In the United States and Australia, the adult population spends on average more than 50% of waking hours sedentary (15,22). Time spent in sedentary behavior is detrimentally associated with risk biomarkers and health outcomes, independent of meeting physical activity guidelines during leisure time (25). To advance the evidence to address this issue, the development of measures for use in epidemiological, health behavior, and population surveillance studies is essential. Although there are now device-based measures (accelerometers, inclinometers) of sedentary time available to researchers, these devices are not able to identify time spent in specific sedentary behaviors or domains of behavior. Therefore, high-quality self-report measures are necessary to complement the information from device-based measures or provide the best alternative when cost is an issue.
Sedentary behaviors are defined by both posture (sitting or reclining) and low energy expenditure (typically ≤1.5 METs) (25,36). Time spent in sedentary behaviors has primarily been measured by questionnaires requiring recall of sitting time over the past week (or longer) or as typical behavior. Correlations (ρ) of these questionnaires with device-based criterion measures in 31 validity studies (14) usually have been only weak (15 studies, ρ < 0.3) or moderate (10 studies, ρ = 0.3 to <0.6), with the highest correlation being 0.61 (95% confidence interval [CI] = 0.42–0.75).
Shorter-term recall, such as past-day recall, may improve the accuracy of self-reported sedentary time (23). Dietary assessment studies (5,17) and physical activity studies with children (29,34) have used past-day recall. For assessment of sitting behaviors, short-term recall has been used in the form of an activity log or a time-use diary demonstrating correlations (r > 0.50) with device-based measures (12,33), which are higher than most such correlations for sedentary behavior questionnaires (14). However, logs or diaries can have substantial participant burden and have thus not been widely used in sedentary behavior research. To date, there have been no questionnaires that have used short-term recall for capturing time spent in sedentary behaviors.
The modest relationships with criterion measures that have been reported for sedentary behavior questionnaires may be due to the particular criteria (accelerometer-derived sedentary time or a behavioral log) commonly used in validity studies (14). Issues associated with these criterion measures include inaccuracy in detecting true sitting time (waist-worn accelerometer ) and error and bias associated with self-report (behavioral log [3,26]). The thigh-worn activPAL inclinometer device has been shown to be highly accurate at measuring sitting time (between 0.2% and 2.8% difference in detecting sitting compared with direct observation) (11,19) and to be more sensitive to reductions in sitting time than the accelerometer (19). The inclinometer device may therefore be a more appropriate criterion measure for validation of sedentary behavior questionnaires.
To assess the use of questionnaires in intervention studies, determining the responsiveness of an instrument to detect change over time is necessary. Findings from the very few studies to assess the responsiveness of sedentary time questionnaires have shown mixed results. Responsiveness was shown to be similar to an accelerometer measure of sedentary time in one study (ActiGraph, GT1M; <100 cpm) (10), although questionnaires showed no correlation with change detected by the accelerometer (ActiGraph, GT1M, <100 cpm) (16) and were not able to detect change in the presence of significant change in sit/lie time on the activPAL (20) in other studies.
We examined the Past-day Adults’ Sedentary Time (PAST) questionnaire, a seven-item instrument that uses past-day recall of sedentary time. The questionnaire was assessed in a study taking place in the Cancer Prevention Research Centre, a weight-loss trial in breast cancer patients that included reducing sedentary time among the behavioral targets. Specifically, we examined test–retest reliability of the PAST for a single day (weekday) administration, criterion validity of PAST total sedentary time as a single day (weekday) administration and a 2-d (weekday and Sunday) administration, and responsiveness to change. The activPAL inclinometer was used as the primary referent measure. The ActiGraph accelerometer was also used to allow comparison with previous validity studies.
Data for these analyses were derived from the Living Well after Breast Cancer feasibility trial, carried out between January 2011 and March 2012. This study was a randomized controlled trial evaluating a weight loss intervention delivered via the telephone, compared with a control group, on changes in weight, body composition, targeted behaviors (diet, physical activity, and sedentary behavior), and patient reported outcomes (quality of life, fatigue, and body image). After baseline data collection, participants were randomized, using a computer-generated random number table by staff not involved with the study, to either the intervention (n = 45) or control group (n = 45). Follow-up data were collected after 6 months. Ethical approval was obtained through the University of Queensland Behavioral and Social Sciences Ethical Review Committee and Queensland Health Research and Governance Unit, and participants provided written informed consent.
Women eligible for recruitment into the study were age 18–75 yr, had been diagnosed with breast cancer (stages I–III) in the previous 9–15 months, and resided in an area within a 50-km radius of Brisbane, Australia. Potential participants were required to have completed initial cancer treatment (surgery, radiation, or chemotherapy), speak sufficient English to undertake the assessments, and have a body mass index of 25–40 kg·m−2. They were excluded if they had distant metastases, had a previous diagnosis of invasive breast cancer, had been diagnosed with any other cancer in the past 5 yr, or had contraindications to participation in unsupervised exercise.
