Validity of questionnaire-based assessment of sedentary behaviour and physical activity in a population-based cohort of older men; comparisons with objectively measured physical activity data
Jefferis et al. International Journal of Behavioral Nutrition
and Physical Activity
Validity of questionnaire-based assessment of sedentary behaviour and physical activity in a population-based cohort of older men; comparisons with objectively measured physical activity data
Barbara J. Jefferis 0 1
Claudio Sartini 0 1
Sarah Ash 1
Lucy T. Lennon 1
S. Goya Wannamethee 1
Peter H. Whincup 2
0 UCL Physical Activity Research Group, University College London , London , UK
1 UCL Department of Primary Care & Population Health, UCL Medical School , Rowland Hill Street, London NW3 2PF , UK
2 Population Health Research Institute, St George's University of London , Cranmer Terrace, London SW17 0RE , UK
Background: Older adults are the most inactive age group and self-reporting of activities may be complicated by age-related reductions in structured activities and misclassification or recall biases. We investigate the validity of simple questionnaires about sedentary behaviour (SB), (including the widely used proxy television (TV) viewing), and physical activity (PA) in comparison with objective measures. Methods: Community dwelling men aged 71-93 years, from a UK population-based cohort wore a GT3X accelerometer over the right hip for 7 days and self-completed a questionnaire including information about SB (TV, reading, computer use and car use) and PA (leisure and sporting domains). Results: 1566/3137 surviving men (mean age 79 years) attended. 1377 ambulatory men provided questionnaire and accelerometer data. Questionnaires under-estimated mean daily sedentary time; 317 minutes total SB (TV, computer use, reading or driving), 176 minutes (TV) vs 619 minutes (objectively measured). Correlations between objective measures and self-reports were 0.18 (total SB) and 0.17 (TV), both P < 0.001. Objective SB levels were similar across the lowest three quartiles of self-reported SB but raised in the highest quartile. Correlations between steps/day or moderate to vigorous PA with self-reported total PA were both 0.49, P < 0.001 and measured PA levels were progressively higher at higher levels of self-reported PA. Conclusions: Among older men, simple SB questions performed poorly for identifying total SB time, although simple PA questions were associated with a graded increase with objectively measured PA. Future studies of health effects of SB in older men would benefit from objective measures of SB.
Older adults; Sedentary behaviour; Physical activity; Questionnaire; Accelerometer; Cohort study
Background
Both physical activity (PA) and sedentary behaviour (SB)
are core determinants of health and are associated with
mortality and CVD risk in prospective cohort studies [
1,
2
]. UK guidelines include recommendations about both
PA and SB [3]. At older ages levels of PA are lowest and
SB are highest [
4
] and burdens of PA preventable disease
and disability rise steeply with age [
5
]. Accurate
assessment of PA and SB is therefore essential in this age
group, yet there is little data about validity of both PA
and SB in the oldest old, aged up to 90 years. Until
recently all studies relating PA and SB to morbidity and
mortality used self-reports (with television (TV) viewing
the most commonly used proxy for SB) [
6
]. TV viewing
is consistently associated with elevated risks of mortality,
CVD and diabetes events [
7
]. Yet some studies reporting
detrimental association between self-reported TV time
and CVD risk markers, fail to find associations with
accelerometer measured SB [
8
]. This may be because TV
viewing accounts for only a small part of SB and does
not reflect other domains (leisure, occupational and
transport) [
9, 10
]. Furthermore, there are concerns about
questionnaire-based assessment of both PA and SB in
older people, in whom misclassification bias and recall
bias (which could be exacerbated by memory loss) [11]
are particularly likely and who report difficulty in
answering questions about typical SB [
12
]. Moreover, this
age-group undertakes fewer sporting and structured
exercise and more light activities (including functional
tasks such as walking for transport, household tasks,
caring and gardening [
13
]), which may be harder to recall,
or not be included in questionnaires designed for younger
age groups. Sensor technology permits objective
measurements of PA and SB in population-based studies,
providing estimates of time spent at different intensities of
activity. In order to better understand the health effects of
SB and to design effective interventions to change patterns
of SB in older adults, it is important to assess the validity
of commonly used instruments to measure SB in large
scale studies of community-dwelling older adults.
Likewise, it is important to understand the validity of existing
PA questionnaires which have been used to generate
estimates of PA, from which the dose–response curves
between PA and clinical outcomes are estimated
This paper therefore aims first to inv (...truncated)