Energy expenditure

I. Literature Review

A. Introduction of energy expenditure (EE)

Energy is defined as the origin of life, being expended to support all of our metabolic functions such as growth, reproduction, muscular activity, and the biosynthesis of many metabolites (1). Specifically, energy is required for breath, blood circulation, food digestion, and all other types of physical activities. Therefore, total energy expenditure (TEE) is mainly a sum of internally produced heat including the thermic effect of food (TEF) and resting metabolic rate (RMR) which is also called basal metabolic rate (BMR), and external work-induced caloric expenditure called activity energy expenditure (AEE) (Figure 1).

The RMR is the largest single element of total daily caloric expenditure (2), accounting for 60-70% of TEE (3). RMR is the amount of energy produced while the digestive system is inactive and the whole body is at rest, staying in a neutrally temperate environment. RMR decreases with age and the loss of lean body mass, but it elevates with increased muscle mass (4). RMR was able to be measured in clinic trials by gas analysis through either direct or indirect calorimetry since the late 19th and early 20th centuries (5), although predictive equations for RMR had been adopted as the major method (2,3,5). Along with RMR, TEF acts as an increment to TEE due to the cost of digesting and absorbing food for use and/or storage (6). It normally accounts for 10% of TEE (7). Moreover, AEE, providing 15-30% of TEE (3), is produced while undertaking all kinds of postures as well as spontaneous and voluntary physical activities with various intensities; therefore, AEE is considered to contribute the most to the variation of TEE (12). Internally, AEE can be divided into two major domains: one is the energy expended while undertaking structured and planned physical activities such as sports and workout in gyms; and the other one is EE induced by activities with relative lower intensity such as walking, housework and even fidgeting. In fact, both domains are quite influential to the total amount of AEE, although sport-like exercise is conventionally promoted to elevate TEE (8). Non-exercise activity thermogenesis (NEAT), although individually contributing little to the total AEE, tends to be appreciable when it is considered accumulatively because non-exercise activities take place throughout the day. However, energy expended for higher-intensity activities may just count for a small proportion of daily AEE because the duration of exercising for a normal person occupies only a small fraction of the entire day (8). Therefore, individuals with different lifestyles or habits could have significant variance in their AEE, especially in NEAT. Dr. Levine of the Mayo Clinic, for instance, has estimated that NEAT could vary as much as 2000 kcal per day interpersonally (9), which further substantially demonstrated the crucial role of AEE in weight maintenance.

B.Applications of energy expenditure assessment

B.1. Assessment of energy requirements

In 1985, energy requirements were recommended to be based on assessments of EE rather than energy intake, according to the report of the joint Food and Agriculture Organization/World Health Organization/United Nations University (FAO/WHO/ UNU) Expert Committee on Energy Requirements (10). Great effort was made to validate the existing recommendations for energy requirements for different target groups. Black et al., for instance, compiled 574 studies on TEE measurements in 1996 and demonstrated that the existing recommendations for energy intake for infants and children in U.S. were high, whereas those for adolescents and adults were usually low (15). Consistently, Davis (1998) reviewed TEE estimations for infants using doubly labeled water (DLW) method (n>400) (16), Bratteby et al. (1998) studied TEE in 46 adolescents (18), and Carpenter et al. (1998) published a TEE study of 164 subjects aged 65-70 y (19), whose results all supported those of Black et al.. Meanwhile, DLW method has been used to assess energy requirements for healthy individuals under abnormal or fascinating situations, such as 11 sailing men in Branth et al.'s study (27), 18 men in Van Etten et al.' resistance training program (28), and 7 soccer players in Ebine et al.'s investigation (30). Furthermore, energy requirements for unhealthy populations can be estimated on the basis of their EE. For example, infants with cyanotic congenital heart disease (31) or broncopulmonary dysplasia (34) were documented to have low survival rate because it was investigated that they tended to have increases in TEE of 30 and 15%, respectively. Therefore, these enhance the risk of under-nutrition and require special attention to balance with energy intake in these infants.

