Systematic literature review of determinants of sedentary behaviour in older adults: a DEDIPAC study

A common protocol for the three DEDIPAC systematic reviews across the life course
(youth, adults, older adults) was developed and is available from PROSPERO (PROSPERO
2014: CRD42014009823).

Search strategy

The literature search was performed in August 2014 in five electronic databases (Pubmed,
Embase, CINAHL with full text, PsycINFO and Web of Science).

The search strategy was based on search terms within four key elements: (a) sedentary
behaviour and its synonyms (e.g. sedentariness); (b) determinants and its synonyms
(e.g. correlates, factors); (c) types of sedentary behaviour (e.g. TV viewing, gaming)
and (d) possible determinants of sedentary behaviour (e.g. environmental, behavioural).
Terms referring to these four elements were used as MESH-headings and title or abstract
words in all databases. The complete list of search terms is shown in Additional file
1: Table S1. In addition, the reference lists of all included articles were scanned
for articles that met the inclusion criteria.

Selection of studies

To be included in the review, articles had to be published in English l and published
between January 2000 and 1
^st
of May 2014. The following study designs were eligible for inclusion; observational
studies (cross sectional, case control and prospective), experimental studies (randomized
controlled trials, quasi-experimental trials) and qualitative studies. These had to
provide information about sedentary time and associated factors for participants aged
65 and over. Articles were included if they measured one or more of the following;
total sedentary or sitting time (e.g. minutes per day) or time spent in one or more
of the following specific domains of sedentary behaviour; time spent watching TV,
screen time, occupational sitting time or motorized transport time. Both objective
and subjective measurement outcomes were included. Articles which recruited only specific
patient groups or samples identified by diseases were excluded.

The study selection process consisted of three phases: In the initial phase, two reviewers
(SC and EF) independently screened articles based on title. In the case of doubt or
disagreements, the articles were included in the abstract review phase. In phase two,
the abstracts of all articles selected from the initial phase were reviewed and assessed
by two independent reviewers (JMO, CB). Any disagreement was resolved by the third
reviewer (SC). In the final phase, the remaining articles were fully reviewed by two
teams of two reviewers (SC, EF and JMO, CB) using the pre-determined inclusion criteria,
and assessed by two independent researchers. Any disagreement between reviewers was
solved by discussion in the wider team.

Data extraction

A standardized template was used to extract data from the included studies using the
following heading; General Information – title of article, main author and publication year, sample characteristics, study characteristics, measurement of sedentary behaviour
and determinants, statistical methods and main results. The four reviewers involved
in article selection, extracted data independently (SC,EF and JMO,CB). A quality assurance
process enabled cross checking of the data extraction. Discrepancies were resolved
through discussion.

Quality assessment

The quality assessment tool employed was the QUALSYST from the “Standard Quality Assessment Criteria for Evaluating Primary Research Papers from
a Variety of Fields” (Alberta Heritage Foundation for Medical Research). This pragmatic tool incorporates two scoring systems, allowing quality assessment
to be conducted on both quantitative and qualitative research 15]. The Qualsyst score is based on eight criteria such as appropriate study design and
research question, definition of outcomes and exposures, reporting of bias and confounding,
and sufficient reporting of results and limitations. Criteria can be answered as ‘yes’
(2), ‘partial’ (1), ‘no’ (0), and ‘NA’. The Qualsyst score is calculated as sum of
ratings of applicable criteria divided by the maximum scores of applicable criteria.

The four reviewers involved in articles selection and data extraction, assessed quality
independently (SC, EF and JMO, CB). A quality assurance process enabled cross checking
of quality assessment. Discrepancies were resolved through discussion. Articles were
not selected based on a threshold of the Qualsyst score.

Results

Searches of the five databases (Pubmed, Embase, CINAHL with full text, PsycINFO and
Web of Science) yielded 4472 records. After duplicates were removed 4050 titles and
abstracts were screened against the inclusion criteria. 3877 were excluded for the
following reasons; relevance (2780) exercise interventions (108) did not include older
adults (327) measured inactivity rather than sedentary behaviour (341) were conducted
in special populations (318) or were incorrect records (3). 171 full-text articles
were assessed for eligibility with 22 studies meeting the inclusion criteria. Figure 1 illustrates selection of studies from search to inclusion. Table 1 provides an overview of the main characteristics of the studies included.