Participants were recruited through the Queensland Cancer Registry, which collects information on cancer incidence and mortality in Queensland, Australia. Potentially eligible participants were identified from the registry, and a letter was sent to their treating doctor, explaining the study and inquiring as to whether the patient was appropriate to participate. If deemed appropriate, the potential participants were sent a letter from their doctor seeking consent for them to be contacted about the study. Those consenting to contact then received a telephone call from study staff, who provided more information about the study and screened for eligibility. Those women who were eligible to participate and were interested in taking part signed and returned the consent form. From 1077 patients identified in the Queensland Cancer Registry, 90 participants were included in the study. The reasons for exclusion were doctor consent to contact not obtained (n = 153), refusal (n = 173), ineligible (n = 139), no response or uncontactable (n = 456), sample size reached before contact (n = 58), and no baseline assessment (n = 8).
Participants allocated to the intervention group received a detailed workbook, self-monitoring diary, pedometer, calorie counter book, and up to 16 calls for the 6-month period from a lifestyle coach. Lifestyle coaches were dieticians with at least undergraduate bachelor degree in nutrition and dietetics who received study-specific training in exercise promotion. The intervention aimed to achieve 5%–10% weight loss and targeted reducing dietary energy intake, increasing physical activity and reducing sedentary behavior using behavior change strategies. Participants were encouraged to reduce their sedentary behavior by aiming to get up every 30 min and limiting screen time outside work to no more than 2 h·d−1. Participants were provided with feedback from their baseline assessment at the beginning of the intervention, including their self-reported sitting and lying time across behaviors from the PAST and total amount of sitting/lying time during waking hours from the activPAL for each monitored day.
Control participants received brief feedback after their baseline assessment, including the total amount of sitting/lying time during waking hours from the activPAL for each monitored day; they received no further contact between their baseline and 6-month follow-up assessments.
Data collection procedure.
Data were collected from all participants at baseline and at the 6-month follow-up. At each assessment, participants attended the research center for measurement of height, body weight, body composition (bioelectrical impedance spectroscopy), and waist circumference by trained research staff members blinded to participants study group. At this visit, they were provided with an activPAL (activPAL version 3; PAL Technologies, Glasgow, UK) device, an accelerometer (ActiGraph GT3X+; ActiGraph, Pensacola, FL) device and a log book to record monitor wear time and waking hours.
Participants were instructed to wear the activPAL inclinometer, attached with a hypoallergenic flexible adhesive dressing, continuously for the following 7 d on the right, front mid-thigh. The device was waterproofed; thus, removal was only to be undertaken for swimming in the sea (in case of loss) or for changing the adhesive dressing. For the same 7 d, the GT3X+ accelerometer was to be worn during all waking hours on an elastic belt around the waist positioned on the right midaxillary line. The GT3X+ accelerometer could be removed for sleep, swimming, and bathing.
Participants also completed a telephone interview during the week that they were wearing the monitors. This telephone interview collected information on dietary intake (24-h dietary recall), leisure-time physical activity (Active Australia Survey ), and demographic and health-related information (education, marital status, occupation, work status, income level [optional], and details of breast cancer treatment). Participants completed the PAST questionnaire on four occasions during the study: at baseline during the standard telephone interview, recalling a weekday during device wear (T1); at retest 7 d later, recalling the same weekday (T2); at follow-up, recalling a weekday during device wear (T3); and, finally, at follow-up, recalling a Sunday (T4). The T4 administration recalled a Sunday for pragmatic reasons to fit in with dietary recalls and the working hours of research staff. The T1 and T2 assessments were designed to establish criterion validity and test–retest reliability of a single (weekday) administration of the PAST, based on evidence that recall of weekday sitting is more reliable than weekend (21). The T3 assessment enabled testing criterion validity after intervention and coupled with the T4 assessment enabled validity to be examined for a 2-d (weekday and Sunday, unweighted) administration. The T1 and T3 assessments were used in the analyses for responsiveness. The length of time taken to answer the PAST questionnaire was recorded for the T2 administration to provide information on usability in population surveys.
The PAST questionnaire (see Past-day Adults’ Sedentary Time Questionnaire, Supplemental Digital Content 1, http://links.lww.com/MSS/A252) asked about the time spent sitting or lying (while awake) on the previous day with questions about time spent sitting or lying while at work, travelling, watching television, using the computer (excluding work), reading (excluding work), hobbies, and any other purposes not reported in the previous items. Continuous time reported in the seven items was summed to provide a composite measure of overall sitting and lying time, termed sedentary time.
The activPAL continuously recorded posture (sitting/lying, standing, or stepping). Time spent in a sitting or lying position recorded by the activPAL device during waking hours is called sit/lie time. Using SAS 9.2, each activPAL bout of sit/lie time was compared with the overlapping log-reported sleeping and removal periods. When participants failed to indicate a sleep/wake time in the log, this was estimated from when movement or standing began/ceased according to the device. ActivPAL bouts that were predominantly (≥50%) sleep or removed were excluded as sleep or nonwear time. The comparison of a visual representation of the activPAL data was used to check for any very long uninterrupted movement that could reflect unreported removal and for consistency of the combined activPAL-log awake/worn time against the monitor.
