Reliability and validity of a modified self-administered version of the Active Australia physical activity survey in a sample of mid-age women

Article (PDF Available)inAustralian and New Zealand Journal of Public Health 32(6):535-41 · January 2009with87 Reads
DOI: 10.1111/j.1753-6405.2008.00305.x · Source: PubMed
Abstract
To assess the test-retest reliability and validity of a modified self-administered version of the Active Australia physical activity survey. One hundred and fifty-nine mid-age women (54-59 years) completed a mailed physical activity questionnaire before recording daily pedometer step counts for seven consecutive days. A random subsample (n=44) also wore an accelerometer during this period. Participants then completed the physical activity questionnaire again. Spearman's rho and per cent agreement were used to assess test-retest reliability. Self-reported physical activity data (time 2) were compared with pedometer and accelerometer data using box plots and Spearman's correlations to assess validity. Median time between surveys was 13 days. Median frequency and duration of moderate and vigorous physical activity were the same at both surveys, but median walking frequency was slightly higher at time 2 than time 1. Reliability coefficients for frequency/time in each domain of physical activity ranged from 0.56-0.64 and per cent agreement scores ranged from 40% to 65% for the physical activity categories; agreement was 76% for 'meeting guidelines'. Correlations (p) between self-reported physical activity and 1) weekly pedometer steps and 2) accelerometer data for duration of at least moderate intensity physical activity were 0.43 and 0.52 respectively. The measurement properties of this modified self-administered physical activity survey are similar to those reported for the original computer assisted telephone interview survey. This modified version of the Active Australia survey is suitable for use in self-administered format.
2008 vol. 32 no. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 535
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Reliability and validity of a modified self-administered
version of the Active Australia physical activity
survey in a sample of mid-age women
Wendy J. Brown, Nicola W. Burton
School of Human Movement Studies, The University of Queensland
Alison L. Marshall
School of Human Movement Studies, The University of Queensland, and School of
Public Health, Queensland University of Technology
Yvette D. Miller
School of Human Movement Studies and School of Psychology,
The University of Queensland
Abstract
Objective: To assess the test-retest
reliability and validity of a modified self-
administered version of the Active Australia
physical activity survey.
Methods: One hundred and fty-nine
mid-age women (54-59 years) completed
a mailed physical activity questionnaire
before recording daily pedometer step
counts for seven consecutive days. A
random subsample (n=44) also wore
an accelerometer during this period.
Participants then completed the physical
activity questionnaire again. Spearman’s
ρ and per cent agreement were used
to assess test-retest reliability. Self-
reported physical activity data (time 2)
were compared with pedometer and
accelerometer data using box plots and
Spearman’s correlations to assess validity.
Results: Median time between surveys was
13 days. Median frequency and duration
of moderate and vigorous physical activity
were the same at both surveys, but median
walking frequency was slightly higher at
time 2 than time 1. Reliability coefcients for
frequency/time in each domain of physical
activity ranged from 0.56-0.64 and per
cent agreement scores ranged from 40%
to 65% for the physical activity categories;
agreement was 76% for ‘meeting guidelines’.
Correlations (ρ) between self-reported
physical activity and 1) weekly pedometer
steps and 2) accelerometer data for duration
of at least moderate intensity physical activity
were 0.43 and 0.52 respectively.
Conclusions: The measurement properties
of this modified self-administered physical
activity survey are similar to those reported
for the original computer assisted telephone
interview survey.
Implications: This modified version of the
Active Australia survey is suitable for use in
self-administered format.
Key words: physical activity, questionnaire,
validation study, reproducibility
Aust N Z Public Health. 2008; 32:535-41
doi: 10.1111/j.1753-6405.2008.00305.x
D
uring the past 10 years, the Active
Australia survey
1
has been widely
used to measure physical activity in
Australian national and State surveys.
2,3
The
measurement properties of this survey have
been assessed and are known to be as good
or better than other commonly used physical
activity surveys.
4-7
Because of its strong association with
both the prevention
8
and management of
several chronic health problems
9
and health
service use and costs,
10
physical activity is an
important research theme in the Australian
Longitudinal Study of Women’s Health. In
1998, researchers modified the physical
activity items used in the Active Australia
survey for inclusion in the longitudinal
study surveys.
11-13
The modified items:
1) specify ‘briskly’ in the description of
walking to ensure that women do not report
time spent in lower intensity walking which
is characteristic of activities such as shopping
and walking with young children; 2) specify
‘leisure activity’ in the descriptors for both
moderate and vigorous activity, while the
Active Australia questions ask about all
activities excluding vigorous household
and gardening activity; 3) ask participants
Submitted: February 2008 Revision requested: July 2008 Accepted: September 2008
Correspondence to:
Wendy J. Brown, PhD, School of Human Movement Studies, University of Queensland, St Lucia,
QLD 4072. Fax: (+ 61 7) 3365 6877; e-mail: wbrown@hms.uq.edu.au
to report activity that lasted 10 minutes or
more, while the Active Australia questions
only specify this for walking; and 4) re-order
the items to ask about moderate intensity
activity second (after walking) instead of
last (after walking, household/gardening,
and vigorous activity).
The Active Australia survey was originally
developed for use in a computer assisted
telephone interview (CATI) format.
