Validation and application of objective measures of obesity and physical activity: studies in pregnant and non-pregnant adults and in infants

(1)

Umeå University Medical Dissertations, New Series No 1422

Validation and application of objective measures of obesity and physical activity: studies in pregnant and non-pregnant adults and in infants

Swedish title: Validering och tillämpning av objektiva mätmetoder för obesitas och fysisk aktivitet: studie av gravida och icke-gravida vuxna och av spädbarn

Anna Gradmark

'HSDUWPHQWRI3XEOLF+HDOWKDQG&OLQLFDO0HGLFLQH

8PHn

(2)

Responsible publisher under Swedish law: the Dean of the Medical Faculty This work is protected by the Swedish copyright legislation (Act 1960:729) ISBN:978-91-7459-216-0

ISSN: 0346-6612

Cover by: Therese Fara Gradmark

E-version available at http://http://umu.diva-portal.org/

Printed by: Print & Media, Umeå University Umeå, Sweden

(3)

Till min familj

Men, kan vi? vill vi? törs vi?

Ja, vi kan! Vi vill! Vi törs!

(Margareta Garpe och Suzanne Osten)

(4)

^dZd

Background Excess adipose tissue and low levels of physical activity are two major determinants for chronic diseases such as type 2 diabetes and cardiovascular disease.

Details of these relationships require accurate measures of body composition and physical activity, and most existing observational studies lack such measures. Paper I to III in this thesis address the validity of measures of physical activity and abdominal adipose mass. In paper IV and V, we explore the relationships between obesity and physical activity on metabolic health in non-pregnant and pregnant women and their offspring.

Methods and Results Two hundred men and women representative of the Northern Sweden EPIC cohort were recruited for Paper I. A questionnaire on physical activity (PAQ) was validated against objectively measured physical activity energy expenditure (PAEE) using a combined heart rate monitor and accelerometer (Actiheart) worn for at least four days. A categorical physical activity index (Cambridge index) calculated from the PAQ using information on occupational activity and time spent in cycling and sports showed the strongest correlation with PAEE (r=0.32 p<0.05) and moderate-to vigorous physical activity (r=0.21, p<0.05), respectively. In Paper II, body composition was assessed in 29 adult men and women using anthropometric measurements, dual energy x-ray absorptiometry (DXA) and ultrasound measurements which, were compared to the gold-standard method of computed tomography (CT). Waist circumference showed the highest correlation with CT-assessed visceral (r=0.85, p<0.0001) and subcutaneous adipose tissue (r=0.86, p<0.0001) followed by BMI (r=0.67 and 0.71 respectively, p<

0.001). Adipose thickness was best assessed with ultrasound (visceral r=0.89, subcutaneous r=0.93, deep subcutaneous r=0.84, all p<0.0001). In Paper III, the validity of a wrist-worn triaxial accelerometer (GENEA) was assessed in 32 pregnant and 74 non-pregnant women using doubly-labelled water (DLW) as the criterion measure. The output from the accelerometer explained 24% (p <0.001) of the variation in PAEE in non-pregnant and 11% (p<0.05) in the pregnant women. In Paper IV, 34 pregnant and 74 non-pregnant women underwent a 75g oral glucose tolerance test and habitual energy expenditure and physical activity were assessed for 10 days by DLW and the Actiheart monitor. The amount of total physical activity was inversely associated with early insulin response in both pregnant (r=-0.47, p=0.007) and non-pregnant (r=-0.36 p=0.004) women. Pregnant women spent more time sedentary (p<0.0001) and had lower PAEE than non-pregnant women (p=0.045). In, Paper V, 32 women and their offspring (n=33) were studied 4 months post-partum. Body composition was quantified using DLW in the women and air-displacement plethysmography in the infants, and measures of glucose tolerance from OGTT during gestational week 28-32 (paper IV) were obtained. Mid- pregnancy weight gain was significantly associated with infant fat mass (r=0.41,

(8)

p=0.022), whereas late-pregnancy weight gain was associated with infant fat-free mass (r=0.37, p=0.04).

Conclusion This work describes new methods as well as conventional anthropometric estimates and a questionnaire, that provide relatively strong estimates of body composition and physical activity which could be used in larger studies. Pregnant women were shown to have more sedentary behavior than non- pregnant women but physical activity appeared to have equal effects on glucose homeostasis in both groups, which may help guide lifestyle interventions in pregnancy. The impact of weight gain during the different trimesters seems to differentially affect the offsprings´ body composition in early infancy, which might give us clues to when different aspects of fetal development and growth occur and how modifiable lifestyle behaviors might be intervened upon to improve long-term health.

Key words: physical activity, validation, body composition, abdominal adipose tissue, methods, accelerometer, pregnancy, glucose, insulin sensitivity, offspring

(9)

^hDDZz/E^t/^,ʹ^DDE&ddE/E'W

^sE^<

Bakgrund: Livstilsfaktorer som övervikt och fysisk aktivitet utgör viktiga delar av vår hälsa både som vuxna och redan som barn. Exakta mätmetoder för att uppskatta både bukfetma och fysisk aktivtet är viktigt när man vill utvärdera deras effekter på hälsan. Denna avhandling baseras på fem arbeten varav de tre första fokuserar på mätmetoder och de andra två undersöker förhållandet mellan övervikt, fysisk aktivitet och metabol hälsa hos icke-gravida och gravida kvinnor och deras avkomma.

