91�ȱ�

Being overweight or obese puts men at a greater risk of dying prematurely than women, the largest ever study on obesity and death suggests.

Scientists say though the reasons behind the trend are unclear, the study supports others that suggest obese men are at higher risk of diabetes and have higher levels of dangerous liver fat.

The authors say second to smoking, obesity is the most significant cause of death in Europe and North America.

The report appears in the Lancet.

'Obesity challenges'

A global consortium of researchers pulled together data from 189 studies across the world, involving almost four million people.

They focused on people who had never smoked and did not have a long-term illnesses when the studies started - in an attempt to exclude people who had lost weight through heavy smoking or serious ill health.

Overall, they found the risk of death increased the more overweight a person was.

And the links between obesity and death were strongest for men.

Giving an example of an obese but otherwise healthy man in North America, they estimate the risk of death before the age of 70 to be about 29%, compared with 19% for a man of normal weight.

Meanwhile, they say the risks for a woman in North America would rise from 11% at a healthy weight to 14.6% if she were moderately obese.

Prof Sir Richard Peto, said: "Smoking causes about a a quarter of all premature deaths in Europe and North America, and smokers can halve their risk of premature death by stopping.

"But overweight and obesity now cause about one in seven of all premature deaths in Europe and one in five of all premature deaths in North America."

The data also suggests the risk of death increases at the other end of the spectrum, for people with lower body masses.

But in a linked commentary, researchers from the United States say there are challenges when translating this type of evidence into public health guidelines.

In particular, they point out that removing smokers and people with chronic health problems from the studies excludes a large proportion adults - making it difficult to be sure how widely applicable the results will be.

They call for new approaches to study design and data collection to help overcome these challenges.