 0.53) = 0.6349. "
    [Show abstract] [Hide abstract] ABSTRACT: Estimators for the variance between treatment effects from randomized clinical trials (RCTs) in a meta-analysis may yield divergent or even contradictory results. In a sequential meta-analysis (SMA), their properties are even more important, as they influence the point in time at which definite conclusions are drawn. In this study, we evaluated the properties of estimators of heterogeneity to be used in an SMA. We conducted an extensive simulation study with dichotomous and continuous outcome data and applied the estimators in real life examples. Bias and variance of the estimators were used as primary evaluation criteria, as well as the number of RCTs and patients from the accumulating trials needed to get stable estimates. The simulation studies showed that the well-known DerSimonian-Laird (DL) estimator largely underestimates the true value for dichotomous outcomes. The two-step DL (DL2) significantly improves this behaviour. In general, the DL2 and Paule-Mandel (PM) estimators are recommended for both dichotomous and continuous outcome data for use in an SMA.
    Full-text · Article · Dec 2013
Show more