Metoder och resultat: Tvåhundrafyra män och kvinnor representativa för norra Sveriges EPIC-kohort rekryterades till Delarbete I. Ett frågeformulär om fysisk aktivitet validerades mot en objektiv mätmetod- en kombinerad hjärtfrekvensmätare och accelerometer (Actiheart). Denna bars i minst fyra dagar och energiförbrukningen från fysisk aktivitet räknades ut. Tre aktivitetsindex räknades ut från frågeformuläret genom informationen angående aktivitet på arbetet och tidsförbrukning för sportaktiviteter. Av dessa tre korrelerade Cambridge indexet starkast med energiförbrukningen (r=0.33 p<0.05) och med medelintensiv- till intensiv fysisk aktivitet från Actihearten (r=0.21, p<0.05). I Delarbete II mättes bukfettet med antropometri, dual energy x-ray absorptiometry (DXA) och ultraljud hos 29 medelålders män och kvinnor och jämfördes mot gold standard datortomografi (DT). Midjeomfång korrelerade starkast med DT-uppmätt area av visceralt (r=0.85, p<0.0001) och subkutant fett (r=0.86, p<0.0001), följt av BMI (r=0.67 viceralt, 0.71 subkutant, alla p<0.001). Ultraljudsmätning av tjocklek av de olika fettdepåerna korrelerade starkast med mätningar från DT jämfört med de andra metoderna (visceralt (r=0.89), subkutant (r=0.93), och djupt subkutant fett (r=0.84, alla p<0.0001)). I Delarbete III undersöktes validiteten av en handledsburen triaxial accelerometer (GENEA) hos 32 gravida och 74 icke-gravida kvinnor gentemot dubbelmärkt vatten (DLW) som användes som jämförelsemetod. Accelerometern förklarade 24% (p<0.001) av variationen av energiförbrukningen från fysisk aktivitet hos icke gravida och 11% (p<0.05) hos gravida kvinnor. I Delarbete IV genomgick 34 gravida och 74 icke gravida kvinnor ett glukosbelastningstest och habituell energiförbrukning och fysisk aktivitet mättes under 10 dagar med DLW och Actiheart. Mängden total fysisk aktvitet associerade negativt med måttet på akuta insulinsvaret hos både gravida (r=-0.47, p=0.007) och icke-gravida kvinnor (r=-0.36; p=0.004). Gravida kvinnor spenderade mer tid stillasittande (p<0.0001) och hade lägre energiförbrukning från fysisk aktivitet än icke-gravida kvinnor (p=0.045). I Delarbete V (uppföljning till studie IV) studerades 32 kvinnor och deras avkomma (n=33) 4 månader post-partum. Kvinnornas kroppssammansättning uppmättes med DLW och barnens med luftpletysmografi. Glukosbelastningstestet under gravididitetsecka 28-32 (delarbete IV) användes i analyserna. Viktuppgång

(10)

under andra trimester associerade significant med andelen kroppsfett hos barnet (r=0.41, p=0.022), medan viktuppgång under tredje trimestern associerade med barnets fettfria massa (r=0.37, p=0.04).

Slutsatser: Denna avhandling beskriver såväl nya som konventionella mätmetoder för att mäta bukfett och fysisk aktivitet. Antropometriska mätningar och ett frågeformulär om fysisk aktivitet ger relativt korrekta uppskattningar av både kroppssammansättning och energiförbrukningen från fysisk aktivitet, vilka kan användas i större studier. Gravida kvinnor var mer stillasittande än icke-gravida men fysisk aktivitet verkade ha samma effekt på glukosbalansen hos båda grupperna vilket kan utgöra en grund för livstilsförändringar gravida kvinnor. Viktuppgång under olika delar av graviditeten verkar påverka spädbarnets kroppsammansättning på olika sätt, vilket kanske kan ge oss ledtrådar när olika aspekter av fosterutveckling och tillväxt sker och hur modifierbara livstilsfaktorer kan påverkas för att förbättra hälsan på lång sikt.

Nyckelord: fysisk aktivitet, validering, antropometri, frågeformulär kroppssammansättning, abdominell adipositas, mätmetoder, accelerometer, graviditet, glukos, insulinkänslighet, avkomma, , fosterutveckling

(11)

>/^dK&KZ/'/E>WWZ^

I. Peters T, Brage S, Westgate K, Franks P.W, Tormo Diaz M.J, Bendinelli B, Jose Maria Huerta, Vigl M, Wendel-Vos W, Gradmark A, Benjaminsen-Borch K, de Lauzon Guillain B, Sharp S, Kerrison N, Langenberg C, Arriola L, Clavel-Chapelon F, Barricarte A, Boeing H, Gonzales C, Kaaks R, Key T, Tee Khaw K, May A, Nilsson P, Norat T, Overvad K, Palli D, Panico S, Quirós J.R, Sanchez MJ, Slimani N, Spijkerman A, Tjonneland A, Tumino R, Feskens E, Riboli E, Ekelund U, Wareham N. Validity of the Short European Investigation into Cancer and Nutrition (EPIC) Questionnaire to Assess Physical Activity across EPIC countries. Submitted II. Gradmark AM, Rydh A, Renstrom F, De Lucia-Rolfe E, Sleigh A, Nordstrom P, Brage, S, Franks P.W. Computed tomography-based validation of abdominal adiposity measurements from ultrasonography, dual-energy X-ray absorptiometry and anthropometry. Br J Nutr Aug 2010;104(4):582-8.

