Conceived and designed the experiments: HP CB JAS TILN LG. Analyzed the data: HP CB TILN. Wrote the paper: HP JAS LG CB TILN.
The authors have declared that no competing interests exist.
Distribution of body fat is more important than the amount of fat as a prognostic factor for life expectancy. Despite that, body mass index (BMI) still holds its status as the most used indicator of obesity in clinical work.
We assessed the association of five different anthropometric measures with mortality in general and cardiovascular disease (CVD) mortality in particular using Cox proportional hazards models. Predictive properties were compared by computing integrated discrimination improvement and net reclassification improvement for two different prediction models. The measures studied were BMI, waist circumference, hip circumference, waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR). The study population was a prospective cohort of 62,223 Norwegians, age 20–79, followed up for mortality from 1995–1997 to the end of 2008 (mean follow-up 12.0 years) in the Nord-Trøndelag Health Study (HUNT 2).
After adjusting for age, smoking and physical activity WHR and WHtR were found to be the strongest predictors of death. Hazard ratios (HRs) for CVD mortality per increase in WHR of one standard deviation were 1.23 for men and 1.27 for women. For WHtR, these HRs were 1.24 for men and 1.23 for women. WHR offered the greatest integrated discrimination improvement to the prediction models studied, followed by WHtR and waist circumference. Hip circumference was in strong inverse association with mortality when adjusting for waist circumference. In all analyses, BMI had weaker association with mortality than three of the other four measures studied.
Our study adds further knowledge to the evidence that BMI is not the most appropriate measure of obesity in everyday clinical practice. WHR can reliably be measured and is as easy to calculate as BMI and is currently better documented than WHtR. It appears reasonable to recommend WHR as the primary measure of body composition and obesity.
It is well documented that distribution of body fat is more important than the amount of fat as a prognostic factor for metabolic disturbance, cardiovascular diseases (CVD) and life expectancy
In recent years an increasing amount of knowledge has been gathered regarding the metabolic basis for the special importance of central fat distribution. Various metabolic, endocrine and neural factors appear to influence where the body fat accumulates and how this affects the individual's physiology and disease risk
In the wake of the increasing prevalence of obesity in most parts of the world
The aim of this study was to further clarify the associations of anthropometric indicators of obesity and body composition (BMI, WHR, WHtR, waist circumference, and hip circumference) with overall mortality, and specifically with CVD mortality.
All adults aged 20 years or older and living in Nord-Trøndelag County in Norway were in 1995 to 1997 invited to participate at the second wave of the Nord-Trøndelag Health Study (HUNT 2). Overall, 74% of women (34,786) and 65% of men (30,575) chose to participate. The HUNT 2 population is ethnically homogenous (dominated by individuals of Nordic origin) and has been considered representative of the total Norwegian population regarding demography, socio-economic factors, morbidity and mortality, including mortality from CVD
For the purpose of the present analysis, 3,138 participants aged 80 years or more at baseline (1,231 men and 1,907 women) were excluded. Individuals with established CVD at baseline (self-reported myocardial infarction, stroke or angina pectoris) were excluded, 4,571 in total (2,780 men and 1,791 women), as well as 681 person with missing data on one or more of the following variables: height, weight, waist circumference, and hip circumference. Our calculations are thus based on information from 56,971 individuals (26,461 men and 30,510 women) aged 20–79 years who were without any known CVD at baseline. Baseline characteristics are depicted in the supporting information (
In the HUNT 2 study, height and weight were measured with participants wearing light clothes without shoes; height to the nearest 1.0 cm and weight to the nearest 0.5 kg. Based on these measures we calculated BMI as weight in kg divided by the squared value of height in meters. Waist and hip circumferences were measured with a steel band to the nearest 1.0 cm with the participant standing and with the arms hanging relaxed. The waist circumference was measured horizontally at the height of the umbilicus, and the hip circumference was measured likewise at the thickest part of the hip
In the present analysis smoking was defined as daily smoking of cigarettes, cigars or a pipe. Smoking status was defined as unknown, current smoker, former, or never smoker. Levels of recreational physical activity were defined as self-reported number of hours spent on hard or light activity during one week: no activity; <3 h light activity; ≥3 h light activity or <1 h hard activity; ≥1 h hard activity; unknown.
