Hostname: page-component-cd9895bd7-8ctnn Total loading time: 0 Render date: 2024-12-22T14:54:54.941Z Has data issue: false hasContentIssue false

Assessment of physical activity using accelerometry, an activity diary, the heart rate method and the Indian Migration Study questionnaire in South Indian adults

Published online by Cambridge University Press:  05 August 2009

Ankalmadagu V Bharathi*
Affiliation:
St. John’s Research Institute, St. John’s National Academy of Health Sciences, Opp. Koramangala BDA complex, Bangalore – 560034, Karnataka, India
Rebecca Kuriyan
Affiliation:
St. John’s Research Institute, St. John’s National Academy of Health Sciences, Opp. Koramangala BDA complex, Bangalore – 560034, Karnataka, India
Anura V Kurpad
Affiliation:
St. John’s Research Institute, St. John’s National Academy of Health Sciences, Opp. Koramangala BDA complex, Bangalore – 560034, Karnataka, India
Tinku Thomas
Affiliation:
St. John’s Research Institute, St. John’s National Academy of Health Sciences, Opp. Koramangala BDA complex, Bangalore – 560034, Karnataka, India
Shah Ebrahim
Affiliation:
Department of Epidemiology and Population Health, London School of Hygiene & Tropical Medicine, London, UK
Sanjay Kinra
Affiliation:
Department of Epidemiology and Population Health, London School of Hygiene & Tropical Medicine, London, UK
Tanica Lyngdoh
Affiliation:
Centre for Chronic Disease Control, Delhi, India
Srinath K Reddy
Affiliation:
Public Health Foundation of India, New Delhi, India Department of Cardiology, All India Institute of Medical Sciences, New Delhi, India
Prabhakaran Dorairaj
Affiliation:
Department of Cardiology, All India Institute of Medical Sciences, New Delhi, India
Mario Vaz
Affiliation:
St. John’s Research Institute, St. John’s National Academy of Health Sciences, Opp. Koramangala BDA complex, Bangalore – 560034, Karnataka, India
*
*Corresponding author: Email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Objective

To validate questionnaire-based physical activity level (PAL) against accelerometry and a 24 h physical activity diary (24 h AD) as reference methods (Protocol 2), after validating these reference methods against the heart rate–oxygen consumption (HRVO2) method (Protocol 1).

Design

Cross-sectional study.

Setting

Two villages in Andhra Pradesh state and Bangalore city, South India.

Subjects

Ninety-four participants (fifty males, forty-four females) for Protocol 2; thirteen males for Protocol 1.

Results

In Protocol 2, mean PAL derived from the questionnaire (1·72 (sd 0·20)) was comparable to that from the 24 h AD (1·78 (sd 0·20)) but significantly higher than the mean PAL derived from accelerometry (1·36 (sd 0·20); P < 0·001). Mean bias of PAL from the questionnaire was larger against the accelerometer (0·36) than against the 24 h AD (−0·06), but with large limits of agreement against both. Correlations of PAL from the questionnaire with that of the accelerometer (r = 0·28; P = 0·01) and the 24 h AD (r = 0·30; P = 0·006) were modest. In Protocol 1, mean PAL from the 24 h AD (1·65 (sd 0·18)) was comparable, while that from the accelerometer (1·51 (sd 0·23)) was significantly lower (P < 0·001), than mean PAL obtained from the HRVO2 method (1·69 (sd 0·21)).

Conclusions

The questionnaire showed acceptable validity with the reference methods in a group with a wide range of physical activity levels. The accelerometer underestimated PAL in comparison with the HRVO2 method.

