Health Reports
Comparison of self-reported and accelerometer-measured physical activity in Canadian adults

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

by Rachel C. Colley, Gregory Butler, Didier Garriguet, Stephanie A. Prince and Karen C. Roberts

The Canadian Physical Activity Guidelines (PAG) for adults (⋝ 150 minutes per week of moderate-to-vigorous physical activity [MVPA] in bouts of ⋝ 10 minutes)^{Note 1} are supported by a large body of evidence, which indicates that physical activity is associated with a reduced risk of chronic disease and all-cause mortality.^{Note 2Note 3} The inclusion of devices to objectively measure physical activity as part of national population health surveys in Canada has broadened the scope of options for physical activity surveillance.^{Note 4} However, the lack of agreement between measured and self-reported estimates of physical activity^{Note 5Note 6Note 7Note 8} has created a surveillance challenge. Data from the 2016 Canadian Community Health Survey (CCHS)indicate that almost half of Canadian adults report that they were at least moderately active in their leisure time,^{Note 9} whereas accelerometer-measured data from the 2014-to-2015 Canadian Health Measures Survey (CHMS) indicate that only 17% meet the current PAG.^{Note 10} Differences in the presence of, and degree of association between, self-reported and accelerometer-measured physical activity as this relates to health add an additional challenge to reconciling the differences between methods.^{Note 11}

Questionnaires are relatively easier and cheaper to implement within a population health surveillance system than objective measurement tools and are therefore more commonly used. For this reason, it is important to develop and sustain valid and reliable questionnaires that capture this health behaviour. From 2007 to 2011, the CHMS used the Minnesota Leisure-Time Physical Activity Questionnaire (MLTPAQ). An analysis of CHMS data observed large differences between the MLTPAQ and the accelerometer in average daily minutes of physical activity, as well as in the classification with respect to meeting the PAG. For example, differences between methods were as high as 37.5 minutes per day in one direction or the other, and about 40% of the population met the PAG according to one method and not the other.^{Note 5}

The 2012-to-2013 CHMS adopted the International Physical Activity Questionnaire (IPAQ). The IPAQ addressed some of the key limitations of the MLTPAQ, accounting for the 10-minute-bout stipulation set out in the PAG and assessing MVPA across transportation, recreational, occupational and household domains in accordance with an emerging global consensus.^{Note 12} However, the IPAQ exhibited a low correlation with accelerometer data in the CHMS and classified almost all Canadians (90%) as meeting the PAG^{Note 6}; a finding consistent with a previous study.^{Note 13} In the 2012-to-2013 CHMS, a newly developed questionnaire module, the Physical Activity Adult Questionnaire (PAAQ), was tested alongside the IPAQ. A limited analysis on a sub-sample of the CHMS (n = 112) indicated that physical activity data from the PAAQ related more strongly to the accelerometer-measured data than did the IPAQ data (R=0.44 versus R=0.20) and yielded more plausible results for adherence to the current PAG (percentage meeting the PAG: 70% when all minutes of measured MVPA were used, 61% when self-reported data from the PAAQ were used, and 90% when self-reported data from the IPAQ were used).^{Note 6} Consequently, the PAAQ was fully implemented in both the CHMS (the 2014-to-2015 cycle) and the CCHS (2015 and 2016).

Resolving the differences between self-reported and objectively measured physical activity is an important surveillance challenge currently facing population health experts in Canada. The objective of this paper is to compare measured and reported physical activity data by using data from the 2014-to-2015 CHMS. A secondary objective is to compare associations with obesity markers between self-reported and accelerometer-measured physical activity. Finally, the self-reported estimates from the CHMS are compared with those obtained from the larger sample of respondents from the 2015 and 2016 CCHS.

