UBC Faculty Research and Publications

Meta-analysis of internet-delivered interventions to increase physical activity levels Davies, Cally A; Spence, John C; Vandelanotte, Corneel; Caperchione, Cristina M; Mummery, W K Apr 30, 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12966_2011_Article_610.pdf [ 328.56kB ]
JSON: 52383-1.0223966.json
JSON-LD: 52383-1.0223966-ld.json
RDF/XML (Pretty): 52383-1.0223966-rdf.xml
RDF/JSON: 52383-1.0223966-rdf.json
Turtle: 52383-1.0223966-turtle.txt
N-Triples: 52383-1.0223966-rdf-ntriples.txt
Original Record: 52383-1.0223966-source.json
Full Text

Full Text

REVIEW Open AccessMeta-analysis of internet-delivered interventionsto increase physical activity levelsCally A Davies1*, John C Spence2, Corneel Vandelanotte1, Cristina M Caperchione1,3 and W Kerry Mummery1,4AbstractMany internet-delivered physical activity behaviour change programs have been developed and evaluated.However, further evidence is required to ascertain the overall effectiveness of such interventions. The objective ofthe present review was to evaluate the effectiveness of internet-delivered interventions to increase physical activity,whilst also examining the effect of intervention moderators. A systematic search strategy identified relevant studiespublished in the English-language from Pubmed, Proquest, Scopus, PsychINFO, CINHAL, and Sport Discuss (January1990 – June 2011). Eligible studies were required to include an internet-delivered intervention, target an adultpopulation, measure and target physical activity as an outcome variable, and include a comparison group that didnot receive internet-delivered materials. Studies were coded independently by two investigators. Overall effect sizeswere combined based on the fixed effect model. Homogeneity and subsequent exploratory moderator analysis wasundertaken. A total of 34 articles were identified for inclusion. The overall mean effect of internet-deliveredinterventions on physical activity was d= 0.14 (p= 0.00). Fixed-effect analysis revealed significant heterogeneityacross studies (Q= 73.75; p= 0.00). Moderating variables such as larger sample size, screening for baseline physicalactivity levels and the inclusion of educational components significantly increased intervention effectiveness. Resultsof the meta-analysis support the delivery of internet-delivered interventions in producing positive changes inphysical activity, however effect sizes were small. The ability of internet-delivered interventions to producemeaningful change in long-term physical activity remains unclear.Keywords: Physical activity, Internet, Intervention, Meta-analysisIntroductionEstimates from the World Health Organisation [1] sug-gest that approximately 60% of the world’s populationare classified as inactive or insufficiently active to receivehealth benefits. With the increasing burden caused byphysical inactivity and chronic disease, new ways fordelivering behaviour change programs to large numbersof people at low cost are needed. In particular, the internetoffers an innovative medium to produce health behaviourchange in terms of reach, availability and opportunities forinteractive approaches [2]. Statistics demonstrate morethan a 300% increase in internet usage since 2000, withover 1.5 billion internet users worldwide, representing ap-proximately 23.5% of the world’s population [3]. Mostimportant, internet access provides an alternate means tohealth care promotion for individuals who cannot accessstandard care due to physical disability or living in remoteareas [2]. Already a large number of individuals are uti-lising the internet to access health-related information [4],creating an opportunity to develop and deliver health-related behaviour change interventions via the internet.Furthermore, internet-delivered behaviour change inter-ventions are becoming increasingly common for physicalactivity [2,5,6] particularly over the last 10 years [7].Several reviews have examined the effectiveness ofinternet-delivered interventions to produce health relatedbehaviour change among adults [2,6,7] and one has specif-ically examined the components and operationalisation ofcomputer tailored programs across all ages [8]. Further,two meta-analysis were conducted on general healthbehaviours in both children and adults; one focusing oncomputer delivery [9] and another comparing web-basedto non-web-based delivery [5]. The main findings indicatethat: a) short-term behaviour change is more often* Correspondence: cally1@ualberta.ca1Centre for Physical Activity Studies, Institute for Health and Social ScienceResearch, CQ University Australia, Rockhampton, QLD, AustraliaFull list of author information is available at the end of the article© 2012 Davies et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52http://www.ijbnpa.org/content/9/1/52reported than long-term behaviour change; [2,7-9] b)specific intervention elements such as website compo-nents (e.g., tailored content, theoretical design) and inter-active features need to be further evaluated and exploredfor their role in both short-and long-term behaviourchange and increasing website usage [2,6,7], and, c) inter-net-delivered physical activity interventions are moreeffective than true control groups [6]. This meta-analysisexpands upon what is currently known through compre-hensively synthesizing the effect of internet interventionson physical activity levels and variations in physical activityoutcomes due to potential moderating variables. Thesefindings will be useful to determine the current standingof internet interventions and to identify future directionsfor these types of interventions.MethodsInclusion criteriaTo be included in the review, studies were required toprovide an internet-delivered intervention with a focuson increasing physical activity. More specific, studieswere included if they met the following criteria: a) parti-cipants were ≥ 18 years of age; b) the main form of inter-vention delivery was via the internet with either the useof a web page for the delivery and/or exchange of infor-mation, or in the form of email communication; c) phys-ical activity and sample size measures were reported forboth intervention and comparison groups; d) studiescomprised of an experimental design, such as a rando-mized or quasi-experimental design; e) studies includeda non-internet comparison group; and f) articles pub-lished in the English language. Studies that did not meetall inclusion criteria were deemed ineligible and wereexcluded. Additionally, studies that did not provideenough data to allow for the calculation of effect sizeswere deemed ineligible.Search methodA comprehensive search strategy was undertaken toidentify all possible studies for inclusion. The followingelectronic databases were searched: Pubmed, Proquest,Scopus, PsychINFO, CINHAL, and Sport Discuss. Thesearch process was limited to articles published or pro-vided ahead of publication access between January 1990and June 2011. To locate potential