Is There a Blind Spot in the Oversight of Human Subject Research?

HumanResearch

A scientist examining samples. (© lily/Fotolia)

Human experimentation has come a long way since congressional hearings in the 1970s exposed patterns of abuse. Where yesterday’s patients were protected only by the good conscience of physician-researchers, today’s patients are spirited past hazards through an elaborate system of oversight and informed consent. Yet in many ways, the project of grounding human research on ethical foundations remains incomplete.

As human research has become a mainstay of career and commercial advancement among academics, research centers, and industry, new threats to research integrity have emerged.

To be sure, much of the medical research we do meets exceedingly high standards. Progress in cancer immunotherapy, or infectious disease, reflects the best of what can be accomplished when medical scientists and patients collaborate productively. And abuses of the earlier part of the 20th century–like those perpetrated by the U.S. Public Health Service in Guatemala–are for the history books.

Yet as human research has become a mainstay of career and commercial advancement among academics, research centers, and industry, new threats to research integrity have emerged. Many flourish in the blind spot of current oversight systems.

Take, for example, the tendency to publish only “positive” findings (“publication bias”). When patients participate in studies, they are told that their contributions will promote medical discovery. That can’t happen if results of experiments never get beyond the hard drives of researchers. While researchers are often eager to publish trials showing a drug works, according to a study my own team conducted, fewer than 4 in 10 trials of drugs that never receive FDA approval get published. This tendency- which occurs in academia as well as industry- deprives other scientists of opportunities to build on these failures and make good on the sacrifice of patients. It also means the trials may be inadvertently repeated by other researchers, subjecting more patients to risks.

On the other hand, many clinical trials test treatments that have already been proven effective beyond a shadow of doubt. Consider the drug aprotinin, used for the management of bleeding during surgery. An analysis in 2005 showed that, not long after the drug was proven effective, researchers launched dozens of additional placebo-controlled trials.  These redundant trials are far in excess of what regulators required for drug approval, and deprived patients in placebo arms of a proven effective therapy. Whether because of an oversight or deliberately (does it matter?), researchers conducting these trials often failed in publications to describe previous evidence of efficacy. What’s the point of running a trial if no one reads the results?

It is surprisingly easy for companies to hijack research to market their treatments.

At the other extreme are trials that are little more than shots in the dark. In one case, patients with spinal cord injury were enrolled in a safety trial testing a cell-based regenerative medicine treatment. After the trial stopped (results were negative), laboratory scientists revealed that the cells had been shown ineffective in animal experiments. Though this information had been available to the company and FDA, researchers pursued the trial anyway.

It is surprisingly easy for companies to hijack research to market their treatments. One way this happens is through “seeding trials”- studies that are designed not to address a research question, but instead to habituate doctors to using a new drug and to generate publications that serve as advertisements. Such trials flood the medical literature with findings that are unreliable because studies are small and not well designed. They also use the prestige of science to pursue goals that are purely commercial. Yet because they harm science- not patients (many such studies are minimally risky because all patients receive proven effective medications)- ethics committees rarely block them.

Closely related is the phenomenon of small uninformative trials. After drugs get approved by the FDA, companies often launch dozens of small trials in new diseases other than the one the drug was approved to treat. Because these studies are small, they often overestimate efficacy. Indeed, the way trials are often set up, if a company tests an ineffective drug in 40 different studies, one will typically produce a false positive by chance alone. Because companies are free to run as many trials as they like and to circulate “positive” results, they have incentives to run lots of small trials that don’t provide a definitive test of their drug’s efficacy.

Universities, funding bodies, and companies should be scored by a neutral third-party based on the impact of their trials -- like Moody’s for credit ratings.

Don’t think public agencies are much better. Funders like the National Institutes of Health secure their appropriations by gratifying Congress. This means that NIH gets more by spreading its funding among small studies in different Congressional districts than by concentrating budgets among a few research institutions pursuing large trials.  The result is that some NIH-funded clinical trials are not especially equipped to inform medical practice.

It’s tempting to think that FDA, medical journals, ethics committees, and funding agencies can fix these problems.  However, these practices continue in part because FDA, ethics committees, and researchers often do not see what is at stake for patients by acquiescing to low scientific standards. This behavior dishonors the patients who volunteer for research, and also threatens the welfare of downstream patients, whose care will be determined by the output of research.

To fix this, deficiencies in study design and reporting need to be rendered visible. Universities, funding bodies, and companies should be scored by a neutral third-party based on the impact of their trials, or the extent to which their trials are published in full — like Moody’s for credit ratings, or the Kelley Blue Book for cars. This system of accountability would allow everyone to see which institutions make the most of the contributions of research subjects. It could also harness the competitive instincts of institutions to improve research quality.

Another step would be for researchers to level with patients when they enroll in studies. Patients who agree to research are usually offered bromides about how their participation may help future patients. However, not all studies are created equal with respect to merit. Patients have a right to know when they are entering studies that are unlikely to have a meaningful impact on medicine.

Ethics committees and drug regulators have done a good job protecting research volunteers from unchecked scientific ambition. However, today’s research is plagued by studies that have poor scientific credentials. Such studies free-ride on the well-earned reputation of serious medical science. They also potentially distort the evidence available to physicians and healthcare systems. Regulators, academic medical centers, and others should establish policies that better protect human research volunteers by protecting the quality of the research itself.

What do you think?

We welcome all thoughts, feedback and constructive critiques: editor@leapsmag.com.
A curated selection of responses are collected here.