Article Text

Download PDFPDF
Effects of waist to height ratio, waist circumference, body mass index on the risk of chronic diseases, all-cause, cardiovascular and cancer mortality
  1. Kenneth Lo1,2,
  2. Yu-Qing Huang1,
  3. Geng Shen1,
  4. Jia-Yi Huang1,
  5. Lin Liu1,
  6. Yu-Ling Yu1,
  7. Chao-Lei Chen1,
  8. Ying Qing Feng1
  1. 1 Department of Cardiology, Guangdong Cardiovascular Institute, Hypertension Research Laboratory, Guangdong Provincial People's Hospital, Guangdong Provincial Key Laboratory of Coronary Heart Disease Prevention, Guangdong Academy of Medical Sciences, South China University of Technology School of Medicine, Guangzhou, China
  2. 2 Centre for Global Cardiometabolic Health, Department of Epidemiology, Brown University, Providence, Rhode Island, USA
  1. Correspondence to Dr Kenneth Lo, Centre for Global Cardiometabolic Health, Department of Epidemiology, Brown University, Providence, RI 02912, USA; kenneth_lo{at}brown.edu

Abstract

Background Given the fat redistribution in later stages of life, how the associations between abdominal obesity and the risk of morbidity and mortality have changed with age have not been elucidated, especially for waist to height ratio (WHtR).

Objective To compare the strength of association between obesity indices and chronic diseases at baseline, and the subsequent mortality risk among US adults.

Methods We included 21 109 participants from National Health and Nutrition Examination Survey 1999–2014. We performed logistic regression and receiver operating curve analysis to examine the discriminatory power of obesity indicators on cardiometabolic diseases and cancer at baseline. Sex-stratified and age-stratified Cox models were constructed to explore the prospective association between obesity indices and all-cause, cardiovascular and cancer mortality.

Results Elevated WHtR, elevated waist circumference (WC) and body mass index (BMI)-classified obesity are associated with higher odds of hypertension (OR: 1.37-2.13), dyslipidemia (OR: 1.06 to 1.75, all p<0.05) and diabetes (OR: 1.40-3.16, all p<0.05). WHtR had significantly better discriminatory power to predict cardiometabolic health than BMI, especially for diabetes (area under the curve: 0.709 vs 0.654). After multivariable adjustment, all obesity indicators are associated with lower risk of all-cause mortality among females aged ≥65 years (HR: 0.64 to 0.85), but the association was only significant for BMI when obesity indicators were mutually adjusted (HR: 0.79).

Conclusions WHtR and WC appeared to be the better indicators for cardiometabolic health than BMI. However, BMI had a stronger and inverse association with a greater risk of all-cause mortality among older females.

  • epidemiology

Data availability statement

Data are available in a public, open access repository.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Cardiovascular diseases (CVDs) and cancer confer a significant global public health burden.1 2 Valid and easy-to-measure indicators are needed to screen for people with elevated cardiometabolic and cancer risk. Anthropometric indices are easily measured in clinical setting and attract much attention in evaluating their screening power. Although body mass index (BMI) is widely used due to easy measurement, accumulating evidence suggested that surrogates for abdominal obesity, such as waist to height ratio (WHtR) and waist circumference (WC), have better screening power for cardiometabolic risk.3 Meta-analyses of receiver operating characteristic curves showed that WHtR had better power than BMI and WC in classifying CVD risk factors among adults and children.3 4 After the publication of these two meta-analyses, an analysis of baseline data from a Chinese prospective cohort also suggested WHtR to be best indicator for dyslipidaemia and hyperglycaemia when compared with BMI and WC.5

Moreover, prospective studies have demonstrated the predictive power of obesity indices on long-term disease risks. For example, a higher WC has been associated with higher all-cause and CVD mortality regardless of BMI categories.6 7 Results from the Nurses’ Health Study also supported that elevated WC positively associated with all-cause, CVD and cancer mortality independently of BMI.8 Among participants from the Alberta’s Tomorrow Project cohort, WC attenuates the association between BMI and the risk of all-cancer incidence, especially among females.9 Meanwhile, higher WHtR may associate with a higher risk of diabetes,10 CVD morbidity11 and liver cancer.12 When compared with WC, WHtR has accounted for the variations in height. Height has found to associate with a lower risk of cardiometabolic morbidity and mortality.13 Recently, a cohort analysis among 109 536 postmenopausal women has shown a higher risk for CVD events when WHtR was >0.5 (HR: 1.29). The magnitude of association was comparable to WC (HR: 1.23) and probably stronger than BMI-classified obesity (HR: 1.19).14

