Revised approach for assessing obesity improves on BMI

Body mass index (BMI), which divides a person’s weight by their height squared, comes with fundamental and well-acknowledged problems for diagnosing obesity. It can’t distinguish fat from muscle, tell us where fat is distributed in the body or determine whether excess body fat is actually affecting someone’s health.

Now, a very large international study suggests that a more-nuanced approach could substantially improve obesity diagnosis, identify people at elevated risk of illness and mortality, and help clinicians determine who needs treatment – and how much.

The study tested a framework proposed by a 2025 Lancet Commission. It combines measurements of abdominal fat with assessments of organ function to distinguish between those who are harmed by obesity and those who aren’t. (The health effects of obesity vary among people for reasons not yet known.)

The framework has two steps. First, clinicians used waist circumference and waist-to-height ratio alongside BMI to establish whether someone actually has excess body fat, known as adiposity. “These measures of adiposity in the abdomen are more relevant from a clinical standpoint, because we know that fat has an impact on health,” says study co-author Francesco Rubino at King’s College London.

The second part of the framework determines whether that excess fat is actually causing harm. To develop this framework, the researchers assessed specific criteria for organ dysfunction across multiple systems, including heart failure, chronic obstructive sleep apnea, osteoarthritis, chronic kidney disease or other impairments. If dysfunction is present, the diagnosis is that the individual has clinical obesity. If the individual has excess body fat but organ function remains intact, they are diagnosed with preclinical obesity, a state associated with possible health risks rather than established disease.

To develop this new approach to diagnosing obesity, Rubino and his colleagues analysed health data from 844,108 adults across four cohorts in China, the UK, the US and Spain. They compared conventional BMI-based classifications with the combined measurements and organ-function assessments, examining mortality and the progression from preclinical to clinical obesity.

The results revealed that BMI alone misdiagnosed a substantial number of the adults. Between 1.2 and 3.1 per cent were diagnosed with obesity when the new framework suggests they don’t have the condition. An even larger number of the adults – between 5.2 and 8.6 per cent – had excess adiposity that was missed by using their BMI alone.

In other words, BMI was simultaneously labelling some people as having obesity when they didn’t have excess adiposity, while missing others who did.

“In countries with large populations, you’re talking about misdiagnosing tens of millions of people – this is not a trivial difference,” says Rubino.

Crucially, the study found that mortality risk was better predicted by excess body fat, not BMI alone.

The study also uncovered an important distinction between clinical and preclinical obesity. Clinical obesity accounted for approximately 23 to 31 per cent of obesity cases, with a population prevalence of 5.6 to 10.9 per cent. People with clinical obesity had substantially higher mortality than those with preclinical obesity, the latter of which Rubino called a “risk state”, not a disease itself.

Not only were people with preclinical obesity otherwise healthy, the study showed that the condition didn’t inevitably progress to obesity. Over 10 years, only around 10 to 30 per cent of people with preclinical obesity developed clinical obesity.

These findings could have significant implications for treatment. Rather than assuming that everyone meeting any obesity criteria needs the same level of intervention, clinicians could tailor approaches to the individual. That could include simple interventions and monitoring, while reserving more intensive treatments for people with greater risk or established organ dysfunction.

This more targeted treatment could ensure that limited healthcare resources go to where they can do the most good, instead of attempting to treat all 1.1 billion humans on Earth who have obesity according to BMI alone.

The new study also accounted for well-known differences in BMI and waist measurement thresholds for some ethnic groups. For instance, in populations from South and East Asia, thresholds are lower.

However, Sonia Anand, an epidemiologist and obesity expert at McMaster University in Toronto, says she is less certain that the ethnicity-specific thresholds used in the study represent the optimal global definition of obesity. “Further validation in South Asian, Indigenous and other underrepresented populations using incident diabetes, cardiovascular disease and organ-specific outcomes is still needed,” she says.

The researchers acknowledge that their cohorts relied on diagnostic codes as proxies for organ dysfunction, rather than clinical assessments. Still, Rubino says that the findings again show how BMI isn’t an ideal measure of health. He says they also reframe obesity as a spectrum, separating a risk factor from an established disease and potentially avoiding both undertreatment and unnecessary medical intervention.