General and abdominal adiposity and hypertension in eight world regions: a pooled analysis of 837 population-based studies with 7·5 million participants

Abstract

Background: Adiposity can be measured using BMI (which is based on weight and height) as well as indices of abdominal adiposity. We examined the association between BMI and waist-to-height ratio (WHtR) within and across populations of different world regions and quantified how well these two metrics discriminate between people with and without hypertension.

Methods: We used data from studies carried out from 1990 to 2023 on BMI, WHtR and hypertension in people aged 20-64 years in representative samples of the general population in eight world regions. We graphically compared the regional distributions of BMI and WHtR, and calculated Pearson's correlation coefficients between BMI and WHtR within each region. We used mixed-effects linear regression to estimate the extent to which WHtR varies across regions at the same BMI. We graphically examined the prevalence of hypertension and the distribution of people who have hypertension both in relation to BMI and WHtR, and we assessed how closely BMI and WHtR discriminate between participants with and without hypertension using C-statistic and net reclassification improvement (NRI).

Findings: The correlation between BMI and WHtR ranged from 0·76 to 0·89 within different regions. After adjusting for age and BMI, mean WHtR was highest in south Asia for both sexes, followed by Latin America and the Caribbean and the region of central Asia, Middle East and north Africa. Mean WHtR was lowest in central and eastern Europe for both sexes, in the high-income western region for women, and in Oceania for men. Conversely, to achieve an equivalent WHtR, the BMI of the population of south Asia would need to be, on average, 2·79 kg/m² (95% CI 2·31-3·28) lower for women and 1·28 kg/m² (1·02-1·54) lower for men than in the high-income western region. In every region, hypertension prevalence increased with both BMI and WHtR. Models with either of these two adiposity metrics had virtually identical C-statistics and NRIs for every region and sex, with C-statistics ranging from 0·72 to 0·81 and NRIs ranging from 0·34 to 0·57 in different region and sex combinations. When both BMI and WHtR were used, performance improved only slightly compared with using either adiposity measure alone.

Interpretation: BMI can distinguish young and middle-aged adults with higher versus lower amounts of abdominal adiposity with moderate-to-high accuracy, and both BMI and WHtR distinguish people with or without hypertension. However, at the same BMI level, people in south Asia, Latin America and the Caribbean, and the region of central Asia, Middle East and north Africa, have higher WHtR than in the other regions.

Funding: UK Medical Research Council and UK Research and Innovation (Innovate UK).

Figure 1. Distributions of BMI and waist-to-height ratio, by region

The black lines below each distribution show the 2·5%, 25·0%, 75·0%, and 97·5% quantiles of the distributions and the points show the median. The dashed lines show medians across all participants. Regions are ordered by their sex-specific median BMI. See appendix (p 55) for numerical summaries.

Figure 2. Relationship between waist-to-height ratio and BMI, by region

The shading indicates the density of participants in each region, with darker shades corresponding to more participants and vice versa. Pearson correlation coefficients between BMI and waist-to-height ratio in each region are shown in the panels. The vertical dashed line shows median BMI for all participants globally, the horizontal dashed line shows median waist-to-height ratio for all participants globally. See appendix (pp 85–87) for results using waist circumference.

Figure 3. Regional BMI adjustment

The BMI adjustment shows how much lower BMI in each region should be to achieve an equivalent waist-to-height ratio. The adjustment is shown relative to the population of the high-income western region where most current epidemiological studies have been done; regional ordering and differences across regions would be unchanged if a different reference were used. The bars show 95% CIs of the BMI adjustments. See appendix (pp 90–91) for results using waist circumference.

Figure 4. Prevalence of hypertension at different levels of waist-to-height ratio and BMI, by region

Cells with 30 or fewer participants have been excluded from the figure because the results are less stable than at larger numbers. The number on each panel indicates the crude prevalence of hypertension among all participants in each region. See appendix for separate results for untreated and treated hypertension (pp 98–103) and for results using waist circumference (pp 92–94, 104–109).

Figure 5. Distribution of participants with hypertension in relation to BMI and waist-to-height ratio, by region

The shading indicates the density of participants with hypertension in each region, with darker shades corresponding to more participants. The vertical lines show median BMI and horizontal lines show median waist-to-height ratio for all participants (dashed lines) and those with hypertension (solid lines) globally. See appendix for separate results for untreated and treated hypertension (pp 110–115) and for results using waist circumference (pp 95–97, 116–121).