Regular physical activity has been widely mentioned as contributing to several health benefits in all age ranges, namely for mental health [1], bone health [2], diabetes [3], cardiovascular disease [4] and obesity [5]. Physical activity (PA) assessment is therefore essential in surveillance, screening, programme evaluation and intervention studies. In order to obtain valid and reliable measures of PA, objective and improved methods of evaluation are recommended [6, 7], such as those from accelerometers, although they have a high cost and frequently are unavailable. In children and adolescents the difficulties of use have been referred [8-10]. And considering the recommended protocols, particularly in larger sample studies, questionnaires have been used as an alternative [11-13].

The International Physical Activity Questionnaire (IPAQ) is the more widely used and accepted questionnaire; it has proven to be valid and reliable for adults, in several countries and in different formats: long version, short version, self-reported and telephone interview [14].

Researchers detected the need to validate a PA questionnaire for adolescents, because the type and duration of activities are unique for this age group [15], either for intensity, frequency, type and location of practice. Schools, commuting and organized physical activity have major importance for this age group. Also the adolescents’ reasons for being physically active differ from those of children and adults and elderly. For adolescents body shape, weight management, new social networks and support from family and peers are significant motivations. Even so, adolescents’ PA is more similar to that of adults than of children [15]. Hence, adaptations and validations of IPAQ for Adolescents (IPAQA) have been published in some countries [16-19]. Although a validation of IPAQ for Portuguese adults has been done [20], the Adolescents version, IPAQA, is not yet validated for Portuguese adolescents.

The present work aims to determine the validity of the Portuguese version of the IPAQA using GT3X+ Actigraph accelerometers.

Concurrent validity is a form of criterion validity, that uses a correlation with a criterion, ideally a gold standard, that is administered simultaneously [22]. In this study, the accelerometers were used as criterion and the IPAQA completed regarding the same time period.

The Actigraph GT3X+ considers acceleration in 3 axes: vertical, medio-lateral and antero-posterior. This device measures and records time varying accelerations ranging in magnitude from +/- 6g, which are subsequently digitized by a 12-bit analogue-to-digital converter at a rate between 30 and 100 Hz. The raw data are stored for subsequent analysis.

The accelerometer was used with an elastic belt in the waistline, positioned in the anterior axillary line of the non-dominant side. The device was initialized, selecting the 3-second epoch. Adolescents received both verbal and written information on how to use the accelerometer, based on previous researches [7]; instructions on use were given during 7 consecutive days, starting immediately after waking up and until going to bed, except for water activities (bath and swimming).

After usage, accelerometer data were processed using Actilife (version 6.9, Actigraph, Florida). The data was reduced to one-minute periods (epochs) and wear and non-wear time was determined according to previous recommendations [23]. Time periods with at least 90 consecutive minutes of zero counts recorded were excluded from analysis assuming that the monitor was not worn. A minimum recording of 8-hours/day (480-minutes/day) was the criteria to accept daily PA data as valid, as previously suggested [24, 25]. Participants were required to have a minimum of 3 days recorded data, in accordance with previous researchers [25, 26].

The outcome variables were time (min/day) spent in each of the following categories: sedentary PA (0-100 Counts/min); light activity (101-2295 Counts/min); moderate activity (2296-4011 Counts/min) or vigorous activity (≥ 4012 Counts/min), according to Evenson cut-points [27], as previously suggested for studies with adolescents [28]. Moderate-to-vigorous PA (MVPA) was defined as the sum of Moderate PA and Vigorous PA.

Descriptive statistics [mean, standard deviation, inter-quartile range (IQR) and frequencies] were used to describe the sample characteristics and the main outcomes of IPAQA and accelerometer.

After performing Kolmogorov-Smirnov normality tests, with Lilliefors significance correction, Mann-Whitney U test was used to compare differences between boys and girls, and between the two age groups: early adolescents, from 10 to 14 years, and the late adolescents from 15 to 19 years [29]. Spearman correlation coefficient was performed to assess the relation between time spent at each intensity level and total PA obtained by IPAQA and the accelerometer [30].

Finally, tertiles of IPAQA and accelerometer total PA and moderate to vigorous PA were calculated and the agreement between the two measurements was tested using Cohen’s Kappa [30].

Data analysis was performed using IBM Statistical Package for Social Sciences, version 22 (SPSS Inc; Chicago, IL, USA). Statistical significance was set at P<0.05.