B.2.Validation of energy intake measurements

The balance between energy intake and EE is the key to our health, so both sides of balance are extensively studied. Self-reported energy intake was commonly employed to estimate TEE before the development of accurate techniques (43). Nowadays, modern technique-derived TEE, combined with adjustment due to changes in body weight and composition, has often been used to predict energy intake. Dietary food intake, unfortunately, was usually documented to be underreporting (35), which was a troublesome problem in studying the relationship between diet and health; hence, many studies have focused on characterizing, understanding, and minimizing the underreporting. Commonly, self-reported dietary food intake was unreliable and underreported, compared with DLW-generated TEE (43). For instance, Champagne et al. found that children (n=118) tended to underreport their daily energy intake by 17 to 33% of TEE measured by DLW (46); and Martin et al. demonstrated that nonobese, middle aged women underreported their energy intake by 20% of TEE (47). In the progress of understanding the underreporting, Johnson et al. found that the underreporting of energy intake by 24-hour dietary recall method strongly correlated with being overweight or fat, or having poor reading, spelling and understanding skills (48). Therefore, big consideration has to been taken when self-reporting energy intake is involved in the study.

B.3. Obesity research in the context of energy imbalance

Accurate and precise measurement of EE for healthy and unhealthy individuals has many useful clinical applications. One of the most popular uses is in the prevention and treatment of obesity and obesity-related chronic diseases since obesity is a rapidly growing epidemic worldwide and the related diseases dramatically increase the risk of morbidity and mortality (11,17). In Western countries, overweight and obesity had been widespread since 1978 (29). It was estimated that between 1980 and 2002 obese patients in U.S. doubled in adults aged 20 years or older, and overweight children and adolescents aged 6 to 19 years tripled (29) (Figure 2&3). Similarly, Canadian obesity prevalence in adults increased from 6% (1985) to16% (2003) as well (32). Meanwhile, the chronic diseases associated with obesity, such as hypertension, type II diabetes and cardiovascular disease, are increasing so quickly that related costs have exceeded or at least equalized those resulting from smoking and drinking (33). Considering one of the leanest populations, China is fast in catching up with the West. Recent estimates in the China Health and Nutrition Survey indicated that the population of overweight adults increased by nearly 40% and that of obesity doubled from 1992 to 2002, approximately 215 million Chinese affected (36) (Figure 4). The growth in population who has or will have obesity-related chronic diseases is even out of evaluation. As a result, both obese patients and physicians are trying all possible strategies to achieve weight loss, including reduction of energy intake by intervention of diets, drugs, and bariatric surgery, and elevation of energy consumption by exercise and non-exercise promotion (17), aiming to improve health conditions.

RMR and TEF are fairly constant within and between persons (12) and do not correlate with obesity significantly, even though it is widely believed that genetic variance in RMR and TEF may contribute to weight gain susceptibility (14). Hence, enhancement of AEE plays the major role in making TEE overweigh energy intake. Despite the promotion of moderate to vigorous activities to prevent and treat overweight and obesity (13), the effectiveness of exercise alone on weight loss appears to be modest unless it is combined with diet (17). Many prospective and observational studies have demonstrated that continuing regular vigorous physical activities over a long time period without diet change (>1 year) might only reach weight maintenance in both children and adults (20,21), but weight loss is not guaranteed.