Another issue is the sex-specific and age-specific association with survival risk for elevated WHtR, WC and BMI. Researchers proposed a universal 0.5 value as cut-off point for WHtR regardless of age, sex and ethnicity.15 Although sex-specific thresholds do exist, the recommendations on elevated WC and BMI are independent on age for adults aged ≥18 years. However, people aged >60 years face the decrease of total body mass and redistribution of body fat.16 Some studies suggested that BMI-classified overweight or obesity is associated with a lower morality risk among elderly aged over 70 years.17 18 This controversial relationship has been termed as ‘obesity paradox’.19 Although WC is a better indicator for visceral fat than BMI,20 few studies have examined the relationship between obesity and mortality among elderly using WC, needless to say the use of WHtR. For the studies that were conducted among elderly, comparison with the results among middle-aged participants in the same population is lacking.

To the best of our knowledge, few studies have examined the predictive power of WHtR on long-term survival status and compared it with BMI and WC among general population. How the associations differed by sex and age groups are also unclear. In the present study, we addressed the research gap by examining the association of WHtR, WC, BMI with the odds of chronic diseases at baseline and the risk of all-cause, cardiovascular and cancer mortality among US adults who participated in 1999–2014 National Health and Nutrition Examination Survey (NHANES).

Methods

Study design

In the present study, we have used the data from the 1999 to 2014 NHANES, which is an ongoing survey conducted by the Center for Disease Control and Prevention designed to evaluate the health status of US adults. Participants in this study were followed through 31 December 2015. Written informed consent was obtained from all participants. We excluded participants with incomplete exposure and covariate data, did not have available mortality status and had unreliable dietary recall status. We also excluded few participants (<2%) with BMI-classified underweight.

Exposure assessment

The exposure variables were WHtR, WC and BMI. Anthropometric measurements were performed using standardised procedures.21 In brief, a stadiometer was used for height measurement, and an electronic digital scale to measure body weight. WC was measured at the high point of the iliac crest at minimal respiration. BMI was also calculated for statistical adjustment in regression model. WC and WHtR were dichotomised by the common cut-off points among adult population. In specific, the threshold values for elevated WC was 102 cm for men and 88 cm for women,22 while the value for WHtR ratio was 0.5.15 Participants with BMI greater than 25 were classified as overweight, while those with BMI greater than 30 were classified as obese (https://www.cdc.gov/obesity/adult/defining.html). BMI was also treated as continuous variable to evaluate the effect estimates for every 5-unit increment.

Outcomes ascertainment

Mortality data of the people participating in NHANES 1999–2014 survey was linked to death certificate data from the National Death Index. The causes of death were classified into ‘Diseases of heart’, ‘Malignant neoplasms’, ‘Chronic lower respiratory diseases’, ‘Accidents (unintentional injuries)’, ‘Cerebrovascular diseases’, ‘Alzheimer’s disease’, ‘Diabetes mellitus’, ‘Influenza and pneumonia’, ‘Nephritis, nephrotic syndrome and nephrosis’ and ‘All other causes (residual)’. The outcomes of interest in the present study were all-cause mortality, cardiovascular mortality (combining deaths due to diseases of the heart and cerebrovascular diseases) and cancer mortality (Malignant neoplasms). Mortality data for participants younger than 18 years were not available for public release.

Covariates for analysis

Covariates were assessed at baseline for each survey, which were chosen based on established associations with mortality. These included age, gender, race/ethnicity (dichotomised into non-Hispanic white vs others), education level (dichotomised into college graduate or above vs others), marital status (married vs others), had at least 100 cigarettes in lifetime (yes vs no), time spent on physical activity in hours per week, family poverty income ratio, systolic and diastolic blood pressure (average values of four measurements), self-report of whether a doctor had informed the respondent of a diagnosis of hypertension, dyslipidaemia, diabetes, prior diagnosed congestive heart failure, coronary heart disease, angina/angina pectoris, heart attack or stroke, and/or cancer, dietary energy in kilocalories, protein/carbohydrates/fat/alcohol use in grams (all assessed by 1-day food record). Same sets of covariates were included for both logistic and Cox regression analyses.

