The MPAQ was developed (Additional file 1) after reviewing various published validated physical activity questionnaires both in India and abroad. In addition, 24 hour physical activity recalls encompassing a weekday and weekend were collected from 50 volunteers across all ages and occupations. From these 24 hr recalls, the various activities reported across all domains were listed in the MPAQ and similar activities were grouped together and further truncated based on the average energy cost, as the Physical Activity Ratio (PAR) provided by the WHO/FAO 2001 [7]. The MPAQ was designed to capture frequency and duration of habitual obligatory and discretional activities by means of a mix of open and closed-ended questions arranged in four domains viz. work-related activity (work domain), activities of daily living [general activity domain which includes sleep (daytime napping and sleep at night), personal care and domestic chores], transport-related activities (transport domain) and recreational activities (recreational domain). In all domains, options are provided to capture both seasonal and non-seasonal activities. The questionnaire captures details of up to two jobs and elicits information on time spent sitting, standing, walking and climbing stairs in each of these jobs, providing insight into the nature of the job and intensity of work activity.
In the recreational domain, the questionnaire elicits information on sedentary behavior (including TV viewing, chatting with friends, listening to music etc.) as well as light, moderate and vigorous activities on a daily, weekly or monthly basis. In addition, there is provision for recording the extra activities or extra hours of sedentary behavior that happen during the weekend. The questionnaire enables calculation of physical activity for an “average” day by summing up activities in various domains for a 24-hour period. Similarly, weekly and monthly calculations can also be done and information aggregated to compute activity for a year.
Total energy expenditure can be estimated through factorial calculations recommended by a joint FAO/WHO/UNU expert consultation [7]. The factorial calculations are based on the time spent on various activities in the multiple domains and the energy cost of these activities. Energy cost is reported as a multiple of Basal Metabolic Rate (BMR) and called Physical Activity Ratio (PAR). Total time spent on habitual activities is multiplied by PAR to derive the total energy expenditure (TEE) of 24 hours. The physical activity level (PAL) can then be calculated as TEE/BMR for 24 hours. Based on the PAL values [7], individuals can be divided into three categories: Sedentary (1.40 – 1.69), moderately active (1.70-1.99) and vigorously active (2.00-2.40) [7].
Written informed consent was obtained from each participant before start of the reliability and validity studies. Institutional Ethics Committee approval was obtained from the Ethics Committee at MDRF.
Reliability study
The MPAQ was administered by trained interviewers to individuals of either gender aged 20 years and above from 10 states in India namely Tamilnadu, Gujarat, Maharashtra, Jharkhand, Haryana, Bihar, Chandigarh, Assam, Tripura and Arunachal Pradesh. The states were so chosen as to be representative of the country in terms of geography, socioeconomic status, variability of occupations and climatic conditions. From one district in each state, two census enumeration blocks (CEBs) in urban areas and three villages in rural areas were randomly selected (Figure 1). In each CEB or village, 10 households were randomly selected and in each household, one individual was selected to participate in the study. Thus, 50 individuals were selected from each state, and 500 for the entire study. In addition, five individuals from each state were recruited to allow for non-response over time. Hence a total of 550 individuals were initially invited for the study, of whom 543 individuals participated and had all required information at baseline. The participants were sampled so as to obtain individuals across all age categories and both genders with varying literacy levels and engaged in a wide range of occupations so as to test the ability of the MPAQ to measure physical activity of individuals from all walks of life.
Demographic details and information on smoking and alcohol use were obtained from all participants as well as height, weight, waist and blood pressure (BP) measurements, assessed using standardized techniques. Weight (in kilograms-kg) was measured with the subjects wearing light clothing after having removed shoes and heavy jewelry. Height was measured to the nearest centimeter (cm) using a stadiometer with the subjects standing erect without shoes. Body mass index (BMI) was calculated as the weight (kg) divided by the height (in meters squared). Waist circumference was measured using a non-stretchable tape, as the mean of two measurements of the smallest horizontal girth between the costal margins and the iliac crests at minimal respiration. BP was recorded in the sitting position in the right arm using an electronic instrument (Model: HEM- 7101, Omron Corporation, Tokyo, Japan). Two readings were taken 5 minutes apart and the mean of 2 readings was taken as the blood pressure.
The baseline administration of MPAQ was performed from May to August 2011 in all the states. This was followed by a repeat administration within a month for assessing reliability. The interval of one month was chosen based on a previously published study from India [6], and was deemed most appropriate to eliminate recall of previous responses by the participants as well as any possibility of physical activity patterns having significantly altered in the interim.
