Dr Trisha Greenhalgh.

In 1988, 1,596 previously healthy infants in England and Wales were found inexplicably dead by a parent or carer. By 2005 the number had reduced to 191, and the numbers continue to fall. A leading US coroner was reported in the British Medical Journal last month to have said: 'I simply don't believe in Sudden Infant Death Syndrome any more.' So where has this uniquely tragic condition disappeared to?

Cot death, or SIDS, is by definition a death that remains unexplained even after full investigation. It has affected some high-profile families, notably the BBC reporter Anne Diamond, whose son Sebastian was found dead in 1991. She went on to figurehead a major public education programme 'Back to Sleep' which educated parents against putting their babies to sleep in the more dangerous prone position.

This move is said to have halved the incidence of cot death - but sleeping position alone is only a part of the solution, because SIDS is a multifactorial condition. As the BMJ article illustrates, cot death is essentially caused by an inherent weakness (perhaps genetic, perhaps acquired in the womb), coupled with one or more environmental triggers. The fight against cot death has been led, by and large, not by laboratory scientists or pathologists but by epidemiologists, psychologists and sociologists - as well as by parents like Diamond, who play a crucial role in public education.

The first task in reducing the incidence of a multifactorial disease is to gain a detailed understanding of the factors that combine to produce it. This is done by collecting detailed data in a standardised format on every death, and entering these on to a specialised epidemiological database. Complex statistical analysis allows us to quantify the impact of different risk factors, how these interact, and which ones are 'real' risk factors rather than confounders.

Social class, for example, is a very strong risk factor for cot death, but if you control for maternal smoking, co-sleeping, alcohol, infections and other established risk factors, the actual impact of class per se is minuscule - and will probably one day be fully explained by specific behavioural and material differences. Co-sleeping in the parents' bed (common in the middle classes) is associated with a marginally increased risk of cot death, but co-sleeping on a sofa (much commoner in lower socio-economic classes) is many times more dangerous.

Putting your baby to sleep on its front is easy enough to change if you are advised against it. But because of the success of the Back to Sleep campaign, the proportion of cot deaths accounted for by sleeping position has fallen from 50% to around 10%. Many residual risk factors - especially smoking, which is now implicated in 90% of cot deaths and the leading explanation in 60% - are much more deeply entrenched in our identity and lifestyle.

Sociological research shows that young women from poor backgrounds who smoke do so because it's one of the few things in life they have control over. All this makes for a complex health education challenge.

The key to further progress lies in more information, systematically collected and meticulously analysed. Why, for example, is a sofa more dangerous than a bed for co-sleeping? Is it that sofas have sides so suffocation is more likely, or that the sofa is more likely to have animal hairs on, or that the incumbent of the sofa is more likely to be drunk or on sleeping pills? In order to differentiate between these hypotheses (and explore others) we need to know a lot about the circumstances of the co-sleeping arrangement.

This is why the US Centre for Disease Control has produced an eight-page form with 79 questions which it recommends that every cot death family should complete with sensitive professional help. The questions include details of the pregnancy and birth, a dietary and medical history, a 'witness report' of the scene where the infant was found, social issues such as domestic violence, prior sibling deaths, and religious, cultural or ethnic remedies.

Should bereaved parents be made to go through such an extensive analysis? Of course not, but the evidence suggest that most of them want to. Not only does it structure and contain the heart-searching that they are going through anyway (Why our baby? Could anything have been done to prevent it?), but it offers hope that they are contributing to an important research endeavour that will one day consign the spectre of cot death to history.