Type
Research Paper
Copyright
Copyright © The Authors 2009

Most methods that assess physical activity in individuals are not feasible in large epidemiological studies due to their high cost, need for technical expertise or their limited ability to capture habitual activity in free-living conditions(Reference Lamonte and Ainsworth1, Reference Haskell and Kiernan2). Despite certain limitations(Reference Shephard3), physical activity questionnaires are among the most widely used methods to assess physical activity in large epidemiological studies. Most questionnaires assess specific domains, in particular, leisure-time discretionary activity(Reference Richardson, Ainsworth, Wu, Jacobs and Leon4, Reference Helmert, Herman and Shea5). In rural India, leisure-time discretionary activity (exercise and games) may not be a major physical activity domain. Manual labour in rural populations is high in occupational and household activities(Reference Kanade, Rao, Yajnik, Margetts and Fall6). Use of either job description or reported occupation as a measure of overall physical activity has been shown to be inadequate(Reference Vaz and Bharathi7). The frequency and intensity of activities across the various physical activity domains are also likely to vary between men and women(Reference Vaz and Bharathi8). Hence, in India, questionnaires that capture information across multiple domains of physical activity would be ideal. Some published questionnaires do assess multiple domains of physical activity, but list activities that are uncommon to India(Reference Chasen Taber, Erickson, Nasca, Chasan-Taber and Freedson9). Other questionnaires require individuals to determine the intensity of various activities and categorize them into moderate or vigorous intensity(Reference Craig, Marshall, Sjostrom, Bauman, Booth and Ainsworth10). This may be a problem, since we have demonstrated earlier that perception of intensity of an activity is dependent on age and familiarity with the activity(Reference Vaz and Bharathi11).

Previously, we developed a questionnaire specific for an urban Indian community that assessed physical activity across multiple domains(Reference Bharathi, Sandhya and Vaz12). This questionnaire was subsequently modified for the Indian Migration Study (IMS) in order to obtain an estimate of physical activity in both rural and urban communities(Reference Lyngdoh, Kinra and Shlomo13).

The main aim of the present study was to examine the validity of a physical activity questionnaire to assess physical activity level (PAL; the ratio of total energy expenditure to estimated BMR)(Reference James and Schofield14) using accelerometry recordings and a 24 h physical activity diary (24 h AD) as reference. Prior to the study addressing the main aim, we also performed a validation study of the accelerometer and 24 h AD against the heart rate–oxygen consumption (HRVO2) method(Reference Kurpad, Raj, Maruthy and Vaz15) and data from this are also provided.

Methods

Physical activity assessment methods

Physical activity questionnaire

The questionnaire assessed physical activity of the past month across multiple domains including discretionary leisure time, household chores, work, sleep, sedentary activities and other common daily activities. The frequency and average duration for each activity were documented. Frequencies were ascertained using fixed categories of ‘daily’, ‘once a week’, ‘2–4 times a week’, ‘5–6 times a week’, ‘once a month’ and ‘2–3 times a month’. When all reported activities did not cumulatively account for 24 h, a standard MET (metabolic equivalent) of 1·4 was applied to the ‘residual time’, as in previous studies(Reference Bharathi, Sandhya and Vaz12). For manual occupational activity, the integrated energy index (IEI) of the activity was applied instead of the MET value. Unlike MET, IEI accounts for ‘rest’ periods that participants are likely to take when engaged in manual activities(Reference James and Schofield14). PAL cut-offs have been described to classify physical activity patterns into sedentary/light, moderately and vigorously active lifestyles(16).

24 h activity diary

Participants documented their activities in blocks of 10 min. Illiterate participants were included in the study if an educated family member working in close proximity with the participant could document the activities. Participants and family members were instructed to document additional details of activities such as posture (sitting, standing, walking, etc.), walking speed (slow, normal, brisk) and the average weight of loads, where required. A pen attached to a string, worn around the neck, reminded participants to document the 24 h AD regularly. The 24 h AD was scrutinized after completion and the ambiguities clarified directly with the participants.

Accelerometry

Twenty uniaxial accelerometers (model 7164; Manufacturing Technology Incorporated, Shalimar, FL, USA) were used in the study. The accelerometers were initialized for 1 min epochs. Between-instrument variation was assessed by comparing counts after they were fixed in batches (ten in each batch) to a simple barrel mixer and rotated for a period of 15 min at 45 rpm. One accelerometer produced counts out of range in comparison with the other accelerometers and was not used in subsequent studies. The CV of the counts for the remaining nineteen accelerometers was 1 %. The workings of the accelerometer and the computations used to arrive at physical activity measures have been described previously(Reference Bassett, Ainsworth, Swartz, Strath, O’Brien and King17). PAL was calculated as the average of the hourly MET over 24 h obtained using the customized software ActiGraph Software Analysis version 3·2 (ActiGraph LCC, Pensacola, FL, USA).