studies the followingsearch string was used: (“physical activity” OR exerciseOR “physical fitness” OR walking) AND (internet OR“website delivered” OR “web based” OR,” world wideweb”) AND (education OR behavio* OR intervention). Allreferences including duplicates were then imported intoEndNote (bibliographic software). Reference lists of allrelevant review articles [2,5-13] were manually searchedfor potential studies not yet identified [14].Screening of articlesAfter the removal of duplicates, articles underwent twophases of screening to identify the final sample. Phaseone involved scanning article abstracts for inclusion cri-teria to rule out literature that clearly did not meet theinclusion criteria. In phase two, full text versions of theremaining articles were obtained and further screened toidentify the final set of articles for inclusion.The initial search strategy (excluding duplicates) identified2651 potentially relevant articles. Following title/abstractscreening and screening of relevant review articles, a refer-ence list of 172 potentially relevant articles remained. Afterassessing the full text articles, the final set of articles for in-clusion in this meta-analysis resulted in 34 primary articles(Figure 1) representing 34 unique interventions [15-48].Data extractionOnce finalised for inclusion, studies were collated andcoded independently by two of the researchers (CD andCV), any discrepancies in coding were resolved throughdiscussions [8,49]. The coding framework was pilottested (i.e., both researchers independently coded twotest articles) and refined prior to the first article beingcoded. Characteristics were coded under four generalcategories including: study design (e.g., sample size,physical activity mode targeted [i.e. leisure time or total],duration of intervention), participant characteristics (e.g.,age, gender, population health status, and baseline phys-ical activity levels), intervention features (e.g., number ofintervention contacts, type of tailoring, presence of atheoretical underpinning, interactive features [e.g. goalsetting, quizzes]) and intervention results (sample sizes,physical activity measures and any additional informa-tion to allow for the calculation of effect sizes). The listwas developed based on previous reviews [2,5-8,10,50]and perusal of original research articles published on thetopic. Articles were coded to provide descriptive infor-mation and to allow for subsequent moderator analysis.Tables 1 & 2 contain information of coded characteristicsfor the first three categories. The coding framework isnot exhaustive of all intervention aspects and only char-acteristics reported in sufficient detail across studies aresubsequently reported on.The majority of characteristics included in the firstthree categories, study design, participant characteristicsand intervention features were also included in themoderator analysis. Potential moderating variables (seeTable 3 for the list of moderator variables) were includedif able to effectively extract data and code the variable.For example, the presence of a theoretical underpinningas a moderating variable of intervention effectivenesscould not be examined, as only two studies did not spe-cify a theoretical framework. Additionally, a number ofcontinuous moderating variables were recoded intoDavies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 2 of 13http://www.ijbnpa.org/content/9/1/52categorical variables to enable a more meaningful ana-lysis to be undertaken. For example, although attrition wasinitially coded as a continuous variable it was re-coded as(1) above, or (2) below the average cross-study attritionrate of 23% for the internet intervention groups. Coding ofarticles resulted in an agreement rate of 92% between theresearchers with any discrepancies being resolved.Study quality was assessed based on a previously devel-oped methodological assessment tool, [51] which wasmodified to specifically address quality assessment forinternet-delivered interventions [5]. This evaluation wasbased on five main criteria: a) study design; b) selectionand specification of the study sample; c) specification ofillness/conditions; d) reproducibility of the study; and e)outcomes specification and measurement. During thequality assessment process, studies could receive a scoreup to 18 points; the score obtained by each study wasdivided by 18 and multiplied by 100 to provide a “StudyQuality Percentage”. Study Quality Percentages werethen classified as good (66.7% or higher), fair (between50 to 66.6%) and poor (less than 50%) [52].Data analysisEffect sizes (d) were computed to represent the impactof the internet-delivered interventions on physicalactivity [53]. The effect size (d) is defined as the standar-dised mean difference and allows meaningful compari-sons across measurement instruments. A positive effectsize indicates a more favourable change in physical activityfor the intervention condition. If studies reported statisticsother than means and standard deviations (e.g., F, p),efforts were made to estimate d from the information pro-vided [54]. For studies that used more than one followup measure, effect sizes were calculated using data fromthe time point closest after intervention completion. Thistime-point was used in order to best determine the ac-tual effects of the intervention. A separate effect sizeanalysis was calculated for studies that reported sixmonths or greater post-intervention follow up data;this was done to investigate the effectiveness of inter-net-delivered interventions in producing long-term be-haviour change. To assess the possibility of publicationbias the Egger test was used [55].The fixed effects model was explored in relation tothe summary effect size to estimate the mean distribu-tion of effects. The heterogeneity statistic (Q) was calcu-lated to determine whether studies shared a commoneffect size. Q represents the observed weighted sum ofsquares and df is the expected weighted sum of squares[53]. If heterogeneity was present among the effectSearch of electronic databases using key words 664 Pubmed 280  Proquest 1828 Scopus 293  PsychINFO 585  CINHAL 281  Sports Discus Results: 3931 2479    Excluded based on title and abstract a) Focused on children b) Did not include a physical activity component c) No internet delivery d) Focused on exercise as a form of fitness Removal of duplicates Results: 2651 Screening of title and abstract Results: 172 Manual search of review articles reference list Results: 177 Screening of full text abstract Results: 34 143    Excluded a) Did not meet inclusion criteria b) Review article c) Connected to already included studies d) Unusable outcome measures 5    Additional articles were included from manual search of relevant review articles reference list Figure 1 Selection for studies of internet delivered physical activity interventions.