Methods

Data sources

The Canadian Health Measures Survey (CHMS) is an ongoing survey conducted by Statistics Canada that collects reported and measured health information from a representative sample of the Canadian household-dwelling population aged 3 to 79 years. Residents of Indian reserves, institutions, certain remote regions, and the territories, and full-time members of the Canadian Forces were excluded. This analysis includes data on adults aged 18 to 79 years from Cycle 4 of the CHMS that were collected in 2014 and 2015. A total of 2,388 adult respondents had valid accelerometer and PAAQ data. A further 16 respondents were excluded on the basis of an outlier analysis. Specifically, respondents who reported more than 3.5 hours per day of recreation- (n=2) or transportation-based (n=8) physical activity or who reported more than 100 minutes per day of vigorous activity (n=6) were excluded. The excluded values were clear outliers as determined by means of a visual examination of the distributions and were all more than 6 standard deviations above the mean. These outlier rules were adapted from an approach employed in a previous analysis using CHMS data.^{Note 6} This analysis is therefore based on 2,372 respondents. CHMS respondents completed an interviewer-administered questionnaire in their home and visited a mobile examination centre (MEC) within the next six weeks to undergo a series of physical measurements. The questionnaire and accelerometer measurements were not taken during the exact same week. This therefore means that all analyses herein are making the assumption that both methods are capturing “typical” physical activity habits. Further detail about the CHMS, including ethics approval information, is available in previous publications.^{Note 14Note 15Note 16}

The Canadian Community Health Survey (CCHS) is an ongoing cross-sectional survey conducted by Statistics Canada that collects information related to health status, health care utilization and determinants of health for the Canadian population aged 12 and older. Physical activity is measured in the CCHS via the household questionnaire only (i.e., no accelerometer measurement) and is therefore included here to assess reliability of the estimates obtained from the questionnaire. The CCHS covers approximately 97% of the Canadian population aged 12 and older. Excluded from the survey’s coverage are people living on reserves and other Aboriginal settlements, full-time members of the Canadian Forces, people living in institutions, and people living in the Quebec health regions of Nunavik and Région des Terres-Cries-de-la-Baie-James. This analysis includes data collected in 2015 and 2016 from a subsample aged 18 to 79 years (n = 90,080) to match the age of respondents from the CHMS. The same outlier exclusions were applied to the CCHS analysis as were used in the CHMS (n=1,425).

Physical activity measured by accelerometer (CHMS only)

Upon completion of the MEC visit, ambulatory respondents were asked to wear an Actical accelerometer (Philips Respironics, Oregon, United States) over their right hip on an elasticized belt during waking hours for seven consecutive days. All respondents were blind to the data while they wore the device. The Actical measures and records time-stamped acceleration in all directions, providing an index of physical activity intensity via a count value for each minute. A valid day was defined as having 10 or more hours of wear time, and a valid respondent was defined as having a minimum of four valid days.^{Note 17} Wear time was determined by subtracting nonwear time from 24 hours. Nonwear time was defined as at least 60 consecutive minutes of 0 counts, with allowance for one to two minutes of counts between 0 and 100.^{Note 17} Published movement intensity thresholds were applied to the data for the purpose of deriving time spent in light-intensity physical activity (LPA) and MVPA.^{Note 18} Analyses included two MVPA variables: one that included all minutes of MVPA (MVPA_ALL) and another that included only minutes accumulated in bouts of 10 minutes or more (MVPA_BOUTS). A complete description of the accelerometer data reduction procedures is available elsewhere.^{Note 4Note 15Note 17}

Physical activity measured by questionnaire (CHMS and CCHS)

As part of the household questionnaire, CHMS and CCHS respondents were asked to provide estimates of time spent in the last seven days engaged in transportation (PAAQ_TRA), recreational (PAAQ_REC), or occupational or household (PAAQ_OCC) physical activity (Appendix A). These values were summed to give total physical activity (PAAQ_TOTAL). Respondents were then asked to estimate the number of minutes in the last seven days during which they engaged in vigorous-intensity physical activity (PAAQ_VPA). Average daily values of each domain of physical activity were calculated by dividing the total values by seven.

Obesity measures (CHMS only)

Body mass index (BMI) was calculated as measured weight in kilograms divided by measured height in metres squared (kilograms per metre squared; kg·m^-2). A ProScale M150 digital stadiometer (Accurate Technology Inc., Fletcher, United States) was used to measure height to the nearest 0.1 centimetre, and a Mettler Toledo VLC with Panther Plus terminal scale (Mettler Toledo Canada, Mississauga, Canada) was used to measure weight to the nearest 0.1 kilogram. A flexible tape was used to measure waist circumference to the nearest 0.1 centimetre.