Statistical analysis

Continuous data were reported as means and SD, and categorical data as frequencies with percentages. Means and percentages by mortality status (alive vs deceased) were compared using independent t-tests and χ2 tests, respectively. To examine the cross-sectional association between obesity indicators and the likelihood of chronic diseases at baseline, we built logistic regression models for each of the diseases (hypertension, dyslipidaemia, diabetes, heart disease or cancer) and were adjusted for other diseases. For instance, in the regression model that evaluated the relationship between WC and hypertension, diagnosed dyslipidaemia, diabetes, heart disease or cancer were treated as covariates in the same model. When heart disease was treated as outcome, it was defined by any prior diagnosed congestive heart failure, coronary heart disease, angina/angina pectoris, heart attack or stroke. For each obesity indicators, we used receiver operator characteristic curve analysis to identify the cut-off value with the maximum sum of sensitivity and specificity to predict each chronic disease, as reflected by the highest Youden Index (sensitivity+specificity-1). In addition, we have used multivariable Cox proportional hazard models to examine the associations between each obesity indicators, namely WC and WHtR with all-cause, cardiovascular and cancer mortality. Person-years was calculated from completion of the baseline examination to the confirmation of mortality, loss to follow-up or 31 December 2015, whichever came first. We have adjusted for the association between obesity indices and mortality using the covariates mentioned in previous paragraph.

We conducted stratified analysis for association analyses by genders and groups (males aged <65 years, males aged ≥65 years, females aged <65 years, females aged ≥65 years). To examine if the association between abdominal obesity and the risk of disease was dependent on general obesity, sensitivity analysis was performed by mutually adjusting WHtR and BMI or WHtR and BMI in the same logistic regression or Cox regression models. All analysis was conducted by SPSS V.23.0 (SPSS, Chicago, Illinois, USA) with complex sampling to avoid sampling bias, unequal probability of selection and oversampling. Statistical significance level was set at 0.05.

Results

Participant characteristics

A total of 21 109 participants in NHANES with complete data and reliable dietary recall status were included in our analysis. Table 1 has compared the characteristics between participants who were alive versus deceased. In general, significant differences were found across all variables. For example, people who were deceased had shorter follow-up period, older age, higher obesity rate and higher rate of chronic diseases at baseline. The only exception was the insignificant difference in the rate of WC-classified obesity.

Table 1

Characteristics of included participants

Table 2 has demonstrated the association between each obesity index and the presence of chronic diseases at baseline. Among all participants, elevated WHtR (OR: 2.13, 95% CI=1.90 to 2.40), elevated WC (OR: 2.01, 95% CI=1.87 to 2.17) and BMI-classified obesity (OR: 2.82, 95% CI=2.59 to 3.10) are associated with higher chance of hypertension. Similarly, elevated WHtR (OR: 1.75 and 3.16) and WC (OR: 1.29 and 2.27), as well as BMI-classified obesity (OR: 1.40 and 2.88) are associated with a significantly higher likelihood of dyslipidaemia and diabetes among all participants. None of the indicators have significant association with heart diseases, but BMI-classified obesity is associated with lower likelihood of cancer among overall participants (OR: 0.88, 95% CI=0.77 to 0.99). Moreover, per 5-unit increment in BMI is associated with higher chance of all chronic diseases except for cancer. In stratified analysis (online supplementary table 1), WHtR, WC and BMI remained significant associations with hypertension, dyslipidaemia and diabetes among most subgroups. Elevated WHtR (OR: 0.68, 95% CI=0.48 to 0.97) and BMI-classified obesity (OR: 0.57, 95% CI=0.42 to 0.78) are associated with lower chance of cancer among males aged <65 years. Every 5-unit increment in BMI is associated with higher chance of chronic diseases in most subgroups, except for lower likelihood of cancer among males aged <65 years.

