Study design
The current study was conducted as a randomized crossover trial. Participants were randomized to answer the German version of the GPAQ with show cards depicting examples of moderate and vigorous physical activities (“GPAQ+”) or without such show cards (“GPAQ-“) after 7 days. After the next 7 days, the two study groups were reversed according to the randomized crossover design (see Fig. 1). In addition, objective physical activity data of all participants were recorded by accelerometry (ActiGraph GT3X+) over the whole investigation period of 14 days (2 × 7 days).
Participants
A convenience sample was recruited mainly via bulletins at the German Sport University Cologne and different sport clubs in Cologne, Germany. Interested individuals were contacted via email or telephone and given detailed instructions about the study.
Recruitment and data collection were conducted between May and July 2017. Recruitment was closed after more than 50 usable datasets were collected [23].
Eligibility criteria
Inclusion criteria were the participants’ age of at least 18 years, understanding German, and access to the internet on day seven and day 14 to answer the respective version of the GPAQ online. Individuals with acute musculoskeletal diseases or orthopaedic injuries were excluded.
The study was approved by the ethical committee of German Sport University Cologne (reference: 084/2017). Participants received a report on their individual accelerometer measured physical activity after the end of the study. No other incentives for participating in the study were provided. All participants provided written informed consent.
Measures
Global Physical Activity Questionnaire (GPAQ)
The German version of the GPAQ (available at https://www.who.int/ncds/surveillance/steps/GPAQ/en/) was adapted to a self-administration format by rephrasing the first paragraph, which includes sentences only appropriate in an interview [18]. Consistent with the instruction manual of the GPAQ [24], culturally adapted examples of different physical activities were inserted in the respective questions.
For this study, the GPAQ was delivered in two different versions. GPAQ+ included pictures of work or leisure time activities of moderate or vigorous intensity, while GPAQ- included no pictures. In line with the instruction manual calling for cultural adaptation [24], pictures in GPAQ+ were selected to represent typical physical activities of the German population. In order to match the pictures to the respective intensities, the” Compendium of Physical Activities “[25] was used. Activities with 2–6 MET (e.g., working as a mailman, Nordic walking, yoga) were considered moderate intensity, activities with more than 6 MET were considered vigorous intensity (e.g., working as a construction worker, running, basketball). For sedentary behaviour – in line with the instruction manual - no pictures were included in either version. The resulting show cards can be seen in Additional file 1.
For practical reasons, the two versions of the GPAQ were administered via the online survey tool Unipark (Questback GmbH, Cologne, Germany). In this way, transcription errors were avoided and data were automatically available in digital form. A link to the respective GPAQ version was automatically sent by the system to the participants’ individual email-addresses at the morning of day 7 and day 14. Reminder emails were automatically sent if participants did not answer the questionnaire within 2 days after receiving the link.
Actigraph GT3X+
Objective physical activity was recorded over 14 days with Actigraph GT3X+ accelerometers (ActiGraph, Pensacola, Florida, USA). This type of accelerometer has shown valid results for the assessment of physical activity in adults [26,27,28,29].
Participants were instructed to wear the accelerometer on the right-side of their waist for 14 consecutive days, removing them only whilst sleeping or participating in any water-based activity (e.g., showering or swimming). A picture of the application of the accelerometer can be seen in Additional file 2.
Data was collected with a sample rate of 30 Hz and saved in 30-s epochs.
In line with initializing and handing out the accelerometers, demographic variables (sex, age, body mass index) of all participants were recorded.
Statistical analyses
Data preparation
GPAQ data were cleaned according to the GPAQ analysis guide [24]. In a next step, daily averages (minutes/day) were calculated for moderate and vigorous physical activity, respectively. For this purpose, physical activity at work, in leisure time and transportation were combined. This procedure was applied to both GPAQ versions, resulting in daily averages of moderate and vigorous physical activity for each version.
The accelerometer data were processed using the ActiLife software (version 6.10.2, ActiGraph, Pensacola, Florida, USA).
Since the data were recorded as counts per minute (CPM), the following cut-points [30] were applied to classify CPM into intensities of physical activity:
-
0–99 CPM: sedentary behaviour,
-
100–1951 CPM: light intensity,
-
1952–5724 CPM: moderate intensity,
-
> 5724 CPM: vigorous intensity.
Sixty or more minutes of consecutive zero CPM were marked as non-wear time (allowing for an interruption of maximum 2 minutes with 1–100 CPM) [31]. Individuals were included in the analysis if valid data of at least 3 days with a minimum of 10 hours wear time were available for each of the 2 weeks.
In line with the processing of the GPAQ data, daily averages (min/day) of valid days were calculated for moderate and vigorous physical activity (minutes in each activity category divided by the number of recorded days). Data of week one and two were handled separately.
Data analyses
All statistical analyses were conducted using IBM SPSS Statistics 24.
Descriptive statistics (means, standard deviations, frequencies and percentages) were used to describe demographic characteristics and the data from questionnaires and accelerometers. Chi-squared tests and Mann-Whitney U tests were used to examine statistically significant differences between groups at baseline and between the included and excluded sample.
Statistical significant differences regarding the validity of the two GPAQ versions were examined using Wilcoxon signed-rank tests. For this purpose, the differences of each GPAQ version to the accelerometer data were compared (i.e., GPAQ+ data minus accelerometer data, and GPAQ- data minus accelerometer data).
Spearman’s rank correlation coefficients were calculated for the correlation between the daily averages of the accelerometer data and the data from each GPAQ version. The resulting coefficients were interpreted as no correlation (rho = 0–0.09), poor (rho = 0.10–0.29), fair (rho = 0.30–0.59), moderately strong (rho = 0.6–0.79), very strong (rho = 0.8–0.99), and perfect correlation (rho = 1) [32].
In addition, Bland-Altman plots were built to illustrate the agreement of accelerometer and GPAQ data regarding moderate and vigorous physical activity. In a Bland-Altman plot, the difference of two measures (y-axis) is plotted against the data of the accelerometer (x-axis) for each participant [33, 34]. The limits of agreement were set at mean difference ± 1.96 standard deviations. Plots were designed for moderate and vigorous physical activity of both GPAQ versions individually.
The significance level for all analyses was set at p < 0.05.