Fig. 1. Prisma diagram of the study selection process

Table 1. Characteristics of studies on determinants of sedentary behaviour in older adults

Three of the studies were conducted in North America, 12 in Europe (6 in the UK),
five in Asia and two in Australia. The majority of studies employed a cross-sectional
design (19) with two prospective cohort studies and one qualitative study.

Number and range of participants

Participant numbers ranged from nine in a small qualitative study Chastin et al. 29], 36] to over 460,000 in a cross-sectional study Du et al. 32] and included male and female participants. Most studies used samples of convenience
with only one study Ku et al. 24] reporting on a nationally representative sample. Participants were included from
a range of SES but tended to be of higher SES and higher educational background. Similarly
participants tended to be from urban environments and while they were from a range
of ethnicity but no study specifically looked at ethnic minorities. Participants were
mostly healthy and community dwelling older adults, however, medical conditions were
never mentioned as an exclusion criteria and the samples are therefore likely to include
participants with pre-existing morbidities.

Quality of studies

Quality scores, expressed as percentage of maximum quality score, ranged from 46–96
% are presented in Table 1.

Measurement of sedentary behaviour

Studies included in the review used objective (n?=?8) (3 activPAL inclinometer, 5 accelerometer) or subjective (self report; n?=?14) measures of sedentary time. Several studies used proxy self-report measures
of sedentary behaviour, including TV or screen viewing (n?=?4), leisure time sitting, (n?=?2) sitting in a car (n?=?1) Sugiyama et al. 23].

Table 2 provides a summary of correlates identified by the selected studies mapped onto Owen’s
ecological model. These are discussed in detail below.

Table 2. Mapping of the results of the 22 studies onto the ecological model per level and categories
and according on whether or not a significant association was found

Individual factors

Age

Ten studies examined the association between age and sedentary behaviour. All but
two of these studies 16], 17] noted a significant effect of age. van Cauwenberg et al. 18] reported lower TV viewing time by 0.5 min per day for every year after the age of
65. Kikuchi et al. 19] also observed that older adults aged over 70 were less likely to watch TV compared
to those aged less than 70 (?0.11 difference in odd ratio). Godfrey et al. et al.
20] reported non-linear association with age, sedentary time was higher by 5 % at 70 years
of age but similar at 80 years compared to 65. Shiroma et al., 2013 21] also observed around 5 % increase in total daily sedentary time per year after the
age of 65. Compared to adults Hamrik 22] reported higher sedentary time in older adults (1 h per day). Finally Sugiyama et
al., 2012 23] estimated that older adults spend half the time sitting in cars compared to adults
aged between 55–65 years.

Sex

Nine studies considered the association between sex and sedentary behaviour with five
reporting an association 17], 18], 20], 23], 24] and four failing to note any association 16], 22], 25], 26]. Ku and colleagues 24] and Arnardottir et al. 17] found that males were more sedentary than females. van Cauwenberg et al. 18] found a statistical significant but trivial difference in TV viewing with men reporting
2.9 mins less viewing per day, but Kikuchi et al. 19] reported that men are 21 % more likely to watch TV compared to female. Godfrey et
al. 20] found no significant association between gender and total sedentary time but patterns
of accumulation of SB differed by gender.

Marital status

The association of marital status with sedentary behaviour was equally inconsistent.
In a cross-sectional study by Van Der Berg et al. 27], unmarried older adults had higher levels (15.3 mins per day more) of self-reported
and objectively measured sedentary behaviour than their married counterparts. This
is partially supported by the findings of Van Cauwenberg et al. 18] who reported the highest levels of sedentary behaviour determined by self-reported
time spent viewing TV among widows and widowers compared to those who had never married/divorced
or were married/cohabiting (6.5 mins per day and 11 mins per day less respectively
compared to widowed individuals). However, two studies failed to find a significant
association between marital status and sedentary behaviour 16], 26].