The personal identity number of Norwegian citizens enabled linkage of HUNT 2 participant data to the Cause of Death Registry at Statistics Norway (information on
Each participant in the HUNT study signed a written consent regarding the screening and the use of data for research purposes as well as to linking their data to other registers (subject to approval of the Norwegian Data Inspectorate). The study was approved by the Norwegian Data Inspectorate and by the Regional Committee for Ethics in Medical Research.
We used Cox proportional hazard models to compute hazard ratios (HRs) for overall mortality and CVD mortality associated with different levels of each anthropometric measure. Precision of the estimated associations was assessed by a 95% confidence interval. Departure from the proportional hazards assumption was evaluated by Schoenfeld residuals and log-minus-log plots. An interaction term between time and the appropriate variables was added to the model if the proportional hazards assumption did not hold.
We analysed the HR for participants with BMI below 18.5 kg/m2 (104 men and 314 women) for comparison with the other BMI categories but excluded them from further analysis due to the potential of reverse causality (a J-shaped mortality curve)
We calculated sex specific standard deviation (SD) scores for each of the anthropometric variables and estimated the HR associated with an increase of one SD.
We analysed the data separately for men and women, and all associations were adjusted for potential confounding effects of age, smoking status and recreational physical activity. We conducted sensitivity analyses involving three additional models (Model 2–4). Model 2 included the same covariates as the main model but excluded participants with unknown smoking status. Model 3 was adjusted for age, smoking, and physical activity (as our main model) in addition to self-reported diabetes mellitus and weekly alcohol consumption (abstinence, 0–2 glasses [units], 2.1–5 glasses, 5.1–8 glasses, >8 glasses). Model 4 was identical to our main model but excluded the first three years of follow-up to limit the reverse causality effect of undiagnosed diseases.
The “relative informativeness” of each anthropometric measure was evaluated by examining the contributions made to the χ2 likelihood ratio statistic in the Cox regression model compared with a model that only contained the confounders, as the χ2 statistic can be used as a measure of the improvement of goodness of fit
To further compare the predictive properties of the different anthropometric measures for CVD death, sex-specific net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were computed when adding each anthropometric measure to two different prediction models. Model A included age as the only predictive variable, while Model B included age, smoking status, systolic blood pressure, and total cholesterol. For each model three different NRI calculations were done, using two (<5%, ≥5%), three (<1%, 1–9%, ≥10%), and four (<1%, 1–4%, 5–9%, ≥10%) levels of risk of CVD death, respectively.
In addition, we conducted an analysis of the anthropometric measures stratified by age (above and below 60 years). Finally, mutually adjusted analyses were conducted for waist and hip circumference, as well as for BMI and WHR.
All statistical tests were two-sided and all analyses were performed using Stata for Windows (Version 11 StataCorp LP, TX, USA).
