Registering Clinical Trials

SPECIAL COMMUNICATION

Kay Dickersin, PhD, MA That it is not possible to find information about all initiated clinical trials is Drummond Rennie, MD of international concern. This is a particular worry because scientists tend N 1974, SHORTLY AFTER PRESIDENT to publish their positive findings more often than their negative findings (pub- Nixon had called for a “War Against lication bias). A comprehensive register of initiated clinical trials, with each Cancer,” Mary Lasker, patroness and advocate of clinical research, trial assigned a unique identifier, would inform reviewers, physicians, and atI a meeting of the President’s Cancer others (eg, consumers) about which trials had been started and directly ad- Panel, asked the National Cancer In- dress the problem of publication bias. Patients and their clinicians could also stitute to publish a book, to be up- know which trials are open for enrollment, thus speeding medical ad- dated every 6 months, listing all ongo- vances. Individuals who participate in clinical trials typically provide con- ing cancer treatment protocols in the sent in the belief that they are contributing to medical knowledge. But if the 1 United States. The idea was that phy- knowledge gained is never reported, the trust between patients and inves- sicians would be able to identify open tigators and that between patients and research ethics review boards are both trials in which their patients could en- roll. Just a few years later, Tom Chalm- damaged. Ethical issues are of particular concern if industry is gaining fi- ers, former director of the National In- nancially from public involvement in trials, but refusing to reciprocate by stitutes of Health (NIH) Clinical Center, making information from industry-sponsored trials generally available. All and president and dean at Mt Sinai stakeholders—investigators, research organizations and institutions, jour- Medical Center, extended this con- nal editors, lawmakers, consumers, and others—must act now, together and cept to include registers of clinical trials in their own domains, to ensure comprehensive registration of clinical trials. in all areas.2 Lasker and Chalmers had different aims. Lasker’s aim was to speed JAMA. 2003;290:516-523 www.jama.com a “cure for cancer” by disseminating in- ent countries, making energetic and me- seemed irrefutable evidence gathered formation to physicians and their pa- ticulous efforts to perform systematic from observational studies that hor- tients, so that there would be no short- reviews of the evidence for different mone therapy was effective in prevent- age of participants in clinical trials. types of treatments, have each re- ing cardiovascular events.12-15 Since the Chalmers’ aim was at least as impor- ported substantial difficulties in ob- British Medical Research Council’s tant in any such “war”; it was to re- taining a good account of the exis- landmark RCT of streptomycin in pul- duce bias in the reporting of trials. Ul- tence and number of trials, the numbers monary tuberculosis was reported in timately, both recognized an enormous of patients included, the number of re- 1948,16 perhaps as many as 1 million gap in the dissemination of good infor- ports associated with each trial, and the controlled trials have been carried out; mation and both hoped to speed the de- investigators involved.4-10 it is estimated that only about half of livery of the best new treatments to the which have been reported. This esti- patient. The Importance and Number 3 mate is derived by using the current Yet Manheimer and Anderson, of Randomized Clinical Trials nearly 30 years after Lasker, could write, The study design conferring the best Author Affiliations: Center for Clinical Trials and Evi- “No comprehensive system for track- evidence for effects of interventions is dence-based Healthcare, Brown Medical School, Provi- ing, organizing, and disseminating in- dence, RI, and Director, US Cochrane Center, San Fran- the randomized clinical trial (RCT).11 formation about ongoing clinical trials cisco, Calif (Dr Dickersin). Dr Rennie is Deputy Editor, This has been borne out recently by the JAMA, and co-director of US Cochrane Center, San currently exists.” And decades after Francisco, Calif. Heart and Estrogen/progestin Replace- Chalmers, different groups in differ- Corresponding Author and Reprints: Kay Dickersin, ment Study and the Women’s Health PhD, MA, Center for Clinical Trials and Evidence- Initiative Study, both of which are clini- based Healthcare, Brown Medical School, 169 See also p 495. Angell St, Box G-S2, Providence, RI 02912 (e-mail: cal trials that have overturned what had [email protected]).

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS total number of trials in the Cochrane Unpublished Trials detect any beneficial effect on substan- Central Register of Controlled Trials,17 and Publication Bias tive outcomes but did show an increase the proportion of total initiated trials Even if there were a perfect system for in sudden death for patients with ven- that are estimated to be published,17-19 indexing published trials, it is likely that tricular arrhythmias.31 Despite the fact and the total number of biomedical a sizeable proportion of trials would re- that more than 50 trials involving more journals that remain to be hand main unpublished.18,19 Studies in nu- than 23000 patients had been con- searched.17 In addition, many trials are merous areas of medicine have shown ducted, the effects observed were appar- terminated early and are never pub- that about 50% of presentations at sci- ently not large enough to convince sci- lished.20 entific meetings never result in pub- entists and physicians to abandon new Systematic reviews of randomized lished articles, and these numbers in- trials or to stop using the drugs in prac- trials are considered the highest level of clude clinical trials.19 Efforts to try to tice until the 1990s. One study com- evidence, better than a single RCT. Many contact experts and authors after the fact pleted in 1980 was not published until probably believe that all published RCTs have invariably proved inefficient.25,26 1993.32 If the existence of this study had for systematic reviews are readily avail- The fact that some trial results are never been known earlier and its results gen- able through various large biblio- published would not be a problem except erally available, the use of this danger- graphic databases, notably the Na- that there is good evidence that the results ous although biologically rational drug tional Library of Medicine’s MEDLINE, from unpublished trials are systemati- may have been halted.33 Indeed, there are the Institute for Scientific Informa- cally different from those of published estimates that 20000 to 75000 lives were tion’s Science Citation Index, and Elsevi- trials (publication bias). For decades, sci- lost each year in the 