Statistical analysis

Descriptive statistics were used to calculate means, 95% confidence intervals, the standard error of the estimate, and the coefficient of variation. Pairwise contrasts were used to assess differences between sex and age groups. Pearson correlation coefficients were used to assess the relationship between measured and reported estimates of physical activity. To provide context, typical correlation coefficients obtained when comparing self-reported and accelerometer-measured physical activity are low-to-moderate and range from -0.71 to 0.96.^{Note 7} Weighted histograms were used to present the distribution of the mean difference (calculated as measured estimate – reported estimate) between measured and reported physical activity variables. Measured (MVPA_ALL and MVPA_BOUTS) and reported variables (PAAQ_REC+TRA and PAAQ_TOTAL) were used to assess the percentage of respondents meeting the current PAG (⋝ 150 minutes of MVPA per week). A classification analysis was conducted to assess differences in how respondents were classified as meeting or not meeting the PAG. Given that seven complete days of accelerometer data were not available for all respondents (respondents were required to have four or more valid days of data to be included), respondents were deemed adherent if their average daily MVPA was greater than or equal to 21.43 minutes per day (150 minutes / 7 days). Association with obesity measures was assessed by means of linear regression controlling for age and sex.

To account for the complex survey design and non-response bias and to correctly estimate variance, all analyses were weighted by means of the survey weights generated by Statistics Canada for Cycle 4 of the CHMS^{Note 15} and the 2015 and 2016 CCHS.^{Note 19} The data were analyzed using SAS 9.3 (SAS Institute, Cary, United States) and SUDAAN 11.0 using denominator degrees of freedom (DDF=11) in the SUDAAN procedure statements for the CHMS analyses. To account for survey design effects, the bootstrap technique was used to estimate 95% confidence intervals.^{Note 15} Response rates were 40% for the CHMS (reflecting the analytical requirement of at least four valid days of accelerometer data) and 57.5% for the 2015 and 2016 CCHS.

Results

Descriptive statistics

Adults in the CHMS sample accumulated, on average, 23 minutes per day of measured MVPA_ALL, 11 minutes per day of MVPA_BOUTS, and 200 minutes per day of LPA (Table 1). Estimates were higher in males than in females for all types of physical activity, but these differences rarely reached statistical significance. Adults in the CHMS reported, on average, that they accumulated 15 minutes per day of transportation activity (PAAQ_TRA), 11 minutes per day of recreational activity (PAAQ_REC), and 22 minutes per day of occupational or household activity (PAAQ_OCC) (Table 2). Average daily vigorous physical activity (VPA) was low (< 5 min·d^-1) according to both the measured and the reported data (Tables 1 and 2).

Correlation analysis

MVPA_ALL and MVPA_BOUTS were positively correlated with PAAQ_REC, PAAQ_TRA, PAAQ_REC+TRA and PAAQ_TOTAL (Figure 1). The highest correlation coefficient observed was between the accelerometer-measured variables (MVPA_ALL and MVPA_BOUTS) and PAAQ_REC+TRA (R = 0.34/0.36, p < 0.0001). The strength of the correlation between PAAQ_TOTAL and the measured variables was weakened by the negative direction of the relationship between the accelerometer-measured variables and PAAQ_OCC (R = -0.038 [not significant] for MVPA_ALL and -0.094 [p < 0.0001] for MVPA_BOUTS) (Figure 1). Measured LPA was positively correlated with PAAQ_OCC (R = 0.20, p < 0.0001) and PAAQ_TOTAL (R = 0.15, p < 0.0001), but was negatively and weakly correlated with PAAQ_REC and PAAQ_TRA. The correlation between some accelerometer-measured and self-reported variables was slightly stronger in males than in females (e.g., MVPA_BOUTS and PAAQ_REC+TRA: R = 0.39 in males and R = 0.31 in females, both p < 0.0001). The strength of correlation between MVPA_BOUTS and PAAQ_REC+TRA was relatively stable across age groups for both males and females (data not shown).

Self-reported vigorous physical activity (PAAQ_VPA) and measured VPA were positively and weakly correlated (VPA_ALL: R = 0.21, p < 0.0001; VPA_BOUTS: R = 0.24, p < 0.0001) (data not shown). The correlation between measured and reported vigorous physical activity was stronger when using measured data accumulated in 10 minute bouts and was significant in 18 to 59 year olds (e.g., 18 to 39 year olds: R = 0.24, p < 0.0001; 40-59 year olds: R = 0.29, p < 0.0001) but not in respondents aged 60 and older. The strongest correlation for vigorous physical activity was observed in males aged 40 to 59 years (R=0.35, p < 0.0001).