For the daily data to be used, participants needed to have worn the activPAL for ≥10 h and ≥80% of waking hours, or simply for ≥ 80% of waking hours if the participant reported being awake for <10 h. Participants had to have worn the device for sufficient time on at least 5 d for data to be included in the analysis of weekly data. Sit/lie time (h·d−1) during waking hours was recorded for both the day recalled in the PAST questionnaire, and also as average daily sit/lie time over all the days monitored. Participants are unlikely to recall and report very short bouts of sitting, and no respondent reported a sitting activity/domain of less than 5 min. Thus, sit/lie time that occurred in continuous bouts of 5 min or longer (sit/lie 5 min+) was examined as the main activPAL validity criterion measure in addition to all sitting/lying time (sit/lie total).
ActiGraph GT3X+ accelerometer device.
ActiGraph GT3X+ accelerometer data were taken from the vertical axis only to provide comparison with previous validity studies (6,10,30). Wear time was estimated using an automated algorithm with nonwear identified by bouts of ≥60 min of 0 cpm with two interruptions or less of <50 cpm (modification made to allow nonwear bouts to span midnight) (35). For the daily data to be used, participants needed to have worn the accelerometer for close to the self-reported waking hours identified in the activity log (≤2-h difference). For the included participants at baseline, the mean ± SD accelerometer wear time was 14.8 ± 1.1 h on the day of recall and 14.8 ± 0.9 h on average over the monitored days. Participants had to have worn the device for sufficient time on at least 5 d for data to be included in the analysis of weekly data. A minimum of 5 d of ActiGraph accelerometer monitoring has been shown to be necessary to predict average time in sedentary behavior (13). Sedentary time ( <100 cpm, h·d−1) was recorded for the day recalled in the PAST and as an average of daily sedentary time over all the eligible monitored days.
Analyses were conducted in SPSS version 20.0 (IBM Corporation, Armonk, NY) with statistical significance set at P < 0.05. The characteristics of the sample were described as n (%), median (25th–75th percentile) or mean ± SD. Test–retest reliability of total sedentary time and each item from the PAST was examined using intraclass correlation coefficients (ICC) or Spearman’s rank order correlations depending on the distribution of the data. The strength of correlation was interpreted as follows: ICC <0.4 poor repeatability, 0.4–0.75 fair to good repeatability, and >0.75 excellent repeatability (27). Test–retest was examined in the 86 women with retest data on the relevant day, 7 d after the T1 assessment. Women who did not answer the T2 questionnaire (n = 2) or whose assessments were not 7 d apart (n = 2) were excluded.
Pearson’s correlations (r) were used to examine the correlation of a single (weekday) administration of PAST total sedentary time with activPAL sit/lie time (5+ min and total time) on the day recalled and average activPAL sit/lie time over all monitored days, both before intervention (T1) and after intervention (T3). Analyses also included the correlation of a 2-d (weekday and Sunday) administration of PAST total sedentary time (average of T3 and T4, unweighted) with the average activPAL sit/lie time over monitored days at follow-up. The 2-d administration was not weighted for weekday and weekend, and the findings are intended to be indicative of multiple administrations rather than a week of sitting. The 95% CI for the correlations were calculated using Fisher’s transformation.
Comparisons were also made in terms of agreement between PAST total sedentary time test and retest (T1 and T2) and PAST total sedentary time (T1) and activPAL sit/lie time (5+ min and total time) on the day recalled and average activPAL sit/lie time over all monitored days. Agreement was examined using the method outlined by Bland and Altman (2). Plots with mean difference (MD) and limits of agreement (LoA) (±1.96 SD) are presented for test–retest reliability and comparison of PAST total sedentary time with the main criterion measure. Linear regression was used to check whether the MD and LoA varied across average values of PAST sedentary time and activPAL sit/lie time ([total sedentary + acitvPAL sit/lie] / 2) (4) after visual examination of the plots.
To provide a comparison with existing validity studies (8,14), the correlations of PAST total sedentary time (T1) were also made with GT3X+ accelerometer-derived sedentary time on the same day and as average over monitored days using data from the baseline assessment (n = 59).
Criterion validity at baseline was examined in the 72 women with sufficient activPAL data available for the day of the T1 questionnaire and at follow-up in 64 women (intervention n = 35, usual care n = 29). Data loss at baseline and follow-up from the original 90 participants is shown in the participant flow chart (Participant flow chart for the PAST measurement study in the Living Well after Breast Cancer feasibility trial, Supplemental Digital Content 2, http://links.lww.com/MSS/A253). A total of 57 participants provided complete data for the PAST and activPAL at both baseline and follow-up (intervention n = 31, usual care n = 26).
