behavioral determinants of physical activity across the life ......across the life course were...
TRANSCRIPT
-
REVIEW Open Access
Behavioral determinants of physical activityacross the life course: a “DEterminants ofDIet and Physical ACtivity” (DEDIPAC)umbrella systematic literature reviewGiancarlo Condello1*† , Anna Puggina2†, Katina Aleksovska2†, Christoph Buck3, Con Burns4, Greet Cardon5,Angela Carlin6, Chantal Simon7, Donatella Ciarapica8, Tara Coppinger4, Cristina Cortis9, Sara D’Haese5,Marieke De Craemer5, Andrea Di Blasio10, Sylvia Hansen11, Licia Iacoviello12, Johann Issartel13, Pascal Izzicupo10,Lina Jaeschke14, Martina Kanning11, Aileen Kennedy15, Fiona Chun Man Ling6,16, Agnes Luzak17,Giorgio Napolitano10, Julie-Anne Nazare7, Camille Perchoux7, Caterina Pesce1, Tobias Pischon14, Angela Polito8,Alessandra Sannella9, Holger Schulz17, Rhoda Sohun6, Astrid Steinbrecher14, Wolfgang Schlicht11,Walter Ricciardi18,19, Ciaran MacDonncha6†, Laura Capranica1†, Stefania Boccia18†
and on behalf of the DEDIPAC consortium
Abstract
Background: Low levels of physical activity (PA) are a global concern and increasing PA engagement is becominga priority in current public health policies. Despite the large number of studies and reviews available, the evidenceregarding the behavioral determinants of PA is still inconclusive. Thus, the aim of this umbrella systematic literaturereview (SLR) was to summarize the evidence on the behavioral determinants of PA across the life course.
Methods: A systematic online search was conducted on MEDLINE, ISI Web of Science, Scopus, and SPORTDiscusdatabases. The search was limited to studies published in English from January, 2004 to April, 2016. SLRs andmeta-analyses (MAs) of observational studies that investigated the behavioral determinants of PA were consideredeligible. The extracted data were assessed based on the importance of the determinants, the strength of evidence,and the methodological quality. The full protocol is available from PROSPERO (PROSPERO 2014:CRD42015010616).
Results: Seventeen reviews on 35 behavioral determinants of PA were eligible for this umbrella SLR. Regardless ofage, the most investigated determinants were those related with ‘screen use’ and ‘smoking’. For youth, probablepositive evidence emerged for ‘previous PA’ and ‘independent mobility and active transport’ among children andadolescents. For the adult population, ‘transition to university’ and ‘pregnancy/having a child’ showed probablenegative associations.(Continued on next page)
* Correspondence: [email protected]†Equal contributors1Department of Movement, Human and Health Sciences, University of RomeForo Italico, P.za Lauro de Bosis, 15, 00135 Rome, ItalyFull list of author information is available at the end of the article
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Condello et al. International Journal of Behavioral Nutritionand Physical Activity (2017) 14:58 DOI 10.1186/s12966-017-0510-2
http://crossmark.crossref.org/dialog/?doi=10.1186/s12966-017-0510-2&domain=pdfhttp://orcid.org/0000-0001-5092-3258mailto:[email protected]://creativecommons.org/licenses/by/4.0/http://creativecommons.org/publicdomain/zero/1.0/
-
(Continued from previous page)
Conclusions: Although the majority of the evidence was limited and most of the determinants were notassociated with PA, this umbrella SLR provided a comprehensive overview of the associations between behavioraldeterminants and PA. Youth should be physically active in the early years and increase active transportation to/fromschool, independent mobility, and ‘free-range activities’ without adult supervision, whilst adult PA behaviors aremostly influenced by the life events. Finally, more research is needed that incorporates prospective study designs,standardized definitions of PA, objective measurement methods of PA assessment, and the use of interactionist andmediational approaches for the evaluation of different behavioral determinants influencing PA behaviors.
Keywords: Physical activity, Behavioral determinants, Umbrella systematic literature review, Health promotion
BackgroundPhysical activity (PA) is widely recognized, together withappropriate nutritional and mental status, to be an im-portant component of healthy lifestyle [1]. Since thebenefits of being physically active for the prevention andthe management of the non-communicable diseases arewell known [2], the European Union (EU) is strongly en-gaged in promoting health-enhancing physical activity[3] for all individuals independently from age and socialstatus. However, in Europe, a large proportion of chil-dren, adolescents, adults, and older adults still adoptssedentary lifestyles [4] and, consequently, are more atrisk of non-communicable diseases.Several ecological models for the exploration of the
lifestyle choice have been proposed [5–8], including in-dividual (e.g., biological, psychological, and behavioralaspects), interpersonal (e.g., relationships with parents,relatives, peers, and socio-cultural networks), environ-mental (e.g., access/availability of tools/services, andproximal/distal built/natural surroundings), and policy(e.g., organizational and governmental aspects) dimen-sions. Within those dimensions, positive, negative, incon-clusive, or no associations might exist between severaldeterminants and PA. However, both determinants andPA present a great diversity in research designs, measure-ment approaches, populations studied, types of measure-ment, terminologies, which still make difficult to draw acomprehensive understanding. In general, the term ‘deter-minant’ is used to address causal variables also includingcorrelates (i.e., multiple variables intervening in cause-effect relationships), whilst mediators (i.e., variables influ-encing a cause-effect relationship between variables),moderators (i.e., variables effecting the strength of a rela-tionship between variables), and/or confounders (i.e., vari-ables associated with the outcome that distort theobserved relationships) are considered different variables[9, 10]. Furthermore, a lack of commonality exists in thePA terminology applied in the studies and different formsof PA are considered, ranging from unstructured daily ac-tivities, occupational PA, leisure time PA to structured PA(e.g., exercise, grassroots sports, and competitive sports)and considering the most relevant parameters of PA, such
as frequency (e.g., daily, weekly, monthly), duration (e.g.,total time of activity, rest intervals), and intensity (e.g., low,moderate, moderate-vigorous, vigorous, maximal efforts).Within the individual dimension, the determinants re-
lated to behaviour are widely investigated. Commonly,behavioral determinants have been already defined as“those that can be eliminated or reduces through life-style or behavioral changes” [11]. Thus, among all ages,investigations into the behavioral determinants consid-ered those related with major life changing events (e.g.,pregnancy and retirement) [12], habits (e.g., smokingand alcohol consumption) [13], sedentary behaviors (e.g.,TV viewing and video/computer game use) [14], andavailable mobility (e.g., active school transport and inde-pendent mobility) [15, 16]. They have been studied in anumber of primary epidemiological studies, systematicliterature reviews (SLRs), and meta-analyses (MAs).However, the diversity in research designs, theoreticaland measurement approaches, population groups, deter-minants investigated, and PA outcomes, across the lit-erature, makes it difficult to understand the evidenceand to draw appropriate conclusions on the importanceof these determinants in influencing PA behaviors.Recently, the European Commission endorsed a Joint
Programming Initiative to increase research capacityacross Member States to engage in a common researchagenda on a Healthy Diet for a Healthy Life [17] and theDEterminants of DIet and Physical ACtivity-KnowledgeHub (DEDIPAC-KH) project was created [18]. In orderto expand knowledge and to develop new insights andsolutions in PA, the Thematic Area 2 of the DEDIPAC-KH project organized and carried out an umbrella system-atic literature review (SLR) [19] on all the possible deter-minants of PA. Overall, seven categories of determinantsof PA have been identified: biological, psychological, be-havioral, physical, socio-cultural, socio-economic, and pol-icy determinants. Due to the amount of researchesavailable, the DEDIPAC-KH Management Team decidedto organize the findings in seven separated umbrella SLRs,each focused on a single category. The DEDIPAC-KHManagement Team is conscious that splitting the categor-ies may cause a lost in the analysis of the interactions
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 2 of 23
-
between those categories of determinants which mayshare commonalities. However, this strategy was consid-ered necessary for a clear dissemination of insights on thedeterminants of PA.The aim of this umbrella SLR was to provide a system-
atic overview of studies investigating behavioral determi-nants of PA across the life course by evaluating existingSLRs and MAs. The summary also captures the differentstudy designs and definitions applied for behavioral de-terminants and PA. Additionally, overall results of theretrieved SLRs and MAs have been evaluated in terms ofthe importance of the determinants, the strength of theevidence, and the methodological quality.
MaterialsThe manuscript has been drafted following the PRISMAchecklist [20]. A common protocol of the seven um-brella SLRs (biological, psychological, behavioral, phys-ical, socio-cultural, socio-economic, and policy) wasregistered and is available on PROSPERO (Record ID:CRD42015010616), the international prospective registerof systematic reviews [21]. Review title, timescale, teamdetails, methods, and general information were all re-corded in the PROSPERO register prior completing dataextraction.