We present the risk of death from all causes and from CVD among men and women aged 20–79 (
All causes | Cardiovascular disease | ||||||
Anthropometric measures | No. of persons | No. of deaths | Adjusted |
|
No. of deaths | Adjusted |
|
|
|||||||
<18.5 |
104 | 31 | 2.48 (1.73–3.54) | 9 | 2.23 (1.15–4.33) | ||
18.5–24.9 | 9,575 | 970 | 1.00 (Reference) | 300 | 1.00 (Reference) | ||
25.0–29.9 | 13,138 | 1,320 | 0.86 (0.79–0.93) | 492 | 1.04 (0.90–1.21) | ||
30.0–34.9 | 3,154 | 445 | 1.10 (0.98–1.23) | 175 | 1.42 (1.17–1.71) | ||
≥35.0 | 490 | 70 | 1.34 (1.05–1.72) | 0.16 | 28 | 1.78 (1.20–2.64) | <0.001 |
per 5 kg/m2 | 26,357 | 2,805 | 1.04 (0.98–1.10) | 0.21 | 995 | 1.19 (1.08–1.30) | <0.001 |
per SD (3.4 kg/m2) | 26,357 | 2,805 | 1.02 (0.99–1.06) | 0.21 | 995 | 1.12 (1.06–1.20) | <0.001 |
|
|||||||
<80 | 1,882 | 116 | 1.17 (0.96–1.42) | 33 | 1.06 (0.74–1.53) | ||
80–89 | 9,466 | 723 | 1.00 (Reference) | 233 | 1.00 (Reference) | ||
90–99 | 10,378 | 1,134 | 1.01 (0.92–1.11) | 404 | 1.10 (0.93–1.29) | ||
100–109 | 3,625 | 588 | 1.11 (0.99–1.24) | 226 | 1.27 (1.05–1.53) | ||
≥110 | 1,006 | 244 | 1.64 (1.41–1.90) | <0.001 | 99 | 1.99 (1.56–2.53) | <0.001 |
per 10 cm | 26,357 | 2,805 | 1.11 (1.07–1.16) | <0.001 | 995 | 1.21 (1.13–1.29) | <0.001 |
per SD (9.1 cm) | 26,357 | 2,805 | 1.10 (1.06–1.14) | <0.001 | 995 | 1.19 (1.12–1.26) | <0.001 |
|
|||||||
<95 | 2,360 | 275 | 1.18 (1.02–1.36) | 80 | 0.96 (0.75–1.25) | ||
95–99 | 6,158 | 639 | 1.00 (Reference) | 225 | 1.00 (Reference) | ||
100–104 | 9,203 | 925 | 0.89 (0.80–0.99) | 335 | 0.92 (0.77–1.09) | ||
105–109 | 5,471 | 546 | 0.86 (0.76–0.96) | 200 | 0.89 (0.73–1.08) | ||
≥110 | 3,165 | 420 | 1.17 (1.03–1.33) | 0.52 | 155 | 1.23 (1.00–1.51) | 0.24 |
per 10 cm | 26,357 | 2,805 | 1.01 (0.95–1.07) | 0.76 | 995 | 1.11 (1.00–1.23) | 0.05 |
per SD (6.2 cm) | 26,357 | 2,805 | 1.01 (0.97–1.05) | 0.76 | 995 | 1.06 (1.00–1.13) | 0.05 |
|
|||||||
<0.85 | 5,301 | 254 | 1.07 (0.90–1.26) | 75 | 1.11 (0.82–1.50) | ||
0.86–0.87 | 5,126 | 328 | 1.00 (Reference) | 97 | 1.00 (Reference) | ||
0.88–0.89 | 5,287 | 493 | 1.12 (0.97–1.29) | 167 | 1.25 (0.97–1.60) | ||
0.90–0.93 | 5,367 | 646 | 1.14 (0.99–1.30) | 233 | 1.33 (1.05–1.69) | ||
≥0.94 | 5,276 | 1,084 | 1.38 (1.21–1.56) | <0.001 | 423 | 1.70 (1.36–2.13) | <0.001 |
per 0.1 unit | 26,357 | 2,805 | 1.28 (1.20–1.36) | <0.001 | 995 | 1.43 (1.29–1.59) | <0.001 |
per SD (0.06) | 26,357 | 2,805 | 1.15 (1.11–1.19) | <0.001 | 995 | 1.23 (1.16–1.30) | <0.001 |
|
|||||||
<0.47 | 5,286 | 239 | 1.10 (0.93–1.30) | 63 | 1.09 (0.79–1.50) | ||
0.48–0.49 | 5,219 | 334 | 1.00 (Reference) | 94 | 1.00 (Reference) | ||
0.50–0.51 | 5,360 | 501 | 1.11 (0.96–1.27) | 173 | 1.34 (1.04–1.72) | ||
0.52–0.54 | 5,264 | 663 | 1.07 (0.94–1.23) | 238 | 1.31 (1.03–1.67) | ||
≥0.55 | 5,228 | 1,068 | 1.24 (1.09–1.40) | 0.005 | 427 | 1.65 (1.32–2.08) | <0.001 |
per 0.1 unit | 26,357 | 2,805 | 1.24 (1.15–1.33) | <0.001 | 995 | 1.50 (1.33–1.68) | <0.001 |
per SD (0.05) | 26,357 | 2,805 | 1.12 (1.08–1.16) | <0.001 | 995 | 1.24 (1.16–1.31) | <0.001 |
Abbreviations: HR = hazard ratio, CI = confidence interval, SD = standard deviation.
Adjusted for age (in the time scale), smoking (never, former, current, unknown) and physical activity per week (no, <3 hours light, ≥3 hours light or <1 hour hard, ≥1 hour hard, unknown).
This category was excluded from the remainder of the analysis presented in the table.
