This paper is in the following e-collection/theme issue:

Digital Health Reviews (1289) Fitness Trackers and Smart Pedometers/Accelerometers (561) mHealth for Symptom and Disease Monitoring, Chronic Disease Management (1399) Wearable Devices and Sensors (888)

Published on in Vol 24, No 8 (2022): August

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/36337, first published .
Association Between Patient Factors and the Effectiveness of Wearable Trackers at Increasing the Number of Steps per Day Among Adults With Cardiometabolic Conditions: Meta-analysis of Individual Patient Data From Randomized Controlled Trials

Association Between Patient Factors and the Effectiveness of Wearable Trackers at Increasing the Number of Steps per Day Among Adults With Cardiometabolic Conditions: Meta-analysis of Individual Patient Data From Randomized Controlled Trials

Association Between Patient Factors and the Effectiveness of Wearable Trackers at Increasing the Number of Steps per Day Among Adults With Cardiometabolic Conditions: Meta-analysis of Individual Patient Data From Randomized Controlled Trials

Authors of this article:

Alexander Hodkinson1, 2 Author Orcid Image ;   Evangelos Kontopantelis1, 2, 3 Author Orcid Image ;   Salwa S Zghebi1, 2 Author Orcid Image ;   Christos Grigoroglou1, 2 Author Orcid Image ;   Brian McMillan1, 2 Author Orcid Image ;   Harm van Marwijk4 Author Orcid Image ;   Peter Bower1, 2 Author Orcid Image ;   Dialechti Tsimpida1, 2 Author Orcid Image ;   Charles F Emery5 Author Orcid Image ;   Mark R Burge6 Author Orcid Image ;   Hunter Esmiol6 Author Orcid Image ;   Margaret E Cupples7 Author Orcid Image ;   Mark A Tully8 Author Orcid Image ;   Kaberi Dasgupta9, 10 Author Orcid Image ;   Stella S Daskalopoulou9, 11 Author Orcid Image ;   Alexandra B Cooke9 Author Orcid Image ;   Ayorinde F Fayehun12 Author Orcid Image ;   Julie Houle13 Author Orcid Image ;   Paul Poirier14 Author Orcid Image ;   Thomas Yates15 Author Orcid Image ;   Joseph Henson15 Author Orcid Image ;   Derek R Anderson5 Author Orcid Image ;   Elisabeth B Grey16 Author Orcid Image ;   Maria Panagioti1, 2 Author Orcid Image
Article Authors Cited by (3) Tweetations (19) Metrics

    Review

    1Division of Population Health, School of Health Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, United Kingdom

    2Health Services Research and Primary Care, National Institute for Health Research School for Primary Care Research, Manchester, United Kingdom

    3Division of Informatics, Imaging & Data Sciences, Faculty of Biology, Medicine & Health, University of Manchester, Manchester, United Kingdom

    4Department of Primary Care and Public Health, Brighton and Sussex Medical School, University of Brighton, Brighton, United Kingdom

    5Department of Psychology, The Ohio State University College of Arts and Sciences, Columbus, OH, United States

    6Department of Medicine, Endocrinology and Metabolism, University of New Mexico Health Sciences Center, Albuquerque, NM, United States

    7Department of General Practice and Primary Care, Centre for Public Heath, Queen’s University Belfast, Belfast, United Kingdom

    8School of Medicine, Ulster University, Londonderry, United Kingdom

    9Department of Medicine, McGill University, Montreal, QC, Canada

    10Centre for Outcomes Research and Evaluation, Research Institute of the McGill University Health Centre, Montreal, QC, Canada

    11Centre for Translational Biology, Research Institute of the McGill University Health Centre, Montreal, QC, Canada

    12Department of Family Medicine, University College Hospital, Ibadan, Nigeria

    13Department of Nursing, Université du Québec à Trois-Rivières, Trois-Rivières, QC, Canada

    14Institut Universitaire de Cardiologie et de Pneumologie de Québec, Université Laval, Laval, QC, Canada

    15Diabetes Research Centre, University of Leicester, Leicester, United Kingdom

    16Centre for Motivation and Health Behaviour Change, Department for Health, University of Bath, Bath, United Kingdom

    Corresponding Author:

    Alexander Hodkinson, BSc, MSc, PhD

    Division of Population Health

    School of Health Sciences, Faculty of Biology, Medicine and Health

    University of Manchester

    Oxford Road

    Manchester, M13 9PL

    United Kingdom

    Phone: 44 (0)161 2753535

    Email: alexander.hodkinson@manchester.ac.uk


    Background: Current evidence supports the use of wearable trackers by people with cardiometabolic conditions. However, as the health benefits are small and confounded by heterogeneity, there remains uncertainty as to which patient groups are most helped by wearable trackers.

