Skinfold thickness measurements and mortality in white males during 27.7 years of

follow-up

Wann Jia Loh1,2, Desmond G Johnston1, Nick Oliver1, Ian F. Godsland1

1Diabetes Endocrinology and Metabolic Medicine, Faculty of Medicine, Imperial College London,

St. Mary's Campus, London, UK

2Department of Endocrinology, Changi General Hospital, Singapore

Short title: iliac skinfold, obesity and mortality

Conflicts of interest: there are no conflicts

Corresponding author:

Wann Jia Loh,

Division of Diabetes, Endocrinology and Metabolism, Department of Medicine,

Imperial College London, St Mary’s Campus,

Room G1, Norfolk Place, London W2 1NH, UK.

email: wann_jia_loh@cgh.com.sg, l.wann-jia@imperial.ac.uk,

1

Abstract

Introduction: Obesity is a major risk factor for mortality from a range of causes. We investigated

whether skinfold measurements were associated with mortality independently of variation in

body mass index (BMI).

Methods: A prospective analysis of mortality in 870 apparently healthy adult Caucasian men

participating in an occupational health cohort was undertaken. At baseline, skinfold

measurements were taken at biceps, triceps, iliac and subscapular sites. Derived

measurements included the sum of all 4 skinfolds and subscapular to triceps, subscapular to

iliac and BMI to iliac ratios. All-cause mortality was analysed by Cox proportional hazards

modelling and death in specific mortality subcategories by competing risks analysis.

Results: During a mean of 27.7 years follow up, there were 303 deaths (119 cancer, 101

arteriovascular, 40 infection, 43 other). In univariable analysis, BMI was associated with all-

cause, cancer, arteriovascular and other mortality and subscapular skinfold with all-cause and

arteriovascular mortality. On bivariable analysis, with inclusion of BMI, subscapular skinfold

ceased to be a associated with mortality but iliac skinfold emerged as strongly, negatively

associated with all-cause and arteriovascular mortality. In multivariable analysis, with inclusion

of age, BMI, smoking, alcohol and exercise, iliac skinfold was negatively associated with all-

cause (Hazard ratio HR 0.77, 95% confidence interval CI 0.66-0.90, p=0.002), arteriovascular

(HR 0.75, 95%CI 0.58,0.97, p=0.02) and infection (HR 0.63, 95%CI 0.42,0.94, p=0.02) death.

Among obese participants (BMI ≥30kg/m2), iliac skinfold of ≤9.7mm was associated with a six-

fold increase in all-cause mortality risk.

Conclusion: Low iliac skinfold thickness is an independent risk factor for all-cause mortality in

adult white males with risk apparently concentrated among people who are obese.

2

Introduction

Concern over the rapidly increasing prevalence of obesity throughout the world (1) has

led to many studies correlating indices of obesity to long-term health. In particular, body mass

index(BMI) has been found to be strongly associated with all-cause mortality (2, 3) and the

development of chronic diseases including cardiovascular disease (4), diabetes (5), cancer

(6),gallbladder diseases (7) and gout (8). BMI is a commonly used measure of overall adiposity

in clinical and research settings because of ease of measurement. However, there may be

appreciable heterogeneity in risk at any given level of BMI. This is exemplified by the concept of

‘metabolically healthy obesity’, according to which overweight or obese people with a relatively

normal metabolic risk factor profile are at no greater cardiovascular risk than their normal weight

equivalents (9, 10). However, the metabolically healthy obese may show greater risk of

progression to metabolic abnormality (11, 12) and, although risk of CHD in metabolically healthy

overweight or obese people may be less than in their metabolically unhealthy equivalents, with

long follow-up (13) or in a sufficiently large sample (14), an increase in cardiovascular risk may,

nevertheless, be detected. Such heterogeneity in risk, independent of overall adiposity, may

result from variation in the distribution of body fat. Obese people with a large depot of visceral

fat appear to be at particularly high risk (15-19) and ectopic deposition of fat in liver, muscle or

other tissues may also be important (20). However, clinical evaluation of specific fat depots is

limited by the need for computed tomography (CT) or magnetic resonance imaging (MRI) (21,

22). Beyond measurement of BMI, clinical evaluation of adiposity-related risks may be

enhanced by using alternative anthropometric measurements. Skinfold thickness

measurements offer one such alternative but, hitherto, studies have not provided strong

evidence to support their use in clinical practice. An increase in subscapular skinfold may not be

associated with all-cause mortality in men (23) but has been found to be linked to ischaemic

heart disease or variation in its risk factors in some studies (24, 25), although not in all(26). In

the Paris Prospective Study, preferential localisation of fat in the abdominal component, as

measured by sagittal diameter adjusted for the sum of trunk skinfolds was associated with

3

increased cancer mortality (27). Subscapular to triceps ratio has been found to be associated

with stroke (28), ischaemic heart disease (29), and blood pressure (30) and has been proposed

as a marker of central to peripheral fat distribution.