All causes | Cardiovascular disease | ||||||
Anthropometric measures | No. of persons | No. of deaths | Adjusted |
|
No. of deaths | Adjusted |
|
|
|||||||
<18.5 |
314 | 44 | 2.02 (1.49–2.74) | 9 | 1.39 (0.71–2.71) | ||
18.5–24.9 | 13,895 | 819 | 1.00 (Reference) | 230 | 1.00 (Reference) | ||
25.0–29.9 | 10,947 | 872 | 0.81 (0.74–0.90) | 308 | 0.93 (0.78–1.10) | ||
30.0–34.9 | 3,961 | 469 | 0.93 (0.83–1.05) | 181 | 1.10 (0.90–1.35) | ||
≥35.0 | 1,393 | 204 | 1.24 (1.06–1.45) | 0.26 | 74 | 1.41 (1.08–1.85) | 0.02 |
per 5 kg/m2 | 30,196 | 2,364 | 1.03 (0.98–1.07) | 0.27 | 793 | 1.10 (1.03–1.19) | 0.009 |
per SD (4.5 kg/m2) | 30,196 | 2,364 | 1.02 (0.98–1.07) | 0.27 | 793 | 1.09 (1.02–1.17) | 0.009 |
|
|||||||
<70 | 3,981 | 126 | 1.11 (0.92–1.35) | 25 | 0.94 (0.62–1.44) | ||
70–79 | 11,122 | 566 | 1.00 (Reference) | 152 | 1.00 (Reference) | ||
80–89 | 8,589 | 761 | 1.00 (0.90–1.12) | 265 | 1.14 (0.93–1.40) | ||
90–99 | 4,330 | 537 | 1.11 (0.99–1.23) | 207 | 1.36 (1.10–1.68) | ||
≥100 | 2,174 | 374 | 1.48 (1.30–1.70) | <0.001 | 144 | 1.80 (1.43–2.27) | <0.001 |
per 10 cm | 30,196 | 2,364 | 1.11 (1.07–1.16) | <0.001 | 793 | 1.20 (1.12–1.27) | <0.001 |
per SD (11.3 cm) | 30,196 | 2,364 | 1.13 (1.09–1.18) | <0.001 | 793 | 1.22 (1.14–1.31) | <0.001 |
|
|||||||
<95 | 6,457 | 348 | 1.10 (0.95–1.27) | 96 | 1.17 (0.89–1.54) | ||
95–99 | 6,639 | 428 | 1.00 (Reference) | 115 | 1.00 (Reference) | ||
100–104 | 6,840 | 499 | 0.86 (0.75–0.98) | 173 | 1.04 (0.82–1.31) | ||
105–109 | 4,498 | 410 | 0.87 (0.76–1.00) | 151 | 1.07 (0.84–1.36) | ||
≥110 | 5,762 | 679 | 1.02 (0.90–1.15) | 0.33 | 258 | 1.27 (1.01–1.58) | 0.14 |
per 10 cm | 30,196 | 2,364 | 1.03 (0.98–1.07) | 0.20 | 793 | 1.10 (1.02–1.18) | 0.01 |
per SD (9.4 cm) | 30,196 | 2,364 | 1.03 (0.99–1.07) | 0.20 | 793 | 1.09 (1.02–1.17) | 0.01 |
|
|||||||
<0.74 | 6,040 | 191 | 1.01 (0.84–1.22) | 46 | 0.94 (0.65–1.35) | ||
0.74–0.77 | 6,011 | 282 | 1.00 (Reference) | 83 | 1.00 (Reference) | ||
0.78–0.79 | 5,988 | 413 | 1.08 (0.93–1.26) | 134 | 1.11 (0.84–1.46) | ||
0.80–0.83 | 6,125 | 572 | 1.16 (1.00–1.34) | 189 | 1.17 (0.90–1.51) | ||
≥0.84 | 6,032 | 906 | 1.48 (1.29–1.69) | <0.001 | 341 | 1.65 (1.30–2.10) | <0.001 |
per 0.1 unit | 30,196 | 2,364 | 1.34 (1.25–1.43) | <0.001 | 793 | 1.49 (1.33–1.66) | <0.001 |
per SD (0.06) | 30,196 | 2,364 | 1.19 (1.15–1.24) | <0.001 | 793 | 1.27 (1.18–1.36) | <0.001 |
|
|||||||
<0.43 | 6,001 | 156 | 1.29 (1.05–1.59) | 30 | 1.29 (0.83–2.02) | ||
0.43–0.46 | 6,114 | 235 | 1.00 (Reference) | 55 | 1.00 (Reference) | ||
0.47–0.49 | 6,010 | 407 | 1.19 (1.01–1.40) | 121 | 1.35 (0.98–1.86) | ||
0.50–0.54 | 6,014 | 606 | 1.15 (0.99–1.34) | 218 | 1.42 (1.05–1.91) | ||
≥0.55 | 6,057 | 960 | 1.35 (1.16–1.56) | 0.005 | 369 | 1.71 (1.28–2.28) | <0.001 |
per 0.1 unit | 30,196 | 2,364 | 1.20 (1.14–1.27) | <0.001 | 793 | 1.34 (1.21–1.47) | <0.001 |
per SD (0.07) | 30,196 | 2,364 | 1.14 (1.10–1.19) | <0.001 | 793 | 1.23 (1.15–1.32) | <0.001 |
Abbreviations: HR = hazard ratio, SD = standard deviation, CI = confidence interval.
