Journal of Science and Medicine in Sport (2007) 10, 418—427
Trends in physical activity behaviours andattitudes among South Australian youthbetween 1985 and 2004
Nicole Lewisa, James Dollmanb,∗, Michael Dalea
a School of Health Sciences, University of South Australia, Australiab Nutritional Physiology Research Centre, University of South Australia, Australia
Received 11 May 2006; received in revised form 4 October 2006; accepted 12 October 2006
KEYWORDSPhysical activity;Children;Trends
Summary Rising youth obesity is a serious public health concern. There is awidespread view that declining physical activity is contributing to this trend. A totalof 929 young South Australians (age = 9—15 years) were surveyed in 1985 and 2004on usual physical activity in several contexts, including attitudes to physical activ-ity. Eight of 10 South Australian schools participating in the 1985 Schools Healthand Fitness Survey were revisited in 2004. Comparisons were made on: organisedsport, active transport, physical education (PE), playground activity, vigorous phys-ical activity (VPA), total leisure-time physical activity (LTPA), and attitudes to PEand school sport. The questionnaire and method of administration were identical inboth surveys. There were no differences between surveys in club and school sportparticipation, walking to school, and reported enjoyment of PE and school sport. In2004 fewer children rode to school, but PE classes were more frequent. The per-centage of children who ‘sit and talk’ during school breaks had increased, with adecreased percentage of older girls who ‘run around’ during school breaks. Therewas a significantly higher LTPA in MET.min in 2004 for boys, which was particularlyevident at higher percentiles. There were no changes in mean or distribution of LTPAfor the whole sample or girls. The percentage of respondents reporting ≥3 bouts of
VPA in the previous week rose from 51% (1985) to 76% (2004). There is no consistentevidence of declining physical activity among South Australian youth. It is apparentthat physical activity in some contexts has declined, while in other contexts levelsare the same or higher than in 1985. This underscores the complex nature of physicalactivity and the influences on this behaviour.
rends in physical activity behaviours and attitudes
onsequences in childhood and adolescence, butlso because of the greater risk of obesity indulthood.1 There is a widespread view that declin-ng physical activity is contributing to this trend.owever few empirical studies have been con-ucted to test this assumption.2,3 Areas whereuantifiable data are available include enrolmentnd active participation in physical education (PE),articipation in organised sport, and active trans-ort.
Studies in the USA,4 England,5 New Zealand6 andustralia7 indicate that participation in PE at schoolas declined during the past 25 years. Declines inrganised sport participation have been noted inouth Australian children between 1985 and 1997,7
hile increases in school sport across a similar timepan have been reported among children in Vic-oria, Australia.8 Data on active transport trendsppear to be more consistent, with studies from theSA,9 UK,10 and Australia8,12 indicating that fewerhildren are walking or riding to school.
In order to determine whether a decline inhysical activity has occurred, there is a need toave data collected with the same assessment andnterpretation techniques in demographically sim-lar samples.13 The purpose of this study was toxamine trends between 1985 and 2004 in thehysical activity contexts of school PE, organisedport, active transport, playground activity, vig-rous physical activity (VPA), total leisure timehysical activity (LTPA) and attitudes to PE andchool sport among 9—15 year old South Aus-ralians. These trends were assessed according toex and age specific sub-groups.
ethods
he Australian Schools Health and Fitnessurvey (ASHFS), 1985
n 1985, a nationally representative sample of484 children aged 7—15 years participated in the
bdco
Table 1 Demographic characteristics of respondents, by su
ustralian Schools Health and Fitness Survey, pro-iding extensive data on health and fitness througharious field and technical tests, questionnairesnd blood samples. Participants aged 9—15 yearsn = 6659) completed the questionnaire on healthehaviours and attitudes. The sampling design haseen described in detail elsewhere.14 Briefly, a self-eighted sample was drawn by selecting schoolsith probability proportional to enrolment num-ers, and then using simple random sampling of 10hildren per age/sex category within each school.inety percent of invited schools agreed to partic-
pate (n = 109), and 78% of students selected in thenitial sample received parental consent to partici-ate.