The 222 adolescents included mainly females (123; 55.4%) and adolescents above 15 years of age (147; 66.2%). The mean BMI was 22.2 (SD=3.47) kg/m², with 77.0% having normal weight for height, age and gender [21]. There were no statistically significant differences in the younger age group (≤ 14 years) between boys and girls, regarding height, weight, BMI, and waist circumference Table 1. In older age group (15 years or older), boys had a significantly higher weight (Z=-3.313; P<0.001) and height (Z=-7.012; P<0.001), but no significant differences were found for BMI Table 2.

According to IPAQA, adolescents engaged on average 138.2 minutes (SD=106.95) in total PA daily. The total daily PA was divided up into 42.4 minutes (SD=48.61) moderate PA, 34.4 minutes (SD=36.07) vigorous PA, and in 61.4 minutes (SD=54.26) walking. No statistical differences between boys and girls were found in the reported PA for adolescents aged 14 years or under Table 1, and for the older adolescents differences were significant (Z=-2.796; P<0,01) only for the time reported in vigorous PA, with boys reporting more time than girls Table 2.

The accelerometer was used on average for 809.4 (SD=79.63) minutes/ day; sedentary, vigorous, moderate and light activity represented on average 506.9 (SD=117.61), 12.9 (SD=12.01), 35.1 (SD=19.27) and 277.7 (SD=72.40) minutes per day, respectively. In both age groups, boys spent significantly more time in vigorous and moderate PA than girls Tables (1 and 2).

Considering accelerometer measured time, younger boys (Z=-4.450; P<0.001) and girls (Z=-4.244; P<0.001) spent significantly more time in moderate PA than older boys and girls. The same was observed for MVPA: younger boys (Z=-2.733; P<0.01) and younger girls (Z=-3.054; P<0.01) spent more time in that intensity when compared to the older adolescents.

Spearman correlations between reported PA time (IPAQA) and measured time (accelerometer) revealed that, for the total sample Table 3: the total time reported with IPAQA had no significant correlation to the total time assessed with accelerometer; the minutes per day reported in vigorous PA according to IPAQA was poorly correlated to the time measured with accelerometer in the same intensity level (ρ=0.214; P<0.01), and moderately correlated to the time spent in moderate PA (ρ=0.338; P<0.01); reported and measured time in MVPA had a poor correlation (ρ=0.250; P<0.01); and the total PA (MET/day) according to IPAQA was significantly correlated (ρ=0.237; P<0.01) to the total PA (counts/min) according to accelerometer, however this was a poor correlation.

Younger adolescent girls Table 4 show moderate correlations between measured MVPA time and walking time reported with IPAQA (ρ=0.407; P<0.05). In this group, walking time had also a moderate correlation with total PA counts from the accelerometer (ρ=0.412; P<0.05).

There were no meaningful correlations in the younger boys group Table 4.

For adolescent girls 15 years and older, the total PA (steps/day) measured with accelerometer was significantly correlated with walking time (ρ=0.323; P<0.01), with vigorous PA (ρ=0.286; P<0.01), with MVPA (ρ=0.251; P<0.05) and with IPAQA total PA (ρ=0.286; P<0.01) Table 4.

In the older adolescent boys group, the minutes per day reported in vigorous PA according to IPAQA were significantly correlated to the time measured with accelerometer on the same intensity level (ρ=0.428; P<0.01), and in the moderate intensity (ρ=0.363; P<0.01); when considering MVPA, the time per day according to IPAQA was significantly correlated (ρ=0.428; P<0.01) to the one measured with accelerometer; the total PA (MET.min.day^-1) according to IPAQA was significantly correlated (ρ=0.372; P<0.01) to the total PA (counts/min) according to accelerometer, though this was a poor correlation; there was no significant correlation between moderate PA level obtained by the two methods, however reported moderate PA was correlated with measured vigorous PA (ρ=0.321; P<0.05) Table 4.

The tertiles agreement Table 5 between the accelerometer and the questionnaire for total PA was 42.3% (K=0.135; P<0.01); and for MVPA tertiles there was a 39.6% agreement (K=0.094; P<0.05).

The authors and the translation experts felt no need to change the content and structure of the questionnaire, finding it suitable for the age range and culture.

As with previous research on PA questionnaire validation, the accelerometers were used for concurrent validity [16, 17, 19, 31, 32].

This study has the advantage of using a GT3X+ model, which considers acceleration in 3 axes. By doing so, it guarantees the measurement of PA in the three dimensions of space, making it more accurate for PA measurement in free living conditions [33].

Data from the present study corroborate previous epidemiological studies [34, 35], where boys were significantly more active than girls and younger adolescents engaged significantly more time in MVPA than older adolescents, although IPAQA data failed to show these differences between sexes.