Search strategy and eligibility criteriaThe present study applied the same search strategy as wasused for the other umbrella SLRs of the DEDIPAC-KH.SLRs and MAs investigating the determinants of PAacross the life course were systematically searched onMEDLINE, ISI Web of Science, Scopus, and SPORTDis-cus. The search was limited to SLRs and MAs publishedin English, between January 2004 and April 2016. To avoidduplications of the earliest individual studies, reviews pub-lished before 2004 were not included. Table 1 shows theMEDLINE search strategy, and Fig. 1 summarizes the
process of the literature research, common to the subse-quent 7 umbrella SLRs. Thus, Table 1 provides the overalllist of searched terms, whilst Fig. 1 shows the count of theoverall excluded/included reviews, related and not-relatedwith behavioral determinants. SLRs or MAs of observa-tional primary studies on the association between any de-terminant and PA, exercise (Ex) or sport as the mainoutcome, were included in the seven umbrella SLRs. Thefollowing studies were excluded: i) SLRs and MAs ofintervention studies; ii) SLRs and MAs that focused onspecific population groups (e.g., chronic diseases); and iii)umbrella SLR’s on the same topic (e.g., reviews of SLRs orMAs of epidemiological studies on variables in associationwith PA). In line with Peel and colleagues [10], behavioraldeterminants were defined as “those that can be elimi-nated or reduced through lifestyle or behavioral changes”.
Selection processRelevant articles were independently screened and assessedby two reviewers belonging to the DEDIPAC-KH, whoscreened the titles, the abstracts, and the full texts. Beforethe final study inclusion or exclusion, a common decisionwas reached for each study. Any uncertainty and disagree-ment was resolved by consulting three further authors ofthe DEDIPAC-KH (BS, CL, PA).
Data extractionFor each included review, data was extracted on a prede-fined data extraction form, developed by the DEDIPAC-KH and checked by two authors (AK, PA). In reportingdata, authors agreed to use the terms “reviews” as thoseSLRs and MAs found eligible for the umbrella SLR, and‘primary studies’ as those studies included in the eligibleSLRs and MAs. Moreover, authors agreed to consider allof the terminologies and forms of PA, including unstruc-tured (i.e., PA linked with daily life) and structured (i.e.,
Table 1 Search strategy: key words used for the literature research
Set Search terms
#1 “physical activit*” OR “physical exercise*” OR sport OR “motor activit*” OR “locomotor activit*” OR athletic* OR fitness OR “physical movement*”OR “physical performance*” OR “aerobic exercise*” OR “physical effort*” OR “physical exertion*”
#2 determinant OR determinants OR correlate OR correlates OR mediator OR mediators OR moderator OR moderators OR contributor ORcontributors OR factor OR factors OR association OR modifier OR modifiers OR confounder OR confounders OR pattern OR patterns ORpredictor*
#3 demographic* OR motivation OR cognition OR emotion* OR attitude* OR “self-perception” OR “self-confidence” OR “self-efficacy” ORcompetence OR reward* OR success* OR challenge* OR knowledge OR belief* OR “personal trait*” OR “body image” OR satisfaction OR “timeavailability” OR “perceived environment” OR family OR peer* OR school* OR leader* OR coach* OR group* OR “climate” OR network* ORemployment OR retirement OR “educational level” OR SES OR “socioeconomic status” OR “local identity” OR “national identity” OR value* ORtradition* OR “social expectation*” OR “social trend*” OR “social barriere*” OR “availability of tool*” OR “availability of service*” OR “access to tool*”OR “access to service*” OR neighborhood OR “community route*” OR “school environment” OR “work environment” OR architecture ORurbanization OR transport OR traffic OR “facilit* in public space*” OR advertisement OR “availability of sport club*” OR “availability of fitnesscenter*” OR advocacy OR lobbying OR “corporate social responsibility” OR “physical activity promotion initiative*” OR legislation OR health OReducation OR tourism OR environment OR “urban planning” OR transport* OR sport OR sports OR culture OR dance OR theater OR “gendermainstreaming” OR “social inclusion” OR “fiscal measure*” OR program* OR plan OR plans OR communication OR media OR guideline*
#4 “systematic literature review” OR “meta-analysis”
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 3 of 23
-
exercise and sports) independently from their frequency,duration, and intensity.The following information was extracted from each in-
cluded review: year of publication, type of review (eitherSLR or MA), number of eligible primary studies in-cluded in the behavioral umbrella SLR over the totalnumber of primary studies included in the review, con-tinent/s of the included primary studies, primary studydesign, overall sample size, age range or mean age, gen-der proportion, year of publication range of included pri-mary studies, outcome details, type of determinant/s,aim of the review, overall results (qualitative or quantita-tive), overall recommendations, and limitations as pro-vided by the review itself.
Evaluation of importance of determinants andstrength of the evidenceThe results retrieved from the eligible primary studiesincluded in the reviews were summarized combining twoslightly modified grading scales, previously used bySleddens et al. [22]. The first scale, grades the importanceof the determinants, referring only to the consistency anddirection of the associations among the reviews, or the in-dividual primary studies. The second scale, grades thestrength of evidence, referring to the study design usedamong individual primary studies. For its importance, a de-terminant scored a “–” if all reviews, without exception, re-ported a negative association between the determinant and
the outcome and a “-” if the negative association was foundin more than 75% of the reviews or of the original primarystudies. The importance of the determinant was scored a“0” if the results were mixed, or more specifically, if thevariable was found to be a determinant and/or reported anassociation (either positive or negative) in 25 to 75% ofavailable reviews or of the primary studies of these reviews,but not in others. Furthermore, the importance of the de-terminant scored a “+” if a positive association was foundin more than 75% of the reviews or of the included primarystudies and a “++” if a positive association was found in allreviews, without exception. Despite in the literature thecodes “+” and “++” were used in presence of an associ-ation, independently from its direction (e.g., positive ornegative) [22], in the present SLR these codes denote boththe strength and positive direction of the association.The strength of the evidence was described as ‘convin-
cing’ (Ce) if it was based on a substantial (n > 10) numberof longitudinal observational studies showing associationsbetween the determinant and PA. The strength of the evi-dence was defined as ‘probable’ (Pe) if it was based on atleast two cohort studies or five cross-control studiesshowing associations between the determinant and PA.Furthermore, the strength of the evidence was given as‘limited suggestive evidence’ (Ls) if it was based mainly onfindings from cross-sectional studies showing associationsbetween the determinant and PA. Evidence was labelled as‘limited, no conclusive evidence’ (Lns) if the study findings
Fig. 1 Flowchart of the literature research by database
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 4 of 23
-
were suggestive but insufficient to provide an associationbetween the determinant and PA [and if no longitudinaldata available].
Quality assessmentThe methodological quality of the included reviews wasassessed using a modified version of the AMSTAR Check-list [23]. After a consensus between the DEDIPAC-KHpartners, the question (number 11) referring to the pres-ence of any conflict of interest was modified to allow forany conflict of interest to be evaluated in any of the in-cluded reviews but not in the primary studies included ineach review.Two authors belonging to the DEDIPAC-KH inde-
pendently evaluated the included reviews, using thesame methodology of Sleddens et al. [22]. Any uncer-tainty and disagreement was resolved by consultingthree further authors (SB, LC, AP). The eleven criteriawere evaluated and scored with 1 when the criterion wasapplicable to the analyzed review or with 0 when the cri-terion was not fulfilled, not applicable to the analyzed re-view, or could not be answered based on the informationprovided by the review. As a consequence, the total qual-ity score for each included review ranged from 0 to 11.The quality of the review was labelled as weak (score ran-ging from 0 to 3), moderate (score ranging from 4 to 7),or strong (score ranging from 8 to 11).
ResultsSLRs and MAs selection processAs summarized in Fig. 1, the systematic literature searchidentified 17,941 reviews that were potentially relevantfor inclusion in our umbrella SLR. After the removal ofduplicates, 15,147 reviews remained for screening. Afterreading title and abstract, 12,414 and 2,198 reviews, re-spectively, were excluded because they did not meet theinclusion criteria. Thus, a total of 535 full-text reviewswere assessed for eligibility. From these, another 472 re-views were removed because they did not meet the in-clusion criteria. Thus, after the full-text reading phase,the number of reviews eligible for the umbrella reviewwas 63. Of these, 46 reviews did not include behavioraldeterminants of PA. Therefore, the final number of re-views included in the present umbrella SLR on behav-ioral determinants of PA was 17 (15 SLRs and 2 MAs).
Characteristics of the reviews and quality assessmentThe characteristics of the 17 included reviews are summa-rized in Table 2. All the primary studies included in the in-dividual SLRs or MAs referring to non-behavioraldeterminants and PA were not considered. Most of the re-views included primary studies from multiple continents,mostly from Europe, North America, and Australia. Thepredominant study design used among the primary studies
was cross-sectional [12, 13, 15, 16, 24–33]. Nine reviewsincluded prospective and cohort studies, either as the onlyeligible study design [34, 35] or as part of the includedstudies [12, 15, 16, 25, 26, 29, 32]. In six reviews, it was notpossible to retrieve the total population sample size of theincluded primary studies because it was either not reportedor not complete [26, 27, 29, 30, 32, 33]. In the remainingstudies, the total population sample size ranged from 26[12] to 82,918 [13].Eleven reviews referred to primary studies that in-
cluded young people only. Among these, preschool chil-dren aged between 4 and 6 years old were assessed inone review [26], whilst children and adolescents (8–18years) were included in ten of the reviews [14–16, 28–30, 32–35]. Three reviews considered adults older than40 years [24, 25, 27], while three other reviews consid-ered the population as a whole [12, 13, 31]. Finally, thepercentage of the female participants, when reported,ranged from 0 [25] to 100% [16, 25, 35], though thatdata was absent in the majority of the studies [12–14,26–30, 32–34].
