AN ULTRASOUND-BASED ADIPOSITY ASSESSMENT SYSTEM. Nicholas Robert Anderson B.Sc. Hons (Human ~iology) Loughborough, England, 1982

Transcription

1 AN ULTRASOUND-BASED ADIPOSITY ASSESSMENT SYSTEM Nicholas Robert Anderson B.Sc. Hons (Human ~iology) Loughborough, England, 1982 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in the Department of NicholasRobertAnderson 1985 SIMON FRASER UNIVERSITY December 1985 All rights reserved. This work may not be reproduced in whole or in part, by photocopy or other means, without permission of the author.

2 APPROVAL Name : Degree: Nicholas Robert Anderson Master of Science in Kinesiology Title of thesis: An Ultrasound Based Adiposity Assessment System Examining Committee: Chairman : Professor M.V. Savage Dr. W.D. Ross Senior Supervisor - Dr. D. Hedges Dr. R.L. Mirwald External Examiner Date Approved: 7 January 1986

3 PARTIAL COPYRIGHT LICENSE I hereby grant to Simon Fraser University the right to my thesis, project or extended essay (the title of which is shown lend below to users of the Simon Fraser University L ibrary, and to make part ial or single copies only for such users or in response to a request from the library of any other university, or other educational institution, on its own behalf or for one of its users. I further agree that permission for multiple copying of this work for scholarly purposes may be granted by me or the Dean of Graduate Studies. It is understood that copying or publlcation of this work for financial gain shall not be allowed. without my written permission. Title of Thesis/Project/Extended Essay Author: (s ignaturk) (name 1 (date)

4 ABSTRACT The prediction of percent body fat from anthropometric variables based on a density criterion has been challenged. The inherent assumption of constant density of a fat and a non-fat compartment is vulnerable to the extent that there is variation - in the density of the non-fat compartment, An alternate method, the 0-Scale system, simply rates adiposity from the size-adjusted total of six skinfolds. However, this too is vulnerable since there are individual differences in the extent to which tissue is compressed by the caliper. Moreover, the caliper method becomes impractical when measuring obese subjects. The thickness measured is limited to the maximum opening of the caliper's jaws. The purpose of the thesis was to overcome these problems by developing an ultrasound based assessment system. Three sets of measurements using each of: 1) The Echoscan 1502 ultrasound system, 2) Harpenden calipers and 3) Slimguide calipers were made at six sites on thirty males and thirty-four females. Intra-observer reliability correlation coefficients for all instruments, at all sites, were R = 0.99 and above. Mean intra-observer technical errors of measurement were 0.66mm for ultrasound, 0.82mm for the Harpenden caliper and 0.78mm for the Slimguide caliper. The same measurement protocol was completed on a sample of ten subjects by each of three investigators. Mean inter-observer iii

5 reliability coefficients were R = 0.98 for all three instruments. All instruments were equally objective in terms of inter-observer error at a given measured thickness. Males and females had similar levels of adipose tissue compressibility. There was a significant site difference in compressibility. Mean compressibility across six sites ranged from 20.7 percent to 47.5 percent. Equations relating ultrasound to caliper measures of adipose tissue thickness at each site were developed for males and females and used to transform skinfold norms to ultrasound equivalents. A micro-computer program was developed to facilitate classification of a subject's ultrasonically assessed adiposity, proportional weight and proportional skinfold profiles with respect to size-adjusted same age and sex norms. In conclusion, ultrasound is at least as reliable and objective in adipose tissue thickness measurement as either caliper and has the advantage of not compressing the tissue.

6 ACKNOWLEDGMENTS I would like to express my appreciation and gratitude to all those who helped complete this thesis. Special thanks are due first to Dr. W.D. Ross for his infectious enthusiasm,. unfailing support and continuous guidance from the conception of the thesis to the completion of the final dhft. I am also indebted to Dr. D. Hedges and Dan Hale who meticuously proofread and edited the earlier drafts. I would also like to show my gratitude to Dr. D. Goodman and Dr. M.A. Stephens for the time and trouble they took in giving me statistical advice. I am particularly grateful to Richard Ward for his general support and counsel. i would especially iike to thank Mavis Nordstrom and Paul Verlaan for their part in data collection, data recording, and recruitment of subjects. I also owe a debt of gratitude to Dr. Al-Kassim and his colleagues, PAR scientific Instruments APS, Herluf Trollesvej 8, 5220 Odense, Denmark, for the development of the ultrasound instrumentation and for their role in facilitating the study. I would like to thank all my subjects and those who helped find them, especially Suzanne Bell of Fitness and Fashion Enlarged.

7 I am appreciative of grant support from the Natural Sciences and Engineering Research council of Canada and of earlier support to Dr. W.D. Ross by Fitness Canada which aided in the development of the 0-Scale system.

8 TABLE OF CONTENTS Approval... ii Abstract... iii Acknowledgments... v List of Figures... xi. List of Tables... xiii Preface... xv 1. CHAPTER 1: LITERATURE REVIEW... \ 1 HEIGHT-WEIGHT RATIOS 1.2 DENSITOMETRY AND OTHER LABORATORY METHODS OF BODY COMPOSITION ANALYSIS... 2 The measurement of density... 3 The conversion of density to percent fat... 3 Variation in the density of the fat compartment... 4 Variation in the density of the fat free compartment ANTHROPOMETRIC TECHNIQUES... 8 Caliper measurement of skinfold thickness... 8 Ultrasonic measurement of subcutaneous adipose tissue (SAT) thickness measurement PREDICTION OF PERCENT FAT FROM ANTHROPOMETRY LIMITATIONS OF THE PREDICTION OF PERCENT FAT Densitometric validation criterion Precision of measurement.....l2 Assumptions in converting skinfold thicknesses to total body fat..., THE 0-SCALE SYSTEM: AN ALTERNATIVE TO PERCENT FAT PREDICTION I6 vii

