The effectiveness of pedometers to increase physical activity: a systematic review and meta-analysis.

The effectiveness of pedometers to increase physical activity: a systematic review and meta-analysis. Dan Mason (1), Laura Lamming, Ed Wilson, Vijay Singh GC, Sally Pears, Katie Morton, Maaike Bijker, Stephen Sutton, Wendy Hardeman. (1) The Behavioural Science Group, Institute of Public Health, Cambridge, UK

Pedometers and physical activity Small, inexpensive Popular and usable

Current evidence Systematic review of pedometers to increase physical activity Bravata et al (2007) JAMA; 298:19; 2296-2304 8 RCTs: pedometers increased steps by 2419±1394 per day 18 observational studies: increased steps by 2183±613 per day Lots of different study designs Interventions typically multi-component; do not isolate pedometer effect

Case for a review A doubling of studies of pedometers and physical activity since 2007 Many more even if we only focus include RCTs More studies = greater power to examine heterogeneity Isolate pedometer effect Examine effects of different intervention components (e.g. step goals)

Additional vs residual components Additional components Substantial addition to pedometer e.g. PA counselling, access to PA website Pedometer NOT isolated if these vary between groups Residual components Instructions to increase PA Given a PA goal Asked to record daily PA Pedometer IS isolated if only these things vary between groups

Aims of the review Aim #1: identify pooled effect size for pedometer intervention compared with non-pedometer control on free-living physical activity Aim #2: identify pooled effect size as above, but for the case where the pedometer is isolated from other intervention components Aim #3: identify whether effect size varies with the presence or otherwise of residual intervention components: Telling participants to increase their physical activity Giving participants a physical activity goal Asking participants to record their daily physical activity

Inclusion criteria Intervention group receives pedometer as intervention tool Participants are adults (>18yo) Free-living: pedometer worn in everyday life (e.g. not during prescribed exercise class; not inpatients; not lab studies) A physical activity or fitness outcome is reported More than one group; i.e. no cohort studies, no within-subjects experiments Exclude if: Controls wear an open pedometer throughout the intervention period

Search MEDLINE: 1471 PsycINFO: 760 SCI-EXPANDED/SSCI/CPCI- S/CPCI-SSH: 1791 EMBASE: 1540 Cochrane: 371 ERIC: 34 CINAHL: 678 SCOPUS: 3107

Preliminary subgroup: select by measurement Objective physical activity measure Self-report physical activity measure N=27 N=15 N=26 Fitness outcomes only N=6

Preliminary subgroup: select by measurement Objective physical activity measure Self-report physical activity measure N=27 N=15 N=26 Fitness outcomes only N=6

Preliminary subgroup: select by design Pedometer isolated e.g. ped+x vs X Pedometer not isolated e.g. ped+x vs Y N=23 N=6 N=45

Preliminary subgroup: select by design Pedometer isolated e.g. ped+x vs X N=23 N=6 Pedometer not isolated e.g. ped+x vs Y N=45

Preliminary subgroup: final selection N=13 Objective physical activity measure N=29 N=13 Pedometer isolated e.g. ped+x vs X N=16

Study characteristics 13 studies had objective measures and isolated the pedometer 1397 participants randomised, 1053 analysed Mean age 55.2yo; study means from 20.6 to 77.3yo, but mostly >40yo 88.6% female (6 studies only recruited women) Follow-up for physical activity outcome typically around 4-12 weeks; some longer term Note: very preliminary, no double coding beyond initial selection of 74 studies

Residual intervention components Instructions vary Logging PA varies Goals vary Baker 2011 Carr 2008 X X X Du Vall 2004 X X X Gray 2009 X X Hultquist 2007 McMurdo 2010 Ornes 2007 X Samuels 2011 Sugden 2008 Vallance 2007 X Yates 2009 Eastep 2004 X Strath 2011 X X