Heart rate–oxygen consumption method

Oxygen consumption (VO2) was measured using indirect calorimetry (model VMax 29; SensorMedics Corp., Yorba Linda, CA, USA). Heart rate (HR) was measured using a heart-rate monitor (Polar S720; Polar Electro Oy, Kempele, Finland). Details of the HRVO2 protocol measurements have been described elsewhere(Reference Horner, Lampe, Patterson, Neuhouser, Beresford and Prentice18). Briefly, steady-state resting VO2 (4 h after breakfast) was measured for each individual, after which they performed a set of standard activities for 5–6 min which included lying down at rest, sitting quietly, walking at 2·4 and 4·8 km/h on a treadmill (no gradient) and spot jogging of 120 steps/min. Steady state was defined as ‘10 minutes during which the volume of oxygen consumed, ventilatory rate, and respiratory quotient does not vary by greater than 10 %’(Reference Horner, Lampe, Patterson, Neuhouser, Beresford and Prentice18). After a steady state was achieved, the mean value of VO2 over the time period of each task was used to determine the linear relationship between VO2 and HR. This relationship was in turn used to predict the VO2 from the HR(Reference Kurpad, Raj, Maruthy and Vaz15).

As it is possible that there is a breakpoint in the relationship between HR and oxygen consumed per minute (VO2), a critical HR called the FHFLEX was identified below which the RMR was used to represent the metabolic rate(Reference Spurr, Prentice, Murgatroyd, Goldberg, Reina and Christman19). In that study, which compared total daily energy expenditure (TEE) from the HRVO2 method with that from whole-body calorimetry, the closest estimates of TEE were obtained when an arbitrary value of FHFLEX + 10 beats/min was used as the breakpoint in the relationship between HR and VO2. All recorded HR below the breakpoint were assigned a metabolic rate that was equivalent to the measured VO2 at rest, while all HR above the breakpoint were used in an equation relating HR to VO2 for activities equal to and above slow walk, obtained by calibrating these two variables for each individual. For this study, ‘FHFLEX + 10’ was used as the method to obtain TEE using the Weirs equation (TEE = [3·941 + (1·106 × 0·9)] × VO2, where 0·9 is the assumed respiratory quotient and VO2 is volume of oxygen consumed) to subsequently derive PAL(Reference Weirs20).

Protocol of experimental studies

Protocol 1: preliminary study to validate physical activity level derived from accelerometry and the 24 h activity diary against the heart rate–oxygen consumption method

All subjects recruited for the experimental protocols below completed a sociodemographic questionnaire. Height and weight of each participant were recorded.

A convenience sample of thirteen participants who were employees from our academic institute (St. John’s Medical College, Bangalore, India) was recruited. The study was conducted in the metabolic laboratory of the institute. Participants were required to stay overnight in the metabolic ward. This posed a problem in recruiting women as they were unable to participate due to social commitments or restrictions. Hence all participants were males, aged between 19 and 49 years (mean 28 (sd 8) years).

The accelerometer and heart-rate monitor were strapped to the participants on the experimental day. Steady-state VO2 was measured and the protocol for the HRVO2 method was followed as described above. After completion of the standard activities, participants continued with their daily routine activities while continuing to wear the accelerometer and the heart-rate monitor for a period of 24 h. During this period, the participants also maintained the 24 h AD. Participants were instructed to have a sponge bath instead of a regular bath and to contact the study coordinator if the accelerometer or the heart-rate monitor became dislodged from its normal position.

Protocol 2: validation of the questionnaire against accelerometry and the 24 h activity diary

A convenience sample of ninety-four participants (fifty males and forty-four females; forty-five urban and forty-nine rural) were recruited in this protocol. Of the eligible participants initially contacted, 95 % agreed to participate in the study. The urban participants included employees from our academic institute (teaching staff, clerks, attenders, cleaners, etc.) and residents living in nearby urban localities. The rural participants were recruited from two villages in Palamner Taluk in Chittoor district, Andhra Pradesh, about 140 km from Bangalore, and consisted of housewives, agricultural labourers and farmers, among others. Participants unwilling to wear the instrument for the entire 24 h period and those with lower-limb deformities were excluded from the study. The recruitment of urban participants was restricted to weekdays. The rural participants were recruited all through the week as their activities were similar throughout the week.