1980s in the United er’s EMBASE. But these databases cap- entists have complained about the over- States alone from inappropriate admin- ture mainly publications from the reporting of “positive” results (results that istration of antiarrhythmic drugs for sec- journals they index, the majority of them favor the new therapy).27 Publication bias ondary prevention of myocardial infarc- in the English language. MEDLINE for is common and is almost always because tion.34 It is not difficult to see why the example contained only 229160 cita- of failure of the investigators to com- failure to publish studies has been called tions to controlled trials at the time of plete, write up, and submit to journals scientific misconduct.35 writing this article (January 2003). findings from trials with negative In baseball, it is easy to find out just A more comprehensive source for results.18,28 Ioannidis29 has shown that how well Cal Ripken has hit against vari- published reports of controlled trials is even when all trials are published, those ous pitchers in the past, at home or away the Cochrane Central Register of Con- with positive results tend to be submit- games, in recent weeks or during his ca- trolled Trials,17 coordinated by the US ted much sooner after completion. Thus, reer.36,37 Yet in medicine, there is no com- Cochrane Center, which currently con- news about new (and perhaps more prehensive source for finding out simi- tains more than 350000 reports of trials. expensive) treatments is disseminated lar, accurate statistics for medical The Cochrane Central Register of Con- faster, and these interventions may be interventions. How can baseball be bet- trolled Trials database has been devel- adopted at the expense of those that are ter organized and keep better records oped as a result of a massive and con- cheaper, just as effective, and safer. than medical science? The Cochrane tinuing effort on the part of the Collaboration is now assembling regu- Cochrane Collaboration to search by Harms Resulting From Trials larly updated systematic reviews of the hand issues from 1948 to the present Disappearing Without Trace results of clinical trials of health care in- of more than 2200 journals to find Patients who agree to participate in terventions. But nowhere is there reli- relevant articles and trials otherwise clinical research do so with the under- able, comprehensive information on all lost to medicine and systematic re- standing that they are contributing to initiated clinical trials that a systematic views.21,22 This time-consuming, inef- medical knowledge. If the knowledge review would bring together to arrive at ficient, and costly effort has shown that gained in a trial is never communi- an estimate of an intervention’s effec- at the very least one third of RCTs cated to others, then their contribu- tiveness and safety. Depending solely on since 1966 have not been indexed in tion is unrealized and the covenant be- the accessible published medical litera- MEDLINE, and thus are effectively lost tween researcher and patient, indeed ture for assessing a treatment’s efficacy to the majority of searchers who limit between ethical review boards and pa- is akin to using only information from their searching in this way.23,24 tients, is broken. Ripken’s home games to calculate his bat- This article focuses on issues re- A crucial question is whether the dis- ting average. lated to registration of controlled clini- tortion of available evidence, aside from The result of not knowing who has cal trials designed to test the efficacy or being unethical, actually harms patients. performed what is loss and distortion of other effects of an intervention, and not There is evidence that it does. Starting the evidence, waste and duplication of clinical trials, such as phase 1 studies, with Furberg’s 1983 systematic review trials, inability of funding agencies to designed primarily to test dosage and of 14 trials of class 1 antiarrhythmic plan, and a chaotic system from which safety. drugs,30 meta-analyses have failed to only certain sponsors might benefit, and

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS which is invariably against the interest reported, one would be able to exam- be inexpensive and simple to contrib- of those who offered to participate in ine a trials register and observe that the ute to for those registering trials. Trial trials and of patients in general. trial had taken place; this in turn could entries would need to include at least lead to finding out more about the tri- the condition under study and contact Trial Registers al’s design and outcome. details for obtaining additional infor- It has been known since the seminal The objective in establishing a com- mation, but trial results could be op- work of Simes in 198638 that basing re- prehensive clinical trials register would tional. The ideal register would also views on the results of trials registered be to make information about all trials need to use a unique identifier for each in advance is likely to produce an es- undertaken publicly available. The ideal trial conducted because trials may have timate of effect free of publication bias. register would be comprehensive, more than 1 site or funder. This re- The idea is that even if a trial were never accurate, and easy to access and it must quirement has been recognized since the 1960s39 and can be compared with Table 1. Selected Events Supporting and Leading to Trials Registration* use of the International Standard Book Examples Number, the unique identification Publications recommending/supporting number assigned internationally to registration books. 2 Letter to the editor Opinion leader Trials registers have been in exis- 40-42 Editorials Journal editors tence at least since the 1960s.39 A Special panels Society for Clinical Trials Panel on Trial Registration43 International Collaborative Group44 searchable computerized interna- Institute of Medicine committees45,46 tional register of therapeutic trials of Special columns in journal Controlled Clinical Trials47 psychopharmacological agents, first op- Special columns in lay magazine MAMM−Women, Cancer and the Community48 erational in 1967, was developed by the Research articles Demonstration of publication bias favoring positive US National Institute of Mental Health. results among published trials compared with registered trials38 Participants in its Early Clinical Drug Demonstration that even the best trials registers are Evaluation Units Program recognized 3 incomplete the problems associated with finding Registration efforts Registration of trial protocols Major biomedical journals49,50 out about unpublished trials and ob- Assignment of unique number to Various registers39,44,51,52 taining an unduplicated count of trials, individual trials and the effect of publishing delays on Publication of unique registration Major biomedical journals53,54 