1
The
Australian Longitudinal Study of Women’s
Health however, uses a print version of the
survey, which is mailed to three large cohorts
of Australian women. Compared with
telephone interviews, mailed physical activity
surveys are less costly, minimally intrusive,
able to be completed at the participants
convenience, and are less susceptible
to interviewer effects.
14
However, mail
surveys rely on the participants being able
to comprehend the intention of questions
and respond accordingly. Questions with
multiple parts (linked items), as in the
Active Australia survey, can be particularly
problematic for respondents.
15
As this modified version of the Active
Australia
1
questions is used in national
Australian Longitudinal Study of Women’s
Health surveys to monitor physical activity
Methods Article
536 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2008 vol. 32 no. 6
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Brown et al. Article
levels and to assess relationships with various health outcomes, it is
important to know whether the changes to the wording and order of
the items, and the mode of administration, effect the measurement
properties of this measure. As the prevalence of ‘meeting physical
activity guidelines’ is lower in mid-age than in other age groups,
2
mid-age women were chosen as the focus for this study.
Objective
The objective of this study was to assess the test-retest reliability
(repeatability) and validity of the modified version of the Active
Australia survey
1
as used in the Australian Longitudinal Study of
Women’s Health, in a sub-sample of the mid-age cohort.
Methods
Ethical clearance for this study was obtained from the Human
Research Ethics Committee at the University of Newcastle (H-
963-0205) and the Medical Research Ethics Committee at the
University of Queensland (2005000201).
Design
The overall design of the study is presented in Figure 1.
Participants
The Australian Longitudinal Study of Womens Health is a
longitudinal study of factors affecting the health and well-being of
three national cohorts of women who were selected randomly from
the national Medicare health insurance database (which includes all
permanent residents of Australia). The original respondents were
reasonably representative of the general population of Australian
women in this age group, although there was over-representation
of Australian-born, employed and university-educated women.
11,12
Details of the study can be found at www.alswh.org.au. The
sampling frame for this sub-study was all mid-age participants
who were living in the Brisbane Statistical Local Area (Australia)
in 2005. They were aged 54-59 years in 2005.
Procedure
All eligible participants (N=228) were initially contacted
in March 2005. A sub-sample of 90 participants (40%) was
randomly selected using a random number generator to wear an
accelerometer as well as a pedometer. The women were initially
advised of the study by letter, and given the option to decline
participation by telephone (freecall 1800 number). Those who did
not opt out at this stage were telephoned with up to 10 attempted
contacts. Once contacted, participants were provided with more
information about the study. If they provided verbal consent to
participate, a time and place for a face-to-face visit was organised
to deliver the activity monitor(s).
Consenting participants were mailed the physical activity
questionnaire, with a request that it be completed before the
face-to-face visit. The questionnaire included items asking about
leisure time physical activity as well as socio-demographic and
health characteristics. The questionnaire took approximately 10
minutes to complete and included 24 items.
Within three weeks of receiving this questionnaire (range 0-20
days, with one respondent rescheduling to 38 days), respondents
were visited by a research assistant who ensured the questionnaire
was completed before discussing the study protocol. Participants
received verbal and written instructions on how to wear the
pedometer (Yamax Digiwalker SW700) and (if assigned) the
accelerometer (Computer Science and Applications [CSA]
accelerometer WAM 7164, Shalimar, Florida, US), and were
shown how to record their daily pedometer step counts in a step
log. They were asked to wear the activity monitor(s) during all
waking hours for seven consecutive days, and to remove them only
during water-based activities, such as bathing and swimming. At
this visit, participants were also given the second physical activity
questionnaire, and asked to complete it at the end of the week of
activity monitoring. Participants were asked not to refer to the step
log when they completed the second questionnaire. The activity
monitors, questionnaire and step log were returned using reply
paid mail. Participants who did not return the materials after three
weeks were reminded by telephone.
Figure 1: Study design and the number of participants at each stage.
Mail invitation to participate
(N=228)
Telephone invitation to
participate (n=210)
Mail time 1 questionnaire
(n=160)
Participants complete time 1
questionnaire
Visit to collect time 1
questionnaire and deliver
monitors and time 2
questionnaire (n=160)
Activity monitoring
(pedometer n=159;
accelerometer n=44)
Participants complete time 2
questionnaire (n=159)
Return questionnaire by mail
Notes:
(a) for all except one participant (38 days).
(b) Median time 13 days (range 7-28 days, for all except two participants, 37 and 45 days).
4 weeks <3 weeks
a
<1 week
4 weeks
b
2008 vol. 32 no. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 537
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Measures
Self reported physical activity
The questions asked about frequency and duration of walking
briskly (‘for recreation or exercise or to get to or from places’),
moderate leisure activity (‘like social tennis, moderate exercise
classes, recreational swimming, dancing’) and vigorous leisure
activity (‘that makes you breathe harder or puff and pant like
aerobics, competitive sport, vigorous cycling, running, swimming’)
in the last week. Participants were asked to only report activity that
lasted 10 minutes or more. A physical activity score was calculated
as the sum of the products of total weekly minutes in each of the
three categories of activity and the metabolic equivalent value
(MET) assigned to each category: [(walking minutes x 3.0 METs)
+ (moderate-intensity activity minutes x 4.0 METs) + (vigorous-
intensity activity minutes x 7.5 METs)].
16
Physical activity was
then categorised based on total MET.minutes per week as: none
(<40); low (40-<600); moderate (600-<1200); or high (1200+).
