JOURNAL OF SPORTS SCIENCE & MEDICINE
Department of Physiology, University of Kuopio, Kuopio, Finland
(Online): 01 December 2003
Body composition assessment is an important factor in weight management,
exercise science and clinical health care. Bioelectrical impedance analysis
(BIA) is widely used method for estimating body composition. The purpose
of this study was to evaluate segmental multi-frequency bioimpedance method
(SMFBIA) in body composition assessment with underwater weighing (UWW)
and whole body dual energy x-ray absorptiometry (DXA) in healthy obese
middle-aged male subjects. The measurements were carried out at the UKK
Institute for Health Promotion Research in Tampere, Finland according
to standard procedures of BIA, UWW and DXA. Fifty-eight (n=58) male subjects,
aged 36-53 years, body mass index (BMI) 24.9-40.7, were studied. Of them
forty (n=40) underwent also DXA measurement. Fat mass (FM), fat-percentage
(F%) and fat free mass (FFM) were the primary outcome variables. The mean
whole body FM (±SD) from UWW was 31.5 kg (±7.3). By DXA it was 29.9 kg
(±8.1) and by SMFBIA it was 25.5 kg (±7.6), respectively. The Pearson
correlation coefficients (r) were 0.91 between UWW and SMFBIA, 0.94 between
DXA and SMFBIA and 0.91 between UWW and DXA, respectively. The mean segmental
FFM (±SD) from DXA was 7.7 kg (±1.0) for arms, 41.7 kg (±4.6) for trunk
and 21.9 kg (±2.2) for legs. By SMFBIA, it was 8.5 kg (±0.9), 31.7 kg
(±2.5) and 20.3 kg (±1.6), respectively. Pearson correlation coefficients
were 0.75 for arms, 0.72 for legs and 0.77 for trunk. This study demonstrates
that SMFBIA is usefull method to evaluate fat mass (FM), fat free mass
(FFM) and fat percentage (F%) from whole body. Moreover, SMFBIA is suitable
method for assessing segmental distribution of fat free mass (FFM) compared
to whole body DXA. The results of this study indicate that the SMFBIA
method may be particularly advantageous in large epidemiological studies
as being a simple, rapid and inexpensive method for field use of whole
body and segmental body composition assessment.
Obesity is global (WHO, 2000) and national (Aromaa and Koskinen, 2002) epidemic worldwide. Body composition reflects the balance of physical activity and nutritional habits. Body weight alone can be very misleading. The scale cannot tell the difference between the amount of fat and muscle (Heitmann and Garby, 2002). Ageing people tend to gain fat and lose muscle without an obvious change in their weight (Baumgartner, 2000; Wang et al., 2000). And, even though we need a certain amount of fat in our bodies to insure good health, excess body fat has been found to increase the risk of diseases such as type II diabetes (Smith and Ravussin, 2002), cardiovascular disease (Rashid et al., 2003) and cancer (Calle et al., 2003). On the other hand, too little body fat can also pose a number of health risks, especially for women.
When assessing the overweight or obesity the self-reported measures can mislead the results as well. The results from self-reported measures may lead to around 30% misclassification of body composition (Wang et al., 2002). Therefore objective body composition assessments are essential for the whole healthcare field.
Only by accurately analysing body composition, you will learn the amount of fat and lean tissue that makes up your weight, enabling sensible decisions regarding nutrition and exercise programs. Body composition assessment is an advanced method to investigate the status of the human body.
Bioelectrical impedance analysis (BIA) is widely used method for estimating body composition. Wide range of measuring apparata exists for field and clinical use. BIA and its accuracy are dependent on apparatus and valid choice of prediction equation used (Heitmann, 1994; Ellis, 2001).
This thesis describes body composition assessment principles and specific methods. Furthermore, the purpose of this thesis is to compare the method of segmental multi-frequency bioimpedance analysis (SMFBIA) to underwater weighing (UWW) and dual energy x-ray absorptiometry (DXA).
Body composition analysis can be classified according to several models, depending on the purpose or analysis methods and devices available. These include mainly atomic model, molecular model, cellular model and tissue-system model and of course the whole body model as well (Figure 1, Table 1) (Pietrobelli et al., 2001). Most of these models are based on the constant relationship between different components inside the model. Body composition models with most relevant measurement methods are presented at the end of this chapter (Table 2).
The classical two-component (2-C) model assessing of fat and fat-free mass (compartments) by measuring body density (Siri, 1956; Brozek, 1966) is probably the most studied and used reference method for newer body composition assessment methods.