One the other hand, the importance of NEAT assessment to obesity treatment study is beginning to receive more attention. Theoretically, spending 2 min/hour on watching television or playing computer rather than walking may result in 5 kg of weight gain in 10 years (22). Clinically, the effect of increasing NEAT on weight loss has been established in several trails. Robinson et al. (1996-1997), in an randomized controlled school-based trial (n=192), analyzed the questionnaires reported by students and their parents, finding that children who reduced television, videotape, or video game use did have significant decrease in BMI, waist circumference and waist-to-hip ratio compared with controls (23). Consistently, some other evidence indicated that reducing sedentary activities such as screen-based entertainment and use of motorized transport is a promising, population-based approach to prevent obesity (23,26). A subsequent study measured EE with various postures and movements, demonstrating that compared with lying down position, EE increased while sitting motionless by 4%, while seat-fidgeting by 54%, while stand-fidgeting by 94%, and while walking at 1.6, 3.2 and 4.8 km/h by 154%, 202% and 292%, respectively (24). Therefore, the thermogenic potential of fidgeting-like or low-intensive activities is great to substantively contribute to EE side of energy balance. Levine et al. (2005), for instance, recruited 20 non-exercising subjects (1/2 lean and 1/2 obese) to study the impact of postures on NEAT inter-individually. They exhibited that the thin subjects stood at least 2 hrs/day longer than the obese, which was translated into an EE difference of 350 kcal/day (9). Indeed, accumulative evidence suggests that enhancing daily NEAT substantially contributes to the increment of AEE, sequentially to TEE, opening a new window for patients and physicians to treat obesity practically.

Furthermore, the morbidity and mortality associated with obesity-related chronic diseases such as cardiovascular disease, type II diabetes, and some cancers (11,17) may be controlled to some extent if morbidly obese patients succeed in losing reasonable amount of body weight by increasing their low-intensity activities. Unfortunately, the effects of EE on the reduction of morbidity or mortality related to obesity so far could not be concluded due to inconsistency and uncertainty among existing studies. One of main contributors would be the crude assessment of AEE, with entailing variation in measurement accuracy and precision (12). Given this considerable heterogeneity, a need exists for accurate and precise measurement of AEE to decrease the chance of misclassification or the influence of confounders or other bias, which would further enhance our understanding the nature and strength/weakness of the association between health and physical activity/inactivity within populations, internationally, or cross-culturally.

New developed technologies and instrumentation have been smartly adopted for EE measurements, such as calorimetry, DLW method, heart rate monitors and accelerators. However, no method is suitable for every situation or every population, so in addition to understanding the principles of the available methods, clarifying the similarities versus differences or advantages versus disadvantages is necessary for a proper operation of research.

C.Methods for energy expenditure measurement

C.1.Direct versus indirect calorimetry

Direct calorimetry measures total heat dissipated by evaporation, radiation, conduction and convection from the body (37). While this technique is able to measure TEE accurately and precisely as well as AEE, the subject has to be placed in a thermally-isolated chamber. Therefore, it does not exhibit a free-living AEE produced by daily physical activities (54). Besides, only one individual can be monitored at one time, so it is very time-consuming if applied to large populations (12,37). Indirect calorimetry, on the other hand, measures oxygen consumption (VO2) or CO2 production, but subjects have to wear a mouthpiece, hood or a whole body chamber, which also substantially interferes with people's daily activities (37).

C.2.Doubly labeled water (DLW) method

DLW method was invented by Lifson et al. in 1955 (39) and applied to small animals until the early 1980s (38). After long-lasting extensive validation and evaluation studies (40-42), its first application in humans to measure free-living EE was published in 1985 (49). An oral dose of DLW contains deuterium (2H) and 18O which are non-radioactive and both occur naturally in humans. It is deemed safe because the typically given doses of the two stable isotopes fall well below the levels able to cause side-effects, damage or toxicity (43,44). To date, DLW method is known as the gold standard for TEE measurements in free-living individuals (45) and is utilized in a variety of human groups, including premature infants, children, adolescents, adult, obese people, pregnant and lactating women, the elderly and hospitalized patients. The applications of DLW include assessing energy requirements, validating measurements of dietary intake or physical activity through other available methods, and understanding effects of dietary and/or physical activity intervention on health (53).

