BMI categories may need adjusting, argue researchers

Wednesday May 11 2016

"Being overweight may not be as unhealthy as it was 40 years ago," BBC News reports.

New research has found a body mass index (BMI) of 27 is linked to the lowest rate of death – but someone with a BMI of 27 is currently classed as being overweight.

BMI is a score calculated by dividing your weight (usually in kilograms) by the square of your height (usually in metres and centimetres). Currently, a BMI of 25 to 29.9 is classified as being overweight.

Researchers looked at 120,528 people from Copenhagen, recruited from 1976 to 2013, and separately compared those recruited during the 1970s, 1990s and 2000s. They were followed up until they died, emigrated, or the study finished.

The BMI linked to the lowest risk of having died from any cause was 23.7 in the 1970s group, 24.6 in the 1990s group, and had further risen to 27 in the 2003-13 group.

It may be the case that the suggested upward shift in optimal BMI is the result of improvements in preventative treatments for weight-related conditions such as type 2 diabetes.

But this is just an estimate based on averages – it doesn't mean that having a "healthy" BMI is bad for you. Similarly, it shouldn't be assumed that it's now best to be in the overweight category. People often gain weight as they age, so there is the risk you could move from being overweight to obese.

Where did the story come from?

The study was carried out by researchers from Copenhagen University Hospital.

It was funded by the Danish Heart Foundation, the Danish Medical Research Council, Copenhagen County Foundation, Herlev and Gentofte Hospital, and Copenhagen University Hospital. 

The study was published in the peer-reviewed Journal of the American Medical Association (JAMA).

The study was covered by the UK media with a certain amount of glee, with the Daily Mail suggesting that the BMI system was a "blunt instrument".

It also said this study showed that, "Millions of Britons who are currently classed as overweight, actually have the optimal BMI and the lowest chance of death."

However, the study was reported on accurately, and the reports included expert views saying that people still need to keep an eye on their weight.

What kind of research was this?

This cohort study compared results from three large previous cohort studies in the same part of Denmark, starting at different times.

Researchers wanted to see if there had been a change over time in the optimal BMI score – that is, the BMI shared by people with the lowest rate of death from any cause.

While this type of study can show trends of this nature, it cannot explain why the changes happen.

What did the research involve?

Groups of adults in Copenhagen had their height and weight measured as part of three studies carried out in the city in 1976-78, then 1991-94, and the final study in 2003-13.

Researchers followed them up, then looked to see at which BMI people had the lowest chance of dying. They compared the numbers for the three studies to see if that number changed over time.

The first two studies were linked. Participants for the first study were invited back for a second round of measurements over the period from 1991-94, although younger people were recruited to add to the numbers. People in the third study had not taken part in either of the first two.

As well as weight and height, researchers checked whether people smoked, how much exercise they did, whether they'd been diagnosed with any medical conditions, including cancer or heart disease, and how much alcohol they drank.

They carried out sensitivity checks by including or excluding people with different risk factors to see whether any of them explained the overall results.

The researchers also looked at whether length of follow-up made a difference. They did this by carrying out their calculations with a much shorter follow-up period to see if the longer follow-up from the older studies distorted the results.

What were the basic results?

The average BMI at which fewest people in the studies died from any cause increased by three points over the three decades:

  • 23.7 (95% confidence interval [CI] 23.4 to 24.3) in 1976-78
  • 24.6 (95% CI 24 to 26.3) in 1991-94
  • 27 (95% CI 26.5 to 27.6) in 2003-13

The results showed a similar shift when researchers looked at just deaths from cardiovascular disease for non-smokers who had not been diagnosed with diabetes, cardiovascular disease or cancer, as well as for shorter periods of follow-up. None of the sensitivity analyses explained the trend.

In addition, researchers found the increased risk of death linked to being obese – a BMI of 30 or above – compared with a "healthy" BMI has gradually decreased to zero.

In the 1970s obese people had a 31% increased risk of death. By the 1990s it had reduced to a 13% increased risk, and by 2003-13 there was no longer a statistically significant link (adjusted hazard ratio 0.99, 95% CI 0.92 to 1.07).

How did the researchers interpret the results?

The researchers say their findings were "robust" and cannot be explained by confounding factors such as age, sex, smoking status and disease at the start of the study.

They said that, "If this finding is confirmed in other studies, it would indicate a need to revise the World Health Organization (WHO) categories presently used to define overweight."

They also said cohort studies cannot address the causes of the results, but speculated that their finding may reflect improvements in treatments for diseases affecting people with higher BMIs, such as heart disease and diabetes.

This would make it less risky to be overweight than in the 1970s, when more people died of these diseases. The reduction in smoking and increase in exercise they found could also have helped mitigate the effects of being overweight, they said.


The link between weight and health is not straightforward. We've known for years that if you plot death rates against BMI categories on a graph, you get a U-shaped curve, where people who are very underweight or very overweight are at higher risk of dying, while people in the middle have a lower risk.

This makes sense: extremes of weight are linked to illness, both as a cause or result. Many people with cancer or lung disease, for example, are underweight, which is one reason why lower BMIs are linked to higher death rates. That's why doctors talk about people having a "healthy" BMI.

What this study seems to show is that the lowest point of that U-shaped curve has shifted to the right, towards higher BMIs. But it doesn't mean that slimmer people are at a higher risk of death.

The study shows that in the period 2003-13, there was no difference between the death rates of people with a BMI of 18.5 to 24.9 (healthy) and those with a BMI of 25 to 29.9 (overweight), which were 4 per 1,000 per year for both groups.

The rate for obese people was 5 per 1,000 per year, despite this being a non-significant increased risk of death. There's certainly no need to try to put on weight if you are already at a healthy weight for your height.

The potential reasons for the shift are interesting. It may be, as the researchers suggest, that the diseases which killed more overweight people in the 1970s are now better treated and controlled, meaning that the risks of being overweight are smaller than they once were.

It's possible that the risks associated with being underweight have not decreased in the same way, which would automatically shift the "optimal" point towards overweight.

Also, despite a general increase in the population's BMI over the decades, health awareness has improved. Though the results have taken smoking status into account in the analyses, other factors, such as improvements in physical activity and alcohol moderation, could be having an influence.

However, this study has some limitations. Importantly, it was only carried out among white Danish people, which means it may not apply to other ethnic groups.

We know that some groups, such as people of south Asian origin, are more likely to have problems such as diabetes at lower BMIs than white people, so this study might not apply to everyone. And the follow-up for the most recent group studied was, on average, four years, so we don't yet know if this is a long-term trend.

The criticisms of the BMI system are not unfounded, though. BMI doesn't take into account the increased weight of muscle compared with fat – some athletes have high BMIs, despite being very fit, for example.

Waist circumference and waist-to-hip ratio can give a good indication of body "fatness". Regardless of your height or BMI, you should try to lose weight if your waist is:

  • 94cm (37in) or more for men
  • 80cm (31.5in) or more for women

You are at very high risk and should contact your GP if your waist is:

  • 102cm (40in) or more for men
  • 88cm (34in) or more for women

Read more about why waist size is important.

Analysis by Bazian
Edited by NHS Choices