Waist Measure Improved Mortality Prediction in Older Women

By using waist circumference along with BMI, modest gains were made in predicting which postmenopausal women were at a higher risk for premature death, a prospective cohort study has found.

While obesity remains a pressing public health threat, prevention and treatment are made difficult by the limits of the BMI to accurately measure excess adiposity, according to the study’s authors.

“Waist circumference is a simple, inexpensive method to assess visceral adiposity and correlates well with visceral fat as assessed by imaging,” Aaron K. Aragaki, MS, and his colleagues wrote. Aragaki is a researcher at the Fred Hutchinson Cancer Center in Seattle.

The problem, wrote Aragaki and his colleagues, is that waist circumference is not often measured in the clinical setting. Further, “Waist circumference would be especially valuable for staging of obesity risk if BMI-specific thresholds were available,” they wrote.

The investigators said that the current recommended waist circumference thresholds do not complement BMI because most adults diagnosed with obesity are already beyond these thresholds. Currently accepted waist circumference thresholds for overweight are 88 cm or above for women and 102 cm or above for men.

To stratify risk more accurately, new BMI-specific waist circumference threshold recommendations have been made recently by the International Atherosclerosis Society (IAS) and the International Chair on Cardiometabolic Risk (ICCR) Working Group on Visceral Obesity. They have proposed thresholds in women of 80 cm or greater, 90 cm or greater, 105 cm or greater, 115 cm or greater, and 115 cm or greater to correlate, respectively, with BMI categories: normal weight (18.5 to < 25), overweight (25 to < 30), obesity 1 (30 to < 35), obesity 2 (35 to < 40), and obesity 3 (≥ 40).

In men, they propose the corresponding waist measurements of 90 cm or greater, 100 cm or greater, 110 cm or greater, 125 cm or greater, and 125 cm or greater to correlate with the respective BMI categories.

A consensus statement issued in 2020 by the IAS and the ICCR suggested that prospective data were needed to further stratify BMI categories by waist circumference thresholds to improve mortality risk prediction compared with BMI categories.

To that end, Aragaki and his colleagues examined all-cause mortality data from the Women’s Health Initiative (WHI), a large, national, multicenter, population-based study of generally healthy postmenopausal women (aged between 50 and 79 years), with enrollment from 1993 to 1998 and follow-up through 2021.

Data of 139,213 women from the WHI were distributed across three groups. There was a development cohort (N = 67,774) and two validation cohorts, one according to overweight and obesity prevalence, and another according to geography and diversity.

Validation Cohort 1 (N = 48,335) had half the prevalence of prior cardiovascular disease and cancer as the development group, but a higher rate of overweight or obesity (72.6% vs 59.1%). Validation Cohort 2 (N = 23,104) had nearly twice the number of women who identified as Black compared with the development group (12.7% vs 6.4%) and nearly three times the number of women who identified as Hispanic (9.1% vs 3.1%).

Aragaki and his colleagues analyzed annual all-cause mortality in the datasets and the National Death Index. They found that respective death rates at 10- and 20-year follow-up marks were 5808 and 22,124 in the development cohort; 3418 and 14,252 in Validation Cohort 1, and 2123 and 7668 in Validation Cohort 2.

Validation Cohort 1 had a higher prevalence of large waist circumference (21.9%) than Validation Cohort 2 (18.2%), according to currently accepted BMI-specific thresholds. Nearly all women with obesity 2 or obesity 3 had a waist circumference of 88 cm or larger, whereas women with a normal weight did not much exceed this waist circumference threshold.

Differences in the percentage of trunk fat between women with normal waist circumference and those with large waist circumference within the same BMI categories across centers were notable, ranging from 0.5 to 1.0 SDs.

Baseline risk for mortality at 10 and 20 years was highest in those with a prior history of disease and was lowest in those with no prior disease or preexisting condition across 5-year age groups, Aragaki and his colleagues found.

All estimated hazard ratios in the mortality model were associated with higher mortality risk compared with the baseline risk. When BMI categories > 30 were added to this model, they were positively associated with higher mortality risk in a dose-dependent manner compared with the BMI category of normal weight.

After stratification of BMI categories by waist circumference thresholds, hazard ratios were consistently greater for BMI categories with large waist circumference than their counterparts with normal ones. The mortality risk in women with normal weight or overweight and large waist circumferences was similar to that in women with obesity 1 and a normal waist circumference.

Mortality risk in those with obesity 1 and a large waist circumference was similar to that in women with obesity 3 and normal waist circumference.

Stratifying BMI categories by waist circumference thresholds improved mortality discrimination over BMI alone at 10 years for Validation Cohort 1, with the c-statistic increasing from 60.7% (95% CI, 59.5%-61.9%) to 61.3% (95% CI, 60.2%-62.5%), an improvement of 0.7% (95% CI, 0.3%-1.0%). The c-statistic is the agreement between the number of observed outcomes and predicted risk at 10 and 20 years, according to Aragaki and his colleagues.

Discrimination was not significantly improved for Validation Cohort 2, however, with a difference in c-statistics of 0.3% (95% CI, 0.2%-0.7%).

“Combining BMI and waist circumference assessments allows for more personalized decision making,” Aragaki and his colleagues wrote. “Patients with a large waist circumference in any BMI category are at elevated cardiometabolic and mortality risk and may benefit from more aggressive interventions than those with similar BMIs but lower waist circumferences. Conversely, those with a normal waist circumference may require less aggressive care.”

The results of the study reinforce other data in favor of supplementing BMI with a second metric to more completely assess cardiometabolic health, a clinical obesity specialist told Medscape Medical News.

“BMI is a flawed criteria for diagnosis of obesity as it reflects total body mass but does not distinguish between fat and lean mass or where fat is distributed,” said Reema Hamid Dbouk, MD. “Waist circumference is a surrogate for visceral fat, which is more metabolically active and linked to cardiometabolic risk and mortality.”

Dbouk is an assistant professor of obesity medicine at Emory School of Medicine and a Rollins Distinguished Clinician at the Emory Clinic, both in Atlanta.

“Waist circumference measurement is inexpensive, quick, and requires minimal training. The paper notes that embarrassment or discomfort among patients is low, particularly among older adults. Integration into EMRs and vital sign protocols could normalize its use. Patients may understand waist size more intuitively than BMI. Tracking waist circumference over time may be more encouraging, especially if muscle gain masks fat loss on the scale,” said Dbouk.

Dbouk said she thought the study was thorough, even if more data are still needed.

“Waist circumference-based thresholds may not fully account for differences in fat distribution by race/ethnicity or body habitus. For example, Asians may have higher metabolic risk at lower waist circumference. The thresholds are a step toward personalized risk assessment, but they may still require refinement to account for racial/ethnic differences in fat distribution and risk. More research is needed to validate waist circumference thresholds in broader populations, including men and younger individuals.”

The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services, through contracts 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, and 75N92021D00005. Conflicts of interest are listed online with the published paper. Dbouk reported having no disclosures.