Duke: Industry-sponsored studies may be trouble

Few agreements met standards

Academic medical centers frequently engage in industry-sponsored research that does not adhere to basic standards needed to protect the independence and objectivity of the investigators and the interests of patients who consent to be subjects, a study by researchers at Duke University in Durham, NC, has found.

A survey of research agreements entered into by academic medical centers and sponsors between November 2001 and January 2002 indicated that few agreements meet the standards established by the International Committee of Medical Journal Editors (ICMJE).

The Duke researchers surveyed 108 member institutions of the Association of American Medical Colleges (AAMC) to determine the extent with which their contracts with clinical trial sponsors met the three main provisions of the ICMJE guidelines:

  • whether authors of reports of multicenter trials had access to all trial data;
  • whether the authors or an independent committee controlled editorial and publication;
  • whether investigators were fully involved in the design and conduct of the trials.

The researchers found that the agreements rarely included such provisions. Only 1% required that authors of multicenter studies have independent access to all data from the trial, though institutions at coordinating centers for multicenter trials stipulated this level of access at a higher rate (50% of contracts).

The industry sponsors also typically control decisions regarding the publication of data from clinical studies. Provisions addressing the editorial content of the multicenter manuscripts appeared in 40% of contracts, while independent committees that would guarantee publication were specified in less than 1%. The agreements also did not ensure dissemination of findings from research on stored biological materials collected during clinical trials — these materials are collected by the industry sponsors for use in future research.

"These contracts are important because the truth of the matter is research is governed by the contracts between the sponsors and institutions, not by the study protocols approved by the IRB [institutional review board]," says Kevin Schulman, MD, lead author of the study and professor of medicine at Duke University Medical Center.

The research agreements are the only documents that legally bind the industry sponsor in terms of their conduct related to the research, Schulman explains. "Items like trial governance and publication of results are contained in the study protocol, but rarely in the research agreement, and the sponsor is not legally bound to follow the study protocol. Separately, the research agreement is named as the superseding document in the event of a disagreement between researchers and sponsors in most cases."

For example, a pharmaceutical company many initiate a multicenter trial of a new drug and sign research agreements with a number of different sites. These individual sites recruit participants into the trial and submit data to a central source. Study protocols are approved by each site’s IRB, says Schulman. But those protocols do not guarantee that data from all sites will be published, researchers writing up the study results for publication will have access to all data or that the sponsoring company will not simply just decide to pull funding if the initial results of the trial do not prove favorable.

"The agreements generally allow access to site-level data, but that is scientifically invalid in a multicenter trial," Schulman explains. "If I have 50 patients at Duke in a 10,000-person trial, what does it mean if I try to analyze the data on those 50 people? It doesn’t mean anything."

Guaranteed publication necessary

The most troubling aspect of the survey’s findings was the lack of assurance that all study results would be reported, Schulman says.

A basic tenet of clinical research is humans participate to yield information that will help advance medical science and benefit society. A key facet is that results of the research will be made public — good or bad.

"If you cannot guarantee that results of a research trial will be published, then how can you ethically tell a person that what they are participating in is research?" Schulman asks. "If a company comes in and sponsors a study, the IRB signs off on the protocol, you recruit patients at that site and tell them they are participating in a research trial, but later the study doesn’t go the way the company hoped it to, and the results are never published? What are you recruiting people to at the site level?"

Conducting experiments involving humans, telling them it is a research study, with no guarantee the results will ever be shared with others is not research, he emphasizes.

"If clinicians told participants up front that they were being asked to participate in a marketing study for a private company and that they would be exposed to a potentially toxic medication for the purpose of the manufacturer finding out whether the medication was effective and they could make money at their discretion, we would probably get a lot less participation," Schulman notes.

Not only do many research agreements not guarantee the objectivity of the research, but some investigators have been severely pressured to discard or not reveal study results that were not favorable to the sponsoring company, says David G. Nathan, MD, president emeritus of the Dana-Farber Cancer Institute in Boston.

