« Back to blog

The accidental epidemiologist

5 June 2018

The University of Bristol's Professor Peter Fleming describes the work that led to the Back to Sleep campaign, which advised parents to put their babies to sleep on their backs to prevent Sudden Infant Death Syndrome (SIDS). This simple measure has saved around 25,000 babies' lives in the UK, and perhaps as many as 250,000 worldwide. This blog is part of the Bristol Firsts series, celebrating Bristol-based innovations in the NHS’s 70th year. If you have an idea for a Bristol Firsts blog, get in touch with Zoe Trinder-Widdess on

“He was so still - when I turned him over I knew he was dead…”

The description by so many mothers of how they found their precious babies, unexpectedly and inexplicably dead as cot deaths, contains the essential information to allow prevention of many of those deaths – but it took a long time to recognise the importance of what the mothers were telling us.

As a paediatrician and physiologist in the 1980s, trying to understand what might be causing such tragedies, there was urgency to this work. Almost every week another baby, unexpectedly dead or dying, would be brought to the Children’s Hospital.

Research in our lab in Bristol had made great progress. We were beginning to understand the development of breathing, and how it might be vulnerable in early infancy, and of how heat stress might be a further factor in causing catastrophic failure. But somehow, we were missing the point – our careful studies of how breathing control systems might operate or fail seemed to be getting us no nearer to understanding how to prevent these deaths.

Every week, as on-call paediatrician in the emergency department I would meet yet another family whose precious, loved and treasured baby had died suddenly and unexpectedly – and despite my “knowledge” of babies I could give them no explanation, and no advice on how to ensure this tragedy would not strike again.

I could, however, listen to what the families had to say, and a recurrent theme was “I don’t want my baby to have died in vain – please use our experience to help prevent other families from going through this…”.

With this entreaty in our minds, we therefore started systematic collection of information from the families of all babies who died unexpectedly, including the (then) revolutionary concept of visiting the place where the infant had died.

Initially all we could do was to listen to what they had to say, empathise and offer what professional support we could. On its own this information didn't take us much further forward. We needed to know how the experiences of families whose baby had died differed from those whose baby survived – basic epidemiology.

In 1987, with the support of an expert in childhood infections (Peter Rudd) and an enthusiastic trainee paediatrician (Ruth Gilbert) together with a small team including a pathologist (Jem Berry) and an epidemiologist (Jean Golding), we started a study of unexpected baby deaths, investigating the factors that might have contributed.

For each baby who died, via the health visitor, we approached two families in the same neighbourhood with babies of similar age and collected similar information. As a physiologist, my particular interest was in the possible interactions between temperature, heavy wrapping, breathing control, infection and the risk of death.

A series of chance meetings with an Australian paediatrician, Susan Beale, an outspoken advocate of the opinion that putting babies on their tummies caused cot deaths, led me to include detailed collection of information on infant sleeping position in the study. This seemed such an unlikely factor – after all most babies in the UK slept this way, and we had shown that premature infants had higher levels of blood oxygen when sleeping on their fronts – that it would be very easy to disprove, and allow us to concentrate on the other, more likely and more physiologically plausible potential contributory factors.

Two years later the results of the study were becoming apparent. Our initial ideas of the possible importance of heavy wrapping, heat stress and infection had indeed been confirmed, but astonishingly (to us) the single most important contributory factor seemed to be putting babies to sleep on their fronts. Babies who slept this way were almost 10 times as likely to die as those who slept on their backs.

We presented our results to a highly sceptical audience of scientists, health care professionals and bereaved parents at the first conference of SIDS International in London in April 1989, and then spent the next several months trying to identify the fault in our study or data collection that had led to such an unexpected and almost unbelievably powerful finding. When we could not find any such faults we submitted the study for publication – though found the reviewers even more sceptical than we had been, resulting in more than a year’s delay in publication.

As part of our agreement with the families, health visitors, midwives and GPs in Avon involved in the collection of the data, we shared our results with them in a series of small local meetings at health centres throughout the county during the summer and early Autumn of 1989. Although we did not understand the results, and could not be sure that we were correct – it looked likely that sleeping babies on the front led to an increased risk of cot death, and perhaps we should not be advising families to do so.

