Smoking

A draft report was later modified to say the links were 'difficult to quantify', raising concerns that the company tried to influence the study findings.

The link between cot death and babies who live in homes where adults smoke has been strengthening over the past decade. This year, a report by the government's scientific committee on tobacco and health said children exposed to cigarette smoke were at 'greater risk' of cot death.

The first drafts of the study emerged after researchers at the University of California trawled through company documents released as a result of lawsuits in the US. Their findings have just been published in the journal Pediatrics.

The report was written by the British toxicology consultant Dr Frank Sullivan. When he concluded that passive smoking was a possible cause of cot death, or sudden infant death syndrome, company scientists tried to persuaded him to change his mind, supplying him with evidence to suggest that the links were 'unproven'.

As a result of interventions by Philip Morris - the world's biggest tobacco firm - Sullivan modified his findings. His final study was published in 2001 in an influential scientific journal, without any indication of the full role played by the tobacco giant.

The authors claim that the company documents reveal that the Sullivan study was part of an attempt by Philip Morris to sway public policy about smoking around children.

In the early 1990s, Philip Morris was trying to counter the view that tobacco smoke could cause cot death. Tobacco bosses feared that such evidence would create a powerful argument for smoke-free areas in the home and public places. Stan Glantz, professor of medicine at the University of California, said: 'Their aim was to use this study to subtly manipulate public opinion. By [using] the scientific literature, they can create doubt and confusion among doctors and health officials over the debate about smoking around children.'

In 1997, Philip Morris commissioned Sullivan to write a review of all the known risk factors of sudden infant death syndrome. Sullivan was an expert on birth deformities and had advised the British government on the impact of chemicals on human reproduction. There is no suggestion that he was knowingly part of any plan improperly to shift public policy.

For Philip Morris, Sullivan was an ideal author because he had also been a consultant to the cigarette firm Rothmans. A memo said he had 'excellent credentials'.

Sullivan's consulting fees from Philip Morris were £800 a day and the company budgeted between £25,000 and £50,000 for his study. The contract stated that Sullivan would remain free to publish the final version without restriction by Philip Morris.

Sullivan's original draft said that not only does a mother smoking during pregnancy increase the risk of cot death, but so also does exposing a newborn to smoke: 'The evidence overall suggests that both pre-natal and post-natal maternal smoking exert independent effects.'

His final draft included amendments suggested by company scientists which played down the role of smoking and particularly the role of second-hand smoke.

At the suggestion of Philip Morris, he added two paragraphs about a 1999 Tasmanian study of cot death that found no conclusive link between second-hand smoke and cot death. A company scientist told Sullivan: 'I think that this paper is extremely important and I also think that, if at all possible, it ought to be included in your review.'

After Philip Morris's interventions, Sullivan changed his conclusion that passive smoking was an independent risk factor for cot death. He argued that, although smoking during pregnancy raised the risk of cot death, the threat of passive smoking was 'less well-established'. His final article, published in the Journal of Paediatric and Perinatal Epidemiology in 2001, stated the relationship was 'difficult to quantify'.

Sullivan said of Philip Morris: 'It is true they suggested I looked at some other studies and made comments on my work, but they were entitled to do that. Any changes that were subsequently made were down to me and me alone.'

He said he stood by his conclusions that the risks posed by passive smoking were 'difficult to quantify'. The problem was that babies who came into contact with second-hand smoke were normally exposed to it because their mothers smoked during pregnancy. In cot death, it was difficult to establish the precise cause.

Nobody at Philip Morris could be contacted, but a statement said people should be 'guided by the conclusions of public health officials' regarding passive smoking.

· Additional research by Jason Rodrigues.

Cot death: the facts

Cot death is the industrial world's most common cause of infant death.

In 2003, 175 babies died of cot death in England and Wales.

According to the Office for National Statistics cot deaths are four times more likely where babies are born outside marriage. Within marriage, cot death is three times more likely where the father is a manual worker. It also occurs more among mothers under 20. Low-weight babies are also at greater risk.

A European study found a range of risk factors including babies' heads being covered in bedding and sleeping under a duvet. Out of 10 deaths, six could probably be attributed to lying babies on their front or side.