Extra Pounds May Defy Expectations in Heart Failure, With or Without Diabetes

For decades, patients with heart failure have been told that carrying extra weight is a liability, a metabolic burden that accelerates the very disease threatening their hearts. Yet a growing body of evidence keeps pointing in a direction that unsettles conventional wisdom: among people already living with heart failure, those with a higher body mass index often outlive their leaner counterparts. Now, one of the largest analyses ever conducted in an East Asian population suggests that this so-called obesity paradox holds firm even in the sickest and most vulnerable subgroup of patients—those who also have diabetes. The finding, published in Clinical Research in Cardiology, comes from a pooled retrospective cohort of more than 8,300 hospitalized heart failure patients in South Korea and carries implications for how clinicians weigh the risks and benefits of weight-lowering therapies in a population where every kilogram may matter more than previously assumed.

The study drew on two prospective, multicenter Korean registries: the Korea Heart Failure Registry, which enrolled 3,200 patients hospitalized for acute heart failure at 24 tertiary centers between 2004 and 2009, and the Korean Acute Heart Failure Registry, which captured 5,625 patients across 10 tertiary centers between 2011 and 2014. After excluding 442 patients with missing body mass index or diabetes data, the researchers analyzed 8,383 individuals with a mean age of 68.2 years, 52 percent of whom were men. Importantly, the team applied an Asian-specific threshold for overweight, classifying patients as lean if their body mass index fell below 23 kilograms per square meter and overweight if it met or exceeded that value—a cutoff grounded in evidence that cardiometabolic risk rises at lower body weights in Asian populations than in Western ones. Roughly a third of the cohort had diabetes, and the patients were sorted into four groups: overweight non-diabetic, overweight diabetic, lean non-diabetic, and lean diabetic.

The baseline profiles of these groups told a striking story before any outcome data were even examined. Overweight patients without diabetes had the most favorable characteristics: they were younger, less likely to have ischemic heart disease, and displayed lower creatinine, C-reactive protein, and natriuretic peptide levels, along with better echocardiographic measures of cardiac function. At the opposite extreme, lean patients with diabetes carried the heaviest burden of disease, with more advanced heart failure symptoms, greater renal dysfunction, and a higher prevalence of ischemic etiology. That gradient—healthiest at the heavier, non-diabetic end of the spectrum and sickest at the lean, diabetic end—set the stage for outcome differences that were as dramatic as they were consistent.

During the index hospitalization, 5.3 percent of all patients died, with in-hospital mortality peaking at 6.9 percent among lean diabetic patients and falling as low as 4.2 percent among overweight diabetic patients. But it was the post-discharge picture that revealed the paradox in its sharpest form. Over one year of follow-up, 14.9 percent of patients died overall, yet the mortality gradient across the four groups was steep: 9.3 percent among overweight non-diabetic patients, 12.7 percent among overweight diabetic patients, 16.4 percent among lean non-diabetic patients, and a sobering 20.7 percent among lean diabetic patients. Diabetes alone raised the one-year mortality rate from 13.8 to 17.1 percent, but leanness exerted an even larger penalty, lifting mortality from 11.1 to 18.8 percent.

After multivariable adjustment for age, sex, systolic blood pressure, symptom severity, hemoglobin, renal function, ejection fraction, ischemic etiology, and other confounders, the survival advantage of overweight status remained statistically robust for one-year mortality. Overweight diabetic patients faced a 35 percent lower risk of death within a year compared with their lean diabetic counterparts, with an adjusted hazard ratio of 0.65 and a 95 percent confidence interval of 0.53 to 0.79. The same directional benefit appeared among overweight patients without diabetes. Critically, when the researchers tested for an interaction between body mass index and diabetes status, none was found—the protective association held with equal force whether or not a patient had diabetes. When body mass index was modeled as a continuous variable, each single-unit increase was associated with a 7 percent reduction in one-year mortality risk, and restricted cubic spline analysis revealed a smooth, monotonic decline in hazard across the entire observed range of body weight.