    Objective: This study examined the effects of wearable trackers in patients with cardiometabolic conditions to identify subgroups of patients who most benefited and to understand interventional differences.

    Methods: We obtained individual participant data from randomized controlled trials of wearable trackers that were conducted before December 2020 and measured steps per day as the primary outcome in participants with cardiometabolic conditions including diabetes, overweight or obesity, and cardiovascular disease. We used statistical models to account for clustering of participants within trials and heterogeneity across trials to estimate mean differences with the 95% CI.

    Results: Individual participant data were obtained from 9 of 25 eligible randomized controlled trials, which included 1481 of 3178 (47%) total participants. The wearable trackers revealed that over the median duration of 12 weeks, steps per day increased by 1656 (95% CI 918-2395), a significant change. Greater increases in steps per day from interventions using wearable trackers were observed in men (interaction coefficient –668, 95% CI –1157 to –180), patients in age categories over 50 years (50-59 years: interaction coefficient 1175, 95% CI 377-1973; 60-69 years: interaction coefficient 981, 95% CI 222-1740; 70-90 years: interaction coefficient 1060, 95% CI 200-1920), White patients (interaction coefficient 995, 95% CI 360-1631), and patients with fewer comorbidities (interaction coefficient –517, 95% CI –1188 to –11) compared to women, those aged below 50, non-White patients, and patients with multimorbidity. In terms of interventional differences, only face-to-face delivery of the tracker impacted the effectiveness of the interventions by increasing steps per day.

    Conclusions: In patients with cardiometabolic conditions, interventions using wearable trackers to improve steps per day mostly benefited older White men without multimorbidity.

    Trial Registration: PROSPERO CRD42019143012; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=143012

    J Med Internet Res 2022;24(8):e36337

    doi:10.2196/36337

    Keywords

    systematic review; individual patient data; meta-analysis; steps/day; wearable tracker; cardiometabolic conditions; diabetes; obesity; cardiovascular disease 


    Background

    Cardiometabolic conditions are the leading cause of death worldwide, accounting for more than 41 million deaths annually [1]. These conditions include obesity, diabetes mellitus, and cardiovascular disease (CVD); these 3 common, intersecting noncommunicable diseases affect almost two-thirds of the global population [2,3].

    The World Health Organization recently recognized physical inactivity as a serious and growing public health problem and has set out to reduce it by 10% before 2025 [4]. The United Kingdom’s National Institute for Health and Care Excellence also highlights the importance of physical activity for obesity management, successful aging, CVD prevention, and weight management during pregnancy [5]. The consequences of being physically inactive include unhealthy weight gain, dyslipidemia, and elevated blood pressure and blood glucose levels, all of which heighten the risk of developing a cardiometabolic condition [6].

    Wearable physical activity trackers, such as accelerometers, pedometers, and the Fitbit (Fitbit Inc), are portable electrical or electromechanical devices that count each step a person takes by detecting the motion of the person along the body’s long axis [7,8]. They have become very popular for motivating and monitoring (thereby increasing) physical activity in general and in people with cardiometabolic conditions in particular [9,10]. Systematic reviews have suggested that the validity of various wearable trackers, especially those measuring steps, is high, and these reviews have found that they are useful for tracking ambulatory physical activity in clinical populations [11-13].

    Since many wearable trackers are affordable and user-friendly, they are viewed as a good practical method for monitoring basic physical activity [14], such as the number of steps per day, in high-risk people with chronic cardiometabolic conditions [15-17]. However, their long-term effectiveness in achieving the desired behavior changes (ie, increasing steps per day) in specific patient subgroups with cardiometabolic conditions is unclear, and they may only succeed in the short term as “quick fixes” [18].

    Our recent meta-analysis of 38 randomized trials [19] suggested that interventions using wearable trackers are moderately effective at increasing physical activity, including steps per day, in people with cardiometabolic conditions. The most promising interventions were those that focused on the number of steps per day. However, without individual participant data (IPD), we could not conduct an assessment of patient factors, baseline effects, and interventional differences, nor could we analyze their importance [20]. This was a major constraint of the review findings.

    In IPD meta-analysis, rather than extracting summary (ie, aggregate) data from study publications or from investigators, original research data are obtained directly from the researchers responsible for each study. These data are then re-analyzed centrally and combined in the meta-analysis. The IPD approach is becoming an increasingly popular tool compared to traditional aggregate-data meta-analysis, especially as the IPD approach avoids reliance on published results and provides an opportunity to investigate individual-level interactions, such as treatment-effect modifiers [20].

    Objectives

    In the present study, we undertook an IPD meta-analysis to identify whether belonging to certain subgroups of patients with cardiometabolic conditions, including groups with differing age, sex, ethnicity, and number or combination of cardiometabolic conditions, moderated the effectiveness of interventions using wearable trackers in improving physical activity, measured by the number of steps per day. We also examined the impact of interventional differences, such as behavior change, device placement, delivery method, and performance over time, on the effectiveness of interventions using wearable trackers in improving steps per day.