Results pedometer effect isolated, N=13 Study or Subgroup Carr 2008 Eastep 2004 Strath 2011 DuVall 2004 Samuels 2011 Ornes 2007 Hultquist 2007 Gray 2009 Sugden 2008 Baker 2011 Yates 2009 McMurdo 2010 Vallance 2007 Pedometer Control Std. Mean Difference Std. Mean Difference Mean 9,668 63,421 5,754 322.93 8,877 8,890 8,491 10,182 108,738 9,573 8,995 147,142 8,109 SD 1,556 23,265 1,756 88 2,394 1,172 2,187 4,081 54,728 2,587 2,402 56,458 4,302 Total 5 12 16 17 14 30 23 24 27 23 33 60 172 Mean 6,618 52,505 5,000 318.33 7,921 6,673 9,073 6,709 113,822 10,279 7,922 139,714 8,070 SD 1,779 25,787 1,756 88 1,808 1,093 2,513 2,918 62,337 2,615 4,424 57,041 3,606 Total 5 9 15 16 29 29 20 24 18 23 29 53 166 Weight 2.9% 5.9% 7.1% 7.3% 7.6% 7.8% 7.9% 8.0% 8.0% 8.1% 8.8% 9.8% 10.8% IV, Random, 95% CI 1.65 [0.10, 3.20] 0.43 [-0.45, 1.31] 0.42 [-0.29, 1.13] 0.05 [-0.63, 0.73] 0.47 [-0.18, 1.11] 1.93 [1.30, 2.55] -0.24 [-0.85, 0.36] 0.96 [0.36, 1.56] -0.09 [-0.68, 0.51] -0.27 [-0.85, 0.31] 0.30 [-0.20, 0.81] 0.13 [-0.24, 0.50] 0.01 [-0.20, 0.22] IV, Random, 95% CI Total (95% CI) 456 436 Heterogeneity: Tau² = 0.22; Chi² = 49.25, df = 12 (P < 0.00001); I² = 76% Test for overall effect: Z = 2.30 (P = 0.02) 100.0% 0.36 [0.05, 0.67] -4-2 0 2 4 Favours control Favours pedometer

Results pooled steps per day (1000s), N=10 Study or Subgroup Eastep 2004 Carr 2008 Gray 2009 Yates 2009 Baker 2011 Hultquist 2007 Samuels 2011 Strath 2011 Vallance 2007 Ornes 2007 Pedometer Control Mean Difference Mean Difference Mean 9.06 9.668 10.182 8.995 9.573 8.491 8.877 5.754 8.109 8.89 SD 3.324 1.556 4.081 2.402 2.587 2.187 2.394 1.756 4.302 1.172 Total 12 5 24 33 23 23 14 16 172 30 Mean 7.501 6.618 6.709 7.922 10.279 9.073 7.921 5 8.07 6.673 SD 3.684 1.779 2.918 4.424 2.615 2.513 1.808 1.756 3.606 1.093 Total 9 5 24 29 23 20 29 15 166 29 Weight 5.3% 8.1% 8.3% 9.1% 10.2% 10.6% 10.6% 11.3% 12.8% 13.7% IV, Random, 95% CI 1.56 [-1.50, 4.61] 3.05 [0.98, 5.12] 3.47 [1.47, 5.48] 1.07 [-0.73, 2.88] -0.71 [-2.21, 0.80] -0.58 [-2.00, 0.84] 0.96 [-0.46, 2.37] 0.75 [-0.48, 1.99] 0.04 [-0.81, 0.88] 2.22 [1.64, 2.80] IV, Random, 95% CI Total (95% CI) 352 349 100.0% Heterogeneity: Tau² = 1.41; Chi² = 40.02, df = 9 (P < 0.00001); I² = 78% Test for overall effect: Z = 2.38 (P = 0.02) 1.08 [0.19, 1.96] -4-2 0 2 4 Favours control Favours pedometer Estimate 1080 steps per day advantage with pedometer