he South Australian Physical Activityurvey, 2004
n 1985, 10 South Australian schools were includedn the ASHFS, and 8 of these agreed to partici-ate again in 2004. One of the 10 schools from985 no longer exists, while in 2004 a single sexon-government school chose not to participate.he data from the un-matched 1985 schools wereemoved from the analyses. Within participatingchools, random selection of participants repli-ated the 1985 process. In 2004, 82% of parentsrovided consent for their children to completehe survey. The decision to re-visit the Southustralian schools from the ASHFS, rather thaneplicate the 1985 school sampling strategy in004, was founded on the relatively small num-er of schools, and the likelihood that 10 randomlyelected schools in 2004 could result in a substan-ially different demographic profile to the 1985outh Australian sample. For instance, the selec-ion of an additional single sex private schoolould bias the 2004 sample in terms of gender
alance and socioeconomic status (SES). As theemographic characteristics of the schools had nothanged between 1985 and 2004, the proportionsf urban/rural (6/2 schools), government/non-
government (6/2 schools), primary and secondary(4/4 schools), and large and small schools wereretained (see Table 1). Records of school-card reg-ister percentages (SCR; the percentage of childrenin a school receiving means-tested government sup-port to attend school) for the participating SouthAustralian schools suggest there has been no dis-cernable shift for each school from the earliest(1997) to the most recent (2004) records. Between1997 and 2004, the coefficient of variation ofSCR varied between 0.2 and 0.6% for participat-ing schools. This suggests that the socioeconomiccharacteristics of the schools’ feeder neighbour-hoods have changed little between survey years.Data collection in both the 1985 and 2004 surveysoccurred during the winter months, minimising theimpact of seasonal variation in physical activity asa confounder of trends.
The questionnaire
The full ASHFS questionnaire has been publishedpreviously.14 It comprises items on: organised andnon-organised physical activity in contexts such asschool and club sport; PE; free play at school;active transport to and from school; attitudes tohealth-related behaviours, including physical activ-ity, diet, smoking and alcohol consumption; andperceived activity levels of parents. In both sur-veys, trained data collectors read instructions togroups of 4—6 respondents and supervised comple-tion of the questionnaire. Responses to questionsrelated specifically to physical activity behaviourand attitudes were compared in this study (seeTable 2 for questions). Neither survey includedassessment of test—retest reliability of responses,nor is there published evidence of validity of thequestionnaire.
Approval for the 1985 survey was grantedthrough the then State Directors General of Edu-cation, while the 2004 survey was approved by theSouth Australian Department of Education and Chil-dren’s Services and the University of South AustraliaHuman Research Ethics Committee.
Data transformation
Leisure time physical activityQuestion 2 of the questionnaire asked for organisedand non-organised activities performed outside ofschool hours in the previous week. The LTPA vari-able was calculated from responses to the active
transport questions (bicycle riding and walking toand from school) and the Other Activities describedin Question 2. A rate of energy expenditure (EE)was ascribed to these activities based on the com-
acuD
N. Lewis et al.
endium of Ainsworth and colleagues.15 Where theate of EE has not been published or could note determined, a MET.min value was allocated onhe basis of activities in the compendium deemedimilar in intensity. For example, where the respon-ent did not specify what form of ‘athletics’ wasndertaken, a mean of the three available ‘ath-etics’ options (running, jumping, and throwing) atach intensity (light, medium or hard) was used touantify EE. Energy expenditure for each specificctivity was calculated as the rate of EE in METsisted for that activity multiplied by the frequencyf participation in the week and the average dura-ion of participation in each episode of activity.hese were summed to provide an estimate of totaleekly LTPA.
ctive transport
n the questionnaire children reported the numberf times they walked or rode to and from schooln the previous week. These data were reduced towo categories: those who engaged in active trans-ort (three or more occasions per week) and thoseho did not (two or less times per week). Trends inalking and cycling were analysed separately as it
s reasonable to believe that these are influencedy separate environmental factors.
tatistical analyses
he sampling design raises the possibility of vari-nce inflation due to clustering of subjects inchools. To test this, log binomial generalised linearodelling was used with each physical activity vari-
ble in the study (STATA, version 8, College Station,X). Models were run with and without the ‘cluster-
ng’ function. As the standard errors were changedegligibly (<5%) in all cases, it was concluded thathere was a minimal clustering by school effect.
