Solving the Challenges of Large Multicenter Trials in Anesthesia.

2009·1.2 MB·English

by  Wallace, S

pprroocceeeeddiinnggss
Cardiovascinu Ilnatre Annseivset hCeasriea
ORIGINAL ARTICLE
Endorsed by
46
Solving the Challenges of Large
multicenter Trials in Anesthesia
S. Wallace¹, P.S. Myles²
¹Department of Anaesthesia and Perioperative Medicine, Alfred Hospital; Melbourne, Australia;
² Department of Anaesthesia and Perioperative Medicine, Alfred Hospital;
Professor, Academic Board of Anaesthesia and Perioperative Medicine, Monash University; NHMRC Practitioner Fellow;
Melbourne, Australia.
ABSTRACT
This paper describes many of the challenges encountered when establishing a large multicentre trial in cardiac
anesthesia. We address funding, authorship, multisite ethics review, patient recruitment, data quality manage-
ment, communication with individual sites, and strategies to enhance cooperation and patient recruitment.
Keywords: cardiac anesthesia, anesthesia, multicenter trials, methodology.
INTRODuCTION in coronary artery surgery (5). The aspirin
and tranexamic acid for coronary artery sur-
Large randomized controlled trials, test- gery (ATACAS) trial is a factorial designed
ing new treatments in routine clinical trial in 4600 patients, designed to detect
practice, can optimize generalizability and thrombotic (principally myocardial infarc-
so are clinically relevant and reliable (1). tion (MI), stroke, and death) and bleeding
They thus provide the best evidence of complications – see www.atacas.org.au.
effectiveness (2, 3). Most large trials are We reasoned that although aspirin may in-
multicenter studies, and often conducted crease bleeding, there is some evidence that
in many countries. Despite being labelled it could reduce thrombotic complications
as “simple” or “pragmatic” trials (1, 3, 4), after coronary artery surgery. The opposite
reflecting their focus on easy-to-administer can be said for antifibrinolytic therapy. In
treatments in routine settings, they create both cases there are insufficient randomized
a number of difficult challenges for those trials to address these questions unequivo-
involved. However the rewards are great cally. A large multicenter trial is required
and include the opportunity to answer im- (5). Pharmaceutical companies are unlikely
portant clinical questions reliably, to pub- to fund such reaearch, and so specialty or
lish in top-ranked journals, and to be rec- government research bodies must provide
ognized by your peers. We would like to financial support.
share our experience of establishing a large
multicenter trial testing two interventions
PROTOCOL DEvELOPmENT
Corresponding author: AND PLANNED SuB-STuDIES
Sophie Wallace
BHlthSc, MPH , Research Manager
Department of Anaesthesia & Perioperative Medicine The effort and commitment to undertake
Alfred Hospital, Commercial Road, Melbourne, Vic, 3004, Australia
e.mail: [email protected] or contribute to a large multicenter trial
Solving the Challenges of Large multicenter Trials in Anesthesia
is substantial. Before embarking on such in the trial (6). The trial steering committee 47
a project, the aims and study hypothesis should meet at regular intervals throughout
should be clearly outlined, hopefully ad- the life of the trial to discuss overall man-
dressing a clinically important question. A agement, progress and policy decisions.
supportive literature review will provide a Trial management includes data manage-
background and justification for conducting ment, data security and back-up, quality
such a trial. There are often opportunities checks, review of patient safety and includ-
to design small sub-studies at selected cen- ing consideration of reports from the tri-
tres, requiring additional data collection, als’ data and safety monitoring board. Each
increasing opportunities for authorship individual site reports via the study chief
and additional publications. The explana- investigators to the steering committee. Ide-
tory data can be used to link the effects of ally each site should have a lead investigator
an intervention to selected intermediate who takes responsibility for overseeing the
outcomes that may correlate with the main study at their site, for which they should be
study aims. For the ATACAS trial we are acknowledged in the final publication. Fi-
conducting substudies to investigate aspi- nancial management should be continually
rin non-responsiveness in a subset of our assessed throughout the trial (7).