Mean difference analysis

The mean difference between MVPA_ALL and PAAQ_REC+TRA was less (-4.4 min·d^-1, 95% confidence interval [CI]: -7.5 to -1.3) than the difference between MVPA_BOUTS and PAAQ_REC+TRA (-16.3 min·d^-1, 95% CI: -18.7 to -14.0) (data not shown). The mean difference between MVPA_ALL and PAAQ_TOTAL was less (-26.2 min·d^-1, 95% CI: -33.0 to -19.4) than the difference between MVPA_BOUTS and PAAQ_TOTAL (-38.2 min·d^-1, 95% CI: -44.6 to -31.7) (data not shown). The mean difference (accelerometer-measured minus self-reported) in minutes for PAAQ_TOTAL was within +/- 12.5 minutes in 50% of respondents when MVPA_ALL was used (Figure 2) and 55% of respondents when MVPA_BOUTS was used (data not shown). The mean difference (accelerometer-measured minus self-reported) in minutes for PAAQ_REC+TRA was within +/- 12.5 minutes in 43% of respondents when MVPA_ALL was used (Figure 2) and 44% of respondents when MVPA_BOUTS was used (data not shown). The mean difference between self-reported VPA was -1.6 minutes per day [95% CI: -2.7 to -0.5] for VPA_ALL and -3.0 minutes per day [95% CI: -3.96 to -1.96] for VPA_BOUTS. The difference between measured (VPAALL) and reported vigorous physical activity was within +/- 5 minutes in 77% of people (data not shown). No notable differences in mean difference existed for age, sex and obesity status (underweight and healthy weight versus overweight or obese); however, differences were evident by quintile of accelerometer-measured minutes of MVPA_ALL (Figure 3). Less active people were more likely than the most active people to report greater physical activity than accumulated on the accelerometer (i.e., mean difference was a negative number). On average, the most active people reported less activity than was measured on the accelerometer (i.e., mean difference was a positive value). The contrast in mean difference between MVPA_ALL and PAAQ_REC+TRA between the lowest and highest quintiles was significant (-9.2 min·d^-1 [95% CI: -12.9 to -5.5] for the lowest quintile, versus +5.1 min·d^-1 [95% CI: -4.5 to +14.7] for the highest quintile) (Figure 3).

Adherence to physical activity guidelines

Percentage adherence to the current PAG (⋝ 150 minutes of MVPA per week) varied according to which variable was used. The percentage was lower for the accelerometer-measured variables (17% for MVPA_BOUTS and 39% for MVPA_ALL) than for the self-reported variables (46% for PAAQ_REC+TRA and 60% for PAAQ_TOTAL) (Figure 4). A classification analysis showed that accelerometer-measured MVPA_ALL and self-reported (PAAQ_REC+TRA) were in agreement 67% of the time (i.e., both classifying respondents as either meeting or not meeting the PAG) (Figure 5). The remaining 33% was split between the accelerometer classifying respondents as meeting the PAG but not the PAAQ (20%), and the PAAQ classifying respondents as meeting the PAG but not the accelerometer (13%).

Association with obesity markers

According to linear regression models adjusted for age and sex, MVPA_ALL, MVPA_BOUTS, PAAQ_REC, PAAQ_TRA and PAAQ_REC+TRA were all negatively associated with BMI and WC, while measured LPA and PAAQ_OCC were not associated with either obesity measure (Figure 6). The effect size (beta) for the association was greater for the measured variables than for the self-reported variables. While the level of significance for PAAQ_REC+TRA (p = 0.0005 for BMI and p = 0.0006 for WC) was greater than for either PAAQ_REC (p = 0.016 for BMI and p = 0.01 for WC) or PAAQ_TRA (p = 0.046 for BMI and p = 0.040 for WC) on their own, the effect size (beta) was not any greater for PAAQ_REC+TRA. This was explained by wider confidence intervals around the association for the individual variables than for the combination of recreation and transportation (data not shown).