The isotopes, 2H and 18O, mix with body water within a few hours. As generating energy, CO2 and water are produced and eliminated from the body in different manners. The principle of DLW method is based on the general concept that 18O is eliminated as both CO2 and H2O, whereas 2H is excreted solely as H2O (55); hence, the elimination rate of 18O is higher than that of 2H, and the difference between the two rates can be further used to estimate CO2 output. Based on the same physiological principle of indirect calorimetry, CO2 production can be used to predict TEE. However, different from calorimetry, DLW method allows subjects to perform their daily living activities because it relies on collection of urine and saliva samples rather than heat or respiratory gases. Briefly speaking of the procedure, a specific dose of DLW is given to an individual based on weight; after a few hours, saliva samples are collected for total body water calculations. Post-dose urine samples are collected at different time points for later measurements of isotope enrichments that remain in the body. 2H and 18O enrichments in all samples are measured against Standard Mean Ocean Water (SMOW) international standard by Thermal Conversion-Elemental Analyzer/ Isotope Ratio Mass Spectrometry (TC-EA/IRMS). The length of the study period varies from 4-21 days depending on the intensity of activities of study groups (53). To avoid imprecise results relative to analytic errors, the study should be long enough for sufficient elimination of isotopes but short enough so that enrichments of the final urine samples exceed pre-dose isotopic abundance sufficiently (56).

Several assumptions were made with which DLW method becomes valid (57,58): 1) the volume of water pool for labeled isotope dilution is constant; 2) the flux rates of water and CO2 are constant; 3) deuterium is washed out only as water whereas 18O is lost both as water and as CO2; 4) enrichments of water and CO2 exiting the body are the same as those remain in body water; and finally 5) no isotopic exchange between inside body and environmental water and CO2 through skin or lungs. Therefore, estimations of TEE may get inaccurate or even wrong when applied to persons with disorders that can alter total body water pool or fluxes of water and CO2, such as kidney and respiratory disorders. Any environmental factors that influence ambient temperatures, water turnover rates, or background isotope levels will also lower the precision of DLW method (61).

Generally, DLW method is an accurate and precise technique for TEE estimations in free-living individuals who engage in various intensity levels of physical activities (51). The typical advantage of this technique is its noninvasiveness which is least likely to interfere with individual behavior, and subjects can participate freely in their daily activities during the study period (52). Unfortunately, relatively high cost of DLW purchase and high demand for specialized expertise for mass spectrometry instrumentation limit its widespread utilization in large population studies.

C.3. Heart rate monitors/Accelerometers

Heart rate monitors measure TEE based on the linear relationship between heart rate and O2 consumption (VO2). However, they are less accurate and precise for EE measurements of low-intensity activities (59), and they have many influential factors including emotional stress, body temperature and medication (60). With respect to accelerometers, they detect acceleration and deceleration movements electronically with varying degrees of sensitivity (53). However, detectability of accelerometers is restricted to some activities, such as weight lifting, carrying and pushing, which do not involve variation in acceleration (62), so they are not sensitive enough to capture all types of activities in daily life. Even though these two devices are noninvasive and cost-effective to record duration, intensity and frequency of some physical activities, they are only suitable for small to medium sized population studies (65). Problems with compliance and behavior changes are also big issues (12).

C.4. Activity questionnaires (AQs)

Compared to other techniques, questionnaires are the easiest to distribute and administer, and they do not require much motivation or time from subjects (12,53). With lower investment on time and money, AQs also provide a lot of information on physical activity/inactivity as well as other factors about a great number of study participants. Hence, it is of clear advantage for large population-based and/or epidemiological studies. There are a range of AQs designed to capture various activity parameters, most of which were developed after the Minnesota Leisure Time Physical Activity Questionnaire published in 1978 (74). The standard format of a questionnaire consists of diverse activities which are generally categorized by the intensity, including very hard-, hard-, moderate-, and light-intensity and their corresponding frequency and duration questions (12). Normally, AEE calculation process follows: 1) each activity is first assigned a metabolic equivalent value (MET) based on the intensity; 2) total AEE is then computed by multiplying the reported hours (frequency Ã- duration) spent in each activity by the corresponding MET; 3) and the weight factor will be considered finally to estimate AEE of an individual during the study period (Figure 5).