Measurements of PAFrom the 17 reviews included, 461 primary studies werefound eligible. Among these, 218 studies from 15 re-views used non-objective measurement methods of PAassessment (e.g., self-report, parental report, direct PAobservation) [12, 13, 15, 16, 24, 25, 27–35]. Objectivemeasurements of PA, either assessed by accelerometeror pedometer, were used in 201 of the eligible primarystudies and were included in nine of the included re-views [15, 16, 28–30, 32–35]. Eleven primary studies in-cluded in four reviews combined objective with non-objective measures of PA [15, 28, 29, 34]. Finally, 31 pri-mary studies from two reviews did not report the exactnumber of the studies that used objective and non-objective measures [14, 26].As reported in Table 3, the majority of the included re-
views evaluated overall PA as an outcome (n = 13) [12–16,24–26, 28, 29, 32, 34, 35]. One review also measuredmoderate-to-vigorous physical activity (MVPA) and activetransport [26] next to overall PA. Two reviews measuredoverall PA/Ex [27, 31], one review measured time-specificPA (i.e., school break time PA and after-school PA) [30]and one review measured recess PA [33].
Categorization of the included determinantsDuring the preliminary phase, 52 behavioral determi-nants of PA were identified. Among those, similar deter-minants were identified to form a single determinant.For example, the determinants ‘TV viewing’, ‘TV viewing/playing video games’, ‘television viewing/sedentary’, ‘enjoytelevision viewing’, ‘video/computer game use’, ‘computer’,‘screen time’, and ‘video games’ were merged into the
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 5 of 23
-
Table
2Characteristicsof
theinclud
edreview
s
Autho
r,Date
(Typeof
review
)[Ref]
Num
berof
eligiblestud
ies
includ
edin
theum
brella
review
/totalnu
mbe
rof
stud
iesinclud
edin
thereview
Con
tinen
t/sof
includ
edstud
ies
Stud
yde
sign
ofinclud
edstud
ies
Totalsam
plesize
ofinclud
edstud
y(Sam
plerang
e)
Age
rang
eor
mean(years)
ofeligiblestud
ies
Gen
der(female,%
rang
e)of
eligible
stud
ies
Year
rang
eof
includ
edstud
ies
BabakusWS,2012
(SLR)[24]
6/38
Europe
(n=6)
In-dep
thinterviews(n=2)
Focusgrou
p(n=1)
Semi-structuredinterviews
(n=1)
N.A.(n=2)
276(30–109)
N.A.(n=2)
40–83
N.A.(n=2)
39–53
N.A.(n=2)
1980–2012
BarnettI,2012
(SLR)[25]
19/19
North
America(n=11)
Europe
(n=6)
Australia(n=2)
Coh
ort(n=14)
Cross-sectio
nal(n=5)
62,455
(51–11,469)
45–99
0–100
1985–2010
Cragg
sC,2011
(SLR)[34]
13/46
North
America(n=11)
Europe
(n=2)
Prospe
ctive(n=13)
13,332
(40–3,878)
4–9(n=1)
10–13(n=8)
14–18(n=4)
N.A.
1986–2009
DeCraem
erM,
2012
(SLR)[26]
7/43
N.A.
Cross-sectio
nal(n=6)
Long
itudinal(n=1)
N.A.
4–6
N.A.
2003–2010
Engb
ergE,2012
(SLR)[12]
32/34
Europe
(n=7)
North
America(n=22)
Australia(n=3)
Cross-sectio
nalretrospective
(n=7)
Prospe
ctivelong
itudinal
(n=25)
276,558
(26–80,944)
17–70
N.A.
1992–2012
KaczynskiA
T,2008
(SLR)[13]
50/50
North
America(n=23)
Europe
(n=18)
Australia(n=3)
Asia(n=1)
Multip
leCon
tinen
ts(n=
1) N.A.(n=4)
Mostly
cross-sectional
381,807
(120–82,918)
6–89
N.A.
1970–2005
Koen
eman
MA,
2011
(SLR)[27]
2/34
Australia(n=1)
Asia(n=1)
Observatio
nal(n=2)
N.A.
40–80
N.A.
2007
Larouche
R,2014
(SLR)[15]
46/73
Europe
(n=25)
North
America(n=9)
SouthAmerica(n=1)
Asia(n=1)
Australia(n=9)
Russia(n=1)
Cross-sectio
nal(n=41)
Prospe
ctive(n=5)
66,489
(103–7,023)
5–17.9
532002–2012
LeeMC,2008
(SLR)[28]
24/32
North
America(n=9)
Europe
(n=10)
Australia(n=5)
Cross-sectio
nal(n=24)
33,756
(88–10,771)
5–18
N.A.
2002–2007
MarshallSJ,2004
(MA)[14]
24/54
N.A.
N.A.
1443,235
(36–20,766)
0–18
N.A.
1990–2002
PearsonN,2014
(MA)[29]
163/163
Multip
leCon
tinen
tsProspe
ctive(n=12)
Cross-sectio
nal(n=147)
N.A.
0–18
N.A.
1987–2013
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 6 of 23
-
Table
2Characteristicsof
theinclud
edreview
s(Con
tinued)
Prospe
ctiveandCross-sectio
nal
(n=4)
Ridg
ersND,2012
(SLR)[33]
3/53
Europe
(n=2)
Australia(n=1)
Cross-sectio
nal(n=2)
N.A.(n=1)
N.A.
5–18
N.A.
2006–2010
Scho
eppe
S,2013
(SLR)[16]
42/52
Europe
(n=25)
North
America(n=9)
Asia(n=1)
Australia(n=6)
Russia(n=1)
Cross-sectio
nal(n=39)
Long
itudinal(n=3)
55,896
(103–6,085)
3–18
39–100
2002–2012
StanleyRM
,2012
(SLR)[30]
6/22
Europe
(n=1)
North
America(n=4)
Australia(n=1)
Cross-sectio
nal(n=6)
N.A.
8–14
N.A.
1997–2010
TzormpatzakisN,
2007
(SLR)[31]
3/36
Europe
(n=3)
Cross-sectio
nal(n=3)
4,213
(171–1,000)
15–89
50–54
2002–2004
Uijtde
willinge
nL,
2014
(SLR)[35]
11/30
Europe
(n=5)
North
America(n=5)
Asia(n=1)
Prospe
ctive(n=11)
11,259
(155–5,451)
4–18
51–100
(n=10)
N.A.(n=1)
2005–2010
vande
rHorstK,
2007
(SLR)[32]
10/57
N.A.
Cross-sectio
nal(n=8)
Prospe
ctive(n=2)
N.A.
4–12
(n=3)
13–18(n=7)
N.A.
1999–2004
Notes:M
A:M
eta-Ana
lysis;SLR:
System
aticLiterature
Review
;N.A.:Not
App
licab
le
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 7 of 23
-
Table
3Results
oftheinclud
edreview
s
Autho
r,Date(Type
ofreview
)[Ref]
Outcome(s)
Determinant(s)
Review
aim
Overallqu
alitativeresults
ofthereview
Overall
quantitative
results
ofthereview
Overalllim
itatio
nsof
the
stud
yOverallRecommen
datio
ns
BabakusWS,2012
(SLR)[24]
OverallPA
Lang
uage
difficulties;lackof
time
Toassess
whatisknow
nabou
tthelevelsof
PAand
sede
ntarytim
eandto
contextualizethese
behaviorsam
ongSouth
Asian
wom
enwith
anim
migrant
backgrou
nd.
SouthAsian
wom
enareless
activethan
theothe
rethn
icgrou
psas
wellascompared
toSouthAsian
males;
know
ledg
eof
PAandits
bene
fitswas
foun
dto
belackingam
ongsouthAsian.
N.A.
Nostandardized
metho
dfor
quality
evaluatio
n;lack
ofde
tails
from
someof
the
includ
edpape
rs;
measuremen
tandde
finition
ofPA
variedwidely;
publicationandresearcher
bias
possibility;significant
heteroge
neity
amon
gstud
ies.
Moreresearch
shou
ldbe
dedicatedto
standardize
objectivePA
measuremen
tandto
unde
rstand
how
toutilize
theresourcesof
the
individu
alsandcommun
ities
toincrease
PAlevelsand
overallh
ealth
ofSouthAsian
wom
en;futureresearch
isne
eded
toassess
levelsof
sede
ntarytim
eand
contextualizesede
ntary
behaviors.