Responsiveness to change of PAST sedentary time and activPAL (5+ min and total time) was evaluated using the responsiveness index and the difference in responsiveness between instruments was tested (32). The responsiveness index, that is, intervention group change divided by the comparison group SD of change, and its 95% CI (based on an assumed normal distribution) were calculated for the PAST and for each activPAL measure. The responsiveness of the PAST was compared with responsiveness of activPAL. For the responsiveness analysis, activPAL sit/lie time was standardized to 16 observed waking hours ([sit/lie time/observed hours] × 16) at both baseline and follow-up as observed waking hours differed significantly between baseline and follow-up. We adopted this approach as the least-complex practical solution eliminating the potential error that could have been associated with wear-time and sleep-time variations. Responsiveness was assessed and compared across instruments for the 57 participants who provided coinciding PAST and monitor data at both baseline and follow-up.
The characteristics of the participants who undertook the baseline assessment (including the T1 PAST administration), which provided sufficient activPAL data at the baseline assessment, and the follow-up assessment are presented in Table 1. Most participants were married or living together, had completed post high school education, and were working part time or full time. They were middle to older age (range = 33–75 yr) and, as per the eligibility criteria, all were overweight or obese.
Reported sitting times
The median time taken to complete the questionnaire was 7 min (25th–75th percentile = 6–8 min). Table 2 shows the reported times for the total sedentary time and the individual items at the single (weekday) administrations at T1 and T2 of the PAST. Total sedentary time was normally distributed while the individual sedentary item times were not. Participants reported approximately 8 h·d−1 on average in sedentary time on the day of recall at both T1 (mean ± SD = 8.20 ± 2.81 h) and T2 (mean ± SD = >7.78 ± 2.77 h). The highest median sedentary time reported for an individual item was for television viewing followed by transport. However, for those who worked on the day of recall (n = 47), sitting for work represented the highest median sedentary time (median = 4.5 h, interquartile range = 2.5–7.0 h).
The correlations between T1 and T2 administrations of the PAST questionnaire are shown in Table 2. Reliability for the composite measure of sedentary time was fair to good (27) (ICC = 0.50; 95% CI = 0.32–0.64). For the individual items, work sedentary time showed the highest test–retest correlation. However, when the data set was limited to those who had reported working on the recalled day (n = 47, 55%), the correlation was lower but still statistically significant (ρ = 0.37; 95% CI = 0.08–0.60). The questions asking about time spent sedentary for other purposes (such as eating and socializing) showed the lowest correlation between test and retest.
The Bland–Altman plot for total sedentary time from the T1 and T2 administrations of the questionnaire is shown in Figure 1. Here, an MD of −25 min (95% CI = −61 to 11 min) was observed, which was not statistically significant (t-test, P = 0.17) and only 5.2% of the average sedentary time of both measures, indicating reasonable agreement at the group level. However, the 95% LoA were wide (−5.9 to 5.0 h), suggesting agreement is poor at the individual level.
Table 3 shows the correlations with activPAL criterion measures of PAST total sedentary time as a single (weekday) administration at baseline (T1) and after intervention (T3) and as a 2-d (weekday and Sunday) administration after intervention. A single (weekday) administration of PAST showed strong (i.e., ≥0.5) correlations at baseline with the main activPAL criterion measure (r = 0.57) and with all sit/lie time (r = 0.58) measured on the recall day; these correlations remained strong after intervention (both r = 0.66). The 95% CI place the true correlations as likely to be at least moderate (i.e., ≥0.3 to <0.5) and possibly reaching the level desired for criterion validity (i.e., ≥0.7). Correlations with average activPAL sit/lie time were only moderate (>0.3 to <0.4). While statistically significant, the 95% CI ruled out as unlikely a desirable level of criterion validity (i.e., ≥0.7) but did not rule out as unlikely that the true correlation may be weak ( <0.3) or strong (≥0.5). By contrast, average sedentary time from the two (weekday and Sunday) PAST administrations showed strong correlation with average activPAL sit/lie time (r = 0.54; 95% CI = 0.34–0.69 for the main criterion measure and r = 0.53; 95% CI = 0.32–0.69 for all sit/lie time) with the 95% CI placing the correlation as at least moderate and potentially close to the desired level.
Additional analyses showed that on the day of recall, the correlations with activPAL sit/lie time for those who reported working (n = 36, r = 0.64, 95% CI = 0.40–0.80) were higher than for those not working (n = 36, r = 0.40, 95% CI = 0.08–0.64). At follow-up, 54 participants recalled sitting on a Sunday and had device-measures for that day. Here, the correlation between the PAST and activPAL sit/lie time was r = 0.57 (95% CI = 0.36–0.73).
