Background
Increasing physical activity is associated with reduction in chronic illnesses, such as hypertension and type 2-diabetes [
1‐
3]. The 2008 Physical Activity Guidelines for Americans recommends U.S. adults to engage in a total 150 min of moderate-intensity aerobic activity (i.e., brisk walking) every week or 75 min of vigorous-intensity aerobic activity every week, to be done with at least 10 min bouts of activity [
4]. The self-reported data from the national surveys suggested that approximately 50 % of all adults met the 2008 Physical Activity Guidelines [
5]. In contrast, the accelerometer-based objectively measured data indicated that only a small proportion of the adults met the guidelines [
5]. A large gap between self-reported and objectively measured physical activity levels exists at population levels [
6].
Accurate measurements of physical activity are important to evaluate the efficacy or effectiveness of interventions designed to increase physical activity levels over time. Self-reported questionnaires are the most commonly used tools to assess changes in physical activity [
5] because they are easy to complete within a short time period and can be administered at relatively low cost. However, a recent systematic review pointed out that self-report measures are susceptible to both overestimation and underestimation of true physical activity levels [
6]. Recall bias, response bias, social desirability, and inability to understand levels of intensity are often considered to be the sources of the inaccuracy [
7,
8]. Understanding the degree of discrepancy in physical activity levels between self-report and objective measures has significant public and scientific implications for designing physical activity intervention studies.
A recent systematic review comparing direct versus self-reported measures for assessing physical activity in adults reported several limitations of the current evidence [
6]. First, the majority of the reviewed papers (148 out of 173 papers) only examined a correlation between the two measures, but did not report the level of agreement and systematic bias [
6]. Second, inconsistency in the number of days measured, measurement time lag, and the unit of physical activity reported between the self-reported and objective measures made it difficult to make a direct comparison of these two measures. Lastly, a small sample size appeared to be an issue. Only 3.4 % and 1.2 % of the papers included a sample size between 100 and 200 and greater than 200 participants, respectively. Therefore, it is important to take into account all of these issues when designing a study to compare self-reported and objectively measured physical activity.
This
mobile phone based
physical activity
education (mPED) study is a randomized controlled clinical trial (RCT) with a run-in procedure and is designed to evaluate the efficacy of a mobile app and accelerometer delivered physical activity intervention for physically inactive women [
9,
10]. The baseline physical activity data measured by the 7-Day Physical Activity Recall Questionnaire (7D-PAR) (self-report) and accelerometer (objective measure) provide a unique opportunity to explore the level of agreement between the two measures by addressing all of the limitations described above. The aims of this paper are: 1) to describe the duration of moderate to vigorous physical activity (MVPA) and the proportion of participants meeting the recommended criterion of at least 150 min of MVPA per week as measured by the 7D-PAR and accelerometer among women who were enrolled in the mPED trial; 2) to assess the level of agreement of the two measures using a Bland-Altman plot; and 3) to describe the positive predictive value (PPV) of meeting the guidelines by calculating the percentage of women meeting the physical activity recommendation by the 7D-PAR who also met this recommendation based on data from the accelerometer. We believe that the findings of this study can assist in designing physical activity measures and interventions for physically inactive women in the near future.
Discussion
This study investigated the agreement between self-report and objective measures of MVPA among 215 women who completed the screening/baseline and randomization visits of the mPED trial. We found large differences in the median weekly total MVPA between the self-reported and objective measures when a 10-min bout of MVPA criteria with was applied. The agreement of the weekly duration of at least 10 min in bout of MVPA between 7D-PAR and accelerometer is poor despite allowing for a 1 or 2-min interruption in accelerometer data. Only 8.4 % of women who met the recommended criterion of 150 min of MVPA per week according to the 7D-PAR also met this criterion by the objective accelerometer-based measure. In contrast, all of the women who reported fewer than 150 min on the 7D-PAR had concordant qualitative accelerometer results. According to a recent systematic review of the comparison of direct versus self-report measures for assessing physical activity, self-report measures of physical activity were generally higher than those measured by accelerometers, in particular female adults [
6]. A reported mean percent difference was 138 %, ranging from 100 % to 4024 % in the female adult only samples.
When at least 10 min in duration of MVPA allowing for a 1 or 2 min interruption was applied, the differences in weekly total minutes of MVPA reported on the 7D-APR and presented on the accelerometers increased, with an increase in the average weekly minutes of MVPA of the two measures (Fig.
2). The 7D-PAR assesses at least 10-min in duration and frequency spent in MVPA using each subject’s relative intensity, while the accelerometer uses absolute intensity across all subjects. This may partially explain the large average difference (+145 min with a 95 % limit of agreement at + 471 to −181 min) between the two measures, since the original aim of the RCT was to enroll physically inactive women who tended to have lower aerobic capacity. Interestingly, when at least a 3-min bout of MVPA criteria was used in the accelerometer data, the average difference between the two measures was the smallest (+59 min with a 95 % limit of agreement at + 386 to −268 min). The trend seen in Fig.