    This IPD meta-analysis followed a registered (PROSPERO CRD42019143012) protocol. A statistical analysis plan was produced in advance of analysis and the findings are reported in accordance with the PRISMA-IPD statement [21].

    Literature Search and Study Identification

    We searched MEDLINE (Ovid), EMBASE (EBSCO), CENTRAL, CINAHL, and PsycINFO from inception until August 2018, without language restriction; this was updated in December 2020 (Multimedia Appendix 1, pages 3-11). Additional studies were obtained by citation tracking, extraction from previous systematic reviews, and searches of trial registers (ie, ClinicalTrials.gov and ICTRP). A list of all the search sources and the data collection and management process are detailed in the protocol for this paper.

    Two researchers (AH and MP) independently identified the citations and then fully screened the relevant manuscripts according to the eligibility criteria. We included randomized controlled trials (RCTs) or cluster RCTs involving adults (aged 16 years or older) with a cardiometabolic condition, defined as a diagnosis (or high risk) of type 2 diabetes, CVD, or obesity or overweight.

    We included studies with an intervention program designed around the daily usage of wearable trackers, such as pedometers, accelerometers, and fitness trackers, rather than studies that only measured performance at the beginning and end of the study. Studies were required to have a usual-care comparator. Participants’ steps per day in both intervention and usual-care groups were measured in parallel using the same device. We further required that studies reported a physical activity assessment (ie, step count) at baseline and follow up using a separate wearable tracker that all participants received independently of their allocation (ie, both the intervention and control groups). We adopted this eligibility criterion to be able to reliably pool the results across the studies.

    The primary outcome was the number of steps per day, measured with any wearable tracker. Wearable trackers, either mechanical (eg, spring levered) or electronic (eg, using GPS or actigraph functionality), mostly monitored daily steps as the main outcome of interest; intervention programs involving wearable trackers often set goals for increasing the number of daily steps incrementally over time and estimated and logged the total distance travelled [22]. We excluded wearable trackers that used other types of measurement output, because these vary considerably in terms of their performance, choice of activity measurement (ie, light, moderate, or vigorous activity, energy expenditure, sedentary time, or stationary pattern) and intensity and frequency of the physical activity (ie, bouts of exercise). This would have made their pooling more problematic.

    Secondary outcomes included anthropometric measures, glycated hemoglobin level (mmols/mol), blood pressure, and cholesterol level.

    Risk of bias (RoB) was independently assessed by 2 reviewers (AH and MP) using the Cochrane RoB tool [23], alongside the completeness and quality of the provided IPD. Results of IPD studies were also compared with studies that did not supply IPD.

    Data Extraction and Assessment of IPD Integrity

    We used IPD (obtained from November 2019 to October 2021) to determine demographic characteristics that we chose a priori in the protocol, such as age, sex, ethnicity, and comorbidity; intervention characteristics, such as objective, duration, use of a behavior change framework, delivery method and placement of wearable tracker; and primary and secondary outcomes.

    Continuous variables were kept continuous, but some were also categorized when this was considered to be more clinically meaningful. For instance, patient groups were split by median age and by age range (20-49, 50-59, 60-69, and 70-90 years); the total number of comorbidities, predefined as cardiometabolic conditions including type 2 diabetes, hypertension, obesity or overweight, metabolic syndrome, and any cardiovascular condition; and BMI (normal 18.5-24.9 kg/m2; overweight, 25-29.9 kg/m2; obese, ≥30 kg/m2). Descriptive characteristics (eg, age, comorbidity, education, and ethnicity) were analyzed for intervention and control groups using ANOVA for continuous variables and the chi-square test for categorical variables. Following this, ethnicity was classified into 2 groups, White European/North American and other ethnicities, to improve the analysis of covariance (ANCOVA), due to the limited number of patients in other ethnic groups.

    Since all the trials provided above 70% of the IPD for the corresponding primary outcome, we imputed any missing values using the R package MICE: Multivariate Imputation by Chained Equations [24], following Rubin’s principle for imputation [25]. The range of imputed values was bounded by the observed values of the primary outcome, and baseline covariates (study, intervention, age, sex, and baseline) were used to predict missing data. The algorithms’ convergence was assessed, and sensitivity analyses were performed using only cases with available data (ie, complete cases).

    Data Synthesis

    Primary analyses used a 1-stage linear mixed effect model that incorporated random effects to allow for heterogeneity across trials [20,26,27], fitted using the Stata (version 16.1; StataCorp LLC) commands mixed and ipdforest to summarize the evidence by study and obtain forest plots [28]. Restricted maximum likelihood was used for model estimation, and centering of covariates by study-specific means was performed to avoid aggregation bias [29].