Initially, the most considered was to utilize an appropriate questionnaire/recall method according to the purpose and the target group of the study. Therefore, numerous AQs were developed with emphasis on different study groups, investigating different aspects of activity at different time periods (12,50,53). As inactivity being a globally growing health concern, more questionnaires have extended their questions asking about not only structured and planned exercise, but also occupational and leisure-time activities (63-65,74). Nowadays, the issue of AQ validity has been raised (66) since significant heterogeneity exists between AQ-derived AEE and AEE derived by other accurate techniques. A lot of validation and comparison studies boosted to explore the reliability, validity and sensitivity of existing AQs, showing that although the developed AQs try to capture various activity parameters, they still exhibit large limitation (12,50,53). A good review paper of Neilson et al. (50), for instance, evaluated >20 literatures about validating various AQs against DLW. Giving a summary of study designs and other factors, they appraised each AQ's validity using mean difference and correlation coefficients (TEEAQ versus TEEDLW or AEEAQ versus AEEDLW). In summary, 23 AQs covering both genders, all levels of age, obese and non-obese people, and international measurements were extracted and validated by DLW method, concluding that only 2 out of 23 reviewed AQs, the Tecumseh Occupational Activity and past month Minnesota Leisure Time Questionnaire (63) and the Tecumseh Community Health Survey (67), showed acceptable criterion validity with mean differences of 10% and 2%, and correlation coefficients of 0.62 and 0.63, respectively. The vast majority of AQs did not have sufficient face validity for EE estimation, which might be attributable to 1) ignoring of non-exercise activities, 2) different time periods covered by AQs and DLW, 3) potential errors in assignment of MET to self-reported activities, 4) relatively small sample size, and 5) the mode of AQ administration (62) and respondent characteristics (68).

With respect to possible influencing factors, a lack of consideration of NEAT has been extensively discussed. In theory, low-intensity activities accumulatively occupy majority of our time each day, so TEE or AEE yielded from them could contribute to large variance in EE-derived from AQs (50). According to Westerterp KR (70), a profound underestimate of AEE could be generated if AQs and/or participants do not recall low-intensity activities such as personal care, climbing stairs, walking, transportation or sedentary activities. Likewise, various postures during sedentary activities have been indicated to contribute differently to AEE which could cause the discrepancy between AEEAQ and AEEDLW (25). By and large the limitation of AQs precludes us from drawing firm and reliable conclusions from research which involves EE as an influential parameter. However, as pointed out by Sesso (71), self-reported methods of data collection will still remain the primary way in which activity must be quantified in large-population and/or epidemiological studies. Therefore, the development of a comprehensive questionnaire which can more accurately and consistently measure and describe subtle components of lifestyle activities associated with day-to-day living is emergent, especially for understanding the essential role of physical activity/inactivity in chronic disease.

II. Rationale

DLW method has been recognized as the gold standard to measure free-living TEE at individual level (12). Not only does DLW accurately estimate individual TEE with minimal behavioral changes, it also allows subjects to live normally and freely even during the experimental period. If AEE is interested, it can be yielded by subtracting both RMR and TEF from TEE (Figure 1). Specifically, RMR can be calculated using mathematical prediction equations, four of which are commonly used in clinical studies (Harris-Benedict, Owen, Mifflin-St Jeor, and World Health Organization/Food and Agriculture Organization/United Nations University [WHO/FAO/UNU]) (2). TEF is usually considered to account for 10% of TEE (7). However, high cost of DLW and high technical demands make it infeasible and impractical in population-based health research unless the subject group is small (i.e. <200) (72). Likewise, other available techniques are also impractical in large health studies because of cost, burden on subjects and investigators, invasiveness, and/or potential behavioral changes. Hence, self-administered questionnaires are so far the best choice to assess AEE in large population studies (71).