The correlations of PAST total sedentary time with accelerometer-derived sedentary time (mean ± SD; sedentary time on day of recall = 8.9 ± 1.4, monitored days = 8.9 ± 1.0) tended to be lower than those seen with activPAL. As had been the case with activPAL, correlations for the single (weekday) administration were higher for the day of recall (r = 0.51; 95% CI = 0.29–0.68) than for average sedentary time (r = 0.45; 95% CI = 0.22–0.63). Correlations between activPAL sit/lie time and accelerometer-derived sedentary time were r = 0.53 (0.33–0.69) for the day of recall and r = 0.62 (0.44–0.75) on average over the monitored days.
The Bland–Altman plot for the agreement between total sedentary time (PAST) and the main criterion measure of activPAL sit/lie time on the recall day can be seen in Figure 2a. When sit/lie time was examined only in bouts of at least 5 min, the MD was small (−0.15 h, 95% CI = −0.72 to 0.42), not statistically significant (P = 0.61), and represented a small proportion of the average time from both measures (1.8%). However, the LoA were wide, indicating that for individuals, differences may be incorrect by several hours (LoA = −4.90 to 4.60 h). The PAST questionnaire produced estimates of sedentary time that were one 1 h than activPAL total sit/lie time on the day of recall (MD = −0.96h, 95% CI = −1.52 to 0.39, P = 0.001), which was 11.0% of the average of the PAST and activPAL total sit/lie time. As seen with the main criterion measure, the LoA were wide (LoA = −5.67 to 3.76 h; plot not shown as similar to main criterion measure).
Linear regression analysis showed a significant positive association for the difference between the PAST and the activPAL sit/lie time per monitored day over the week and the average of the two measures, but no association of variability with the average of the two measures. At lower levels of sedentary time, the PAST recorded less time than the activPAL recorded sit/lie time per monitored day and at higher levels of sedentary time the PAST recorded more time than the activPAL sit/lie time per monitored day. This was the case for the main criterion measure (MD: B = 0.84, SE 0.15, P < 0.001 LoA=MD ± 5.01 h; Fig. 2b) and for all sit/lie time (MD: B = 0.85, SE 0.15, P < 0.001 LoA=MD ± 4.97 h). At the mean levels seen in this sample (average of the two measures= 8.1 h·d−1 for the main criterion measure and 8.5 h total sit/lie time), these MD were −0.19 h (95% CI = −0.45 to 0.82) for the main criterion measure and −0.57 h (95% CI = −1.12 to 0.06) for total sit/lie time.
Changes over the intervention are shown in Table 4. Only PAST total sedentary time showed a significant difference in change within the intervention group (P = 0.04), with the change being large according to the questionnaire (−77 min) and negligible (≤15 min) according to the activPAL. No statistically significant changes in PAST sedentary time or sit/lie time were observed in the control group on any measure, with the change reported being small according to the questionnaire (−32 min) and negligible ( <15 min) according to the activPAL. The changes observed in the intervention group were exceeded by variation in change observed in the control group; hence, the responsiveness index was consistently less than one. The responsiveness index was only significant for the PAST questionnaire (−0.44; 95% CI = −0.92 to −0.04); however, the questionnaire was not shown to be more responsive than activPAL sit/lie time as the CI for the differences in responsiveness all included zero.
As with the validity findings, interinstrument correlations in change scores were stronger for the recalled day than for average sit/lie time (Table 4). For both groups, change in PAST sedentary time had a strong correlation with change in activPAL sit/lie on the recalled day and a lower and nonsignificant correlation with average sit/lie time (Table 4).
This study provides some of the first findings on the reliability, validity, and responsiveness to change of a questionnaire measure of past-day recall of sedentary time in adults. A novel element of this study was the use of past-day recall rather than recall of the past week or a usual/typical day: an approach that has been successfully used in dietary and physical activity research (5,17,29) but has not yet been applied in the sedentary behavior field. The findings suggest that the PAST questionnaire, which has a short administration time (approximately 7 min), provides fair to good test–retest reliability. The criterion validity showed a strong correlation with activPAL sit/lie time (r = 0.57), which is high in terms of the correlations that have been generated from other questionnaires with device based criterion measures (14). There was good agreement at the group level, with minimal mean differences that did not indicate a statistically significant bias at baseline, although agreement was poor at the individual level. The PAST questionnaire was responsive to change; however, as the behavioral changes were not verified by the objective monitoring, this may be attributable to biases in reporting.
A major strength of this study is the use of the activPAL device as a criterion measure for questionnaire measures of sedentary time since the activPAL has shown high agreement compared with direct observation (11), thus providing a good criterion for assessing both correlation and absolute agreement. Our findings compare favorably to a small study that used activPAL as criterion measure for the International Physical Activity Questionnaire (IPAQ) single item for sitting question (weekdays r = 0.41, weekends r = 0.55) and a composite measure of sitting time (weekdays r = 0.30, weekends r = 0.17) (20).
