Med journals: make clinical trial data public

via MedScape Dermatology

Call for Mandatory Clinical Trial Registration, Open Access to Results
Laurie Barclay, MD

              Sept. 14, 2004 — The International Committee of Medical Journal Editors (ICMJE) has mandated registration of all clinical trials, and other groups, such as the Pharmaceutical Research and Manufacturers of America (PhRMA), are requiring that both negative and positive results from pharmaceutical trials be reported in detail and made publicly available. Manufacturers of medical devices would also be affected by requirements for trial registration.

Without access to all such data, the public and healthcare providers are forced to rely on drug companies’ interpretations, along with the potential bias inherent in presenting their product in the most favorable light, some proponents of the new measures say.

“We know that some clinical trials, especially those with results unfavorable to drug companies, are never published,” Gregory D. Curfman, MD, executive editor of the New England Journal of Medicine (NEJM), told Medscape. “This means we hear the good news, but not the bad news. This is not in the best interest of patients.”

According to the ICMJE statement, which was published in the Sept. 16 issue of the NEJM, patients who volunteer for drug trials are motivated by altruism, hoping that their participation will increase medical knowledge and thereby contribute to improved health for others. They have a right to trust that investigators will not only minimize risks to the participants but also will conduct research ethically and report it completely and honestly. A crucial first step, best served by a registry, is revealing the existence of all clinical trials, even those with unfavorable marketing and financial implications for the sponsoring company’s drug.

“Citizens and their doctors have a right to know all information about clinical trials, not just information that is favorable to the company’s bottom line,” Dr. Curfman said. “For example, if there are adverse effects of drugs, people need to know about them. Otherwise, the information that we get is biased in a positive direction and may not reflect the truth.”

The drug companies are not the only ones to blame for selective reporting and publication, according to the ICMJE statement. Journal editors also tend to favor positive trials that show a large treatment effect of a new drug or noninferiority trials that show a new drug offers equivalent results to an accepted treatment. Negative or inconclusive trials are typically much less likely to be selected for publication.

For drugs or other interventions likely to enter mainstream clinical practice, availability of all trial results is especially important so that practitioners can consider the entire body of scientific evidence before changing their prescribing patterns, and so that policymakers determining insurance coverage can make well-informed decisions.

Before a given drug is accepted as effective, 10 or more trials may be needed, according to Torben V. Schroeder, MD, DMSc, editor of the Journal of the Danish Medical Association and a professor of vascular surgery at Rigshospitalet in Copenhagen, Denmark.

“It is therefore crucial that those people who sum up available evidence have access to all existing data,” Dr. Schroeder told Medscape. “You can imagine that the perception may get skewed if all trials that end up with negative results are kept secret, and only the positive trials are at hand. This will leave the impression that the drug is effective, although the facts may be that five trials ended up being positive and five ended up being negative, and the truth in fact is that the drug does not work.”

The ICMJE editors call for registration of all trials in a public repository at the trials’ inception, so that the existence of every trial is part of the public record. Currently, trial registration is largely voluntary, registries contain only a small percentage of trials, and registry data sets and public access to them vary.

Before considering publication of results from any clinical trial starting enrollment after July 1, 2005, the ICMJE member journals will require registration in a public trials registry at or before the onset of patient enrollment. Trials beginning enrollment before this date must register before Sept. 13, 2005, to be considered for publication.

“This is not the same as making the results available, but it is a step in the right direction,” Dr. Schroeder said. “If the public is aware that a trial has been performed it is easier to go after the drug company, and more difficult for them to continue concealing the results.”

This policy, which the ICMJE hopes will be adopted by editors of other medical journals, applies to clinical trials prospectively assigning human subjects to intervention or comparison groups to study the cause-and-effect relationship between a medical intervention and a health outcome. Studies of pharmacokinetics or of major toxicity, such as phase 1 trials, would be exempt.

Recommended criteria for a drug trials registry include free public access, acceptance of all prospective registrants, management by a not-for-profit organization, a viable mechanism to ensure validity of the registration data, and electronic searchability. At minimum, data should include a unique identifying number, description of the intervention(s) and/or comparison(s) studied, hypothesis, primary and secondary outcomes, eligibility criteria, key trial dates, target number of participants, funding source, and contact information.

Although the ICMJE does not recommend one particular registry, they suggest that only, sponsored by the U.S. National Library of Medicine (NLM), currently meets all of these requirements.

Editors who signed the ICMJE editorial hailed from JAMA, NEJM, The New Zealand Medical Journal, the Norwegian Medical Journal, the Canadian Medical Association Journal, The Lancet, the Annals of Internal Medicine, the Croatian Medical Journal, Nederlands Tijdschrift voor Geneeskunde (Dutch Journal of Medicine), the Journal of the Danish Medical Association, the Annals of Internal Medicine, and The Medical Journal of Australia.

