Richard Peto, who worked with Richard Doll to link smoking and cancer, tells Huw Richards their rewarding collaboration continues despite Doll's death

When does an academic collaboration end? Sir Richard Doll died at the age of 92 in July 2005, but Sir Richard Peto, his longtime working partner on studies of cancer that established beyond question the connection between the disease and smoking, says: "I feel that we are still collaborating."

This should not be taken as meaning that Peto, who is as rational as one would expect of a man who is Oxford University's professor of medical statistics and epidemiology and co-director of the Clinical Services Trial Unit, has taken to the Ouija board. He explains: "There are ways of thinking and general standards of work that he set that have passed into my thinking and that of the unit."

It is hardly surprising that Peto should bear the imprint of a connection that lasted the length of many academic careers. They met in 1967 when Peto, a recent Cambridge University graduate, was interviewed for a job as a statistician in Doll's unit. "I had no idea what epidemiology was and had never heard of him. He said he'd like to offer me the job but it had already been offered to somebody better qualified. But this was what he said to all interviewees - and a couple of days later his secretary rang to offer me the job."

³ÉÈËVRÊÓƵ

Doll was always scrupulously concerned to describe him as a "colleague", but Peto says: "Inevitably, in the early days, it was a rather asymmetric relationship. He knew so much more than I did about the subject. It was not until the late 1970s that we were collaborating as equals with different perspectives, although he remained the senior figure."

The centrepiece of their joint work was the long-term study of cancer among British doctors, begun in 1951, with snapshots taken every ten years. Peto recalls the 20-year follow-up in 1971: "It was hugely demanding. We had nothing like the computing capacity today, and it took until 1976 to publish the results."

³ÉÈËVRÊÓƵ

Each decennial survey has further underlined the connection between smoking and premature deaths from cancer. Peto recalls being asked in the late 1970s by Nature to review a publication called The Politics of Cancer .

"There had been a wave of hysteria about cancer having environmental or occupational causes. It did make us think and wonder whether we'd got it right; but the more we looked, the more it was clear that smoking was overwhelmingly the most important cause of premature deaths. This book argued that it was down to industrial villainy. I ended up writing a four-page critique that was regarded as something of a declaration of war. The US Congress asked me to do a report on the causes of cancer, and I said I would as long as they allowed me to do it with Richard."

The result - Avoidable Causes of Cancer in the United States - was "one of the best things we ever did", Peto says, and "much better as a collaboration than it could ever have been written by one or other of us".

Although their joint work was fruitful, it was rarely easy. "We argued about every line." In part this related to differences in interests and academic training. "Richard was always more interested in cancer, whereas my strongest interest was in premature deaths. He was not a mathematician, but he had the ability to think like one."

³ÉÈËVRÊÓƵ

More often, though, the arguments arose from a shared belief in the importance of their work to the millions of people whose lives it might help prolong. "It was so difficult and so desperately important that we get it right. When we got it right both of us knew that we had, but it was an incredible struggle to get there. Richard wrote beautifully and with great clarity, but there is hardly a line in our work that I can point to and say 'that's his or that's mine'," Peto says.

Nor was there ever any question of Doll - who had received an MBE for his research work while Peto was still a schoolboy - pulling rank to impose his views. "I found that he rather liked being criticised, provided this led to something being done better than it would have been."

There was always a tension between the need for precision and the pressure of deadlines. "I managed to get a six-month extension from the US Congress. We knew that the report was needed urgently, but there was so much that needed saying and was still unsaid."

Collaboration continued long after Doll had reached academic retirement age. "I employed him - in theory at least. It seemed crazy to let one of the greatest epidemiologists leave us. He had done 22 years of epidemiology before becoming regius professor, reforming medical education at Oxford and setting up Green College, and then returned for another 22 years of extraordinary vigour and productivity."

³ÉÈËVRÊÓƵ

The partnership extended to mutual support when they lost their life partners. "The funerals for his wife and my partner were in the same month. Each of us spoke at the other memorial service, and for a couple of years we were both oppressed by grief and could only work superficially," Peto recalls.

Doll recovered and was at work until the end. "He wrote one of his best papers in the year before he died." In hospital in his final days, he remained sharp enough to read a 600-page book and leave an indelible impression. "Having seen how he coped with it, I'm now sure that I won't be afraid of dying," Peto says.

³ÉÈËVRÊÓƵ

Doll died while they were working on a report on the fifth survey (2001). "Typically, he'd got his contribution in before going into hospital," says Peto, who points to a graph showing the dramatic decline in premature deaths as a result of the reduction in smoking. He adds: "If you want a monument to your life's work, that's not a bad one."