Appreciable covariation may be expected between skinfold thicknesses and overall

adiposity and previous analyses may not have fully discriminated between risks associated with

different measures of adiposity. The primary aim of the present analysis, therefore, was to

investigate whether skinfold measurements were associated with mortality independently of BMI

as an index of overall adiposity.

Subjects and Methods

Participants:

The Heart Disease and Diabetes Risk Indicators in a Screened Cohort (HDDRISC) study

is a prospective study of 1192 white males recruited as part of a company health screening

program undertaken in London, U.K.. At enrolment, the participants were apparently healthy

senior executives, working or recently retired. The study started in 1971 with participants invited

for follow-up every 2-3 years; data collection ended in 2000. Skinfold thickness measurements

were taken from the beginning of the study until 1996 and for the purposes of this analysis, only

the participants with skinfold thickness measures (n=888) were considered for inclusion. The

present analysis is restricted to participants’ earliest records with skinfold information. The study

was approved by the ethics committee. All participants gave their written informed consent.

Measurements:

Details of data recording have been described previously (31). In brief, measurements

were carried out at a dedicated metabolic day ward, in the morning following an overnight fast.

Each participant provided a full medical history including details of smoking, alcohol

consumption, exercise and medication and also underwent a range of clinical and laboratory

measurements. Skinfold thicknesses were measured using the procedure of Durnin and

Womersley (32). The biceps skinfold was taken over the mid-point of biceps muscle with the

arm resting supinated on the thigh. The triceps skinfold was taken over the mid-point of the

4

triceps muscle, at the posterior line between the olecranon and tip of acromion. The

subscapular skinfold was taken at an angle of about 45 degrees to the vertical, at the inferior

angle of scapular and the iliac skinfold was taken just above the iliac crest in the mid-axillary line

(32). Skinfold thicknesses were measured 3 times at each of the 4 sites using Harpenden

callipers. The mean of the 3 readings for each site was calculated.

Mortality ascertainment:

Mortality information was obtained through the United Kingdom NHS Information Centre

for Health and Social Care. Mortality information for participants who died overseas was notified

by family members of the deceased or their employer. The mortality information was complete

except for 18 participants who were no longer traceable. The present analysis includes deaths

up to 1 January 2014 and follow-up time was calculated as the time between the earliest visit

with skinfold information and death, or up till 1 January 2014 for those who were still alive. The

primary end- point was all cause mortality. The mortality causes were also subdivided into 4

categories; cancer deaths, arteriovascular deaths, infection deaths (predominantly pneumonia)

and other deaths. Arteriovascular mortality comprised death from coronary heart disease, heart

failure, cerebrovascular disease and aneurysm. Causes in the ‘other death’ category included

motor neurone disease, respiratory disease, gastrointestinal ulceration, alcohol and accident.

Data and statistical analysis:

The following skinfold ratios were calculated: subscapular to triceps, subscapular to iliac,

and BMI to iliac. In addition, the sum of all 4 skinfold thicknesses was calculated. Smoking was

coded as 0, <15 (light) or ≥ 15 (heavy) cigarettes per day. Alcohol consumption was coded as 0,

<28units per week (light) or ≥28 units per week (heavy). Exercise habits were coded as none,

moderate and regular (≥3 periods of ≥20min of exercise to breathlessness per week). All data

was analysed using STATA 13 for Windows (Stata, College Station, TX, USA). Baseline

characteristics were expressed in median and interquartile range (IQR) for continuous variables

and percentages and number of observations (n) for categorical variables. Significant

differences between the survivors and non-survivors for continuous variables were assessed

5

using Mann-Whitney test and for categorical variables by chi-square test. Triceps, subscapular,

iliac, subscapular tricep ratio, subscapular iliac ratio, BMI iliac ratio and sum of skinfold

thicknesses were logarithmically transformed to reduce heteroscedasticity in regression

analyses. Continuous variables were standardised as z-scores for entry in univariable and

multivariable analyses. The Cox proportional hazards model was used to analyse survival time

in relation to all-cause mortality (STATA command: ‘stcox’) whereas competing-risks survival

regression (STATA command: ‘stcrreg’) was used to analyse survival in the specific mortality

subgroups: cancer, arteriovascular, infection and other. The Cox analysis proportional hazards

assumption was checked using Schoenfeld residuals and linearity using Matingale residuals.