17
The lower cut-off for the moderate category is the equivalent of
150 minutes per week of moderate-intensity physical activity, the
minimum required to meet Australian and US guidelines; women
with scores 600 MET.mins per week were therefore categorised
as ‘meeting guidelines’.
18,19
Steps
Participants recorded daily step counts in the step log for seven
consecutive days. Total weekly steps were calculated by summing
the daily entries. Where data were missing for one (three people)
or two days (one person), that participant’s weekday or weekend
average was imputed.
Accelerometer
The accelerometer was set to record movement counts per
minute. These were transformed into minutes of moderate-
(1,952 and <5,724 counts.min-1), and vigorous-intensity activity
(5,725 counts.min-1) based on criteria developed by Freedson et
al.
20
Consistent with physical activity guidelines and self-report
measures used in this study, participants’ data were only included
in the estimates of moderate- and/or vigorous-intensity activity
when the accelerometer data showed at least 10-minute periods
of activity, with at least eight minutes of recorded movement
counts within the ranges set for moderate and vigorous activity.
If participants did not record at least 10 hours of accelerometer
data on a minimum of five days, their data were excluded from
the analyses.
21
Demographic, health and physical activity characteristics
Socio-demographic, health and physical activity characteristics
of the sample were extracted from the most recent survey of the
mid-aged cohort (2004), with the exception of country of birth
and education which were extracted from the 1996 survey, as these
questions were only asked at that time.
Statistical methods
Descriptive statistics (means, standard deviations, proportions)
were used to compare the distribution of characteristics of the 228
women eligible for this study, and in each analysis subsample
(reliability/pedometer validity and accelerometer validity) with
those of the 10,905 participants in the entire mid-age cohort
who completed the 2004 survey. One sample t-tests and Chi-
Square Goodness-of-Fit analyses were conducted, with expected
values derived from the entire mid-age cohort dataset. For
test-retest reliability, means, medians and inter-quartile ranges
were calculated for walking, moderate- and vigorous intensity
activity reported at time 1 and time 2, and Spearman’s rank order
correlations were calculated. Repeatability of the categorical
physical activity measure was assessed using per cent agreement
for each category of physical activity.
Box plots were used to compare the categorical time 2
self-report physical activity data with pedometer data (total
weekly steps) and with accelerometer data (for total time in
moderate- and vigorous-intensity activity). Spearman’s rank order
correlations were used to examine the relationships between both
the pedometer and accelerometer data and the self reported time
in walking/moderate/vigorous activity at time 1 and time 2. All
analyses were conducted using SPSS 13.0 for Windows (SPSS
Inc., Chicago IL, 2004).
Results
Of the 228 women eligible to participate, 18 were unable to
be contacted, 45 declined, and five cancelled their appointments.
Of the 160 who agreed to participate (76% of those contacted),
159 provided physical activity data for time 1 and time 2, and
completed the pedometer step logs. Almost all the women (n=155)
provided pedometer data for seven days; three provided step counts
for six days and one for five days. Of the 56 women assigned to
wear an accelerometer, 48 agreed, 45 returned the monitors and
44 provided data for at least 10 hours on five days.
Demographic, health and physical activity characteristics of
the sample selected for this study and of those who completed
the reliability and validity studies, are shown in Table 1, with
corresponding data from the entire mid-age cohort. The sample
selected for this study (n=228) included a higher proportion of
women with post-school education (χ
2
= 17.26, df = 1, p<0.001)
than in the 1996 baseline survey. There were no significant
differences between this sample and the entire mid-age cohort for
age, BMI, marital status, employment status, self-reported health
status or self-reported physical activity levels. The sample that
provided reliability and validity data also included a significantly
higher proportion of women with post-school education (χ
2
=
17.39, df = 1, p<0.001) and of women reporting excellent/very
good health (χ
2
= 7.09, df = 2, p<0.05) than the entire mid-age
cohort. Participants who wore the accelerometer (n=44) were
reasonably representative of the mid-age cohort (see Table 1).
Test-retest reliability
The median time between activity surveys was 13 days, with a
range of 7-28 days for all except two participants (37 and 45 days).
Methods Reliability and validity of Active Australia survey
538 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2008 vol. 32 no. 6
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Brown et al. Article
Table 1: Demographic characteristics of all participants in the Australian Longitudinal Study on Women’s
Health (ALSWH) mid-age cohort; of those selected for this study, and of those who provided data for the reliability
and validity analyses.