Discussion

This study observed large differences in physical activity estimates and modest correlation between self-reported and accelerometer-measured physical activity. The highest strength of correlation between accelerometer-measured and self-reported data in this study was observed between the sum of self-reported recreation and transportation physical activity and measured minutes of MVPA (Figure 1). Associations with health markers existed between both measured and reported physical activity variables, but were stronger when measured variables were used. Differences in the percentage of Canadians meeting the PAG were observed between measured and reported methods. This therefore presents a surveillance reporting challenge. While the results suggest that accelerometer-measured and self-reported physical activity estimates should not be used interchangeably, the two methods provided PAG adherence values at the population level that were somewhat aligned (39% versus 46%, Figure 4), and there was also some consistency in the direction of association with markers of obesity. Collectively, the results of this study highlight the importance for data users to understand the differences between methods and to exercise caution in using estimates derived from these methods interchangeably. The level of correlation observed in this study is similar to that obtained in a previous analysis using the same questionnaire in a preliminary CHMS sample^{Note 6} and stronger than previously observed with other questionnaire tools used in the CHMS, such as the IPAQ (R = 0.20)^{Note 6} or MLTPAQ (R = 0.22 to 0.26).^{Note 5} The previous analysis of a preliminary CHMS sample (n = 112) lacked the sample size to investigate correlations by domain, but reported that PAAQ_TOTAL was moderately correlated with measured MVPA_BOUTS at a level of R = 0.44.^{Note 6} The degree of correlation between the same variables in the present study was weaker (R = 0.14) and likely a reflection of some bias in the small sample used in the preliminary study. The present study did observe higher correlations (e.g., R = 0.42) between PAAQ_REC+TRA and MVPA in 40- to 59-year-old men. PAAQ_REC and PAAQ_TRA were positively correlated, while PAAQ_OCC was negatively correlated with measured MVPA. This resulted in a weakening of the level of correlation observed between PAAQ_TOTAL and accelerometer-measured MVPA.

A comparison between self-reported and accelerometer-measured physical activity in the 2005 and 2006 National Health and Nutrition Examination Survey (NHANES)^{Note 20} reported similar correlation coefficients by domain: overall total (R = 0.27 in NHANES and R = 0.22 in CHMS), recreational activity (R = 0.29 in NHANES and R = 0.26 in CHMS), transportation activity (R = 0.20 in NHANES and R = 0.23 in CHMS), and occupational or household activity (R = 0.08 in NHANES and R = -0.04 in CHMS). The questionnaires used in the NHANES and present analyses were different (Tucker and colleagues assigned metabolic equivalent values to activities from the Compendium of Physical Activities^{Note 21}), but both were designed to provide an estimate of time spent in the various domains of physical activity. The NHANES results also suggest that questions about household and occupational physical activity are likely capturing a combination of LPA and MVPA. Systematic reviews have reported wide variation between studies in the degree and direction of correlation between measured and reported physical activity.^{Note 7Note 8} Further, the strength of correlation varies within studies by age and sex,^{Note 7Note 13} obesity status^{Note 22} and physical activity level.^{Note 23} The present study observed only modest differences in correlation and mean difference by age, sex and obesity status; however, a pattern was evident by level of physical activity. Less active people were more likely than the most active quintile of respondents to report more activity than they accumulated on the accelerometer. This may be a reflection that more active respondents were more likely to report energetic activities that are easier to recall accurately (e.g., sport participation or exercise class) and are likely to be predominantly MVPA, while less active respondents may have reported less energetic activities (e.g., gardening, chores or incidental walking) that are harder to recall accurately and are likely to comprise both LPA and MVPA. Previous research has shown that more energetic or intense activities relate better to accelerometer-measured data.^{Note 23} The agreement between measured and reported vigorous physical activity was quite good in this study (the mean difference was within +/- 5 minutes per day in 77% of respondents) and may provide further evidence of more accurate reporting by the most active people in the study.