study population, perioperative genomics
with the iPEGASUS group, and the effects
of tranexamic acid on seizure risk. FuNDING
The study protocol describes the science of
the research project, and the study proce- Large trials require substantial funding.
dures manual the structure and processes The ATACAS trial is primarily funded by
that allow it to be properly conducted. the Australian National Health and Medi-
cal Research Council (NHMRC). Being gov-
ernment provided, such funding is usually
STuDy mANAGEmENT limited, and when considering the costs
and demands on clinicians and research
Experienced trial management and leader- staff, it is usually insufficient to properly
ship are vital for successful large scale clini- fund all aspects of the trial. Most centers
cal trials. Numerous individual centers, have other cardiothoracic research projects
sometimes with their own research inter- which may compete for patients, research
ests and studies, must arrive at a consensus staff availability, and interest from local
regarding study procedures and data collec- clinicians. There may be competition with
tion, inclusion of other clinicians (not just pharmaceutical company-funded projects
anaesthesiologists) and language and cultur- which typically provide much higher rates
al differences, all of which test goodwill and of remuneration (8, 9). The ATACAS trial
cooperation on a multinational scale. Trials is an investigator-initiated trial, funding
should have a core group of co-investigators individual sites Australian Dollars (AUD)
responsible for the overall management and 700 (about Euro 390) per patient enrolled;
running of the trial, headed by a Principal we have been involved in some pharma-
Investigator (PI). The PI, co-investigators, ceutical company-funded studies providing
and perhaps other experts, constitute the funds at 5-10 times that rate.
trial steering group. Some bodies recom- Large clinical trials aim to address clinically
mend that the chairman of the trial steering important questions, often testing simple in-
committee should not otherwise be involved expensive interventions. There is a compelling
S. Wallace, P.S. Myles
48 argument that such trials ought to be funded Another mandatory step is informed con-
by the health (not medical research) budget sent (14), for which local expectations and
because of the opportunities to immediately requirements can vary, as well as some-
improve outcomes of healthcare (1, 10). times introducing a need for translation of
such documents.
The time line for this process from begin-
PROCuREmENT OF STuDy DRuG ning (initial contact with site) to end (man-
agement receiving the approval letter) av-
Initial management hurdles can include erages six months. This is a major barrier
sourcing of study drug and matched place- for many sites who may otherwise be inter-
bo, and these issues can vary across coun- ested in collaboration (15).
tries because of differences in the status of
the study drug licensing. For ATACAS, we
approached the pharmaceutical companies AuThORShIP AGREEmENT
that produce aspirin and tranexamic acid
to assist with free supply of study drug and Researchers are rated according to the qual-
matched placebo. ity and quantity of their publication record.
For aspirin, this proved to be relatively Large trials involve many individuals, but
straight-forward and positive, but for only some deserve authorship on the main
tranexamic acid it resulted in a two year publication(s). Others may share in au-
delay and eventual disappointment. We thorship of subsidiary publications. In any
subsequently arranged our own purchase case, all of those involved in the conduct of
of tranexamic acid from the UK, leaving us a large trial should be acknowledged, and
with the cost-burden for supply of this drug this is typically published as an appendix to
to most ATACAS sites around the world. the main publications.
This of course also delayed the commence- For this reason acknowledged site leaders
ment of the trial. ought to be given credit for their leading
Following public announcement of the re- role within their own institutions.
sults of the BART trial (11), and the market Authorship is a vexed topic, and it cannot
withdrawal of aprotinin around the world be overstated: who and under what circum-
(12), the initial purchase price of tranexam- stances each collaborator is included in the
ic acid went from AUD30 (about € 17) per authorship or acknowledgement lists must
box to AUD100 overnight. be outlined at the beginning of the trial,
This added a new and unexpected cost bur- and ideally a signed authorship agreement
den to the study. be completed in order to avoid disappoint-
Fortunately this did not interrupt recruit- ment and conflict.
ment although it highlights how trial bud-
gets can suddenly be tested.