Participants were administered the physical activity questionnaire, strapped with the accelerometer and were instructed to continue with their daily routine activities and simultaneously maintain the 24 h AD. Eighty-three participants were finally included for analysis as data of eleven participants were excluded due to inadequate documentation of the detailed 24 h AD (n 9) or malfunctioning of the accelerometer (n 2). The mean age of participants in this protocol was 39 (sd 13) years (range 19–61 years).

Ethical approval

The studies were approved by the local institution ethics review board. Written informed consent was obtained from the participants after a detailed explanatory statement of the study was provided.

Statistical analyses

Data for continuous variables are presented as means and standard deviations. The mean PAL values estimated from the accelerometer, questionnaire, HRVO2 method and the 24 h AD were compared using repeated-measures ANOVA and post hoc evaluation using the t test with Bonferroni correction. Pearson’s correlations were used to evaluate the relationship between the PAL values estimated from the physical activity methods. The Bland–Altman method with limits of agreement was used to assess the bias in the mean PAL estimated using the physical activity methods(Reference Bland and Altman21). In Protocol 2, the physical activity methods were administered on weekdays and weekends in the rural group as opposed to only during weekdays in the urban group. The mean bias of PAL estimated between the methods in the two groups was compared using an independent t test. The mean bias between the questionnaire with the detailed 24 h AD and with the accelerometer in the urban and rural groups was not significantly different; hence the urban and the rural data were combined for all analyses. A linear regression was performed to assess if age, gender or BMI predicted the mean bias estimated from the above methods. The validity of the questionnaire in ranking participants into sedentary/light, moderate and vigorously active lifestyles using standard cut-offs was assessed by evaluating the proportion of participants falling into the same and extreme categories when compared with the detailed 24 h AD(16).

The model that best predicted the relationship between the accelerometer counts for specific activities reported in this community with their intensities obtained from published sources was assessed using a linear and a curve fit model. For all tests, the level of significance of two-sided tests was set at the 5 % level. All analyses were performed using the SPSS statistical software package version 13·0 (SPSS Inc., Chicago, IL, USA).

Results

Protocol 1: preliminary study to validate physical activity level derived from accelerometry and the 24 h activity diary against the heart rate–oxygen consumption method

Mean BMI of the participants was 19·5 (sd 2·9) kg/m2. The mean PAL of the HRVO2 method (1·69 (sd 0·21)) was not significantly different from the mean PAL of the 24 h AD (1·65 (sd 0·18)), but was significantly higher than that derived from the accelerometer (1·51 (sd 0·23); P < 0·001; Table 1).

Table 1 Comparison of physical activity level (PAL) derived from accelerometry, the detailed 24 h activity diary (24 h AD), the heart rate–oxygen consumption (HRVO2) method and the physical activity questionnaire

*Subjects were males from Bangalore city, South India.

†Subjects were males and females from two villages in Andhra Pradesh state and Bangalore city, South India.

‡Mean value was significantly different from that of the detailed 24 h AD and the HRVO2 method (P < 0·001).

§Mean value was significantly different from that of the detailed 24 h AD and the questionnaire (P < 0·001).

The mean bias and limits of agreement of PAL derived from the accelerometer were larger (−0·17; −0·28, −0·06) than those obtained for the 24 h AD (0·04; −0·10, 0·01) when compared with the HRVO2 method. The PAL derived from the accelerometer showed a correlation of 0·64 (P = 0·018), while with that from the 24 h AD was higher at 0·91 (P < 0·001), when compared with the HRVO2 method.

Protocol 2: validation of the questionnaire against accelerometry and the 24 h activity diary

The mean BMI of this group was 22 (sd 3) kg/m2 (13 % underweight, 15 % overweight). The mean PAL of the questionnaire (1·72 (sd 0·20)) was not significantly different from the mean PAL of the 24 h AD (1·78 (sd 0·20)), but was significantly higher than the mean PAL of the accelerometer (1·36 (sd 0·20); P < 0·001; Table 1). Correlations of PAL from the questionnaire with that of the accelerometer (r = 0·28; P = 0·01) and the 24 h AD (r = 0·30; P = 0·006) were modest.