BarnettI,2012
(SLR)[25]
OverallPA
Workafterretirem
entfro
mmainoccupatio
n;retirem
ent
from
strenu
ousoccupatio
n;be
ingretired
for<5years;
lifelon
gparticipationin
PA;
beingmarried
Toexam
inechange
sand
pred
ictorsof
change
sin
PAacross
thetransitio
nto
retirem
ent;whe
ther
these
change
svary
bySES;what
isknow
nabou
tpred
ictors
ofchange
sin
PAacross
the
retirem
enttransitio
n.
Exercise
andleisure-tim
ePA
increasedafterthetransitio
nto
retirem
ent,whe
reas
the
finding
sregardingchange
sin
totalP
Awere
inconclusive;m
enincrease
theirPA
morethan
wom
en;
lower
SESisassociated
with
ade
crease
andhigh
SES
with
anincrease
inPA
.Eviden
ceon
othe
rpred
ictors
was
scarce,o
ften
inconsistent,and
metho
dologically
weak.
N.A.
Eviden
ceon
pred
ictorsof
change
was
scarce
and
metho
dologically
weak;no
lang
uage
orcoun
try
restrictio
ns;p
ublishe
dpe
er-
review
edjournalarticlesas
wellasgray
research
literaturewereinclud
ed;
multid
isciplinaryapproach
contrib
uted
tothe
heteroge
neity
oftheresults
andto
theun
feasibility
ofmeta-analysis.
Furthe
rstud
iesshou
ldinclud
eothe
rmeasuresof
SES,approp
riate
andvalid
PAmeasures,applyclear
andrelevant
defın
ition
sof
retirem
ent,andstud
ypred
ictorsof
PAchange
across
thetransitio
nto
retirem
ent;qu
alitative
stud
iesandlong
itudinal
stud
ieswith
long
erfollow-
uparene
eded
.
Cragg
sC,2011
(SLR)[34]
OverallPA
Vigo
rous
PA;p
articipationin
sportteam
sou
tsidescho
ol;
previous
PA;alcoh
olconsum
ption;
dietaryhabits;
smokingstatus;sed
entary
behavior
Tocollate
thecurren
teviden
cebase,h
ighlight
research
tren
dsand
limitatio
nsin
physical
activity
determ
inants
research,and
synthe
size
the
existin
geviden
ce.
Inconclusive
associations
wererepo
rted
forlarge
prop
ortio
nof
the
determ
inantsexam
ined
;girls
consistentlyrepo
rted
larger
declines
inPA
than
boys
inyoun
gerchildren;
parental
maritalstatuswas
consistentlyshow
nno
tto
beassociated
with
change
inactivity;h
ighe
rlevelsof
self-efficacywereassociated
with
smallerde
clines
comparedto
lower
levelsof
selfefficacyin
olde
rchildren
andadolescents.
N.A.
Possibility
ofpu
blicationbias
(includ
edpu
blishe
dstud
ies
only);he
teroge
neity
instud
ysamples,exposureand
outcom
emeasuresinclud
edin
thisreview
;som
estud
ies
draw
data
from
thesame
coho
rts;semi-q
uantitative
repo
rtingused
inthereview
that
limits
theclassification
oftheassociations.
Furthe
rresearch
shou
ldinclud
eob
jectivemeasures
ofPA
andusepreviously
validated
questio
nnairesto
assess
theinvestigated
determ
inants;m
orehigh
quality
research
isne
eded
inallage
grou
ps,especially
inyoun
gerchildren;
investigationinto
determ
inantsof
change
shou
ldtake
into
accoun
tof
specifıcph
ysicalintensities
such
asminutes
spen
tin
mod
erateor
vigo
rous
physicalactivity.
DeCraem
erM,
2012
(SLR)[26]
OverallPA
,MVPA,
activetransport
Participationin
organized
sports;televisionview
ing/
sede
ntary;en
joytelevision
view
ing;
child
hasno
energy
touseactivetransport
Tosystem
aticallyreview
the
correlates
ofPA
,sed
entary
andeatin
gbe
havior
inpreschoo
lchildren4-6years
old.
Atten
ding
aruralp
rescho
olwas
positivelyassociated
with
physicalactivity;
gend
er,age
and
socioe
cono
micstatus
were
notassociated
with
physical
N.A.
Somelim
itatio
nsregarding
thecoding
ofthe
associationof
thevariables;
new
catego
riesforeach
behavior
weremade.
Future
research
shou
ldfocus
oniden
tifying
thecommon
correlates
ofph
ysicalactivity,
sede
ntarybe
havior
and
eatin
gbe
havior
inpreschoo
l-age
dchildrenso
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 8 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
activity,w
hilean
inde
term
inateresultwas
foun
dforethn
icity;g
ende
randethn
icity
wereno
tassociated
with
sede
ntary
behavior
andinde
term
inate
results
werefoun
dforage
andsocioe
cono
micstatus;
preschoo
lersweremore
physicallyactiveas
wellas
sede
ntaryon
weekdays;
watchingtelevision
was
associated
with
ahigh
erconsum
ptionof
snacks
and
sweetbe
verage
s.
that
better
tailored
interven
tions
couldbe
develope
d.Furthe
rmore,
morelong
itudinalstudies
couldcontrib
utein
draw
ing
strong
erconclusion
son
determ
inantsof
theseEBRBs.
Engb
ergE,2012
(SLR)[12]
OverallPA
Transitio
nto
university;
change
inem
ploymen
tstatus;m
aritaltransition
sandchange
sin
relatio
nships;p
regn
ancy/
having
achild;experiencing
harassmen
tat
work,
violen
ceor
disaster;m
oving
into
aninstitu
tion
Toexam
ineliterature
concerning
theeffectsof
lifeeven
tson
change
sin
PA.
Mostof
thestud
iesreview
edshow
edstatistically
sign
ificant
change
sin
PAassociated
with
certainlife
even
ts;transition
toun
iversity,havingachild,
remarriage
andmassurban
disaster
wereassociated
with
decreasedPA
levels,
whileretirem
entwas
associated
with
increased
PA;experiencingmultip
lesimultane
ouslifeeven
tswereassociated
with
decreasedPA
inmen
and
wom
en;PAisoftenused
aspartof
arehabilitation
prog
rammefordiseases.
N.A.
Self-repo
rted
PAdata
are
likelyto
besomew
hat
limited;
anothe
rlim
itatio
nis
thepo
ssiblecross
contam
inationof
respon
ses
whe
nPA
levelsbe
fore
and
afteralifeeven
twere
assessed
atthesametim
e;thege
neralizationof
stud
yfinding
sto
lower
socioe
cono
micstatus
and
ethn
icminority
popu
latio
ns,
aswellasto
othe
rcoun
tries,
may
belim
itedbe
causePA
andlifeeven
tsvary;
limitatio
nsof
thestud
ies
includ
ethat
PAdata
were
notalwaysrepo
rted
inde
tail;somestud
iesdidno
tanalyze,statistically,the
magnitude
ofchange
inPA
;anothe
rim
portantlim
itatio
nisthat
lifeeven
tstend
tooverlap.
Future
stud
iesshou
ldexam
inege
nder
differences
intheeffectsof
lifeeven
tsandusevalidated
metho
dsin
assessingleisurePA
.Long
itudinalcoh
ortstud
ies
andclinicaltrialsfro
mdifferent
coun
triesand
cultu
resarene
eded
.Studies
with
long
erfollow-ups
are
necessaryto
exam
ineho
wlong
theeffectsof
life
even
tson
PApe
rsist.
KaczynskiA
T,2008
(SLR)[13]
OverallPA
Smoking
Tobe
tter
unde
rstand
the
co-occurrenceof
smoking
andph
ysicalinactivity
inbo
thadultsandyouth.
Inapproxim
ately61%
ofthe
stud
iesof
adultpo
pulatio
ns,
theassociationbe
tween
smokingandPA
was
negative,less
pron
ounced
foryouthandadolescents,
andam
ongmales
than
females.
N.A.
Someauthorsde
fined
asm
oker
assomeo
newho
smoked
atleaston
cein
the
past30
days
whe
reas
othe
rstud
iesused
thecriterio
nof
atleaston
ecigarettepe
rdayover
thepastmon
th.
Furthe
rcomplicating
interpretatio
nsisthefact
that
controlling
foror
includ
ingdifferent
variables
change
stheim
pact
ofthe
"predictor"variables
onthe
"outcome"
measures.Alm
ost
allo
ftheaforem
entio
ned
Researchersho
ping
toim
proveun
derstand
ingof
thejointrelatio
nbe
tween
PAandsm
okingne
edto
collect
data
atbo
ththe
individu
alandorganizatio
n(i.e.,enviro
nmen
t)levels.
Und
erstanding
the
mechanism
sby
which
depression
,lun
gfunctio
n,scho
olsetting,
orothe
rfactorsmed
iate
ormod
erate
thesm
okingandPA
relatio
nshiprequ
ires
measuremen
tof
these
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 9 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
stud
ieswerecross-sectional,
andso
inferringcausality
orthedirectionof
relatio
nships
isne
arim
possible.
variables
atmultip
le(preferablythreeor
more)
pointsin
time.By
expand
ing
thearrayof
investigative
metho
dsem
ployed
,researcherswillbe
better
equipp
edto
unde
rstand
linkage
sbe
tweensm
oking
andPA
andto
design
effectiveinterven
tions.