INDEmNITy
GOvERNANCE Multicenter trials should have a clinical trial
agreement (CTA) signed with each individ-
Before enrolling participants in a clinical ual site. This pertains to both pharmaceu-
trial individual sites must gain approval by tical-sponsored and investigator-initiated
their hospital’s institutional review board trials.
or ethics committee (13). The CTA document requires legal review
Solving the Challenges of Large multicenter Trials in Anesthesia
and comment from each site. This adds cost surgeon preference have proven to be ob- 49
and poses another potential for delaying stacles to large multicenter trials in surgery
start up of the trial. (8), but Devereaux et al. (22) have suggest-
Pharmaceutical-sponsored trials have the ed solutions.
resources to provide their own indemnity
insurance, but investigator-initiated tri-
als rarely can because such funding is not RESEARCh
included in most research funding bodies’ NuRSES/COORDINATORS
budgets. AND PATIENT RECRuITmENT
In such cases it is usual to ask that individ-
ual sites cover their own indemnity costs, It is very important that the infrastructure
because the study procedures usually only and staff to conduct research at each site
involve currently established therapies. are actively sought, available and most im-
As we, and others (10), argue, investigator- portantly supported (23-26). Sites that have
initiated large pragmatic trials ought to be limited infrastructure in place to conduct
considered “public good” research and so research must commence with recruitment
individual institutions should support such of a research nurse or coordinator, and this
trials. takes time (recruitment, training). The co-
ordinating center for the trial can assist in
this regard.
SITE SELECTION Constant communication and availability
of assistance has proved to be important in
Site selection is vital to a successful trial. It facilitating this role. The research nurse is
relies on some research infrastructure and responsible for the screening, recruitment,
staffing, to identify eligible patients for re- consent, data collection, data storage, sub-
cruitment, study interventions and follow- ject logs, data entry, and protection of hu-
up (16). man subjects in clinical trials (Figure 1)
Initial site investigators invited to join the (27).
ATACAS trial were those previously in- It is vital that the research nurse be sup-
volved in other multicenter trials (17-20), ported by the site investigator, and partici-
and have proven track records. All sites pating units (28), as this will be the main
were asked to discuss the feasibility of un- contributing factor to the success or failure
dertaking the trial with their respective of patient recruitment (25). It has been
cardiothoracic surgical colleagues. Support previously reported that the individual un-
from the surgeons at each institution was dertaking the recruitment can influence re-
an essential component for the trial. New cruiting patients to the trial (29). No differ-
sites were also sought. ence was found when doctors or research
Publicity for the trial occurred via presen- nurses were examined, but there was a
tation at scientific meetings, establishment statistically significant difference when re-
of a trial website (www.atacas.org.au), and cruitment was undertaken by the operating
journal publication (5). surgeon (29).
Rahbari et al (21) challenge the surgical This therefore highlights the importance of
community to optimize study power us- having the support from all disciplines in-
ing properly conducted, pragmatic (multi- volved in the research.
center) trials with large sample sizes. Varia- A recent survey of trials published in the
tion in surgical practice, surgical skill and Lancet or BMJ found that nearly 60% of
S. Wallace, P.S. Myles
50
Figure 1
Day-to-day role
of the research nurse.
trials had either failed their recruitment proving recruitment (29). The management
target or required an extension of their team provide the following process to assist
planned recruitment period (29). Recruit- with site recruitment (Figure 2).
ment of participants to trials is one of the Newsletters are used to disseminate infor-
most important aspects to a successful trial mation to all sites and focus on recruitment
(23, 24, 30). techniques, addressing frequently asked
It has long been recognized that recruit- questions, current and new sites, future
ment is a much greater problem than is per- meetings, changes to the database and re-
ceived by the investigator when instigating cently published literature relevant to the
and designing of the trial (8, 24). During study.
the course of the trial it is important to im-
plement and identify strategies to overcome
barriers to recruitment (31). DATA mANAGEmENT
Delays in recruitment lead to important AND mONITORING
scientific answers being left unanswered,
increased unidentified costs, early clo- Data collection from multiple sites, in vari-
sure of trial (8, 23-25, 30, 31), statistical ous time zones, needs to be streamlined
power may be reduced (29, 31), poor mo- and secure.