Using Bland–Altman plots (Fig. 1a and 1b), the mean bias of the PAL from the questionnaire was larger against the PAL from the accelerometer (0·36) than against the 24 h AD (−0·06), but with large limits of agreement against both (−0·14, 0·85 and −0·59, 0·47). Age, gender or BMI did not predict the mean biases that were obtained between the above methods.

Fig. 1 Bland–Altman plots showing the mean bias (—) and limits of agreement (- - -) for physical activity level (PAL) measured between (a) the questionnaire and the accelerometer and (b) the questionnaire and the detailed 24 h physical activity diary (24 h AD) among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India

Participants were ranked into three categories of PAL (sedentary/light activity = 1·40–1·69, moderately active = 1·70–1·99, vigorously active = 2·00–2·40) obtained from published sources(16). Fifty-five per cent of the participants were correctly categorized by the questionnaire (using 24 h AD as the reference method) and 8 % were misclassified into extreme categories (sedentary/light activity to vigorously active lifestyle or vice versa).

Assessment of the relative validity of the accelerometer counts for individual activities

Various individual activities reported by the participants were extracted out of the 24 h AD along with the simultaneously recorded accelerometer counts from those activities. For each individual reporting an activity, the average 2 min counts were computed, and the median of all the means of individuals reporting the same activity were used in the analysis.

First, the accelerometer counts were compared with known accelerometer count cut-offs from published sources(Reference Freedson, Melanson and Sirard22). For example, an accelerometer count of less than 1952 corresponds to a MET of 3, i.e. ‘light’ intensity category (Fig. 2). Second, the relationship of the accelerometer-derived counts with the intensity of these activities was determined using MET values of these activities from published sources(16, Reference Ainsworth, Haskell and Whitt23).

Fig. 2 Median accelerometer counts of specific activities: (a) activities with intensity between 1·0 and 3·0 MET (metabolic equivalents) and (b) activities with intensity between 3·1 and 8·0 MET, among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India. The horizontal line at 1952 counts represents the upper limit for light activities based on Freedson’s equation

There was a significant positive correlation between the accelerometer counts of the specific activities with their published MET values (r = 0·56; P < 0·001). The scatter plot of this relationship indicated that activities at higher intensities did not fit into a linear relationship. A cubic curve fit model between the accelerometer counts and the MET values (R 2 = 0·41; P < 0·001) predicted the relationship better than a linear model (R 2 = 0·31; P < 0·001; Fig. 3).

Fig. 3 Regression line for median accelerometer counts of reported individual activities (determined among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India) and published MET (metabolic equivalents). – – –, linear fit regression model (R 2 = 0·31, P < 0·001); ——, cubic curve fit model (R 2 = 0·41, P < 0·001)

Discussion

The questionnaire-derived PAL had significant, modest correlations with the PAL derived from the accelerometer and the detailed 24 h AD. The strength of the correlations is similar to those observed in the validation of questionnaire-derived physical activity parameters elsewhere(Reference Ainsworth, Bassett, Strath, Swartz, O’Brien, Thompson, Jones, Macera and Kimsey24, Reference Taber, Schmidt, Roberts, Hosmer, Markenson and Freedson25). It is conceivable that correlations between the questionnaire and other methods would be stronger if a larger number of measurement days of the reference methods (accelerometer, 24 h AD) were obtained to better reflect the one month recall period captured by the questionnaire. It is also likely that the questionnaire may in fact have stronger correlations with the 24 h AD than obtained in the present study (Protocol 2), since the 24 h AD was filled out in some instances by a family member rather than the individual (because of illiteracy). This is suggested by the rather higher correlation between accelerometer-derived PAL and 24 h AD when the 24 h AD was filled in by the participant alone (r = 0·64, Protocol 1) as compared with a modest correlation of r = 0·37 (Protocol 2) when the 24 h AD was filled in by a combination of participants and family members. In the present study, PAL from the accelerometer was significantly lower than that from the HRVO2 method and the detailed 24 h AD. The underestimate of PAL by the accelerometer was confirmed when compared with the 24 h AD in the study group where the questionnaire was validated. The underestimate of PAL by the accelerometer may in part be due to the underestimation of accelerometer counts for the various moderate to vigorous-intensity activities observed in our study. Similar problems have been reported for this accelerometer (formerly Computer Science and Application Inc.) and for other accelerometers as well(Reference Leenders, Sherman and Nagaraja26Reference Hendelman, Miller, Baggett, Debold and Freedson28). In contrast, the mean PAL from the questionnaire compared well with that obtained with the 24 h AD although 24 h AD is prone to over-reporting as it relies on self-report(Reference Irwin, Ainsworth and Conway29).