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 3 of 13http://www.ijbnpa.org/content/9/1/52Table 1 Study design and intervention characteristicsSource Country StudyQualityIntention toTreatControlGroupBaselineSampleaAvgAgea%FemaleaHealth Status AdditionalBehavioursBaselineMeasureIntroSessionDuration(weeks)Bosak and Yates,2009 [15]USA Fair Yes SC 22 50.9 27 MetabolicSyndromeNil Phone Face 6Carr et al.,2008 [16] USA Good No Control 67 45 81 Overweight Nil Face Face 16Cook et al.,2007 [17] USA Fair No Int 480 42.01 72 General Nutrition Internet NR 12Dunton andRobertson, 2008 [18]USA Good No Control 156 42.8 100 General Nil Internet NR 12Glasgow et al.,2010b [19]USA Good Yes SC 463 58.4 49.8 Diabetes Self-M Phone Internet 16Grim et al., 2011b [20] USA Fair No SC 233 21.2 72 General Nil NS NR 10Hager et al., 2002b [21] USA Fair No Control 525 42 56 General Nil Internet NR 6Haung et al., 2009 b [22] Taiwan Good No Minimal 146 18 100 General Nil Face NR NRHurling et al.,2007 [23]England Fair Yes Control 77 40.4 67 General Nil Face Face 9Kim and Kang,2006 b [24]South Korea Good No SC 73 55.1 46.6 Diabetes Nil Face Internet 12Kosma et al., 2005 [25] USA Good No Control 151 38.7 72 PhysicalDisabilitiesNil Internet NR 4Leibreich et al.,2009 [26]Canada Good Yes Minimal 49 54.1 59 Diabetes Nil Internet NR 12Lorig et al.,2006 [27] USA Good No SC 958 57.5 71.4 Chronic Disease Self-M Internet NR 6Lorig et al.,2008 [28] USA Good No SC 855 52.35 90.2 COPD Self-M Internet NR 6Lorig et al., 2010 [29] USA Good Yes SC 761 54.3 73 General Self-M Internet NR 6Mailey et al.,2011 [30]USA Good No SC 51 25 68.1 Mental Illness Nil Face Face 10Marshall et al.,2003 [31]Australia Good Yes Int 655 43 51 General Nil Phone NR 8McConnon et al.,2007 [32]England Fair No SC 221 45.8 77 Overweight Weight loss Face Face 52McKay et al.,2001 [33]USA Good No Minimal 78 52.3 53 Diabetes Nil Internet Internet 8Morgan et al.,2009 [34]Australia Fair Yes Control 65 35.9 0 Overweight Weight loss Face Face 12Morgan et al.,2011 [35]Australia Good Yes True 110 44.4 0 Overweight Weight loss Face Face 14Motl et al., 2011 [36] USA Good No Control 54 45.85 90 MultipleSclerosis Nil Mail Internet 12Napolitano et al.,2003 [37]USA Good No Control 65 42.8 86 General Nil Phone NR 12Nguyen et al.,2008 [38]USA Good Yes Int 50 69.5 44 COPD Self-M Face Face 26Daviesetal.InternationalJournalofBehavioralNutritionandPhysicalActivity2012,9:52Page4of13http://www.ijbnpa.org/content/9/1/52Table 1 Study design and intervention characteristics (Continued)Ornes and Randsell,2007b [39]USA Fair Yes Minimal 112 20.6 100 General Nil Face Face 4Parrott et al., 2008 [40] USA Good Yes Control 170 20.2 38 General Nil Face NR 2Plotnikoff et al.,2005 [41]Canada Good No Control 2121 44.9 73.5 General Nutrition Internet NR 12Skar et al.,2011 b [42]Scotland Fair Yes True 1273 22.8 63.7 General Nil Internet Internet 8Smith et al.,2009 [43] USA Good No Control 41 43.5 80.5 Overweight Nutrition Face Face 16Spittaels et al.,2007 b [44] Belgium Good Yes Control 434 42.4 66.1 General Nil Internet NR 26Steele et al.,2007 b [45]Australia Good Yes Int 192 38.7 86 General Nil Face NR 12Wadsworth andHallam, 2010 [46]USA Fair No Control 91 NS 100 General Nil Face Face 26Winnett et al.,2007 b [47] USA Good No Control 1071 52.17 67 General Nutrition Face NR 12Zutz et al.,2007 [48] Canada Good No Control 15 58.5 20 Cardiac Rehab Nutrition Face Face 12Abbreviations: avg, average; COPD, chronic obstructive pulmonary disease; Self-M, self-management; Int, intervention; Intro, introductory; NR, not reported; SC, standard care.a If studies include more than two groups, information presented is inclusive of all groups included in the study; b Studies include more than two groups.Daviesetal.InternationalJournalofBehavioralNutritionandPhysicalActivity2012,9:52Page5of13http://www.ijbnpa.org/content/9/1/52sizes, further analysis was undertaken to examine studylevel moderating factors of physical activity outcomes[53]. The Bonferroni correction factor was applied toadjust the alpha value required for statistical signifi-cance within each of the three moderator categories(study design, participant characteristics, interventionfeatures).ResultsDescription of included studiesArticles were published from 2001 to 2011, with themodal year being 2007 (n = 8). Of the 34 studies, 21 werefrom the United States, 4 from Australia, 3 from Canada,2 from England, and 1 each from Belgium, Taiwan,Scotland and South Korea. At baseline the 34 studiesTable 2 Intervention featuresSource Tailored Theory Interactive Features Attrition (%) Logins PsycImpAll IntBosak and Yates, 2009 [15] Limited SCT AC, Edu, ER, Fac, FB, GS, Q, SM, UC 14 17 NR YesCarr et al., 2008 [16] Limited TTM Edu, ER, Fac, FB, GS, Q, SM, 52 62 NR NRCook et al., 2007 [17] Nil SCT, SOC GS 13 15 NR YesDunton and Robertson, 2008 [18] Full TTM, HBM Edu, ER, FB 15 16 NR NoGlasgow et al., 2010b [19] Nil SCT, Self-M, SEM Edu, ER, Fac, FB, GS, Q SM, UC 17 20 28 YesGrim et al., 2011b [20] Nil SCT, Edu Q, UC 28 24 NR YesHager et al., 2002b [21] Limited TTM FB 23 24 NR YesHaung et al., 2009 b [22] Limited TTM AC, Edu, ER, FB, Q, SC, SM, UC 12 NR NR YesHurling et al., 2007 [23] Full Other AC, Edu, ER, FB, GS, SC,SM, UC, NR NR 26.1 YesKim and Kang, 2006 b [24] Limited TTM AC, FB, GS, UC, NR NR NR NoKosma et al., 2005 [25] Limited TTM AC, Edu, FB, ER, UC, 50 54 NR NoLeibreich et al., 2009 [26] Limited SCT AC, Edu, ER, Fac, FB, SM, UC, 10 8 NR YesLorig et al., 2006 [27] Limited Self-M AC, Edu, ER, Fac, FB, GS, Q, UC 19 22 26.5 NoLorig et al., 2008 [28] Full Other AC, Edu, ER, Fac, FB, GS, Q, SM, UC, 24 29 31.5 YesLorig et al., 2010 [29] Nil Self-M AC, Edu, ER, Fac, FB, GS, SC, SM, UC 15 20 NR YesMailey et al., 2011 [30] NIL SCT FB, GS, SM, ER, UC, Edu 9 13 NR YesMarshall et al., 2003 [31] Limited SOC ER, FB, GS, Q 22 24 NR NRMcConnon et al., 2007 [32] Limited Nil ER, FB 31 51 15.8 NRMcKay et al., 2001 [33] Full Self-M, SEM AC, Fac, FB, GS, SM 13 8 8.9 NoMorgan et al., 2009 [34] Nil SCT AC, Fac, FB, GS, SM 17 18 120 NRMorgan et al., 2011 [35] Nil SCT Edu, FB, GS, SM 19 19 NR YesMotl et al., 2011 [36] Nil SCT AC, Edu, ER, Fac, FB, GS, SC, SM, UC, 11 15 8.6 YesNapolitano et al., 2003 [37] Limited SCT, SOC Edu, ER, Q 12 30 NR YesNguyen et al., 2008 [38] Full SCT, Self-M, Other Edu, ER, Fac, FB, GS, SC, SM 24 31 59 YesOrnes and Randsell, 2007b [39] Nil SCT Edu, ER, FB, GS, SM, 7 NR NR NRParrott et al., 2008 [40] Nil TPB AC 0 0 NA YesPlotnikoff et al., 2005 [41] Nil SCT, TPB, TTM, PMT Nil 18 NR NA YesSkar et al., 2011 b [42] Nil TPB FB 42 44 NR YesSmith et al., 2009 [43] Limited TTM Edu, ER, Fac, FB, GS, Q, SM NR NR NR NRSpittaels et al., 2007 b [44] Full TPB, SOC AC, Edu, ER, FB, GS 34 40 NR NRSteele et al., 2007 b [45] Nil SCT, Self-M AC, Edu, ER, Fac, Q, SM, UC, 15 10 11.8 NRWadsworth and Hallam, 2010 [46] Nil SCT Edu, ER, Fac, FB, GS, Q, SM, UC 22 24 NR YesWinnett et al., 2007 b [47] Nil SCT Edu, FB, GS, SM, UC, 15 15 NR NoZutz et al., 2007 [48] Nil Nil Edu, Fac, FB, Q, SC, UC 13 0 50 YesAbbreviations: Psyc Imp, psychological improvements; All, overall attrition