Adjusted for age (in the time scale), smoking (never, former, current, unknown) and physical activity per week (no, <3 hours light, ≥3 hours light or <1 hour hard, ≥1 hour hard, unknown).
This category was excluded from the remainder of the analysis presented in the table.
All cause mortality was for both sexes statistically significantly lower in the BMI range 25.0–29.9 compared to the reference group (BMI 18.5–24.9), given the above adjustments.
Overall the results were similar for both sexes except for WHR appearing as a somewhat stronger predictor among women, as compared to men, while HRs for WHtR seemed more comparable with that of waist circumference than WHR (
The sensitivity analyses did not deviate considerably from the primary results. Among men, the HRs per one SD increase in anthropometric measures never differed more than 0.02 from the main model (
Informativeness | ||
Anthropometric measures | All cause mortality | Cardiovascular disease mortality |
|
||
Body mass index | 1.5 (3%) | 13.3 (31%) |
Waist circumference | 26.3 (48%) | 30.4 (70%) |
Hip circumference | 0.1 (0.2%) | 3.7 (8%) |
Waist-to-hip ratio | 54.7 (100%) | 43.5 (100%) |
Waist-to-height ratio | 34.4 (63%) | 45.0 (104%) |
|
||
Body mass index | 1.2 (2%) | 6.6 (15%) |
Waist circumference | 33.4 (47%) | 30.7 (69%) |
Hip circumference | 1.6 (2%) | 6.3 (14%) |
Waist-to-hip ratio | 71.5 (100%) | 44.4 (100%) |
Waist-to-height ratio | 38.7 (54%) | 33.2 (75%) |
The results from our analysis of reclassification and discrimination improvement are shown in
Anthropometric measures | IDI (‰) |
|
NRI 1 |
|
NRI 2 |
|
NRI 3 |
|
|
||||||||
Body mass index | 0.59 | 0.20 | 1.50 | 0.64 | 1.64 | 0.78 | 5.74 | 0.39 |
Waist circumference | 1.99 | 0.009 | 4.32 | 0.28 | 1.09 | 0.88 | 9.62 | 0.23 |
Hip circumference | 0.10 | 0.58 | 1.24 | 0.48 | −0.79 | 0.81 | 0.73 | 0.84 |
Waist-to-hip ratio | 3.45 | <0.001 | 4.20 | 0.35 | 5.88 | 0.42 | 15.44 | 0.07 |
Waist-to-height ratio | 3.64 | <0.001 | 2.86 | 0.52 | 5.39 | 0.47 | 13.37 | 0.12 |
|
||||||||
Body mass index | 0.40 | 0.42 | −1.94 | 0.61 | −4.16 | 0.39 | −2.41 | 0.69 |
Waist circumference | 1.59 | 0.04 | 7.33 | 0.14 | 0.67 | 0.91 | 12.76 | 0.10 |
Hip circumference | 0.09 | 0.72 | −0.04 | 0.99 | 4.20 | 0.27 | 5.78 | 0.21 |
Waist-to-hip ratio | 2.63 | 0.007 | 3.69 | 0.46 | 4.32 | 0.53 | 13.64 | 0.11 |
Waist-to-height ratio | 2.77 | 0.005 | 7.23 | 0.17 | −6.18 | 0.36 | 6.84 | 0.43 |
Abbreviations: IDI = integrated discrimination improvement, NRI = net reclassification improvement.
Participants with body mass index <18.5 kg/m2 were excluded from the analysis.