Koen
eman
MA,
2011
(SLR)[27]
OverallPA
,OverallEx,
OverallPA
/Ex
Baselineactivity
level;
smoking
Tosystem
aticallyreview
determ
inantsof
PAand
exercise
amon
ghe
althy
olde
radults.
Insufficien
teviden
cefor
mostassociations
betw
een
possiblede
term
inantsand
PAor
Ex.
N.A.
Theremay
bepo
ssibility
ofpu
blicationbias;a
wideage
rang
eisappliedthat
might
have
maskedsomeof
the
differences
betw
een
subsam
ples
inside
that
popu
latio
n;they
exclud
edsomespecificsubsam
ples
oftheolde
rpo
pulatio
n;overall
low
quality
ofthestud
ies
includ
ed
Thede
term
inantsof
PAne
edfurthe
rstud
ythat
includ
etheuseof
objective
measuresof
PAandexercise
andvalid
andreliable
measuresof
determ
inants.
Larouche
R,2014
(SLR)[15]
OverallPA
Activescho
oltransport
Tosystem
aticallyassess
the
quality
ofeviden
cewith
respectto
therelatio
nship
betw
eenactivescho
oltransportanddaily
PA,
cardiovascular
fitne
ss,and
body
compo
sitio
n.
Overall,81.6%
ofthestud
ies
show
edpo
sitiveassociations
betw
eenactivescho
oltransportandPA
levelswith
mod
eratequ
ality
ofeviden
ce.
N.A.
Meta-analyses
were
preclude
ddu
eto
thewide
heteroge
neity
instud
ymetho
dologies
andanalyses.
Furthe
r,thefunn
elplot
and
relatedstatisticalmetho
dsforde
tectingpu
blication
bias
couldno
tbe
used
becauseaconsistent
measure
ofeffect
across
stud
ieswas
notavailable.
Future
activescho
oltransportstud
iesshou
ldanalyzewalking
andcycling
separately.Futurestud
ies
couldevaluate
theim
pact
ofexistin
gprog
ramsthat
prom
oteactivescho
oltransporton
PAlevelsand
health
relatedou
tcom
es.
Future
stud
iesshou
ldconsider
assessingactive
scho
oltransportas
acontinuo
usvariable(i.e.,
frequ
ency
×distance).
LeeMC,2008
(SLR)[28]
OverallPA
Activecommutingto
scho
olThisarticlepresen
tsa
system
aticreview
ofthe
literatureon
active
commuting(walking
orbicycling)
andthe
outcom
esof
physical
activity
andweigh
t/ob
esity
inscho
ol-age
children.
Inmultip
lestud
iesin
anu
mbe
rof
coun
tries,a
sign
ificant
association
betw
eenph
ysicalactivity
andwalking
orcyclingto
scho
olhasbe
enno
ted
across
manyagegrou
psof
scho
olchildren.
Thestreng
thof
theassociationismixed
andasummaryestim
ateis
notpo
ssiblebe
causeof
the
heteroge
neity
instud
yde
sign
.Reg
ardlessof
this
association,
itremains
unclearas
towhe
ther
the
commutetrip
itselfisa
sign
ificant
compo
nent
oftotalp
hysicalactivity.
N.A.
Thede
finition
ofactive
commutingandthe
potentialfor
misclassification
bias;the
metho
dology
ofmeasurin
gph
ysicalactivity;
themajority
ofcross-
sectionalstudies.
Future
research
shou
ldfocus
onmeasurin
gtheim
pact
ofinterven
tions
that
succeedin
changing
commuting
behavior
andqu
antifying
the
contrib
utionof
the
commuteto
children’sdaily
activity.
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 10 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
MarshallSJ,2004
(MA)[14]
OverallPA
TVview
ing;
vide
o/compu
ter
gameuse
Toexam
inetheeviden
cethat
sede
ntarybe
haviors
displace
physicalactivity.
Possiblemechanism
slack
supp
ortin
geviden
ceand
claimsthat
TVview
ing,
playingvide
ogames
orusingcompu
tersdisplace
physicalactivity
receivevery
limitedem
piricalsupp
ort.
−0.096(95%
CI=
-0.080,-0.112)
-0.104
(95%
CI=
-0.080,-0.128)
Onlystud
ieswrittenin
English.
Moreexpe
rimen
talresearch
isne
eded
toreplicatethese
finding
sandexplore
possiblemechanism
s.
PearsonN,2014
(MA)[29]
OverallPA
Totalsed
entary
behavior;
compu
ter;ho
mew
ork;
internet;reading
;screen
time;television
;video
games;com
posite
sede
ntary
behavior
Theaim
ofthecurren
tstud
ywas
tosystem
atically
review
andmeta-analysis
peer-reviewed
research
describ
ingtheassociation
betw
eensede
ntarybe
havior
andPA
inchildrenand
adolescents.
Asign
ificant,b
utsm
all,
negativeassociationwas
foun
dbe
tweensede
ntary
behaviorsandph
ysical
activity
inchildrenand
adolescents;in
mod
erator
analyses,studies
that
recruitedsm
allersamples,
employed
objective
metho
dsof
measuremen
tor
wereassessed
tobe
ofhigh
ermetho
dological
quality
repo
rted
strong
erassociations,alth
ough
the
magnitude
ofeffect
remaine
dsm
allo
rsm
allto
mod
erate;sm
allinverse
associations
wereob
served
betw
eenspecificsede
ntary
behaviou
rs,including
Internet
use,screen
time
andTV
view
ing,
andph
ysical
activity.
−0.108(95%
CI=
-0.128,-0.087)
-0.018
(95%
CI=-0.038,0.001)
0.014(95%
CI=-0.043,0.095)
-0.051
(95%
CI=
-0.097,-0.006)
-0.009
(95%
CI=-0.039,0.021)
-0.080
(95%
CI=
-0.101,-0.060)
-0.064
(95%
CI=
-0.084,-0.045)
-0.002
(95%
CI=-0.043,0.040)
-0.265
(95%
CI=
-0.364,-0.165)
Dicho
tomizationof
acontinuo
usconstruct;
searches
wereconfined
tostud
iespu
blishe
din
peer-
review
edjournalsandthose
writtenin
English;
evaluatio
nof
associations
betw
een
sede
ntarybe
havior
andall
PAou
tcom
escombine
d,insteadof
different
PAdo
mains.
Furthe
rresearch
isrequ
ired
toclarify
therelativeand
interactingim
pact
ofPA
and
sede
ntarybe
havior
onweigh
tstatus
inthis
popu
latio
n.
Ridg
ersND,2012
(SLR)[33]
Recess
PASportactivities;p
hysical
conflictfre
quen
cy;d
ay-to-
dayvariability;play
ball
games;sed
entary
activities
Toexam
inethecorrelates
ofchildren’sand
adolescent’sPA
durin
gscho
olrecess
perio
ds.
Highe
rpe
rceived
encouragem
entfro
mparents,pe
ers,andthe
scho
olas
awho
lewas
associated
with
high
erself-
repo
rted
physicalactivity
levelsdu
ringrecess
perio
ds,
particularlyin
adolescents;
overallfacility
provision(i.e.,
sum
offacilitiesavailable)
was
positivelyassociated
with
physicalactivity;a
positiveassociationwas
foun
dbe
tweenun
fıxed
equipm
entandrecess
physicalactivity
inchildren;
boys
aremoreph
ysically
activedu
ringrecess.
N.A.
Themajority
aresm
all-sized
andcross-sectionalstudies;
meta-analysisisdifficultto
obtain
giventhelim
ited
numbe
rof
stud
iesandthe
lack
ofconsistencyin
correlates
assessed
;arang
eof
physicalactivity
measures
have
been
used
;differen
tcut-po
intsandob
servation
system
smay
have
influen
cedthestreng
thof
associations
observed
.
Furthe
rresearch
shou
ldinvestigatecorrelates
ofchildren’sandadolescents’
recess
physicalactivity
for
variables
that
wereno
tinvestigated
frequ
ently
but
indicatedpo
sitive
associations
with
physical
activity;further
research
usingob
jectivemeasuresis
need
edto
determ
ine
adolescents’recess
physical
activity
levels;futureresearch
shou
ldexam
inewhe
ther
increasing
access
toscho
olfacilitiesdu
ringrecess
perio
dsincreasesph
ysical
activity
levelsin
childrenand
adolescents;furthe
rresearch
isne
eded
tode
term
ine
whe
ther
specifıctype
sof
equipm
ent,or
theoverall
availabilityof
unfıxed
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 11 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
equipm
ent,areassociated
with
high
erlevelsof
physical
activity;futureresearch
shou
ldexam
inethe
correlates
ofbo
ysandgirls
physicalactivity
separately;
Moreresearch
isne
eded
concerning
correlates
ofPA
inrecess
perio
d,particularly
inadolescents.