An obvious limitation in the current study is that free-living activities and published, rather than actually measured MET were used to compare with the accelerometer-derived counts. It is conceivable that the published MET compiled from various sources may have been different from the true value due to variations in the descriptions of the activities, varied methodology in measuring MET and in the estimation of BMR to derive MET(Reference Vaz, Karolis, Draper and Shetty30). However, it is unlikely that the underestimation of the intensity for the whole range of activities captured by the accelerometer can be explained by these limitations alone. Evidence shows that physical activities that are more complex (having a combination of both dynamic and static movements), static activities and activities involving upper-body movements are poorly captured by accelerometers(Reference Leenders, Sherman and Nagaraja26, Reference Crouter, Clowers and Bassett27). To overcome these, studies have used triaxial accelerometers, but with little improvement in assessing these activities(Reference Welk, Blair, Wood, Jones and Thompson31).

In summary, the present study demonstrates that the questionnaire has reasonable validity, concordant with other published studies that capture the patterns of physical activity across a wide range of behaviours (sedentary to heavily active) in large epidemiological studies(Reference Ainsworth, Bassett, Strath, Swartz, O’Brien, Thompson, Jones, Macera and Kimsey24, Reference Taber, Schmidt, Roberts, Hosmer, Markenson and Freedson25). MET values of the free-living activities, if measured, would have provided accurately the underestimations of the accelerometer and also revealed if existing published MET are valid for this community. The impact of the day-to-day variation of physical activity (within-subject) in the estimation of PAL derived from the accelerometer and the 24 h AD would have been possible if additional days of physical activity recordings were captured. Although the questionnaire is designed to also capture individual physical activity components (occupation, household activities, etc.), its validity for this purpose needs to be ascertained. There is also a need to assess the performance of this questionnaire in other regions in India.

Acknowledgements

Sources of funding: The work was funded by the Wellcome Trust, project grant GR070797MF. Conflict of interest: None declared. Contribution of the authors: A.V.B. and M.V. were involved in the study design, carrying out the study, data analysis and writing of the manuscript. R.K. and A.V.K. were involved in the study design, carrying out the study, data analysis and reviewing the final draft of the manuscript. T.T. analysed the data and reviewed the manuscript. S.E., S.K. and T.L. were involved in data analysis and reviewed the final draft of the manuscript. S.K.R. and D.P. were involved in the original conception of the Indian Migration Study and reviewed the final draft of the manuscript. Acknowledgements: The authors would like to thank Dr A. Jacob (Director of Emmaus Swiss Leprosy Project, Palamaner) who facilitated the study, Jayachitra P. Parimala and Mr Subramanian for supervising the field operations and data entry. The Indian Migration Study Group comprises the following individuals. Regional Principal Investigators: Professor R.C. Ahuja, Professor R.K. Saran (Lucknow), Dr Prashant Joshi (Nagpur), Professor S. Mohan Das and Dr R.K. Jain (Hyderabad) and Dr Murali Mohan (Bangalore). Collaborators: Professor Tulsi Patel, Dr Lakshmy Ramakrishnan, Dr K.V.R. Sarma, Professor Anura V. Kurpad, Dr Mario Vaz, Ms A.V. Bharathi and Professor Kamla Gupta (Mumbai), Dr Chittaranjan Yajnik (Pune) and Dr Ian Baker, Professor George Davey Smith, Professor Yoav Ben Shlomo and Dr Kate Tilling (Bristol). Industrial Medical Officers: Dr N.M. Thakre (Nagpur) and Dr S.S. Potnis (Hyderabad).