For all models, linearity was further checked on the basis of hazard ratios in quartile and quintile

percentiles of each skinfold. Effects significant at p <0.05 were considered for interpretation.

Results

Out of 1192 participants in the HDDRISC study, 870 who had skinfold thicknesses

measured and mortality follow-up data recorded were included for analysis. Of these, 303

participants died before 1st January 2014. There were 119 participants who died from cancer,

101 from arteriovascular disease, and 40 from infection. The remaining 43 participants died

from other causes. The mean mortality follow-up time from the first record with skinfold

information was 27.7 years with a range of 0.5-41.5years.

At baseline, compared with survivors, those who died were older, had higher BMI,

smoked more cigarettes and drank more alcohol (Table 1). These differences generally

extended to the cancer, arteriovascular and infection mortality subgroups. Triceps and biceps

skinfold thicknesses did not differ between survivors and those who died (Table 1). Subscapular

skinfold was higher among those who died of any cause (p=0.01), specifically, those who died

of arteriovascular disease (p=0.02). Iliac skinfold was 16% lower (p<0.001) among those who

died of any cause (median 12.5 (IQR 8.6, 8.7) vs 14.9 (IQR10.1, 21.5) mm in survivors). Iliac

skinfold was also significantly lower among those who died of cancer (16% lower, p=0.02),

arteriovascular disease (13% lower, p=0.008) and infection (31% lower, p=0.001).

6

Variables in Table 1 that differed between survivors and those who died also generally

showed associations with mortality in univariable survival analysis (results not shown). One

exception was iliac skinfold, which were not significantly associated with all-cause, cancer, AVD

and infection death in univariable survival analysis. However, in multivariable survival analysis,

with age, BMI, smoking, alcohol and exercise as covariates, lower iliac skinfold was

significantly, independently associated with all-cause, AVD and infective death (Table 2). No

significant deviations from the Cox model proportionality and linearity assumptions were

detected. Subscapular skinfold ceased to show any associations with mortality in multivariable

analysis and, triceps and biceps skinfolds showed no associations with mortality. Bivariable

survival analyses were undertaken to identify which variable was responsible for iliac skinfold

emerging as a significant associate of mortality in multivariable analysis. The key variable was

BMI: hazard ratios (95%CI, p) for iliac skinfold and BMI in bivariable analysis were: for all-cause

mortality 0.76 (0.66,0.88, p<0.001) and 1.51 (1.35,1.69, p<0.001) respectively; for AVD mortality

0.75 (0.60,0.93, p=0.009) and 1.44 (1.22,1.69, p<0.001) respectively; and for infection mortality

0.64 (0.42,0.96, p=0.03) and 1.27 (0.84,1.92, p=0.2) respectively. Bivariable analysis also

showed that group differences and univariable associations between subscapular skinfold and

all-cause and AVD mortality were dependent on BMI (results not shown). Linearity checks

based on hazard in skinfold percentiles found no evidence for J- or U-shaped relationships

between skinfolds and mortality. Nevertheless, an association was found between low triceps

skinfold and all-cause mortality (lowest vs highest: quartile HR 1.51 (95%CI 1.07, 2.13) p=0.02);

quintile 1.73 (1.18,2.54), p=0.005). However, this association did not extend to specific causes

of mortality and was not enhanced in overweight or obesity.

The emergence of iliac skinfold as a strong negative associate of mortality when the

positive associate of mortality, BMI, was included in survival analysis as an additional

independent variable (plus the strengthening of the positive association between BMI and

mortality) could result from a positive correlation between iliac skinfold and BMI. On Pearson

correlation, BMI was found to explain 21% of the variation in iliac skinfold (R=0.46, p<0.001).

7

There were, therefore, two components of variation in iliac skinfold, one positively related to

mortality and dependent on BMI and one negatively related to mortality and independent of

BMI. This unexpected finding suggested that risk of mortality associated with a low iliac skinfold

might show differential associations with mortality at different BMI. We, therefore, derived the

HR for all-cause mortality in iliac skinfold quartiles in ranges of BMI corresponding to normal

weight (<25 kg/m2), overweight (25 to <30 kg/m2) and obese (≥30 kg/m2). Among normal weight

participants, the lowest quartile of iliac skinfold was not associated with all-cause mortality