for both groups; Int, attrition for internet intervention group only; NS, not reported; SCT,social cognitive theory, TTM, transtheoretical model; SOC, stages of change, HBM, health belief model; Self-M, self-management; SEM, social ecological model; TPB,theory of planned behaviour; PMT protection motivation theory; AC, asynchronous communication; Edu, education; ER, email reminders; Fac, facilitator; FBfeedback; GS, goal setting; Q, quiz; SC, synchronous communication; SM, self-monitoring; UC, updated content,.Note: Limited tailoring was defined as those interventions that mentioned the inclusion of some tailored materials, but did not deliver a comprehensively tailoredintervention as the main component of the intervention. Presence of education material was defined as interventions that delivered structured educationalmaterial targeting physical activity knowledge. Psychological improvements are present where statistically significant improvements on any psychologicalmeasures is reported in the intervention group (e.g. self-efficacy, attitudes).Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 6 of 13http://www.ijbnpa.org/content/9/1/52Table 3 Summary statistics and effect sizes by moderator variable for change in physical activity as a result of internet-delivered interventionsVariable Qb No. d+ SE 95%CI QWStudy DesignPhysical activity 0.07 73.68aMain outcome 25 0.14a 0.03 0.09/0.19 69.95aSecondary outcome 9 0.15a 0.04 0.07/0.23 3.73Design 0.11 73.64aRandomised Trial 9 0.13a 0.04 0.05/0.21 25.36aRandomised Controlled Trial 25 0.16a 0.03 0.09/0.19 48.25aStudy Quality 0.47 73.28aFair 10 0.13 0.05 0.02/0.20 29.18aGood 24 0.15a 0.02 0.10/0.20 44.10aSample Size 13.14a 13.14a<35 per group 15 0.40a 0.08 0.25/0.55 17.92≥35 per group 19 0.12a 0.02 0.07/0.16 42.70aPhysical Activity Mode 0.08 73.68aLeisure time 15 0.14a 0.03 0.08/0.19 20.32Overall 19 0.15a 0.04 0.08/0.22 53.36aAdditional Behaviours 0.05 73.70aNo 21 0.15a 0.04 0.08/0.22 66.05aYes 13 0.14c 0.03 0.08/0.19 7.65Intervention Duration 2.01 66.72a0–6 weeks 8 0.11a 0.04 0.03/0.19 24.63a7–12 weeks 17 0.13a 0.03 0.08/0.19 35.45a13+ weeks 8 0.21a 0.06 0.09/0.33 6.65Internet and/or Email 0.51 73.25aInternet and email 21 0.16a 0.04 0.09/0.23 34.12Only internet OR email 13 0.13a 0.03 0.08/0.18 39.13aComparison Group 10.50 63.25aIntervention group 4 0.03 0.06 −0.08/0.14 1.76Minimal intervention 4 0.43a 0.12 0.21/0.66 5.80Standard care 9 0.16a 0.04 0.09/0.23 23.23aControl group 17 0.14a 0.03 0.07/0.20 32.46Intervention Attrition 4.59 39.82Below average (<23%) 16 0.16a 0.03 0.10/0.23 17.68Above average (>22%) 12 0.06 0.04 −0.01/0.13 22.14Participant CharacteristicsAge 0.42 71.00a< 45 years 19 0.13a 0.03 0.07/0.18 46.38a> 44 years 14 0.15a 0.03 0.09/0.22 24.61Gender 0.92 72.83a<60% female sample 12 0.10 0.05 0.01/0.19 39.57a>59% female sample 22 0.15a 0.02 0.10/0.20 33.26aHealth Status 4.13 69.62aGeneral population 17 0.11a 0.03 0.06/0.17 44.06aDavies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 7 of 13http://www.ijbnpa.org/content/9/1/52Table 3 Summary statistics and effect sizes by moderator variable for change in physical activity as a result of internet-delivered interventions (Continued)Chronic diseased 12 0.19a 0.04 0.11/0.28 22.25Overweight 5 0.28a 0.11 0.07/0.48 3.31Physical Activity Level 8.83a 64.92aNot screened for 25 0.12a 0.02 0.08/0.16 54.29aSedentary 9 0.37c 0.08 0.21/0.52 10.63Intervention FeaturesIntervention Contacts 1.06 72.57aLess than 10 22 0.13a 0.03 0.07/0.18 63.03a10 or more 10 0.18a 0.04 0.10/0.25 9.54Tailored 1.61 72.14aComprehensive tailoring 6 0.13 0.06 0.02/0.24 1.92Limited tailoring 12 0.09 0.04 0.02/0.18 39.71aNo tailoring 16 0.16a 0.03 0.11/0.22 30.51SCT 6.85 66.91aYes 16 0.20a 0.03 0.14/0.27 20.54No 18 0.09a 0.03 0.03/0.15 46.37aTTM 0.80 72.95aYes 9 0.11a 0.03 0.04/0.19 34.90aNo 25 0.15a 0.03 0.10/0.21 38.05Education Components 8.02a 65.73aYes 24 0.20a 0.03 0.14/0.26 32.50No 10 0.08 0.03 0.01/0.14 33.23aGoal Setting 1.05 72.70aYes 19 0.16a 0.03 0.10/0.22 40.23aNo 15 0.12a 0.03 0.06/0.12 32.47aSelf-Monitoring 3.85 69.91aYes 18 0.20a 0.04 0.13/0.27 25.70No 16 0.11a 0.03 0.06/0.16 44.21aEmail Reminders 0.11 73.64aYes 22 0.15a 0.03 0.09/0.21 34.27No 12 0.13a 0.03 0.07/0.19 39.37aUpdated Content 4.79 68.96aYes 17 0.19a 0.03 0.13/0.26 34.68aNo 17 0.10a 0.03 0.04/0.16 34.28aQuizzes 0.10 73.66aYes 12 0.15a 0.04 0.08/0.22 21.16No 22 0.14a 0.03 0.08/0.19 52.49aAsynchronous Communication 0.58Yes 15 0.16 0.04 0.09/0.23 32.37 aNo 19 0.13 0.03 0.08/0.18 40.79 aAbbreviations: Qb, heterogeneity between; No., number of studies; d+, weighted average effect size; SE, standard error; CI, confidence interval; Qw, heterogeneitywithin; SCT, social cognitive theory; TTM, transtheoretical model.a represents significance according to Bonferroni correction factor adjusted for each category (study design: P=< .005; participant characteristics: P=< .013 andintervention features P=< .005).Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 8 of 13http://www.ijbnpa.org/content/9/1/52included a total sample size of 11,885 across all groups(including sample size from studies that included morethan two groups) or 9,638 participants for the twocomparison groups being examined in the current meta-analysis. Physical activity was the primary behaviourtargeted in 25 (74%) of the studies.The average duration of interventions was 12.64 weekswith a range of 2–52 weeks. Overall attrition (20%) wasreported in 31 of the included studies with an attrition of23% experienced in the intervention groups as reported in28 studies. Although all studies used internet delivery, vari-ation existed in the type of delivery employed. Specifically,21 studies (62%) included a combination of internet andemail, 9 (26%) used internet only and the remaining 4(12%) used email only. The type of comparison group usedalso varied; with 26 studies (76%) using a true/standardcare control group; while 4 studies used an alternate inter-vention (12%) and 4 studies used a minimal interventioncomparison group (12%). In terms of study quality, none ofthe studies were rated as poor; 10 (29%) were rated fair,and 24 (71%) as good quality.Nine (26%) studies used a sample population that wereclassed as inactive (not meeting the national physicalactivity guidelines according to their respective countryrecommendations). The other 25 articles (74%) did notscreen for activity status. The majority of studies repre-sented the general population (n = 17; 50%), with theremaining studies involving population with overweight(n = 5; 15%); Type 2 diabetes (n = 4; 11%); arthritis (n = 1;3%); cardiac rehabilitation (n = 1; 3%); metabolic syn-drome (n = 1; 3%); physical disabilities (n = 1; 3%);chronic disease (n = 1; 3%); multiple sclerosis (n = 1; 3%);a diagnosed mental illness (n = 1; 3%) and cardio ob-structive pulmonary disorder (n = 1; 3%). The averageage represented across studies was 43.06 years, 65% ofthe overall sample was female and, among the 18 articlesthat reported on ethnicity, 