ALSWH Sample selected Reliability and Validity Sample
Mid-age cohort for this study pedometer sample
(accelerometers)
n = 10,905 n = 228 n = 159 n = 44
Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Age (years)
a
55.0 (1.5) 54.9 (1.4) 54.9 (1.4) 54.7 (1.6)
Body Mass Index
a
27.2 (5.5) 27.1 (5.5) 26.9 (5.7) 27.2 (5.6)
n (%) n (%) n (%) n (%)
Country of Birth
b
Australia 8,154 (74.8) 172 (75.4) 121 (76.1) 34 (77.3)
Other English-speaking 1,406 (12.9) 36 (15.8) 24 (15.1) 5 (11.4)
Non-English speaking 918 (8.4) 15 (6.6) 10 (6.3) 5 (11.4)
missing 427 (3.9) 5(2.2) 4 (2.4) 0
Education
b
Up to high school 9,198 (84.3) 169 (74.1) 114(71.7) 36 (81.8)
Post high school 1,618 (14.8) 56 (24.6) 42 (26.4) 6 (13.6)
missing 89 (0.8) 3 (1.3) 3 (1.9) 2 (4.5)
Marital status
a
Married/Defacto 8651 (79.3) 174 (76.3) 126 (79.2) 38 (86.4)
Not married 2,097 (19.2) 50 (21.9) 31 (19.5) 6 (13.6)
missing 157 (1.4) 4 (1.8) 2 (1.3) 0
Employment status
a
Full/Part time paid work 6,609 (60.6) 151 (66.2) 100 (69.2) 28 (63.6)
Home duties 3,843 (35.2) 72 (31.6) 44 (27.7) 15 (34.1)
Retired/Other 453 (4.2) 5 (2.2) 5 (3.1) 1 (2.3) †
Self-reported health status
a
Excellent/Very Good 5,080(46.6) 120 (52.6) 90 (56.6) 28 (63.6)
Good 4,202 (38.5) 79 (34.6) 47 (29.6) 11 (25.0)
Fair/ Poor 1,550 (14.2) 27 (11.8) 20(12.6) 4 (9.1)
missing 73 (0.7) 2 (0.9) 2 (1.3) 1 (2.3)
Self-reported physical activity
a
None 1,671 (15.3) 30 (13.2) 19 (11.9) 4 (9.1)
Low 2,982 (27.3) 68 (29.8) 47 (29.6) 10 (22.7)
Moderate 2,315 (21.2) 59 (25.9) 38 (23.9) 13 (29.5)
High 3,195 (29.3) 62 (27.2) 49 (30.8) 15 (34.1)
missing 742 (6.8) 9 (3.9) 6 (3.8) 2 (4.5)
Notes:
(a) Based on the 2004 ALSWH survey
(b) Based on the 1996 ALSWH survey
Figures in bold indicate a significant difference (p<0.05) between the distribution in the study sample and the ALSWH cohort (Chi square, with cell counts <5
excluded).
Mean (SD) and median (IQR) values for frequency and duration
of walking, moderate and vigorous intensity activity in the two
surveys are shown in Table 2. Median frequency of walking was
slightly higher at time 2, but median frequencies and duration were
the same at both times for all other items. The IQRs for frequency
and duration of walking and vigorous activity were greater at time
2 than time 1, indicating a tendency for some women to report
slightly more activity at the second survey. Spearman’s rho values
ranged from 0.56 for moderate intensity activity (frequency and
duration) to 0.64 for overall minutes of activity (see Table 2).
The per cent agreement scores for the four categories of physical
activity are shown in Table 3. Agreement was highest for the ‘none’
and high’ categories and lowest for the ‘moderate’ category.
Overall, there was 76% agreement on meeting/not meeting the
criterion of 600 MET.mins per week (see Table 3).
Validity
Box plots to show the relationships between the four physical
activity categories (derived from self-report data at time 2)
with weekly pedometer steps, and minutes derived from the
accelerometer data, are shown in Figure 2. Average daily step
counts increased with increasing physical activity category; with
7,708 (SD 2219), 8,415 (SD 3312), 9,019 (SD 2803), 10,676 (SD
2870) mean steps per day in the ‘none’, ‘low’, ‘moderate’ and
‘high’ activity categories respectively.
The 44 women whose accelerometer data were included wore
the device for 15 hours/day on average (SD 1.0, range 12.1-16.9
hours). Box plots for the accelerometer-derived minutes showed
increasing time in physical activity with increasing physical
activity category; from a mean of 20 (SD 31) minutes/week for the
‘none’ category to 186 (110) minutes/week for the high’ category.
Variation was notably lower in the ‘none’ category.
2008 vol. 32 no. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 539
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Table 3: Descriptive profile and test-retest repeatability
of the categorical measures of physical activity (n=159).
Time 1 Time 2 Test – retest
n % n % % agreement
None 18 11.3 20 12.6 61
(<40 MET.min/wk)
Low 59 37.0 58 37.0 56
(40, <600 MET.min/wk)
Moderate 42 26.4 36 22.6 40
(600, <1200 MET.min/wk)
High 40 25.2 45 28.3 65
(1200 MET.min/wk)
‘Meeting guidelines’ 82 51.5 81 50.9 76
(600 MET.min/wk)
Table 2: Descriptive profile and test-retest repeatability of the continuous physical activity measures (n=159).
Time 1 Time 2 Spearman’s rho
Mean (SD)
a
Median IQR
b
Mean (SD)
a
Median IQR
b
Frequency/week
Walking 4.0 (3.3) 3.0 3 4.3 (4.0) 4.0 4.0 .58
Moderate-intensity activity 0.9 (1.8) 0.0 1.0 0.7 (1.3) 0.0 1.0 .56
Vigorous-intensity activity 0.8 (1.7) 0.0 0.0 0.7 (1.6) 0.0 1.0 .60
Total 5.6 (4.6) 5.0 5.0 6.3 (9.3) 5.0 6.0 .58
Minutes/week
Walking 141.8 (128.0) 120 150 166.0 (162.3) 120.0 180.0 .58
Moderate-intensity activity 51.3 (96.7) 0.0 70.0 42.0 (78.0) 0.0 60.0 .56
Vigorous-intensity activity 34.4 (85.8) 0.0 0.0 35.0 (100.2) 0.0 10.0 .61
Total 227.4 (187.4) 180.0 240.0 243.0 (222.4) 180.0 260.0 .64
Notes:
(a) SD – standard deviation
(b) IQR – inter-quartile range
HIGHMODLOWNONE
Physical Activity Category
120000.00
100000.00
80000.00
60000.00
40000.00
20000.00
0.00
Total Weekly Step Count
124
203
69
66
HIGHMODLOWNONE
Physical Activity Category
600.00
500.00
400.00
300.00
200.00
100.00
0.00
Accelerometer recorded min/week in moderate-to-vigorous
activity
69
Figure 2: Box plots to show (top) weekly pedometer steps (N=159) and (bottom) minutes of
moderate and vigorous activity in bouts of at least 10 minutes from the accelerometer, in
relation to the four activity categories derived from time 2 questionnaire data (N=44).
