Reliability and validity of the French version of the global physical activity questionnaire.

BACKGROUND: The Global Physical Activity Questionnaire (GPAQ) has been used to measure physical activity (PA) and sedentary time in France, but no study has assessed its psychometric properties. This study aimed to compare the reliability as well as criterion and concurrent validity of the French version of the GPAQ with the French International Physical Activity Questionnaire long form (IPAQ-LF) and use of an accelerometer in a general adult population.
METHODS: We included 92 participants (students or staff) from the Medicine Campus at the University of Lorraine, Nancy (north-eastern France). The French GPAQ was completed twice, 7 days apart, to study test-retest reliability. The IPAQ-LF was used to assess concurrent validity of the GPAQ, and participants wore an accelerometer (ActiGraph GT3X+) for 7 days to study criterion validity. Reliability as well as concurrent and criterion validity of the GPAQ were tested by the intraclass correlation coefficient (ICC), Spearman correlation coefficient for quantitative variables, and Kappa and Phi coefficients for qualitative variables. Both concurrent and criterion validity of GPAQ were assessed by Bland-Altman plots.
RESULTS: The GPAQ showed poor to good reliability (ICC = 0.37-0.94; Kappa = 0.50-0.62) and concurrent validity (Spearman r = 0.41-0.86), but only poor criterion validity (Spearman r = -0.22-0.42). Limits of agreement for the GPAQ and accelerometer were wide, with differences between 286.5 min/week and 601.3 min/week.
CONCLUSION: The French version of the GPAQ provides limited but acceptable reliability and validity for the measurement of PA and sedentary time. It may be used for assessing PA and sedentary time in a French adult population.

Introduction

Physical activity (PA) surveillance is a public health preoccupation and is considered by the World Health Organisation (WHO) as a protective factor for non-communicable diseases. A high PA level is associated with reduced mortality and the occurrence of diseases or their consequences and improved quality of life.2, 3 Because of its therapeutic role, PA is also used as adjuvant treatment in chronic diseases.4, 5

In this context, the measurement of PA is essential to assess strategies promoting PA and to survey and compare PA levels between countries. Questionnaires are the most commonly used instrument in epidemiologic studies to assess PA because they are relatively inexpensive and easy to use both for a large population and in a short time. They can be self-administered, completed during an interview or administered by phone. Many different questionnaires have been developed and used to measure PA, so international comparison is difficult, and overall, their development lacked methodological quality.

In the late 1990s, the International Physical Activity Questionnaire (IPAQ) was developed in 2 forms (short form (IPAQ-SF) and long form (IPAQ-LF)) to create national and international comparable and standardized measures of PA. The long form of the IPAQ (31 items) was developed to capture information about domains of PA but has been considered too long and too complex to be used in surveillance studies, while the short form (9 items) does not take into account the domains of PA.7, 8 For PA surveillance, the measurement of PA domains is needed to understand the patterns of PA and to develop interventions. Thus, in order to provide an instrument that would address the limits of these questionnaires, the Global Physical Activity Questionnaire (GPAQ) has been developed by the WHO, as part of the WHO STEPwise approach to survey chronic disease risk factors. It is now recommended by the WHO for national surveillance of PA. Since its development, the GPAQ has been translated into and tested in many languages and is used in many countries.9, 10, 11, 12, 13, 14, 15, 16 In France, the GPAQ has been used to describe and analyse PA and sedentary time of the general population. However, it has not been validated in the French language. Evidence for the validity and reliability of the French version of the GPAQ is needed because the results may be affected by the sociocultural specificities of the country.

Rigorous methodology is needed to examine the degree in which an instrument is affected by measurement error (reliability) and measures the construct it intends to measure (validity). Concurrent validity refers to the degree to which the GPAQ measures what it purports to measure, and criterion validity is the degree to which the results of the questionnaire are an adequate reflection of a “gold standard”. Because of no satisfying available gold standard measurement for PA behavior, objective measures such as accelerometers and pedometers are commonly used. To appraise the concurrent validity of the GPAQ, a questionnaire measuring the same construct and with similar structure is considered relevant. Even if the IPAQ-LF is more detailed than the GPAQ, it is the most similar in its construct and its structure. For this reason, the IPAQ-LF has been considered relevant to examine the concurrent validity of the GPAQ.

This study aimed to assess the test–retest reliability as well as criterion and concurrent validity of the French version of the GPAQ by comparison with the IPAQ-LF and use of an accelerometer in a general adult population in France.

