Loading ...

A new scoring tool could complement BMI in identifying individuals with overweight or obesity who are at the highest risk for obesity-related complications and poor health outcomes, according to a study published in Nature Medicine.

The tool, called OBSCORE, is an integrated model developed using interpretable machine learning that incorporates a range of health indicators to predict future risk for obesity-related conditions. The model may provide a framework for prioritizing high-risk individuals for earlier and more intensive interventions.

“Our study developed and validated a risk prediction model that can stratify individuals into high- and low-risk groups based on their likelihood of developing 18 obesity-related health conditions, including type 2 diabetes, heart disease, or kidney disease,” lead author Kamil Demircan, MD, PhD, a postdoctoral researcher at the Precision Healthcare University Research Institute, Queen Mary University of London, London, England, told Medscape Medical News.

Why BMI Alone Falls Short

“Overweight and obesity have become a major global health challenge, affecting roughly 60%-70% of adults in many Western countries, with prevalence continuing to rise in most parts of the world,” Demircan said.

However, health trajectories can vary substantially among individuals with similar body weight or BMI.

“Accurately identifying individuals at highest risk of developing obesity-related complications is therefore a challenge, but essential to ensure earlier monitoring, interventions, and improved health outcomes,” he said.

The authors argue that BMI alone is an incomplete measure of risk and that obesity assessment should incorporate broader measures of health status and disease susceptibility. Yet the parameters required to accurately identify “clinical obesity” and predict future complications remain unclear, and no data-driven framework has existed to guide intervention allocation according to risk.

Developing OBSCORE

To address this gap, the researchers developed OBSCORE, a single, integrated model designed to predict the future onset of 18 important cardiovascular metabolic and mechanical complications associated with obesity.

The model was derived from data on 197,264 UK Biobank participants with a BMI ≥ 27 (median age, 58 years). Using machine learning, investigators evaluated more than 2000 demographic, lifestyle, clinical, biochemical, and health-related variables.

The final model incorporated 20 health-related features, most of which are routinely collected in clinical practice. Twelve were also included in the SURMOUNT-1 trial, including age, sex, waist to height ratio, hypertension status, cholesterol measures, A1c, renal function markers, and smoking status.

Additional factors included family history of heart disease, self-rated health status, gamma-glutamyl transferase, long-standing illness, pain (abdominal, joint, or chest), and current smoking.

The 18 outcomes included type 2 diabetes (T2D), hypertension, coronary artery disease, heart failure, atrial fibrillation, stroke, chronic kidney disease (CKD), gout, obstructive sleep apnea, metabolic dysfunction-associated steatotic liver disease, liver cirrhosis, gallbladder disease, gastroesophageal reflux disease, osteoarthritis/arthropathy, venous thromboembolism, obesity-related cancers, chronic obstructive pulmonary disease, and all-cause mortality.

Approximately 48% of participants were women, nearly 10% had T2D at baseline, and 4.4% had a medical history of a major adverse cardiac events.

A Complement to BMI

The researchers divided participants into five risk quintiles using OBSCORE predictions. Importantly, because all participants had overweight or obesity, even those in the lowest-risk quintile had elevated BMI.

For 12 of the 18 outcomes, OBSCORE demonstrated substantial risk stratification, with rate ratios exceeding 10 when comparing the highest- and lowest-risk quintiles. The largest differences were observed for CKD, gout, and T2D, with rate ratios of 89, 36, and 42, respectively.

For cardiovascular mortality, the rate ratio was 47. The 10-year event rate was 5.7% in the highest-risk quintile compared with 0.1% for the lowest-risk quintile.

When applied to participants in the SURMOUNT-1 trial, OBSCORE showed that weight loss with tirzepatide was similar across baseline risk groups and that predicted risks for obesity-related complications decreased following treatment.

“We found considerable differences in future health risks among individuals with overweight or obesity despite being in similar BMI categories,” Demircan said. “Our findings suggest that BMI-based approaches to obesity management could be complemented by a more risk-centric assessment of obesity.”

Once validated in additional studies, the findings could help clinicians identify high-risk individuals earlier and support more targeted prevention, monitoring, and treatment strategies, he added.

The authors noted that participants were predominately middle-aged and older adults. Given that obesity rates continue to rise in younger adults and adolescents, future studies will need to determine whether OBSCORE performs similarly in these and other populations.

Not Ready for Prime Time in the US

Commenting for Medscape Medical News, Marc-André Cornier, MD, professor of medicine and director of the Division of Endocrinology, Diabetes and Metabolic Diseases at the Medical University of South Carolina, Charleston, South Carolina, called the concept “excellent.”

“We use risk assessment equations for other chronic diseases, like high cholesterol and cardiovascular prevention,” said Cornier, immediate past president of The Obesity Society. “Obesity is similarly more complex than merely having high BMI.”

Because clinicians cannot realistically treat every patient with overweight or obesity using costly medications or surgery, identifying those at highest risk makes sense, he said.

Cornier also praised the inclusion of conditions beyond traditional metabolic complications, such as arthritis and obstructive sleep apnea.

However, he questioned how broadly the findings can be applied.

“The study drew on participants in the UK Biobank, and we’d probably want to have the risk predictor validated in the United States before regarding it as ready for prime time here,” he said.

The study was supported by the British Heart Foundation, the German Centre for Cardiovascular Research, UK Research and Innovation, the UK National Institute for Health and Care Research, the European Research Council, and the German Research Foundation. Additional funding sources are listed in the original article.

Demircan reported no relevant financial relationships. Disclosures for the remaining study authors are available in the original article.

Cornier reported receiving research grants from Novartis, Kaneka, Ionis, and Cleerly; serving as a consultant or advisory board member for AstraZeneca, Biophytis, Bonus Health, Enveda, Kailera Therapeutics, Keros Therapeutics, Lilly, Novo Nordisk, Wave, Zealand Pharma, and ZyVersa; and serving on a data safety monitoring board for Advarra.

Batya Swift Yasgur, MA, LSW is a freelance writer with a counseling practice in Teaneck, New Jersey. She is a regular contributor to numerous medical publications, including Medscape and WebMD, and is the author of several consumer-oriented health books as well as Behind the Burqa: Our Lives in Afghanistan and How We Escaped to Freedom (the memoir of two brave Afghan sisters who told her their story).