knowledge. Since that time, much has number with manuscript been written about the need for coor- Industry registers Individual registers55 Compilation register (CenterWatch)56 dinated comprehensive trials regis- Government registers NIH: ters, but until now no single effort has NIMH register of therapeutic trials of pharmacologic survived in the long term (TABLE 1). agents (1967-1972)39 57 Registers have been recommended by NIH all trials (1975-1979) 2 ClinicalTrials.gov (2000 to present)52 individuals, presidentially appointed HIV/AIDS trials (1989 to present)58 committees,1 and expert panels such as United Kingdom: National Health Service Research Register59 those appointed by the Institute of Medical Research Council Trials Register60 Medicine.45,46 Biomedical journals have Country-wide registers Registration in Spain through ethics committees61 tried to do their part by publishing trial Collaborative efforts Comprehensive registers of published trials: Oxford protocols,49,50 unique trial registration 62 Database of Perinatal Trials (1986-1993) 51,69,70 Cochrane Central Register of Controlled Trials17 numbers, editorials about the im- Prospective registers of initiated trials: International portance of registration,40,41 and regu- Committee on Thrombosis and Haemostasis63 48 Current Controlled Trials Meta-Register (a compilation lar lists of ongoing trials. There have of more than 20 individual registers)64 been country-wide registration ef- TrialsCentral (an online register of more than 200 forts, for example, in Spain,61 and laws US-based trials registers)65 mandating registers.66-68 Some govern- Legislation mandating registers National Cancer Act of 1971, leading to National Cancer Institute International Cancer Research ment funding agencies, such as the UK Data Bank1 Medical Research Council, have re- Health Omnibus Programs Extension Act of 1988, leading to establishment of HIV/AIDS trials register66 quired registration of trials they sup- Food and Drug Administration Modernization Act of port.24 In the United States, the NIH reg- 1997, leading to establishment of 57 ClinicalTrials.gov67,68 istered its trials from 1975 to 1979 but Abbreviations: HIV, human immunodeficiency virus; NIH, National Institutes of Health; NIMH, National Institute of Men- not again until it was legislatively man- tal Health. dated for trials of serious and life- *The events represent a selected nonrandom sample and are not intended to be all-inclusive. threatening diseases in 1997.67,68

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS

Specialty or Regional Registers Harlan, presented a history and up- cluding ClinicalTrials.gov.64 The Meta- An increasing number of small medi- date of the US ClinicalTrials.gov Web Register is the most comprehensive reg- cal center registers or disease-based reg- site, developed through the National Li- ister in existence and may be the best isters exist in many countries, but even brary of Medicine in collaboration with approach if regional registers are more taken together are by no means com- NIH and the Food and Drug Adminis- practical to achieve than a single inter- prehensive.63,65 TrialsCentral (http: tration (FDA).52 ClinicalTrials.gov was national register. If this route is taken, //www.trialscentral.org) attempts to launched in February 2000 and cur- however, there must be a way to en- bring the 200 or more US-based regis- rently contains more than 5000 stud- sure comprehensiveness of the smaller ters under a single umbrella65 but also ies sponsored by federal agencies and registers, as well as compliance of the serves to demonstrate the underlying the pharmaceutical industry in North registers in submitting their contents disorganization of the clinical trials en- America and about 80 countries. to a meta-register. terprise as a whole. ClinicalTrials.gov was developed Meetings about trial registration were largely as a result of breast cancer con- Barriers first held in Europe in 1991,44 leading 10 sumer lobbying, which led to authoriz- There are several major barriers to years later to pan-European support for ing language in the FDA Moderniza- development of a comprehensive reg- trial registration. In May 2001, the Eu- tion Act of 1997.67,68 This effort was ister of clinical trials: industry resis- ropean Science Foundation advised its modeled on the Health Omnibus Pro- tance, the lack of a funding appropria- member organizations to require regis- grams Extension (HOPE) Act of 1988,66 tion for a serious and sustained effort, tration as a condition of funding trials, which led to a successful human immu- lack of a mechanism for enforcement contribute to the Meta-Register of con- nodeficiency virus/AIDS trials register. of policies, and lack of awareness of the trolled trials (http://www.controlled ClinicalTrials.gov is mandated only importance of the problem. -trials.com), and support the use of an to include trials of serious and life- Industry Resistance. The Pharma- international standard randomized con- threatening diseases, but even in stud- ceutical Research and Manufacturers of trolled trial number.71 Even such broad ies of cancer it is deficient. Manhei- America is explicit that it will not com- support did not guarantee action; how- mer and Anderson3 have shown that mit to publish or to a register of trials. ever, when a funding application for such ClinicalTrials.gov included only 17 of The group’s “Public Disclosure of Clini- a register was made last year to the Eu- 32 known and ongoing prostate and co- cal Trial Results” states, “Sponsors do ropean commission by scientific lead- lon cancer controlled trials that were not commit to publish the results of ev- ers from 11 countries, it was unexpect- funded by industry. Given that indus- ery exploratory study performed, or to edly turned down on the dubious basis try funds more than 60% of clinical make the designs of clinical trial pro- that legislation would first be needed to trials,72,73 the fact that the register is far tocols available publicly at inception, make information about trials publicly from comprehensive when it comes to as in a clinical trials registry.”74 In- available (I. Chalmers, written commu- industry-funded trials is a serious deed, high-level industry executives re- nication, January 2, 2003). defect. jected a proposed “Good Publication The most recent development is a Eu- CenterWatch, a trials register recruit- Practice: Guidelines for Pharmaceuti- ropean Science Foundation–convened ing Web site for the drug industry, may cal Companies,” which provides guide- meeting in Frankfurt, Germany, in No- well be the largest single trials regis- lines that would reduce publication bias vember 2002 to discuss ways to initiate ter, claiming to have between 7500 and and define appropriate publishing re- the process. The European Science 41000 trials listed (information on site lationships between industry and