Systematic comparative validation of self-report measures of sedentary time against an objective measure of postural sitting (activPAL)

Abstract

BackgroundSedentary behaviour is a public health concern that requires surveillance and epidemiological research. For such large scale studies, self-report tools are a pragmatic measurement solution. A large number of self-report tools are currently in use, but few have been validated against an objective measure of sedentary time and there is no comparative information between tools to guide choice or to enable comparison between studies. The aim of this study was to provide a systematic comparison, generalisable to all tools, of the validity of self-report measures of sedentary time against a gold standard sedentary time objective monitor.MethodsCross sectional data from three cohorts (N = 700) were used in this validation study. Eighteen self-report measures of sedentary time, based on the TAxonomy of Self-report SB Tools (TASST) framework, were compared against an objective measure of postural sitting (activPAL) to provide information, generalizable to all existing tools, on agreement and precision using Bland-Altman statistics, on criterion validity using Pearson correlation, and on data loss.ResultsAll self-report measures showed poor accuracy compared with the objective measure of sedentary time, with very wide limits of agreement and poor precision (random error > 2.5 h). Most tools under-reported total sedentary time and demonstrated low correlations with objective data. The type of assessment used by the tool, whether direct, proxy, or a composite measure, influenced the measurement characteristics. Proxy measures (TV time) and single item direct measures using a visual analogue scale to assess the proportion of the day spent sitting, showed the best combination of precision and data loss. The recall period (e.g. previous week) had little influence on measurement characteristics.ConclusionSelf-report measures of sedentary time result in large bias, poor precision and low correlation with an objective measure of sedentary time. Choice of tool depends on the research context, design and question. Choice can be guided by this systematic comparative validation and, in the case of population surveillance, it recommends to use a visual analog scale and a 7 day recall period. Comparison between studies and improving population estimates of average sedentary time, is possible with the comparative correction factors provided.Electronic supplementary materialThe online version of this article (10.1186/s12966-018-0652-x) contains supplementary material, which is available to authorized users.

A comparison of self-reported and device measured sedentary behaviour in adults: a systematic review and meta-analysis

“…Interestingly, questionnaires that asked participants to recall the proportion of their workday spent sedentary did not significantly differ from inclinometer-assessed sedentary time ranging from an average over-reporting of 9% to under-reporting by 4%. Chastin et al, compared 18 self-report measures of sedentary time against an activPAL and found tools that used a scale to assess the proportion of daily time spent sitting had the lowest amount of missing data and best agreement with deviceassessed measures [25]. Matsuo et al also found that when Japanese workers were asked to indicate the percentage of time spent sitting at work it improved the reliability and validity of the Workers' Sitting and Walking-time Questionnaire (WSWQ) compared to asking them to recall length of time [136].…”

“…A minimum of two studies were required to conduct a meta-analysis. Several studies [24][25][26][27][28][29][30] included multiple self-report measures compared to one device-assessed outcome (e.g., IPAQ and SBQ compared to activPAL), in this case, the applicable 'arms' were included in the appropriate meta-analysis and the device measure group was split to include half in each comparison as per the Cochrane handbook [23]. If a study reported device-assessed results from both an accelerometer and an inclinometer, the inclinometer data was chosen.…”

“…Author's knowledge and bibliographies identified a further three papers. Of these, 185 unique studies met the criteria for study inclusion [19,20,[24][25][26][27][28][29][30]. Individual study characteristics can be seen in Supplemental table 2. Common reasons for excluding studies included: no measure of SB (n = 114); no comparisons of measures (n = 80); not adult population (n = 5); no self-report measure (n = 66); no device measure (n = 46); measures not comparable (n = 23); no measure of variance (n = 6); duplicate study (n = 31); review paper (n = 15); not English or French (n = 14); study protocol with no results (n = 6); and, unable to obtain full text (n = 3).…”

Detecting prolonged sitting bouts with the ActiGraph GT3X

“…Research has linked SB to a plethora of serious chronic diseases and premature deaths . However, the largest body of evidence is based on imprecise and biased self‐reports possibly underestimating the strength of the relationship . The technological improvements in the past years made it feasible to record SB objectively.…”

“…3,4 However, the largest body of evidence is based on imprecise and biased self-reports possibly underestimating the strength of the relationship. 5,6 The technological improvements in the past years made it feasible to record SB objectively. Nowadays, studies investigating SB use small and lightweight body-worn sensors capable to record free-living behavior over several days.…”