Methods

Patients and study design

A convenient sample was recruited from January 20, 2015 to April 20, 2015, from the Medicine Campus, University of Lorraine, Nancy (north-eastern France), by posting an advertisement on campus and by e-mailing students and staff. Participants had to be ≥18 years old, working or studying at the Medicine Campus, able to read and understand French, and willing to participate in the study. The study protocol was approved by the Legal representative of the French data protection authority (Commission Nationale Informatique et Libertés) of the University of Lorraine, France. All participants were asked to read and sign a consent form. A ratio of 5 subjects per item was used to determine the number of participants to include. Because the GPAQ contained 16 items, a minimum number of 80 participants was required.

Each subject was invited to participate in a face-to-face interview on Day 0 (D0) and received all explanations about the study and its purpose from an interviewer. After giving consent, participants answered sociodemographic and anthropometric questions, then completed the GPAQ and IPAQ-LF. Then, the interviewer gave the participant an accelerometer and explained its use. Participants were asked to wear the accelerometer for 7 consecutive days. Eight days after the first interview (D8), participants returned the accelerometer and completed the GPAQ and IPAQ-LF a second time. They were also asked if they had changed their activity during the week of the study as compared to a typical week.

Instruments

We used the French translation of the GPAQ (Version 2.0) to gather information on the time spent in moderate and vigorous PA and in sedentary behavior. At the WHO level, the GPAQ has been translated in French by a professional translator, and back-translated by 2 independent technical experts. The versions were then compared, and where discrepancies existed, these were discussed and a consensus was found. The GPAQ contains 16 items designed to assess the frequency and duration of PA in 3 domains: during work, transportation, and leisure time as well as time spent sitting during a typical week. It distinguishes PA duration by min/day and min/week for each PA domain, which allows for calculating the energy expenditure scored in metabolic equivalent tasks (METs). One MET corresponds to resting energy expenditure. According to duration and energy expenditure, PA level was classified as low, moderate, and high.

The French IPAQ-LF was used to test the concurrent validity of the GPAQ. It contains 27 items designed to assess the frequency and duration of PA in 4 domains: during work, transportation, household activities, and leisure time, then time spent sitting. The IPAQ-LF scores PA in terms of energy expenditure (MET), intensity (low, moderate, high, and sedentary), and duration (min/day, min/week).

The ActiGraph accelerometer, model GT3X+ (ActiGraph, Pensacola, FL, USA), was used as the criterion measure. The device is worn at the waist and measures and records the changes in acceleration and deceleration movements in 3 axes (antero-posterior, superio-inferior, and medial side). Data for measuring acceleration and deceleration are stored in non-volatile flash memory and can be read by using ActiLife software. Accelerometer data were scored using ActiLife 6 Data Analysis Software (ActiGraph) to assess time spent at various PA intensity levels (moderate and vigorous in min/day). Freedson's Adult VM3 (2011) cut-off points were used to determine several PA levels: light: 0–2690 counts per minute (cpm); moderate: 2691–6166 cpm; vigorous: 6167–9642 cpm; and very vigorous: 9643–∞ cpm. Minutes spent at each intensity level were averaged across valid days. Non-wear periods were identified as 60 consecutive minutes with no movement data (0 counts). All calculations were based on 60 s epochs; an epoch is a user-defined time-sampling interval used to filter the acceleration signal. In this study, we used 7-day PA questionnaires, so only data with ≥10 h of wear time per day for ≥7 days were considered valid and included in the analysis.24, 25

Sociodemographic data such as age, sex, and education (high school or higher education) and socioprofessional status (student or staff) were collected. Anthropometric data including height (in m) and weight (in kg) were reported by each participant for calculating body mass index (BMI, kg/m2), then participants were classified by BMI level: underweight (BMI <18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight 25.0–29.9 kg/m2), and obese (≥30 kg/m2). All data (except accelerometer data directly transferred into ActiLife software) were entered into an electronic case report form (CRF) created with Epidata 3.1 (The EpiData Association, Odense, Denmark).

Statistical analysis

Data analysis involved use of SAS Version 9.4 (SAS Inst., Cary, NC, USA). Qualitative variables were reported as relative frequency and quantitative variables as mean ± SD or median. The Kolmogorov-Smirnov test was used to assess the normality of data distribution. For participants who declared changing their PA, paired Student's t test was used to evaluate the difference in total PA between the 2 visits. Because the activity measured by the GPAQ includes work and household activities, it was compared to the sum of work and household PA measured by the IPAQ-LF.