in- Foundation secretary general urged a varies).56 However, much of the infor- vestigators.75 greater sense of openness about trials and mation (eg, funding source) about the There is evidence that many indus- noted that a system for European reg- trials listed on CenterWatch is not di- try trials are never published. For ex- istration is important for ensuring pub- rectly accessible by the user. In addi- ample, in a systematic review of trials lic trust in the biomedical community. tion, there is no unique identifier to al- of nonsteroidal anti-inflammatory Many of those attending the meeting low the user to determine, for example, drugs, MacLean et al76 found that only noted that public trust cannot possibly whether the 12 sites listing trials of Zo- 1 of 37 studies reported in FDA re- be fostered by the existing European meta (zoledronic acid for injection) for views had been published. Because Clinical Trials Database for medicinals bone metastases are 12 unique trials or there is commercial advantage to be because it is confidential and access is a single trial with 12 recruiting sites. gained by early publication of positive limited to the regulatory agency and fun- results and the suppression of nega- ders, and not to those physicians, pa- “Comprehensive” Registers tive results, industry reluctance to pub- tients, and others who most need ac- Current Controlled Trials, established in lish negative findings would not come cess to the information. 1998 by the publisher Biomed Cen- as a surprise. Nor would it be a sur- At the Frankfurt meeting, a repre- tral, is a composite of 26 (as of June 26, prise if pharmaceutical and industry ex- sentative from the NIH, Dr William 2003) registers from 4 continents, in- ecutives in general were opposed to the

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS idea of registration of trials. Reasons for cently, the Association of the British Who should pay for comprehensive avoiding registration given to each of Pharmaceutical Industry has encour- trials registration? Industry and gov- the authors in various public and pri- aged its members to register trials,78,79 ernment bear the largest responsibil- vate meetings over the years include which undermines the arguments ity for this, because they support the protection of information about prod- against registration put forward by in- trials and stand to gain the most, ei- ucts under development, patents, and dustry leaders elsewhere. ther financially or through better sci- information about good recruiting cen- Lack of Funding. In 2001, a re- ence and a healthier public. Funding ters, and not wanting to be bothered by ported $30.3 billion was spent by the issues have also arisen with the imple- dealing with consumers and others who Pharmaceutical Research and Manufac- mentation of new federal guidelines on contact them for information.77 In ad- turers of America–member companies ethical review of human studies, pro- dition, companies may assess many on pharmaceutical research and devel- tecting research from conflicts of in- trials as performed solely to meet regu- opment, and on average $800 million terest and privacy protection; how- latory requirements rather than pro- was spent to develop a single drug.80 In- ever, there has been general agreement viding information patients have a right dustry spends more on pharmaceutical that ethical concerns outweigh those of to know about. research than was provided to NIH efficiency and cost. Given the minimal information (about $20 billion in 2001) for its en- Lack of Mechanisms for Enforce- (TABLE 2) that would be required to be tire operating budget.81 Current Con- ment. The FDA Modernization Act of publicly disclosed in any register and trolled Trials estimates that since its start 199767 may have mandated that clini- the easy availability of information in 1998 it has incurred expenses of cal trials conducted to test effective- about what is in the pipeline,3 it is dif- £500000 (approximately US $830000) ness of an investigational new drug for ficult to understand the reluctance of (I. Chalmers, written communication, a serious or life-threatening disease or industry to register trials. Two compa- May 27, 2002). The savings to society condition are required to be submit- nies, GlaxoWellcome and Schering effected by a comprehensive register ted to ClinicalTrials.gov for registra- Healthcare UK, agreed to register their would exceed many-fold the money that tion, but the law provided neither fund- trials in 1998,55,77 but these efforts have is at present wasted on inaccessible and ing nor a mechanism of enforcement. been extremely circumscribed. More re- unnecessarily duplicated knowledge. Although industry generally wants to cooperate with the FDA because it ap- proves new drug and device applica- Table 2. Trial Details Requested by Current Controlled Trials and ClinicalTrials.gov tions, industry also knows that any ad- Current Clinical verse consequences of failure to comply Details Requested Controlled Trials Trials.gov fully with the mandate are unlikely. Identifying information Name of organization/individual supplying the record X ClinicalTrials.gov (accessed January 24, Name of trial sponsor X X 2003) lists almost 2700 NIH studies, Protocol number given to the trial by the sponsor, if relevant X X 200 other federally funded studies, Trial details more than 1800 university studies, and Trial purpose X only 618 industry studies recruiting pa- Interventions (all interventions and trial groups) X X tients, a large proportion from biotech- Title of trial X X nology companies. Given other avail- Acronym (if relevant) X able data,3,80,82 it seems unlikely that the Disease or condition X X trials listed are a complete count of Participants (eligibility criteria) X X industry trials in serious and life- Phase of trial (1, 2, 3, 4) X threatening diseases. Trial locations X A poster presented at the FDA Sci- Recruitment status X ence Forum on April 24, 2003, con- Date study started X firms that compliance with the law is Funding 83 Sources of funding (all known funding sources XXlow, at least for cancer trials. The au- and reference numbers given to the trial thors found that 366 commercial can- by each funding agency) cer protocols were submitted to the Contact FDA for approval between January 1, Lead principal investigator or person with overall responsibility Name X X 2002, and September 30, 2002, and 187 Full address X (51%) of these met the criteria for list- Telephone X* X ing in ClinicalTrials.gov. However, only Fax X* 115 (61%) of 187 were submitted to the E-mail X* X register: 61 (48%) of 127 pharmaceu- *Does not have to be displayed but must be provided. tical industry protocols, 52 (91%) of 57