Questionnaire items on physical activityehaviour and attitudes required either cate-orical responses, which were analysed usinghi-square, or descriptive responses, which wereranslated into continuous variables and comparedsing ANOVA. Individual cell Chi-squares wereomputed if the overall Chi-square was significant.TPA was skewed and normalised by log trans-ormation, before comparisons by ANOVA. Theistributions of LTPA in the two survey samplesere compared by Chi-square, based on groupings
ccording to the 1985 quintiles. The frequencies oflub and school sport participation were comparedsing Mann—Whitney U for non-parametric data.ue to the large number of separate analyses,
Trends in physical activity behaviours and attitudes 421
Table 2 Items from the survey of physical activity behaviours and attitudes
bdbrt
R
TosprbifRsega
422
Bonferonni adjustment was conducted (˛ dividedby the number of comparisons).
All analyses were conducted on the whole sam-ples, and separately by sex and primary schoolage (<13 years old) versus secondary school age(≥13 years old). Comparisons in separate age stratawere performed as physical activity opportunitiesof the current generation of adolescents might bemore restricted than in the past, by such fac-tors as increased homework demands and part-timeemployment.
Results
Comparisons between 1985 and 2004 survey sam-ples are provided in Tables 3—7.
Active transport
In 2004, fewer children rode to and from school,
with a decrease of 15.3% for the whole sample,17.6% for boys, and 11.9% for girls (see Table 3).There was no difference between surveys for walk-ing to school among the whole sample, and for
aaal
Table 3 Comparison of responses to categorical questionssurveys, by sex
Response Whole sample (%) Boys (%)
1985 2004 1985
Cycling to and from schoolYes 20.6 5.4 26.6No 79.4 94.6 73.4
�2 = 40.4, p < 0.0001, df = 1 �2 = 21
Walking to and from schoolYes 39.9 36.7 36.9No 60.1 63.3 63.1
�2 = 0.8, p = 0.3627ns, df = 1 �2 = 1.
RecessSit and talk 23.4 36.4 24.7Walk and talk 27.0 25.4 28.6Run around 49.4 38.0 46.2Study 0.3 0.3 0.6
�2 = 15.7, p = 0.0013ns, df = 3 �2 = 6.
LunchSit and talk 16.6 34.5* 20.7Walk and talk 26.8 19.4 29.0Run around 53.8 45.8 47.9Study 2.9 0.3* 2.4
�2 = 35.9, p < 0.0001, df = 3 �2 = 11.
Note: Yes, ≥3 times per week. No, <3 times per week. *IndividualnsModel �2 is not significant after modified Bonferonni adjustment.
N. Lewis et al.
oys and girls separately. When analyses were con-ucted in separate age strata, older boys and girlsoth rode to school less (p < 0.0001 and p = 0.0003,espectively) in 2004. No differences were seen inhe separate age strata for walking.
ecess and lunch time activity
here was a significant increase in the proportionf respondents reporting sedentary activity duringchool breaks in the 2004 survey (see Table 3). Theercentage of children who ‘sit and talk’ at lunchose by 18.0% (p < 0.0001) for the whole sample, andy 23.5% (p < 0.0001) for girls. Other non-significantncreases were seen in ‘sit and talk’ behaviouror boys at lunch and for all students at recess.eported ‘study’ declined at lunch for the wholeample and also for girls, but the number of childrenngaged in study is very small in both surveys. Forirls, the shift to increased ‘sit and talk’ came fromdecline in reported ‘run around’ behaviour. When
nalyses were conducted in separate age strata,mong older children ‘sit and talk’ had increasedt recess (boys, p = 0.0007; girls, p = 0.0009) andunch (boys, p = 0.0008; girls, p < 0.0001) in 2004.