(Figure 1), but was independently associated in the overweight (HR 2.07, p=0.01) and risk was

markedly increased in the obese (HR 6.34, p=0.02). Numbers of deaths were relatively low

among obese participants (n=40). Therefore, variation in risk associated with low iliac skinfold

was explored in successive ranges of overweight, both to include more participants and to

identify at which level of BMI hazard substantially increased. Hazard ratios, HR (95%CI),

associated with the lowest iliac skinfold quartile relative to the highest were for those with

BMI>25 kg/m2: HR 2.09 (1.29,3.38 p=0.003); BMI>26 kg/m2: HR 2.15 (1.27,3.66 p=0.005);

BMI>27 kg/m2: HR 3.15 (1.64,6.04 p=0.001); BMI>28 kg/m2: HR 3.33 (1.30,8.56 p=0.01); and

BMI>29 kg/m2: HR 6.23 (1.66,23.4 p=0.007). Numbers of deaths in each of these BMI ranges

were 179, 142,99, 66 and 50, respectively.

Discussion

Skinfold thickness measures, including iliac skinfold, are positively related to BMI but the

possibility that this relationship might obscure an underlying negative association between a

skinfold measure and mortality appears to have been little appreciated. Our analysis of

associations of skinfold thickness measures with mortality independently of variation in BMI has

revealed that, in the cohort of men we studied, a low iliac skinfold is independently associated

with l mortality over the course of long-term follow-up, in particular, arteriovascular and infection

mortality. Furthermore, risks associated with a low iliac skinfold appear to be particularly

concentrated in the obese.

8

Previous studies have emphasised a positive association between subscapular skinfold

thickness and total or arteriovascular mortality risk (23, 25, 33). Our study found some evidence

for such an association, although not independently of BMI. Neither the Northwick Park Heart

study(33) nor the Seven Countries Study of Cardiovascular Disease (23) took into account

variation in BMI, although this was taken into account in the Caerphilly study. In contrast to our

findings, the Caerphilly Study reported a relationship between subscapular skinfold and incident

ischemic heart disease that was independent of BMI(25). Of these previous studies, only the

Northwick Park Heart study recorded iliac skinfold(33), but BMI was not taken into account in

exploring its associations with mortality and, in accord with our findings in univariate analysis, no

association was apparent. Clearly, further prospective studies with measurement of iliac skinfold

are needed to confirm or refute our findings and resolve existing inconsistencies.

In accord with other studies (2-4, 6), our analysis found that BMI was positively

associated with all-cause, cancer and arteriovascular mortality. However, BMI is a marker of

overall adiposity (15) and skinfold measurements are generally taken to represent adiposity at

different regional adipose tissue depots (34). Truncal measurements (e.g. subscapular,

abdominal and iliac) then represent central adiposity and peripheral measurements (e.g. biceps,

triceps and thigh) peripheral adiposity. The importance given to this distinction stems from

recognition of the role of the visceral fat depot in generating an adverse metabolic risk factor

profile and augmenting long-term risk of chronic disease (21). Given the way they are

measured, variation in skinfold thicknesses necessarily reflect variation in subcutaneous fat and

may relate only indirectly or not at all (35) to the visceral fat depot (34). Nevertheless, skinfolds

as indices of subcutaneous fat storage could be important given that a limited ability to store fat

in subcutaneous depots may lead to redistribution of excess fatty acids to metabolically

disruptive ectopic locations such as viscera, liver or muscle (18, 20). Clinical evidence for an

adverse effect of diminished subcutaneous fat comes from the observation that among people

with type 2 diabetes, those who have high visceral fat with low subcutaneous fat, show

increased carotid intima media thickness(36). This observation is consistent with our finding that

9

obesity combined with low iliac skinfold constitutes a high-risk state and accords with mortality

risk being concentrated in deaths from arteriovascular disease. However, whilst depletion of a

particular subcutaneous fat depot might imply increased ectopic fat deposition and, therefore,

increased arteriovascular risk, there appear to be no precedents for an association between a

low subcutaneous fat depot and infection death. Pneumonia was the most common cause of

infection death which might indicate a contribution from premature ageing or wasting to the

association we observed. However, these considerations do not explain why iliac skinfold alone

was associated with increased mortality. In this respect, it is noteworthy that measurement error

for the iliac skinfold may be less compared to other skinfolds, especially with increasing obesity

levels (37). Accordingly, iliac skinfold might provide a particularly precise index of subcutaneous

fat.

Regardless of underlying mechanisms, it does appear that a low iliac skinfold can

indicate particularly unhealthy obesity but whether the pattern of low iliac skinfold and high BMI

is metabolically unhealthy remains to be established. In our previous analysis of relationships

between skinfold thicknesses and arteriovascular disease risk factors in the HDDRISC cohort,

variations in age, smoking, alcohol and exercise were taken into account but variation in BMI

was not (31). Accordingly, the associations observed between skinfolds and risk factors were

simply those expected with overall adiposity.