Intriguingly, the same pattern did not survive scrutiny for in-hospital mortality. Although the raw death rates favored heavier patients during the admission itself, the association weakened and lost statistical significance once the fully adjusted model accounted for the less favorable baseline profiles of lean patients—including their more advanced heart failure and greater comorbidity burden. This divergence suggests that body mass index carries greater prognostic weight over the months following discharge than during the acute phase, when illness severity and body composition may dominate short-term outcomes more than adiposity itself. In other words, the obesity paradox appears to be primarily a phenomenon of medium-term survival rather than of immediate peri-hospitalization risk.

The research team took considerable care to test whether their findings could be an artifact of statistical fragility. Sensitivity analyses using the World Health Organization overweight cutoff of 25 kilograms per square meter produced nearly identical results, with mortality falling monotonically across body mass index strata from 26.9 percent in the underweight range to just 7.0 percent among those with a body mass index of 30 or higher. Excluding patients with a body mass index below 20—who are most likely to suffer from cardiac cachexia, frailty, or severe malnutrition—did not erase the survival advantage of overweight status, arguing against reverse causality as the sole explanation. Registry-stratified analyses showed consistent associations in both cohorts, with no significant registry-by-body-mass-index interaction, and the pattern held across all three heart failure phenotypes: reduced, mildly reduced, and preserved ejection fraction. Only the smaller mid-range ejection fraction subgroup failed to reach statistical significance, most likely due to insufficient sample size.

Still, the investigators are careful to emphasize the limits of what an observational study can prove. Reverse causality—the idea that leanness reflects advanced disease rather than causing poor outcomes—cannot be dismissed entirely, and the registries lacked longitudinal weight trajectories, body composition data, nutritional markers, and measures of skeletal muscle mass that might distinguish protective adiposity from simply the absence of wasting. The cohort was also exclusively Korean, with a low mean body mass index of 23 kilograms per square meter, raising questions about generalizability to Western populations with very different body composition distributions. HbA1c data were missing in more than three-quarters of patients, diabetes type and duration were not captured, and the cohort predates the widespread use of sodium-glucose cotransporter-2 inhibitors, so the interplay between modern weight-lowering drugs and the paradox remains unresolved. The authors frame their conclusions explicitly as associations, not evidence of a causal protective effect of being overweight.

Yet the clinical implications are difficult to ignore, particularly at a moment when guideline-directed heart failure therapy increasingly includes drugs that cause weight loss. Sodium-glucose cotransporter-2 inhibitors can reduce body weight by 3 to 5 percent, and while trial-level evidence from DAPA-HF and EMPEROR-Reduced suggests these therapies benefit patients across the body mass spectrum without modifying the paradox, the new findings add urgency to the question of whether aggressive weight reduction is appropriate for lean patients with heart failure. Notably, a prespecified analysis of the STEP-HFpEF trial showed that semaglutide improved symptoms and function in overweight patients with preserved ejection fraction in proportion to the weight they lost—hinting that there may be a therapeutic sweet spot, a window of body weight beyond which the prognostic advantage of adiposity diminishes. For now, the message from Korea is unambiguous: among patients hospitalized with heart failure, low body weight—especially when combined with diabetes—identifies a strikingly vulnerable phenotype, and the obesity paradox is no respecter of diabetes status.

Subject of Research: The obesity paradox in heart failure patients with and without diabetes mellitus

Article Title: Obesity paradox in heart failure with and without diabetes mellitus: a retrospective cohort study

Article References: Kwon, O., Yoon, M., & Park, J. J. (2026). Obesity paradox in heart failure with and without diabetes mellitus: a retrospective cohort study. Clinical Research in Cardiology. https://doi.org/10.1007/s00392-026-03007-0

Image Credits: AI Generated

DOI: 10.1007/s00392-026-03007-0

Keywords: obesity paradox, heart failure, diabetes mellitus, body mass index, mortality, KorHF, KorAHF, prognosis, HFrEF, HFpEF, SGLT2 inhibitors, cardiology