The relatively low correlation and large mean differences between accelerometer-measured and self-reported data observed in this study as well as in others is not surprising given that they do not measure the same constructs. Questionnaires capture behaviour or perceived time spent in specific activities or domains (e.g., work or school, play or leisure, or transport) while objective measurement devices capture movement or continuous measures of bodily acceleration above a defined threshold.^{Note 24} Leading experts are now asserting that direct comparisons between estimates from reported and measured methods are unsuitable^{Note 24} and that researchers should stop asking which method is “correct” and rather focus on the richness and complementary information that both methods can offer.^{Note 25} While the present study did use traditional approaches to compare accelerometer-measured and self-reported estimates, such as correlation and mean difference analyses, it also examined whether the differences observed were logical and whether there was any evidence that the two approaches told a similar story. The finding that PAAQ_REC+TRA, and not PAAQ_OCC, was correlated with measured MVPA is notable, particularly because occupational or household activity was correlated with measured LPA. While the strength of association with BMI and WC was weaker when self-reported physical activity was used, the presence and direction of associations were similar for measured MVPA and PAAQ_REC+TRA. This finding is consistent with the literature.^{Note 2Note 3Note 26} The strength of association between reported variables and BMI and WC was higher for PAAQ_REC+TRA than for either domain alone or for PAAQ_TOTAL (Table 3). It appears that PAAQ_OCC was weakening the association between PAAQ_TOTAL and the obesity markers. This finding is consistent with the correlation analysis.

This study has important limitations that could explain, in part, the lack of agreement and low correlation between methods. Firstly, the seven days of accelerometer wear time did not match directly with the self-report response timeframe (i.e., the accelerometer and questionnaire measures occurred during different weeks). Intra-individual variability in activity habits between different days and weeks can be quite high.^{Note 27} Therefore, this analysis is relying on an assumption that both measures are capturing a reflection of a given respondent’s “typical” physical activity habits. This unfortunate mismatch in timing of measurement also means that the correlation strength observed in this study is likely lower than what would have resulted if the two difference measurements occurred simultaneously. Secondly, the inclusion of cycling as an example in the questionnaire for both recreation- and transportation-based activity bears noting, in light of the fact that accelerometry generally does not capture cycling accurately. Finally, asking respondents to report only activities that “make you sweat at least a little and breathe harder” may be interpreted differently by different people, and may not always exclusively capture MVPA. The correlation between LPA and the occupational or household domain suggests that respondents were reporting a combination of LPA and MVPA for this domain.

Both accelerometer-measured physical activity and self-reported measures have been shown by the literature to have beneficial associations with health. However, the cost of accelerometers limits the sample size of surveys that employ them. In Canada, currently, self-reported modules are the best option for reporting across all provincial and territorial jurisdictions. Furthermore, because of their cost, accelerometer data are not collected in many countries. As a result, international comparisons are dependent on self-reported data. In Canada, accelerometers are used to measure physical activity in the CHMS (about 5,000 to 6,000 respondents every two years) but not in the much larger CCHS (about 65,000 respondents every year). The difference is important, as the smaller sample size and clustered sampling frame of the CHMS mean that it is possible to report only national-level estimates every two years while the CCHS data can be reported at the provincial and territorial and health region levels. The ability to examine disparities across the country is an important advantage offered by the self-reported physical activity outcomes obtained via the CCHS. The physical activity estimates by domain and overall were similar between the CHMS and a larger sample from the CCHS (Figure 7). This suggests good reproducibility of the questionnaire module.

Accelerometers are not practical in all settings; however, the richness of accelerometer data cannot be understated. In addition to mitigating concerns about social desirability bias and recall difficulty, the per-minute resolution of actual movement across a seven-day period has led to important contributions to the understanding of how movement relates to health. As the evidence base linking accelerometer-measured physical activity and health grows, the research community may develop complementary PAGs based on device-measured movement.^{Note 24} In the future, these devices may become more readily available and efficient, and consequently facilitate national tracking of adherence to guidelines by means of objective measures.^{Note 24} Schuna and colleagues reported that respondents meeting the PAG according to reported data accumulated only 56 minutes per week of MVPA_BOUTS when an accelerometer was used.^{Note 28} When data from the present study are used, the corresponding average MVPA_BOUTS value for respondents who met the PAG according to PAAQ_REC+TRA was 116 minutes per week (data not shown). The average PAAQ_REC+TRA reported by respondents accumulating ⋝ 150 minutes per week of MVPA_BOUTS was 355 minutes per week. More detailed analyses are needed to clarify whether having different PAGs for objective versus self-reported measures is realistic.