rale (16), and delayed uptake into clinical For the ATACAS trial we use paper-based
practice. case report forms (CRF) at each centre, and
Studies have shown that individual site the data are later transferred onto a web-
training and regular feedback and commu- based form. The online data entry is ac-
nication to staff improve recruitment rates cessed through a password-protected link
(15, 32). Start-up meetings, personalised on the trial website.
education and training visits assist in im- The site also offers a trial summary, recruit-
Solving the Challenges of Large multicenter Trials in Anesthesia
51
Figure 2
Personalising the trial.
ing centres, current randomisation and also tabulated for review by the steering com-
a trial register interest section for any sites mittee. If a site has a lag in recruitment, the
wishing to participate or contact the man- research manager or project officer can ini-
agement team. tiate communication with the site to assist
The web-based database therefore allows in identifying areas requiring assistance.
for the original study CRF to be retained Correct and complete study procedures can
at each site for audit and privacy purposes, be checked, including consent, secure data
as well as reducing the time spent in identi- storage, and verification of trial events.
fying and resolving data queries, and mini-
mising data entry errors.
This has been identified as one factor that CONCLuSIONS
may assist in increasing efficiency (29). We
believe simple study procedures encourage Large multicenter clinical trials are de-
participation in multicentre trials. manding but ultimately rewarding in that
Careful monitoring of the recruitment pro- they provide reliable answers to everyday
cess throughout studies is vital, and enables clinical problems.
the management center to identify problem Clear guidelines on all aspects of the trial
areas at individual sites (25). procedures assist in a teamwork approach
These logs can be sent to the data man- to overcoming the many barriers to suc-
agement center on a monthly basis and cessful completion of such trial.
S. Wallace, P.S. Myles
52 No conflict of interest acknowledged by the 12. Nov. 5, 2007, the U.S. Food and Drug Adminis-
authors tration announced that Bayer Pharmaceuticals
Corp. agreed to an FDA-requested marketing
suspension of Trasylol, a drug used to control
REFERENCES bleeding during heart surgery. http://www.fda.
gov/NewsEvents/Newsroom/PressAnnounce-
1. Tunis SR, Stryer DB, Clancy CM. Practical ments/2007/ucm109021.htm (accessed Aug
clinical trials: increasing the value of clinical 7th, 2007)
research for decision making in clinical and 13. London AJ. Reasonable risks in clinical research:
health policy. JAMA 2003; 290: 1624-1632. A critique and a proposal for the integrative ap-
2. Collins R, MacMahon S. Reliable assessment proach. Stat Med 2006; 25: 2869-2885.
of the effects of treatment on mortality and 14. Karunaratne AS, Myles PS, Ago MJ, Kome-
major morbidity, I: clinical trials. Lancet 2001; saroff PA. Communication deficiencies in re-
357: 373-380. search and monitoring by ethics committees.
3. Myles PS. Why we need large trials in anaes- Intern Med J 2006; 36: 86-91.
thesia and analgesia. In: Tramer MR. An evi- 15. de Salis I, Tomlin Z, Toerien M, Donovan J.
dence based resource in anaesthesia and anal- Qualitative research to improve RCT recruit-
gesia, 2nd ed. London: BMJ Publishing Group ment: issues arising in establishing research
2003, pp 12-21. collaborations. Contemp Clin Trials 2008; 29:
4. Yusuf S, Collins R, Peto R. Why do we need 663-670.
some large, simple randomized trials? Stat Med 16. Hague WE, Gebski VJ, Keech AC. Recruitment
1984; 3: 409-422. to randomized studies. MJA 2003; 178: 579-
5. Myles PS, Smith J, Knight J, et al. Aspirin and 581.
Tranexamic Acid for Coronary Artery Surgery 17. Rigg JR, Jamrozik K, Myles PS, et al. Epidural
(ATACAS) Trial: rationale and design. Am anaesthesia and analgesia and outcome of ma-
Heart J 2008; 155: 224-230. jor surgery: a randomised trial. Lancet 2002;
6. Health Technology Assessment programme: 359: 1276-1282.
research governance. http://www.ncchta.org/ 18. Myles PS, Leslie K, McNeil J, et al. Bispectral
investigators/governance.shtml; (accessed Aug index montiroing to prevent awareness dur-