III. Van Hees V*, Renstrom F*, Wright A, Gradmark A, Catt M, Chen K, Löf M, Bluck L , Wareham NJ , Ulf Ekelund U, Brage S, Franks P.W. Estimation of daily energy expenditure in pregnant and non-pregnant women using a wrist-worn triaxial accelerometer. Submitted

IV. Gradmark A*, Pomeroy J*, Renstrom F, Steiginga S, Persson M, Wright A, Bluck L , Domellöf M, Kahn S.E, Mogren I, Franks P.W. Physical activity, sedentary behaviors, and estimated insulin sensitivity and secretion in pregnant and non-pregnant women. Submitted

V. Pomeroy J*, Renstrom F*, Gradmark A, Steiginga S, Persson M, Wright A, Bluck L , Domellöf M, Kahn S.E, Mogren I, Franks P.W. Metabolic risk-factor profiles in infants in relation to those of their mothers during pregnancy. In manuscript

* Denotes equal contribution.

ZĞůĂƚĞĚWƵďůŝĐĂƚŝŽŶ

Pomeroy J, Soderberg (Gradmark) AM, Franks PW: Gene-lifestyle interactions and their consequences on human health. Med Sport Sci 2009, 54:110-135.

(12)

Zs/d/KE^Θ&/E/d/KE^

ADP air-displacement plethysmography

BMI body mass index

BMR basal metabolic rate

BiA bioimpedance analysis

DXA/DEXA dual energy x-ray absorptiometry

CT computed tomography

DLW doubly-labelled water

DSAT deep subcutaneous adipose tissue

EPIC European prospective investigation into cancer and nutrition

GDM gestational diabetes mellitus

HC hip circumference

HW hydrostatic weighing

IGT impaired glucose tolerance

IPASSU interact physical activity sub-study Umeå MET metabolic equivalent of task

MVPA moderate-to vigorous activity MRI magnetic resonance imaging OGTT oral glucose tolerance test

PA physical activity

PAEE physical activity energy expenditure PAL physical activity level

REE resting energy expenditure

RMR resting metabolic rate

SAD sagittal abdominal diameter

SSAT superficial subcutaneous adipose tissue

TBW total body water

TEE total energy expenditure TEF thermic effect of feeding TG triglycerides TSAT total subcutaneous adipose tissue VAT visceral adipose tissue

VIP Västerbotten intervention program

WC waist circumference

WHR waist/ hip ratio

WtHR waist/height ratio

WRC whole room calorimetry

Chance- “the unknown and unpredictable element in happenings that seems to have no assignable cause” (1).

Bias- systematic deviation (2).

Confounding-“extraneous variable that may affect the dependant and independent variable thus affecting the result” (2).

(13)

Effect measure modification- “the association between a an exposure and an outcome differs according the level of a another variable” (2).

Precision- “made or done in a very exact way” (3).

Accuracy- “the quality of being correct” (3).

Validity- “is based on what is true, to be valid, the instrument must measure what it is intended to measure” (4).

Reliability- “can be trusted or depended on, to be reliable; the instrument must consistently give the same results under the same circumstances. If the instrument is reliable and valid, it is also accurate” (4).

Criterion/construct validity- “a questionnaire is validated against an objective method (usually reported as correlation coefficients)” (Adopted from U. Ekelund, not published).

Absolute validity- “the absolute outcome i.e. energy expenditure is compared to data from an objective measurement which provides the same outcome measure (usually reported with Bland-Altman method)” (Adopted from U. Ekelund, not published).

Concurrent validity- “a questionnaire is reported to another self-report instrument i.e. another questionnaire (correlation reported but can be subject to correlation error)” (Adopted from U. Ekelund, not published).

(14)

/EdZKhd/KE

Obesity is increasing worldwide (5) and in its trail follows diseases such as diabetes and cardiovascular disease (6, 7). Already in the 1950s Vague published (8) results showing that the distribution of adiposity was important for the development of these diseases and he separated two body types; gynoid (fat distributed mainly around the hips) and android (fat mainly centered around the waist and upper part of the body), where the latter is associated with increased risk of metabolic disturbances such as increased insulin resistance and increased risk of coronary heart disease(6, 9). To explore the associations between adiposity and diseases much effort has been placed on measuring abdominal adiposity and recently also dividing it into three separate anatomic compartments with different metabolic profiles (10).

Abdominal obesity or visceral adiposity has been associated with hyperlipidemia (11) and impaired glucose tolerance (12) and myocardial infarction (6).