92% of the sample was Cauca-sian. The number of times participants logged in to thestudy website was reported for only 11 (32%) of theincluded studies. The average number of logins per per-son/per week was 3.08. Tables 2 & 3 outline study char-acteristics in more detail.Indication of resultsThe estimated overall mean effect of internet-deliveredinterventions on physical activity was d= 0.14 (p< 0.001;Figure 2), suggesting that internet-delivered interven-tions had a small but significantly greater impact onphysical activity change than the comparison conditions.The results of the Egger test revealed that publicationbias was present (p< 0.001). Thus, as recommended bySterne and colleagues [56], no statistical methods wereused to correct for publication bias, as corrections wouldbe based on assumptions and therefore could producepotentially flawed results. Homogeneity tests from thefixed-effect analysis revealed significant heterogeneityacross studies (Q= 73.75; p< 0.001). The overall meaneffect for sustained physical activity at least 6 monthspost- intervention (n = 11) resulted in a small but signifi-cant effect size d= 0.11 (p< 0.01).Study DesignAmong the 10 study design variables (Table 3), post-intervention sample size (Qb (1) = 13.1, p< 0.001) wasthe only significant moderator of internet-delivered inter-ventions on physical activity. Specifically, studies thatincluded less than 35 participants per group at the post-intervention follow up measure (d=0.40) demonstratedlarger effect sizes than the studies that had a post-inter-vention sample size of 35 or greater (d=0.12).Participant CharacteristicsOf the four participant characteristic variables (Table 3),initial physical activity level (Qb (1) = 8.83, p< 0.05) wasfound to be significant moderator of physical activitychange. Studies that screened participants and includedonly participants classified as sedentary or insufficientlyactive produced a greater effect size (d= 0.37) than stud-ies that did not screen participants for physical activity(d= 0.12).Intervention FeaturesThe inclusion of educational components was the onlysignificant moderator (Qb (1) = 8.02, p< 0. 005) of phys-ical activity change among the 11 intervention featurescoded (Table 3). Specifically, interventions consisting ofeducational components producing a larger effect size(d= 0.20) than interventions that did not (d= 0.08).DiscussionThe magnitude of the overall effect size indicates thatinternet-delivered programs have a small but positiveeffect on physical activity (d= 0.14). This finding is con-sistent with previous narrative reviews suggesting theability of internet-delivered interventions to producemodest effects on physical activity [7,13]. The results alsosuggest these interventions produce variable effects onphysical activity, as evident by the variance in effectsacross individual studies, which has also been highlightedby previous research [2,7,13]. Despite significant variabil-ity in intervention effects and the presence of possiblemoderator variables, the combined intervention out-comes resulted in improved short-term physical activity,with a smaller effect being found for longer-term behav-iour change. This study presents the first meta-analyticalreview to extensively examine the impact of moderatorson the effectiveness of internet-delivered interventions toincrease physical activity.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 9 of 13http://www.ijbnpa.org/content/9/1/52The effect size is similar to recent findings from ameta-analysis investigating the effect of physical activityinterventions for healthy adults across all modes of de-livery [57]. The review did not specifically examineinternet-delivered interventions but found the effect sizeacross all delivery modes (d = 0.19) was similar to othermediated interventions (email and telephone) to in-crease physical activity levels (d= 0.15) [57]. However,the effect size was smaller than face-to-face interven-tions (d= 0.29) [57]. A previous review also attempted toconduct a meta-analysis on the effects of distance inter-ventions to increase physical activity, but due to hetero-geneity and poor study quality opted to do a systematicnarrative review instead [50]. Since the cut-off date forthe search strategy of that study [50], an additional 26studies were published that were included in the currentmeta-analysis. Furthermore, due to the increase in pub-lications and higher quality of these additional studies, itwas possible to conduct a moderator analysis in thecurrent study.Internet-delivered interventions have the potential toproduce small but significant increases in physical activ-ity levels. Given the potential breadth of delivery, thepublic health impact of producing small changes inphysical activity across a population has the potential forlarge positive changes at the population level [58]. Thisfinding has even greater potential when considering theincrease in effectiveness when specifically targeting asedentary population (d= 0.37). Sedentary individuals areat higher risk of developing a number of chronic condi-tions and facing premature morbidity and mortality [1],hence it is encouraging to observe the effect sizes arelarger for interventions targeting insufficiently activeindividuals. The potential implications for populationlevel change remain tempered by questions as to whetherthese small effect sizes are clinically relevant. Addition-ally, the moderator analysis indicated that studies with alarger sample size have a smaller effect (d= 0.12) and itmay be argued that these effect sizes are more represen-tative of the actual effect of internet-delivered interven-tions on physical activity levels. Furthermore despiteinternet delivered interventions making claims aboutusing the internet for reaching large populations atlow cost, surprisingly few have evaluated their cost-effectiveness. We could not examine this factor in thecurrent meta-analysis, and it is recommended thatfuture studies evaluate the cost-effectiveness in terms ofdevelopment, maintenance and breadth of delivery ofSummary differenceFigure 2 Forrest plot of effect sizes representing effect on physical activity behaviour.