9 . 1.7 THE ULTRASOUND ALTERNATIVE A COMPARISON OF THE VALIDITY OF ULTRASOUND AND CALIPERS AS MEASURES OF SAT THICKNESS CHAPTER 2: GENERAL PROCEDURE SUBJECTS MEASUREMENT PROTOCOL SKINFOLD TECHNIQUE ULTRASOUND TECHNIQUE CHAPTER 3: INSTRUMENTATION SKINFOLD CALIPERS.. $ Specifications ULTRASOUND Specifications Nature of Ultrasound (wells. 1977) Properties of ultrasound (.Wells. 1977) Generation of ultrasound Pulse-echo measuring technique Calibration Validity; a pilot study CHAPTER 4: RELIABILITY ASSESSMENT INTRODUCTION RELIABILITY OF ULTRASOUND AND CALIPER TECHNIQUES (ONE OBSERVER) METHOD DATA TREATMENT... RESULTS RELIABILITY OF ULTRASOUND AND CALIPER TECHNIQUES

10 (THREE OBSERVERS) METHOD DATA TREATMENT RESULTS DISCUSSION OF RESULTS CHAPTER 5: OBJECTIVITY ASSESSMENT INTRODUCTION METHOD DATA TREATMENT RESULTS DISCUSSION OF RESULTS CHAPTER 6: SITE AND SEX SPECIFICITY OF COMPRESSIBILITY INTRODUCTION METHOD DATA TREATMENT AND RESULTS DISCUSSION OF RESULTS CHAPTER 7: DEVELOPMENT OF NORMS INTRODUCTION METHOD RESULTS INTRODUCTION CHAPTER 8: DEVELOPMENT OF A COMPUTERIZED ANALYSIS AND DISPLAY SYSTEM A COMPARISON OF THE ULTRASOUND SYSTEM WITH THE ORIGINAL CALIPER SYSTEM LIMITATIONS OF THE ULTRASONIC ASSESSMENT SYSTEM CHAPTER 9: SUMMARY DISCUSSION

11 REFERENCES APPENDIX A: ABBREVIATIONS APPENDIX B: MEANS AND STANDARD DEVIATIONS FOR SAMPLE APPENDIX C: MEANS AND STANDARD DEVIATIONS FOR SAMPLE APPENDIX D: TRANSFORM EQUATIONS

12 LIST OF FIGURES F I GURE PAGE 1.1 Sources of intra- and inter-observer error in anthropometry Assumptions implicit in the use of skinfold calipers to predict body fat Caliper and Ultrasound measurement sites The Echoscan 1502 Ultrasound scanner Absolute intra-observer errors for one observer Relative intra-observer errors for one observer Absolute intra-observer errors for three observers (averaged over observers) Absolute intra-observer errors for three observers (averaged over sites) Absolute inter-oberver errors Inter-observer relative errors Site and sex differences in compressibility (U. ~/2) Changes in compressibility with tissue thickness Changes in percent compressibility with tissue thickness Site and sex differences in compressibility for a given tissue thickness Regression: caliper vs. ultrasonic SAT thicknesses (size-adjusted sum of six sites. males) Regression: caliper vs. ultrasonic SAT thicknesses (size-adjusted sum of six sites. females) Caliper-based physique analysis printout for subject A

13 Ultrasound-based physique analysis printout for subject A ; Caliper-based physique analysis printout for subject B Ultrasound-based physique analysis printout for subject B Differences between caliper- and ultrasound-based adiposity ratings in sample xii

14 LIST OF TABLES TABLE PAGE 4.1 Caliper test-retest technical errors of measurement Intra-observer correlation coefficients for one observer Intra-observer error scores for one observer a. d - means and standard deviations b. Technical errors c. Coefficients of variation Anova: Intra-observer error scores for 1 observer a. Anova (instrument x site) b. Tukey's HSD test for significant effects Ancova: Intra-observer error scores for one observer(instrument x site) Intra-observer correlation coefficients for three observers Intra-observer errors for three observers a. d - means and standard deviations b. Technical error Anova: Intra-observer error scores (Observer x Instrument x Site) Inter-observer correlapion coefficients Inter-observer errors a. d - means and standard deviations b. Technical errors c. Coefficients of variation Anova: Inter-observer error scores a. Anova (instrument x site) b. Tukey's HSD test for significant effects Ancova: Inter-observer error scores (instrument x site) Compressibility of SAT at six sites xiii

15 a. Compressibility (U - C/2) b. Percent compressibility (comp/u x 100) Anova: Compressibilty (U - C/2) a. Anova (sex x Site) b. Tukey's HSD test for significant effects Allometric analysis: compressibiity changes with tissue thickness (U) a. Changes in Compressibility (U - C/2) with tissue thickness (U) b. Changes in Percent compressibility (~ornp/u x 100) with tissue thickness (u) Ancova: Compressibilty (U - C/2) a. Ancova (Sex x Site) b. Tukey's HSD test for significant effects Caliper and ultrasound norm parameters for individual variables a yr old males b yr old males c yr old females d yr old females Caliper and ultrasound norm parameters for the size adjusted sum of six SAT thicknesses xiv

16 PREFACE Body composition measurements are commonly used to evaluate physique status as well as to monitor changes with respect to growth, ageing, exercise, and nutrition. A greater demand and need for reliable, valid, accurate, practical and informative methods of assessing body composition in recent years has been concomitant with our society's increased concern and awareness of health care in terms of both nutrition and exercise. In reviewing the literature, Lohman (1981) reported over 40 papers advocating the use of some 100 anthropometric equations to determine percent body fat. The problems associated with such a method of composition analysis can be divided into three areas: (1) Assumptions inherent in the use of densitometry as a validation criterion: Limitations of the 2 compartmental model of the body. / In particular, variation in the density of the fat free mass as a result of: a) differing proportions of the constituent tissues in the non-fat mass, and b) varying densities of each of its constituents. (2) Imprecision and inaccuracies inherent in the use of calipers to measure skinfold thickness. (3) s he assumptions and the various steps necessary to derive total body fat from skinfold thickness.