Subcutaneous adipose tissue is separated into two compartments by the fascia superficialis where the deep subcutaneous fat has more similarities with visceral adipose tissue, associating with disturbed glucose metabolism (13) and increased lipolysis (14). This indicates that it might be relevant to divide the different compartments when conducting research on metabolic risk factors.

Physcial activity (PA) is a lifestyle factor which has many beneficial effects on human physiology with improved insulin sensitivity (15) and lower risk for cardiovascular events (16) as examples. Measuring PA can be done in subjective ways with questionnaires and diaries or objective ways with different monitors or the gold standard method of doubly labeled water. The accuracy, feasibility and costs are always factors that affect the method used in studies.

During pregnancy there is an increase in peripheral insulin resistance by up to 50%

(17) which is counterbalanced by an increase in insulin secretion. An excess of adipose tissue increases the risk of impaired glucose tolerance (IGT) and the increased obesity epidemic has led to more pregnant women are being overweight hence an increased risk of gestational diabetes, birth complications such as macrosomia and preeclampsia (18). The beneficial effect of PA that is seen in non- pregnant populations (19) can also be seen in pregnancy (20). Lifestyle factors such as overweight and low PA also effects the in utero environment where an increased risk of childhood obesity have been seen in children born of obese mothers with diabetes (21). This is believed to be due to fetal programming which means that the gene transcription and modulation in the immature organs of the fetus is being affected by the milieu in the uterus (22). Childhood obesity in turn is associated with premature death (23) so to strive for a healthier lifestyle among the population would not just benefit the people today but also our children and their future health.

(15)

<'ZKhE

DĞĂƐƵƌĞŵĞŶƚƐŽĨďŽĚǇĐŽŵƉŽƐŝƚŝŽŶ

For the past two centuries, scientists have worked towards improving the methods for assessing body composition, mainly because of the confirmed associations between excess body weight and health risks. A number of significant breakthroughs were made during the 1900s, namely the development of the isotope dilution method, densitometry, whole body potassium count, and imaging techniques such as dual energy x-ray absorptiometry (DXA) and computed tomography (CT) and magnetic resonance imaging (MRI) (24). The assessment of body composition is of importance for many different types of studies investigating public health and diseases, but the methods used vary in accuracy, precision, and feasibility. Below is an overview (Figure 1) of different commonly used methods for measuring body composition, along with an analysis of respective method’s accuracy and feasibility. An overview of different methods, their benefits and disadvantages are found in the end of this section (Table 1).

Figure 1. Overview of common methods for measuring body composition, along with accuracy in estimation of body fat and feasibility (adapted from U. Ekelund, not published).

(16)

ŶƚŚƌŽƉŽŵĞƚƌǇ

Anthropometry is the most commonly used method to assess body composition because it is possible to use in large-scale studies and it does not require expensive equipment. Anthropometry usually refers to measures of weight, height, circumferences of waist and hip, and various ratios of these measures, as well as sagittal abdominal diameter (SAD). The measures describe body mass, size and shape, and amount of fatness (25). Weight and height is measured using a calibrated scale and a stadiometer, respectively (26). Weight is influenced by changes in body water, amount of lean and adipose tissue, and age. Stature is closely related to weight and therefore it is important to consider height when evaluating body composition. The ratio between weight in kg and height in meters squared (kg/m²), body mass index (BMI), is often calculated to provide an estimation of overweight or underweight. The thresholds for BMI are underweight <18.5, normal weight 18.5–24.9, overweight 25–29.9 and obese 30 (27). Waist circumference (WC) is used as an estimate of abdominal adiposity, but this measurement does not separate subcutaneous from visceral adipose tissue. WC is measured using a non-stretchable tape at the midpoint between the 12^th rib and crista iliaca. Hip circumference (HC) is measured at the position trochanter major or sometimes at the broadest part of the hip. HC and WC are also used in calculations of ratios, such as waist/hip ratio (WHR) or waist/height ratio (WtHR). These ratios are used as crude estimates of body habitus and provide an index of body composition. WHR >0.85 for women and >1.0 for men is associated with an increased risk for cardiovascular diseases and diabetes (28). SAD, which is a fairly new method, is measured at the iliac crest (L4- L5) during expiration with the subject lying down with bent knees. The distance from the examination table to the top of the stomach is measured and this measure has been associated with insulin resistance in obese men (29). WC has been found to associate more strongly with cardiovascular risk than BMI and may thus be a better indicator of disease risk (30). WHtR performed slightly better than WC, but in the same study WC, BMI, and WHtR were highly correlated and all three methods correlated strongly with DXA estimates of body fat %, (r =0.65-0.87). WHtR was considered to be a better method for assessing overweight and obesity when investigating health risk in general (31). Anthropometry, especially WC, seems to better estimate visceral than subcutaneous adipose tissue when compared to CT (32).

In general, these methods are all easy to use as well as portable and fairly inexpensive but they are unable to separate different sources of adipose tissue and results may be affected by the distribution of fat and muscle tissue and also by sex and age (33).