To reduce the limitations in 2-C model it was logical to expand to three-component (3-C) configuration. Typical approach is to include total body water (TBW) to 2-C model, usually with dilution method. Unfortunately, this does not improve the method too much. To the subjects with unstable protein and/or mineral condition the estimated values for solids compartment will be incorrect. That leads to inaccurate body fat estimation as well (Ellis, 2000).
Natural expand to four-component (4-C) model is to add protein or mineral measurement to 3-C model. In 4-C model the body is divided into fat, protein, fluid, and mineral (Figure 3.). These factors are important in clinical use and are helpful in providing nutritional information (e.g. BCM, soft lean mass). The 4-C model can provide more thorough information and efficient classification in health care field.
The six-component (6-C) model is probably the latest and most challenging model of all. It consists from TBW, N, Ca, K, Na, and Cl, where TBW is measured by dilution method and remainder by NAA (Wang et al., 1998; Wang et al., 2002).
The UWW methods were developed mainly as a means to measure body volume to assess body fatness (FM and/or F%). Even if the body weight and volume could be measured without error, there would still be considerable uncertainty regarding the individual's body fatness estimate due to normal variations in body hydration, protein, and mineral content. It has been estimated that the total error for body fatness is ~3-4% of body weight for individuals (Heymsfield et al., 1989). Thus, it has been recommended that without correction for variation in the water and mineral content of FFM, densitometry should not be used as a reference method for heterogenic population.
Dual energy x-ray absorptiometry
When an X-ray or photon source is placed on one side of an object, the intensity of the beam on the opposite side of the object is related to its thickness, density, and chemical composition. This attenuation phenomenon is also dependent on the energy of the incident photons and is dominated by two principles at low energies: the photoelectric effect and Compton scattering. The attenuation response is nonlinear, such that for a homogeneous material, it can be described by the exponential equation. If the absorber is composed of two or more materials, then the composite is the weight sum of the individual mass attenuation coefficients, each weighted for its fractional contribution to the total mass (Heymsfield et al., 1997).
The attenuation through bone, lean tissue, and fat is different, reflecting their differences in densities and chemical composition. With increasing photon energy, the differences in the attenuation properties for these tissues decrease. Thus, if the relative intensity of the transmitted beam can be measured, and the mass attenuation coefficients are accurately known, estimates of the bone mass and overlaying soft tissue mass can be calculated. This 2-C model is also used when the beam passes through body regions without bone. In this case, the appropriate attenuation coefficients are those for fat and lean tissues, respectively (Heymsfield et al., 1997).
It should be evident that DXA is composed of two separate sets of equations, each used to describe a 2-C model. Dual-energy X-ray absorptiometry does not provide three independent measurements, even though three body composition values [bone mineral content (BMC), lean tissue mass (LTM), and fat mass (FM)] are reported (Figure 4.). To accomplish this, the manufacturers must assume that the composition of the soft tissue layer overlaying bone has the same fat-to-lean ratio as that for non-bone pixels in the same scan region. In the case of the whole body scan, ~40-45% of the pixels are classified as containing bone. The remaining pixels are used to estimate the body's fat-to-lean ratio; this value is applied to the soft tissue component in the adjacent bone pixels. Thus, the relative lean-to-fat composition of the total soft tissue mass is based on sampling only one-half of the whole body (Mazess et al., 1990).
Some investigators have expressed their concern about measurement bias related to the impact of a significant change in the hydration of the lean tissues. The latter concern, however, has recently been theoretically shown not to significantly alter the estimates for the bone, lean, or fat mass (Pietrobelli et al., 1998b).
Many reconstruction algorithms have been demonstrated to produce a cross-sectional image of the body region. This basic anatomical image contains information of the tissue density at each pixel. This information together with the anatomical location of the pixel in the image can be used to identify it as adipose, muscle, skin, visceral, or bone tissue. Reconstruction of total body mass and separate organ masses based on scans along the length of the body at 10-cm intervals has been shown to have excellent accuracy (<1% error) and precision (<1%). These reconstructed CT images can be assigned to level 4 or tissue systems level of the multi-component model. The CT images can also be used to separate the total adipose tissue mass into its subcutaneous and visceral components, or the lean tissues into skeletal muscle and visceral or organ mass. Likewise, bone can be identified as cortical or trabecular in nature on the basis of density (Baumgartner et al., 1988; Rossner et al., 1990; Ross et al., 1991).