Scho
eppe
S,2013
(SLR)[16]
OverallPA
Inde
pend
entmob
ility
with
outadultsupe
rvision
(Activetravelto
and/or
from
scho
olandactivetravelto
leisuretim
erelatedplaces);
inde
pend
entmob
ility
with
outadultsupe
rvision
(outdo
orplay);activetravel
with
orwith
outadult
supe
rvision(Activetravelto
and/or
from
scho
ol);active
travelwith
orwith
outadult
supe
rvision(Activetravelto
leisuretim
erelatedplaces)
Tosynthe
size
theeviden
ceforassociations
ofchildren’s
inde
pend
entmob
ility
and
activetravelto
vario
usde
stinations
with
physical
activity,sed
entary
behavior
andweigh
tstatus.
Children’sactivetravelto
and/or
from
scho
olwas
consistentlypo
sitively
associated
with
physical
activity;p
ositive
relatio
nships
werefoun
dbe
tweenactive
travelto
leisure-related
destinations
andph
ysical
activity;traveltovario
usplaces
includ
ingscho
ol,
shop
s,cine
mas
andfrien
d’s
homes
werepo
sitive
associated
with
mod
erate-
to-vigorou
sandoverall
physicalactivity;children
weremorelikelyto
increase
physicalactivity
levelswhe
ntheirou
tdoo
rplay
occurred
inde
pend
ently;associatio
nsbe
tweenactivescho
oltravel
andweigh
tstatus
were
inconsistent
across
the
stud
ies.
N.A
Few
stud
iesmeasurin
gactivetravelto
places
othe
rthan
scho
ol;use
ofdifferent
thresholds
forde
fining
objectivelymeasured
sede
ntarybe
havior
inchildren;
themeasuremen
tof
sede
ntaryactivities
inchildrenstilllacks
standardizationand
validation;
Socialand
physicalen
vironm
ents
cond
uciveto
children’s
inde
pend
entmob
ility
and
activetravelwereno
texplored
butcertainlyplay
aroleas
correlates
ofchildren’sph
ysicalactivity,
sede
ntarybe
havior
and
weigh
tstatus.
Future
stud
iesshou
ldseek
toinvestigateassociations
ofinde
pend
entmob
ility
and
activetravelwith
light
intensity
physicalactivity;
children’sdaily
activetravel
tovario
usde
stinations
beyond
thescho
olsetting
requ
iresfurthe
rattention,
aswellasinvestigatingdiverse
sede
ntarybe
haviorsbe
yond
simplyscreen
-based
activities
andde
fining
approp
riate
thresholds
for
objectivelymeasured
sede
ntarybe
havior
inchildren;
thepo
tentialo
finde
pend
entmob
ility
topreven
tsede
ntarybe
havior
andexcessiveweigh
tshou
ldbe
explored
infuture
observationalstudies;studies
shou
ldmeasure
inde
pend
entmob
ility
and
activetravelmore
thorou
ghlyby
considering
frequ
ency
anddu
ratio
nof
inde
pend
entmob
ility
and
activetravel,aswellas
whe
ther
childrenare
travellingalon
eor
with
peers,or
toproxim
alversus
distantde
stinations,and
whe
ther
theob
served
associations
differby
age
andge
nder.
StanleyRM
,2012
(SLR)[30]
Time-specificPA
(i.e.,schoo
lbreak
timePA
and
after-scho
olPA
)
TVview
ing/playingvide
ogames;use
offacilities;
mem
berof
organized
activities
Toiden
tifythecorrelates
ofchildrens'PA(8-14years)
occurringdu
ringthescho
olbreak-tim
eandafterscho
olpe
riods.
Boys
andyoun
gerchildren
tend
tobe
moreactive
durin
gbreak-tim
eandafter-
scho
ol;BMIinfemales
negativelyassociated
with
after-scho
olPA
,age
was
negativelyassociated
in
N.A.
Smalln
umbe
rof
stud
iesthat
vary
inmetho
dological
aspe
cts;po
ssibility
that
some
stud
iesaremisseddu
ring
thesearch
process;majority
ofcross-sectionalstudies;
somestud
iesstratified
Future
stud
iesusinga
contextspecificapproach
shou
ldiden
tifyandrepo
rtspecificfacilitiesrelevant
tothecontextin
questio
n,which
willcontrib
uteto
aclearerun
derstand
ingof
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 12 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
scho
ol-break
andafter
scho
ol;fam
ilyaffluen
ce,
access
toagym,accessto
four
ormorePA
prog
rams
andthecond
ition
ofa
playingfield
wereall
associated
with
scho
olbreak
timePA
inon
estud
y;access
tolooseandfixed
equipm
ent,playgrou
ndmarking
s,size
ofandaccess
toplay
spaceandtheleng
thof
scho
olbreaktim
ewereall
positivelyassociated
with
change
sin
scho
olbreak
timePA
ininterven
tion
stud
ies;in
theafter-scho
olpe
riod,
gend
er(with
boys
againmoreactive),you
nger
age,lower
body
massinde
x(fo
rfemales),lower
TVview
ing/playingvide
ogames
andgreateraccess
tofacilitieswereassociated
with
high
erlevelsof
after-
scho
olPA
intw
oor
more
stud
ies,whileparent
supe
rvisionwas
negatively
associated
with
females’
after-scho
olPA
inon
estud
y.
analyses
bysalient
variables
such
asage,ge
nder
and
intensity
ofPA
,resultin
gin
anover-rep
resentationof
thesestud
iesin
thereview
;therelativelynarrow
age
rang
especified
inthe
curren
treview
isalim
itatio
n.
context-specificPA
;future
stud
iesshou
ldchoo
semeasuremen
ttoolswith
approp
riate
psycho
metric
prop
erties;Needof
high
quality
eviden
ceup
onwhich
PAprom
otionin
youn
gpe
oplecanbe
tailoredto
specificsettings
and
contexts.
TzormpatzakisN,
2007
(SLR)[31]
OverallPA
/Ex
Lack
oftim
e;tim
econstraints;maritalstatus
Toevaluate
theeviden
cefro
mresearch
relevant
toparticipationin
PAand
exercise
inGreece.
Participationin
physical
activity
andexercise
inGreecewas
gene
rally
low;
men
exercise
more
vigo
rouslyandmoreactively
than
wom
en;the
main
reason
sforparticipation
werehe
alth,w
eigh
tcontrol,
fitne
ss,and
stress
managem
ent.Lack
ofleisure
timewas
themainpe
rceived
barrierto
increasing
levelsof
physicalactivity
andexercise;
men
seem
edto
exercise
morefre
quen
tlyandmore
vigo
rouslythan
wom
en;
sing
lepe
opleweremore
likelyto
beph
ysicallyactive
comparedto
thosewho
weremarriedor
divorced
/widow
ed;p
eoplelivingin
ruralareas
weremorelikely
tobe
physicallyactivethan
thoselivingin
urbanareas.
N.A.
Non
eof
thestud
iesused
objectivemeasuremen
tsand
also
they
used
different
self-
repo
rted
estim
ates
ofPA
;lack
ofapprop
riate
useof
theterm
s'exercise'and'PA';
none
oftheinstrumen
tswerevalidated
inGreece;
season
ality
effect,since
season
alvariatio
nexistsin
physicalactivity
participation.
PAprom
otionshou
ldbe
organizedin
asystem
atic
way;a
clearde
finition
ofvariables
isne
eded
;interven
tionstud
iesshou
ldbe
employed
inorde
rto
test
theefficiencyof
policies,
strategies
andcampaigns;
long
itudinalstudies
shou
ldbe
utilizedin
orde
rto
evaluate
tren
dsin
physical
activity
participationandthe
long
-term
effectsof
campaigns;a
clearpatternof
who
isconsidered
physically
activeandinactiveshou
ldbe
determ
ined
accordingto
internationally
establishe
dcriteria
inorde
rto
have
morevalid
andreliabledata;
stud
iesshou
ldno
tconcen
trateon
lyon
leisure
timeph
ysicalactivities
but
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 13 of 23
-
Table
3Results
oftheinclud
edreview
s(Con
tinued)
onthetotalp
hysicalactivity
profile
ofeach
participant.
Uijtde
willinge
nL,
2014
(SLR)[35]
OverallPA
Pastph
ysicalactivity;screen
time
Tosummarizeandup
date
theexistin
gliteratureon
determ
inantsof
PAand
sede
ntarybe
havior
inyoun
gpe
ople.
Mod
erateeviden
cefor
intentionas
ade
term
inant
ofchildren’sph
ysicalactivity,
andage(i.e.,older
children
weremoreactive),ethnicity
(i.e.,b
eing
ofAfrican–
American
descen
tde
term
ined
beingless
physicallyactive)
and
planning
asde
term
inantof
adolescent
physicalactivity.
N.A.
Includ
edstud
iesassessed
overallP
Aon
ly;usedtw
odatabaseson
ly;the
selected
lang
uage
ofpu
blicationwas
Englishon
ly.