^ŬŝŶĐĂůůŝƉĞƌƐ

Percent body fat is estimated by measuring subcutaneous adipose tissue thickness in folds at specific places on the body. The skin is gently pinched and stretched as a

(17)

fold so only skin and underlying fat is measured. A skinfold calliper (Figure 2) is placed where a standardized pressure per unit of calliper jaw surface is used and the thickness is measured. Repeated measurements are undertaken and the average is used in equations to calculate total body fat percentage or characterisations of the distribution of subcutaneous adipose tissue (26). One side of the body is usually measured and the most common places are: triceps, pectoralis (in front of the armpit), subscapular, mid-axillar line, abdomen, iliac crest, and quadriceps, but biceps and calf are sometimes also measured (34). The prediction of total body fat using this method is influenced by several things, such as age (young people’s skin is more hydrated which makes the compression of the fold easier) and individual and population differences exist (e.g. pregnancy and obesity). In very obese persons it is more difficult to technically achieve reproducible results as skinfolds are much thicker in these individuals and the callipers cannot expand enough to pinch the skinfold. The prediction also depends on which sites have been used as some sites are more related to total body fat than others. There is reference data for different populations, e.g. children, that can be used when evaluating the result from skinfold measurements (26). The benefits of the skin calliper method are that it is inexpensive, easy to teach and learn, and it is portable and thus easy to use in field studies. The method is useful when estimating changes in body composition, especially subcutaneous adipose tissue, in the same individual, for example when evaluating a treatment. The downsides are that it is operator-dependable and that accuracy is affected by factors mentioned earlier (35).

Figure 2. Skin calliper, photo from Alaitz Poveda.

ŝŽŝŵƉĞĚĂŶĐĞ;ŝͿ

BiA is based on the knowledge of passive electric properties of biological tissues (Figure 3). Fat mass is considered a poor conductor whereas fat-free mass with a high water and electrolyte content conducts current easily (36). Depending on the design of the device, the subject is either placed with a handle in each hand or standing with both feet on metal plates, and low current is allowed to pass through the body. The estimated impedance index, stature squared divided by resistance (S²/R) at a constant frequency, is proportional to the volume of total body water.

Fat-free mass is then estimated based on the assumption that 73% of the body’s fat-

(18)

free mass is water (FFM/TBW=0.73) (25). BiA measurements are affected by factors that influence the amount of water retained in the body. Examples of such factors and physiological states are obesity, age, pregnancy, and chronic diseases (37). Several different population-specific equations have been developed for calculating fat mass from BiA; these equations may not be applicable to all populations (38) but when used in the correct population, high correlations (r=0.92 and 0.86 for FFM and FM, respectively) can be achieved when compared to the isotope dilution method (39). Methods based on bioimpedance also require that the individual is fasted and rested as these factors may affect the hydration of the body.

BiA is portable and hence suitable for field studies, it is easy to use, safe (not recommended for patients with pacemaker though), and cost efficient (40).

Figure 3. Bioimpedance equipment, photo from Alaitz Poveda.

/ƐŽƚŽƉĞĚŝůƵƚŝŽŶŵĞƚŚŽĚ

The assessment of total body water (TBW) has been a common method for measuring body composition since the 1930s when isotopes of hydrogen and oxygen were first discovered (41) although the method did not come into use until the 1980s. TBW is measured at molecular level using the dilution principle, which states that the volume of the compartment is equal to the amount of tracer divided by the concentration of the tracer in that compartment (42). Different substances have been used as tracers, e.g. ethanol, urea, and antipyrin, but they are not equally distributed in the body and nowadays is isotope-labelled water most commonly used (24). The isotope dilution method is based on four assumptions:” i) that the tracer is only distributed in the body water; ii) that the tracer is equally distributed in the different compartments of the body that contains water; iii) that the equilibration rate is rapid; and iiii) that neither tracer nor body water are metabolised during the time of equilibration” (24). The individual is given a weight-dependant dose of deuterium after leaving a sample of body fluid (blood, saliva or urine). After a couple of hours when equilibrium is obtained, an additional sample is taken and analysed with spectroscopy. TBW is then calculated based on the dilution of the tracer and the assumed exchange rate of the tracer to non-water compounds. This method is easy to

(19)

distribute among participants in a study, who then only need to leave samples of body fluid at a later time (a simple saliva sample is sufficient). This method has been and is commonly used as a criterion measure when validating bioimpedance and DXA. The accuracy of the isotope dilution method in measuring TBW is very good, it is only affected by the uncertainty of the estimate of nonaqueous exchange, which is about 1%, (24) and the uncertainty concerning the hydration constant, about 2% (43). The analyses and the isotopes are expensive and the calculations of fat-free mass and fat mass are based on the correlation coefficient of 0.738 which may be affected by obesity or pregnancy, and by physiological states that influence how much water is retained within the body. TBW has been shown to vary from approximately 75% during infancy to 40% in obese adults (24).