2 disappeared and the variability became more consistent across the graph in Fig.
4.
A clear strength of the accelerometer is that it can capture minute-by-minute activity. In contrast, most self-report measures, including the 7D-PAR, cannot ascertain MVPA with such fine resolution, and are also subject to reporting bias [
27]. For these reasons, the accelerometer is becoming widely accepted as a gold standard measure of free-living physical activity. However, a disadvantage of the accelerometer measure is that no consensus method for summarizing the data has been established [
28]. For example, in some of studies, an interruption of one or two minutes during the 10 min bout of MVPA has been allowed, since in developed countries intentional walking or running may be interrupted by traffic signals or safety concerns [
24]. In addition, the minimum number of days required to calculate weekly average MVPA has varied across studies. In our study, only participants with validated data on all 7 days were included. In addition, we reported both data allowing and not allowing an interruption of 1 or 2 min during the 10-min bout of MVPA. We noted that more women would have met the 150-min criterion based on the accelerometer data if shorter bouts of MVPA had been allowed.
Troiano and colleagues first examined the 2003–2004 National Health and Nutrition Examination Survey (NHANES) which examined accelerometer measured physical activity data in a nationally representative U.S. sample [
5]. They reported that only 2.3 % to 3.2 % female adults met the adherence of the recommendation of 150 min of MVPA per week by accelerometer, while approximately half of the sample met the recommendation by the self-reported measure. Although our sample differed substantially from the nationally representative NHANES sample, the findings of our study were similar: approximately half of our study sample met the recommendation by self-report, but only 8.4 % met it by accelerometer. As we discussed before, this large difference may be explained by over-estimation of the duration and/or intensity of physical activity in self-reports, with light intensity activities reported as moderate intensity, and/or bouts of less than 10 min included in the subjective totals. And while the accelerometer cannot capture activities such as swimming, bicycling, or weight-lifting, potentially resulting in underestimation of MVPA, these activities are relatively uncommon [
5]. As seen in Table
1, the results of our baseline data confirmed this assumption in this sample. Furthermore, the sample NPV of 100 % indicates that if individuals did not meet with the recommendation by the self-reported measures, they were very unlikely to meet the recommendation by the accelerometer. In contrast, the low PPV suggests the need for caution in interpreting self-report in meeting the 150-min recommendation. Overall, the findings of this report highlight on how self-report might function as a proxy for accelerometer measurements for meeting physical activity recommendations.
Strengths and limitations of the study
The strength of this study is that we were able to use the 7D-PAR and accelerometer measured physical activity data over the same time period, and to evaluate the objective measure to be consistent with self-report with a relatively large sample size. Another strength is that participants were not able to view their steps taken and intensity of physical activity during the run-in period. The Omron Active Style Pro HJA-350IT with triaxial accelerometer had a program option to select what types of information to be displayed. We believe that this blind function during the run-in period helped prevent participants from modifying their MVPA based on real-time feedback. Despite these strengths, some limitations need to be taken into account. First, the findings of this study might not be generalizable to men or children, nor to women who are unwilling to participate in an exercise trial like mPED. Second, the accelerometer used in the mPED trial was not able to capture activities such as swimming, bicycling, and weight lifting, in contrast to the 7D-PAR. However, our baseline data suggest that the prevalence of these activities in this sample is relatively low.
Conclusion
The large discrepancy between the self-reported and objective measures of MVPA meeting the 150-min criterion suggests that self-reported physical activity measures should be used with caution in intervention studies. In particular, self-report of at least 150 min of MVPA had a PPV of only 15.4 %, although its NPV was 100 %. Thus while our data suggest that self-report could be used to identify a physically inactive sample, it could grossly over-estimate the proportions of women who become active in one or both arms of trials of interventions designed to increase MVPA.
Abbreviations
MVPA, Moderate to intensive intensity physical activity; BMI, body mass index; CES-D, Center for epidemiological studies depression scale; kg, kilogram; m2, meter squared; mPED, Mobile phone based physical activity education; OR, odds ratio; PA, physical activity; RCT, randomized controlled trial; S/B, screening baseline; SD, standard deviation; SPSS, Statistical Package for Social Science; UCSF, University of California, San Francisco; US, United States; (7D-PAR), 7 Day Physical Activity Recall Questionnaire
Acknowledgements
The authors thank Shoshana Arai, PhD, RN at University of California (UC), San Francisco and Ms. Mo Zhou at UC, Berkeley for data cleaning and analysis assistance.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.