Our primary objective is to validate a newly developed self-administrated questionnaire, referred to as the past month Sedentary Time and Activity Reporting Questionnaire (STAR-Q), which was designed for the purpose of estimating free-living individual AEE in large populations. By using DLW method, the accuracy and precision with which STAR-Q estimates free-living AEE will be evaluated, assisting further modification and development of STAR-Q. Secondarily, we will test three available predictive equations for RMR calculation, Mifflin-St Jeor (3), WHO/FAO/UNU (10) and Harris-Benedict (73) against DLW method to review and document individual accuracy of equation utilization in healthy, middle-aged, nonobese adults. Moreover, if applicable (n≈100), we will create a new algorithm for RMR estimation unless the existing RMR equation is good enough.

Hence, the null hypothesis is that the mean differences and correlation coefficients of AEEAQ versus AEEDLW will be within the acceptable range of <10% and >0.6, respectively, and so does comparison of RMR derived from each predictive equation and DLW.

III. Sedentary Time and Activity Reporting Questionnaire

(STAR-Q)

STAR-Q is the crystal developed with collaborative effort of University of Calgary and the Alberta Cancer Board. Dr. Ilona Csizmadi is the Nominated Principal Investigator of the Calgary study, working with some investigators of the Alberta Cancer Board's Tomorrow Project®(76). Therefore, my study is a co-investigation of validity of STAR-Q by using DLW as a reference method. During the preliminary development of STAR-Q, the initial draft was designed based on the data already collected from approximately 18,000 participants in Tomorrow Project® who have completed an open-ended Past-Year Total Physical Activity Questionnaire. Differentially, STAR-Q was particularly developed to estimate daily AEE, so other domains of activity that represent day to day living were included and measured in detail. The reference period was decided to cover only past month rather than past year because it has been indicated by cognitive tests that recent events tend to be recalled more accurately and precisely than distant past events (77). To obtain information about how people answer the survey questions and to identify difficulties encountered by respondents when retrieving and reporting activities engaged during past month, two rounds of one-on-one cognitive interviews were carried out between trained, professional and experienced research interviewer and 20 participants. The collected data was used to modify the questionnaire design and re-word the text. Following revision of STAR-Q after cognitive interviews, pilot testing was conducted to re-evaluate the ease of the questionnaire and to test various memory aids for helping participants recall details pertaining to all kinds of activities.

The ultimate START-Q was organized into 15 sections, trying to capture as many activities as possible throughout a 24-hour period during the past 4 weeks (28 days). They are 1) directions of questionnaire completion, 2) sleeping and napping, 3) eating, 4) personal and medical care, 5) occupation and volunteer work, 6) transportation and moving about, 7) household, 8) year work, 9) caregiving, 10) exercise, sports and leisure, 11) light leisure and relaxing, 12) other activities, 13) using the stairs, 14) some final questions asking about mobility level, and lastly 15) comments. In the questionnaire, it ascertains the types of activity and the corresponding frequency, duration, and intensity (4 physical effort levels). Different from many questionnaires, STAR-Q subdivided physical inactivity into many domains, such as occupation, transportation, household, night leisure and relaxing. Moreover, many daily routine activities are considered, such as watching TV, playing computers, reading, and studying; besides, accompanying postures (e.g. reading while lying down, sitting, or standing) are differentiated as well, which are great advances compared with other available AQs. It is anticipated that with list of activities shown in the questionnaire, it could efficiently aid subjects to report most of participated activities during the last month.