on physical activity behaviours between 1985 and 2004
Girls (%)
2004 1985 2004
9.1 13.4 1.691.0 86.6 98.5
.5, p < 0.0001, df = 1 �2 = 19.4, p < 0.0001, df = 1
31.7 43.5 42.068.3 56.5 58.0
3, p = 0.261ns, df = 1 �2 = 0.1, p = 0.7657ns, df = 1
�2 = 53.8, p < 0.0001, df = 1 �2 = 29.3, p < 0.0001, df = 1 �2 = 25.5, p < 0.0001, df = 1
Note: Yes, VPA ≥ 3 times per week. No, VPA < 3 times per week.
Table 5 Comparison of club and school sport participation between 1985 and 2004 surveys, by sex
Whole sample Boys Girls
1985 2004 1985 2004 1985 2004
Club sportMedian 1 1 1 1 1 1IQR 2 2 2 2 2 1
U = 73846.0, p = 0.3165ns U = 20766.5, p = 0.664ns U = 16311.0, p = 0.353ns
School sportMedian 1 1 1 1 1 1IQR 2 2 2 2 1 2
U = 75180.5, p = 0.5608ns U = 18787.5, p = 0.0387ns U = 15860.5, p = 0.1727ns
Note: nsNot significant after modified Bonferonni adjustment.
Table 6 Comparison of participation in, and enjoyment of, school sport and PE between 1985 and 2004, by sex
Response Whole sample (%) Boys (%) Girls (%)
1985 2004 1985 2004 1985 2004
Enjoy PEVery much 33.3 41.6 35.6 42.2 30.7 40.9Quite a lot 27.7 27.3 27.0 31.2 28.4 23.3Sometimes 24.3 24.0 26.5 22.1 21.6 25.9Not much 4.7 4.1 3.3 3.0 6.3 5.2Not at all 1.6 1.8 1.9 0.5 1.1 3.1I do not do PE 8.5 0.8* 5.7 0.5 11.9 1.0*
We do not have PE 0.0 0.5 0.0 0.5 0.0 0.5
�2 = 31.1, p < 0.0001, df = 6 �2 = 14.0, p = 0.0296ns, df = 6 �2 = 24.6, p = 0.0004, df = 6
Enjoy school sportVery much 41.6 40.1 40.7 41.7 42.6 38.3Quite a lot 24.4 23.7 27.8 24.6 20.5 22.8Sometimes 15.3 14.8 10.5 14.1 21.0 15.5Not much 2.3 3.1 2.4 3.0 2.3 3.1Not at all 1.6 1.0 1.4 1.5 1.7 0.5I do not do school
sport13.5 15.6 14.8 14.1 11.9 17.1
We do not haveschool sport
1.3 1.8 2.4 1.0 0.0 2.6
�2 = 1.9, p = 0.9323ns, df = 6 �2 = 2.8, p = 0.8351ns, df = 6 �2 = 9.8, p = 0.1314ns, df = 6
Note: *Individual cell �2 is significant after modified Bonferonni adjustment. nsModel �2 is not significant after modified Bonferonniadjustment.
424 N. Lewis et al.
Table 7 Comparison of the distribution of leisure time physical activity (LTPA) between 1985 and 2004 surveys,by sex
�2 = 13.0, p = 0.0113ns, df = 4 �2 = 23.4, p = 0.0001, df = 4 �2 = 0.4, p = 0.9844ns, df = 4
-timeell �2
P
Rt(spca
L
LimgsaTa
D
Ltcsl
A
T
Note: 1985 quintile ranges in brackets. LTPA, reported leisuresignificant after modified Bonferonni adjustment. *Individual c
In younger children, there was no change in thereported incidence of ‘run around’ during recessand lunch. In older children, reported ‘run around’had decreased in 2004 in both boys (p = 0.0005)and girls (p = 0.0009) at recess and in girls at lunch(p < 0.0001).