NRI when adding a given anthropometric measure to a prediction model using two risk categories (<5%, ≥5%).
Three risk categories (<1%, 1–9%, ≥10%).
Four risk categories (<1%, 1–4%, 5–9%, ≥10%).
Variable included in model: Age.
Variables included in model: Age, smoking status, systolic blood pressure, and total cholesterol.
Anthropometric measures | IDI (‰) |
|
NRI 1 |
|
NRI 2 |
|
NRI 3 |
|
|
||||||||
Body mass index | 0.94 | 0.07 | 0.28 | 0.95 | −8.41 | 0.23 | −6.63 | 0.43 |
Waist circumference | 4.12 | <0.001 | 2.73 | 0.67 | 8.42 | 0.37 | 15.00 | 0.19 |
Hip circumference | 1.12 | 0.03 | −0.98 | 0.81 | −7.35 | 0.28 | −7.38 | 0.35 |
Waist-to-hip ratio | 5.01 | <0.001 | 2.15 | 0.77 | 26.76 | 0.009 | 32.21 | 0.01 |
Waist-to-height ratio | 4.36 | <0.001 | −4.39 | 0.51 | 10.77 | 0.26 | 9.10 | 0.43 |
|
||||||||
Body mass index | 0.84 | 0.15 | −3.27 | 0.42 | 5.90 | 0.35 | 5.36 | 0.48 |
Waist circumference | 3.46 | 0.002 | 7.09 | 0.26 | 30.25 | 0.001 | 43.01 | <0.001 |
Hip circumference | 1.11 | 0.07 | −2.80 | 0.53 | 6.36 | 0.34 | 6.98 | 0.38 |
Waist-to-hip ratio | 3.90 | 0.002 | −3.95 | 0.49 | 33.30 | <0.001 | 36.08 | <0.001 |
Waist-to-height ratio | 3.65 | 0.001 | 4.37 | 0.47 | 25.41 | 0.006 | 35.50 | 0.001 |
Abbreviations: IDI = integrated discrimination improvement, NRI = net reclassification improvement.
Participants with body mass index <18.5 kg/m2 were excluded from the analysis.
NRI when adding a given anthropometric measure to a prediction model using two risk categories (<5%, ≥5%).
Three risk categories (<1%, 1–9%, ≥10%).
Four risk categories (<1%, 1–4%, 5–9%, ≥10%).
Variable included in model: Age.
Variables included in model: Age, smoking status, systolic blood pressure, and total cholesterol.
Risk of death from CVD associated with the measures studied stratified by age is shown in
Adjusted |
||||
Men | Women | |||
Anthropometric measures | 20–59 years | 60–79 years | 20–59 years | 60–79 years |
Body mass index, per 5 kg/m2 | 1.55 (1.27–1.89) | 1.11 (1.00–1.23) | 1.26 (0.97–1.64) | 1.09 (1.01–1.18) |
Waist circumference, per 10 cm | 1.49 (1.28–1.74) | 1.15 (1.07–1.24) | 1.36 (1.12–1.66) | 1.18 (1.10–1.26) |
Hip circumference, per 10 cm | 1.45 (1.14–1.84) | 1.04 (0.93–1.17) | 1.12 (0.86–1.47) | 1.09 (1.01–1.18) |
Waist-to-hip ratio, per 0.1 unit | 1.96 (1.52–2.53) | 1.35 (1.20–1.51) | 2.15 (1.60–2.89) | 1.42 (1.26–1.60) |
Waist-to-height ratio, per 0.1 unit | 2.25 (1.73–2.93) | 1.37 (1.21–1.55) | 1.69 (1.23–2.33) | 1.30 (1.18–1.44) |
|
||||
Body mass index |
1.35 (1.18–1.55) | 1.08 (1.00–1.15) | 1.23 (0.97–1.56) | 1.08 (1.01–1.16) |
Waist circumference |
1.44 (1.26–1.66) | 1.14 (1.06–1.22) | 1.42 (1.14–1.77) | 1.20 (1.12–1.30) |
Hip circumference |
1.25 (1.08–1.45) | 1.03 (0.96–1.10) | 1.11 (0.86–1.44) | 1.09 (1.01–1.17) |
Waist-to-hip ratio |
1.47 (1.27–1.69) | 1.18 (1.11–1.26) | 1.58 (1.33–1.89) | 1.24 (1.15–1.33) |
Waist-to-height ratio |
1.54 (1.34–1.76) | 1.18 (1.10–1.26) | 1.46 (1.16–1.84) | 1.21 (1.13–1.30) |
Abbreviations: HR = hazard ratio, SD = standard deviation, CI = confidence interval.