The BMJ, conspicuous in its absence from journals participating in the ICMJE editorial, does not endorse recommendation of the NLM registry above others, according to acting editor Kamran Abbasi. He states that BMJ policy, which will also require trial registration, will be discussed in an editorial in the next issue.

Registration is only a first, albeit crucial step, the ICMJE maintains, toward complete transparency regarding performance and reporting of clinical trials. Although drug companies may argue that public registration of clinical trials could delay research and blunt their competitive edge by allowing competitors knowledge of their research plans, the ICMJE counters that greater public confidence in pharmaceutical research will compensate for the costs of full disclosure.

“I am sure [drug companies] are reluctant to reveal results, as profit/survival depends on positive results, but it is not acceptable that, after a certain time, results are not made public,” Dr. Schroeder said. “It is an offense towards all those people who volunteered to participate in trials they will never benefit from. The knowledge gained here must be available for future patients.”

According to an article in the Sept. 11 BMJ, GlaxoSmithKline (GSK) is the first pharmaceutical company to set up a comprehensive registry of its drug trials. On Aug. 25, GSK settled for $2.5 million a lawsuit brought by the New York State Attorney General alleging that GSK suppressed data from at least four trials of the antidepressant paroxetine (Paxil) in adolescents, which showed either no benefit over placebo or a slight increase in self-injurious behavior.

A letter in the same issue of the BMJ by Toshi A. Furukawa, MD, PhD, a professor of psychiatry at Nagoya City University Medical School in Japan, analyzed these newly released data from paroxetine trials in childhood depression, and he concluded that GSK mislabeled suicidal tendencies as “emotional lability.”

“Clinical decisions should be based on three factors: patient’s values and preferences, risks and benefits of relevant interventions, and local and global situational considerations,” Dr. Furukawa told Medscape. “Risks and benefits of interventions can be judged only after considering all high-quality evidence, ie, all clinical trials. The latter therefore need to be made publicly accessible.”

Dr. Furukawa found that GSK did not include two cases of suicidal tendencies in the “emotional lability” group. A 14-year-old boy who “punched pictures, broke glass, and sustained lacerations that required six sutures,” “expressed hopelessness and possible suicide thoughts,” and was hospitalized was categorized as a case of aggression rather than of emotional lability. An 11-year-old boy who “threatened to harm himself and was hospitalized with an acute exacerbation” was categorized as a case of exacerbated depression but not of emotional lability.

When all these cases were included, Dr. Furukawa calculated that the pooled odds ratio for suicidality with paroxetine was 2.77 (95% confidence interval, 1.03 - 7.41).

“Unless [pharmaceutical companies make the results of their trials available to the public], they can mislead the physicians and patients,” Dr. Furukawa said. “Drug companies are obliged to make their findings public for an even stronger ethical reason. Their trials are based on the goodwill of trial participants, who believe that they are contributing not only to the drug companies’ profits, but mainly [to the benefit of] future sufferers like them.”

On Sept. 3, the National Institutes of Health (NIH) also proposed free public access to manuscripts and supplemental data from research studies it sponsors, with availability through PubMed Central six months after journal publication. After a 60-day period for public comment, this process is likely to take effect early next year.

“I do not think such a practice would undermine today’s journal publishing business,” Dr. Furukawa said. “An article is typically four to six pages long and nicely summarizes the research findings, while details of research could be 100 pages long and/or enumerations of numbers. Readers would usually prefer the first.”

The NEJM now makes its entire research archive freely available online six months after articles are published, according to Dr. Curfman. “We applaud the initiative proposed by Dr. Zerhouni at NIH, and we have been out front on this important issue,” he said.

On Sept. 7, PhRMA announced the development of a database that would contain the results of hypothesis-tested clinical trials. This database will be online Oct. 1, according to PhRMA spokesperson Court Rosen.

"This was the culmination of a three-year process to make results, both positive and negative, more transparent and available to physicians, Rosen told Medscape. “Physicians should refer to the FDA-approved drug label as a first source. The database is supplementary information if a physician wishes to learn more about the results from clinical trials.”

In a commentary accompanying the ICMJE editorial in the Sept. 15 issue of JAMA, deputy editor Drummond Rennie, MD, notes that the need for a comprehensive tracking system for reporting results is long overdue. This is especially true because the U.S. Food and Drug Administration is not currently required to report trial findings of negative results or adverse drug reactions to the public. Because of inherent conflicts of interest, he doubts that the pharmaceutical industry will ever be able to set up and maintain a complete, useful, trustworthy, and easily accessible registry.

However, Dr. Curfman believes that drug trial registration and availability of all results will have a “very significant” effect both on physician prescribing patterns and on pharmaceutical research.

“Pharmaceutical research will become much more transparent, both to physicians and the public at large,” Dr. Curfman said. “Physicians will have unbiased information about drug efficacy and safety, [including] side effects, which will lead to better prescribing patterns.”

N Engl J Med. 2004;351:1250-1251
BMJ. 2004;329:626
JAMA. 2004;292:1359-1362, 1363-1364

Reviewed by Gary D. Vogin, MD