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS of NIH/National Cancer Institute pro- cine budget (D. Lipman, written com- make registration a condition of ap- tocols, and 2 (67%) of 3 other proto- munication, May 21, 2003). Its budget proval.87,88 The Institute of Medicine cols. Industry compliance with the law is approximately $5.5 million annu- Committee on Assessing the System for was the same before and after the law ally, which covers biological curators Protecting Human Research Partici- went into effect (May 2, 2002). The au- who handle sequence submissions, pro- pants has recommended that the re- thors concluded, “Participation by the grammers and database developers, sponsibility for ensuring ethical con- pharmaceutical industry was less than hardware, software, and networking ex- duct of research be taken at the highest expected one year after the availabil- penses. Data are made available free of level of the organization conducting ity of a final guidance document de- charge to the international public by the that research.46 The responsibility for spite a federal law, a targeted educa- National Library of Medicine via the In- registering trials also lies there. Insti- tional program, and an easy-to-use ternet. Thus, NIH has coordinated a re- tutions, organizations, and research eth- web-based data entry tool.” Indeed, sponsible effort by investigators, jour- ics review boards also have responsi- FDA officials have indicated that in the nals, industry, and others for creating a bility for ensuring that trials registers future they could take steps to facili- system that helps to ensure research data are actually consulted to avoid unnec- tate or even compel registration.84 are not lost and helps to prevent unnec- essary duplication of research. Lack of Awareness. One of the big- essary duplication of effort. Industry Leaders Must Agree to and gest barriers to comprehensive trials reg- Similar strong coordination of com- Insist on Comprehensive Registra- istration is the relative obscurity of this prehensive trials registration in the tion. In the United States, government- issue in the lay and biomedical press. United States is also required. Provid- funded trials are registered through There are almost no articles in medical ing a means whereby the public can find ClinicalTrials.gov. However, the larg- journals touting the existence of a spe- all clinical trials would be at least as im- est funder of clinical trials is industry cific trials register, even those as large portant as GenBank is now, and its im- and ensuring cooperation of this group as the National Health Service’s Na- portance will increase as the products to register all trials they are support- tional Research Register (UK), Current of research in molecular biology are ing is essential. One approach may be Controlled Trials’ Meta-Register, Physi- tested in humans. The NIH is the logi- for patient advocacy groups, which have cian Data Query, or ClinicalTrials.gov. cal place for coordination of registra- more power than ever before, to de- Because there is no news in reporting tion, given its experience in the area over mand that industry be proactive in com- that trials registers exist, there is no writ- the years and the public responsibili- plying with the registration effort. The ten record of it and scientists get the mes- ties it has been entrusted with for re- individual patient should insist that a sage that their community does not search related to health. Given that an trial is registered before enrolling. think registers are important. initial NIH conference on the topic of Another strategy would be for clinician- comprehensive trials registration was scientists to require that a trial is cen- Our Recommendations held almost a decade ago,86 it is not clear trally registered before enrolling pa- The NIH Should Take Responsibility that further workshops and discussion tients, and research ethics review boards for Ensuring Trial Registration in the about “why” or “whether” are needed. should insist trials are registered and United States. An example of a suc- In our opinion, it is long past time to that this information is in the in- cessful registry system exists in molecu- make it happen. formed consent document before trials lar biology. GenBank is a database that Those at the Highest Level at Insti- are approved. has been in existence for more than tutions and Organizations That Con- Leaders at the highest levels in in- 20 years and serves as the electronic re- duct Research Must Require Registra- dustry55 must actively support and be pository in which investigators contrib- tion of Trials for Which They Are involved in trial registration. Those at ute genetic sequences for more than Responsible. Given that patients are lower levels are unlikely to have rea- 100000 different organisms.85 Data are most at risk for harm from the present son to advance the cause and doing so also contributed to GenBank by the US chaos and bias, and also are most likely is unlikely to advance their career. An- Office of Patents and Trademarks for is- to gain from a working, comprehen- ecdotally, it is difficult to find a single sued patents and are exchanged inter- sive, and easily searchable register, fail- person who considers himself/herself nationally on a daily basis with 2 other ure to register trials should be recog- in charge. For instance, when we have international sites to ensure compre- nized as unethical. The fact that industry asked about local trial tracking mecha- hensiveness. Many journals require sub- could be gaining financially from the nisms, industry representatives claim mission of sequence information to Gen- public involvement in trials yet not re- that within a company there is usually Bank and assignment of an accession ciprocating by making scientific infor- not a centralized repository of infor- number before publication.53 mation publicly available is of particu- mation about which trials are ongo- GenBank’s production is supported lar concern. ing. Thus, they have argued, those in and maintained entirely by the NIH as Institutional review boards (re- charge of knowing about ongoing on- part of the National Library of Medi- search ethics review boards) should cology trials are typically not aware of