17 The 0-Scale system (ROSS et al., 1985) utilises the size-adjusted sum of skinfolds obtained at six sites in comparison to age and sex specific norms as an index of adiposity. Such a method avoids the assumptions inherent in converting skinfolds to percent fat. However, two problems persist in using the skinfold thicknesses themselves. They are: (1) Variation in compressibility of skinfolds. (2) Difficulty in obtaining accurate caliper readings when measuring obese subjects. Ultrasonic techniques permit direct measurement of SAT in an uncompressed state and therefore help eliminate the inaccuracy inherent in caiiper skinfold measurement due to variation in the compressibility of the tissue. Moreover, ultrasound is not limited by the thickness of the SAT layer. As such, the problems associated with the use of calipers to measure obese subjects are avoided.., Ultrasound technology hps been used previously in both - investigation and in clinical practice as a method for determining SAT thickness. However, the lack of an overall methodology and appropriate norms has limited its application. The purpose of this study was to develop an iltrasonic assessment system for both research and clinical application. Chapter 1 reviews the literature pertinent to body composition analysis. Chapters 2 and 3 discuss the general procedures and xvi

18 instrumentation used during data collection. Analysis was completed in the following areas: Chapter 4: Reliability of ultrasound and caliper techniques. Chapter 5: Objectivity of ultrasound and caliper techniques. Chapter 6: Specificity of site and sex SAT compressibility. Chapter 7: Development of norms. Chapter 8: Development of a computerised analysis and display system. xvi i

19 1. CHAPTER 1: LITERATURE REVIEW 1.1 HEIGHT-WEIGHT RATIOS Various indices of weight relative to height are used to assess the extent of obesity (Billewicz et al., 1962; Khosla and Lowe, 1967; Keys et al., 1972; Babu and Chuttani, 1979; Florey,1970). Such methods may have some applicability in epidemiological study for assessing group characteristics; however, they can not be used with impunity for individual assessment due to individual variation in body composition. For example, according to a height-weight table an athlete may be classified as 'overweight' when in fact much of his excess weight will be due to deposition of lean d ss (Behnke et al., 1942). Welham and Behnke (1942) studied a group of football players and demonstrated that 17 of their 25 subjects would be considered as not physically qualified for military duty on the basis of being 'overweight'. However the fat content, as assessed by other methods, was very low for all subjects. The extra weight of a heavy individual may be due to the better development of bone or of muscle, both of which are denser than

20 adipose tissue (Florey, 1970). The limitations and inaccuracies of these readily used methods demonstrate the need for more definitive techniques that differentiate between overweight and overfat. 1.2 DENSITOMETRY AND OTHER LABORATORY METHODS OF BODY COMPOSITION ANALYSIS Density has been used as a quantitative index of obesity since the 1940's. Behnke et al. (1942) reasoned that since the density of fat is low in contrast to other body-tissues, there is a relationship between whole body density and the amount of fat in an individual: low values of whole body density indicating obesity and high values suggesting low percent body fat. This is illustrated practically by the fact that fat people float whereas lean people tend to sink. Wilmore (1983) refers to the densitometric method as the "Gold Standard" since it is used by many investigators as a validity criterion for other methods of body composition assessment. However, Ross et a!. (1985) critisised the densitometric method since it is dependent on the perception of the body as consisting of two compartments, each of known and fixed density, and requires assumptions which have never been validated. While whole body density can be measured

21 accurately, the prediction of percent body fat from whole body density involves some questionable assumptions The measurement of density Body density can be determined by applying Archimedes' principle, from the weight of the subject in air and the loss of weight when the subject is completely submerged in water (~ehnke et a!., 1942). Alternatively it may be calculated from the weight and the volume of the body, the volume being measured by displacement of water (Krzywicki and Chinn, 1967) or of air (Gnaedinger et al., 1963), or by helium dilution of air in a chamber of known volume occupied by the subject (Gnaedinger et al., 1963; Siri, 1961). Using such methods and correct techniques, density can be determined accurately. However, many of the assumptions implicit in converting density to percent fat are not tenable The conversion of density to percent fat

22 In ordek to predict percent fat from density it is necessary to conceive of the body as having two compartments, fat and non-fat, each of known density. Behnke and his colleagues (1942) were the first to partition the human body into a "fat free mass1' (FFM) of fixed density (1.10C g/ml) and a "fat mass" (FM). The fat mass is assumed to have a density of g/ml (Fidanza et al., 1953). Other researchers (Siri, 1956; Brozek et al., 1963) have also used the densitometric method to predict percent fat. Different researchers have assigned slightly different values to the densities of the two compartments but have all used the same principle to predict total body fat. According to the two-compartment model, the \ percentage of fat within the body can be determined from whole body density, provided the densities of.the two compartments are known and fixed. The issue, obviously, is whether or not this is true. J Variation in the density of the fat compartment \ The density of human fat is reported to be g/ml (~idanza et al., 1953). Although there appears to be some variation in the density of lipid throughout the body (~llen et al., 1959), inter-subject variation in the density of the fat compartment is negligible and of little concern in comparison to

23 , variation in the density of the fat free compartment Variation in the density of the fat free compartment Originally, Rathbun and Pace (1945) estimated the density of human FFM to be g/ml. This estimation was based on guinea pig tissue. More recently, evidence has suggested that there is considerable variation in the density of FFM arti in, 1984). FFM fails to distinguish between bone, muscle, and various viscera. Therefore, two conditions must prevail if there is no variation in the density of the FFM: (a) The proportion of each of the components (muscle, skin, viscera, bone, and fluids) making up the FFM must be constant; and, I i I (b) The densities of each of thehe components must also be '\ constant. There is evidence that neither of these conditions is satisfied. Martin (19841, in his study on 25 cadavers, revealed that the proportion of bone as a percentage of total adipose-tissue-free-mass varied from 16.3 percent to 25.7