Table 2

Adjusted OR for the association between obesity indices and the odds of chronic disease at baseline

Table 3 presents the area under the curve (AUC) with optimal cut points for obesity indices in relation to cardiometabolic parameters and cancer. WHtR (cut-off: 0.610, sensitivity: 0.665, specificity: 0.649), WC (cut-off: 99.05 cm, sensitivity: 0.720, specificity: 0.559) and BMI (cut-off: 29.14 kg/m2, sensitivity: 0.624, specificity: 0.605) demonstrated the best discriminatory power when screening for diabetes (AUC ranged from 0.654 to 0.709), followed by hypertension (AUC ranged from 0.632 to 0.675). With regard to the differences in 95% CI, WHtR and WC had significantly higher AUC than BMI for all parameters. Although WHtR had higher AUC than WC for all parameters except cancer, the difference was not significant judging from the overlapping of 95% CI.

Table 3

AUC and optimal cut-off points for obesity indices in relation to chronic diseases

Table 4 has demonstrated the association between each obesity index and mortality risk. Among all participants, elevated WHtR (OR: 0.83, 95% CI=0.70 to 0.98), elevated WC (OR: 0.86, 95% CI=0.78 to 0.95) and BMI-classified obesity (OR: 0.77, 95% CI=0.68 to 0.87) are associated with a lower risk of all-cause mortality. In stratified analysis (online supplementary table 2), elevated WHtR (OR: 0.68, 95% CI=0.50 to 0.94), elevated WC (OR: 0.72, 95% CI=0.59 to 0.89) and BMI-classified obesity (OR: 0.64, 95% CI=0.50 to 0.82) are associated with a lower risk of all-cause mortality among females aged ≥65 years. Moreover, BMI-classified obesity was associated with a lower risk of cancer mortality among males aged <65 years (OR: 0.53, 95% CI=0.31 to 0.89). Per 5-unit increment of BMI is associated with lower risk of all-cause mortality among females aged ≥65 years (OR: 0.85, 95% CI=0.78 to 0.93).

Table 4

Association between obesity indicators and the risk of all-cause, cardiovascular and cancer mortality

The results of sensitivity analysis were presented in online supplementary table 3 and online supplementary table 4. Although the strength of most associations attenuated, all obesity indices had significant associations with the odds of hypertension, dyslipidaemia and diabetes. After mutual adjustments, only BMI-classified obesity is associated with a lower risk of all-cause mortality (HR: 0.79, 95% CI=0.66 to 0.93) after adjusting for WC.

Discussion

Using the data from NHANES study 1999–2014, we found moderate discrimination power for obesity indices to predict cardiometabolic outcomes, especially for diabetes. Although WHtR and WC did not differ significantly in screening power, both performed better than BMI. In addition, we examined the association between obesity indices and the risk of all-cause, cardiovascular and cancer mortality among US adults. After adjusting for age, sex, lifestyle factors and disease history of participants, we found that while all obesity indices are associated with a lower risk of all-cause mortality, the magnitude of association was stronger among females aged ≥65 years. After mutual adjustments for obesity indicators, the strength of most associations attenuated but remained statistically significant.

Abdominal obesity is expected to be important in cardiometabolic disease aetiology because of the increased fatty acid release into portal system,23 which might explain the better discriminatory power of cardiovascular risk factors than general obesity (ie, BMI).3 There are several advantages of WHtR compared with BMI and WC in identifying CVD risk. BMI may be less predictive than WHtR because it is a less specific measure of abdominal adiposity, and BMI does not equally reflect the specific changes in body fat, while WHtR specifically assesses abdominal fat.24 25 In addition, WC may be insensitive to differences in cardiovascular risk among people with the same WC but different heights.