(Open circles represent outliers - values that are further than 1.5 times the inter-quartile range
from the nearest edge of the box).
6
HIGHMODLOWNONE
Physical Activity Category
120000.00
100000.00
80000.00
60000.00
40000.00
20000.00
0.00
Total Weekly Step Count
124
203
69
66
HIGHMODLOWNONE
Physical Activity Category
600.00
500.00
400.00
300.00
200.00
100.00
0.00
Accelerometer recorded min/week in moderate-to-vigorous
activity
69
Figure 2: Box plots to show (top) weekly pedometer steps (N=159) and (bottom) minutes of
moderate and vigorous activity in bouts of at least 10 minutes from the accelerometer, in
relation to the four activity categories derived from time 2 questionnaire data (N=44).
(Open circles represent outliers - values that are further than 1.5 times the inter-quartile range
from the nearest edge of the box).
6
Physical Activity Category
Total Weekly Step Count
Accelorometer recorded min/week in
moderate-to-vigorous activity
Physical Activity Category
Figure 2: Box plots to show (top) weekly pedometer
steps (N=159) and (bottom) minutes of moderate and
vigorous activity in bouts of at least 10 minutes from the
accelerometer, in relation to the four activity categories
derived from time 2 questionnaire data (N=44). (Open
circles represent outliers – values that are further than
1.5 times the inter-quartile range from the nearest edge
of the box).
Spearman’s rho values for correlations between the time 2
self-report data and the pedometer data are shown in Table 4. The
correlation between total reported weekly minutes of physical
activity and total weekly step counts was much stronger than the
relationship between self reported time spent walking and total
step counts (see Table 4).
As walking and moderate intensity activities have similar
intensity they cannot be distinguished in accelerometer records.
They are therefore reported together in Table 4 (lower panel). Mean
weekly frequency of reporting these activities was slightly lower
than that recorded on the accelerometer, but the mean reported
time in these activities was remarkably similar to the accelerometer
data for activities in the 3-5.9 MET range. In contrast, mean self-
reported frequency and duration of vigorous intensity activity were
somewhat higher than the corresponding accelerometer data, but
the rho values for the self-report and objective data were higher
for vigorous than for walking/moderate activity.
The average total activity time recorded on the accelerometer
was 2,812 minutes/week, of which 243 was at intensity 3 METs,
leaving 2569 minutes, or about 6 hours/day of activity at intensity
<3 METs (see Table 4).
Discussion
This study assessed the test-retest reliability and validity (using
both pedometer and accelerometer data as criterion measures) of a
Methods Reliability and validity of Active Australia survey
540 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2008 vol. 32 no. 6
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
Brown et al. Article
Table 4: Comparison of pedometer (top panel) and accelerometer (bottom panel) data with self reported physical
activity at time 2.
Pedometer Self report
steps/week minutes/week Spearman’s
p
n=159 n=159
ρ
Mean (SD) Mean (SD)
Walking 63,301 (21,405) 166 (162) 0.29 <0.001
Total activity 63,301 (21,405) 243 (222) 0.43 <0.001
Accelerometer Self report Spearman’s p
n=44 n=44
ρ
Mean (SD) Mean (SD)
Frequency/week
Walk-moderate activity 6.7 (6.3)
a
5.4 (5.4) 0.40 0.007
Vigorous-intensity activity 0.8 (3.4)
b
1.0 (1.9) 0.55 <0.001
Total activity 7.5 (8.9)
c
6.4 (6.1) 0.48 0.001
Minutes/week
Walk-moderate activity 205.4 (111.8)
a
210.8 (210.5) 0.39 0.008
Vigorous-intensity activity 37.6 (159.3)
b
53.5 (118.2) 0.54 <0.001
Total activity ( 3 METs) 243.1 (208.4)
c
264.3 (255.0) 0.52 <0.001
All accelerometer recorded minutes 2,811.6 (555.8)
d
264.3 (255.0) 0.23 0.14
Notes:
(a) Total moderate accelerometer (3-<6.0 METS; 1952-<5724 counts) sessions and minutes/week accumulated in 10 minute bouts with 2 minute hesitation
(b) Total vigorous accelerometer (6 METS; 5725 counts) sessions and minutes/week accumulated in 10 minute bouts with 2 minute hesitation
(c) Total moderate-to-vigorous accelerometer (3 METS; 1952 counts) sessions and minutes/week accumulated in 10 minute bouts with 2 minute hesitation
(d) Total accelerometer minutes/week
modified version of the Active Australia survey, which has been used
in the Australian Longitudinal Study of Women’s Health since 1998.