24 percent, while that of muscle varied from 41.9 to 59.4 percent., The densities of the individual compartments, particularly bone, also vary. Martin (1984) reported variation in density from bone to bone as well as within bones. This cadaver evidenc / suggests that a range in bone density of at least 1.15 g/ml to: 1.34 g/ml is likely. Jones and Corlett (1980) suggested that variation in density of bone is due primarily to differing degrees of bone mineralisation. Mazess et al. (1984) assessed bone mineralisation using dual photon absorptiometry and reported systematic differences for seven areas of the body of 13 young male and 24 female adults with coefficients of variation ranging from 7 to 17 percent. Werdein and Kyle (1960) estimated fat-free densities from simultaneous measurement of bone density and body water. They reported fat-free densities of g/ml for an osteoporotic subject and g/ml for an osteosclero-tic subject and state "the case of either abnormally high or low bone density and abnormal hydration are merely extremes of a broad spectrum seen in normal subjects". I Variation in the density odffm can have a very significant effect on the prediction of percent fat. Very low levels of percent fat have been cited in the literature (~ollock et al., 1977; Michael and Katch, 1968), these levels being so extreme as to question the validity of the densitometric methods used to determine them. Reports of negative percent fat (Adams et al.,

25 1982) leave us without any alternative but to seriously question the assumptions involved in the two compartment-densitometric method for determining percent fat. Inter-subject variation in the density of the FFM explains these otherwise impossible findings. For example, if the density of the FFM is greater than the assumed value of g/ml then a negative percent fat will be predicted for a very lean individual. Apart from this specific case, under-estimation of percent fat due to a FFM density of greater than g/ml and over-estimation due to a FFM density of less than g/ml will go unnoticed. Martin (1984) estimated the range of fat free density to be at least g/ml. This corresponds to a 20 percent error in the percent fat prediction of an obese subject with a whole body density of g/ml. Other laboratory methods for determining body compostion include isotope dilution (~heng and Huggins, 1979), potassium-40 counting (Burkinshaw and Cotes, 1973)~ radiographic analysis (~arn and Gorman, 1956), computerized axial tomography (~orkan et a!., 1982a), total body electrical conductivity (Garrow, 1982), and infra-red interactpnce (Conway et al., 1984). J Although such techniques are relatively precise, they are all validated against the indicted densitometry criterion. Often the equipment is expensive and not readily portable and the techniques employed are not always suitable for all subjects. In addition, the time required for such.techniques limits their I I

26 application in screening large population groups. 1.3 ANTHROPOMETRIC TECHNIQUES Due to the impracticality of laboratory methods for screening large segments of the population, various indirect field techniques have been developed for this purpose. In the following two sections the use of caliper and ultrasonic techniques will be discussed Caliper measurement of skinfold thickness The most widely used indirect method of body composition assessment is measurement of skinfold thickness by means of a caliper. A fold of skin and subcutaneousadipose tissue (SAT) is picked up with the thumb and index finger of one hand. The jaws of the caliper are pladd across the resulting double layer of tissue, and the thickness of the fold is measured. Various calipers have been developed for this purpose. These include the Harpenden caliper (~dwards et al., 1955; Fletcher, 1962; Tanner and whitehouse, 19551, the Lange caliper

27 (Rombeau e t (Donoghue, 1984). al., 1977), and more recently, the Slimguide caliper In trained hands the caliper technique provides a fairly accurate means of measuring skinfold thickness (Tannner and Whitehouse, 1955). The caliper technique has many advantages in the field situation: it is relatively painless, non-invasive and involves no elaborate electronic technology. The calipers are portable, inexpensive, self-contained and reliable when in trained hands. However, the use of skinfold caliper measurements to predict percent fat has been challenged and will be discussed in section Ultrasonic measurement of subcutaneous adipose tissue (SAT) thickness measurement Ultrasound has been proposed as an alternative to caliper as a method of measuring SAT thickness. The development of portable scanners has idcreased the potential of the ultrasonic method (Booth e t a1., 1966; Bullen e t a1., 1965; Haymes e t a1., 1976; Whittingham, 1962). 4 Ultrasound is defined as a sound frequency greater than ( '\ 20,000 Hz. When using ultrasound to measure SAT thickness an, 9

28 ultrasound pulse is generated by a transducer. Tkie sound wave travels through the skin and SAT and is reflected off the SAT-muscle interface; its returning echo is received by the transducer. The time interval between which the pulse leaves the transducer and its echo returns to the transducer is determined.) If the velocity of the sound wave through the tissue is known, a depth reading can be calculated. Ultrasonic echosound has been employed for a number of years in agriculture as a means of evaluating the composition of livestock (Stouffer et al., 1961; Stouffer, 1969; Wallace et al., 1977). Success in the agricultural field initiated interest in the application of ultrasovnd to human assessment. Ultrasonic measurements have shown great potential when compared to other methods such as skinfolds, needle puncture and radiography (~00th et a1., 1966; Bullen et a1., 1965; Haymes e t a1., 1976). Other studies which have examined the feasibility of using A scan ultrasound to measure adipose tissue in humans include Hawes et al., (1972); Sloan (19671, Whittingham (1962) and Glein et al. (1979).

29 1.4 PREDICTION OF PERCENT FAT FROM ANTHROPOWTRY Traditionally, regression equations have been derived to calculate body density and, hence, percent fat from anthropometric measures. Brozek and Keys (1951) were the first to use the relationship between skinfold thickness and body density for assessing fat content. Pascale et al. (1956) produced an equation relating skinfolds to percent fat in ~ritain. Durnin and Rahaman (1967)~ also using ~ritish subjects, correlated four skinfolds with body density. Numerous other anthropometric equations have been developed to predict densitometrically determined total body fat (young et al., 1962; Young et al., 1963; Chinn and Allen, 1960; Edwards and Whyte, 1962; Haisman, 1970; Sloan, 1967; Katch and Mcardle, 1970; Katch and Michael, 1968; Sloan et al., 1962; Wilmore and Behnke, 1969; Jackson and Pollock, 1985; Durnin and Womersley, 1974; Newman, 1955) More recently, ultrasonic measures of SAT thickness have been used to predict densitometr2c percent fat (Glein e t al., 1979; Borkan et al., 7982b; Sloan, 1967).