The recently published 2018 Physical Activity Guidelines Advisory Committee Scientific Report^{Note 29} indicates that physical activity accumulated in bouts of any length is associated with health benefits (i.e., bouts need not be 10 minutes long). Interestingly, the present study found that the inclusion of all minutes of measured MVPA was more closely aligned with self-reported activity. Further, researchers are increasingly arguing against threshold-based guidelines, given that any dose of physical activity (particularly moving from none to a little) is associated with the greatest health benefit.^{Note 2Note 30} This evolution in the evidence base linking physical activity and health in adults will be used to inform the revision of PAG in Canada and around the world. Future versions of the PAAQ and any resultant analyses similar to the present one will have to adapt accordingly.

The newly developed Canadian PAAQ was successfully implemented into two large Canadian health surveillance surveys, overcomes key limitations of previous self-report tools used, and aligns with the current focus on capturing activity across the various domains. This study found that MVPA captured by the newly developed Canadian PAAQ module achieved low correlation and agreement with physical activity measured by accelerometry. The study also found some agreement in population-level adherence to the PAG and in associations with markers of obesity. These findings provide further evidence to support the idea that self-reported and objectively measured physical activity levels should not be used interchangeably. Rather, greater focus should be placed on maximizing the richness and potential that both methods offer to help increase the understanding of how behaviour and movement relate to health.

Appendix A: Physical Activity Adult Questionnaire (PAAQ)

Adapted from: Canadian Health Measures Survey documentation available at : http://www23.statcan.gc.ca/imdb-bmdi/instrument/5071_Q1_V5-eng.pdf

Preamble: The following questions are about various types of physical activities done in the last seven days. I want you to only think of activities you did for a minimum of 10 continuous minutes.

1. Transportation

1. In the last 7 days, did you use active ways like walking or cycling to get to places such as work, school, the bus stop, the shopping centre or to visit friends?
   Interviewer's note: Do not include walking, cycling or other activities done purely for leisure. These activities will be asked about later.
   1. Yes/No
2. In the last seven days, on which days did you do these activities?
   1. Monday/Tuesday/Wednesday/Thursday/Friday/Saturday/Sunday
3. How much time, in total, in the last seven days, did you spend doing these activities? Please only include activities that lasted a minimum of 10 continuous minutes.
   1. ___minutes or hours (min: 0, max: 168)

2. Recreation

1. Not including activities you just reported, in the last 7 days, did you do sports, fitness or recreational physical activities, organized or non-organized, that lasted a minimum of 10 continuous minutes?
   Interviewer's note: Examples are walking, home or gym exercise, swimming, cycling, running, skiing, dancing and all team sports.
   1. Yes/No
2. Did any of these recreational physical activities make you sweat at least a little and breathe harder?
   1. Yes/No
3. In the last seven days, on which days did you do these recreational activities that made you sweat at least a little and breathe harder?
   1. Monday/Tuesday/Wednesday/Thursday/Friday/Saturday/Sunday
4. In the last seven days, how much time in total did you spend doing these activitgies that made you sweat at least a little and breathe harder?
   1. ___minutes or hours (min: 0, max: 168)

3. Occupational/Household

1. In the last seven days, did you do any other physical activities while at work, in or around your home or while volunteering?
   Interviewer's note: Examples include carrying heavy loads, shoveling, and household chores such as vacuuming or washing windows. Please remember to only include activities that lasted a minimum of 10 continuous minutes.
   1. Yes/No
2. Did any of these other physical activities make you sweat at least a little and breathe harder?
   1. Yes/No
3. In the last seven days, on which days did you do these other activities that made you sweat at least a little and breathe harder?
   1. Monday/Tuesday/Wednesday/Thursday/Friday/Saturday/Sunday
4. In the last seven days, how much time in total did you spend doing these activities that made you sweat at least a little and breathe harder?
   1. ___minutes or hours (min: 0, max: 168)

4. Vigorous Physical Activity

1. You have reported a total of ___minutes of physical activity (insert sum of transportation, recreation, and occupational or household physical activity). Of these activities, were there any of vigorous intensity, meaning they caused you to be out of breath?
   1. Yes/No
2. In the last seven days, how much time in total did you spend doing vigorous activities that caused you to be out of breath? Please only include activities that lasted a minimum of 10 continuous minutes.
   1. ___minutes or hours (min: 0, max: 168)