Overall, the total device-measured sedentary time corresponded to 56.6% of the day in the entire sample (men: 59.5%, women: 53.1%), while at work, the participants were sitting for almost half of the time each day. The recorded daily time in SB measured objectively was comparable to other studies conducted on a Chilean adult population.[21, 39] The reliability of the PAST questionnaire was moderate, and its validity was weak compared with the AP to estimate SB, but only during working hours.
The reliability of the PAST showed similar results to those reported previously with the English version in Australian adult and university populations.[24, 25] However, our findings on the concurrent validity of the PAST were dissimilar to those reported previously in Australia, since, in those studies, the validity was moderate to substantial.[24, 25] These differences may be due to our research included adults with various occupations and physical activity requirements compared to the Australian participants, which had patients with cancer and people in a university context.[24, 25] Apart from comparing between sexes, we explored different grouping options using both occupations and latent class analysis based on the movement behaviours as measured with the AP, but no improvements were found in the accuracy properties of the instrument.
The mean bias between the PAST and AP for total SB (54.9 min/day) was considerably lower than the observed with the single question of the GPAQ assessed in Chilean population (-293.9 min/day). Our findings are in line with a recent meta-analysis in which multi-item questionnaires showed smaller mean differences (-10.93 min/day, 95% CI: -51.13, 29.28) with accelerometers compared with single-item instruments (-159.56 min/day, 95% CI: -189.69, -129.44). Although the mean bias was almost an hour, the limits of agreement between the PAST and AP were wide (± 9hrs), indicating that the estimations on the individual level are very limited. These large differences might be because some study participants over or under-reported some sedentary activities without a clear pattern, as found in several studies from a meta-analysis. For example, some people may have reported time on the computer for recreational or work activities separately, while others may have included that as part of the estimation only at work. During the data collection, efforts were made orally to avoid these confusions, as well as in the written instructions. Still, each participant was free to report the times that they estimated.
The PAST questionnaire performed better when it was assessed in the work context, as shown in other studies. The concurrent validity of the PAST was weak (r = 0.37, p < 0.002) for both when measuring time in SB at work and comparing the proportion of their workday spent in SB (r = 0.32, p = 0.007). The use of percentages could be a complementary approach to explore in the future as the current and previous studies have shown no differences between overall recall estimates based on proportions and those derived from devices. Following this principle, it has been suggested that a visual analogue scale could be used to measure the proportion of the day spent sitting as it has shown higher precision and data loss when compared with other tools. The slightly better performance of the PAST in work context could be explained due to the time in SB during the workday being less variable between days, unlike free time activities, which were also observed in the initial validation of the PAST.
Self-report measures are valuable tools as they provide contextual information that could be very difficult to register with accelerometers or other devices. Also, when comparing prospective studies that have been used self-reported and device-measured SB, the direction of the associations with outcomes, such as mortality, are consistent. Although the self-reports performance is usually poor to moderate compared with devices, these methods will remain as relevant tools as more interest has been placed into differentiating the influence of different sedentary domains or activities on outcomes, such as mental health.[8, 41, 42] Due to the lower cost of self-reports compared with accelerometers or inclinometers, these instruments play an important, and sometimes unique, role to conduct studies in settings where resources are limited, or research is incipient, like in Chile. Therefore, this study fills an important gap to better understand sedentary behaviours in a Latin American context.
This study was the first to validate the PAST questionnaire in Spanish. Although the recruitment was not probabilistic, it managed to include participants from various work environments and with a wide age range. This improves the external validity of the findings reported in this study about a working adult population. Although 101 participants were recruited, only 77 were included in the analyses, mainly for not using the accelerometer. Nevertheless, there were no differences in the demographic characteristics between those included and not included in the study. This instrument only evaluates the SB from the last day, which could limit its use to estimate common SB that could be more variable, particularly in free time. Although the PAST assesses SB in different domains, apart from the whole day, we were able to explore only the time spent sedentary at work. This remains an issue in the field due to the difficulty of assessing SB in different contexts, so the use of cameras or logs as reference standards may be useful tools to fill this gap.