Relatively little use has been made of skinfolds in clinical practice, possibly because of

an apparent lack of any strong predictive advantage over basic anthropometric measurements

such as BMI or waist circumference. Use of ratios might be expected to strengthen independent

associations by skinfold measures but our analysis provided little support for this. The

subscapular/triceps ratio has been proposed as a CVD-associated marker of central fat

deposition(29) but our analysis found no evidence for this ratio having an independent

association with cardiovascular mortality. The BMI/iliac ratio might have been expected to

provide a sensitive index of mortality risk. However, the ratio proved uninformative, with only a

weak negative relationship emerging in multivariable analysis. The positive correlation between

10

BMI and iliac skinfold and their contrasting independent associations with mortality made

interpretation of these findings problematic. Therefore, we undertook a more explicit analysis by

exploring risks in successive ranges of BMI. This uncovered a marked increase in mortality risk

associated an iliac skinfold of 9.7mm or less among participants with a BMI of 29 kg/m2 or more,

which offers the possibility for a useful clinical index. Exploring this in other cohorts could be

worthwhile.

Our study has limitations and strengths. Only white males were studied so the results

are not generalisable to other ethnic groups or women. There were relatively small numbers of

obese participants (n=67). Also, there were no direct measurements of adiposity (e.g. MRI or

CT) that would have helped us establish how iliac skinfold related to ectopic fat deposition in

this cohort. A strength of our study is that it concerned apparently healthy individuals, followed

for a long period and with relatively little prescription drug use. Moreover, the restricted selection

of participants may have helped to distinguish significant associations. Besides that, the skinfold

measurements were carried out by a historically continuous research group working in

dedicated clinical research facilities. The long duration of follow-up may have enabled

discrimination of effects that shorter duration prospective studies have missed, although with

greater numbers of events, it might have been possible to detect non-linearities in the

relationships between skinfolds and mortality. Linearity checks revealed a discontinuous

relationship whereby a low triceps skinfold was associated with all-cause mortality but this

association showed little of the strength and consistency we found for low iliac skinfold and may,

therefore have been a chance finding. Our analysis did employ extensive statistical testing,

which suggests that some false-positive significances may have been encountered. In principle,

five primary hypotheses were explored, namely that biceps, triceps, subscapular and iliac

skinfold thicknesses and the subscapular/triceps skinfold ratio were each independently

associated with all-cause mortality. A probability of <0.01 (0.05÷5) is therefore appropriate for

rejecting the null hypothesis. The independent association of iliac skinfold with all-cause

mortality was significant at p=0.002 and is, therefore, unlikely to have been a chance finding.

11

We consider further testing in specific mortality groupings and in different ranges of BMI to have

been exploratory and hypothesis-generating.

In conclusion, we found that in adult white males, low iliac skinfold thickness is

associated with all-cause, arteriovascular and infection mortality independently of BMI. Risk

appears to be particularly concentrated in the very overweight and the obese, with a 6-fold

increase in all-cause mortality risk above a BMI of 30 kg/m2. This observation needs to be

confirmed in other cohorts and metabolic and ectopic fat correlates of low iliac skinfold need to

be explored further.

12

Acknowledgements

The HDDRISC study was initiated the late Professor Victor Wynn, who directed it for much of its

course. Data acquisition was sustained by many clinical, scientific, technical, nursing and

administrative staff, to each of whom we extend our thanks.

Funding

Data acquisition for the HDDRISC study was funded by the Heart Disease and Diabetes

Research Trust and the Rosen Foundation.

13

References

1. Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono C, et al. Global,

regional, and national prevalence of overweight and obesity in children and adults during 1980-

2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet.

2014;384:766-81.

2. Aune D, Sen A, Prasad M, Norat T, Janszky I, Tonstad S, et al. BMI and all cause

mortality: systematic review and non-linear dose-response meta-analysis of 230 cohort studies

with 3.74 million deaths among 30.3 million participants. BMJ. 2016;353:i2156.

3. Prospective Studies Collaboration, Whitlock G, Lewington S, Sherliker P, Clarke R,

Emberson J, et al. Body-mass index and cause-specific mortality in 900 000 adults:

collaborative analyses of 57 prospective studies. Lancet. 2009;373:1083-96.

4. He Y, Lam TH, Jiang B, Li LS, Sun DL, Wu L, et al. Changes in BMI before and during

economic development and subsequent risk of cardiovascular disease and total mortality: a 35-

year follow-up study in China. Diabetes Care. 2014;37:2540-7.