We were planning a second study to start in autumn 1989 - aiming to improve our information collection, and to investigate how the apparent risks of sleeping on the front might have arisen. In this study we would make direct physiological and environmental recordings of infants at home as well as environmental recordings in the homes of infants who died as cot deaths. Our aim was to compare infants who routinely slept on their fronts with those who slept on their sides or on their backs, but to our astonishment, by the late autumn of 1989 we could not find any babies routinely sleeping on their fronts – the families, health visitors and midwives had taken our results at face value, and were no longer placing babies to sleep on the front.

Over the next two years we watched in near-incredulity, as the numbers of cot deaths in Avon fell dramatically. By the time our initial study was published – in July 1990 – we had already seen local cot deaths fall by more than half. The widespread professional and public hostility that this publication produced – including letters to the Times – with accusations of irresponsibility in making such potentially dangerous recommendations, could be countered with the information that we had not seen any adverse effects in Avon.

In the spring of 1991 the first meeting of the European Society for the Study and Prevention of Infant Deaths in France gave us the opportunity to present the preliminary results of our second study. This time the audience, although sceptical, was less hostile, as our results were almost identical to those of a group from New Zealand, who had made similar observations and similar changes in recommended infant sleeping position a few months after we had. Both studies confirmed the importance of sleeping position as a contributory factor to cot deaths, and showed similar marked falls in numbers of cot death following avoidance of sleeping on the front.

In July 1991 Sebastian Diamond died as a cot death, and his mother, the television presenter Ann Diamond, sought answers to how and why he had died. On learning that in Avon (where she had previously lived) and in New Zealand, parents were being advised not to place babies on the front, Ann worked tirelessly, with the Bristol research team, using her media and political contacts to persuade the UK government to publicise the Avon findings and implement a campaign to prevent cot deaths.

The national Back to Sleep campaign, launched in December 1991 with prime-time television and national newspaper advertisements, together with leaflets at all health centres and many shops, was immediately successful. The number of cot deaths fell, and the winter peak in the number of cot deaths – a consistent feature since the recognition of these deaths as an entity 20 years earlier – did not occur that winter, or any subsequent year. Nationally the number of deaths fell from around 1,600 per year to less than 500. By the time the results of the second Avon study (on which the Back to Sleep campaign had been based) were published in February 1992, scores of infants’ lives had been saved by the simple practice of putting them to sleep on the back rather than the front.

Over the subsequent years our understanding of other factors that contribute to cot deaths has improved, and the death rate has fallen further with implementation of this new information. Back to Sleep campaigns have been mounted in many other countries, with similar effects - a fall of 50-80% in the number of cot deaths.

This approach – of using carefully collected epidemiological information to identify potentially modifiable factors that may contribute to infant deaths, and then to change recommended practice with the aim of saving lives – has been widely recognised as one of the most successful public health achievements in infant care of the past two decades, but we remain unsure of exactly how or why sleeping position affects the risk of such deaths. From the viewpoint of the physiologist there are several possible causal mechanisms, but the definitive investigations of infants sleeping on the front or the back – which we intended to include in our second study – were not possible because of the effectiveness of the advice given then by midwives and health visitors, and became unethical once the effectiveness of this intervention became clear.

All of us who were involved in this extraordinary process have learned a great deal: the importance of being open-minded – even when findings are completely unexpected; the importance of working closely with the whole healthcare team and not shutting ourselves away in our laboratories; and most important of all, listening to what patients tell us.


This work was made possible by research grants from the Foundation for the Study of Infant Deaths, Cot Death Research, the Cot Death Society, the US National Institutes of Health, Action Research, the Southwest Regional NHS Research Directorate, and the Charitable Trusts of the United Bristol Hospitals.

In addition to acknowledging the contributions of those mentioned, I would like to express my thanks to Ruth Wigfield, Mike Levine, Michael Purves, Pete Blair, Yehu Azaz and all members of the multidisciplinary research team.

My particular thanks to the bereaved families and the families of infants who took part in these studies.

powered by Hummingbird CMS