IV. Research Design- Procedures and Methods

A. Subjects and recruitment

Adults living in the greater Calgary area (includes farming, ranching and rural area) will be eligible to write informed consent if they are aged between 30 and 60 years and with a BMI of <35. Additionally, men and women will be considered eligible if they: 1) do not have metabolic disorders, any diagnosed cancers or disorders of the kidney, liver, cardiovascular, respiratory, or neurological systems; 2) are not taking medications or natural health products that could potentially influence energy or water balance; 3) do not have a weight fluctuation of >2.5 kg in the previous two years (78) and are not planning any weight gain or loss in the near future; 4) are not breast-feeding, pregnant or planning to be pregnant within the next year and; 6) will not be away from Calgary area for >3 months at a time once the study starts. Subjects will be requested to complete baseline and DLW completion questionnaires at the start and end of the study respectively to confirm eligibility criteria and to assess health status, medications, and intention to travel as well as other factors.

The recruitment mostly relies on the Tomorrow Project® cohort , in which approximately 3000 participants are eligible. The desired sample size for our study is 100 subjects. However, we have only achieved half so far and the recruitment is still continuing.

B. Procedures and Methods

B.1. Doubly labeled water (DLW) method

DLW method in my study is used to estimate individual free-living TEE per day. The DLW protocol will be conducted over a 14-day period (Figure 6). All subjects will be requested to fast overnight on Day 0. Prior to DLW dosing, baseline urine (5ml minimum) and saliva samples (3ml minimum) will be collected to determine the background isotope enrichments, and measurements of height, weight, and waist circumference will be taken. Then each subject will be given a single dose of DLW, consisting of 2.5g 10 atom percent of 18O per kg body weight and 0.18g 99 atom percent of deuterium per kg body weight, followed by a 100 ml rinse of fruit juice. Then enriched saliva samples will be collected at 3 h and 4 h for determining total body water (TBW), and post-dose urines (2nd void of the day) will be collected separately on Day1, 8 and 14. Like Day 0, each subject after fasting will come back on Day 14 and take a dose of D2O (0.18g 99 atom percent of deuterium per kg body weight), followed by 100 ml of rinse. Post-dose saliva samples will be collected at 3 h and 4 h. The Physiology Specialist will aliquot urine and saliva samples for storage in a -20°C freezer until shipment to the Richardson Centre for Functional Foods and Nutraceuticals (RCFFN) in Winnipeg for 2H and 18O enrichment analysis. 2H and 18O enrichments of each sample will be measured five times using automated Thermal Conversion-Elemental Analyzer in combination with Isotope Ratio Mass Spectrometry (TC-EA/IRMS: Finnigan DELTAplus V). Only the reasonable values without memory effect will be picked for mean calculation. Then using the two-point approach, elimination rates (ko and kh) of the two isotopes will be computed by dividing the changes in enrichment (Ef-Ei) by the time difference (tf-ti). Gathering all data, TBW (mol) is calculated using the formula:

TBW = (d/MW) Ã- (APE/100) Ã- 18.02 ÷ (Rstd Ã- ΔδD) Ã-1000 ÷ 18.02

Where d is the dose of 2H2O in grams, MW is the molecular weight of the 2H2O, APE is the atom percent excess of deuterium, Rstd is the ratio of 2H/H in the standard (1.5576Ã-10-4), and ΔδD is the enrichment change between pre- and post-dose saliva samples.

CO2 production rates can be calculated using the formula:

rCO2 (mol/day) = 0.4554 Ã- TBW (mol) Ã- (1.007ko - 1.041kh)

where ko and kh (day-1) are the elimination rates of corresponding isotopes

After getting rCO2, TEE (MJ/day) can be determined with an assumed respiratory quotient (RQ) of 0.85 (79), using a modified Weir's formula (75):

TEE = 22.4 Ã- [3.9 Ã- (rCO2/FQ) + 1.1 Ã- rCO2] Ã- 4.18/1000

To calculate AEE (MJ/day) which is our target, the following equation will be used:

AEE= [TEE Ã- 0.90] - [(RMR/24) Ã- (24 - hours of sleep)] + [0.9 (RMR/24) Ã- hours of sleep]

Where TEF is considered to be 10% of TEE (7), and RMR will be determined using Mifflin-St. Jeor equation (3) which has been evaluated to perform the best for healthy and nonobese individual adults (2). For subjects who have gained or lost weight over the course of the study, adjustment will be made to account for weight change, based on the principle that 1.0 kg of body weight is equivalent to 30 MJ of energy (83).