In the same issue of the New England Medical Journal that features the Duke report, Nathan co-wrote a report on the case of Nancy Oliveri, a Canadian clinical researcher at the Hospital for Sick Children in Toronto.

Oliveri was the lead researcher on a study of the iron chelator, deferiprone. The study was sponsored by Apotex, a Canadian generic drug manufacturer.

As part of the research contract, Oliveri and colleagues signed a confidentiality agreement that was consistent with the policy of the hospital at the time.

However, Oliveri then became concerned about the drug’s potential liver toxicity. She conducted a second study comparing hepatic iron levels in patients taking a similar drug and the test drug. Upon finding that certain patients did not do as well on deferiprone, she decided to submit the findings for publication and present the new data at a scientific meeting.

Apotex subsequently sued Oliveri for violation of the initial confidentiality agreement and the hospital administration referred her for research misconduct to the province’s medical licensing board. The board later found no basis for the misconduct allegations, and she was not disciplined.

Though the case is an egregious example of what can happen, it does illustrate the importance of protecting the independence and integrity of clinical researchers, Nathan says.

"This is a huge problem for the American Association of Medical Colleges," he tells Medical Ethics Advisor. "The deans of the schools must get in agreement."

It is not only the academic medical centers that must pay attention, however, adds Schulman.

Many multicenter studies are conducted with sites that are not teaching hospitals. Frequently, patients may be recruited at smaller hospitals and even in individual physician’s offices.

"There really needs to be one set of research ethics, regardless of where and how the participants were recruited," he says.

Potential solutions

To ensure the integrity of industry-sponsored research, Schulman believes a standard research agreement should be used whenever a medical center or other site contracts with a company to participate in a research trial.

The contract would require publication of the study results, the establishment of an independent board to monitor the progress of the study and would require that the person or group writing the report have access to all site data.

"Everyone doesn’t want or need access to all of the data," he explains. "But if it is a 200-site survey all of the investigators need to know that there is a process in place whereby someone they respect, with access to statisticians and someone who can deal with the numbers, is going to head up the effort to write the report. They need to know that the overall results are going to be published before they recruit their patients into a trial."

Would committee resolve problem?

An independent coordinating committee also would protect the study from undue influence from the company’s corporate objectives, he adds.

"In terms of trial governance issues, what if you’ve exposed all of these people to this medication — particularly in a Phase Four setting — and the sponsor realizes the results aren’t going their way and wants to pull the plug. If you had an independent committee, at least you’d have one check on that," he says. "If you decide a trial isn’t working because you can’t recruit enough people, that is one thing. We make mistakes sometimes in study design and we can’t get the subjects we need. But you would want someone who doesn’t have that outside interest to represent the patients."

Nathan agrees with Schulman’s proposals, but also feels that there should also be some standing authority in place to hear and resolve conflicts between industry sponsors and the institutions or researchers they contract with.

"We suggest in our article that a standing committee of the National Academy of Sciences-Institute of Medicine might be established to consider such conflicts and make recommendations to resolve them," he says.

For example, if a corporate sponsor and study author differ in their interpretation of results, an independent body should be available to resolve the dispute.

"It is most important to interpret the results independently," he says. "If the investigator and company do not agree on the interpretation, there must be away to adjudicate the problem."

It is also important that clinicians recruiting patients into industry-sponsored trials make themselves aware of the need for these protections and that they ensure the integrity of the project, says Schulman.

The center’s institutional review board should also bear some responsibility, he adds.

"IRBs are signing off on this research’ that is not research," he says. "There is still the question if you consent the person to participate in research, and that is not what in fact the project is, what have you done? Some courts have held that a consent agreement is a form of contract. If the contract says you are participating in research, when the institutions or investigators involved have no assurance that’s the case, then you have a hard time fulfilling your end of the contract."


  • David G. Nathan, Dana-Farber Cancer Institute, Office of the President, DFCI D1644, 44 Binney St., Boston, MA 02115.
  • Kevin Schulman, Duke University Medical Center, Duke University Hospital, Erwin Road, Durham, NC 27710.