Vigorous physical activity
In the 2004 sample, a higher proportion of childrenreported engaging in three or more bouts of VPA perweek (responses to the ‘‘huff and puff’’ question,see Table 4). When analyses were conducted in sep-arate age strata, reported VPA was higher in bothage categories for boys (<13 years p < 0.0001; ≥13years p = 0.0006) and also in older girls (≥13 yearsp < 0.0001) in the latter survey.
Organised sport
There were no differences between surveys in thefrequencies of both club and school sport participa-tion, among the whole sample and for boys and girlsseparately (see Table 5). Similarly, when analyseswere conducted in separate age strata, no differ-ences were found.
Enjoyment of PE and school sport
There were no differences between surveys inenjoyment of PE and school sport for the wholesample and for older and younger boys, andyounger girls (see Table 6). Among these subgroups,
greater than 60% of respondents indicated thatthey enjoyed school PE and sport ‘‘very much’’, or‘‘quite a lot’’, in both surveys. Among older girls,there was a significant increase in those enjoyingPE ‘‘very much’’ from 27 to 50% (p = 0.0004).
taAni
physical activity during the previous week. nsModel �2 is notis significant after modified Bonferonni adjustment.
E participation
eported participation in PE at school increased inhe 2004 survey for the whole sample, and for girlssee Table 6). When analyses were conducted ineparate age strata, there was an increase in PEarticipation among older children (p < 0.0001). Nohange in PE participation was evident in both boysnd girls <13 years.
eisure time physical activity
TPA (reported from the previous week) was highern 2004 in boys (1985 median = 805.0 MET.min; 2004edian = 1233.6 MET.min, p = 0.001), but not in
irls. A distributional shift was confirmed for boys,uch that increases in LTPA were more evidentmong the most active respondents (see Table 7).here was no evidence of a distributional changemong girls between surveys.
iscussion
imited standardised data are currently availableo examine temporal trends in physical activity inhildren and adolescents. The findings of this studyupport some and contrast other results from simi-ar studies and reviews.
ctive transport
he finding of the current study that overall activeransport to school had decreased in children and
dolescents is consistent with evidence from otherustralian8 and overseas4—6 studies. Interestingly,o change in active transport by walking was notedn the current study of South Australian children.
T
Tirt‘bens1tad
V
TidbmaUhtywd(aivyfitd
S
TibAslUg1TmslIata
fsaaia2ptePIraat
P
Iddoioittoppai
prbatciCbcaoebtic
rends in physical activity behaviours and attitudes
his observation could be attributed to changesn the local context of active transport. Perhapsecently introduced government sponsored initia-ives to promote walking to school, such as thewalking bus’, have led to stabilisation of walkingehaviours in South Australia. It must be acknowl-dged that the current generation of children doot have the option of walking or riding home fromchool during lunch breaks, as was the case in985. This reduced opportunity for routine activeransport, together with widely reported concernsround traffic safety,11 may be contributing toeclines in cycling frequency.
PA (‘huff and puff’)
he findings of the current study suggest anncrease in the proportion of South Australian chil-ren meeting the guidelines for VPA, as representedy the self-reported frequency of activity whichakes the respondent ‘huff and puff’. Data fromstudy conducted by the Schools Health Educationnit12 on UK children support this finding with aigher prevalence of reported VPA at least threeimes in the previous week in 2003 (49% in 14—15ear olds and 53% in 12—13 year olds) comparedith 1995 (36 and 32%, respectively). In contrast,ata from the USA Youth Risk Behaviour SurveysYRBS)16 showed no change in self-reported VPAmong older adolescents between 1993 and 1997. Its important to acknowledge that questions aboutigorous intensity activities or those that ‘‘makeou huff and puff’’ may be biased by decreases intness and exposure to these types of activities,hereby increasing the level of perceived exertionuring typical sub-maximal tasks.3
chool PE
he current study suggested that participationn PE in South Australian children had increasedetween the two surveys, contrasting with anotherustralian study across a similar time span7 thathowed declines in PE enrolment, particularly inow socioeconomic status schools. Data from theSA show that high school enrolment in PE pro-rammes had declined from 65% in 1984 to 52% in99017 and again from 42% in 1991 to 27% in 1997.16
he results of the current study indicate that enjoy-ent of PE remained high between 1985 and 2004,
uggesting that variance in participation is moreikely to be tied to opportunity than preference.