Future
researches
shou
ldbe
focusedon
determ
inantsof
child
andadolescent
sede
ntarybe
havior
andon
environm
entald
eterminants
ofph
ysicalactivity
inbo
thchildrenandadolescents,
shou
ldusereliableandvalid
measuresof
both
determ
inantsas
wellasthe
actualbe
haviorsandshou
ldcond
uctprospe
ctivestud
ies
ofhigh
metho
dological
quality.
vande
rHorstK,
2007
(SLR)[32]
OverallPA
PE/schoo
lspo
rts;watching
TV;smoking;
television
/sede
ntarytim
e
Tosummarizeandup
date
theliteratureon
correlates
ofPA
,insufficient
PA,and
sede
ntarybe
havior
inyoun
gpe
ople.
Theresults
from
ourreview
sugg
estthat
correlates
ofph
ysicalactivity
forchildren
arege
nder,self-e
fficacy,
parentalph
ysicalactivity
for
boys,and
parentalsupp
ort.
Correlatesforadolescents'
physicalactivity
were
gend
er,p
aren
taledu
catio
n,attitud
e,self-efficacy,go
alorientation/motivation,
physicaled
ucation/scho
olsportsparticipation,
family
influen
ces,andfrien
dsupp
ort.Gen
der,self-efficacy,
andfamily/paren
talsup
port
wereassociated
with
physicalactivity,b
othin
childrenandadolescents.
N.A.
Publicationbias
may
bepresen
t;po
ssibility
ofmissed
stud
iesas
aresultof
the
search
strategy;the
main
outcom
ewas
overallP
Awith
outothe
rclassifications;
mostly
cross-sectional
stud
iesinclud
ed;b
ecause
ofthevariability,itwas
not
possibleto
assess
theoverall
streng
thof
theassociations.
Moreprospe
ctivestud
iesare
need
edandmoreresearch
includ
ingchildren.
Notes:M
Ameta-an
alysis,SLR
system
aticliteraturereview
,PAph
ysical
activ
ity,P
Eph
ysical
education,
MVP
Amod
erate-to-vigorou
sph
ysical
activ
ity,Exexercise,SES
socio-econ
omicstatus,EBR
Ben
ergy
balance-relatedbe
havior
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 14 of 23
-
determinant ‘screen use’. Differently, other similar deter-minants were identified to form a sub-group with a spe-cific label. For example, the determinants ‘being retired’,‘change in employment status’, and ‘moving into an insti-tution’ were part of the sub-group ‘job-related’. Thus, afinal consensus between authors had been achieved forthis categorization and the final number of behavioraldeterminants was 35 (Table 4). Any uncertainty and dis-agreement was resolved by consulting three further au-thors of the DEDIPAC-KH (BS, CL, PA).
Findings of the reviewsThe findings of the included reviews on the associationsbetween the behavioral determinants and PA, consider-ing different age groups and different types of PA, aresummarized in Table 4.
Preschool childrenOne review specifically assessed the behavioral correlatesof PA in preschool children that were aged between 4and 6 years old [26]. Among the correlates investigated,only those related with ‘screen use’ were negatively asso-ciated with MVPA in all the studies included in the re-view with a limited suggestive level of evidence (−−, Ls).Inconclusive findings were found regarding ‘participa-tion in organized sports’ (0, Lns) and ‘screen use’ (0,Lns) with respect to overall PA and no association wasfound between ‘children have no energy to use activetransport’ and active transport (0, Lns).
ChildrenFew determinants were investigated among childrenaged between 10 and 13 years old in respect to overallPA [34]. ‘Previous PA’ was the only determinant posi-tively associated with overall PA in more than 75% ofthe studies included in the review with a probable levelof evidence (+, Pe). Conversely, ‘participation in sportteams outside school’ (0, Ls) and ‘alcohol consumption’(0, Ls) were not associated with overall PA in children.Considering after-school PA, one review investigated thebehavioral correlates of PA in children between 8 and14 years old [30]. Among the correlates investigated,only those related with ‘screen use’ were negative associ-ated with after-school PA in more than 75% of the stud-ies included in the review with a limited suggestive levelof evidence (−, Ls). ‘Member of organized activities’ (0,Lns) was not associated with after-school PA, whilst for‘use of facilities’ the evidence was limited and inconclu-sive (0, Lns).
AdolescentsTwo reviews investigated the behavioral determinants[34] and correlates [32] of PA in adolescents aged be-tween 14 and 18 years [32, 34]. ‘PE/school sports’ [32]
and ‘previous PA’ [34] were both positively associatedwith overall PA in all the studies included in the review,without exception, both with a limited suggestive levelof evidence (++, Ls). ‘Vigorous PA’ was negatively associ-ated with overall PA in all the studies included in the re-view, without exception, with a limited suggestive levelof evidence (−−, Ls) [34]. No association was found be-tween ‘dietary habits’ and overall PA (0, Ls) [34].
Children and adolescentsTen reviews examined the behavioral determinants andcorrelates of PA for children and adolescents (≤18 yearsold) in relation to overall PA [13–16, 28, 29, 32, 34, 35]and recess PA [33]. Independent mobility and active trans-port’ [15, 16, 28] was positively associated with overall PAin more than 75% of the studies included in the reviews,with a probable level of evidence (+, Pe). ‘Homework’ [29]was positively associated with overall PA in all the studiesincluded in the review, without exception, with a limitedsuggestive level of evidence (++, Ls). In particular, a posi-tive effect size (r = 0.014, 95% CI = −0.043, 0.095) emergedfrom the results of the MA [29]. Conversely, ‘Internet use’,‘reading’, and ‘composite sedentary behavior’ were nega-tively associated with overall PA in all the studies includedin the review, without exception, with a limited suggestivelevel of evidence (−−, Ls). The results of the MA [29]showed small to moderate negative effect sizes for thethree determinants (r = −0.051, 95% CI = −0.097, −0.006;r = −0.009, 95% CI = −0.039, 0.021; r = −0.265, 95%CI = −0.364, −0.165, respectively). Limited evidence wasfound regarding ‘smoking’ (0, Ls) [13, 32, 34], ‘screen use’(0, Ls) [14, 29, 32, 35], ‘sedentary behavior’ (0, Ls) [34], and‘past physical activity’ (0, Ls) [35]. ‘Sports activities’, ‘playballgames’, and ‘physical conflict frequency’ were positivelyassociated with recess PA in all the studies included in thereview, without exception, with a limited suggestive level ofevidence (++, Ls) [33]. ‘Sedentary activities’ was negativelyassociated with recess PA in all the studies included in thereview, without exception, with a limited suggestive level ofevidence (−−, Ls) [33]. No association was found between‘day-to-day variability’ and recess PA (0, Lns) [33].
AdultsThree reviews assessed the behavioral determinants ofPA in adults older than 40 years [24, 25, 27]. ‘Baselineactivity level’ was positively associated with overall PA/Ex in all the studies included in the review, without ex-ception, with a limited suggestive level of evidence (++,Ls) [27]. Conversely, a negatively association was foundbetween ‘language difficulties’ [24] and overall PA/Ex inall the studies included in the review, without exception,with a limited suggestive level of evidence (−−, Ls). For‘being retired’ (0, Lns) and ‘lifelong participation in PA’(0, Ls), the evidence was limited and inconclusive [25].
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 15 of 23
-
Older adultsFor older adults (≥70 years), only one review [27] inves-tigated the association between ‘smoking’ and overallPA/Ex suggesting negative associations in all the studiesincluded in the review, without exception, with a limitedsuggestive level of evidence (−−, Ls).
All agesFive reviews examined the behavioral determinants of PAacross all ages [12, 13, 24, 25, 31]. A negative associationwas found between ‘moving into an institution’ and overallPA/Ex in all the studies included in the review, without ex-ception, with a limited suggestive level of evidence (−−, Ls)[12]. ‘Pregnancy/having a child’ and ‘transition to univer-sity’ were negatively associated with overall PA/Ex in morethan 75% of the studies included in the review with a prob-able level of evidence (−, Pe) [12], while ‘time limitation’was negatively associated with overall PA/Ex in more than75% of the studies included in the reviews with a limitedsuggestive level of evidence (−, Ls) [24, 31]. Finally, for‘marital status’ (0, Lns) [12, 25, 31], ‘smoking’ (0, Lns) [13],‘change in employment status’ (0, Ls), and ‘experiencingharassment at work, violence, and disaster’ (0, Ls) [12] theevidence was limited and inconclusive.
Evaluation of the quality of the SLRs and MAsThe results of the quality assessment using theAMSTAR checklist are reported in Table 5. Among the17 included reviews, the majority were of moderatequality (n = 13) [12, 14–16, 24–28, 30, 33–35], threewere weak [13, 31, 32], and only one was considered to
be of strong quality [29]. Among those reviews of mod-erate quality, five were scored with 4 points [12, 26, 28,30, 33], one with 5 points [16], four with 6 points [14,15, 24, 34], and three with 7 points [25, 27, 35].Few reviews provided all the characteristics of the pri-
mary studies [12, 13, 15, 16, 25, 28, 31, 34, 35] and noneof the reviews provided a full list of included and ex-cluded studies.