References

1.Lamonte, MJ & Ainsworth, BE (2001) Quantifying energy expenditure and physical activity in the context of dose response. Med Sci Sports Exerc 33, 6 Suppl., 370S378S.CrossRefGoogle Scholar
2.Haskell, WL & Kiernan, M (2000) Methodologic issues in measuring physical activity and physical fitness when evaluating the role of dietary supplements for physically active people. Am J Clin Nutr 72, 2 Suppl., 541S550S.CrossRefGoogle Scholar
3.Shephard, RJ (2003) Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med 37, 197206.CrossRefGoogle Scholar
4.Richardson, MT, Ainsworth, BE, Wu, HC, Jacobs, DR Jr & Leon, AS (1995) Ability of the Atherosclerosis Risk In Communities (ARIC)/Baecke questionnaire to assess leisure-time physical activity. Int J Epidemiol 24, 685693.CrossRefGoogle Scholar
5.Helmert, U, Herman, B & Shea, S (1994) Moderate and vigorous leisure-time physical activity and cardiovascular disease risk factors in West Germany, 1984–1991. Int J Epidemiol 23, 285292.CrossRefGoogle Scholar
6.Kanade, AN, Rao, S, Yajnik, CS, Margetts, BM & Fall, CHD (2005) Rapid assessment of maternal activity among rural Indian mothers (Pune Maternal Nutrition Study). Public Health Nutr 8, 588595.Google Scholar
7.Vaz, M & Bharathi, AV (2004) How sedentary are people in ‘sedentary’ occupations? The physical activity of teachers in urban South India. Occup Med 54, 369372.CrossRefGoogle Scholar
8.Vaz, M & Bharathi, AV (2000) Practices and perceptions of physical activity in urban, employed, middle-class Indians. Indian Heart J 52, 301306.Google Scholar
9.Chasen Taber, LJ, Erickson, B, Nasca, PC, Chasan-Taber, S & Freedson, PS (2002) Validity and reproducibility of a physical activity questionnaire in women. Med Sci Sports Exerc 34, 987992.CrossRefGoogle Scholar
10.Craig, CL, Marshall, AL, Sjostrom, M, Bauman, AE, Booth, ML & Ainsworth, BE (2003) International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 35, 13811395.CrossRefGoogle Scholar
11.Vaz, M & Bharathi, AV (2004) Perceptions of the intensity of specific physical activities in Bangalore, South India: implications for exercise prescription. J Assoc Physicians India 52, 541544.Google Scholar
12.Bharathi, AV, Sandhya, N & Vaz, M (2000) The development and characteristics of a physical activity questionnaire for epidemiological studies in urban middle class Indians. Indian J Med Res 111, 95102.Google Scholar
13.Lyngdoh, T, Kinra, S, Shlomo, YB et al. (2006) Sib-recruitment for studying migration and its impact on obesity and diabetes. Emerg Themes Epidemiol 3, 2.Google Scholar
14.James, WPT & Schofield, EC (1990) Human Energy Requirements: A Manual for Planners and Nutritionists. Oxford: Oxford University Press.Google Scholar
15.Kurpad, AV, Raj, R, Maruthy, KN & Vaz, M (2006) A simple method of measuring total daily energy expenditure and physical activity level from the heart rate in adult men. Eur J Clin Nutr 60, 3240.CrossRefGoogle Scholar
16.Food and Agriculture Organization of the United Nations/World Health Organization/United Nations University (2001) Human Energy Requirements. Report of a Joint Expert Consultation. FAO Food and Nutrition Technical Report Series no. 1. Rome: FAO.Google Scholar
17.Bassett, DR Jr, Ainsworth, BE, Swartz, AM, Strath, SJ, O’Brien, WL & King, GA (2000) Validity of four motion sensors in measuring moderate intensity physical activity. Med Sci Sports Exerc 32, 9 Suppl., S471S480.CrossRefGoogle Scholar
18.Horner, NK, Lampe, JW, Patterson, RE, Neuhouser, ML, Beresford, SA & Prentice, RL (2001) Indirect calorimetry protocol development for measuring resting metabolic rate as the component of total energy expenditure in free-living postmenopausal women. J Nutr 131, 22152218.CrossRefGoogle Scholar