t must be acknowledged that attitudes toward PEre likely to be formed relative to other subjectshat children undertake at school, and that PE maylways be judged highly relative to these.
is
r
425
Trend data on PE participation may be con-ounded by changing patterns of PE delivery inchools, with some schools currently offering PEs a half-year subject. This, and a lack of data onctive involvement in PE classes may obscure trendsn the overall impact of PE. Further, as the currentnalyses are based on two ‘‘snapshots’’ in 1985 and004, there is no way of determining how PE partici-ation fluctuates year to year. Despite these issues,he findings of this study regarding increased lev-ls of PE participation and continued enjoyment ofE can only be perceived as a positive outcome.
t is reasonable to speculate that school PE canespond quickly and dramatically to policy changesnd recommendations by government departments,nd policies to reinforce and increase PE participa-ion and enjoyment should be implemented.
layground activity
n 2004, more children reported sedentary activityuring school breaks, particularly girls and chil-ren of high school age. No data could be locatedn recess and lunch play trends from other stud-es. The declines in active play, particularly amonglder children, are interesting in light of emerg-ng theories of behavioural choice. Epstein18 positshat the choice to be physically active depends onhe attraction to and access to physical activitypportunities. It might be expected that the schoollayground is no different in schools today com-ared with 1985, and therefore declining physicalctivity during school breaks is somewhat surpris-ng.
The decreases found in this study may reflectolicy changes at the school level. For example,educed yard supervision during school breaks haseen associated with lower playground activity insample of primary age USA children.19 Currently
here is greater pressure on schools to safeguardhildren’s well being, such that playground activitys being limited to an entirely supervised context.hanges in recess and lunchtime physical activityehaviour between surveys might be impacted byhanges to the length of these breaks. There isnecdotal evidence that schools, particularly sec-ndary, now schedule shorter lunch breaks and havearlier dismissal times. This abbreviation of thereaks in the day may influence children’s use ofhe available time. Furthermore if PE time hasncreased, as indicated by the data, children mayompensate for this increased physical activity dur-
ng class time with a reduction during their freechool time.
There is no way of knowing whether other bar-iers, invisible to teachers and researchers, are
oesafmtotadaocf
owaphsdricaio
acwsTnubtttAhs
C
Tjbi
426
having a greater impact on the current generationof children as they confront the school playgroundduring breaks. For example, playground bullying isa barrier which would strongly influence children’schoices to enter the playground, but which is typi-cally ignored by physical activity surveys.
Organised sport
Youth sport represents a substantial portion of dailyenergy expenditure for adolescents.2 In the currentstudy, no differences were found between surveysin the frequencies of club and school sport partic-ipation. This contrasts with a report from Victoria,Australia, of increased school sport participationbetween 1985 and 20018. Further, an earlier SouthAustralian study reported that children’s participa-tion in organised sport had declined between 1985and 1997.7 Why might sport participation appearto be returning to 1985 levels? Perhaps the mean-ing of the term ‘‘sport’’ has changed for childrenwho have participated in recent surveys. In otherwords, there may still be declines in organised sportas it existed in 1985, while less formal versions ofsport such as modified games are now being inter-preted by children as organised sport. If children’sunderstanding of the meaning of organised sport ischanging, then true trends will be clouded by thisaltered perception.