Participants with body mass index <18.5 kg/m2 were excluded from the analyses.
Adjusted for age (in the time scale), smoking (never, former, current, unknown) and physical activity per week (no, <3 hours light, ≥3 hours light or <1 hour hard, ≥1 hour hard, unknown).
One SD: men 3.5 kg/m2, women 4.5 kg/m2.
One SD: men 9.2 cm, women 11.5 cm.
One SD: men 9.2 cm, women 9.4 cm.
One SD: 0.06 for both sexes.
One SD: men 0.05, women 0.07.
The results for CVD mortality from the mutual adjustment analysis of hip and waist circumferences are shown in
Men | Women | ||||||
Anthropometric measures | No. of persons | No. of deaths | Adjusted |
No. of persons | No. of deaths | Adjusted |
|
|
|||||||
|
|
||||||
<80 | <70 | 1,882 | 33 | 0.98 (0.67–1.42) | 3,981 | 25 | 0.86 (0.56–1.32) |
80–89 | 70–79 | 9,466 | 233 | 1.00 (Reference) | 11,122 | 152 | 1.00 (Reference) |
90–99 | 80–89 | 10,378 | 404 | 1.20 (1.00–1.44) | 8,589 | 265 | 1.26 (1.02–1.56) |
100–109 | 90–99 | 3,625 | 226 | 1.52 (1.20–1.93) | 4,330 | 207 | 1.65 (1.28–2.14) |
≥110 | ≥100 | 1,006 | 99 | 2.64 (1.91–3.67) | 2,174 | 144 | 2.54 (1.81–3.58) |
Waist circumference, per 10 cm | 26,357 | 995 | 1.42 (1.28–1.58) | 30,196 | 793 | 1.44 (1.29–1.61) | |
|
|||||||
<95 | 2,360 | 80 | 1.16 (0.89–1.50) | 6,457 | 96 | 1.40 (1.06–1.84) | |
95–99 | 6,158 | 225 | 1.00 (Reference) | 6,639 | 115 | 1.00 (Reference) | |
100–104 | 9,203 | 335 | 0.75 (0.63–0.90) | 6,840 | 173 | 0.88 (0.69–1.12) | |
105–109 | 5,471 | 200 | 0.61 (0.49–0.76) | 4,498 | 151 | 0.77 (0.60–1.01) | |
≥110 | 3,165 | 155 | 0.65 (0.49–0.86) | 5,762 | 258 | 0.67 (0.50–0.91) | |
Hip circumference, per 10 cm | 26,357 | 995 | 0.73 (0.62–0.86) | 30,196 | 793 | 0.77 (0.68–0.88) |
Abbreviations: HR = hazard ratio, CI = confidence interval.
Participants with body mass index <18.5 kg/m2 were excluded from all analyses.
Adjusted for age (in the time scale), smoking (never, former, current, unknown), physical activity per week (no, <3 hours light, ≥3 hours light or <1 hour hard, ≥1 hour hard, unknown), and either waist circumference or hip circumference.
Men | Women | ||||||
Anthropometric measures | No. of persons | No. of deaths | Adjusted |
No. of persons | No. of deaths | Adjusted |
|
|
|||||||
18.5–24.9 | 9,575 | 300 | 1.00 (Reference) | 13,895 | 230 | 1.00 (Reference) | |
25.0–29.9 | 13,138 | 492 | 0.91 (0.78–1.07) | 10,947 | 308 | 0.81 (0.68–0.97) | |
30.0–34.9 | 3,154 | 175 | 1.08 (0.87–1.34) | 3,961 | 181 | 0.88 (0.71–1.08) | |
≥35.0 | 490 | 28 | 1.23 (0.81–1.86) | 1,393 | 74 | 1.07 (0.81–1.42) | |
per SD (M: 3.4, W: 4.5) | 26,357 | 995 | 1.01 (0.94–1.09) | 30,196 | 793 | 1.00 (0.93–1.08) | |
|
|||||||
|
|
||||||
<0.85 | <0.74 | 5,286 | 75 | 1.14 (0.85–1.53) | 6,040 | 46 | 0.82 (0.55–1.21) |
0.86–0.87 | 0.74–0.77 | 5,219 | 97 | 1.00 (Reference) | 6,011 | 83 | 1.00 (Reference) |
0.88–0.89 | 0.78–0.79 | 5,360 | 167 | 1.25 (0.96–1.62) | 5,988 | 134 | 1.15 (0.87–1.51) |
0.90–0.93 | 0.80–0.83 | 5,264 | 233 | 1.35 (1.07–1.70) | 6,125 | 189 | 1.11 (0.88–1.40) |
≥0.94 | ≥0.84 | 5,228 | 423 | 1.64 (1.30–2.07) | 6,032 | 341 | 1.59 (1.27–2.00) |
per SD (both sexes: 0.06) | 26,357 | 995 | 1.22 (1.14–1.31) | 30,196 | 793 | 1.27 (1.18–1.36) |
Abbreviations: HR = hazard ratio, CI = confidence interval, SD = standard deviation.