5. Abdullah A, Peeters A, de Courten M, Stoelwinder J. The magnitude of association

between overweight and obesity and the risk of diabetes: a meta-analysis of prospective cohort

studies. Diabetes Res Clin Prac. 2010;89:309-19.

6. Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M. Body-mass index and

incidence of cancer: a systematic review and meta-analysis of prospective observational

studies. Lancet. 2008;371:569-78.

7. Aune D, Norat T, Vatten LJ. Body mass index, abdominal fatness and the risk of

gallbladder disease. Eur J Epidemiol. 2015;30:1009-19.

8. Aune D, Norat T, Vatten LJ. Body mass index and the risk of gout: a systematic review

and dose-response meta-analysis of prospective studies. Eur J Nutr. 2014;53:1591-601.

9. Kip KE, Marroquin OC, Kelley DE, Johnson BD, Kelsey SF, Shaw LJ, et al. Clinical

importance of obesity versus the metabolic syndrome in cardiovascular risk in women: a report

from the Women's Ischemia Syndrome Evaluation (WISE) study. Circulation. 2004;109:706-13.

14

10. Kaur A, Johnston DG, Godsland IF. Does metabolic health in overweight and obesity

persist? - Individual variation and cardiovascular mortality over two decades. Eur J Endocrinol.

2016;175:133-43.

11. Bradshaw PT, Reynolds KR, Wagenknecht LE, Ndumele CE, Stevens J. Incidence of

components of metabolic syndrome in the metabolically healthy obese over 9 years follow-up:

The atherosclerosis risk in communities study.Int J Obes (Lond). 2017; Epub Oct9.

12. Espinosa De Ycaza AE, Donegan D, Jensen MD. Long-term metabolic risk for the

metabolically healthy overweight/obese phenotype. Int J Obes (Lond). 2017; Epub Sep25.

13. Kramer CK, Zinman B, Retnakaran R. Are metabolically healthy overweight and obesity

benign conditions?: A systematic review and meta-analysis. Ann Intern Med. 2013;159:758-69.

14. Lassale C, Tzoulaki I, Moons KGM, Sweeting M, Boer J, Johnson L, et al. Separate and

combined associations of obesity and metabolic health with coronary heart disease: a pan-

European case-cohort analysis. Eur Heart J. 2017; Epub Aug14

15. Cornier MA, Despres JP, Davis N, Grossniklaus DA, Klein S, Lamarche B, et al.

Assessing adiposity: a scientific statement from the American Heart Association. Circulation.

2011;124:1996-2019.

16. Karelis AD, Faraj M, Bastard JP, St-Pierre DH, Brochu M, Prud'homme D, et al. The

metabolically healthy but obese individual presents a favorable inflammation profile. J Clin

Endocrinol Metab. 2005;90:4145-50.

17. Wildman RP, Muntner P, Reynolds K, McGinn AP, Rajpathak S, Wylie-Rosett J, et al.

The obese without cardiometabolic risk factor clustering and the normal weight with

cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US

population (NHANES 1999-2004). Arch Intern Med. 2008;168:1617-24.

18. Neeland IJ, Ayers CR, Rohatgi AK, Turer AT, Berry JD, Das SR, et al. Associations of

visceral and abdominal subcutaneous adipose tissue with markers of cardiac and metabolic risk

in obese adults. Obesity. 2013;21:E439-47.

15

19. Hamer M, Stamatakis E. Metabolically healthy obesity and risk of all-cause and

cardiovascular disease mortality. J Clin Endocrinol Metab. 2012;97:2482-8.

20. Shimabukuro M, Kozuka C, Taira S, Yabiku K, Dagvasumberel M, Ishida M, et al.

Ectopic fat deposition and global cardiometabolic risk: new paradigm in cardiovascular

medicine. J Med Invest. 2013;60:1-14.

21. Fujimoto WY, Bergstrom RW, Boyko EJ, Chen KW, Leonetti DL, Newell-Morris L, et al.

Visceral adiposity and incident coronary heart disease in Japanese-American men. The 10-year

follow-up results of the Seattle Japanese-American Community Diabetes Study. Diabetes Care.

1999;22:1808-12.

22. Graffy PM, Pickhardt PJ. Quantification of hepatic and visceral fat by CT and MR

imaging: relevance to the obesity epidemic, metabolic syndrome and NAFLD. Br J Radiol.

2016;89:20151024.