Overall, we found the reliability and validity coefficients to be similar
to or slightly better than those previously reported for the Active
Australia and for other commonly used population surveys.
4-7,21-23
Reliability estimates for the continuous measures were
somewhat lower than the ICCs reported by Timperio et al.
7
This
may reflect the different measurement coefficient (intra-class
correlation instead of Spearman’s rho) or the shorter time frame
(three days) between repeated surveys; with four out of seven
days of the recall period for the time 2 survey overlapping with
the time 1 survey in that study. The average test-retest interval in
this survey (13 days) was similar to that used in earlier reliability
studies;
22
overlap of the two one-week recall periods is however
clearly an advantage in reliability surveys, even though there
is a danger that participants may recall their initial responses.
4
Notably, the median values for time in walking, moderate and
vigorous intensity activity in this study were the same at time 1
and time 2, but with slightly higher IQRs for walking and vigorous
activity at time 2. This may be because some women recalled
more activity at time 2, after having previously completed the
time 1 questionnaire and the pedometer logs. Previous research
has shown that the process of self-monitoring physical activity,
through completion of a logbook, may improve subsequent recall
of ‘vigorous’ activity,
6
but in this study, time reported in ‘walking’
was also subsequently higher at time 2. The variability between
time 1 and time 2 responses may also reflect a real fluctuation in
physical activity between the two recall periods.
For the categorical data, agreement was highest for the ‘none’
and ‘high’ physical activity categories. This may be because the
high activity group did more ‘structured’ or ‘routine’ activities,
such as walking for fitness or going to the gym, which are easier
to recall consistently than less well-defined bouts of ‘moderate’
activity such as walking to and from places. Alternatively, it may be
that patterns of both sedentariness and vigorous intensity activity
may be less subject to actual variation than walking and moderate
intensity activity. Agreement on the overall estimate of ‘meeting
activity guidelines’ was very good (76%) and higher than the 66%
reported in our earlier telephone administered reliability study of
the Active Australia questionnaire.
4
The objective data from the pedometers showed surprisingly
little difference in mean daily step counts between the lowest and
highest physical activity categories, with only 2,968 steps/day
(about half an hour of brisk walking) separating the most active
from the least active group. The relatively high step count in the
sedentary (none) group (7,708/day) reflects the fact that women
who report little or no structured activity may still be ‘on their
feet for much of the day, accumulating an average number of steps
that is close to or greater than daily estimates from the US.
24,25
It is
important to note that the pedometer records all steps, regardless
of intensity and duration of movement, so it was not surprising
to find relatively poor concordance between total pedometer step
counts and self-report of at least moderate-intensity activity which
occurs in at least 10 minute bouts.
In contrast, the accelerometer data can be used to discriminate
the time spent in activity at intensity greater than 3 METs, so
there was better agreement between self-report and accelerometer
data, than between self-report and pedometer data. Overall,
the validity coefficients (ρ) reported here were higher for both
walking/moderate (0.39 vs. 0.19), vigorous (0.54 vs. 0.10) and
overall (0.52 vs. 0.25) activity than those reported in an earlier
Australian validity study,
7
but this may reflect the fact that we
used 10 minute bouts of accelerometer data, whereas the earlier
researchers used one minute bouts.
2008 vol. 32 no. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 541
© 2008 The Authors. Journal Compilation © 2008 Public Health Association of Australia
The strengths of this study include the large sample size
and the high level of compliance of the women with the study
requirements, which is typically low in studies of physical activity
monitoring.
24,26,27.
The major limitations of the study relate to the
sample representativeness and to the time sequence of the self-
report and objective measures. The women who participated had
slightly higher education than those of the entire study cohort of
mid-age women, so we cannot be certain of the wider applicability
of the results to women with lower levels of education, or to
younger or older women. Although our intention was to have
women complete the second survey immediately after the period
of objective monitoring, we had less control of the timing of the
second survey than we might have had in a telephone survey.
With hindsight, for future reliability studies we would recommend
a shorter interval between test and re-test surveys, without a
concomitant validity study which may serve to enhance recall.
In conclusion, it would appear that the measurement properties
of this modified self-complete Active Australia survey are as
good (in this sample of mid-age women) as those reported in
our previous study of the telephone administered survey with
Queensland adults
4
and as those reported for other measures of
physical activity.
7,21-23
Overall, the data presented here suggest that
this modified self-administered version of the Active Australia
survey, which is being used in the Australian Longitudinal
Study of Women’s Health and other studies, has measurement
properties that are acceptable for use in large-scale population-
based studies.
Acknowledgements
This study was supported by funding from Brisbane City
Council, Queensland Health and the National Health and Medical
Research Council (ID 301200). Nicola Burton was supported by
a program grant from the National Health and Medical Research
Council (ID 301200) and Yvette Miller was supported by a
capacity building grant from the National Health and Medical
Research Council (ID 252977). The Australian Longitudinal
Study on Womens Health is funded by the Australian Government
Department of Health and Ageing and we are grateful to the
participants in that study who provided additional data for this
study. Dr Stewart Trost wrote the software for converting the raw
accelerometer data into usable summary variables.
References
1. Australian Institute of Health and Welfare. The Active Australia Survey: A Guide
and Manual for Implementation, Analysis and Reporting. Canberra (AUST):
AIHW; 2004.