26th 2009.) ing anaesthesia: the B-Aware randomised con-
7. Snowdon C, Elbourne DR, Garcia J, et al. trolled trial. Lancet 2004: 363: 1757-1763.
Financial considerations in the conduct of 19. Myles PS, Chan MT, Leslie K, et al. Effect of
multi-centre randomised controlled trials: ev- nitrous oxide on plasma homocysteine and fo-
idence from a qualitative study. Trials 2006; late in patients undergoing major surgery. Br J
7: 34. Anaesth 2008; 100: 780-786.
8. Anyanwu AC, Treasure T. Surgical research 20. Devereaux PJ, Yang H, Yusuf S, et al. Effects
revisited: clinical trials in the cardiothoracic of extended-release metoprolol succinate in pa-
surgical literature. Eur J Cardiothorac Surg tients undergoing non-cardiac surgery (POISE
2004; 25: 299-303. trial): a randomised controlled trial. Lancet
9. Frank G. Current challenges in clinical trial 2008; 371: 1839-1847.
patient recruitment and enrollment. SoCRA 21. Rahbari N, Diener M, Wente M, Seiler C. De-
Source 2004; 30. http://www.socra.org/ velopment and perspectives of randomized
pdf/200402_Current_Challenges_Recruit- controlled trials in surgery. Am J Surg 2007:
ment_Enrollment.pdf (accessed Aug 1st 2009) 194: S148-152.
10. McNeil JJ, Nelson MR, Tonkin AM. Public 22. Devereaux PJ, Bhandari M, Clarke M, et al.
funding of large-scale clinical trials in Austra- Need for expertise based randomised con-
lia. Med J Aust 2003; 179: 519-520. trolled trials. BMJ 2005; 330: 88.
11. Fergusson DA, Hébert PC, Mazer CD, et al. A 23. Lovato LC, Hill K, Hertert S, et al. Recruitment
comparison of aprotinin and lysine analogues for controlled clinical trials: literature sum-
in high-risk cardiac surgery. N Engl J Med mary and annotated bibliography. Control Clin
2008; 22: 2319-2331. Trials 1997; 18: 328-352.
Solving the Challenges of Large multicenter Trials in Anesthesia
24. Leathem CS, Cupples ME, Byrne MC, et al. search nurses to clinical trials. J Clin Nurs 53
Identifying strategies to maximise recruit- 2007; 17: 2664-2666.
ment and retention of practices and patients 29. Watson JM, Torgerson DJ. Increasing recruit-
in a multicentre randomised controlled trial of ment to randomized trials: a review of ran-
an intervention to optimise secondary preven- domised controlled trials. BMC Med Res Meth-
tion for coronary heart disease in primary care. odol 2006; 6: 34.
BMC Med Res Method 2009; 9: 40. 30. Toerien M, Brookes ST, Metcalfe C, et al. A
25. Hunninghake DB, Darby CA, Probstfield JL. review of reporting of participant recruitment
Recruitment experience in clinical trials: Lit- and retention in RCTs in six major journals.
erature summary and annotated bibliography. Trials 2009; 10: 1-12.
Control Clin Trials 1987; 8: 6S-30S. 31. Howard L, de Salis I, Tomlin Z, et al. Why is
26. Raja-Jones H. Role boundaries - research nurse recruitment to trials difficult? An investiga-
or clinical nurse specialist? A literature review. tion into recruitment difficulties in an RCT of
J Clin Nurs 2002; 11: 415-420. supported employment in patients with severe
27. Yanagawa H, Akaishi A, Miyamoto T, et al. mental illness. Control Clin Trials 2009; 30:
Role of clinical research coordinators in pro- 40-46.
moting clinical trials of drugs for surgical pa- 32. Pope C, Mays N. Reaching the parts other
tients. Int Archive Med 2008; 1: 26. methods cannot reach: an introduction to qual-
28. Spilsbury K, Petherick E, Cullum N, et al. The itative methods in health and health services
role and potential contribution of clinical re- research. BMJ 1995; 311: 42-45.