,ǇĚƌŽƐƚĂƚŝĐǁĞŝŐŚŝŶŐ;,tͿ

HW (or underwater weighing) was long considered to be the gold standard for estimating body composition and it has been commonly used as a criterion measure when validating other methods. It builds on the principal of Archimedes where the volume of water being pushed away by an object is equal to the volume of that object (44). First, the person is weighed on land and then put in a container with water at a constant temperature. The individual exhales, holds his or her breath and is submerged for 3–5 seconds; subsequently, the weight is registered and the volume of displaced water is measured. Multiple measures are usually taken. Body volume is calculated as the difference between the individuals´ weight on land and the weight of displaced water and then corrected for the density of water. This is put into an equation to calculate lean mass (bone, muscle) and fat mass corrected for residual lung volume, which can be measured using different tests or a fixed value, and gastrointestinal gas using a fixed value (44). This method is dependent on calculations of gastrointestinal gas volumes and assumed relationships between body weight and density; it is further not applicable to small children or individuals with disabilities where submersion under water is difficult. Errors in estimating body fat% are low using the HW method; the combination of biological variations in fat-free mass density and technical errors correspond to approximately a change in 2% fat estimates (24). The HW method involves no radiation and is therefore safe to use, but it requires more advanced equipment and is not portable.

ŝƌͲĚŝƐƉůĂĐĞŵĞŶƚƉůĞƚŚǇƐŵŽŐƌĂƉŚǇ;WͿ

ADP is based on the same principal as HW but instead of water, air is displaced and the body volume is measured. This method utilises the different gas laws (Boyles:

P1/P2=V1/V2 and Poisson: P1/P2= (V1/V2)^Ȗ, P=pressure, V=volume and Ȗ= ratio between the gas heat at constant P/V, for air=1.4). Boyles’ law was initially used but it required isothermic conditions, so from the mid-90s when ADP became available for commercial use, the law of Poisson, which does not require constant temperature

(20)

during measurement (adiabatic condition), was implemented. Bod Pod is the most commonly used devise based on ADP, and it consists of two chambers, i.e. one measuring chamber (450 litres) and one reference chamber (300 litres). After the initial calibration step the individual is measured for about 50 seconds. This procedure is usually repeated and the mean of the measurements is used in calculations. Sources of isothermic gas must be minimised and accounted for. A tight swimsuit and cap can be used to minimise the effect of isothermic gas close to the body surface and hair, and the effect (surface area artefact (SAA)) of that air is also accounted for by the software. Thoracic gas is also a source of isothermic gas which can be measured, alternatively a fixed value can be used (45). The benefits of ADP are several: it is quick and non-invasive and it can be used on various study populations such as children and people with disabilities. Disadvantages include equipment requirements, restricted transportability, and in order to obtain a more accurate measurement, lung volume measurements are also necessary. Studies comparing HW and ADP have found good correlations with an explained variance of r² 0.72–0.94 (46, 47) and of 2–4% under- and overestimations of body fat% (48, 49). Differences between the two methods may depend on if or how lung volume was measured and SAA (45). Bod Pod seems to underestimate body fat% when compared to DXA by approximately 2.5–4% (50, 51). Pea Pod (Figure 4) is an ADP system for measuring body composition in infants. It was first validated in 2004 against the isotope dilution method (deuterium, ²H2O) which showed a high r² of 0.76 and a difference in mean body fat% of 0.07% between the two methods (52). A later study compared Pea Pod and the four-compartment model (four quantities are measured using different techniques: body volume, total body water, bone mineral, and body mass). The study showed an equal r² of 0.73 and further mean body fat%

did not significantly differ between the two methods (53).

Figure 4. Four-month old infant in a Pea Pod during the Lifegene pre-pilot study.

(Published with permission from parent).

(21)

ƵĂůĞŶĞƌŐǇǆͲƌĂǇĂďƐŽƌƉƚŝŽŵĞƚƌǇ;yͬyͿ

This technique is about 20–30 years old (24) and is a non-invasive method based on x-ray attenuation to assess bone mineral density, fat mass, and bone-free lean mass (Figure 5, A and B). Two low-energy beams are projected during approximately 15 minutes while the individual is lying down on the machine wearing no metal parts (jewellery etc.) yielding different attenuation depending on which tissue is scanned.

The ratio (R) of the attenuations measured at these energies is used in calculations of the different tissue compartments. Bone is quite easily separated from soft tissue, and in bone-free tissue calibrations make it possible to separate lean and fat mass (25, 54). Different shapes of the beam, i.e. pencil and fan, are used. The pencil beam was the one originally used and the fan beam has been developed more recently allowing a wider beam thus a faster scanning (55). There appears to be differences in the estimation of body fat between the two techniques but they can also be combined (personal communication Jesper Marmstad, Region Sales Manager Nordic, GE Healthcare Lunar). Furthermore, different brands and software affect the results, making it problematic to compare studies (56, 57). The benefits of DXA are many. Firstly, the low radiation dose corresponds to about 1–10% of an ordinary chest x-ray. Secondly, it provides accurate and reliable estimates of total percent body fat (r approximately 0.85) (58), fat mass ( r²=0.94) (59), and abdominal fat mass (r= 0.86) when compared to the four-compartment model and multislice CT (60). Thirdly, DXA also appears to have good inter-operator reliability for estimating abdominal adipose tissue (60). The radiation, although low, has the disadvantage of making the method inappropriate for use in some study populations, e.g. pregnant women. The machines have limitations when it comes to an individual’s size and weight, which means extremely obese study populations (25) are excluded, and as mentioned previously, results vary between brands and software (56).

