The Guardian reported that the largest ever investigation into how obesity affects mortality has found that obese people “die up to 10 years early”. The newspaper said that “moderate” obesity shortens lives by three years, while people who are severely obese will die 10 years earlier than they should.
This study pooled data from 57 separate studies in 894,576 people. It found that, after taking age and smoking into account, people with a ‘normal’ BMI (22.5–25kg/m²) had the lowest overall mortality. With every 5kg/m² increase in BMI above this range, the risk of death from any cause increased by about 30%.
Obesity is associated with diabetes, high blood pressure and ‘bad’ cholesterol, and it is probably a combination of these associated factors that raises the risk of death. This research is valuable in that it gives actual figures for how much obesity increases risk of death.
Where did the story come from?
The research was carried out by members of the Prospective Studies Collaboration from the Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU), University of Oxford. The Clinical Trial Services Unit receives funding from the Medical Research Council, British Heart Foundation and various pharmacological companies. The study was published in the peer-reviewed medical journal The Lancet .
What kind of scientific study was this?
This meta-analysis combined a large number of individual cohort studies with the aim of assessing the relationship between BMI and cause-specific mortality (death from an identified cause). This sort of study requires long-term follow-up of a large number of people. The researchers included studies that had followed people for over five years.
The researchers included 57 studies, with a total of 894,576 participants. The studies were eligible for inclusion in the study if they looked at BMI and mortality; this was the researchers’ sole criterion for inclusion.
BMI was calculated as weight in kg divided by the square of height in metres. A BMI above 30kg/m² was considered obese. People with missing BMI data were excluded, as were those who were severely underweight (BMI <15kg/m²) or were severely obese (BMI ≥50kg/m²). They also excluded anyone with a history of heart disease or stroke at the beginning of the study, or for whom there was no follow-up between the ages of 35 and 89 years.
Most of the participants across the studies had information available on their blood pressure, total cholesterol in their blood, diabetes and smoking status (although only 57% of current smokers had details on the number of cigarettes smoked per day). Far fewer participants had information on blood levels of HDL and LDL (‘good’ and ‘bad’) cholesterol or alcohol consumption. The researchers obtained cause of death from death certificates.
In each individual study, the researchers looked for associations between BMI and other risk factors with adjustment for age. For example, they looked at whether BMI had any link with smoking status. They also looked at associations between BMI and mortality, adjusting the analyses for age, sex and smoking status. To limit the effects of any diseases on the participants’ BMI at the start of the study, the researchers excluded people from their analyses who died within the first five years of follow-up. Risk of death overall and from individual causes was calculated for different BMI categories.
What were the results of the study?
Of the 57 studies that were identified, 92% of the participants were of European origin, with the remainder from the US, Australia, Israel and Japan. The majority (85%) of the participants were recruited during the 1970s and 80s. The average age of most study members when they enrolled was 46 years, and their average BMI was 24.8kg/m². BMI at enrollment was ‘positively linearly associated’ with blood pressure and non-HDL (‘bad’) cholesterol (i.e. as BMI increased so did the other risk factor).
Of the 894,576 people who gave BMI measurements at the start of the study, 15,996 died in the first five years and were therefore excluded from the mortality analyses. During an average of eight years of further follow-up, there were 6,197 deaths from unknown causes and 66,552 deaths from known causes.
These included 30,416 deaths from vascular conditions, 2,070 deaths related to diabetes, kidney or liver disease, 22,592 cancer-related deaths, 3,770 deaths from respiratory conditions, and 7,704 from other causes. Death rates were lowest in those with BMIs between 22.5 and 25kg/m². Comparing all other BMIs to this category, each 5kg/m² rise in BMI above 25 was associated with a 30% increased risk of death overall compared with people in the normal range.
Looking at death from different causes separately, the increase in risk of dying was greatest for deaths related to diabetes, kidney or liver disease (60-120% increased risk compared to those in the normal BMI range), followed by increased risk of vascular mortality (40% compared to those in the normal range), and respiratory-related mortality (20% increased risk). The lowest increase in risk was for cancer-related mortality (10%). For people with a BMI below 22.5kg/m², the risk of death increased as BMI was reduced, mainly due to the increase in respiratory disease and lung cancer, with associations being much stronger for smokers than for non-smokers.
The researchers used the death rates of 35 to 79-year-olds in Western Europe in the year 2000 to estimate the average reduction in lifespan. They estimated that average lifespan is reduced by up to one year for people who, by about age 60, reach a BMI of 25–27.5kg/m². Lifespan was cut by one to two years for those who reach 27.5–30kg/m², and by two to four years for those who become obese (30–35 kg/m²).
For people with a BMI above 35kg/m², they estimate an eight to 10-year reduction in lifespan, although this accuracy is limited because there is much less information for this BMI category.
What interpretations did the researchers draw from these results?
The researchers conclude that BMI is in itself a strong predictor of overall mortality, both for people who are under the optimum weight (less than 22.5kg/m²) and over it (25kg/m²). The increase in mortality above this range is thought to be due mainly to vascular disease, which may also be increased by other closely-associated risk factors, such as high blood pressure. They say that other anthropometric measures, such as waist circumference and waist-to-hip ratio could add additional information to BMI.
What does the NHS Knowledge Service make of this study?This large pooling of data found that overall mortality is lowest in people whose BMI lies within the normal range of 22.5–25kg/m² (after adjustment for age and smoking). Every 5kg/m² increase in BMI above this range increased the risk of death overall, and variably increased the risk of death from individual causes (as listed above). Underweight BMI below the normal range was also associated with an increased risk of death, mainly due to smoking-related lung disease.
This valuable research is useful in that it gives actual figures for how much obesity increases risk of death. There are a few points to consider:
- In the analyses of BMI and mortality, there were some associated risk factors (cholesterol, blood pressure and diabetes) that were not adjusted for. This is because these factors (along with obesity) are collectively associated with a raised risk of cardiovascular disease. Therefore, the increased death rate cannot be attributed to obesity alone as it is likely to be caused by a combination of associated conditions, particularly the increased risk of vascular mortality with raised BMI. In addition, the effects of diet, exercise and socioeconomic status (also related to BMI and other cardiovascular risk factors) were also not taken into account, and these could have confounded the results.
- The participants’ BMI was measured only once in adulthood. But the researchers address this and say that a single measurement is highly correlated with a person’s long-term BMI. However, it also means that no conclusions can be made about links between obesity and excess weight in childhood and increased mortality. Other measures of waist circumference and body fat distribution may also be helpful.
- By combining the results from a variety of different studies from around the world, there may have been differences in study reliability, methods of data collection and follow-up. This could affect how accurate the estimates are.
Analysis by Bazian
Edited by NHS Website
Links to the headlines
The Guardian, 18 March 2009
BBC News, 18 March 2009
The Daily Telegraph, 18 March 2009
Daily Mail, 18 March 2009
Links to the science
The Lancet 2009; early online publication, March 18