23. Menotti A, Lanti M, Maiani G, Kromhout D. Forty-year mortality from cardiovascular

diseases and their risk factors in men of the Italian rural areas of the Seven Countries Study.

Acta Cardiologica. 2005;60:521-31.

24. Donahue RP, Abbott RD. Central obesity and coronary heart disease in men. Lancet.

1987;2:1215.

25. Yarnell JW, Patterson CC, Thomas HF, Sweetnam PM. Central obesity: predictive value

of skinfold measurements for subsequent ischaemic heart disease at 14 years follow-up in the

Caerphilly Study. Int J Obes Relat Metab Disord. 2001;25:1546-9.

26. Imeson JD, Haines AP, Meade TW. Skinfold thickness, body mass index and ischaemic

heart disease. J Epidemiol Community Health. 1989;43:223-7.

27. Oppert JM, Charles MA, Thibult N, Guy-Grand B, Eschwege E, Ducimetiere P.

Anthropometric estimates of muscle and fat mass in relation to cardiac and cancer mortality in

men: the Paris Prospective Study. Am J Clin Nutr. 2002;75:1107-13.

28. Tanne D, Medalie JH, Goldbourt U. Body fat distribution and long-term risk of stroke

mortality. Stroke. 2005;36:1021-5.

16

29. Freedman DS, Williamson DF, Croft JB, Ballew C, Byers T. Relation of body fat

distribution to ischemic heart disease. The National Health and Nutrition Examination Survey I

(NHANES I) Epidemiologic Follow-up Study. Am J Epidemiol. 1995;142:53-63.

30. van Lenthe FJ, van Mechelen W, Kemper HC, Twisk JW. Association of a central pattern

of body fat with blood pressure and lipoproteins from adolescence into adulthood. The

Amsterdam Growth and Health Study. Am J Epidemiol. 1998;147:686-93.

31. Hariri AA, Oliver NS, Johnston DG, Stevenson JC, Godsland IF. Adiposity

measurements by BMI, skinfolds and dual energy X-ray absorptiometry in relation to risk

markers for cardiovascular disease and diabetes in adult males. Disease Markers. 2013;35:753-

64.

32. Durnin JV, Womersley J. Body fat assessed from total body density and its estimation

from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. The

Br J Nutr. 1974;32:77-97.

33. Kim J, Meade T, Haines A. Skinfold thickness, body mass index, and fatal coronary

heart disease: 30 year follow up of the Northwick Park heart study. J Epidemiol Community

Health. 2006;60:275-9.

34. Willett WaH, F. Chapter 9, Anthropometric Measures and Body Composition, Nutritional

epidemiology. 3rd ed. Oxford: Oxford University Press; 2013. ix, 529 p. p.

35. Ludescher B, Machann J, Eschweiler GW, Vanhofen S, Maenz C, Thamer C, et al.

Correlation of fat distribution in whole body MRI with generally used anthropometric data. Invest

Radiol. 2009;44:712-9.

36. Bouchi R, Takeuchi T, Akihisa M, Ohara N, Nakano Y, Nishitani R, et al. High visceral fat

with low subcutaneous fat accumulation as a determinant of atherosclerosis in patients with

type 2 diabetes. Cardiovasc Diabetol. 2015;14:136.

37. Demura S, Sato S. Suprailiac or abdominal skinfold thickness measured with a skinfold

caliper as a predictor of body density in Japanese adults. Tohoku J Exp Med. 2007;213:51-61.

17

Table 1: Baseline Characteristics of Survivors compared to Non-Survivors. Median (interquartile range) are shown for continuous variables and % (n) for categorical variables. The statistical significances of differences between Non-Survivors and Survivors are shown in superscript.

Survivors Non-survivors - cause of deathAll-cause Cancer Arteriovascular Infective Other causes

n 567 303 119 101 40 43

Follow-up time(years) 31.5(24.6,36.1) 23.3(16.2,30.6)<0.001 20.4(14.7,29.3)<0.001 23.4(16.7,28.9)<0.001 29.1(20.7,33.5)<0.01 25.6(13.4,32.7)<0.001

Age (years) 43.1(27.1,48.3) 51.8(47.8,55.7)<0.001 51.7(47.8,55.6)<0.001 52.6(49.1,56.0)<0.001 51.2(47.0,55.5)<0.001 51.6(47.2,55.4)<0.001

BMI (kg/m2) 25.1(23.5,26.9) 25.9(24.2,27.8)<0.001 25.7(24.4,27.5) 0.004 26.1(24.2,27.8) 0.004 25.9(23.9,26.9) 0.3 26.6(24,28) 0.02