The major disadvantage with CT is the radiation dose required per slice for scanning. However, if the pixel resolution required for routine CT scans can be reduced, then the dose can be significantly reduced. Recent studies have shown that a dose can be reduced to 1/25 that of a routine clinical CT scan dose (Baumgartner et al., 1988; Rossner et al., 1990; Ross et al., 1991; Goodpaster et al., 2000).
Magnetic resonance imaging
The frequency at which nuclei for each element will flip is called the Larmor frequency. When radiofrequency (RF) energy, at the Larmor frequency, is applied perpendicular to the direction of the magnetic field, the nuclei will absorb this energy and change its alignment. When the RF field is turned off, the nuclei will lose their alignment and release the stored energy. The intensity of this signal can be used to measure the number of hydrogen nuclei of the tissue. This process can be repeated at each position along the length of body until the whole body is mapped and cross-sectional images at each slice can be generated. Magnetic resonance imaging (MRI) is successful because hydrogen, found mainly in water, is one of the most abundant non-bound elements in the body. For other elements, their concentrations in the body are lower and the Larmor frequency changes, thus requiring an increased magnetic field strength if imaging is to be considered (Abate et al., 1994; Mitsiopoulos et al., 1998).
If the hydrogen densities of adipose and lean tissues were markedly different, then it would be possible to develop images based solely on their number of nuclei. To see the contrast between lean and fat tissues, a second feature of the nuclei, called relaxation time (T1), is used. This is the time it takes for the nuclei to release the RF-induced energy and return to a random configuration. The T1 for protons in fat is much shorter than that for protons in water. This contrast can be maximized by adjusting the time interval of the RF pulse and the time to detect the induced signal. The total process is often referred to as a pulse sequence or spin-echo sequence (Ross et al., 1991; Baumgartner et al., 1992; Ross et al., 1992; Abate et al., 1994; Mitsiopoulos et al., 1998; Ross et al., 2000).
Neutron activation analysis
Bioelectric impedance analysis
For the BIA measurement, a low alternating current is conducted through the outer pair of electrodes, while the inner pair of electrodes from which the impedance is measured measure the drop of voltage. To convert this information to a volume estimate, two basic assumptions are used. Firstly, the body can be modelled as a cylindrical conductor with its length proportional to the subject's height (Ht). Secondly, the reactance component (X) contributing to the body's impedance (Z) is small, such that the resistance component (R) can be considered equivalent to body impedance. When these two assumptions are combined, it can be shown that the conducting volume is proportional to the term Ht2/R, called the impedance index. In the classical BIA method, it was hypothesized that a human body is a cylinder like the figure below (Figure 5), and BIA measures total body water (TBW). The amount of water is equal to the volume of the cylinder. With BIA, the cell membrane and fat tissue show high impedance with electric current. Electrolytes dissolved in the water of lean body tissue provide good conductivity.
Since BIA is estimating the water content of the body compartments, the stable state of tissue hydration is crucial for the reliable BIA measurement. Furthermore, BIA principle implies that the meal and the weight of food in intestines (Fogelholm et al., 1993; Slinde and Rossander-Hulthen, 2001), clothing, skin temperature, and blood flow (Caton et al., 1988; Liang and Norris, 1993; Gudivaka et al., 1996; Liang et al., 2000), and so forth affects the accuracy of measurement.
From total body water, body fat mass can be estimated. After age, height and weight are taken into account the body water is estimated. Lean body mass is in proportion to total body water. The estimation is based on a constant relationship of 72.3% water in LBM (Wang et al., 1999). When LBM is subtracted from weight, body fat mass is calculated, 2-C model. Body fat mass is typically calculated with a regression analysis equation directly from the impedance index.
The bioelectrical impedance method is easy to use and can be readily repeated. At present, BIA is probably the most frequently used method, due mainly to the relatively inexpensive cost of the basic instrumentation, its ease of operation, and its portability.
Single-frequency bioimpedance analysis
Multi-frequency bioimpedance analysis
Segmental bioimpedance analysis
Segmental multi-frequency bioimpedance analysis
Air displacement plethysmography
The system consists of two chambers: one for the subject and the other serving as a reference volume (Figure 8). With the subject in one chamber, the door is closed and sealed, the pressure increased slightly, and a diaphragm, separating the two chambers, is oscillated to slightly alter the volumes. The classic relationship of pressure versus volume, at a fixed temperature, is used to solve for the volume of the subject chamber.
