Background Global evidence suggests that levels of waist circumference (WC) have increased independent of increases in body mass index (BMI) during the same time period, especially in women. Disease risk, especially those associated with the amount of visceral fat, is therefore underestimated by surveillance efforts that quantify secular change only in BMI. Our aim was to use nationally-representative Health Survey for England (HSE) data to quantify recent secular change in WC before- and after-adjustment for BMI, and explore sex-specific differences.

Methods We analysed data (1997 to 2018) among 125,052 adults aged 25–64 y with non-outlying BMI and WC. Means in BMI and WC were compared between the first and last years. Using all data to explore BMI and WC separately, age-adjusted linear and quantile regressions estimated change in means (linear trend) and at specified centiles of the distribution. BMI was then included as a predictor to examine change in WC independent of BMI. Sex-by-year interaction terms estimated whether secular change varied by sex; BMI-by-year interaction terms estimated whether change in WC varied by BMI level. Analyses adjusted for the complex survey design (StataV17.1).

Results Comparing 1997 and 2018, mean BMI increased more in women (1.3 kg/m2; 95%CI:1.0–1.7) than in men (0.7 kg/m2; 95%CI:0.4–1.0). Mean WC also increased more in women (5.5 cm; 95%CI:4.5–6.5) than in men (1.7 cm; 95%CI:0.6–2.7).

Exploring the distributions separately, BMI and WC increased at the middle- and upper-ends in both sexes, with larger increases at the upper-tails. Comparing 1997 and 2018, BMI at the 75th centile increased more in women (2.1 kg/m2) than in men (1.3 kg/m2; P=0.015 for sex-difference using Wald test); WC at the 75th centile also increased more in women (8.0 cm) than in men (2.8 cm; P<0.001).

After adjustment for BMI, the linear trend in WC reduced, but remained positive, in men at BMI=30 kg/m2 (β=0.03 cm/y; 95%CI:0.02–0.05). In women, the linear trend in WC reduced less strikingly after BMI-adjustment, and it increased monotonically with increasing BMI (e.g. no BMI-adjustment: β=0.27 cm/y; 95%CI:0.24–0.29; at BMI=30 kg/m2: β=0.21 cm/y; 95%CI:0.19–0.22). Sensitivity analyses showed that the same pattern was found using quantile regression: BMI-adjusted WC increased over time at the middle- and upper-tails in women but not in men, with larger increases with increasing BMI.

Conclusion Our analyses of twenty years of anthropometric data confirm emerging global evidence of increasing WC relative to BMI, especially in women. As suggested elsewhere, the potential exists for a higher disease burden than would be predicted by relying on BMI alone.