BMI, waist circumference not a reliable measure of obesity

There appears to be an overall dearth of scientific evidence to support the use of anthropometric tools—such as body mass index (BMI) and waist circumference (WC)—in determining obesity, a recent meta-analysis has found.

From the databases of Ovid Medline, Embase, CINAHL, and PubMed, researchers retrieved 32 studies eligible for analysis. Of these, 27 investigated BMI while 15 assessed waist measurements, such as WC, waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR). Six of the included studies had low risk of bias, while 10 showed a high risk; the remaining 16 had an unclear risk of bias.

A pooled analysis of 16 studies (n=14,008 women) showed that the combined sensitivity of BMI to detect obesity was 51.4 percent, with a specificity of 95.4 percent. For this analysis, BMI thresholds ranged from 25–30 kg/m². Excluding studies with a high risk of bias slightly reduced both sensitivity and specificity, but otherwise had no major impact on the primary findings.

In men, a meta-analysis of 12 studies (n=11,320 men) revealed a combined BMI sensitivity of 49.6 percent for identifying obesity and a specificity of 97.3 percent. Cut-offs for this analysis were similar, at 25–30 kg/m². While sensitivity measures varied greatly across studies, excluding studies with a high risk of bias likewise had minimal effect on the overall estimates.

WC was also underpowered to identify obesity, both in women (sensitivity: 62.4 percent; specificity: 88.1 percent) and in men (sensitivity: 57.0 percent; specificity: 94.8 percent). Cut-off ranges were 65.8–107 and 78.9–105 cm, respectively.

The data for both WHR and WHtR were insufficient for meta-analysis.