Test–retest reliability was tested by the Kappa coefficient for categorical data and the intraclass correlation coefficient (ICC) for quantitative data. Spearman correlation was also calculated for quantitative data to compare with previous studies.11, 12, 13, 14 Non parametric correlation coefficient was used because of non-Gaussian distribution for most of PA-score. For one of the GPAQ's question, one answer modality was overrepresented and the correlation was not concordant with the observed agreement (when visualizing the data, the agreement seems good but it was not observed when assessed with ICC and Spearman correlation). Thus the variable was converted into a discrete variable, and the prevalence-adjusted and bias-adjusted Kappa (PABAK) was used to assess the agreement. Concurrent validity was examined by comparing data for the GPAQ and IPAQ-LF at D0 and D8 with the Spearman correlation coefficient and its 95% confidence interval (CI) for quantitative data and the Phi coefficient for qualitative data. Criterion validity was examined by comparing minutes of PA obtained with the GPAQ to accelerometer-obtained data at D8 by the Spearman correlation coefficient and its 95%CI.

Both the concurrent and criterion validity of the GPAQ were assessed by Bland-Altman plots to measure the agreement and bias for total PA and sedentary time between questionnaire's answers and results from accelerometer. Correlation assesses the degree to which 2 variables are related. However, a high correlation does not necessary imply that there is good agreement between the 2 methods. Thus, Bland-Altman was used to quantify the agreement between 2 measurements by plotting the difference between the 2 measurements against the average obtained with each of the 2 methods.

Kappa and Phi coefficients were classified by the ratings suggested by Landis and Koch: poor: <0.00; slight: 0.00–0.20; fair: 0.21–0.40; moderate: 0.41–0.60; substantial: 0.61–0.80; and almost perfect: 0.81–1.00. ICC and Spearman correlation <0.50 were considered as poor, 0.50–0.75 were considered as moderate, and >0.75 were as good.

Results

Participant characteristics

In total, 92 subjects participated in the study (age 30.1 ± 10.7 years, range 19–58 years; 67 (72.8%) females); 56.5% were students, 95.6% had higher education, 9.8% had chronic disease, and 76.9% had normal BMI (Table 1). Overall, 25% of participants declared having changed their activity between the 2 visits, but the difference between the total PA means measured by the GPAQ was not statistically significant (p = 0.49).

Table 1

Sociodemographic and anthropometric characteristics of participants (n = 92).

	Total sample (%)
Sex
Male	25 (27.2)
Female	67 (72.8)
Socio-professional status
Student	52 (56.5)
Staff	40 (43.5)
Education level
High school	4 (4.4)
Higher education	88 (95.6)
Age (year)a	30.1 ± 10.7
BMI (kg/m2)a	22.6 ± 3.5
BMI classes (kg/m2)a
Underweight <18.5	3 (3.3)
Acceptable weight 18.5–24.9	71 (76.9)
Overweight 25.0–29.9	14 (15.4)
Obese ≥30	4 (4.4)

Abbreviation: BMI = body mass index.

Data are presented as mean ± SD.

Descriptive statistics for the GPAQ, IPAQ, and accelerometer

All descriptive statistics for GPAQ, IPAQ, and accelerometer are presented in Table 2.

Table 2

Data for PA measured by GPAQ, IPAQ, and an accelerometer at Day 0 (D0) and Day 8 (D8) in 92 participants.