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 10 of 13http://www.ijbnpa.org/content/9/1/52internet-delivered interventions to allow comparisons totraditional modes of delivery.Recently, meta-analyses examining the effectiveness ofbehavioural medicine interventions have been critiquedfor including underpowered studies (less than 35 partici-pants per group). It was argued that small studies aremore likely to be published if they find positive results,increasing the likelihood of publication bias [55]. Theoutcomes of the present review confirm the presenceof bias. Studies that do not find statistically significantresults may not be being published either due to author’snot attempting to publish or journals not accepting thearticle for publication. Regardless, the findings should beinterpreted within this limitation. There remains a needto include well-designed, randomised controlled trialsthat include adequate sample size at baseline to allow forattrition and to maintain sufficient power at the sched-uled follow-up periods.Increasing physical activity levels and maintaining thebehaviour are important in terms of generating sustainedhealth benefits [59]. Eleven studies followed participantsfor six months or more post-intervention. This resultedin a small effect size (d= 0.11) and further investigationis needed to determine the effectiveness of internet-delivered interventions to produce long-term change inphysical activity levels. Future interventions should in-clude long-term follow up measures for physical activityto identify overall effectiveness. Additionally, studies thattargeted physical activity only or physical activity andadditional behaviours produced similar effect sizes. Thisfinding is supported by previous research [7] and pro-vides justification for further investigation given the roleof multiple behavioural risk factors in the developmentof non-communicable chronic disease.Identifying factors that enhance intervention effectivenesscan inform the development of future research to producegreater physical activity change. For instance, includingstructured educational materials that involved the exchangeof information intended to influence physical activity wasthe only intervention feature found to moderate interven-tion effectiveness. Providing education has previouslyshown to be an effective behaviour change technique to in-crease physical activity among chronically ill adults [10].Some of the intervention features examined, such as emailreminders and updated content, are not necessarilyincluded in an attempt to optimise intervention effective-ness but are incorporated as part of intervention design toenhance exposure to the program [60]. Previous researchhas demonstrated that intervention features such as thenumber of intervention contacts, tailored content, goal set-ting, self-monitoring and updated content enhance inter-vention effectiveness [7]. Although these interventionfeatures were not found to be significant moderators onphysical activity change in the current analysis, futureresearch should attempt to isolate the impact of specificintervention features on physical activity change throughimplementing high quality study designs (such as rando-mised controlled trials, having adequate sample sizes, usingvalidated instruments to measure study outcomes and ap-propriate reporting of results) that will allow suchinvestigation.Based on 11 of the included studies, the average num-ber of logins was 3.08 per-person-per-week whichexceeds the traditional one-contact–per-week that iscommon among face-to-face interventions. Due to lackof data, it was not possible to analyse the decline in web-site logins over time, however it is an issue often identi-fied throughout the literature [7,60]. Several of thestudies that did track decay of logins over time havereported that the majority of intervention logins oc-curred within the first few weeks of the study durationwith a very steep decline shortly thereafter [31,61,62],hence it is an important issue. Enhancing participantengagement is directly related to increased exposure tothe intervention and research has identified a cleardose–response relationship between the intensity of theintervention and resulting behaviour change [61,63]. Itis therefore apparent that maintaining website engage-ment is an important factor in relation to the potentialeffectiveness [31,33,61,64,65]. Due to limited reportingof login and other website engagement data, the impactof intervention features on program engagement couldnot be evaluated. Results from a recent systematic re-view suggest that intervention features, such as provi-sions for peer or counsellor support, email and/orphone contact with visitors and regular website updates,were related to increased exposure in internet-deliveredhealthy lifestyle interventions [60]. However, theauthors noted significant issues in the consistency ofreporting engagement measures and recommended thatfuture interventions apply consistent engagement mea-sures across studies [60]. Consistent with the currentstudy, Brouwer and colleagues [60] identified that themost common measure of website engagement was theaverage number of logins to the intervention website,therefore future web-based interventions should alsoconsistently report on it. Nevertheless, additional mea-sures of website engagement are required, such as de-cline in website logins over time and time exposed tospecific intervention elements.LimitationsThe results presented should be interpreted within thelimitations of the current meta-analysis. Due to the lownumber of articles we were unable to combine variablesand conduct a meta-regression when examining the im-pact of moderators. This is important as features thatwere significant moderators also displayed significantDavies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 11 of 13http://www.ijbnpa.org/content/9/1/52heterogeneity, which indicated the presence of other fac-tors still influencing the effect size. In this respect mod-erator analysis should be interpreted with caution. Anumber of features were also unable to be examined as partof the moderator analysis due to insufficient reportingamong primary articles. Another limitation is that the ma-jority of studies mostly incorporated the use of self-reportmeasures for physical activity, included largely Caucasianand well education samples. Although the majority werebased on valid and reliable measures, the better measure isstill to use objective measures. In terms of internet deliverythis can prove challenging if programs are being widely dis-seminated. Finally, the effect size cannot be translated torepresent a more meaningful and clinically relevant changein physical activity level (for example minutes of moderateor vigorous physical activity) as studies vary widely in theform of measured used for assessing behaviour.ConclusionOverall, the findings demonstrate internet-deliveredinterventions are effective in producing small but signifi-cant increases in physical activity. Although the effectsare small, producing such changes in behaviour across alarge population can have powerful implications at a pop-ulation level. To fully harness the potential of internet-delivered interventions to produce population-wide effectsit is important that interventions target insufficiently activeindividuals as well as more diverse populations. Programsare continuing to evolve with advances in technology, butit is imperative that rigorous high quality research con-tinues to explore the effectiveness of internet deliveredinterventions. Future research in the area should focus onthe various aspects of internet-delivered interventions thatincrease the engagement and retention of the target audi-ence so as to better understand the elements that willenhance effectiveness of this type of intervention.Competing interestsThe authors declare that they have no competing interests to disclose.AcknowledgementsDr. Vandelanotte was supported by a National Health and Medical ResearchCouncil of Australia (#519778) and National Heart Foundation of Australia(#PH 07B 3303) post-doctoral research fellowship. The other authors reportno financial disclosures. No other funding was received for this study.Author details1Centre for Physical Activity Studies, Institute for Health and Social ScienceResearch, CQ University Australia, Rockhampton, QLD, Australia. 