19.Spurr, GB, Prentice, AM, Murgatroyd, PR, Goldberg, GR, Reina, JC & Christman, NT (1988) Energy expenditure from minute-by-minute heart-rate recording: comparison with indirect calorimetry. Am J Clin Nutr 48, 552559.CrossRefGoogle Scholar
20.Weirs, JBD (1949) New methods for calculating metabolic rate with specific reference to protein metabolism. J Physiol 109, 19.CrossRefGoogle Scholar
21.Bland, JM & Altman, DG (1986) Statistical methods for assessing agreement between methods of clinical measurements. Lancet 1, 307310.CrossRefGoogle Scholar
22.Freedson, PS, Melanson, E & Sirard, J (1998) Calibration of the Computer Science and Applications, Inc. accelerometer. Med Sci Sports Exerc 30, 777781.CrossRefGoogle Scholar
23.Ainsworth, BE, Haskell, WL, Whitt, MC et al. (2000) Compendium of physical activities; an update of activity codes and MET intensities. Med Sci Sports Exerc 32, 9 Suppl., S498S504.Google Scholar
24.Ainsworth, BE, Bassett, DR Jr, Strath, SJ, Swartz, AN, O’Brien, WL, Thompson, RW, Jones, DA, Macera, CA & Kimsey, CD (2000) Comparison of three methods for measuring the time spent in physical activity. Med Sci Sports Exerc 32, 9 Suppl., S457S464.CrossRefGoogle Scholar
25.Taber, LC, Schmidt, MD, Roberts, DE, Hosmer, D, Markenson, G & Freedson, PS (2004) Development and validation of a pregnancy physical activity questionnaire. Med Sci Sports Exerc 36, 17501760.CrossRefGoogle Scholar
26.Leenders, NYJM, Sherman, WM & Nagaraja, HN (2000) Comparison of four methods of estimating physical activity in adult women. Med Sci Sports Exerc 32, 13201326.Google Scholar
27.Crouter, SE, Clowers, KG & Bassett, DR Jr (2005) A novel method for using accelerometer data to predict energy expenditure. J Appl Physiol 100, 13241331.CrossRefGoogle Scholar
28.Hendelman, D, Miller, K, Baggett, C, Debold, E & Freedson, P (2002) Validity of accelerometry for the assessment of moderate intensity physical activity in the field. Med Sci Sports Exerc 32, 9 Suppl., S442S449.Google Scholar
29.Irwin, ML, Ainsworth, BE & Conway, JM (2001) Estimation of energy expenditure from physical activity measures: determinants of accuracy. Obes Res 9, 517525.CrossRefGoogle Scholar
30.Vaz, M, Karolis, N, Draper, A & Shetty, P (2005) A compilation of energy cost of physical activities. Public Health Nutr 8, 11531183.Google Scholar
31.Welk, GJ, Blair, SN, Wood, K, Jones, S & Thompson, RW (2000) A comparative evaluation of three accelerometry based physical activity monitors. Med Sci Sports Exerc 32, 9 Suppl., S489S497.Google Scholar
Figure 0

Table 1 Comparison of physical activity level (PAL) derived from accelerometry, the detailed 24 h activity diary (24 h AD), the heart rate–oxygen consumption (HRVO2) method and the physical activity questionnaire

Figure 1

Fig. 1 Bland–Altman plots showing the mean bias (—) and limits of agreement (- - -) for physical activity level (PAL) measured between (a) the questionnaire and the accelerometer and (b) the questionnaire and the detailed 24 h physical activity diary (24 h AD) among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India

Figure 2

Fig. 2 Median accelerometer counts of specific activities: (a) activities with intensity between 1·0 and 3·0 MET (metabolic equivalents) and (b) activities with intensity between 3·1 and 8·0 MET, among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India. The horizontal line at 1952 counts represents the upper limit for light activities based on Freedson’s equation

Figure 3

Fig. 3 Regression line for median accelerometer counts of reported individual activities (determined among eighty-three participants from two villages in Andhra Pradesh state and Bangalore city, South India) and published MET (metabolic equivalents). – – –, linear fit regression model (R2 = 0·31, P < 0·001); ——, cubic curve fit model (R2 = 0·41, P < 0·001)