In biomedical science, discovery is supposed to lead inexorably to application and changes in medical practice. But things are rarely this simple. The use of prenatal steroids is a case in point. It is an undeniable medical 'success story', but an examination of its history, discussed at a recent 'Witness Seminar' raises troubling questions about how research is translated into practice and benefits people.

The basics

The story begins in the late 1960s, when Graham ('Mont') Liggins returned to New Zealand from the USA, with a Wellcome Trust grant to study pregnancy and birth in sheep. Premature birth was of great interest at the time. Liggins had the idea that the fetus might be producing something that triggered premature labour, possibly steroid hormones. He set about testing the effects of different steroids on pregnancy.

As so often happens in science, serendipity then intervened. Liggins routinely carried out post-mortems, and he noticed a striking difference in the lungs of two fetal sheep. As Liggins's co-worker Ross Howie describes:

"I still remember the excitement I felt[…]when he handed me the lungs of twin lambs for pressure – volume studies. The lambs had been delivered very early: one had been infused with glucocorticoids and the other not. Lungs of the infused lamb were perfectly stable after inflation: pink, fluffy and floated in water. In total contrast, the lungs of the other remained solid and liver-like and sank."

At the time, premature birth was a big problem. An infant's underdeveloped lungs left it gasping for air – respiratory distress that could be fatal or caused permanent brain damage, typically cerebral palsy. Many people were trying to extend pregnancy, to give the infant lungs more time to develop. But what if steroids could accelerate development of the fetal lungs? A baby born prematurely would then have a much greater chance of survival.

So chance was only part of the story. As Pasteur noted, "Fortune favours the prepared mind," and Liggins immediately realised his chance discovery had clinical potential.

As it happened, Auckland was the ideal place for the research to progress. Liggins was eager to try clinical studies, and he needed to work with a clinician. At the time, Ross Howie was the only paediatrician in the whole of New Zealand who could artificially respirate premature babies. The two set about designing a clinical trial to test the effects of a single injection of steroids in mothers undergoing premature labour – a conceptually simple randomised controlled trial, half the mothers getting steroids, half a placebo.

Within a few months (an unthinkably short time now), the trial had begun. The results were crystal clear. More babies were surviving and staying healthy in the treated group. Prenatal steroids were working.

Given that the new treatment was saving babies' lives, one might have expected a rapturous reception for the results. But nothing of the sort happened. The Lancet rejected the paper. Ross Howie recalled the reaction of the Royal College of Obstetricians and Gynaecologists in 1972: "They didn’t want to know."

Not that the work was completely ignored. Avery and others began similar studies, in experimental animals and humans. A large trial was coordinated by the US National Institutes of Health (NIH).

Dissemination

Working in Dublin at the time was a young obstetrician, Patricia Crowley. Just as the steroid work was breaking, a premature baby in her care died due to respiratory distress. She became an early convert to steroid treatment and began collecting papers published on the topic (a laborious manual process in those pre-database days). She published a review in a new publication, the Journal of Obstetrics and Gynaecology, a cost–benefit analysis of prenatal steroids (an approach she puts down to the fact she was dating an economist at the time).

She soon realised that this approach was too ad hoc, and needed to be done systematically to be of real value. At the same time, Iain Chalmers in Oxford was growing convinced of the need for systematic reviews of the literature, to inform clinical practice. He had grown increasingly alarmed at the number of medical calamities that could have been avoided had already published knowledge been applied.

Chalmers established the Cochrane Collaboration to promulgate 'evidence-based medicine’', particularly evidence derived from randomised controlled trials. Moreover, the concept of 'meta-analysis' had emerged in the mid-1970s, a way of pooling data from multiple trials; by themselves, clinical trials (particularly small ones) may throw up spurious associations.

Crowley led the systematic review of prenatal steroid use, covering eight large trials. Again, the results could not have been clearer: steroids had a reproducible beneficial effect. A graphic representation of this study became the Cochrane Centre's logo.

Finally, things began to accelerate. In Birmingham in the 1980s, Professor Richard Lilford was a convert to evidence-based medicine. He suggested to the board of the Royal College of Obstetricians and Gynaecologists that it should promote good practice by producing clinical guidelines. To his surprise, they agreed. He drew together a group that spent a long day trawling through a database on perinatal medicine established in Oxford. The group proposed 21 guidelines, in order of clarity of benefit. Number 2 in the list were prenatal steroids.

The guidelines seemed to do the trick. Published in 1992, almost overnight they led to a huge change in clinical practice. Bizarrely, though, doctors not only took up steroids but began using higher or multiple doses, without any evidence that this was effective or safe (several trials are currently assessing the effects of multiple steroid doses).

Lessons

The fascinating tale of the prenatal steroids raises all kinds of questions about biomedical science and its translation. It is often cited as evidence of the benefits of animal research – it originated with Liggins's observations in sheep. It is also an example of the clinical payback of ‘basic’ research. Yet it also raises questions about how scientific information is used.

Why did it take so long for such a valuable treatment to be accepted? A previous Witness Seminar had highlighted the fact that the location of the research – New Zealand – had lessened its impact: how could key research come from such a backwater?

Perhaps Liggins and Howie could have done more to publicise their work. Howie was very cautious, arguing that decisions should never be based on one trial and that more research was needed. Pharmaceutical interests may also have been significant; companies were more interested in new drugs to prolong pregnancy.

The later research was also damaging, often being of poor quality or raising issues in animals that obscured the findings in people. (Steroid biology is complicated, and some findings in animals seem to apply to humans and some do not.) Perhaps most problematic was the US NIH study, particularly its subgroup analysis, which suggested, for example, that prenatal steroids had no benefits for Caucasians. Reanalysis of the Liggins and Howie data failed to confirm this anomaly.

Undoubtedly, clinicians' own experience had a huge impact. In the 1980s, two hospitals in Cardiff had diametrically opposite approaches, one having seen for itself the benefits of prenatal steroids in a series of local case studies.

The turning point seems to have been the systematic reviews and the clinical guidelines. The evidence had always been in the literature but the sheer mass of papers was obscuring it. The meta-analysis and production of clinical guidelines gave doctors an authoritative source on which to base their clinical decisions.

Perhaps there are other lessons. The value of challenging orthodoxy: most people believed it was the mother that controlled the time of childbirth; Liggins insisted it was the fetus. The negative impact that poor science, no matter how well motivated, can have on important medical issues: bad science is worse than no science. The detrimental effect of scientific snobbery: how could the colonies produce anything significant? And finally, interdisciplinary turf warfare undoubtedly played a part too: obstetricians and paediatricians did not always talk to one another, and the latter saw their professional territory being invaded.

External links

• Witness Seminars
• The US National Institutes of Health (NIH)
• The Cochrane Collaboration