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS ongoing allergy trials, which makes it tion of all trials involving human par- Disclaimer: The views expressed in this article are the authors’ own personal views and do not reflect a po- difficult to talk to industry about agree- ticipants, regardless of the topic or fund- sition taken by the Centers, the Cochrane Collabora- ing to registration generally. There is ing source. Perhaps the most direct way tion, or JAMA. Dr Rennie was not involved in the editorial evaluation or editorial decision to accept this work no alternative but for the leadership of to comprehensively register trials is for publication. pharmaceutical companies to agree to through research ethics review boards41; Acknowledgment: We thank Iain Chalmers, DSc, for comprehensive registration. this approach provides a framework for his critical reading of the manuscript and sugges- tions. It is unlikely that registration of broad ensuring comprehensiveness and com- details about clinical trials will infringe pliance of those conducting trials. on the patent rights of manufacturers. In- Congress should start immediately to REFERENCES formation about ongoing trials is al- consider legislation that would imple- 1. Hubbard S, DeVita VT. PDQ: an innovation in information dissemination linking cancer research and ready readily available to that commu- ment such a process. Legislation man- clinical practice. In: DeVita VT, Hellman S, Rosen- nity but not to the public. In 1999, one dating, funding, and enforcing regis- berg SA, eds. Important Advances in Oncology. Phila- delphia, Pa: JB Lippincott Co; 1987. of us wrote about commercial sponsors tration of all clinical trials initiated in 2. Chalmers TC. Randomize the first patient! N Engl and registration of trials, “It is hard to the United States is needed immedi- J Med. 1977;296:107. 3. Manheimer E, Anderson D. Survey of public infor- think of any step a pharmaceutical com- ately. The mandate must carry with it mation about ongoing clinical trials funded by indus- pany could take that would so reassure sufficient funding for start-up and con- try: evaluation of completeness and accessibility. BMJ. its clients as to its ethical, clinical, and tinuation, as well as assignment of the 2002;325:528-531. 8 4. Gøtzsche PC. Multiple publication of reports of drug scientific good intentions.” In an era authority to monitor the process and to trials. Eur J Clin Pharmacol. 1989;36:429-432. when mistrust of pharmaceutical com- take action against those not comply- 5. Huston P, Moher D. Redundancy, disaggrega- tion, and the integrity of medical research. Lancet. panies is increasing, that statement is ing with the law. Based on previous ex- 1996;347:1024-1026. even truer. It will require the efforts of perience, lobbying by patients’ groups, 6. Trame` r MR, Reynolds DJM, Moore RA, McQuay HJ. Impact of covert duplicate publication on meta- the entire clinical research community clinical researchers, and physicians to analysis: a case study. BMJ. 1997;315:635-640. for registration to become routine. produce this type of legislation ap- 7. Johansen HK, Gøtzsche PC. Problems in the de- Journal Editors Should Require pears to be the best way forward, at least sign and reporting of trials of antifungal agents en- countered during meta-analysis. JAMA. 1999;282: Unique Registration Numbers for Trial in countries where such practices are 1752-1759. Reports. Some major medical jour- used. 8. Rennie D. Fair conduct and fair reporting of clinical trials. JAMA. 1999;282:1766-1768. nals have been actively advocating for 9. Flanagin A, Fontanarosa PB, DeAngelis CD. Au- trial registration, but efforts must be Conclusions thorship for research groups. JAMA. 2002;288:3166- 3168. broader and stronger. At the Euro- The public and private sectors must end 10. Dickersin K, Scherer R, Suci E, Gil-Montero M. pean Science Foundation meeting in the inertia and take responsibility to Problems with indexing and citation of articles with group authorship. JAMA. 2002;287:2772-2774. Frankfurt, Fiona Godlee, instrumen- overcome the problem of wasted and un- 11. Chalmers I. Unbiased, relevant, and reliable as- tal in Current Controlled Trials’ devel- necessarily duplicated research. At least sessments in health care. BMJ. 1998;317:1167- opment of the Meta-Register, outlined initially, it will probably be necessary to 1168. 12. Grady D, Herrington D, Bittner V, et al, for the several specific action areas for jour- approach comprehensive trial registra- HERS Research Group. Cardiovascular disease out- nals related to trial registration. Jour- tion through the merging of several reg- comes during 6.8 years of hormone therapy. JAMA. 2002;288:49-57. nals should encourage a culture of isters (eg, country-wide registers). Cur- 13. Hulley S, Furberg C, Barrett-Connor E, et al. Non- transparency in research and report- rent Controlled Trials’ Meta-Register is cardiovascular disease outcomes during 6.8 years of hormone therapy: Heart and Estrogen/progestin Re- ing, publish study protocols, publish an example of the type of compilation placement Study follow-up (HERS II). JAMA. 2002; negative results, and strongly pro- that can work with international col- 288:58-66. mote trial registration. Perhaps most laboration. The United States can start 14. Writing Group for the Women’s Health Initia- tive Investigators. Risks and benefits of estrogen important, journals should make pro- by funding its own comprehensive reg- plus progestin in healthy postmenopausal women: vision of a unique identifier, such as the ister of initiated trials. Investigators, re- principal results from the Women’s Health Initiative randomized controlled trial. JAMA. 2002;288:321- international standard randomized con- search organizations and institutions, re- 333. trolled trial number assigned by Cur- search sponsors, industry leaders, 15. Petitti DB. Hormone replacement therapy for prevention: more evidence, more pessimism. JAMA. 2002; rent Controlled Trials, a condition of ac- journal editors, lawmakers, and con- 288:99-101. ceptance of an RCT. sumers—it’s time for action. 16. The Medical Research Council. Streptomycin treatment of pulmonary tuberculosis. BMJ. 1948;2:769- Lawmakers Must Protect the Pub- Author Contributions: Both authors take full respon- 782. lic by Requiring Comprehensive Trial sibility for the content of this article and had an equal 17. Dickersin K, Manheimer E, Wieland S, Robinson Registration Through Ethics Commit- role in the concept, the writing, and reviewing of suc- K, Lefebvre C, McDonald S. Development of the Coch- cessive drafts, including the final draft. Study con- rane Collaboration’s CENTRAL register of controlled tees. These actions might not be suc- cept and design, drafting of the manuscript, and criti- clinical trials. Eval Health Prof. 2002;25:38-64. cessful; therefore, the only recourse may cal revision of the manuscript for important intellectual 18. Dickersin K. How important is publication bias? content: Dickersin, Rennie. a synthesis of available data. AIDS Educ Prev. 1997; be to change the law. The public should Funding/Support: No funding was received for prepa- 9(suppl A):15-21. be firm in its insistence that the govern- ration of the manuscript. Dr Dickersin was reim- 19. Scherer RW, Langenberg P. Full publication of re- bursed by the European Science Foundation for her sults initially presented in abstracts [Cochrane Meth- ment has a moral responsibility to fund, travel expenses to Frankfurt, Germany, for the Euro- odology Review]. Oxford, England: Cochrane Li- require, and enforce public registra- pean Science Foundation meeting in November 2002. brary, Update Software; 2003; issue 1.