(22)

A: B:

Figure 5. A) DXA used during IPASSU and Lifegene pre-pilot study. B) A typical x-ray image obtained by DXA.

hůƚƌĂƐŽƵŶĚ

Ultrasound is sound waves above the hearing range (15–20 kHz) but in assessment of body composition or in other clinical practices frequencies of 2–10 MHz is usually used. Sound waves are emitted by a transducer placed on the individual; the echo when the waves hit the tissue generates an image. Higher resolution is obtained with increased frequencies but deeper penetration is achieved with lower frequency and thereby less resolution (61). Ultrasound is used to assess regional adiposity, such as abdominal or pericardial. By using different transducers and frequencies it is possible to separate the different compartments of abdominal adipose tissue (subcutaneous: deep and superficial, and visceral) and to measure their thicknesses.

This can be done with good accuracy when compared to MRI (r=0.90 and 0.73 for visceral and subcutaneous adipose tissue, respectively (62), and also with CT area of intraabdominal fat (r=0.81) (63). Ultrasound estimates of adiposity also correlated strongly when added to a model to calculate total body fat% and compared to DXA (r=0.98) (64). The equipment required for ultrasound measures is portable, less expensive than equipment for other imaging techniques, such as CT and MRI, and gives no radiation exposure. The disadvantage is that it is operator-dependant and can only visualise thickness and is not currently adapted to estimate area or volume.

ŽŵƉƵƚĞĚƚŽŵŽŐƌĂƉŚǇ;dͿĂŶĚDĂŐŶĞƚŝĐZĞƐŽŶĂŶĐĞ/ŵĂŐŝŶŐ;DZ/Ϳ

These techniques have become more popular in measuring adiposity and especially regional adiposity. The methods of multi-slice and single-slice CT and MRI were developed and initially used in the 1990s (24). CT is based on an x-ray tube and a receiver that rotate around the individual who is scanned and two-dimensional images are calculated from the different attenuations generated by the scan. These images are reconstructed with different mathematical techniques (24). The attenuation is expressed as the linear attenuation coefficient (CT number) which is a

(23)

measure of attenuation relative to water and air (CT number= -1.000 and 0 Hounsfield unit (HU)). The CT image consists of pixels (1 mm squares) that are assigned different HU values depending on the composition of the tissue (looks like different shades of grey). Tissue with low density, such as adipose tissue, yield a lower HU than tissues with high density, such as bone. This method provides a fast and accurate image of the different tissues in the body and is considered to be gold standard when assessing abdominal adiposity. However, the high radiation level makes it less suitable for whole body scanning and it is expensive and not portable.

MRI is based on magnetic fields that affect the hydrogen protons in the body. The individual is moved through a cylindrical magnetic coil which makes the body’s protons align, a pulsed radio frequency signal provides energy to the protons which is released when the pulse is switched off making the protons gradually return to their original state. During this process, the energy is released as radio frequency signals that are collected and used to provide a two-dimensional image (24). The benefits of MRI are the fairly quick procedure, approximately less than 30 minutes, and as opposed to CT no radiation is involved which allows for the measurement of small children and pregnant women. However, the equipment is very expensive and not portable and the size and weight restrictions exclude obese populations from being studied.

(24)

Table 1. Summarising the methods previously described and their benefits and disadvantages. Methods How?Benefits Disadvantages Anthropometry (body mass index, waist circumference etc.)Body dimensions are measured. Ratios can be calculatedLow cost. Easy to use. PortableNo detailed info. Indirect measure only Bioimpedance (BiA) Low electric current is passed through the body. The impedance index is proportional to the volume of total body water. Equations used to calculate fat and fat-free mass

Low cost. Easy to use. PortablePA, pregnancy, and obesity can influence measurements due to their effect on body hydrogeneity Skin callipersSkinfolds at specific location on the body is measured using a calliper. Equations used to estimate total body fat %Low cost. PortableOperator-dependant. Technically difficult obese individuals. Dependant on differen Hydrostatic weighing (HW) The person is submerged under water and the volume of water is displaced equal to the volume of the body. Equations used to calculate fat and fat-free mass Historically considered gold standard. Low error rate; errors of approximately 2% when estimating body fat Requires more advanced equipment. Not s for all types of participants, e.g. children. A by measurements or calculations of gas (res and gastrointestinal) Air-displacement plethysmography (ADP) Same principle as HW but air is displaced by the body instead of water. Equations used to calculate fat mass, fat-free mass Possible for small children and disabled. Variation of about 6% in estimated body fat compared to HW

Affected by measurements or calculation (respiratory and gastrointestinal) Isotope dilution methodTotal body water is calculated using an isotope marker that binds to the water molecule. Equations used to calculate fat mass, fat-free mass

Easy to distribute. PortableExpensive. Not possible to separate differe adipose compartments Dual energy x-ray absorptiometry (DXA/DEXA) The person is scanned by low radiation. Body fat, lean mass, and bone mineral content can be extracted from the examination Good estimates of body fat mass. Exposure to low dose of ionizing radiation. Quick