30 r 1.5 LIMITATIONS OF THE PREDICTION OF PERCENT FAT Densitometric validation criterion Lohman (1981) reports that there have been more than 40 studies since 1950 which have produced over 100 equations to predict body fat from anthropometric variables. When developing regression equations relating anthropometric measures to total body fat, percent fat is determined densitometrically. skinfold equations are therefore in error to the extent of, or more so than, the densitometric prediction of percent fat (see section 1.2) Precision of measurement Seperate from the assumptions implicit within the densitometric criterion, the prpdiction of total body fat from skinfolds is initially dependent on the precise measurement of skinfold thickness. A high level of consistency should be present in a series of measures obtained by one observer as well as in measures obtained by different observers. However, as discussed by Ross et al., (1972) random and systematic error

31 needs to be rigorously controlled. Possible sources of intraand inter-observer error in anthropometry are displayed in Figure 1.1. Those errors particularly pertinent to caliper measurement of skinfold thickness are: (1) Imprecise landmarks. (2) Varying techniques: a) Orientation of the skinfold b) Variation of the amount of adipose tissue picked up / within the fold. c) Position of caliper placement on the fold. (4) Reading errors. (5) Recording errors. Intra- and inter-observer error is particularly a problem when measuring the obese. Reliable and interpretable measurements are difficult to obtain in the obese (Booth et al., 1966; Himes et al., 1979; Garrow, 1982; Brozek and Kinsey, 1960; Haymes et al., 1976; Burkinshaw et al., 1973; Durnin e t al., I 1971). Indeed, the maximum opening of the Harpenden caliper's jaws is too small to accept some skinfolds (Booth et al., 1966). d /' Considerable training and practice are both required before I accurate measurements are possible. Burkinshaw et al. (1973) observed that untrained observers had mean readings 2mm higher than an experienced technician. This is most likely attributable to imprecise location of the measurement site, since repeated \

32 Figure 1.1. Sources of intra- and inter-observer error in anthropometry (adapted from Leahy, 1983)...

33 I3 I OLOG I CAL VARI AT1 ON MEASUREMENT EFFECTS 'U - 1 SENSORY LIMITATIONS OTHER RANDOM EFFECTS SYSTEMATIC ERRORS Calibration Flawed technique Inprecise landmarks Inexperience Subject compliance ILLEGITIMATE ERROR BLUNDERS Reading errors Recording errors

34 skinfold measures at marked sites by inexperienced observers were shown to be reasonably accurate, in comparison to those made by a trained anthropometrist (Burkinshaw et al., 1973). Womersley and Durnin (1973) report that an experienced observer produced significantly higher values at the triceps and suprailiac marked sites than less experienced observers. However, when measuring females at the biceps and subscapular sites the experienced observer obtained significantly lower readings than one of the less experienced observers. ) Assumptions in converting skinfold thicknesses to total body fat The prediction of total body fat from skinfold thicknesses requires a series of assumptions which are summarised in Figure 1.2. Durnin and Womersley (1974) reported that many of these assumptions do not hold true. They suggest that the proportion of body fat situated subcutaneously, skinfold compressibility, I i the density of the skeleton aid the composition of the FFM all '1 vary with age and sex. More recently, cadaver evidence suggests that there is inter-subject variation in skinfold compressibility, skin thickness, adipose-tissue patterning, the fraction of lipid within the adipose tissue, and the ratio of internal to external adiposity artin in et al., 1985). Of these i I I

35

36 STEPS FROM CALIPER TO 80-DY FAT ASSUMPTIONS t THICKNESS OF A COMPRESSED DOUBLE LAYER OF SKIN'PLUS SUBCUTANEOUS ADIPOSE TISSUE, n THICKNESS OF UNCOMPRESSED DOUBLE LAYER OF SKlN PLUS i [ THICKNESS OF SUBCUTANEOUS 1 I ADIPOSE LAYER WITHOUT SKlN MASS OF TOTAL SUBCUTANEOUS ADIPOSE TISSUE 2. SKIN THICKNESS NEGLlGlBL E OR A CONSTANT fractlon OF SKINFOLD 3. FIXED ADIPOSE TISSUE PATTERNING 4. CONSTANT FAT FRACTION IN ADIPOSE TISSUE 1 TOTAL MASS 1-5. FIXED PROPORTION OF INTERNAL TO EXTERNAL FAT

37 factors, variation in compressibility appears to be the area of most concern. Compressibility can be considered in terms of dynamic and static compressibility. Dynamic compressibility: As the jaws of the caliper provide pressure, the fold becomes compressed and the caliper reading decreases with time. Various techniques have been used to standardise this effect. While some wait for all needle movement to cease before taking the reading (Booth et al., 1966), others have suggested readings should be taken two seconds after applying the pressure (Weiner and Lourie, 1969). Combining such schools of thought Ross and Marfell-Jones (1982) suggest that the reading be made "approximately two seconds after application, when the needle slows". 1 I Static compressibility: In addition to the'dynamic action of the skinfold caliper which may be controlled in part by technique, it should be appreciated that similar thicknesses of skin and underlying adipose tissue may yield different caliper readings. Variation in the compressibility of the tissue may result from variation 1

38 differing results (Edwards et al., 1955; Brozek and Kinsey, - --* in instrument design or from varying tissue properties. I The pressure exerted by the jaws of the the caliper differs from caliper to caliper. Since the force per surface area Lp,. (g/mm2) will be the effective pressure, calipers with similar spring tensions but differing jaw sizes will also provide "x; 1960; Leger e t a1., 1982). The caliper may not always provide a constant force over a constant area of contact throughout its range of opening. This defect is reduced to a minimum in the design of the Harpenden (Edwards et al., 1955) and the Slimguide (~onoghue, 1984) calipers. Independent of instrument design, there is reported within the literature a large degree of intra- and inter-subject variation in skinfold compressibility due to varying tissue properties. Garn and Gorman (1956) compared radiographic and Harpenden caliper measures of adiposity at the level of the lowest rib in the mid-axillary line. When comparing radiographic \ and caliper values ~t is necessary to double the radiographic value since the caliper measures a double layer of tissue. For his sample of 150 boys the caliper values of the skinfold thickness average 70 percent of their true uncompressed,double thickness. Brozek and Mori (1958) report that.the average biceps