Persons with body mass index <18.5 kg/m2 were excluded from all analyses.
Adjusted for age (in the time scale), smoking (never, former, current, unknown), physical activity per week (no, <3 hours light, ≥3 hours light or <1 hour hard, ≥1 hour hard, unknown), and either body mass index or waist-to-hip ratio.
Of the five anthropometric measures studied, WHR and WHtR were most strongly associated with mortality, after adjusting for confounding factors. This was true both regarding overall mortality and death from CVD specifically. In accordance with other studies, our results show that BMI is a poorer predictor of death than the other measures
In all parts of our analysis, BMI showed weaker associations with both all cause mortality and CVD mortality, when compared to waist circumference, WHR and WHtR. Furthermore, BMI was the only among these four measures which failed to show a statistically significant association with all cause mortality. BMI also contributed less additional information to the prediction models studied (
Waist circumference proved to be a statistically significant risk factor in all analyses, but still showed weaker associations with mortality than both WHR and WHtR. In particular, it emerged as a strong risk factor when adjusting for hip circumference. This underlines the significance of considering body configuration rather than the abdominal girth alone.
Hip circumference showed a weak positive association with mortality. However, when adjusting for waist circumference, it proved to be inversely related to CVD mortality in both genders. This finding is in accordance with previous research
Both in the presence (
Based on the IDIs (
The main strength of our investigation lies in the prospective and comprehensive nature of the HUNT 2 study, its good participation rates, and it being fairly representative for the entire Norwegian nation. The fact that the HUNT population is ethnically homogenous may also be considered a strength in this context, since ethnic differences (genetic and epigenetic factors) may influence the predictive properties of anthropometric measures
The HUNT 2 database lacks comprehensive information on the participants' dietary habits and cancer history. However, the exclusion of participants with BMI <18.5 kg/m2 and the sensitivity analysis which excludes the first three years of follow-up minimise the potential for confounding by cancer. Our sensitivity analysis indicates that the impact of other potential confounders is minimal.
Our study adds further knowledge to the evidence that BMI is not the most appropriate measure of obesity in everyday clinical practice. WHR is as easy to calculate as BMI and is presently better documented than WHtR. It therefore appears reasonable to recommend WHR as the primary measure of body composition and obesity, at least when it comes to assessing risk of CVD. There is, however, need for further clarification before determining whether WHtR should be considered an even better alternative than WHR. Single (waist circumference in isolation) or additional measures (involving weight and/or height) may also be added to nuance estimations of CVD risk when indicated, for instance in relation to clearly obese or under-weight individuals with a favourable WHR. A certain weakness of the approach suggested here is the documented, inter-personal variance in measurement of waist and hip circumferences
It is hard to determine how much effort should be put into training healthcare workers to measure WHR or WHtR in a standardised and reproducible manner, as the potential for predictive improvement will depend on the selected cut-off points and also the choice of prediction model. In relation to combined risk algorithms
Baseline characteristics of the study population.
(DOCX)
Risk of death from all causes and from cardiovascular disease among men aged 20–79; associations with anthropometric measures (hazard ratios per increase in anthropometric measures of one standard deviation). Sensitivity analysis involving different models.
(DOCX)
Risk of death from all causes and from cardiovascular disease among women aged 20–79; associations with anthropometric measures (hazard ratios per increase in anthropometric measures of one standard deviation). Sensitivity analysis involving different models.
(DOCX)
We thank the HUNT Research Centre for contributing HUNT 2 data.