Results step goal subgroup analysis, N=13 x 3 subgroups Results goal subgroups Pedometer Control Std. Mean Difference Std. Mean Difference Study or Subgroup Mean SD Total Mean SD Total Weight IV, Random, 95% CI IV, Random, 95% CI 4.3.1 Ped+goal compared to no-ped no-goal group (isolated) Carr 2008 Strath 2011 DuVall 2004 Gray 2009 Subtotal (95% CI) 9,668 5,754 322.93 10,182 1,556 1,756 88 4,081 5 16 17 24 62 6,618 5,000 318.33 6,709 1,779 1,756 88 2,918 Heterogeneity: Tau² = 0.15; Chi² = 5.89, df = 3 (P = 0.12); I² = 49% Test for overall effect: Z = 2.18 (P = 0.03) 5 15 16 24 60 2.9% 7.1% 7.3% 8.0% 25.3% 1.65 [0.10, 3.20] 0.42 [-0.29, 1.13] 0.05 [-0.63, 0.73] 0.96 [0.36, 1.56] 0.61 [0.06, 1.16] 4.3.2 Ped+goal compared to no-ped+goal group (isolated) Samuels 2011 Hultquist 2007 Sugden 2008 Baker 2011 Yates 2009 McMurdo 2010 Vallance 2007 Subtotal (95% CI) 8,877 8,491 108,738 9,573 8,995 147,142 8,109 2,394 2,187 54,728 2,587 2,402 56,458 4,302 14 23 27 23 33 60 172 352 7,921 9,073 113,822 10,279 7,922 139,714 8,070 1,808 2,513 62,337 2,615 4,424 57,041 3,606 Heterogeneity: Tau² = 0.00; Chi² = 5.13, df = 6 (P = 0.53); I² = 0% Test for overall effect: Z = 0.52 (P = 0.60) 29 20 18 23 29 53 166 338 7.6% 7.9% 8.0% 8.1% 8.8% 9.8% 10.8% 61.0% 0.47 [-0.18, 1.11] -0.24 [-0.85, 0.36] -0.09 [-0.68, 0.51] -0.27 [-0.85, 0.31] 0.30 [-0.20, 0.81] 0.13 [-0.24, 0.50] 0.01 [-0.20, 0.22] 0.04 [-0.11, 0.19] 4.3.3 Ped+no goal compared to no-ped no-goal (isolated) Eastep 2004 Ornes 2007 Subtotal (95% CI) 63,421 8,890 23,265 1,172 12 30 42 52,505 6,673 25,787 1,093 Heterogeneity: Tau² = 0.97; Chi² = 7.45, df = 1 (P = 0.006); I² = 87% Test for overall effect: Z = 1.62 (P = 0.11) 9 29 38 5.9% 7.8% 13.7% 0.43 [-0.45, 1.31] 1.93 [1.30, 2.55] 1.21 [-0.26, 2.68] Total (95% CI) 456 436 Heterogeneity: Tau² = 0.22; Chi² = 49.25, df = 12 (P < 0.00001); I² = 76% Test for overall effect: Z = 2.30 (P = 0.02) Test for subgroup differences: Chi² = 6.14, df = 2 (P = 0.05), I² = 67.4% 100.0% 0.36 [0.05, 0.67] -2-1 0 1 2 Favours control Favours pedometer

Discussion Identified 74 studies with ped vs no-ped RCT designs Mix of self-report and objective measures Wide variety of interventions Sufficient numbers to isolate pedometer effect but not without allowing some residual intervention components Preliminary analysis suggests overall increase in physical activity in pedometer groups when (somewhat) isolated against a non-pedometer control Subgroup analyses possible on PA goals, PA logs, goal review etc. But these components are all correlated to some extent

Next steps Data extraction on the full set ongoing Extracting info on intervention intensity and mode of delivery Additional intervention components (e.g. counselling) Residual intervention components (e.g. step goals) Further analyses will include: Self-report measures Studies that only have pedometer groups with additional components i.e. not isolated

Acknowledgements This presentation presents independent research funded by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research Programme (Grant Reference Number RP-PG-0608-10079). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

Search Disjunction of Examples Notes Device terms pedometer$, accelerometer$, activity monitor$ Physical activity terms (physical* NEAR/5 activ*), (life-style* NEAR/5 activ*), inactiv*, walk* Based on previous pedometer review Based on previous Cochrane physical activity review (Foster et al 2009) Design terms clinical trial, random* Based on Haynes et al (2005) scientifically strong studies filter Search in: MEDLINE, PsycINFO, SCI-EXPANDED, SSCI, CPCI-S, CPCI-SSH, EMBASE, Cochrane Library, ERIC, CINAHL, SCOPUS

Intervention examples Study Group Intervention details Gray 2009 Intervention (pedometer) Hultquist 2007 Control (no pedometer) Intervention (pedometer) Control (no pedometer) 12 week pedometer walking programme plus an individualised walking schedule to gradually increase daily step count by 3000s/d on 5 days pw, by week 6, then maintain for a further 6 weeks. Asked to maintain normal walking levels Instructed to take 10,000 steps per day, keeping daily PA log, reporting to lab weekly for log (step count) collection over 4 weeks Instructed to take 30min walk per day, keeping daily PA log, reporting to lab weekly for log (minutes) collection over 4 weeks

Intervention examples Study Group Intervention details Vallance 2007 SR (standard recommendation) PM (print materials) PED (pedometer) COM (combination pedometer and print materials) Received standard recommendation to perform 30min MVPA on 5d pw; those already hitting this were advised to increase Received SR plus a guidebook "exercise for health" specific to breast cancer survivors Received SR plus a pedometer to wear during waking hours and 12 week step calendar to record daily steps Received SR plus guidebook as PM group and pedometer and step calendar as PED group