An advantage of this technique compared with the UWW measurement is that the subject does not have to be submerged in water; however, all of the technical limitations related to the true volume that were noted for the UWW method remain. These instruments presently are designed for adults and will require significant modifications and improvements if the technique is to become useful for monitoring smaller subjects, including infants (Dempster and Aitkens, 1995; McCrory et al., 1995; McCrory et al., 1998; Levenhagen et al., 1999).
Near-infrared measurement When an object is exposed to infrared light, the object absorbs or reflects the light according to its chemical structural properties. All organic materials (i.e. fat or protein) absorb light in unique portions of the spectrum. Futrex, Inc., MD, USA has developed a device to analyze human body composition by this technique. It estimates body composition by analyzing the spectrum of near infrared light reflected from the skin and underlying tissue (Conway et al., 1984; Elia et al., 1990; McLean and Skinner, 1992).
The Futrex device emits near infrared light at very precise frequencies (938 nm and 948 nm) into body-frequencies at which a body fat absorbs the light and the lean body mass reflect the light. In essence, it is measuring how much light is emitted from the light wand and how much light is reflected back into the light wand. This measurement provides an estimation of the distribution between the body fat and lean body mass (Conway et al., 1984; Elia et al., 1990; McLean and Skinner, 1992).
Infrared measurements from the biceps have been the primary site shown to correlate best with a criterion method However, it overestimates the fat of a thin person and underestimates the fat in an obese person compared to underwater weighing, skinfold measurement or BIA (Conway et al., 1984; Elia et al., 1990; McLean and Skinner, 1992).
Circumferences and other measurement methods
Sagittal abdominal diameter
Body mass index
The objective of this study was to evaluate segmental multi-frequency bioimpedance method (SMFBIA) in body composition assessment. The aim was to cross-validate SMFBIA against under water weighing (UWW) and whole body dual energy x-ray absorptiometry (DXA).
1. The primary outcome was to compare fat mass (FM), fat free mass (FFM)
and fat percentage (F%) from all three methods.
The measurements were carried out at the UKK Institute for Health Promotion Research in Tampere, Finland between March 27th - June 15th, 2000 according to standard procedures of SMFBIA, UWW and DXA.
In addition to the measurement of body composition using bioimpedance analysis, UWW and whole body DXA the measurements performed recording of age, weight, height and waist-hip ratio (WHR) as well.
The study was approved by an independent Research Ethical Committee of UKK Institute. Written informed consent was obtained from the participants.
The subject's body density was assessed by underwater weighing, after full exhalation (presumably at residual lung volume) in a sitting position submerged to the chest. Eight to ten submerge trials was performed. (Fogelholm et al., 1996; Fogelholm et al., 1997).
Body composition was calculated from the body density by a two-component model in which the body was divided into fat and fat-free compartments with assumed densities of 0.9 and 1.1 g·cm-3, respectively (Siri, 1956).
Dual energy x-ray absorptiometry
In addition to whole body data, the analysis provides data on regional BMC, LBM, and FM from head, trunk, abdomen, arms (both), and legs (both). The segmental FFM was calculated by a sum of regional BMC and LBM.
Segmental multi-frequency bioimpedance analysis
The measurement takes about two minutes time, where after the device prints the result sheet (Appendix 1) through a standard personal computer printer connected to the InBody 3.0 measurement device.
InBody 3.0 device report gives total body FFM, FM, and F% values calculated from impedance values, equation reported earlier (Cha et al., 1995). The segmental FFM was calculated from segmental fluid distribution with assumption of constant body water content of FFM equals 0.732 L per kg (Wang et al., 1999).
SMFBIA measurements were carried out according to general recommendations. The measurements were performed after 12-hour fasting and within 30 minutes of voiding the urinary bladder. No physical exercise was allowed before 4 hours of the measurement (NIH, 1996a; 1996b).
Total body composition
The Pearson correlation coefficients between UWW, DXA and SMFBIA from FM, FFM and F% are presented in Table 8. All the correlations are statistically significant (p <0.01). Furthermore, the correlation regressions are plotted from F% in Figure 9, FM in Figure 10 and FFM in Figure 11.
Bland-Altman analysis was calculated as a mean difference against average values with +2SD for F%, FM, and FFM, from UWW, DXA and, SMFBIA, respectively. The values are presented in Table 9 and plotted in Figures 9, 10 and, 11.
Segmental body composition
Pearson correlation coefficient between DXA and SMFBIA of FFM in different
body segments are presented in Table
11. All the correlations are statistically significant (p <0.01).
Furthermore, the correlation regressions from different body segments
are plotted in Figure 9.