    As the primary outcomes were all objectively measured using the standard unit of steps per day, the analysis was performed using the mean difference (MD) between intervention and control groups with the 95% CI. Differential effects were then investigated by adding patient covariate parameters (ie, age, sex, ethnicity, and number of cardiometabolic conditions) and interactions between covariates (ie, treatment-covariate interaction terms) to the linear mixed model for the primary outcome. Important and significant differential effects were displayed through subgroup analysis plots.

    Results from a 2-stage random-effects (DerSimonian-Laird) meta-analysis were compared to results from 1-stage analyses using the ipdmetan command, for consistency. The Hartung-Knapp CI was used to account for uncertainty in the variance estimate [30].

    Heterogeneity was examined by visually inspecting forest plots and using the I2 statistic with the 95% CI [31]. Publication bias was assessed visually by using contour-enhanced funnel plots and the statistical Egger test for asymmetry [32]. The Egger test is performed using the following hypothesis, testing by P value: the null hypothesis is symmetry in the funnel plot, and the alternative hypothesis is asymmetry in the funnel plot. If P≤.05, we reject the null hypothesis. If P>.05, we accept the null hypothesis and reject the alternative hypothesis [33]. We assessed for availability bias by comparing summary results of the non-IPD studies with those from IPD studies [34].

    Sensitivity analyses of study-level factors were performed by (1) comparing studies that used a social cognitive theory framework as a guide for behavior change, (2) comparing studies that used a goal or goals as part of the intervention, (3) comparing the placement of the wearable tracker (ie, on the waist or wrist), (4) comparing the delivery of the intervention (ie, face to face or self-managed), (5) removing studies with high or unclear RoB based on allocation concealment, and (6) assessing the long-term performance (ie, beyond 6 months).

    Ethical Considerations

    Ethical approval and patient consent was not required as it had already been obtain from the primary authors during their trial period. Ethical waiver was provided and acknowledged by our University international review board.


    Of 25 eligible RCTs (including 3178 participants) that used wearable trackers and measured steps per day, we found that 9 studies (36%) had a median intervention duration of 12 (range 7-52) weeks, providing IPD for 47% of the total participants (1481/3178) (Figure 1). A list of the eligible studies is provided in Multimedia Appendix 1 (pages 12-14).

    ‎
    Figure 1. Identification and selection of studies providing individual participant data for meta-analysis of interventions involving wearable trackers for measuring steps per day in patients with cardiometabolic conditions. Con: control group; CVD: cardiovascular disease; int: intervention group; IPD: individual participant data; RCT: randomized controlled trial.
    View this figure

    Characteristics of Studies

    Of the 9 included IPD studies, 5 were from North America [35-40], 3 were from the United Kingdom [41-43], and 1 was from Nigeria [44]. Wearable trackers were used in all 9 studies [35,37-44] and the primary outcome was steps per day. Within the IPD sample, 822 of 1481 patients were men (56%) and 907 of 1481 were 60 years or older (61%; range 25-86 years). A total of 1231 of 1481 patients had (or were at risk of) type 2 diabetes (83%), 1262 of 1481 were obese or overweight (85%), and 495 of 1481 patients had CVD (33%). Characteristics of the studies that provided IPD and did not provide IPD are presented in Multimedia Appendix 1 (pages 15-22); baseline characteristics of the IPD are provided in Table 1.

    Table 1. Baseline characteristics of the individual participant data and imbalance assessment between treatment arms. Percentages are proportions of observations to intervention or control arms, as applicable.
    CharacteristicsInterventionControlP valuea
    Steps per day (in 1481 patients in 9 studies) (n), mean (SD)6071.25 (3060.72)6072.11 (3064.40).99
    Age (in 1481 patients in 9 studies) (years), mean (SD)60.53 (9.70)60.73 (10.06).68
    Height (in 986 patients in 5 studies) (cm), mean (SD)124.00 (74.12)126.73 (72.35).56
    BMI (in 1325 patients in 7 studies) (kg/cm2), mean (SD)32.03 (5.44)32.11 (5.03).77
    BMI by classification (in 1325 patients in 7 studies).40

    		Patients, N680645

    		Normal (18.5-24.9 kg/m2), n (%)37 (5.4)26 (4)

    		Overweight (25-29.9 kg/m2), n (%)266 (39.1)267 (41.4)

    		Obese (≥30 kg/m2), n (%)377 (55.4)352 (54.6)
    Ethnicity (in 1414 patients in 9 studies).27

    		Patients, N721693

    		White European or North American, n (%)534 (74.1)510 (73.6)

    		African American, n (%)105 (14.6)116 (16.7)

    		Hispanic or Latino, n (%)40 (5.5)23 (3.3)