The inclusion of an accelerometer criterion measure permitted the comparison to previous studies reporting validity of sedentary behavior questionnaires. The correlation between PAST and accelerometer-derived sedentary time (r = 0.51 for recalled day and r = 0.45 for average monitored time over the week) was within the upper range of those reported for the IPAQ (ρ = 0.07–0.61) (8), and other composite measures of sitting time (ρ = −0.01 to 0.54) (7,14). The validity findings for the PAST were similar to those of time-use diaries compared with accelerometer-derived sedentary time (nonoccupational sedentary time, ρ = 0.57–0.59) (33), which, while they provide more information on timing of behaviors, are more time consuming and have a higher participant burden than the PAST questionnaire. The correlations between the PAST and accelerometer-derived sedentary time were lower than those observed for the activPAL sit/lie time on the day of recall. This suggests that previous studies that have used accelerometer-derived sedentary time as the criterion measure may not have correctly estimated validity correlations of those questionnaires. Future validity studies should attempt to use the most accurate available measure of sedentary behavior as criterion measure.
Because sedentary behavior is likely to vary from day to day, it is acknowledged that it may be difficult to capture habitual sedentary behavior with a single past-day recall. Thus, as expected, the correlations of a single (weekday) recall of sedentary time with average sit/lie time were weaker than correlations with the recalled day. Nevertheless, the correlation of the single administration of the PAST with average accelerometer-derived sedentary time over the week (r = 0.45) is comparable with those for existing sedentary behavior questionnaires that measure past week or typical behavior (14). Two administrations (weekday and Sunday) of the PAST questionnaire were sufficient to show a stronger correlation (r = 0.54) with average activPAL sit/lie time than the single administration (r = 0.34). Further research should examine how many administrations of the PAST would provide a reliable habitual estimate (ICC ≥ 0.8 or 0.9) and have good agreement with average monitored sit/lie time, and whether multiple administrations can successfully model usual behavior (24). Examination of further measurement qualities, such as responsiveness and reliability, of multiple administrations of the PAST is also recommended.
Examining the agreement on test–retest and between a questionnaire and appropriate criterion measures is important to evaluate the questionnaire’s use as a surveillance tool. The difference in actual time recorded on repeat administrations of the PAST was small (25 min) and not statistically significant. Therefore, in surveillance studies where estimates of the actual time are required at a group level, this questionnaire may provide a reliable measure of sedentary time. The PAST questionnaire provided a close estimate of sedentary time compared with sit/lie time examined in bouts of at least 5 min (9 min less on average). However, it underestimated sedentary time relative to all sit/lie time by almost 1 h, which is consistent with a previous study that used activPAL as criterion measure and showed MD of −41 min for IPAQ and +176 min for a composite measure of sitting time (20). As the activPAL records sitting/lying continuously, it is likely that difficulty recalling all time in these postures, including very short bouts, leads to the large differences that are no longer present when examining sitting/lying that occurs in longer bouts (5 min or more, consistent with the shortest duration reported for any of the PAST sitting items). For all comparisons between the PAST and activPAL sit/lie time, the LoA were wide, indicating that the measure may not accurately estimate an individual’s sedentary time.
To examine longitudinal changes within individuals (e.g., in the context of an intervention trial), instruments must be sensitive or responsive to change (31). This study, one of the few studies to have examined the responsiveness to change of a sedentary time questionnaire, showed the PAST questionnaire to be responsive. The activPAL device has been shown to be highly accurate (11,19) and to detect significant intervention changes (20). The changes in activPAL sit/lie time between baseline and follow-up, although not significant, were negative on the day of recall but not on average over the monitored days, raising the possibility of reactivity. However, we would not expect reactivity to be a significant factor influencing our findings. The administration of the PAST interviews was through unscheduled calls to study participants, who did not know that they would be being asked to recall that sitting for that particular day at the time when they were telephoned. Given that no meaningful or significant change in activPAL-derived sit/lie time was observed for either group, this suggests that the change detected by the PAST may be due to social-desirability bias or to participants’ intention to change rather than representing a true change in sedentary time. The lack of activPAL changes and responsiveness in this study could also indicate that this intervention, in which weight loss was the main target and other diet and physical activity behaviors were included, simply did not affect sedentary behaviors. It is difficult to determine whether past-day recall provides better or worse responsiveness than other types of sedentary behavior questionnaires. A past-week recall questionnaire showed significant responsiveness to change, similar to the responsiveness of accelerometer-derived sedentary time (10). However, questionnaires requiring recall of total sitting time on a typical day did not show responsiveness in interventions that resulted in change in accelerometer-derived sedentary time (16) and activPAL sit/lie time (20). Likewise, a questionnaire requiring recall of typical time spent sitting in domains did not show change in sitting despite change in sit/lie time being recorded by the activPAL (20). Given that the reliability of the PAST was only fair, it is likely that a single administration of this measure may not be a good indicator for intervention studies of change in sedentary time.