2SedentaryLiving Lab, Faculty of Physical Education and Recreation, University ofAlberta, Edmonton, AB, Canada. 3Faculty of Health and Social Development,University of British Columbia, Kelowna, BC, Canada. 4Faculty of PhysicalEducation and Recreation, University of Alberta, Edmonton, AB, Canada.Authors’ contributionsCD led the search strategy and was assisted by CV in screening and codingarticles. CD, JS and CV were responsible for the analysis and interpretation ofthe data. All authors participated in the study design, drafting of themanuscript and critical revisions. All authors read and approved the finalmanuscript.Received: 22 August 2011 Accepted: 30 April 2012Published: 30 April 2012References1. Physical inactivity: A global public health problem [http://www.who.int/dietphysicalactivity/factsheet_inactivity/en/]2. Norman GJ, Zabinski MF, Adams MA, Rosenberg DE, Yaroch AL, Atienza AA:A review of eHealth interventions for physical activity and dietarybehavior change. Am J Prev Med 2007, 33:336–345. e316.3. Internet usage statistics: The big picture [http://www.internetworldstats.com/stats.htm]4. Fox S: The social life of health informaiton. In Book The social life of healthinformaiton. City: Pew Internet & American Life Project; 2011.5. Wantland DJ, Portillo CJ, Holzemer WL, Slaughter R, McGhee E: Theeffectiveness of web-based vs. non-web-based interventions: a meta-analysis of behavioral change outcomes. J Med Internet Res 2004, 6:e40.6. Van den Berg MH, Schoones JW, Vliet Vlieland TPM: Internet-based physicalactivity interventions: a systematic review of the literature. J Med InternetRes 2007, 9:e26.7. Vandelanotte C, Spathonis KM, Eakin EG, Owen N: Website-deliveredphysical activity interventions: a review of the literature. Am J Prev Med2007, 33:54–64.8. Lustria MLA, Cortese J, Noar S, Glueckauf R: Computer-tailored healthinterventions delivered over the web: Review and analysis of keycomponents. Patient Educ Couns 2009, 74:156–173.9. Portnoy D, Scott-Sheldon L, Johnson B, Carey M: Computer-deliveredinterventions for health promotion and behavioural risk reduction: Ameta-analysis of 75 randomised controlled trials, 1988–2007. Prev Med2008, 47:3–16.10. Conn VS, Hafdahl AR, Brown SA, Brown LM: Meta-analysis of patienteducation interventions to increase physical activity among chronicallyill adults. Patient Educ Couns 2008, 70:157–172.11. Kroeze W, Werkman A, Brug J: A systematic review of randomized trials onthe effectiveness of computer-tailored education on physical activity anddietary behaviors. Ann Behav Med 2006, 31:205–223.12. Noar S, Benac C, Harris M: Does tailoring matter? Meta-analytic review oftailored print health behavior change interventions. Psychol Bull 2007,133:673–693.13. Webb T, Joseph J, Yardley L, Michie S: Using the internet to promotehealth behavior change: A systematic review and meta-analysis of theimpact of theoretical basis, use of behvaior change techniques, andmode of delivery. J Med Internet Res 2010, 12:e4.14. Higgins J, Green S: Cochrane handbook for systematic reviews ofinterventions. In Book Cochrane handbook for systematic reviews ofinterventions (Editor ed.^eds.), vol. 5.0.1, 5.0.1 edition. City: The CochraneCollaboration; 2008.15. Bosak KA, Yates B, Pozehl B: Effects of an internet physical activityintervention in adults with metabolic syndrome. West J Nurs Res 2009,32:5–22.16. Carr LJ, Bartee RT, Dorozynski C, Broomfield JF, Smith ML, Smith DT:Internet-delivered behavior change program increases physical activityand improves cardiometabolic disease risk factors in sedentary adults:Results of a randomized controlled trial. Prev Med 2008, 46:431–438.17. Cook RF, Billings DW, Hersch RK, Back AS, Hendrickson A: A field test of aweb-based workplace health promotion program to improve dietarypractices, reduce stress, and increase physical activity: Randomizedcontrolled trial. J Med Internet Res 2007, 9:e17.18. Dunton GF, Robertson TP: A tailored internet-plus-email intervention forincreasing physical activity among ethnically-diverse women. Prev Med2008, 47:605–611.19. Glasgow RE, Kurz D, King D, Dickman JM, Faber AJ, Halterman E, Wooley T,Toobert DJ, Strycker LA, Estabrooks PA, et al: Outcomes of Minimal andModerate Support Versions of an Internet-Based Diabetes Self-ManagementSupport Program. J Gen Intern Med 2010, 25:1315-1322..20. Grim M, Hortz B, Petosa R: Impact evaluation of a pilot web-basedintervention to increase physical activity. Am J Health Promot 2011,25:227–230.21. Hager R, Hardy A, Aldana S, George J: Evaluation of an Internet, stage-based physical activity intervention. American Journal of Health Education2002, 33:329.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 12 of 13http://www.ijbnpa.org/content/9/1/5222. Huang SJ, Hung WC, Chang M, Chang J: The effect of an internet-based,stage-matched message intervention on young taiwanese women’sphysical activity. J Health Commun 2009, 14:210–227.23. Hurling R, Catt M, De Boni M, Fairley BW, Hurst T, Murray P, Richardson A,Sodhi JS: Using internet and mobile phone technology to deliver anautomated physical activity program: Randomized controlled trial. J MedInternet Res 2007, 9:e7.24. Kim C-J, Kang D-H: Utility of a web-based intervention for individuals withtype 2 diabetes. CIN: Computers, Informatics, Nursing 2006, 24:337–345.25. Kosma M, Cardinal B, McCubbin J: Longitudinal effects of a web-basedphysical activity motivational program among adults with physicaldisabilities. Res Q Exerc Sport 2005, 76:A116.26. Liebreich T, Plotnikoff R, Courneya K, Boule N: Diabetes NetPLAY: Aphysical activity website and linked email counselling randomizedintervention for individuals with type 2 diabetes. Int J Behav Nutr Phys Act2009, 6:18.27. Lorig KR, Ritter PL, Laurent DD, Plant K: Internet-based chronic disease self-management: a randomized trial. Med Care 2006, 44:964–971.28. Lorig KR, Ritter PL, Laurent DD, Plant K: The internet-based arthritis self-management program: A one-year randomized trial for patients witharthritis or fibromyalgia. Arthritis Care Res 2008, 59:1009–1017.29. Lorig K, Ritter PL, Laurent DD, Plant K, Green M, Jernigan VB, Case S: Onlinediabetes self-management program: a randomized study. Diabetes Care2010, 33:1275–1281.30. Mailey EL, Wójcicki TR, Motl RW, Hu L, Strauser DR, Collins KD, McAuley E:Internet-delivered physical activity intervention for college students withmental