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006 REGISTERING CLINICAL TRIALS

20. Pich J, Carne X, Arnaiz JA, Gomez B, Trilla A, Rodes 40. Horton R, Smith R. Time to register randomised tunity to help improve the quality of clinical research. J. Role of a research ethics committee in follow-up and trials: the case is now unanswerable. BMJ. 1999;319: Curr Control Trials Cardiovasc Med. 2000;1:3-8. publication of results. Lancet. 2003;361:1015-1016. 865-866. 65. Anderson D, Costa I, Dickersin K. Building 21. Hayward S, Brunton G, Thomas K, Ciliska D. 41. Horton R, Smith R. Time to register randomised TrialsCentral, an online register of clinical trials regis- Searching for the evidence: source, time and yield. In: trials. Lancet. 1999;354:1138-1139. ters. Control Clin Trials. 2001;22(suppl 2):40S. Proceedings of the Second International Conference, 42. Meinert CL. Toward prospective registration of 66. Pub L No. 100-607 (1988). Scientific Basis of Health Services and the 5th Annual clinical trials. Control Clin Trials. 1988;9:1-5. 67. Pub L No. 105-115, §113 Stat 2296 (1997). Cochrane Colloquium; October 5-12, 1997; Amster- 43. Dickersin K. Report from the panel on the case 68. DeMarinis AD. Current considerations: regula- dam, the Netherlands. for registers of clinical trials at the Eighth Annual Meet- tory: clinical trials databank. Res Practitioner. 2002; 22. Brazier H. Poorly executed and inadequately docu- ing of the Society for Clinical Trials. Control Clin Trials. 3:105-106. mented? an analysis of the literature searches on which 1988;9:76-81. 69. Rowland D, Di Guiseppi C, Roberts I, et al. Preva- systemic reviews are based. In: Proceedings of the Sec- 44. Ad Hoc Working Party of the International Col- lence of working smoke alarms in local authority in- ond Symposium on Systematic Reviews: Beyond the laborative Group on Clinical Trials Registries. Posi- ner city housing: randomized clinical trial. BMJ. 2002; Basics; January 5-7, 1999; Oxford, England. tion paper and consensus recommendations on 325:998-1001. 23. Hopewell S, Clarke M, Lusher A, Lefebvre C, clinical trial registries. Clin Trials Metaanal. 1993;28: 70. Information for authors. Ann Intern Med. 1995; Westby M. A comparison of handsearching versus 199-201. 123:I-10. MEDLINE searching to identify reports of random- 45. Institute of Medicine Committee on Routine Pa- 71. European Science Foundation Policy Briefing No. ized controlled trials. Stat Med. 2002;21:1625-1634. tient Care Costs in Clinical Trials for Medicare Ben- 13, May 2001. Available at: http://www.esf.org 24. Jull A, Chalmers I, Rodgers A. Clinical trials in NZ: eficiaries. Extending Medicare Reimbursement in Clini- /publication/90/ESPB13.pdf. Accessibility verified June does anybody know what’s going on? N Z Med J. cal Trials. Washington, DC: National Academy Press; 27, 2003. 2002;115:U269. 2000. 72. Moses H 3rd, Martin JB. Academic relationships 25. Hetherington J, Dickersin K, Chalmers I, Meinert 46. Institute of Medicine Committee on Assessing the with industry: a new model for biomedical research. CL. Retrospective and prospective identification of un- System for Protecting Human Research Participants. JAMA. 2001;285:933-935. published controlled trials: lessons from a survey of Responsible Research: A Systems Approach to Pro- 73. Bekelman JE, Li Y, Gross CP. Scope and impact of obstetricians and pediatricians. Pediatrics. 1989;84: tecting Research Participants. Washington, DC: Na- financial conflicts of interest in biomedical research: a 374-380. tional Academy Press; 2002. systematic review. JAMA. 2003;289:454-465. 26. McManus RJ, Wilson S, Delaney BC, et al. Re- 47. Dickersin K. Why register clinical trials? revis- 74. PhRMA. Public Disclosure of Clinical Trial Re- view of the usefulness of contacting other experts when ited. Control Clin Trials. 1992;13:170-177. sults. June 24, 2002. Available at: http://www.phrma conducting a literature search for systematic reviews. 48. Schiff H, Haran C. On the trials trail: clinical trial .org/publications/policy//2002-06-24.430.pdf. Ac- BMJ. 1998;317:1562-1563. watch. MAMM. October/November. 1998:86. cessed May 21, 2003. 27. Sterling TD. Publication decisions and their pos- 49. Horton R. Pardonable revisions and protocol re- 75. Wager E, Field EA, Grossman L. Good publica- sible effects on inferences drawn from tests of signifi- views. Lancet. 1997;349:6. tion practice for pharmaceutical companies. Curr Med cance: or vice versa. J Am Stat Assoc. 1959;54:30-34. 50. BioMed Central. Information for authors: pub- Res Opin. 2003;19:149-154. 28. Dickersin K, Chan S, Chalmers TC, Sacks HS, Smith lish your study protocols. Available at: http://www 76. MacLean CH, Morton SC, Ofman JJ, et al, for the HJR. Publication bias and clinical trials. Control Clin .biomedcentral.com/info/authors/protocols. Ac- Southern California Evidence-Based Practice Center. Trials. 1987;8:343-353. cessed January 24, 2003. How useful are unpublished data from the Food and 29. Ioannidis JP. Effect of the statistical significance 51. Vale CL, Stewart LA. The International Standard Drug Administration in meta-analysis? J Clin Epide- of results on the time to completion and publication Controlled Trial Number (ISRCTN): what does it mean miol. 2003;56:44-51. of randomized efficacy trials. JAMA. 1998;279:281- for reviewers? In: Proceedings of the 9th Interna- 77. Gibbs TG, Wager E. Clinical trial registration: the 286. tional Cochrane Colloquium; 2001; Lyon, France. Glaxo Wellcome experience. Int J Pharm Med. 2000; 30. Furberg CD. Effect of antiarrhythmic drugs on mor- 52. McCray AT. Better access to information about 14:203-205. tality after myocardial infarction. Am J Cardiol. 