Validity studies
For assessing relative validity, the MPAQ and the GPAQ were administered in a randomized order by trained interviewers to all selected participants across the 10 states one day apart so as to avoid questionnaire fatigue. The GPAQ was chosen because it is a widely used global PAQ which has been validated in India [8]. The test questionnaire took on an average, 10 minutes (±5) to administer. Subject acceptability and co-operation were good with both questionnaires but the subject understanding was better with the test questionnaire.
Construct validity indicates the consistency or the relationship between the activity instrument (MPAQ) and the physiological variable such as BMI. This was tested by plotting time spent in sitting and moderate and vigorous physical activity (MVPA) (measured as minutes/day) against BMI and waist circumference measured at baseline.
For assessing criterion validity, 107 individuals of either gender aged 20 years and above were recruited from Chennai city in Tamilnadu. The sample was so chosen as to get individuals across a wide age range, both genders and all categories of activity: At the start of the study, information on demographic parameters, height, weight and occupation were obtained as described above. Criterion validity was assessed using the Actigraph (Actilife 5) GT3X+ Triaxial Accelerometer (Actigraph, Pensacola, Florida, USA). Participants were asked to wear the accelerometer for 7 days during waking hours; however, the device was allowed to be removed while bathing or swimming. The device was worn on the hip of the dominant side (right in most cases). The device was worn either above or beneath clothing and not necessarily in contact with skin; however, a snug fit against the body was ensured to avoid erroneous readings.
Accelerometers were initialized to monitor and record data in 60- second “epochs” as “activity counts” and sample frequency at 100 Hz. The start date and time and stop date and time were used for the start and stop of data collection. While initializing, each device was given a unique number denoting the individual participant with their age, gender, height, weight, date of birth and race.
The GT3X+ device collects data from all three axes of movement regardless of the configuration, with Axis 1 collecting the vertical axis acceleration activity data, Axis 2 the horizontal axis data and Axis 3 the perpendicular axis data. The duration (minutes per day) spent in different intensity activities- light (1.5-3 METS, 100 ≤ 1951 counts), moderate (3–6 METS, 1952–5724 counts) and vigorous (>6 METS, ≥ 5725 counts) were determined according to published data [9,10].
The MPAQ was administered anytime during the period the individual was wearing the accelerometer. Data from the MPAQ was computed for a typical week, and then converted to minutes/week, so as to make comparisons with the accelerometer data more realistic.
Inter-rater reliability
MPAQ being an interviewer administered questionnaire, inter rater reliability was measured to assess the agreement between the interviewers. One interviewer administered the questionnaire to the participant while the rest of the interviewers passively observed and rated participant’s response independently. This procedure was completed for a total of 20 participants by all 20 interviewers who collected the questionnaires across the 10 states. A kappa value of 0.83 indicated good agreement among the interviewers.
Statistical analyses
Statistical analyses were performed using a SAS (Statistical Analysis System) statistical package (version 9.0; SAS Institute, Inc., Cary, NC). The results are expressed as mean ± standard deviation or proportions.
Reliability of the MPAQ was examined by calculating the intra class correlation coefficient (ICC) of the activities reported and presented by urban/rural status and gender. ICC values of <0.40 were considered as poor agreement, 0.40-0.59 as fair, 0.60-0.74 as good and 0.75-1.0 as excellent agreement [11]. Relative validity between the GPAQ and MPAQ was also assessed using ICC.
Construct validity was used to assess the degree to which a measure (in this case, the MPAQ) compares with an underlying theoretical construct (a latent variable in this context such as BMI and waist circumference). Linear regression models were used to assess the association of deciles/tertiles of activity with BMI and waist circumference after adjusting for age and gender.
For assessing criterion validity, the MPAQ was compared against the triaxial accelerometer as a criterion. Spearman’s correlation coefficients and 95% CI were used for comparisons. Total duration (minutes/week) of time spent in sedentary and MVPA as estimated from the MPAQ were compared against those recorded by the accelerometer using established cut-points [10]. As the accelerometer measured data for a week, the data obtained from the MPAQ was also computed for a week so as to make it comparable. Accelerometer data were initially downloaded and processed using customized software viz. Actilife Data Analysis Software [Version 5.0], prepared by Actigraph R&D and Software department (Florida, USA). For the purposes of this study, correlation coefficient values: < 0.20, 0.21-0.40, 0.41-0.60, 0.61-0.80 and 0.81-1.0 were considered as weak, fair, moderate, strong and very strong correlation respectively [6].
Bland and Altman plots were used to assess the agreement between data obtained using the MPAQ and accelerometer (within the 95% limits). In addition this plot also indicates the random and systematic errors of the data. The mean difference (bias) between accelerometer and the MPAQ, of sedentary activity/ week were plotted (y- axis) against the mean of estimated sedentary minutes/week obtained from the accelerometer and MPAQ [12]. A similar plot was constructed for MVPA minutes as well. A p-value <0.05 was considered significant for all statistical measures.