Matt Morgan and Peter Brindley: Why negative trials are usually positive

There are many benefits to patients from negative trial results

We’re going to let you in on a secret. Although intensive care medicine often looks complex, usually it is not. In fact, it is likely one of the simplest medical specialties. Despite a myriad of journals, shockingly expensive therapies, and all the latest shiny machines, we have a meagre four evidence-based curative interventions: antibiotics, steroids, surgery, and most importantly, time. Thus, when a large clinical trial comes along that examines any one of these four pillars it is a big deal. The ADRENAL trial justifies the hype and is, to quote Kathy Rowen, “a wow moment. ADRENAL is the world’s largest trial examining the putative role of corticosteroids in fighting the scourge of critical illness: sepsis. The researchers deserve unfettered praise and a well-earned rest. The researchers also signalled how information dissemination has changed by releasing the data in Belfast, away from the mega-conference circus, but streamed live worldwide thanks to Rob McSweeney. While the trial’s intimate details will be dissected ad nauseum by the Critical Care Medicine community, there are three general themes of relevance for the wider medical community.

The ADRENAL trial could be called another “negative” trial. After all, the primary end point—mortality at 90 days—was no different in the steroid group compared with placebo. For a trial that consumed millions of pounds and a comparable number of research hours, this could be described as a disaster. We would argue instead for more salutary words: important, exciting, provocative. Medics are biased towards positive trials and may assume that these are more beneficial. Importantly, however, negative trials allow interventions NOT to be used. This translates into time savings, cost savings, and time focused on interventions likely to work rather than unlikely to work. Moreover, despite all the resources available to medicine in general, “time” is often our most precious resource. Time is, after all, both a great healer, and a reliable prognosticator. Removing interventions with no proven benefit limits distractions, and focuses our energies. This also obviates harmful side effects. Given the publication bias towards “positive” trials—especially those with pharmaceutical backing—negative trials may actually be a better reflection of scientific truth. If you agree with our logic then perhaps you will accept our conclusion: overall negative trials may actually be more beneficial: to professionals, to health systems, and most importantly to our patients.

ADRENAL interrogated many important secondary outcomes, several of which reached statistical significance. On average, the steroid recipients spent less time hooked up to mechanical ventilators, had less blood transfused, and spent less time in the intensive care unit. However, secondary outcomes and surrogate measures have been treated with disdain in the medical literature for at least twenty years. Scepticism is appropriate, but cynicism is not. Nor is intransigence in the face of large study data. The danger is that the “secondary outcome excuse” can also be a way to hold on to unscientific hunches or unsupported biases. Interestingly, in a holistic way, secondary outcome measures tend to be far more patient focused than raw, binary outcomes: dead versus not dead. For the patient, less time in a scary critical care environment, or fewer days attached to an annoying ventilator, may be personally hugely meaningful. Furthermore, reducing costly or dangerous interventions, such as blood transfusions, is  prudent medicine. Reducing costs could also allow funds to be reinvested. Sometimes, the endless striving towards a p value showing small improvements in death can hide noteworthy benefits that have a positive impact on a patient’s life.

Finally, and perhaps most importantly, the ADRENAL trial casts a jaundiced eye towards trial safety monitoring. Trials are stopped early when interim analyses show strong signals of either harm or benefit. On first blush this seems appropriate as it both safeguards participants, while allowing early dissemination. However well intentioned, interim analyses can also be premature and therefore wrong. Had the ADRENAL trial been stopped during the interim analysis, we might have lost the major take home point: no proven benefit. This begs the question of how many other trials have been stopped before their best-by date, and how often have we landed on an “alternative truth”. Perhaps this “early warning system” is not as safe as finding the enduring truth, which will benefit a far greater number of future patients. In short, in addition to congratulating the researchers, we wish to emphasize that this negative trial was actually extremely positive. The search continues…

Matt Morgan, Honorary Senior Research Fellow at Cardiff University, Consultant in Intensive Care Medicine and Head of Research and Development at University Hospital of Wales, and an editor of BMJ OnExamination. He is on twitter: @Matrix_Mania 

Peter Brindley, professor of Critical Care Medicine, Anesthesiology and Pain Medicine, and Health Ethics, University of Alberta, Edmonton, Canada@docpgb