    		Mixed ethnicity, n (%)27 (3.7)29 (4.2)

    		Asian/Middle Eastern, n (%)15 (2.1)15 (2.2)
    Education statusa (in 407 patients in 2 studies).60

    		Patients, N210183

    		Low (not completed secondary education to A level), n (%)42 (20)33 (18)

    		Medium (completed secondary education; ie, A level equivalent), n (%)36 (17.1)36 (19.7)

    		High (any further or higher education), n (%)94 (45.8)71 (38.8)
    Preexisting CVD (in 495 patients in 3 studies), n/N (%)44/230 (19.1)84/265 (32).07
    Preexisting type 2 diabetes (in 1231 patients in 4 studies), n/N (%)370/619 (59.8)403/612 (65.8).21
    Preexisting hypertension (in 642 patients in 3 studies), n/N (%)250/311 (80.4)264/331 (79.8).83
    Preexisting metabolic syndrome (in 471 patients in 2 studies), n/N (%)194/240 (80.8)201/231 (87).73
    Depression score (in 347 patients in 2 studies), mean (SD)2.24 (4.13)2.29 (4.10).99
    Smokers (in 578 patients in 3 studies), n/N (%)87/282 (30.9)84/296 (28.4).48

    aMean values were compared with a 2-tailed t test and categorical covariates were compared with the chi-squared test or ANOVA.

    Risk of Bias Assessments

    The RoB assessment of studies that contributed IPD compared with those that did not showed that the former had lower RoB across all domains (Multimedia Appendix 1, page 22). The assessments for each of the RoB domains of the IPD studies are available in Multimedia Appendix 1 (page 23), and the results of RoB assessments for non-IPD studies are available in our earlier systematic review [19].

    Efficacy of Wearable Trackers on Increasing Steps per Day

    In the 1-stage analysis involving all 9 studies and 1481 participants, wearable trackers were associated with small but significantly improved levels of physical activity over the median intervention duration of 12 (range 6-52) weeks. The number of steps was 1656 (95% CI 918-2395) greater in the intervention than the control group (Figure 2). These results are consistent with the 2-stage analysis (Multimedia Appendix 1, page 32). The mean number of steps per day at the end of treatment, without adjusting for baseline scores, was 6561 (SD 3336) in the wearable activity tracker intervention group and 6561 (SD 3340) in the control group; this was not a significant difference (P=.99). Visual and statistical evidence (P=.002; Egger test) of a small study effect was found in the funnel plots (Multimedia Appendix 1, pages 24-30). After removing the studies with high or unclear RoB from the analysis, the symmetry of the funnel plot improved somewhat (P=.06). Secondary outcomes showed insignificant differences (Multimedia Appendix 1, page 31).

    ‎
    Figure 2. Forest plot showing 1-stage meta-analysis of individual participant data from studies using wearable trackers to measure steps per day; the mean postintervention difference in steps per day is also shown. MD: mean difference; REML: restricted maximum likelihood.
    View this figure

    Covariate interaction effects for the primary outcome showed that patients older than 50 years benefited more from using wearable trackers. The interaction coefficient for patients aged 50 to 59 years was 1175.16 (95% CI 377.46 to 1972.86) steps per day, with an individual group effect of 2006.83 (95% CI 1163.83 to 2849.82); in patients aged 60 to 69 years, the interaction coefficient was 981.37 (95% CI 222.39 to 1740.35), with an individual group effect of 1813.04 (95% CI 986.40 to 2639.68); and in patients aged 70 to 90 years, the interaction coefficient was 1059.98 (95% CI 200.29 to 1919.66), with an individual group effect of 1891.65 (95% CI 963.98 to 2819.31; I2 15.5%). In contrast, for patients aged under 50 years the interaction coefficient was 831.67 (95% CI –97.00 to 1760.33) (Table 2).

    The number of steps per day after using wearable trackers was greater among men than women (interaction coefficient –668.3, 95% CI –1156.8 to –179.93). For men, the mean number of steps per day was 2006 (95% CI 1204.4 to 2807.6), while for women, it was 1337.65 (95% CI 538.92 to 2136.37; I2 16%) (Figure 3 and Figure 4). Patients with at least 2 comorbidities showed a significantly lower number of steps per day; the interaction coefficient for 2 comorbidities was –516.80 (95% CI –1188.34 to –10.74), and the mean number of steps was 1344.70 (95% CI 421.62 to 1843.87). For patients with 3 comorbidities, the interaction coefficient was –876.44 (95% CI –2071.88 to 509.41), and the mean number of steps was 570.17 (95% CI –304.66 to 870.08; I2 15.7%). In contrast, for patients with only 1 comorbidity, the interaction coefficient was 1861.55 (95% CI 1061.6 to 2661.5). White patients also displayed a higher step count after using wearable trackers (2189, 95% CI 1276 to 3102) compared to other ethnic groups (1194, 95% CI 280 to 2107); the interaction coefficient was 995 (95% CI 360 to 1631) steps per day (I2 21%).