This thesis describes body composition assessment principles and methods. One important decision when selecting the method is to understand the model behind it. The whole body model includes total body weight, height and other anthropometric measures. Probably the most used body composition assessment model in clinical and scientific use is the molecular model. It is valuable in most cases when assessing fat and fat free ratios against total body weight. In research and clinical diagnostic purposes, one can benefit about a data from atomic or cellular models. In addition, in clinical use the functional model can be beneficial. If high accuracy of body composition estimates in individuals is essential, multi-component models are recommended to be used (Fogelholm and van Marken Lichtenbelt, 1997).
In our study, we evaluated SMFBIA in 2-C model and estimated FM, F% and FFM compared to UWW and DXA. The study sample represented a sample of 40 overweighed male subjects. The correlations between methods were high in all cases. The highest correlations were found between SMFBIA and DXA. Similar results have been demonstrated earlier, for example by Bracco et al. reported a comparison between BIA and DXA with a same trend in the results (Bracco et al., 1996). The known fact of overestimation FFM by BIA existed also in our results. That is the primary reason for, despite the high correlation between the methods, a relatively poor result from Bland-Altman analysis. This is especially seen in comparison between SMFBIA and DXA.
The second objective in our study was to compare segmental distribution of FFM between SMFBIA and DXA. Similar studies have been performed for adults (Bracco et al., 1996) and children (Fuller et al., 2002), before. In segmental body composition assessment by SMFBIA, we faced similar fears as described earlier about problems with segmental assessment. By accepting these facts, we made our analysis with device manufacturer based equations and software. Our fears became true and we found a nice correlation and an acceptable range of difference in Bland-Altman analysis for leg and arm sections of the body. The correlation for trunk area was also high, but a big difference was seen in Bland-Altman analysis. Moreover, the Bland-Altman plot shows a remarkable bias in the ends of the measurement range. In the other words, subjects with low or high FFM are showing bigger error in analysis. This is probably a reason for low impedance at the trunk area from SMFBIA that cannot recognise differences accurately at this body segment.
However, the trend is good enough to take advance of the benefit what a segmental analysis can provide, especially in upper and lower extremities, e.g. for assessment of possible side differences in clinical conditions. In addition, the feasibility of the measurement procedure may enhance the use of SMFBIA in epidemiological studies.
BIA is a widely used method for estimating body composition. Wide range of measuring apparatus exists for field and clinical use. BIA and its accuracy are dependent on apparatus and valid choice of prediction equation used (Heitmann, 1994; Ellis, 2001). BIA have been used successfully in large epidemiological studies for example in United States (Chumlea et al., 2002), Switzerland (Pichard et al., 2000; Kyle et al., 2001) and Denmark (Heitmann, 1991). Segmental and multi-frequency features for BIA have been developed for increasing accuracy and more individual assessment of the body composition. Multi-frequency capability of the BIA measurement has definitely brought some advance in accuracy of the assessment (Ellis et al., 1999a). Furthermore, differentiation of ICF and ECF has been possible along with multi-frequency measurement (Gudivaka et al., 1999). Segmental measurement feature on the other hand seems to be interesting but challenging task according to the accuracy. Of course, individual differences in body shape and tissue distribution give a good reason for developing segmental procedure. However, the method is hard to validate against other methods because of different body segment differentiation between measuring devices. Besides, exact body segment differentiation by SMFBIA is probably the most difficult. In BIA, electrical current pathway is against lowest resistance available in human body. Furthermore, by knowing, that SMFBIA gives highest impedance values from extremities and lowest from the trunk area while the volumes in these body parts are opposite, which makes the valid assessment difficult.
Whatever body composition model or method is used the knowledge about them is a must. Furthermore, understanding the confounding factors, sources of error and other possible biases inherent in these methods is crucial for valid body composition estimate. Hydration state of the subject seems to play an important role in most of the methods. MRI is based on the position of the hydrogen atoms. DXA is also sensitive of the hydration of FFM and BIA is primarily measuring total body water content or ratio of ICF and ECF, where hydration state of the subject can mislead the analysis remarkably (Montagnani et al., 1998; Pietrobelli et al., 1998b).
Very lean or obese subjects can cause great error in their body composition estimate. Most of the equations used with these different methods are based on population based studies with subjects close to the average body composition values. Therefore, the error in the body composition assessment seems to grow when lean or obese subjects are in question. Here, also, the different hydration status within these marginal groups may explain some of the bigger error (Fogelholm et al., 1996; Fogelholm and van Marken Lichtenbelt, 1997).