Variable	GPAQ				IPAQ				Accelerometer
	D0		D8		D0		D8
	Mean ± SD	Median	Mean ± SD	Median	Mean ± SD	Median	Mean ± SD	Median	Mean ± SD	Median
Total PA (MET min/week)	2011.1 ± 1940.5	1580.0	1818.0 ± 1478.2	40.7	2648.3 ± 2099.8	2251.5	2484.1 ± 2268.0	1777.5
PA by domain
Work
Vigorous	31.3 ± 300.3	0	33.0 ± 230.6	0	34.8 ± 300.8	0	15.6 ± 85.7	0
Moderate	467.4 ± 1575.3	0	321.1 ± 965.4	0	203.5 ± 758.1	0	212.4 ± 871.2	0
Transport	375.9 ± 410.8	240.0	378.5 ± 426.2	250.0	306.8 ± 295.5	242.5	351.3 ± 414.0	260.7
Household
Vigorous	n/a	n/a	n/a	n/a	4.5 ± 35.4	0	22.1 ± 126.1	0
Moderate	n/a	n/a	n/a	n/a	475.9 ± 785.2	150.0	356.7 ± 594.3	160.0
Work + household
Vigorous	n/a	n/a	n/a	n/a	39.3 ± 302.4	0	37.8 ± 171.8	0
Moderate	n/a	n/a	n/a	n/a	695.0 ± 1080.1	240.0	596.1 ± 1189.1	190.0
Leisure
Vigorous	852.2 ± 1073.3	680.0	772.6 ± 955.9	480.0	868.7 ± 1085.9	600.0	691.3 ± 1011.5	0
Moderate	284.3 ± 366.0	240.0	312.8 ± 382.9	240.0	193.9 ± 265.4	0	218.9 ± 415.4	340.0
Sitting time (min/day)	570.0 ± 152.8	600.0	588.6 ± 146.4	600.0	554.5 ± 138.5	584.3	583.6 ± 143.2	597.1	843.6 ± 134.5	814.0
PA duration by intensity (min/week)
Vigorous	883.5 ± 1090.1	720.0	805.6 ± 977.7	480.0	903.5 ± 1102.4	720.0	707.0 ± 1015.6	360.0	72.0 ± 67.2	46.7
Moderate	751.7 ± 1659.8	360.0	633.9 ± 990.3	360.0	903.8 ± 1131.4	480.0	860.6 ± 1266.9	370.0	426.2 ± 139.5	429.4
PA level (%)
Low	29.4		22.8		8.7		15.2
Moderate	44.6		45.6		60.9		54.3
High	26.1		22.8		30.4		30.4

Abbreviations: GPAQ = Global Physical Activity Questionnaire; IPAQ = International Physical Activity Questionnaire; MET = metabolic equivalent task; n/a = not assessed by the questionnaire; PA = physical activity.

Test–retest reliability

The ICCs ranged from 0.37 to 0.94, with the highest ICC for vigorous leisure PA. Only total vigorous and vigorous leisure PA showed good reliability, whereas all other PA scores were poor to moderate, with the lowest value for moderate leisure PA (ICC = 0.37, 95%CI: 0.15–0.56). A good reliability for total sitting time was also observed (ICC = 0.80, 95%CI: 0.69–0.87) whereas it was moderate for total PA (ICC = 0.58, 95%CI: 0.40–0.72). For PA level, the Kappa coefficient showed moderate to substantial correlation, varying from 0.50 to 0.62 for moderate and low PA levels, respectively. For vigorous activity at work, the GPAQ showed an almost perfect reliability (PABAK = 0.91). Except for total PA, with ICC = 0.58, 95%CI: 0.40–0.72 and Spearman's r = 0.82, 95%CI: 0.72–0.88, most Spearman values were similar to the ICC (Table 3).

Table 3

Test–retest reliability of the GPAQ (n = 68).

Variables	ICC(95%CI)	Spearman's Rho(95%CI)	Kappa coefficient
Total PA	0.58 (0.40–0.72)	0.82 (0.72–0.88)
PA by domain
Work
Vigorous			0.91(+)
Moderate	0.48 (0.28–0.64)	0.52 (0.33–0.68)
Transport	0.67 (0.52–0.79)	0.69 (0.53–0.79)
Leisure
Vigorous	0.94 (0.91–0.96)	0.89 (0.84–0.94)
Moderate	0.37 (0.15–0.56)	0.53 (0.33–0.68)
Sitting time	0.80 (0.69–0.87)	0.78 (0.67–0.86)
PA by intensity
Total vigorous	0.84 (0.76–0.90)	0.80 (0.70–0.88)
Total moderate	0.48 (0.28–0.65)	0.56 (0.38–0.71)
PA level
Low			0.62
Moderate			0.50
High			0.57

(+): Adjusted Kappa (PABAK).

Abbreviations: CI =  confidence interval, GPAQ = Global Physical Activity Questionnaire; ICC = intraclass correlation coefficient; PA = physical activity.