    Table 2. Differential effects of wearable trackers on physical activity measured by steps per day among specific subgroups of patients.
    CharacteristicMean difference in steps per day,a n (95% CI)Treatment covariate interaction


    		Interaction coefficient (95% CI)P valueI2, % (95% CI)
    Age

    		≥60 years1814.39 (996.51 to 2632.28)1N/Ab16.1 (5.0 to 41.1)

    		<60 yearsc1566.83 (766.98 to 2366.68)–247.56 (–762.0 to 266.92).35
    Age category

    		20-49 years831.67 (–97.00 to 1760.33)1N/A15.5 (4.8 to 40.2)

    		50-59 years2006.83 (1163.83 to 2849.82)1175.16 (377.46 to 1972.86).004

    		60-69 years1813.04 (986.40 to 2639.68)981.37 (222.39 to 1740.35).01

    		70-90 years1891.65 (963.98 to 2819.31)1059.98 (200.29 to 1919.66).02
    Sex

    		Men2006 (1204.4 to 2807.6)1N/A16.03 (5.0 to 40.91)

    		Women1337.65 (538.92 to 2136.37)–668.3 (–1156.8 to –179.93).01
    Ethnicity

    		Other1193.65 (280.31 to 2106.99)1N/A20.5 (7.0 to 46.8)

    		Whited2189.0 (1276.3 to 3101.65)995.30 (359.80 to 1630.80).002
    Comorbiditiese

    		11861.55 (1061.6 to 2661.5)1N/A15.7 (4.6 to 41.7)

    		21344.70 (421.62 to 1843.87)–516.80 (–1188.34 to –10.74).04

    		3570.17 (–304.66 to 870.08)–876.44 (–2071.88 to –509.41).01

    		41078.28 (468.72 to 2077.31)N/AN/A
    Cardiometabolic condition focus

    		Other conditionsf1535.28 (–557.35 to 3627.91)1N/A16.2 (0.5 to 87.3)

    		Type II diabetes1942.47 (47.24 to 3837.70)407.19 (–785.09 to 1599.47).50

    aModel accounted for baseline steps per day scores with analysis of covariance.

    bN/A: not applicable.

    cPer year of age.

    dWhite versus all other ethnicities.

    eIncluding type II diabetes, hypertension, angina, obese or overweight, and any other cardiovascular condition (excluding stroke).

    fIncluding hypertension.

    ‎
    Figure 3. Gender effect for women. Forest plot showing 1-stage meta-analysis of individual participant data from women only, derived from studies using wearable trackers to measure steps per day; the mean postintervention difference in steps per day is also shown. MD: mean difference; REML: restricted maximum likelihood.
    View this figure
    ‎
    Figure 4. Gender effect for men. Forest plot showing 1-stage meta-analysis of individual participant data from men only, derived from studies using wearable trackers to measure steps per day; the mean postintervention difference in steps per day is also shown. MD: mean difference; REML: restricted maximum likelihood.
    View this figure

    Further Sensitivity Analysis and Exploratory Analysis

    Behavior change frameworks were used in 3 studies [35,37,38,45], but did not appear to improve the number of steps per day (the interaction coefficient was –1054, 95% CI –2785 to 677). Nevertheless, groups that did or did not use the program still showed a statistically significant change (part of program: 2476, 95% CI 935 to 4017; not part of program: 1422, 95% CI 634 to 2211). In addition, studies that set goals for reaching specific steps per day also did not appear to improve the number of steps per day, nor did the placement of the wearable tracker (ie, waist vs wrist) (Multimedia Appendix 1, page 33).

    Small but nonsignificant improvements in steps per day were seen in studies with low RoB compared to studies with high or unclear RoB. Studies that assessed the number of steps per day beyond 26 weeks showed a lower (but nonsignificant) number of steps per day compared to studies that assessed performance in the short term (ie, less than 6 months). While both groups showed statistically significant changes (<26 weeks: 2000, 95% CI 1068 to 2932; ≥26 weeks: 1143, 95% CI 33 to 2254) when they used trackers over the longer term (ie, beyond 6 months) it was clear that performance waned by almost 1000 steps per day compared to short-term use (ie, up to 6 months). Studies that involved face-to-face delivery of the wearable tracker significantly improved the number of steps per day (2630, 95% CI 1835 to 3425) when compared to studies that involved only self-regulated use of the tracker (850, 95% CI 325 to 1375) (the interaction coefficient was 1780, 95% CI 826 to 2733).