health disorders: A randomized pilot trial. Psychol Health Med2010, 15:646–659.31. Marshall A, Leslie ER, Bauman AE, Marcus BH, Owen N: Print versus websitephysical activity programs: A randomized trial. Am J Prev Med 2003, 25:88–94.32. McConnon A, Kirk SF, Cockroft JE, Harvey EL, Greenwood DC, Thomas JD,Ransley JK, Bojke L: The Internet for weight control in an obese sample:results of a randomised controlled trial. BMC Health Serv Res 2007, 7:206.33. McKay HG, King D, Eakin EG, Seeley JR, Glasgow RE: The Diabetes NetworkInternet-Based Physical Activity Intervention: A randomized pilot study.Diabetes Care 2001, 24:1328–1334.34. Morgan PJ, Lubans DR, Collins CE, Warren JM, Callister R: The SHED-ITrandomized controlled trial: Evaluation of an internet-based weight-lossprogram for men. Obesity 2009, 17:2025–2032.35. Morgan PJ, Collins CE, Plotnikoff RC, Cook AT, Berthon B, Mitchell S, Callister R:Efficacy of a workplace-based weight loss program for overweight maleshift workers: The Workplace POWER (Preventing Obesity Without Eatinglike a Rabbit) randomized controlled trial. Prev Med 2011, 52:317–325.36. Motl RW, Dlugonski D, Wojcicki TR, McAuley E, Mohr DC: Internetintervention for increasing physical activity in persons with multiplesclerosis. Mult Scler 2011, 17:116–128.37. Napolitano MA, Fotheringham M, Tate D, Sciamanna C, Leslie E, Owen N,Bauman A, Marcus B: Evaluation of an internet-based physical activityintervention: A preliminary investigation. Ann Behav Med 2003, 25:92–99.38. Nguyen HQ, Donesky-Cuenco D, Wolpin S, Reinke LF, Benditt JO, Paul SM,Carrieri-Kohlman V: Randomized controlled trial of an internet-based versusface-to-face dyspnea self-management program for patients with chronicobstructive pulmonary disease: Pilot study. J Med Internet Res 2008, 10:e9.39. Ornes L, Ransdell LB: Web-based physical activity intervention for college-aged women. Int Elec J Health Educ 2007, 10:126–137.40. Parrott MW, Tennant LK, Olejnik S, Poudevigne MS: Theory of plannedbehavior: Implications for an email-based physical activity intervention.Psychol Sport Exerc 2008, 9:511–526.41. Plotnikoff R, McCargar LJ, Wilson PM, Loucaides CA: Efficacy of an e-mailintervention for the promotion of physical activity and nutrition behaviorin the workplace context. Am J Health Promot 2005, 19:422–429.42. Skar S, Sniehotta FF, Molloy GJ, Prestwich A, Araújo-Soares V: Do briefonline planning interventions increase physical activity amongstuniversity students? A randomised controlled trial. Psychol Health 2011,26:399–417.43. Smith DT, Carr LJ, Dorozynski C, Gomashe C: Internet-delivered lifestylephysical activity intervention: Limited inflammation and antioxidantcapacity efficacy in overweight adults. J Appl Physiol 2009, 106:49–56.44. Spittaels H, De Bourdeaudhuij I, Vandelanotte C: Evaluation of a website-delivered computer-tailored intervention for increasing physical activityin the general population. Prev Med 2007, 44:209–217.45. Steele R, Mummery K, Dwyer T: Using the internet to promote physicalactivity: A randomized trial of intervention delivery modes. J Phys ActHealth 2007, 4:245–260.46. Wadsworth DD, Hallam JS: Effect of a web site intervention on physicalactivity of college females. Am J Health Behav 2010, 34:60–69.47. Winett RA, Anderson ES, Wojcik JR, Winett SG, Bowden T: Guide to health:Nutrition and physical activity outcomes of a group-randomized trial ofan internet-based intervention in churches. Ann Behav Med 2007, 33:251–261.48. Zutz A, Ignaszewski A, Bates J, Lear SA: Utilization of the internet to delivercardiac rehabilitation at a distance: A pilot study. Telemed J E Health 2007,13:323–330.49. de Boer A, Taskila T, Ojajarvi A, van Dijk F, Verbeek J: Cancer Survivors andUnemployment: A Meta-analysis and Meta-regression. JAMA 2009,301:753–762.50. Jenkins A, Christensen H, Walker JG, Dear K: The effectiveness of distanceinterventions for increasing physical activity: a review. American journal ofhealth promotion: AJHP 2009, 24:102–117.51. Haynes RB, Taylor DW, Sackett DL: Compliance in health care. Baltimore:Johns Hopkins University Press; 1979.52. Newell S, Bowman JA, Cockburn JD: Can compliance with non-pharmacological treatments for cardiovascular disease be improved? PrevMed 2000, 18:253–261.53. Borenstein M, Hedges L, Higgins J, Rothstein: Introduction to Meta-Analysis.United Kingdom: John Wiley & Sons, Ltd; 2009.54. Ray JW, Shadish WR: How interchangeable are different estimators ofeffect size? J Consult Clin Psychol 1996, 64:1316–1325.55. Egger M, Smith GD, Schneider M, Minder C: Bias in meta-analysis detectedby a simple, graphical test. BMJ 1997, 315:629–634.56. Sterne JAC, Egger M, Smith GD: Investigating and dealing with publicationand other biases in meta-analysis. BMJ 2001, 323:101.57. Conn VS, Hafdahl AR, Mehr DR: Interventions to increase physical activityamong healthy adults: meta-analysis of outcomes. Am J Public Health2011, 101:751–758.58. Rose G: Sick individuals and sick populations. Int J Epidemiol 2001, 30:427.59. Goldberg JH, King AC: Physical activity and weight management acrossthe lifespan. Annu Rev Public Health 2007, 28:145–170.60. Brouwer W, Kroeze W, Crutzen R, De Nooijer J, De Vries N, Brug J, OenemaA: Which Intervention Characteristics are Related to More Exposure toInternet-Delivered Healthy Lifestyle Promotion Interventions? ASystematic Review. J Med Internet Res 2011, 13:e2.61. Van den Berg MH, Ronday HK, Peeters AJ, der Harst EM Voogt-van,Munneke M, Breedveld FC, Vliet Vlieland TPM: Engagement andsatisfaction with an Internet-based physical activity intervention inpatients with rheumatoid arthritis. Rheumatology 2007, 46:545–552.62. Webber KH, Tate DF, Ward DS, Bowling JM: Motivation and Its Relationshipto Adherence to Self-monitoring and Weight Loss in a 16-week InternetBehavioral Weight Loss Intervention. J Nutr Educ Behav 2010, 42:161–167.63. Draper M, Jennings J, Barón A, Erdur O, Shankar L: Dose-effect relationshipsin brief therapy based on a nationwide college counseling center sample. InBook Dose-effect relationships in brief therapy based on a nationwide collegecounseling center sample (Editor ed.^eds.), vol. 1. City: The Counseling &Mental Health Center, The University of Texas at Austin; 2000.64. Glasgow RE, Boles SM, McKay HG, Feil EG, Barrera M: The D-Net diabetesself-management program: Long-term implementation, outcomes, andgeneralization results. Prev Med 2003, 36:410–419.65. Leslie E, Marshall A, Owen N, Bauman A: Engagement and retention ofparticipants in a physical activity website. Prev Med 2005, 40:54–59.doi:10.1186/1479-5868-9-52Cite this article as: Davies et al: Meta-analysis of internet-deliveredinterventions to increase physical activity levels. International Journal ofBehavioral Nutrition and Physical Activity 2012 9:52.Davies et al. International Journal of Behavioral Nutrition and Physical Activity 2012, 9:52 Page 13 of 13http://www.ijbnpa.org/content/9/1/52


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items