Concurrent validity

For both measurement times, we observed good correlations between the GPAQ and IPAQ for vigorous activity during leisure, total vigorous activity, and sitting time (r = 0.76–0.89) (Table 4). The values at D0 and D8 seemed almost identical, but important discrepancies were observed between vigorous work at D0 (r = 0.58, 95%CI: 0.43–0.70) and at D8 (r = 0.81, 95%CI: 0.73–0.87). Overall, total PA showed moderate correlation at both D0 (r = 0.66, 95%CI: 0.53–0.76) and D8 (r = 0.67, 95%CI: 0.54–0.77). Results of Bland-Altman analysis (Fig. 1A, C) for the GPAQ and IPAQ demonstrated a mean difference of 637.2 ± 1641.5 MET min/week. The limits of agreement for the 2 instruments were wide, with the difference between 1004.3 MET min/week and 2580.1 MET min/week. For sedentary time, the mean difference of sedentary time was −15.5 ± 79.2 min/day. Overall, the classification by level of PA with the 2 questionnaires, at both times, was only poorly to moderately correlated, with a Phi coefficient ranged from 0.22 to 0.57 (Table 4).

Table 4

Concurrent validity between the GPAQ and IPAQ-LF data at Day 0 (D0) and Day 8 (D8) (n = 92).

Variable	D0		D8
	Spearman's Rho (95%CI)	Phi coefficient	Spearman's Rho (95%CI)	Phi coefficient
Total PA	0.66 (0.53–0.76)		0.67 ((0.54–0.77)
PA by domain
Work
Vigorous	0.58 (0.43–0.70)		0.81 (0.73–0.87)
Moderate	0.56 (0.40–0.68)		0.61 (0.46–0.72)
Transport	0.52 (0.35–0.65)		0.69 (0.57–0.79)
Leisure
Vigorous	0.86 (0.79–0.90)		0.79 (0.70–0.85)
Moderate	0.46 (0.28–0.61)		0.53 (0.36–0.66)
Sitting time	0.85 (0.78–0.90)		0.89 (0.84–0.93)
PA by intensity
Total vigorous	0.86 (0.79–0.90)		0.76 (0.66–0.84)
Total moderate	0.41 (0.22–0.56)		0.58 (0.42–0.70)
PA level
Low		0.22		0.49
Moderate		0.27		0.27
High		0.57		0.54

Abbreviations: GPAQ = Global Physical Activity Auestionnaire; IPAQ-LF = International Physical Activity Questionnaire–Long Form; PA = physical activity.

Fig. 1

Bland-Altman plots of the validity of the Global Physical Activity Questionnaire (GPAQ). A and C: Agreement of GPAQ with IPAQ for total PA and sitting time at D0; B and D: Agreement of GPAQ with accelerometer for total PA and sitting time at D8. IPAQ = International Physical Activity Questionnaire; PA = physical activity.

Criterion validity

Accelerometer data were considered valid for 87 of the 92 participants (5 participants did not wear an accelerometer for at least 10 h per day over 7 days). Criterion validity was assessed by comparing total PA time spent in vigorous-intensity activity, or in moderate-intensity activity, or sitting per day reported with the GPAQ and derived from accelerometer counts. Poor but significant correlations for sedentary time (r = 0.42, p < 0.01) and total vigorous PA (r = 0.38, p < 0.01) were observed (Table 5).

Table 5

Criterion validity of the GPAQ: Spearman's correlation between the GPAQ and accelerometer data at day 8 (D8) (n = 87).

GPAQ	Accelerometer
	Average sedentary counts/day	Average moderate counts/day	Average vigorous counts/day
Total vigorous PA (min)	0.02	0.19	0.38**
Total moderate PA (min)	−0.20	0.10	−0.10
Total PA across all domains (min)	−0.20	0.40**	0.24*
Time spent sitting (min)	0.42**	−0.22*	0.30**

Abbreviations: GPAQ = Global Physical Activity Questionnaire; PA = physical activity.

p < 0.05, **p < 0.01, compared GPAQ with accelerometer's values.

Bland-Altman findings revealed that the GPAQ underreported total PA, with a mean difference between the GPAQ and accelerometer data of 443.95 ± 157.46 min/week (Fig. 1B, D). Limits of agreement for the 2 instruments were wide, with the difference between 286.5 min/week and 601.3 min/week GPAQ underestimated sedentary time as compared with the accelerometer, with a mean difference between the 2 instruments of 251.2 ± 161.1 min/day. Limits of agreement for the 2 instruments ranged from 90.1 min/day to 412.3 min/day.

Discussion

This study provides results, for the first time in a French population, for the reliability and validity of the GPAQ.