Employment and retirement status

Employment status showed significant associations with sedentary behaviour in four
studies included in this review 19], 20], 27], 28]. Not being in full-time employment (?35 h per week) almost double the odd ratio of
watching TV according to Kikuchi et al. 19]. Mid-life occupation is also associated with sedentary time in old age 27]. Barnett et al. estimated that retirement is associated with higher TV viewing time
by 2.6 h per week in white-collars and 3.9 for manual workers 28]. By contrast Godfrey and colleagues 20] reported lower levels of sedentary time in unemployed compared to employed older
adults, largely attributable to a greater number of longer bouts of sedentary behaviour
among those who were in employment. Similarly van Cauwenberg et al. 18] estimated that having an occupational function through volunteering was associated
with almost 15 min less TV time per day. This is consistent with qualitative evidence
29].

Educational attainment

Four of the five studies that considered educational attainment found a significant
inverse association between level of education and time spent in sedentary behaviours
(Table 2) . Estimated effect size reported were 42 min less per day for those with higher
education level 18] or 37 % 19] increase in odd ratio for TV time for those with less than university. Ishii et al.
26] did not report a significant association between educational attainments and time
spent in sedentary behaviour in Japanese older adults.

Health

Eleven of the studies included in this review reported on the relationship between
health and sedentary behaviour (Table 2). The majority (n?=?8) reported inverse associations between a measure of health (psychological, behavioural
or functional) and sedentary time. After controlling for socio-demographic factors
Ku et al. 24] found an inverse correlation between self-reported sedentary behaviour and subjective
well-being with those reporting less sedentary time having higher levels of well-being.
This finding was reported by Vallance and colleagues 30] for weekend days but not weekdays with those with the lowest levels of sitting reporting
better physical mental and global health than those reporting most sitting at weekends.
Seven studies that considered the relationship between sedentary behaviour and obesity
found that obese adults reported greater levels of objectively measured 17], 21], 31] and self-reported 27], 32], 33] sedentary behaviour or TV viewing 19]. Estimated effect size reported were 2.5 % 21] and 3.5 % 27] more sedentary time and 50 % higher odd ratio of TV time 19] for obese individuals. van Der Berg et al. 27] is the only study which investigated mid-life health and found that cardiovascular
disease have the strongest effect associated with around 7 % higher sedentary time
in older age.

The only qualitative study to explore individual reasons for sitting reported physical
health problems including pain felt in the standing position, fatigue experienced
while standing and functional limitations as the most important reasons for sitting
time (Chastin et al. 29], 36].

Two studies showed no association between sedentary behaviour and aspects of self-reported
health. Lord et al. 16] found no association between objectively measured sedentary behaviour and self-reported
depression, anxiety of cognitive function and Kikuchi et al. 19] found no association between self-reported TV viewing time and self-rated health.

Interpersonal factors

Only two studies considered interpersonal factors (Table 2). Loneliness was reported as associated with around 2 extra minutes per day of TV
time 18] and living alone increase the odd ratio for TV time by 26 % compared to living in
shared accommodation 19]. The perception in living in a neighbourhood with not too many older adults but not
too many youth or migrant was reported as associated with 5 to 8 min less TV time
per day 18].

Environmental factors

Four of the studies included in this review considered the relationship between environmental
factors and sedentary behaviour (Table 2). Kikuchi et al. 19] reported 48 % higher odd ratio for TV time for older adults living in rural area
compared to those living in urban settings. On the contrary, the larger study by van
Cauwenberg et al. 18] reported 10 min per day higher TV time in urban compared to rural area. In the same
study of almost 51,000 Belgian adults it was estimated that the presence of cultural
facilities or green spaces in the neighbourhood was associated with less TV viewing
by 3 and 3.5 min per day respectively, while perceptions that the environment was
unsafe was associated with 3.5 min per day more TV viewing. Living in an apartment
or a duplex was associated with 2 % higher levels of objectively measures sedentary
behaviour than living in villa our house 18]. When asked about their reasons for sitting, individuals in the study by Chastin
et al. 29] reported that a lack of facilities for stimulation and a lack of resting places in
the environment encouraged more sitting.