39 skinfold value for 52 men was 82.3 percent of the double roentgenogrammetric width (magnification corrected) of the adipose tissue at the same site. Comparison of this study with that of Garn and Gorman's (1956) suggests that either compressibility decreases with age or that compressibility varies at different sites throughout the body. In a study of age changes in skinfold compressibility Brozek and Kinsey (1960) measured men between the ages of 20 and 69. The results suggest that compressibilty decreases with age and that significant site differences are apparent. Maximum compression was found at the calf site and minimum compression at the subscapular site. They suggested that the observed differences were due in part to changes in elastic properties of the skin and adipose tissue, and to different degrees of tissue hydration. Mean tissue thickness was also found to decrease with age. Therefore, it is possible that the decrease in compressibility with age could be explained purely as a function of decreasing tissue thickness. 1 Other researchers have compared skinfold values with ultrasonic equivalents. Bullen et al. (1965) determined the j median value of skinfolds at the abdominal site to be 66 percent of the uncompressed value for both men (n=51) and women (n=49); whereas at the triceps site, the compressed 61 percent of the ultrasonic value for men, skinfold value was and 67 percent for

40 women. Although no test of statistical significance was reported, this suggests that at least in some sites there might be a sex difference in compressibility. Clegg and Kent (1967) concluded that female adipose tissue is more compressible than that of males. However, the increased compressibility could be directly related to the increased subcutaneous tissue thickness of the more obese females. Indeed, Jones (1970) accounted for variation in tissue thickness and reported that adipose tissue is more compressible in males than in females for a given tissue thickness. Ward (1979) also allowed for the effect of tissue, thickness on skinfold compressiblity and reported no sex difference within his sample group. Himes et al. (1979) reported no difference between male and female compressibility for a given tissue thickness, while noting a trend toward higher compressibility in males. In a sample of 124 white men, skinfold values at the triceps, biceps, subscapular, waist, suprailiac, thigh and calf sites ranged from 60 to 90 percent of the doubled ultrasonic measurement. Thigh, waist and triceps sites demonstrated the least amount of variation in percent compression (Fanelli and Kuezmarski, 1984). Haymes yt al. (1976) reported half the median skinfold value to be 85, 88, and 85 percent of the ultrasound measure at the abdominal, suprailiac, and subscapular sites respectively, while the median value for half the triceps skinfold was 106 percent of the ultrasonic measure.

41 In a comparative study, Voltz and Ostrove (1984) reported that the compressibility of skinfolds was significantly different at all of the seven sites used with the exception of the triceps and the subscapular. The halved skinfold value /' i ranged from 87 percent (suprailiac) to 141 percent (thigh) of \ I / the ultrasound equivalent. The percent value at the thigh suggests that half the skinfold thickness, which is a double layer of compressed adipose tissue and skin, was greater than the single layer of uncompressed adipose tissue as measured by ultrasound. This is impossible, since calipers by nature of, their design, always compress the tissue to some extent. Therefore, these results suggest that the difficulty in obtaining skinfolds at specific sites, especially at the thigh, result in inaccurate measurement. They also, probably, reflect a site difference in compressibility. Cadaver evidence (Martin, 1984) suggests that there is no significant sex difference in compressibility. However there is considerable site variation, the front thigh and the medial calf showing the lowest compressibility (33.6 and 34.4 percent) and the supraspinale and the biceps showing the highest (64.9 and 63.8 percent). The mean cornpre sibility over all sites for eachb- 5 j of the thirteen cadavers ranged from 38.2 to 69.3 percent. This suggests considerable inter-subject variation. This is further illustrated by data from two of the cadavers, These cadavers had very similar adiposity levels of 27.1 percent and 27.8 percent, j

42 yet the value for one of the cadaver's sum of seven skinfolds 1 / was 97 percent higher than the others. When adipose tissue, thickness was measured directly by incision, the difference \ between the two cadaver's adipose tissue thickness sums was only,, I 19 percent. Therefore the discrepancy between total adiposity \ I I l and skinfold thickness is largely due to differences in skinfold \ compressibiity. The preceding evidence emphasises that individuals do not always conform to the assumptions necessary to predict total body fat from skinfold thicknesses. This explains why skinfold equations are sample specific. There is substantial evidence that such equations are only effective when applied to samples similar to the population from which they were derived (~alina, 1979; Wilmore and Behnke, 1969; Haisman, 1970). Cumming and --3 \ Rebar (1984) reported that different equations applied to the 1 same sample of subjects result in significantly different estimates of body fat. Age, ethnic affiliation, sex, economic status, and health must be closely matched as well as the anthropometric techni,que used.

43 1.6 THE O-SC'ALE SYSTEM: AN ALTERNATIVE TO PERCENT FAT PREDICTION Johnston (1982) stated: "At present it seems that human biologists are better off to continue to use anthropometry itself, rather than to attempt to make estimates of whole body composition from available equations. Even if such equations could provide usable estimates of mean parameters for samples, it seems clear that they are not sufficiently reliable for individual prediction." The development of the 0-Scale system (ROSS e t ae., 1985) echoes Johnston's statement. Rather than converting skinfold thickness to percent fat, Ross and his colleagues believe the skinfold values themselves should be used as a measure of obesity. Since adipose tissue is not uniformly distributed throughout the body (~dwards, 1951; Borkan and Norris, 1977; Martin, 1984) the 0-Scale System samples skinfold thicknesses at all major adipose tissue storage regions. The six skinfold sites used are: triceps, subscapular, supraspinale, abdominal, front thigh and ) medial calf. The sum of these six skinfolds is then scaled to what it would be if the subject was the standard height of 1 ' Y cm, since geometrically a taller person will have thicker1 / I ' skinfolds purely as a dunction of his size, not as a result of, / increased adiposity. Depending on how the value of the height-corrected sum of six skinfolds compares to the same age and sex sample distribution, an individual is rated for I

44 adiposity using a standard nine (staninel scale. 1.7 THE ULTRASOUND ALTERNATIVE Two problems persist in using the skinfolds themselves as an index of adiposity: (1) Variation in compressibility of skinfold thickness, and (2) Difficulty in obtaining accurate caliper readings when measuring obese subjects. Ultrasound provides an attractive alternative to caliper skinfold measurement and has the following advantages: (1) It is non-invasive, involves no radiation exposure and1 I i is safe and painless to use (Whittingham, 1962; Devey and Wells, 1978; Donald and Brown, 1961; Williams, 1983; American Institute of Ultrasound, 1984). (2) Access to sites where skinfolds are unobtainable is possible (~lein et,a1., 1979). i (3) Ultrasound measures the uncompressed adipose tissue t thickness. Deformation of tissue is minimal in comparison to skinfold techniques. Varying elastic properties of skin I