DiscussionThis umbrella SLR aimed to provide a comprehensive andsystematic overview of behavioral determinants of PAacross the life course, evaluating importance, strength ofthe evidence, and methodological quality. Through asystematic evaluation of the existing SLRs and MAs, anoverview of the associations between the considered be-havioral determinants and PA is presented to promote ef-fective health enhancing policies and identify gaps forfuture research strategies. To our knowledge, this is thefirst umbrella SLR that examined various potential behav-ioral determinants of PA across the life course.Whether or not individuals choose to engage in PA
behaviors is affected by a number of inter-dependentand multilevel factors. A relevant contribution of theDEDIPAC-KH project was the development of a dynamicframework of determinants of PA, which identified twomain themes related to the person and to the society,respectively [36]. Behavioral determinants are part ofthe individual theme, together with the biological andpsychological determinants, whereas the other themecomprises physical (e.g., environmental), socio-cultural,
Table 4 Summary of the results of the included reviews: the importance of a determinant and its strength of evidence
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 16 of 23
-
socio-economic, and policy determinants. Through thesystematic analysis of scientific evidence, the present find-ings expand the knowledge and the understanding of be-havioral determinants of PA to prepare the ground for acoherent approach towards PA behaviors research and in-terventions to favor active lifestyle choices.Overall, for the majority of determinants, the evidence
was limited and inconclusive or there was not an associ-ation with the different forms of PA. Furthermore, forseveral potential determinants it was not possible to as-certain definite conclusions on their association with PA.These findings urge researchers to consider new andinnovative approaches when investigating factors influ-encing PA behaviors.In relation to the investigated populations, more studies
(n = 11) focused on youth (e.g., ≤18 years) with respect tothose on individuals older than 18 years (n = 5). The rea-son for a greater focus on young populations might bedue to the fact that being physically active at the youngeststages of age is crucial to maintain high levels of PA at theadulthood [32, 37], decreasing the risk factors for non-communicable diseases [2] and having a favorable effecton public health [37]. Nonetheless, the understanding ofthe relationships between behavioral determinants and PAis also crucial among older adults as they are more vulner-able and more likely to have decreasing level of PA and in-creasing levels of sedentary behaviors.Among all the young population groups, 7 determi-
nants were positively associated and 6 determinantswere negatively associated with PA. A probable positiveassociation was evident between ‘previous PA’ and over-all PA in children, with the review showing a moderatequality [34], confirming the importance of PA in child-hood. Moreover, a probable positive association wasfound in three reviews [15, 16, 28] between ‘independentmobility and active transport’ and overall PA in childrenand adolescents, with the reviews showing a moderatequality. Larouche and colleagues [15] argued that it islikely that an active traveler is also more inclined to beactive throughout the day. However, other determinants(e.g., low socioeconomic status and lower income house-holds), which were not considered in their review, mayinfluence the relationship between ‘independent mobilityand active transport’ and PA. Furthermore, potentialbiases due to classification, such as active school trans-port only once per week and direction of active trans-port (e.g., only to school and not from school) could alsoplay a role. Other relevant aspects proposed bySchoeppe and colleagues [16] concern the positive asso-ciations of active travel to destination other than school,the health benefits for children of being involved in‘free-range’ activities, like active travel or play outdoorwithout adult supervision, and the higher frequency ofactive travel from school to home rather than from
home to school. Other determinants related to ‘sport/PAin and out of school’ [32, 33] and to ‘other daily activ-ities’ [29] were positively associated with PA but with alimited suggestive level of evidence. The present findingsare in line with the determinants related to school phys-ical activity and youth sport participation (e.g., “PA Pro-grams in School”, “Availability/Access/Proximity of PAOrganized Sport Facilities/Tools”, “PA Education (atSchool/Work)/Knowledge of Effects of PA”, “GroupActivities (Outdoor/Indoor)”, “Involvement in OrganizedSport”, “Time Spent Outdoor/Playing Spaces”) highlightedby the European PA determinants (EU-PAD) frameworkas having a high priority for research [36]. To improve thestrength of evidence, future research adopting longitudinalresearch designs is strongly needed. Indeed, motor skillsin early childhood, school physical education, and youthsports are important aspects for a holistic development ofchildren and adolescents, so an effective interaction be-tween researchers and policy makers should be establishedto promote of health-enhancing lifestyles in early years, inline with the recommendations of the Expert Group onHealth-Enhancing Physical Activity of the EuropeanCommission [38].The determinants ‘composite sedentary behavior’,
‘internet use’, ‘reading’, ‘sedentary activities’, and ‘vigorousPA’ [26, 29, 33, 34] were negatively associated with PAwith a limited suggestive level of evidence. Even thoughmainly based on cross-sectional studies, the negative as-sociations of ‘composite sedentary behavior’, ‘internetuse’, and ‘reading’ are corroborated by the results of theMA, showing small to moderate negative effect sizes[29]. Regarding ‘vigorous PA’, Bruner and colleagues [39]suggested that a decline in PA over the course of theschool year, in adolescents, is mainly due to a decreasein level of vigorous intensity activity rather than moder-ate intensity activity. Despite a clear dose-response spe-cificity of low, moderate, and vigorous PA intensities oncardiovascular risk, metabolic health, osteoporosis, im-mune function, and mental health [40], the relationshipbetween intensity (but also frequency, duration, andmode of activity) and adherence of PA is controversial[35, 41]. To deepen our understanding of the quality ofthe PA and to provide sound guidelines for PA prescrip-tions, further research on this area is recommended toconsider the intensity of PA on both absolute (e.g., oxy-gen uptake, oxygen uptake relative to body mass, kcal orkJ per minute, and METs) and relative (e.g., percentagesof maximal oxygen uptake, oxygen uptake reserve, heartrate reserve, maximal heart rate, Borg’s Rating ofPerceived Exertion) terms [42].The investigation of determinants in adults showed a
higher proportion of negative associations (n = 6) with re-spect to positive associations (n = 1). Regarding the nega-tive associations, a probable level of evidence was found
Condello et al. International Journal of Behavioral Nutrition and Physical Activity (2017) 14:58 Page 17 of 23
-
only for ‘transition to university’ and ‘pregnancy/having achild’ with respect to overall PA/Ex in all ages >18 years,with the review showing a moderate quality [12]. Further-more, ‘moving into an institution’ [12], ‘time limitation’[24, 31], ‘language difficulties’ [24], and ‘smoking’ [27] werenegatively associated but with a limited suggestive level ofevidence. Conversely, ‘baseline activity level’ was positivelyassociated with overall PA/Ex in adults ≥40 years but witha limited suggestive level of evidence and the quality ofthe review was moderate [27]. Although without a consist-ent convincing evidence among all the determinants, thesefindings highlight the possible impact of specific lifeevents (i.e., ‘transition to university’, ‘pregnancy/having achild’, and ‘moving into an institution’) on the relationshipbetween behavioral determinants and PA and the need foreffective interventions and planning to increase health-enhancing PA.An interactionist approach may help further our un-
derstanding of the role played by specific life events. Dif-ferent life events with shorter-term (e.g., pregnancy) orlonger-term effects (e.g., retirement) might exert theireffects on PA behaviors separately, in different lifephases. Differently, life events, whose shorter- andlonger-lasting effects partially overlap in time (e.g.,pregnancy and marital status) might generate inter-active effects. Moreover, the impact of life events on PAbehaviors might be moderated by other behavioral factors.For instance, the extent to which ‘retirement’ impacts ac-tual PA behavior may be moderated by past PA habits.Moreover, the relationship linking the influence of differ-ent behavioral determinants of PA may be mediational innature, with ‘time limitation’ probably mediating most ofthe life event effects on PA behaviors.However, not only people older than 18 years should
represent a target group population for the investigationof those life events and for considering health promotioninterventions. Adequate attention should be placed also tothose related to ‘job’, such as temporary or long-term un-employment, retirement, change in employment status, aswell as to those concerning the ‘marital status’, such asstarting a new close personal relationship, starting to livewith someone, marriage, separation, divorce, widowhood,interpersonal loss, which are frequent during adulthoodand may interfere and influence PA behaviors.This umbrella SLR demonstrated that some potential
determinants were found to be specific for each agegroup, whilst other determinants can be consideredacross all ages. For the latter, ‘screen use’ and ‘smoking’were the most often investigated determinants, with sixand four reviews examining the associations with severalforms of PA, respectively. Actually, ‘screen use’ comprisesseveral determinants, such as ‘TV viewing’, ‘TV viewing/playing video games’, ‘television viewing/sedentary’, ‘enjoytelevision viewing’, ‘video/computer game use’, ‘computer’,
‘screen time’, ‘video games’, which have been investigatedfor their potential associations with PA from preschoolchildren to adolescents. The negative association be-tween ‘enjoy television viewing’ and MVPA in preschoolchildren emphasized the need for intervention programsto decrease screen related behaviors in the early yearsage group [26]. A similar negative association was alsofound in children with respect to after-school time pe-riods (i.e., after-school PA) [30]. Particularly due to thefact that ‘TV viewing’ may negatively influence PA dur-ing time spent at home with high TV accessibility, inter-vention programs should be encouraged for increasingafter-school PA [30]. Future research should adopt longi-tudinal study designs to investigate the effectiveness oflong-term PA promoting strategies in children afterschool for several components, like fitness, health,cognitive functioning