Radiation. Not so easily portable. Brands an software variations. Body size can be a lim not possible to use with extremely obese individ UltrasoundUsing a probe that emits ultrasound and creates an image from the echo to measure, e.g. abdominal fat, cardiac fatGood estimates of regional adiposity. No radiation. PortableNo whole body scan. Operator-dependant measure of thickness is obtained Computed tomography (CT) Radiation technique. Clear image of body content. Area/volume calculations of adipose tissueAccurate estimation of body tissue. QuickHigh radiation dose. Not suitable for childr pregnant women. Expensive. Not portable Magnetic resonance imaging (MRI)Non-radiation technique where whole body scan is possible. Clear image of body content. Area/volume calculations of adipose tissue

No radiation. Safe. Accurate estimations of body tissue. Fairly quick Expensive, not portable. Size limitation, n possible for very obese individuals

(25)

DĞĂƐƵƌĞŵĞŶƚƐŽĨƉŚǇƐŝĐĂůĂĐƚŝǀŝƚǇĂŶĚĞŶĞƌŐǇĞǆƉĞŶĚŝƚƵƌĞ

It is impossible to assess habitual PA with absolute precision or accuracy. PA must be measured using valid methods as more and more studies are exploring the relationship between lifestyle factors, such as low PA and obesity, and associated diseases, such as cardiovascular disease and diabetes. To date, in excess of 30 different methods have been used to assess PA in population studies (4). PA is defined as “any bodily movement produced by skeletal muscles which results in energy expenditure” (65) and is not synonymous with energy expenditure, which relates to energy expended from PA but also from factors such as digestion and other basic bodily functions. Physical activity energy expenditure (PAEE) is measured in kJ or kcal where 4.184 kJ equals 1 kcal. While PA is related to movement, physical fitness is “a set of attributes that people have or achieve” that are associated with cardiorespiratory and muscular endurance (65). As fitness is influenced by factors such as gender and age, it is not an adequate measure of PA.

Basal metabolic rate (BMR) refers to the minimum amount of energy required for all basic bodily functions (e.g. breathing and organ function) and constitutes about 45 to 75% of total energy expenditure (TEE) in adults. BMR is generally determined by gender, age, body size, and body composition (66). Resting metabolic rate (RMR) is comparable to BMR and is measured through indirect (Figure 6) or direct calorimetry with the subject awake, in a supine position, fasted and rested for 8–10 hours prior to the measurment (66).

Figure 6. Shows measurement of RMR using indirect calorimetry, with a ventilated hood. Research nurse Monica Holmgren during the IPASSU study.

Metabolic equivalent of task (MET) or just metabolic equivalent is a concept of turning modes of activity into an estimate of PA. MET is defined as the ratio between metabolic rate during a specified activity and at rest or multiples of RMR;

(26)

it equals 3.5 ml O2/kg x min or 1 kcal/kg x min. One MET is also equal to the metabolic rate obtained by a subject when he or she is sitting still quietly and 2 METs is equal to twice the amount of energy a person would consume at rest. MET ranges from 0.9 at sleeping to 18, which corresponds to vigorous activity such as fast running (67). These estimates of MET during different activities are often derived in adult lean men and may not reflect the true value of expenditure in other populations (68). METs are commonly used in epidemiological research for converting subjective measures of PA, i.e. questions on PA, into estimates of energy expenditure. One of the problems with METs is generalizability; an activity does not necessarily consume the same amount of energy in all individuals as factors such as fitness and lean mass influence metabolic rate (68). It has also been shown that the assumed RMR value of 3.5 ml/kg/min is inaccurate for many individuals and that a value of 2.5 ml/kg/min appears to be more true (68).

Physical activity level (PAL) is an index which is often used to categorise individuals into different levels of activity and it corresponds to the energy required for usual daily activity (TEE/BMR during 24 hours). A daily PAL of 1.75 and higher is considered desirable in order to maintain a healthy lifestyle and minimize an individual’s risk of becoming overweight or obese (66), and an individual with a recorded PAL of 1.6 or lower is considered sedentary (69). PAL usually ranges from 1.4–2.4 (66) and is an imprecise measure since it is based on measurements of TEE and BMR, which both include random measurement errors (70, 71). Both MET and PAL are based on ratios that require the two variables in the ratio to be proportionally related in order to avoid bias. This proportionality is fairly rare in biology and the use of ratios in MET and PAL may bias energy expenditure measures and make interpretation difficult (72).

PA can be classified and studied in different entities: duration, intensity, frequency, volume, and mode/type. Duration marks how long the individual engages in PA, intensity is sometimes measured as a percent of maximum volume of oxygen consumed (VO2max), frequency is how often PA is undertaken, e.g. times per day or week, volume is often estimated as an average of TEE per day, time in activity or steps/day, and mode or type is which type of activity is performed, e.g. sports or gardening. It is interesting to examine all of these different aspects of PA and it is often necessary to combine the methods available for estimating PA to measure the entire spectrum. If this is unrealistic, it is important that the right method is selected based on the hypothesis of the study. The methods utilised to measure PA are divided into subjective and objective or indirect and direct. Table 2 below gives a short description of common methods, along with their advantages and disadvantages.

Validation and application of objective measures of obesity and physical activity: studies in pregnant and non-pregnant adults and in infants