45 and adipose tissue do not become a confounding factor (Booth et a!., 1966). Uncompressed tissue thickness, as opposed to the compressed thickness measured by calipers, is a better indicator of total body adipose tissue mass (Martin, 1984). Ultrasonically measured SAT thickness is highly related to total body insulation (~ayward and Keatinge, 1981). A preliminary study by Burke (1985) revealed that uncompressed SAT thickness is better related to cooling rate than is compressed tissue thickness. A correlation of r=-0.71 was obtained between ultrasonically assessed SAT tissue thickness and cooling rate in the calf region; hereas as, a lower correlation of r=-0.64 was obtained when calipers were used to measure SAT tissue thickness. (4) Development of a light, self-contained, portable unit is evidence of its versatility (Borkan e t al., 1982b). The ultrasound technique is advantageous over the caliper system; however, it is not without problems. These are enumerated below: (1) Multiple echohs have been noted at the abdominal site (Booth et al., 1966; Voltz and Ostrove, 1984). Haymes et al. ('1976) reported similar findings at both the abdominal and suprailiac sites in obese patients. They also reported

46 finding a corresponding additional interface on the roentgenogram, which they postulated to be a fascia near the skin. Two distinct layers of subcutaneous fascia exist in the lower abdomen (Gray 1980). These two distinct layers may extend laterally to the iliac crest in some individuals. Therefore training and subjective judgment is necessary to ensure correct interpretation of multiple echoes in the waist area. (2) Operator errors can be introduced by applying pressure with the transducer and compressing the fat layer (~aymes et a1., 1976; Sanchez and Jacobson, 1975). I (3) The cost of an ultrasound scanner is substantially \ I higher than that of a skinfold caliper. The Echoscan 1502 system used in this thesis costs in excess of $4,000. (4) When ultrasound is used to measure SAT thickness it is 1 I assumed that the velocity of ultrasound through adipose tissue is constant. Although there is some variation in the velocity of ultrasound through adipose tissue it has little effect on the calculated thickness. For example, according to Wells (1977)~he range of reported velocities is 30 m/sec about an average velocity of 1450 m/sec. This corresponds to an error in thickness measurement of +I percent of true thickness. I I I J

47 (5) Although there is no quantitative evidence cited within the literature it is likely that the training required for an observer to become proficient in ultrasonic measurement is greater than that for the observer to become equally proficient in caliper measurement. - (6) The ultrasonic measurement technique is relatively new. Procedures have not been standardised. Knowledge and experience in the method is limited to a relatively small number of experimenters. 1.8 A COMPARISON OF THE VALIDITY OF ULTRASOUND AND CALIPERS AS MEASURES OF SAT THICKNESS Baumgartner and Jackson (1982) define validity as "the degree to which a test measures what it is supposed to measure1'. The most direct method of measuring SAT thickness is by \ incision and measuredent of the distance between the adipose-muscle tissue fascia and the adipose tissue surfaceav~ee and Ng (1965) measured SAT thickness both directly and with skinfold calipers at nine sites on 43 male and 28 female black I i

48 /- I autopsy subjects. They reported correlation coefficients ranging ' from r=0.61 to 0.92 (mean=0.83) between the direct caliper reading and actual fat thickness. I Similar studies have been used to assess the validity of 7 I ultrasonography, and suggest its superiority over skinfolds as a' valid measure. Sanchez and Jacobson i 1975) used human cadavers to validate ultrasound and reported that at all sites the ultrasound readings were within 2mm of direct measures when the fat layer was greater than licm thick. > I Balta et a)., (1981) report a correlation of r-0.99 between ultrasound values and direct steel ruler measures (N = 13 laparotomy patients). Animal studies reveal correlations between actual carcass fat thickness and ultrasonic measures of r=0.92 for pigs (Stouffer et al., 1961). Earlier studies report lower correlations, ranging from r=0.32 to r=.89, for cattle (Stouffer et a1., However more recently Wallace et al. ( 1 977) obtained higher correlations for sites on cattle varying between r=0.70 to r=0.89. r' Soft tissue radiography has also been used as a direct measure of tissue thickness. Correlation coefficients between skinfold and roentgenographic values at various sites have been

49 reported to range between r=0.47 and r=0.89 (~letcher, 1962; Brozek and Mori, 1958; Garn and Gorman, 1956). Correlations between ultrasound and radiographic values have generally been higher, ranging from r=0.78 to r=0.97 (Haymes et al., 1976; Hawes et a1., 1972; Ward, 1979). Correlations of skinfold and ultrasound values with two other indirect methods, electrical conductivity and needle puncture, also suggest that ultrasound is the more valid of the two methods. Bullen et al. (1965) reported a correlation of, r=0.98 when relating needle puncture to ultrasonic values at the abdominal site. Booth et al. (1966) reported a correlation of r=0.98 (standard error=c0.24mm) between ultrasound and electrical conductivity readings of adominal adipose tissue thickness; whereas, a lower correlation of r=0.81 (standard error=t0.57mm) was reported between caliper and electrical conductivity readings. Weiss and Clark (1985a) reported correlations of r=0.87 for males and r=0.50 for females between caliper and ultrasonic measures of adipose tissue thickness in the calf region. At the biceps and triceps sites the association between skinfolds and sonograms was lower in men (r=0.55 and r.0.63) than in women (r=0.84 and r=0.81) '(~eiss and Clark, l985b). It appears that the degree of association between caliper and ultrasound measures of SAT thickness varies with site and sex and is

50 probably due to differences in compressibility. In conclusion, comparison of both ultrasound and skinfold values with various direct measures suggests that ultrasonic measurements of SAT thickness are more accurate than caliper measurements.