2. Armstrong T, Bauman A, Davies J. Physical Activity Patterns of Australian
Adults. Results of the 1999 National Physical Activity Survey. Canberra
(AUST): Australian Institute of Health and Welfare; 2000.
3. Queensland Health. Physical Activity Patterns of Queensland Adults. Canberra
(AUST): Australian Institute of Health and Welfare; 2003.
4. Brown WJ, Trost SG, Bauman A, Mummery K, Owen N. Test-retest reliability
of four physical activity measures in population surveys. J Sci Med Sport.
2004;7(2):205-15.
5. Brown WJ, Bauman A, Chey T, Trost S, Mummery K: Comparison of surveys
used to measure physical activity. Aust N Z J Public Health. 2004;28:128-34.
6. Timperio A, Salmon J, Rosenberg M, Bull FC. Do logbooks influence
recall of physical activity in validation studies? Med Sci Sports Exerc.
2004;36:1181-6.
7. Timperio A, Salmon J, Crawford D. Validity and reliability of a physical activity
recall instrument among overweight and non-overweight men and women. J
Sci Med Sport. 2003;6(4):477-91.
8. US Department of Health and Human Services. Physical Activity and Health:
A Report of the Surgeon General. Atlanta (GA): US Department of Health and
Human Services, Centres for Disease Control and Prevention, National Center
for Chronic Disease Prevention and Heath Promotion; 1996.
9. Pedersen BK, Saltin B. Evidence for prescribing exercise as therapy in chronic
disease. Scand J Med Sci Sports. 2006;16 Suppl 1:3-63.
10. Brown WJ, Hockey R, Dobson AJ. Relationships between body mass index,
physical activity and health care costs in mid-age and older Australian women.
Aust N Z J Public Health. In press 2008.
11. Brown WJ, Bryson L, Byles J, Dobson A, Manderson L, Schofield M, et al.
Women’s Health Australia: establishment of the Australian Longitudinal Study
on Women’s Health. J Womens Health (Larchmt). 1996,5(5):467-72.
12. Brown WJ, Bryson L, Byles JE, Dobson AJ, Lee C, Mishra G, et al. Women’s
Health Australia: Recruitment for a national longitudinal cohort study. Women
Health. 1998;28(1):23-40.
13. Lee C, Dobson AJ, Brown WJ, Bryson L, Byles J, Warner-Smith P, et al. Cohort
profile: The Australian Longitudinal Study of Women’s Health. Int J Epidemiol.
2005;34:987-91.
14. Dishman RK, Washburn RA, Schoeller DA. Measurement of physical activity.
Quest. 2001;53 Suppl 1:295-309.
15. Dengler R, Roberts H, Rushton L. Lifestyle surveys - the complete answer? J
Epidemiol Community Health. 1997;51(1):46-51.
16. Brown WJ, Bauman A. Comparison of estimates of population levels of physical
activity using two measures. Aust N Z J Public Health. 2000;24:520-5.
17. Brown W, Ford J, Burton N, Marshall A, Dobson A. Prospective study of
physical activity and depressive symptoms in mid-age women. Am J Prev Med.
2005;29(4):265-72.
18. Commonwealth Department of Health and Aged Care. An Active Way to Better
Health: National Physical Activity Guidelines for Adults. Canberra (AUST):
AGPS; 1999.
19. Haskell WL, Lee IM, Pate RR, Powell KE, Blair SN, Franklin BA, et al.
Physical activity and public health: updated recommendation for adults from
the American College of Sports Medicine and the American Heart Association.
Circulation. 2007;116:1081-93.
20. Freedson PS, Melanson E, Sirard J. Calibration of the Computer Science and
Applications, Inc. accelerometer. Med Sci Sports Exerc. 1998;30:777-81.
21. Craig C, Marshall AL, Sjostrom M, Bauman A, Booth M, Ainsworth B, et al.
International physical activity questionnaire: 12-country reliability and validity.
Med Sci Sports Exerc. 2003;35(8):1381-95.
22. Booth M, Owen N, Buamna AE, Gore CJ. Retest reliability of recall measures
of leisure time physical activity in Australian adults. Int J Epidemiol.
1996;25:153-9.
23. Washburn RA, Heath GW, Jackson AW. Reliability and validity of issues
concerning large scale surveillance of physical activity. Res Q Exerc Sport.
2000;71:104-13.
24. Tudor-Locke CE, Ham SA, Macera C, Ainsworth BE, Kirtland KA, Reis JP, et
al. Descriptive epidemiology of pedometer-determined physical activity. Med
Sci Sports Exerc. 2004;36(9):1567-73.
25. Wyatt HR, Peters JC, Reed GW, et al. Using electronic step counters to increase
lifestyle physical activity: Colorado on the MoveTM. Journal of Physical
Activity and Health. 2004;1:181–90.
26. McCormack G, Giles-Corti B, Milligan R. Demographic and individual
correlates of achieving 10,000 steps/day: use of pedometers in a population-
based study. Health Promot J Aust. 2006;17(1):43-7.
27. Wyatt HR, Peters JC, Reed GW, Barry M, Hill JO. A Colorado statewide
survey of walking and its relation to excessive weight. Med Sci Sports Exerc.
2005;37(5):724-30.