    Improvement in steps per day among the non-IPD studies was significantly higher with the wearable trackers; results showed almost double the number of steps per day (2854, 95% CI 1944 to 3763) compared to the IPD meta-analysis estimate (Multimedia Appendix 1, page 34). However, including the non-IPD studies in the meta-analysis increased bias from small studies and worsened the funnel plot asymmetry (Multimedia Appendix 1, pages 26-27).


    Principal Findings

    This IPD meta-analysis confirms that interventions using wearable trackers were effective at increasing physical activity, as measured by total number of steps per day, in participants with cardiometabolic conditions, compared to control groups, over a median duration of 12 weeks. These improvements were slightly lower than estimated by our aggregate-data meta-analysis [19]. In this IPD meta-analysis, we identified differential effects in relation to age, sex, and ethnicity of the participants and the number of comorbidities present. Consistent with our aggregate-data meta-analysis [19], we observed that interventions that used wearable trackers with face-to-face delivery by a professional were more effective at increasing steps per day than patient self-managed interventions.

    The benefits in terms of the mean difference in steps per day after using wearable trackers in the short term were not as large as seen in other meta-analyses [46,47], but at the end of treatment, the activity tracker mean score increased to 6561 steps per day, which does appear to meet the recommended number of daily steps as outlined in public health guidelines [48,49]. For example, the average daily steps recorded were just above the recommendation of 3000 to 6000 steps per day made by the United Kingdom’s National Obesity Forum and the recommendation of 3000 steps per day made by Northern Ireland’s Public Health Agency [50]. However, the large standard deviation (3336) from this mean score suggests that some patients were still underachieving and did not meet the goal for steps per day. Still, recent evidence [51] has shown that for each daily increase of 1000 steps in physically inactive individuals at baseline, the estimated risk reduction in all-cause mortality is 6% to 36%, while the reduction for CVD is 5% to 21%. An increase of 500 steps per day or the equivalent (eg, 5 minutes of brisk walking) is also considered the minimum clinically important difference in steps per day in inactive adults [52]. Further, a recent dose-response meta-analysis [53] of the association between steps per day and all-cause mortality risk indicated a strong inverse association; the risk decreased linearly from 2700 to 17,000 steps per day. More specifically, the hazard ratio for 10,000 steps per day was 0.44 (95% CI 0.31 to 0.63). Even the most extremely physically inactive patients, such as the ones in our study, are still likely to benefit from small gains in steps per day.

    Wearable trackers were predominately more effective in White men, but were still somewhat effective in non-White men and women. Evidence gathered from various countries shows that women are less active than men (there is a global average of 31.7% inactive women vs 23.4% inactive men) and that barriers to women’s involvement in sports are numerous and complex [54-56]. Men tend to have more intrinsic motivators linked to and leading to physical activity, such as the need or desire to improve health, prevent disease onset, and improve body shape, and are also more competitive [57]. In contrast, different stimuli appear to motivate women of various ages to undertake physical exercise, such as emotional involvement, socialization, mental and physical well-being, and the achievement and maintenance of a positive self-image [58]. Women may also have less time due to daily household chores. Therefore, policies that address the sex gap in physical activity could start with better access and investment and by altering sociocultural norms. In relation to ethnicity, there is evidence that non-White participants have lower levels of physical activity and that their participation in and benefit from physical activity programs are suboptimal, due to lower access and socioeconomic and sociocultural bias [59]. Our findings suggest that non-White women in particular are less likely to achieve significant benefits from interventions involving wearable trackers. This may be because of the increased likelihood that they encounter social, economic, and cultural barriers to physical activity that are unique to them.

    Wearable trackers were also most effective for improving physical activity in participants aged 49 years or more, and were surprisingly ineffective in the 365 participants aged 50 years or less. Although this could be a sample size issue, there are some possible explanations for this finding. People under 50 years old may be less likely to engage with wearable trackers due to caring and work responsibilities, and they may have less time to participate in daily physical activity [60]. Conversely, patients aged between 50 and 60 years generally maintained better daily physical activity levels than those over the age of 60. This may be due to multiple comorbidities, which are likely to be the leading cause of reduced physical activity in the older age group [61].

    Participants with a cardiometabolic condition have a relatively higher likelihood of developing additional comorbidities, meaning that many of these high-risk patients are often diagnosed with 2 or more diseases [62]. As we have clearly shown in this study, most patients do not meet the international clinical guidelines for recommended steps per day, and when cardiometabolic risk is combined with a comorbidity, the steps-per-day performance was reduced even further. This result reinforces several key studies and policy guidelines, which show that a worsening steps-per-day performance is highly associated with multimorbidity [1,2,5,6,63]. Moreover, it is well known that low physical activity increases weight, BMI, and waist circumference, which are all key predictors for further exacerbation of comorbid chronic diseases, including diabetes, hypertension, and dyslipidemia [64]. For health care professionals, effective and practical management of patients with multimorbidity is important. A more sensitive understanding of their lifestyles and their tendency toward extremely low levels of physical activity will facilitate the support of those most in need of it.