51 2. CHAPTER 2: GENERAL PROCEDURE 2.1 SUBJECTS Two groups of subjects were used for the experimental work, sample 1 and sample 2. Sample 1: A sample of 64 subjects (30 males and 34 females) was selected from the student population at Simon Fraser University and the Vancouver area. All subjects volunteered and were given a physique analysis printout upon completion of the measurements. The age range was restricted to between twenty and thirty years. Subjects were chosen such that the range of adiposity of the selected sample was similar to that of the skinfold data base to be transformed. The subject population was stratified such that tilde adiposity level (size-adjusted sum of six skinfolds) of at least eight subjects of each sex was within each of the following groups: (1) the bottom 23 percent, (2) the middle 54 percent, and (3) the top 23 percent of the present

52 Lifestyle skinfold data base. The size-adjusted sum of six skinfolds ranged from 27.5 mm to mm. Data collected from this sample were used to compare the reliability of the ultrasound and caliper SAT measurement techniques, to evaluate sex and site SAT compressibility and to determine the relationship between caliper and ultrasound values so that caliper norms could be transformed to ultrasound equivalents. Sample 2: Ten subjects volunteered from the student population at Simon Fraser University. The subjects' ages ranged from 20.5 to 35.9 years. The adiposity level (size-adjusted sum of six skinfolds) ranged from 27.5 mm to mm. Data collected from this group were used to compare the objectivity of the ultrasound and caliper measurement techniques MEASUREMENT PROTOCOL STATURE: Stretch stature was obtained using a stadiometer according to the technique described by Ross and Marfell-Jones BODY WEIGHT: Body weight was obtained with subjects in minimal clothing using a beam-type balance.

53 LIMB GIRTHS: Relaxed arm girth (AG) and calf girth (CG) measurements were obtained using a flexible steel tape as described by Ross and Marfell-Jones (1982). ADIPOSITY MEASUREMENTS: Adiposity measurements were replicated 3 times at each site using each instrument. Triplicate measures were taken with each of the following instruments: 1) The Echoscan 1502 ultrasound measuring system, 2) The Harpenden caliper, and 3) the Slimguide caliper Adiposity measurements were taken with each instrument at the following sites: TRICEPS (TR) - the point midway between the acromiale and radiale, on the midline of the back of the arm. SUBSCAPULAR (ss) - the point 1 of the scapular. cm below the inferior angle SUPRASPINALE (SI) - the point at the level of the iliac crest on an imaginary line from the spinale to the anterior axillary border. ABDOMINAL (AB) - the point 5 cm lateral to, and at the level of the omphalion. FRONT THIGH (FT) - the point on the midline of the anterior r' surface of the thigh halfway between the inguinal crease and the superior anterior margin of the patella. MEDIAL CALF (MC) - the point on the midline of the medial

54 surface of the calf at the level of the greatest circumference. The sites are illustrated in figure 2.1. The sites were clearly marked using a dermagraphic pen. Means and standard deviations of the raw data collected from sample 1 are tabulated in Appendix B. Sample 2's means and standard deviations are tabulated in Appendix C.

55

56

57 2. 3 SK INFOLD TECHNIQUE Skinfolds were measured following the protocol suggested by the International Working Group in Kinanthropometry (Ross and Marfell-Jones, 1982). 2.4 ULTRASOUND TECHNIQUE The pulse-echo technique was used to determine SAT thickness (see section 3.2.5). A pilot study reinforced the suggestion by other researchers (~00th el a1., 1966; Haymes et al., 1976; Voltz and Ostrove, 1984; Ward, 1979) that practice and training are required before optimal measures can be obtained using ultrasound. The three major areas of possible error are: (1) Local compression of tissue due to pressure exerted by the transducer. /- (2) The positioning of the transducer. Manipulation of the transducer is necessary so that the sound waves strike the tissue interface at right angles.

58 (3) Interpretation of the returning echoes. In areas of low adiposity, deflections associated with muscle-bone interface reflection were sometimes present. Also, the occasional presence of a secondary interface within the adipose tissue itself produced confusing echoes. During data collection these sources of error were controlled for by adopting the following procedures: (1) After the initial positioning of the transducer and detection of the interface, the pressure exerted on the tissue by the transducer was reduced until acoustic conduction was lost. The measurement was taken immediately prior to the loss of the signal. (2) At all sites except the abdomen and the supraspinale the thickness measurements were taken from the first interface detected. At the abdominal and supraspinale sites the procedure was as follows: a) In lean subjects the procedure was the same as that at other sites. b) In relatively obese subjects where a secondary fascia is likely to be present, the second interface /- was selected if the depth associated with the first interfaceowas unduely small. This discernment was easy to teach to anthropometrists who had an appreciation

59 of. the rangeof measures involved. The investigator practiced until his technique was optimal before data collection proceeded. Calibration of the instrument was checked prior to the testing of each subject (see section 3.2.6). An acoustic gel was applied in order to aid the conduction of the sound wave between the probe and the skin.

60 3. CHAPTER 3: INSTRUMENTATION 3. I SKINFOLD CALIPERS All skinfold measures were taken using two different calipers: the Harpenden and the Slimguide calipers Specifications. The Harpenden caliper (Tanner and whitehouse, 1955): Jaw pressure 10 g/mm2 Area or jaw face 90 mm2 Shape of jaw face. Range of measurement rectangular 0-60 mm Reading to the nearest 0.1 mm

61 The Slimguide caliper (Donoghue, 1984) Jaw pressure Area of jaw face Shape of jaw face Range of measurement Reading to the nearest 10 g/mm2 90 mm2 rectangular 0-85 mm 0.5 mm 3. 2 ULTRASOUND Uncompressed SAT thickness was measured using an ultrasonic scanner. The scanner used in this study was the Echoscan 1502, developed by PAR Scientific Instruments APS, Herluf Trollesvej 8, 5220 Odense, Denmark (see Figure 3.1). No independent literature has been previously cited refering to the validity, reliability or objectivity of this particular instrument Specifications i Range of measurement Reading to the nearest mm 0.1 mm

62 ... Figure The Echoscan 1502 Ultrasound Scanner

63 PA- SCIENTIFIC INBTRUMCNT~,