Methods Reliability and validity of Active Australia survey
    • "However, HRQL is a subjective judgement, and the SF-36 is a validated, wellused measure (Mishra and Schofield, 1998;). The PA survey has adequate reliability and validity (Brown et al., 2004; Brown et al., 2008 ). We excluded women who reported on more than one survey an inability to walk 100 m; therefore, results may not be generalisable to non-ambulatory PA. "
    [Show abstract] [Hide abstract] ABSTRACT: Physical activity (PA) is positively associated with health-related quality of life (HRQL) in older adults. It is not evident whether this association applies to older adults with poor mental health. This study examined associations between PA and HRQL in older women with a history of depressive symptoms. Participants were 555 Australian women born in 1921-1926 who reported depressive symptoms in 1999 on a postal survey for the Australian Longitudinal Study on Women's Health. They completed additional surveys in 2002, 2005 and 2008 that assessed HRQL and weekly minutes walking, in moderate PA, and in vigorous PA. Random effects mixed models were used to examine concurrent and prospective associations between PA and each of 10 HRQL measures (eight SF-36 subscales; two composite scales). In concurrent models, higher levels of PA were associated with better HRQL (p<0.001). The strongest associations were found for the bodily pain, physical functioning, general health perceptions, social functioning and vitality measures. Associations were attenuated in prospective models, more so for mental HRQL-related scales than for physical HRQL-related scales. However, strong associations (>3 point differences) were evident for physical functioning, general health, vitality and social functioning. For women in their 70s-80s with a history of depressive symptoms, PA is positively associated with HRQL concurrently, and to a lesser extent prospectively. This study extends previous work by showing significant associations in older women with a history of depressive symptoms. Incorporating PA into depression management of older women may improve their HRQL.
    Article · Sep 2016
    • "The main outcome, weekly physical activity, was assessed at all time points via the Active Australia Survey, (AAS), which has a high percentage agreement with other physical activity measures (67%-75%) [36] and has a good test-retest reliability (kappa = .52) [37] including when self-administered [38]. Quality of life was measured at all 3 time points by the SF-12 version 2, which is valid [39] and reliable [40] including when self-administered on the Web [41]. "
    [Show abstract] [Hide abstract] ABSTRACT: Background: Web-based physical activity interventions that apply computer tailoring have shown to improve engagement and behavioral outcomes but provide limited accountability and social support for participants. It is unknown how video calls with a behavioral expert in a Web-based intervention will be received and whether they improve the effectiveness of computer-tailored advice. Objective: The purpose of this study was to determine the feasibility and effectiveness of brief video-based coaching in addition to fully automated computer-tailored advice in a Web-based physical activity intervention for inactive adults. Methods: Participants were assigned to one of the three groups: (1) tailoring + video-coaching where participants received an 8-week computer-tailored Web-based physical activity intervention ("My Activity Coach") including 4 10-minute coaching sessions with a behavioral expert using a Web-based video-calling program (eg, Skype; n=52); (2) tailoring-only where participants received the same intervention without the coaching sessions (n=54); and (3) a waitlist control group (n=45). Demographics were measured at baseline, intervention satisfaction at week 9, and physical activity at baseline, week 9, and 6 months by Web-based self-report surveys. Feasibility was analyzed by comparing intervention groups on retention, adherence, engagement, and satisfaction using t tests and chi-square tests. Effectiveness was assessed using linear mixed models to compare physical activity changes between groups. Results: A total of 23 tailoring + video-coaching participants, 30 tailoring-only participants, and 30 control participants completed the postintervention survey (83/151, 55.0% retention). A low percentage of tailoring + video-coaching completers participated in the coaching calls (11/23, 48%). However, the majority of those who participated in the video calls were satisfied with them (5/8, 71%) and had improved intervention adherence (9/11, 82% completed 3 or 4 modules vs 18/42, 43%, P=.01) and engagement (110 minutes spent on the website vs 78 minutes, P=.02) compared with other participants. There were no overall retention, adherence, engagement, and satisfaction differences between tailoring + video-coaching and tailoring-only participants. At 9 weeks, physical activity increased from baseline to postintervention in all groups (tailoring + video-coaching: +150 minutes/week; tailoring only: +123 minutes/week; waitlist control: +34 minutes/week). The increase was significantly higher in the tailoring + video-coaching group compared with the control group (P=.01). No significant difference was found between intervention groups and no significant between-group differences were found for physical activity change at 6 months. Conclusions: Only small improvements were observed when video-coaching was added to computer-tailored advice in a Web-based physical activity intervention. However, combined Web-based video-coaching and computer-tailored advice was effective in comparison with a control group. More research is needed to determine whether Web-based coaching is more effective than stand-alone computer-tailored advice. Trial registration: Australian New Zealand Clinical Trials Registry (ACTRN): 12614000339651; http://www.anzctr.org.au/TrialSearch.aspx?searchTxt=ACTRN12614000339651+&isBasic=True (Archived by WebCite at http://www.webcitation.org/6jTnOv0Ld).
    Full-text · Article · Aug 2016
    • "Physical activity was assessed using the Active Australia Survey (Australian Institute of Health and Welfare, 2003), which measures frequency and duration of walking, moderate activity and vigorous activity in the past 2 weeks, and has been shown to have good reliability and validity. Minutes per week of total physical activity was calculated using the recommended algorithm, with amounts <150 minutes per week classified as insufficient physical activity (Brown et al, 2008). Psychological data included self-reported anxiety and depression at baseline and prior to the cardiac event, each assessed using a single-item question. "
    Full-text · Article · Aug 2016
Show more