B.2. Activity questionnaires

The past-month STAR-Q will be completed by subjects at the test facility on Day 14 of the DLW study to avoid behavioral changes within the two weeks. The project assistant will provide help if the participant have any problem about the questionnaire while filling. Participants will be requested to complete STAR-Q in 3 and 6 months for evaluating the consistency. Response will be scanned using TELEform®software (TELEform V9.1; Cardiff, USA).

Activities appears on the questionnaires will be assigned codes and metabolic equivalent (MET) values according to the Compendium of Physical Activities (80) before scanning. The reported values of frequency, duration and intensity for each separate activity will be used to calculate total MET-hours. The summation of MET-hours for all reported activities will yield an overall estimate of metabolic output per day to reflect individual AEE (Figure 5). AEE derived from STAR-Q will be directly compared with AEEDLW.

In addition to STAR-Q, participants will be requested to fill a 7-day Activity Diary after DLW study. The dairy is adapted from Conway et al (81) and designed to ascertain all activities and posture every 20 minutes while awake. It was indicated by Levine et al (82) that non-exercise activity thermogenesis gained from a 7-day diary was generally representative of several months of NEAT or even until the lifestyle was changed. Therefore, the captured activities will be used to compare with those obtained from STAR-Q, in terms of activity types, MET-hour/day and duration, and then it will be used to assist the development of SATR-Q.

B.3. Predictive equations for resting metabolic rate (RMR)

Three commonly used predictive equations for RMR estimation, Mifflin-St Jeor (3), WHO/FAO/UNU (10) and Harris-Benedict (73), are picked to test validity against DLW method in healthy, middle-aged, nonobese adults. For each participant, his/her RMR will be calculated following three equations and determine if the values fall into the TEE range that is counted as 60-75% of TEE yielded by DLW method. Finally, the percentages of individuals who fall into the range will be compared. Age factor influencing RMR will be considered. For example, the TEE of adults closer to 60 y is likely to have lower RMR (i.e. ~60% of TEEDLW), but younger adults are likely to have higher RMR (i.e. 70-75% of TEEDLW).

B.4. Data analysis

Means and standard deviation (SD) will be calculated with significance level set at 5%. Average daily AEEs measured using DLW, STAR-Q and 7-day Activity Diary will be calculated for each participant. Bland-Altman plots will be used to evaluate the level of agreement in activity-related AEE between DLW and questionnaires. Spearman correlation coefficients will be calculated to assess the degree of linear association between measurements. Regression analysis will be finally conducted to explore the relationships between the dependent variable (DLW derived AEE) and independent variable (AEE derived from STAR-Q and 7-day Activity Diary, respectively). Potential covariates of interest (age, sex, weight, fat free mass) will be considered and kept in the regression model if they are statistically significant.

V. Anticipated results

Aiming at our primary objective, we expect to a good match in AEEs generated from STAR-Q and DLW method with mean difference and correlation coefficient of <10% and >0.6, respectively. It is also anticipated that the agreement between AEEDLW and AEE yielded from 7-day Activity Diary will not be worse than STAR-Q's because the former activity diary tends to obtain all activities and postures every 20 minutes while awake. In terms of our secondary objective, we expect to see Mifflin-St Jeor equation, in this case, has highest accuracy in calculating RMR for healthy, middle-aged, nonobese adults against RMRDLW compared with Harris-Benedict and WHO/FAO/UNU equations. If our goal is met well, the validated STAR-Q will facilitate much needed studies of AEE and its etiologic roles in disorders of energy imbalance, an epidemic of which is emerging in Canada and worldwide, along with the steady rise in the prevalence of obesity-related disease.