Skinfold measurements

triceps (mm) 10.8(8.6,13.4) 11.2(8.6,14.2) 0.2 11.0 (8.3,13.8) 0.7 11.4(9.2,15.0) 0.06 11.9(8.4,13.7) 0.8 11.0(8.4,13.7) 0.8

biceps (mm) 6.0(4.6,7.4) 6.0(4.9,7.5) 0.7 6.4(5.1,7.5) 0.1 6.0(4.8,7.5) 0.7 5.5(4.1,7.2) 0.4 5.5(5.0,7.2) 0.7

subscapular(mm) 15.4(12.1,20.0) 16.3(13.3,20.5) 0.01 16.1(13.2,20.3) 0.1 16.7(13.9,21.9) 0.02 15.0 (13.2,18.8) 0.6 17.1(13.8,20.3) 0.1

iliac (mm) 14.9(10.1,21.5) 12.5(8.6,18.7) <0.001 12.5(8.6,19.6) 0.02 12.9(9.0,16.4) 0.008 10.3(8.9,15.0) 0.001 13.5(8.2,19.3) 0.1

subscapular/triceps ratio 1.46(1.18,1.81) 1.51(1.2,1.89) 0.07 1.49(1.25,1.86) 0.2 1.54(1.19,1.91) 0.2 1.48(1.12,1.89) 0.6 1.57(1.20,2.03) 0.1

Smoking

non-smoker 73(411) 49(148) 50(59) 45(45) 50(20) 56(23)

<15 cig/day 19(108) 29(86) 21(25) 36(36) 33(13) 29(12)

≥15 cig/day 8(48) 22(67) <0.001 29(35) <0.001 19(19) <0.001 17(7) 0.009 15(6) 0.07

Alcohol

non-drinker 1(8) 1(4) 1(1) 3(3) 0 0

<28 units/wk 64(360) 46(140) 44(52) 47(47) 45(18) 53(23)

≥28 units/wk 35(198) 53(159) <0.001 55(66) <0.001 50(51) 0.004 55(22) 0.03 47(20) 0.2

Exercise

sedentary 38(215) 44(131) 41(49) 51(50) 34(14) 44(18)

moderate 44(248) 42(126) 46(54) 36(36) 51(21) 39(16)

regular 18(100) 14(41) 0.1 13(15) 0.4 13(13) 0.06 15(6) 0.6 17(7) 0. 7

18

Table 2. Multivariate hazard ratios of associations of skinfold measurements and all-cause and cause-specific mortality.

Cause of deathAll-cause Cancer Arteriovascular Infective Other causes

Skinfold measurements

triceps 0.89(0.78,1.02) 0.1 0.90 (0.74,1.10) 0.3 1.14 (0.87,1.50) 0.3 0.94 (0.59,1.50) 0.8 0.83 (0.56,1.21) 0.3

biceps 1.94 (0.82,1.06) 0.5 1.10 (0.91,1.33) 0.3 0.78 (0.59,1.02) 0.08 0.87 (0.58,1.30) 0.5 0.89 (0.62,1.2) 0.5

subscapular 0.93 (0.79,1.09) 0.4 1.00 (0.76,1.33) 0.9 0.91 (0.68,1.21) 0.5 1.03 (0.70,1.51) 0.9 0.92 (0.62,1.37) 0.7

iliac 0.77 (0.66,0.90) 0.002 0.90 (0.70,1.16) 0.4 0.75 (0.58,0.97) 0.03 0.63 (0.42,0.94) 0.02 0.92 (0.59,1.44) 0.7

subscapular/triceps ratio 1.07 (0.95,1.23) 0.2 1.02 (0.84,1.25) 0.7 0.93 (0.75,1.16) 0.5 1.14 (0.78,1.65) 0.5 1.19 (0.84,1.70) 0.3

Analysis of factors associated with all-cause mortality by Cox proportional hazards modelling, and competing risk analysis for specific causes with inclusion of BMI, age, smoking, alcohol and exercise into multivariable analysis. Hazard ratios (95% confidence intervals) and significances in superscript are shown. Standardised data was entered for continuous variables.

19

Legend to Figure 1

Increased risk of death in overweight and obese white males with low iliac skinfold thickness. Hazard

ratios (HR) and 95% confidence intervals (95%ci) are shown for mortality stratified by quartiles of iliac

skinfold: q1 3.3-9.7mm; q2 9.8-13.7mm; q3 13.8-20.0mm relative to q4 >20.0mm in normal weight

(diamonds), overweight (circles) and obese (squares) individuals. P values are shown for quartiles in

which there was a significant effect.

20

Figure 1

21