At ageing, people tend to gain fat and lose muscle without an obvious change in their weight. This is also in part originated from the reason of decreasing hydration in the muscle tissue in aging (Baumgartner, 2000; Wang et al., 2000). Furthermore, gender differences occur when estimating body composition. The differences in hormonal secretion have impacts in body composition as well (Heitmann, 1991). Besides, changes in hormonal state in females during the menstrual cycle should be taken in the consideration when analysing body constitution (Gleichauf and Roe, 1989). Hence, one has to be aware about these varying situations when performing valid body composition assessment.
One of the major confounding factor in the field of body composition estimation, despite what method is in question, is the wide variety of prediction equations. It is understandable because of individual differences between genders, ages, races, etc. Nevertheless, it makes comparison between different studies and interventions, if not impossible, at least difficult and unreliable. That could be the general task for body composition research in the future. Moreover, body composition assessment in childhood is crucial task in the follow up for breaking the trend of growing obesity (James et al., 2001). Therefore, generally accepted prediction equations at least for children are needed for all the major body composition assessment methods.
Body composition assessment is an important method needed in the field of fighting global epidemic of obesity. Body composition reflects the balance of physical activity (energy consumption) and nutritional habits (energy supply). The scale can be very misleading. Body weight alone cannot tell the difference between the amount of fat and muscle (Heitmann and Garby, 2002). And, even though we need a certain amount of fat in our bodies to insure good health, excess body fat has been found to increase the risk of diseases such as type II diabetes (Smith and Ravussin, 2002), cardiovascular disease (Rashid et al., 2003) and cancer (Calle et al., 2003). Only by accurately analysing body composition, you will find the precise amount of fat and lean tissue that makes up your weight, enabling right decisions regarding nutrition and exercise programs. In addition, when studying the overweight or obesity from self-reported data it can mislead the results dramatically. The results from self-reported measures may lead to around 30% misclassification of body composition (Wang et al., 2002). Therefore accurate and objective body composition assessment is essential for the weight management.
At present population based reference values are not available for any body composition assessment method, only for some anthropometric measures. As a feasible field method, SMFBIA is a potential also for large population based studies. In this situation, there is a need for body composition values and trends for different populations. These should include percentiles of FM, FFM and F% in different age groups and genders. Also, further investigation is needed for finding out the predictive value of segmental body composition and its distribution when assessing the risk of e.g. chronic obesity related disorders; metabolic syndrome, type II diabetes, high blood pressure, cardio vascular diseases and, arthritis.
The purpose of this study was to evaluate segmental multi-frequency bioimpedance method (SMFBIA) in body composition assessment.
The present study confirms that SMFBIA is usefull method to evaluate whole body fat mass (FM), fat percentage (F%) and fat free mass (FFM).
Furthermore, this study shows some new evidence that SMFBIA is suitable method for assessing segmental distribution of FFM from upper and lower extremities compared to whole body DXA. Assessment from trunk area remained uncertain in our study.
Further research is needed for creating segmental population based reference values from SMFBIA. These should include percentiles of FM, FFM and F% in different age groups and genders.
Also further research is required for finding out the predictive value
of segmental body composition for diseases e.g. chronic obesity related
disorders. In addition, to evaluate strength of the value of segmental
body composition assessment alone and in association with general anthropometric
measures when predicting the risk of these pathological conditions.
This study was carried out in the UKK Institute for Health Promotion Research in Tampere, Finland. Therefore I want to thank the people there, Katriina Kukkonen-Harjula, MD, PhD, Harri Sievänen, PhD, Patrik Borg, MSc and Mikael Fogelholm, PhD, for the possibility to for me to participate in their research project.
Especially I want to thank my supervisor's docent Harri Sievänen, ScD and docent Heikki Pekkarinen, MD, PhD for their patience during my work.
Humble thanks belong also to the reviewers of this MSc thesis professor Leo Niskanen, MD, PhD and docent Harri Sievänen, ScD.
I am very thankful to the laboratory staff at the UKK Institute for conducting all the measurements.
This study was partially supported by the companies Biospace Co. Ltd. and Mega Electronics Ltd. Also, special thanks go to Mega Electronics Ltd. and President Arto Remes for supporting the printing of this thesis.
Kuopio, September 2003
Jukka A. Salmi