One limitation of this study is the generalizability of the findings: participants were all overweight or obese women with a recent history of breast cancer. However, breast cancer survivors have been shown to be similar to the general female population in terms of body mass index, fruit and vegetable intake, smoking status, alcohol intake, and meeting physical activity guidelines (9). Many test–retest reliability studies and validity studies are undertaken in nongeneralizable samples (14). Embedding such studies in this intervention trial has allowed examination of responsiveness and establishing of validity both pre- and postintervention, which is important to identify the types of biases that can occur in interventions. An additional consideration is that the PAST questionnaire was interviewer administered (by telephone). It may have less reliability and validity in a self-completed format, which would be useful in large-scale mailed surveys.
In the current study, the criterion validity of the individual sedentary time items in the PAST questionnaire could not be examined as the device-based measures were not able to determine the context of behavior. Examination of the criterion validity of the particular items is necessary to determine which items may be more accurate and therefore appropriate for epidemiological and surveillance studies. Test–retest reliability was better for some sedentary time items in the PAST (work, transport, and computer use) than others (sedentary time for other purposes). The validity of individual questionnaire items may also vary, as shown in studies that have used activity logs as criteria (21,28). Future studies should use device-based criterion measures that are able to determine context of behavior, such as the new SenseCam technology, which offers potential in identifying behaviors in context, such as travel behaviors (18).
The PAST questionnaire, which uses a single past-day recall method, shows promise for several applications in sedentary behavior research. It may provide accurate estimates of overall sedentary time in the context of population surveillance. The measure may also be useful in epidemiological studies, as correlations with criterion measures were in the upper range of those seen in previous validity studies on sedentary behavior questionnaires. However, the measure requires evaluation of reliability and validity in a more representative population sample, including men, and in self-completion format. Examination of the validity of the individual sedentary items and responsiveness to change in an intervention that specifically targets sedentary time should also be undertaken. Although further development is required, the PAST questionnaire may be an easy-to-use self-report method for measuring sedentary time in population health research.
This study was funded by a University of Queensland Early Career Researcher Award, Australian National Health and Medical Research Council (NHMRC) Program and Equipment Grants (NHMRC no. 569940) and Queensland Health Core Infrastructure funding. Clark is supported by an NHMRC Program Grant (NHMRC no. 569940). Healy was supported by an NHMRC Training Fellowship (NHMRC no. 569861). Dunstan was supported by an Australian Research Council Future Fellowship. Owen is supported by an NHMRC program grant (NHMRC no. 569940) and a Senior Principal Research Fellowship (NHMRC no. 1003960) and by the Victorian Government’s Operational Infrastructure Support Program. Reeves was supported by an NHMRC Training Fellowship [NHMRC no. 389500].
The authors declare that they have no competing interests.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Australian Institute of Health and Welfare A. The Active Australia Survey: A guide and Manual for Implementation Analysis and Reporting, AIHW, Editor 2003. Canberra (Australia): AIHW; pp. 7–9.
2. Bland JM, Altman GA. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986: 307–11.
3. Bouchard C. A method to assess energy expenditure in children and adults. Am J Clin Nutr. 1983; 37: 461–7.
4. Brown R, Richmond S. An update on the analysis of agreement for orthodontic indices. Eur J Orthod. 2005; 27 (3): 286–91.
5. Burrows TL, Martin RJ, Collins CE. A systematic review of the validity of dietary assessment methods in children when compared with the method of doubly labeled water. J Am Diet Assoc. 2010; 110 (10): 1501–10.
6. Clark BK, Thorp AA, AH Winkler E, et al. Validity of self-reported measures of workplace sitting time and breaks in sitting time. Med Sci Sports Exerc. 2011; 43 (10): 1907–12.
7. Clemes SA, David BM, Zhao Y, Han X, Brown WJ. Validity of two self-report measures of sitting time. J Phys Act Health. 2012; 9: 533–9.
8. Craig CL, Marshall AL, Sjostrom M, et al. International Physical Activity Questionnaire
: 12 country reliability and validity. Med Sci Sports Exerc. 2003; 35 (8): 1381–95.
9. Eakin EG, Youlden DR, Baade PD, et al. Health behaviors of cancer survivors: data from an Australian population-based survey. Cancer Causes Control. 2007; 18: 881–94.
10. Gardiner PG, Clark BK, Healy GN, Eakin EG, Winkler AE, Owen N. Measuring older adults’ sedentary time: reliability, validity and responsiveness. Med Sci Sports Exerc. 2011; 43 (11): 2127–33.
11. Grant PM, Ryan CG, Tigbe WW, Granat MH. The validation of a novel activity monitor in the measurement of posture and motion during everyday activities. Br J Sports Med. 2006; 40: 992–7.
12. Hart TL, Ainsworth BE, Tudor-Locke C. Objective and subjective measures of sedentary behavior
and physical activity. Med Sci Sports Exerc. 2011; 43 (3): 449–56.