1983; clinical trials. Ann Intern Med. 2001;133:609-614. 78. Kmietowicz Z. UK drugs industry sets up trials reg- 52:32C-36C. 53. Information for Authors. Writing for the Lancet. ister. BMJ. 2000;321:850. 31. Teo KK, Yusuf S, Furberg CD. Effects of pro- Available at: http://thelancet.com.authorinfo. Ac- 79. Tonks A. A clinical trials register for Europe. BMJ. phylactic antiarrhythmic drug therapy in acute cessed January 24, 2003. 2002;325:1314-1315. myocardial infarction: an overview of results from 54. Information for Authors. Manuscript format and 80. Pharmaceutical Research and Manufacturers of randomized controlled trials. JAMA. 1993;270:1589- style: trial registry information. Ann Intern Med. Avail- America. 2002 Industry Profile: Research and Devel- 1595. able at: http://www.annals.org/shared/manu_format opment—The Key to Innovation. Washington, DC: 32. Cowley AJ, Skene A, Stainer K, Hampton JR. .html. Accessed January 24, 2003. PhRMA; 2002. The effect of lorcainide on arrhythmias and survival 55. Sykes R. Being a modern pharmaceutical com- 81. Pharmaceutical Research and Manufacturers of in patients with acute myocardial infarction: an pany: involves making information available on clini- America. 2003 Industry Profile: Dramatic Growth of example of publication bias. Int J Cardiol. 1993;40: cal trial programmes. BMJ. 1998;317:1172. Research and Development. Available at: http://www 161-166. 56. CenterWatch. Clinical Trials Listing Service. Avail- .phrma.org/publications/profile02/2003%20chapter 33. Chalmers I. Using systematic reviews and regis- able at: http://www.centerwatch.com. Accessed May %202.pdf. Accessed June 4, 2003. ters of ongoing trials for scientific and ethical trials de- 20, 2003. 82. Research! America. How much is really spent on sign, monitoring, and reporting. In: Egger M, Smith 57. Dickersin K, Min YI. NIH clinical trials and publi- medical and health research? Available at: http://www GD, Altman DG, eds. Systematic Reviews in Health cation bias. Online J Curr Clin Trials. 1993;Doc No. .reseacrhamerica.org/members/4.6_cents.html. Ac- Care: Meta-analysis in Context. London, England: BMJ 50. cessed October 17, 2001. Books; 2001:429-443. 58. AIDSinfo. Available at: http://www.aidsinfo.nih 83. Derbis J, Toigo T, Woods J, Evelyn B, Banks D. 34. Moore T. Deadly Medicine: Why Tens of Thou- .gov. Accessed January 28, 2003. FDAMA section 113: information program on clini- sands of Heart Patients Died in America’s Worst Drug 59. The UK National Health Service National Re- cal trials for serious and life-threatening diseases. Poster Disaster. New York, NY: Simon & Schuster; 1995. search Register (NRR). Available at: http://www.doh presented at: 9th Annual FDA Science Forum; April 35. Chalmers I. Underreporting research is scientific .gov.uk/research. Accessed January 30, 2003. 24, 2003; Washington, DC. misconduct. JAMA. 1990;263:1405-1408. 60. MRC Trials Register. Available at: http://www 84. Reynolds T. Researchers push for publication, reg- 36. The Macmillan Publishing Staff. The Baseball En- .controlledtrials.com. Accessed February 5, 2003. istration of all clinical trials. J Natl Cancer Inst. 2003; cyclopedia: The Complete and Definitive Record of 61. Dickersin K, Garcia-Lopez F. Regulatory process 95:772-774. Major League Baseball. 10th ed. New York, NY: effects clinical trial registration in Spain. Control Clin 85. Benson DA, Karsch-Mizrachi I, Lipman DJ, et al. MacMillan; 1996. Trials. 1992;13:507-512. GenBank. Nucleic Acids Res. 2002;30:17-20. 37. Thorn J, Palmer P, Gershman M, Silverman M, Lah- 62. Chalmers I, Hetherington J, Newdick M, et al. The 86. Harlan WR. Creating an NIH Clinical Trials Reg- man S, Spira G, eds. Total Baseball: The Official En- Oxford Database of Perinatal Trials: developing a registry: a user-friendly approach to health care. JAMA. cyclopedia of Major League Baseball. 7th ed. King- ister of published reports of controlled trials. Control 1994;271:1729. ston, NY: Total Sports Publishing; 2001. Clin Trials. 1986;7:306-324. 87. Savulescu J, Chalmers I, Blunt J. Are research eth- 38. Simes RJ. The case for an international registry of 63. Verstraete M. International Committee on Throm- ics committees behaving unethically? some sugges- clinical trials. J Clin Oncol. 1986;4:1529-1541. bosis and Haemostasis Basil, 8-11: September 1974 tions for improving performance and accountability. 39. Levine J, Guy W, Cleary P. Therapeutic trials of subcommittee on clinical investigations: registry of pro- BMJ. 1996;313:1390-1393. psychopharmacologic agents: 1968-1972. In: spective clinical trials. Thromb Diath Haemorrh. 1975; 88. Blunt J. Research ethics committees must ensure McMahon G, ed. Psychopharmacological Agents. Ar- 33:655-663. registration of research and accessibility of results. BMJ. monk, NY: Futura Publishing Co; 1974. 64. Chalmers I. Current Controlled Trials: an oppor- 1998;316:312.

Downloaded from www.jama.com at Johns Hopkins University, on March 27, 2006