Moreover, we found that elevated WHtR, WC and BMI-classified obesity are associated with a lower risk of all-cause mortality among overall participants. In stratified analysis, the association was only significant among females aged ≥65 years in stratified analysis. This finding seemed contrast to the fact that it has better screening power for adulthood cardiometabolic risk factors than BMI and WC.3 This seemingly paradoxical relationship between obesity and disease risk agrees with the aforementioned ‘obesity paradox’. In a study conducted among 4361 Chinese those aged ≥80 years,26 the lowest tertile of WC is associated with a 41% and 55% greater risk of 3-year all-cause mortality among men and women, respectively. Authors postulated that lower BMI and WC could be the risk factors of sarcopaenia,27 and less adipose tissues to release hormones for anti-inflammatory effect28 and tissue regeneration.29 These effects were collectively unfavourable to survival, and therefore the obesity-associated risk might decline with age.30 It is notable that the associations were not significant among other subgroups. It is possible that elderly is more vulnerable to diseases, hence obesity status can have a greater impact on their mortality risk. Another explanation is that the female-specific disorders, including gestational diabetes, polycystic ovary syndrome and menopause, make them more vulnerable to cardiometabolic diseases than males.31 Furthermore, when WHtR or WC were mutually adjusted for BMI, no associations with mortality were significant. In contrast, BMI remained an inverse association with the risk of all-cause mortality (online supplementary table 4). The results suggested that the overall fat distribution accounted for the obesity paradox instead of abdominal fat only, which warrants further studies to verify this explanation.

One interesting observation is that BMI-classified obesity is associated with a lower odd of cancer at baseline (OR: 0.57) and a lower risk of cancer mortality (OR: 0.53) among males aged <65 years. It is different from the findings as mentioned above, where obesity appears to be beneficial for survival among elderly. For cross-sectional analysis, reverse causation might exist since patients with cancer are likely to suffer from weight loss.32 Since NHANES only collected data from participants at baseline, we were not able to examine the weight changes and how it might affect the length of survival time.

The strength of this study was to include a prospective study design with linkage to national mortality data, which helped to elucidate prospective relationship. However, there were several limitations that should be aware of. First, most of the covariates were self-reported, which might introduce inaccuracy and misclassification. Second, there were residual confounding effects that we might not be able to account for, such as mental health and medication use. Third, we were not able to examine the predictive ability of hip circumference and waist to hip ratio on mortality risk in NHANES 1999–2014 because it was not examined in every year. Fourth, all NHANES data were collected once at baseline, and it is unclear how the changes in exposure over time may influence the association. Despite these limitations, this study adds important evidence to understanding of the usefulness of obesity indices in identifying people at higher mortality risk.

Conclusions

WHtR and WC had better discriminatory power than BMI in predicting cardiometabolic risk factors, especially diabetes. However, BMI is associated with a lower incidence of all-cause mortality among females aged ≥65 years. Excess adiposity may not be as detrimental among elderly comparing with middle-aged population.

Main messages

  • Elevated waist to height ratio (WHtR), waist circumference (WC) and body mass index (BMI) are associated with higher odds of hypertension, dyslipidaemia and diabetes.

  • WHtR had significantly better discriminatory power to predict cardiometabolic health than BMI, especially for diabetes.

  • All obesity indicators are associated with lower risk of all-cause mortality among females aged ≥65 years, but the association was only significant for BMI when obesity indicators were mutually adjusted.

Current research questions

  • Are elevated WHtR, WC and BMI associated with chronic diseases at baseline?

  • Are elevated WHtR, WC and BMI associated with long-term mortality risk?

  • Is there an age and gender difference in the relationship between obesity indices and disease risk?

What is already known on the subject

  • Obesity is closely related to the risk of chronic diseases.

  • Abdominal obesity may have more adverse effects on health than general obesity.

  • WHtR is an emerging indicator to screen for individuals with chronic diseases.

Data availability statement

Data are available in a public, open access repository.

Ethics statements

Ethics approval

Ethical approval was obtained from National Center for Health Statistics Ethics Review Board (Protocol Nos 98-12, 2005-06 and 2011-17).

Acknowledgments

We sincerely thank all the NHANES participants for their important contributions.

References

Footnotes

  • Contributors KL, Y-QH, GS searched the literature, analysed and interpreted the data, and wrote the manuscript. J-YH, LL, Y-LY, C-LC, Y-QF participated in the study design; collected, analysed and interpreted the data; and wrote the manuscript.

  • Funding This work was supported by the Natural Science Foundation of Guangdong Province (No 2015A030313660), the Science and Technology Plan Project of Guangdong Province (No 2014B020212008), the Science and Technology Program of Guangzhou (No 201604020143, No 201604020018, No 201604020186, No 201510010254 and No 201803040012), and the National Key Research and Development Program of China (No 2017FYC1307603, No 2016YFC1301305 and No 2017YFC0909303).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.