13. Hart TL, Swartz AM, Cashin SE, Strath SJ. How many days of monitoring predict physical activity and sedentary behavior
in older adults? Int J Behav Nutr Phys Act. 2011; 8: 62.
14. Healy GN, Clark BK, Winkler AE, Gardiner PG, Brown WJ, Matthews CE. Measurement of adults’ sedentary time in population-based studies. Am J Prev Med. 2011; 41 (2): 216–27.
15. Healy GN, Wijndaele K, Dunstan DW, et al. Objectively measured sedentary time, physical activity, and metabolic risk: the Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Diabetes Care. 2008; 31 (2): 369–71.
16. Hoos T, Espinoza N, Marshall SJ, Arredondo EM. Validity of the Global Physical Activity Questionnaire
(GPAQ) in adult Latinas. J Phys Act Health. 2012;9(5):698–705.
17. Johnson RK. Dietary intake–how do we measure what people are really eating? Obes Res. 2002; 10 (1 Suppl): S63–8.
18. Kelly P, Doherty A, Berry E, Hodges S, Batterham AM, Foster C. Can we use digital life-log images to investigate active and sedentary travel behavior? Results from a pilot study. Int J Behav Nutr Phys Act. 2011; 8: 44.
19. Kozey-Keadle S, Libertine A, Lyden K, Staudenmayer J, Freedson P. Validation of wearable monitors for assessing sedentary behavior
. Med Sci Sports Exerc. 2011; 43 (8): 1561–7.
20. Kozey-Keadle S, Libertine A, Staudenmayer J, Freedson P. The feasibility of reducing and measuring sedentary time among overweight, non-exercising office workers. J Obes. 2012: 1–10.
21. Marshall AL, Miller YD, Burton NW, Brown WJ. Measuring total and domain-specific sitting: a study of reliability and validity. Med Sci Sports Exerc. 2010; 42 (6): 1094–1102.
22. Matthews CE, Chen KY, Freedson PS, et al. Amount of time spent in sedentary behaviors in the United States, 2003–2004. Am J Epidemiol. 2008; 167 (7): 875–81.
23. Matthews CE, Moore SC, George SM, Sampson J, Bowles HR. Improving self-reports of active and sedentary behaviors in large epidemiologic studies. Exerc Sports Sci Rev. 2012; 40 (3): 118–26.
24. Nusser S, Beyler N, Welk G, Carriquiry AL, Fuller WA, King BMN. Modeling errors in physical activity recall data. J Phys Act Health. 2012; 9 (1 Suppl): S56–67.
25. Owen N, Healy GN, Matthews CE, Dunstan DW. Too much sitting: the population health science of sedentary behavior
. Exerc Sports Sci Rev. 2010; 38 (3): 105–13.
26. Rennie KL, Wareham NJ. The validation of physical activity instruments for measuring energy expenditure: problems and pitfalls. Public Health Nutr. 1998; 1 (4): 265–71.
27. Rosner BA. Fundamentals of Biostatistics. 6th ed. Belmont (CA): Thomson Higher Education; 2006. 569.
28. Salmon J, Owen N, Crawford D, Bauman A, Sallis JF. Physical activity and sedentary behavior
: a population-based study of barriers, enjoyment, and preference. Health Psychol. 2003; 22 (2): 178–88.
29. Sirard JR, Pate RR. Physical activity assessment in children and adolescents. Sports Med. 2001; 31 (6): 439–54.
30. Trinh OT, Nguyen ND, van der Ploeg HP, Dibley MJ, Bauman A. Test–retest repeatability and relative validity of the Global Physical Activity Questionnaire
in a developing country context. J Phys Act Health. 2009; 6 (1 Suppl): S46–53.
31. Tudor-Locke C. A preliminary study to determine instrument responsiveness to change with a walking program: physical activity logs versus pedometers. Res Q Exerc Sport. 2001; 72 (3): 288–92.
32. Tuley MR, Mulrow CD, McMahan CA. Estimating and testing an index of responsiveness and the relationship of the index to power. J Clin Epidemiol. 1991; 44 (4–5): 417–21.
33. Van der Ploeg HP, Merom D, Chau JY, Bittman M, Trost SG, Bauman A. Advances on population surveillance
for physical activity and sedentary behavior
: reliability and validity of time use surveys. Am J Epidemiol. 2010; 172 (10): 1199–1206.
34. Weston AT, Petosa R, Pate RR. Validation of an instrument for measurement of physical activity in youth. Med Sci Sports Exerc. 1997; 29 (1): 138–43.
35. Winkler EA, Gardiner PA, Clark BK, Matthews CE, Owen N, Healy GN. Identifying sedentary time using automated estimates of accelerometer wear time. Br J Sports Med. 2012;46(6):436–42.
36. Yates T, Wilmot EG, Khunti K, Biddle S, Gorely T, Davies MJ. Stand up for your health: Is it time to rethink the physical activity paradigm? Diabetes Res Clin Pract. 2011; 93 (2): 292–4.