Leisure time physical activity
The results of the current study suggest that boys’LTPA had increased between the two surveys, par-ticularly at the highest MET.min quintile. Findingsin the international literature are equivocal. Eisen-mann et al.2 examined leisure time physical activityenergy expenditure (AEE) in 12—19 year olds fromnational Canadian surveys conducted between 1981and 1998, finding that AEE had increased between1981 and 1988, and then remained relatively stablebetween 1988 and 1998. Conversely, findings fromthe Amsterdam Growth and Health Study suggestthat overall EE declined between 1977 and 1992 by42% in males and 17% in females between ages 13and 27 years.20 In contrast to the current study onSouth Australian children, there was a significantreduction in EE at the higher percentiles.
One of the strengths of this study is the method-ological consistency in the two surveys. Samplingmethods and protocols were replicated betweenthe two studies. However, some limitations of this
approach must be acknowledged. The 2004 surveytargeted children of the same age range from 1985.Trend studies of this type are plagued by concur-rent trends in maturational status, with the onset
tiaf
N. Lewis et al.
f puberty estimated to present about 3 monthsarlier per decade.21 Consequently, the childrentudied in the latter survey are likely to be moredvanced in maturation than their counterpartsrom 1985. As declines in physical activity may beore aligned with increases in biological rather
han chronological age, this will introduce method-logical error. However, this effect is likely to leado general declines in physical activity, particularlymong girls,22 and yet these declines were not evi-ent. On the other hand, if more mature childrenre better able to handle the cognitive demandsf self-report,23 more thorough reporting of physi-al activity would be expected in the latter survey,urther obscuring true trends.
There is no published evidence of the validityf the component of the survey from which LTPAas derived. While self-report instruments provideconvenient way to assess activity patterns in largeopulations, most validation studies with childrenave reported only moderate correlations betweenelf-report and objective criteria.24 Self-reportedata are reliant on the ability of the child to accu-ately recall activity levels and are plagued bynfluences such as social desirability. An increasedommunity awareness of the importance of physicalctivity among respondents in latter surveys maynfluence the data by introducing a systematic biasf over-reporting.25
The 1985 survey was designed around physicalctivity in discrete episodes, as are most physi-al activity recalls. Spontaneous, incidental activityas missed in the surveys, and this can repre-
ent a substantial portion of a child’s daily EE.he inability to recall salient details of sponta-eous behaviours in a child’s daily life deniess the resolution to detect small but meaningfulehavioural shifts.26 Furthermore, as surveys areypically delivered to classes in schools there is aendency to ignore physical activity during holidayime that represents approximately 25% of the year.ny change in physical activity habits during schoololidays will not be detectable by available trendtudies.
onclusion
here is rising acceptance that the observed tra-ectory of overweight prevalence among adults cane explained by quite small imbalances of energyntake and energy expenditure.27 This is also likely
o apply to young people, in whom the methodolog-cal error associated with self-reported physicalctivity is relatively large. Improved instrumentsor measuring physical activity will provide the
T
rUhathpvupi
A
Tc
R
rends in physical activity behaviours and attitudes
esolution to detect small changes in behaviour.ntil such instruments can be developed, publicealth policies and intervention strategies shouldim to promote adequate physical activity levels. Inhe mean time our data do not support the widelyeld perception of ‘across the board’ declines inhysical activity. Therefore, it is likely that inter-ention strategies need to be based on a betternderstanding of factors that uniquely influencehysical activity in the range of contexts in whicht takes place.
cknowledgements
he authors would like to thank the principals andhildren of the schools involved in the project.
eferences
1. Guo SS, Chumley WC. Tracking of body mass index in chil-dren in relation to overweight in adulthood. Am J Clin Nutr1999;70:S145—8.
2. Eisenmann JC, Katzmarzyk PT, Tremblay MS. Leisure-time physical activity levels among Canadian adolescents,1981—1998. J Phys Act Health 2004;1:154—62.
3. Dollman J, Norton K, Norton L. Evidence for secular trendsin children’s physical activity behaviours. Br J Sports Med2005;39:892—7.
4. Lowry R, Wechsler H, Kann L, et al. Recent trends in partici-pation in physical education among US high school students.J School Health 2001;71:145—52.
5. Harris J. Physical education in the national curriculum:is there enough time to be effective? Br J Phys Educ1994;25:34—8.
