Publish or Perish

Neglecting to report health research can lead to distorted clinical-practice recommendations, suffering, and death. Yet failure to publish results remains by far the most common form of scientific and ethical misconduct in health research.

OXFORD – A few years ago, two Norwegian researchers presented their version of a scientific-research slippery slope, with honest errors (mistaken observations and analyses) at the top and intentional fraud (plagiarism, falsification, and fabrication) at the bottom. They placed “publication bias” about halfway down the scale – but it should have been much lower. Indeed, failure to publish research results is by far the most common and worrying form of scientific and ethical misconduct in health research – and it has had lethal consequences.

Anecdotal evidence of publication bias has existed for a long time. But, as requirements for registering clinical trials have become more stringent, its magnitude has become quantifiable: the results of at least half of the clinical trials involving patients and healthy volunteers remain unpublished years after completion. Studies with “negative” results are particularly unlikely to see the light of day. But neglecting to report research can lead to distorted clinical-practice recommendations, suffering, and death.

For example, anti-arrhythmic drugs were prescribed widely – with the United States Food and Drug Administration’s approval – to heart-attack victims for more than a decade, under the assumption that reducing heart-rhythm abnormalities would decrease mortality rates. But, as the investigative journalist Thomas J. Moore reported in his book Deadly Medicine, at the peak of their use, these drugs were killing more Americans each year than were killed during the entire Vietnam War.

Following this revelation, a British research team reported a clinical trial that they had carried out more than a decade earlier. Their study had found a higher death rate among patients receiving a new anti-arrhythmic drug than among those receiving a placebo. They had not published the results, they explained, because the drug’s development had been abandoned for commercial reasons. In retrospect, however, they observed that their results “might have provided early warning of trouble ahead.”

Beyond avoidable suffering and death, failure to report research also slows the pace of scientific discovery. It is widely acknowledged that the high current level of investment in pharmaceutical research is yielding too few substantive discoveries.

At a meeting on “translational research,” a senior member of a contract-research organization – a company that provides outsourced research services to the pharmaceutical and biotechnology industries – articulated the problem. “What really frustrates us,” he commented, “is to be commissioned by a pharmaceutical company to design an early-phase trial when we know already that it is going up a blind alley. And the reason we know that is that we’ve been up the blind alley with another company; but that experience remains unreported for commercial reasons.”

SPRING SALE: Save 40% on all new Digital or Digital Plus subscriptions

SPRING SALE: Save 40% on all new Digital or Digital Plus subscriptions

Subscribe now to gain greater access to Project Syndicate – including every commentary and our entire On Point suite of subscriber-exclusive content – starting at just $49.99.

Subscribe Now

In other words, the drug-discovery process suffers from built-in inefficiency, owing to inadequate reporting. But, while waste may be tolerable, avoidable harm to patients is not. Drug-trial volunteers should not accept this abuse of their participation in research. Failing to publish the results of clinical trials is a betrayal of the implicit trust that they have placed in researchers to use their contribution to increase and improve the stock of scientific knowledge.

Nor should the public accept this squandering of resources. There is simply no justification for withholding the results of health-research studies. Publication is a moral imperative.