In this study no results are presented regarding the comparison between total time reported by IPAQA and measured by the accelerometer, as authors believe that the measures are not directly equivalent. One minute of IPAQA reported PA is not directly equivalent to one minute of measured PA. In fact this same conclusion was pointed out in studies with the adults version of IPAQ [36]. Additionally, the IQR also reveals that while IPAQA and accelerometer give information on physical activity, expressed in the same units, the two methods do have different scales. The present data are also in accordance with previous research that indicated that questionnaires overestimate time [32, 36], particularly in the higher intensity levels [19]. Indeed, in this sample a systematic error of over-reporting seems to exist. However, a previous attempt to validate a Swedish version of IPAQA pointed out the amount of unreported time as the explanation for the non-validation of the questionnaire [37]. Thus, considering the systematic error on the activities duration, IPAQA should not be used to evaluate compliance with the guidelines, expressed as minutes per day in a specific intensity level.

Similarly to this study, previous attempts to validate PA reporting methods for children and adolescents have consistently mentioned weak associations with accelerometer findings, especially if considering recall methods, and for light and moderate PA levels the associations tend to be particularly weak or inexistent [16, 17, 19, 32].

Sports participation, consistent with MVPA intensity, seems to be consistent over time (every week on the same day, with the same duration), making it easier to report and estimate duration, which may explain the significant correlation coefficient for physical activity in this intensity level, reinforcing the results from previous studies [16, 17, 19]. Considering that boys and older adolescents engage more in sports activities, while younger adolescents have more moments of spontaneous movement and PA, this has been mentioned [16] as a possible explanation for IPAQA having moderate correlation coefficients in the older adolescent boys group, but not in the other groups.

Difficulties in the full understanding of the concepts, as previously mentioned [16], is another possible reason for the low or non-existent correlations. Studies about the cognitive development of children and adolescents referred to the difficulty of reporting time duration of a certain activity, although there is an improvement in time sensitivity throughout childhood [38]. Considering that IPAQA asked specifically about the duration of specific physical activities, this might be one of the reasons why poor correlations were found, specifically for the younger adolescents. This notion is reinforced by research that concluded positively about the effect of motion on time perception [39].

Also related to the IPAQA concepts misunderstanding, the authors believe that adolescents cannot successfully distinguish between moderate and vigorous PA when filling in a questionnaire, which may explain why older boys IPAQA moderate time had a significant correlation coefficient (ρ=0.321; P<0.05) with accelerometer vigorous PA, but no correlation with accelerometer moderate PA. Using a single category of moderate to vigorous physical activity (MVPA) may be useful to overcome this issue, and is in accordance with international guidelines for children and adolescents PA [40].

The concurrent validity proved that the IPAQA questionnaire might be valid when used to determine PA for older adolescent boys (≥15 years), but not with younger adolescents. The same conclusion was reached by the original version [16] and by other countries validation [19].

It is important to emphasize that although there was a weak correlation coefficient, the tertiles percentage of agreement showed that IPAQA could divide a sample in groups of PA, which is of particular interest to epidemiologists that frequently use categories of PA, instead of a quantitative approach. In addition, questionnaires are often a more practical data collection method than the high non-usage rates of instruments that require several day evaluations, such as accelerometers and pedometers [8-10].

The low accelerometer usage rate and the sample size, particularly when comparing groups, are assumed as limitations to this work.

IPAQA may be used in adolescent boys 15 years of age and older, to assess PA in a daily life context. For studies with large samples and budget constraints with no accelerometers availability, IPAQA can be a valid method of data collection. However, the authors would suggest to researchers that choose to use IPAQA that instead of working with time per day in each PA level, creating tertiles of PA may be a better option. This method would enable them to distinguish between more and less active adolescents; or alternatively, it would also be valid to combine moderate and vigorous physical activities into one single category (MVPA). Nevertheless, at this point the authors strongly discourage the use of IPAQA to assess PA in girls or in adolescents under 15 years old. In fact, the authors suggest that, whenever possible and available, objective methods should be used to assess PA in all adolescents. These findings call into question the conclusions of previous studies that used this questionnaire with Portuguese adolescents of all ages and sexes, enhancing the importance of validated tools as the only way of obtaining correct conclusions.

This study confirmed the validation of the first version of IPAQA for Portuguese older adolescent boys, nevertheless, further studies and additional efforts are needed to improve IPAQA and to make it valid for girls, younger adolescents and children.