For reliability, we found poor to good correlation, with highest value obtained for vigorous leisure PA, which indicates the stability of this type of PA. This result is consistent with the findings by Matthews et al. who observed no significant variation in vigorous leisure time activity over 1 year in 580 healthy adults. Overall, our results are comparable to other studies testing the psychometric properties of the GPAQ. Herrmann et al. demonstrated short- and long-term reliability with ICC values from 0.54 to 0.92. Bull et al. reported test–retest correlation coefficients from 0.67 to 0.81 and Kappa coefficients from 0.67 to 0.73 for pooled data.

Whereas Bull et al. and Herrmann et al. showed a poor to moderate correlation between the GPAQ and IPAQ (with coefficients 0.45–0.57 and 0.26–0.63, respectively), our results indicate a poor to good concurrent validity. A reason of this difference could be the use, by the former studies, of the IPAQ short-form (IPAQ-SF) as compared with our use of the long form. Unlike the GPAQ and IPAQ-LF, which measure PA in different domains, the IPAQ-SF measures overall PA duration and frequency, which may explain the differences. In measuring the concurrent validity of the GPAQ, the IPAQ-LF may be more relevant than the IPAQ-SF. However, despite an acceptable concurrent validity, the agreement between the GPAQ and the IPAQ-LF to classify participants by PA levels was only poor to moderate (Phi coefficients: 0.22–0.57), with the highest agreement attributable to high PA level. In addition, the Bland-Altman analysis revealed wide discrepancies in total PA measured by the 2 questionnaires, with a mean difference of 637.2 ± 1641.5 MET min/week. A possible explanation could be that the IPAQ-LF contains detailed items dedicated to household activities, whereas in the GPAQ, household activities are included in work activities. Also, the IPAQ-LF measures time spent walking, which is not considered by the GPAQ if it is not brisk walking (considered moderate activity). These differences may explain the gap in total PA measured by the 2 questionnaires. These results indicate the difficulty in comparing different questionnaires and thus the need to use the same questionnaire in a population surveillance study to be able to interpret the pattern of PA over the years.

A poor criterion-related validity for the GPAQ as compared with accelerometer data was shown. These results are comparable to Cleland et al. and Bull et al., who demonstrated correlations with accelerometer data ranging from 0.19 to 0.48 and −0.20 to 0.40, respectively, whereas results from Hoos et al. showed correlations from 0.32 to 0.52. According to Bland-Altman analysis, the GPAQ seems to underestimate total PA as compared with the accelerometer. This finding can be explained by the GPAQ including only PA that lasts at least 10 min, whereas the accelerometer measures all activities regardless of duration. This result was already found in studies comparing questionnaires to objective measures of PA. In this study and according to Bland-Altman analysis, the GPAQ seemed to underestimate sedentary time as measured by the accelerometer. This finding can be justified most likely by difficulty to accurately recall sitting time as well as by a response bias due to social desirability, which may affect the degree of reporting the time spent sitting by subjects. Future research is needed to identify whether a bias does exist and if so, whether it differs by gender or socioprofessional status, and to what extent.

This study had several strengths, beginning with the adherence to standardized WHO protocols in administering questionnaires (GPAQ was always administered before the IPAQ) and the concordant measurement period (the same 7 days) for both questionnaires and the accelerometer. Also, we used Bland-Altman analysis, a useful and recommended approach to assess the level of agreement, as compared with usual correlation coefficients assessing only the strength of the relationship between the measures. Finally, the use of the IPAQ-LF seems relevant because it induced better concurrent validity with the GPAQ than in previous studies.

The major limitation of this study was the use of accelerometer as an alternative to the gold standard. However, in the absence of a gold standard, accelerometer may be used to measure PA in daily life.32, 33

Conclusion

This study adds important and new information in testing the psychometric properties of the GPAQ in France. The results suggest that the GPAQ is a reliable questionnaire for use in the French population. The overall validity was poor to good but remained acceptable and was similar to previous studies.11, 12 Another important highlight is the need to use the same questionnaire in surveillance studies to allow for comparison and follow-up of the PA level of the study population and for PA surveillance in general.

Authors' contributions

FZW participated in the design of the study, contributed to data collection and data reduction/analysis; AV, ES participated in the design of the study; FR participated in the design of the study and contributed to data collection; MLE contributed to data reduction/analysis; HE contributed to data analysis and interpretation of results. All authors contributed to the manuscript writing. All authors have read and approved the final version of the manuscript, and agree with the order of presentation of the authors.

Competing interests

The authors declare that they have no competing interests.