6. Ross J, Hargreaves J, Cowley V. Assessing the demise ofschool physical education with deregulation. In: Simpson C,Gidlow B, editors. Proceedings of the ANZALS Conference1995. Canterbury: Lincoln University; 1995. p. 242—7.
7. Norton KI, Dollman J, Klanarong S, et al. Kids sport: who’splaying what? Sport Health 2001;19:12—4.
8. Salmon J, Timperio A. Trends in children’s physical activ-ity and weight status in high and low socioeconomic statusatreas of Melbourne, Victoria, 1985—2001. Aust N Z J PublicHealth 2005;29:337—42.
9. Rowe N, Champion R. Young people and sport national sur-vey 1999. London: Sport England; 2000.
10. Tudor-Locke C, Ainsworth B, Popkin BM. Active commut-ing to school, an overlooked source of children’s physicalactivity? Sports Med 2001;31:305—9.
Available online at www.
427
11. Black C, Collins A, Snell M. Encouraging walking: the caseof journey-to-school trips in compact urban areas. UrbanStud 2001;38:1121—41.
12. Schools Health Education Unit. Trends: young people andphysical activity attitudes and participation in exercise andsport 1987—2003. SHEU Publications; 2004.
13. Cale L, Harris J. Young people’s physical activity and fitnessstatus, Chapter 2. In: Cale L, Harris J, editors. Exerciseand young people. Basingstoke: Palgrave MacMillan; 2005.p. 9—40.
14. Pyke J. Australian health and fitness survey. Adelaide:Australian Council for Health, Physical Education and Recre-ation (ACHPER); 1987.
15. Ainsworth BE, Haskell WL, Whitt MC, et al. Com-pendium of physical activities: An update of activity codesand MET intensities. Med Sci Sport Exerc 2000;32(9):S498—504.
16. Pratt M, Macera CA, Blanton C. Levels of physical activityand inactivity in children and adults in the United States,current evidence and research issues. Med Sci Sports Exerc1999;31:S526—33.
17. Heath GW, Pratt M, Warren C. Physical activity patterns inAmerican high school students. Arch Pediatr Adolesc Med1994;14:1131—6.
18. Epstein LH. Integrating theoretical approaches to promotephysical activity. Am J Prev Med 1998;15(4):257—65.
19. Sallis J, Conway TL, Prochaska JJ, et al. The associationof school environments with youth physical activity. Am JPublic Health 2001;91(4):618—20.
20. van Mechelen W, Twisk JWR, Post GB, et al. Physical activityof young people; the Amsterdam longitudinal growth andhealth study. Med Sci Sport Exerc 2000;32:1610—6.
21. Norton K, Olds T, Olive S. Anthropometry and sports per-formance, Chapter 11. In: Norton K, Olds T, editors.Anthropometrica. Sydney: UNSW Press; 1996. p. 287—364.
22. Sallis JF, Prochaska JJ, Taylor WC. A review of correlates ofphysical activity of children and adolescents. Med Sci SportsExerc 2000;32(5):963—75.
23. Matthews CE. Use of self-report instruments to assessphysical activity, Chapter 7. In: Welk GJ, editor. Physicalactivity assessments for health-related research.. Cham-paign: Human Kinetics; 2002. p. 107—24.
24. Welk GJ, Corbin CB, Dale D. Measurement issues in assess-ment of physical activity in children. Res Q Exerc Sport2000;71(2):59—73.
25. Trost SG, Sallis JF, Pate RR, et al. Evaluating a model ofparental influence on youth physical activity. Am J Prev Med2003;25(4):277—82.
26. Dollman J. Trends and sociodemographic distribution ofchildren’s health-related fitness and behaviours. Unpub-
lished Ph.D. thesis. Adelaide: University of South Australia;2003.
27. Brown W, Williams L, Ford JH, et al. Identifying the energygap: magnitude and determinants of five year weight gainin mid-age women. Obes Res 2005;13(8):1431—41.