    Secondary cardiometabolic biomarkers, such as blood sugar glucose, measured as hemoglobin A1c (HbA1c), blood pressure, cholesterol, and BMI or weight, were all found to not be statistically significant, which may be unsurprising given the short duration of the interventions (only 12 weeks). However, one study [65] did find small but significant reductions in BMI and systolic blood pressure when using pedometers, although the patient population in this study involved a variety of outpatients, who may not have been suffering from the same severe underlying cardiometabolic health conditions. Another study [47] investigated the effectiveness of setting physical activity goals in patients with type 2 diabetes who used step counters and did not report any significant reduction in HbA1c level. Similarly, a meta-analysis [46] that compared accelerometers and pedometers for improving physical activity levels and HbA1c levels in people with type 2 diabetes showed no significant differences between either type of tracker. While our original review [19] did show significant reductions in HbA1c levels, this can only be considered a small reduction in effect size, and is not dissimilar to our findings based on IPD (–0.19 vs –0.13).

    Strengths and Limitations

    This is the first IPD meta-analysis to assess the differential effects of wearable trackers for important physical activity and clinically relevant outcomes in participants with cardiometabolic conditions. Strengths include a clear standardization of definitions and outcomes, imputation of missing data, a robust analysis that included ANCOVA [66], exploration of the potential for differential effects [29], extensive data checking and work with study investigators to ensure the quality of the data set, and the inclusion of studies that mostly had low RoB. Nevertheless, we were unable to obtain IPD from 16 studies, which meant that 1697 of 3178 (53%) of potential patients were missing. However, these studies were mainly small and generally had a higher RoB than the included studies. Over two-thirds of the IPD were from White participants, meaning that all other ethnic groups (Black African, Asian, Hispanic, and others) had to be combined into one category to allow for adequate power in the subgroup analysis. Nevertheless, we found that White patients had significantly higher step counts than other ethnic groups. This result is similar to findings from a recent prospective cohort study that assessed the association of steps per day with premature all-cause mortality among Black and White men and women with coronary artery risk and showed that Black participants took fewer steps than White participants (median 8670 steps/day, IQR 6810 to 10,811, vs median 9441 steps/day, IQR 7704 to 11,329, respectively; P<.001) [67]. As 83% of the patients had or were at risk of type 2 diabetes and 85% of the patients were at least overweight or obese, this meant that these 2 conditions largely overlapped in the patient population, meaning that it was not possible to properly adjust for this in the analysis. In addition, as only 33% of the patients had CVD, and no other condition data were provided in the IPD, these 2 conditions could only be compared with CVD. Thus, we urge that these results are interpreted with some caution and encourage better coding of condition data, which would allow for more detailed analyses. Only 3 studies [37,38,42,43] used a behavior change framework as part of their intervention design, and only 3 studies [36-38,43] collected data over the longer term (ie, at least 1 year). Both are clear weaknesses that limit our understanding of sustained effects over time, which is an important gap in knowledge [68]. Differences in tracker functionality may also have significant effects on their performance. For instance, trackers are often criticized for not measuring daily steps precisely enough, particularly when the tracker becomes tilted below the waistline in overweight or obese individuals [69-71]. However, some more expensive wearable trackers that can sense movement in a tilted position have shown promise [72]. Diversity in wearing time and self-monitoring across the studies made it impossible to effectively categorize this information for meaningful subgroup analyses.

    Moreover, following an update to our aggregate review, we found that 8 studies, which included 414 participants, used wearable trackers to measure physical activity with variable measurement outputs rather than steps per day. We excluded these 8 studies from our IPD meta-analysis because the assessment of the differential effects would have been underpowered, with only 414 participants available. Once the evidence base is more developed, we strongly encourage future efforts to compare our findings with those of studies using additional measurement outputs or other wearable technologies (eg, smartphone apps, smartwatches, and wristbands). Finally, our last search update (in December 2020) could be considered marginally outdated for an aggregate-data meta-analysis. However, data acquisition, preparation, and analysis for IPD meta-analyses requires considerably more time and resources than aggregate-data meta-analyses. We strongly recommend universal open access to trial data to speed access to IPD in the future.

    Conclusion

    Interventions using wearable trackers were effective at providing a small mean improvement in steps per day over short periods of use in participants with a cardiometabolic condition. Interventions with wearable trackers that were delivered and guided by a professional were most effective in White men and in those aged 50 years and older with only one comorbidity. Future research should look at ways to extend the beneficial effects of interventions with wearable trackers to other patients (particularly women) with cardiometabolic conditions and for longer periods.

    Acknowledgments

    We thank the authors who agreed to participate and supply their individual participant data: CFE, MRB, HE, MEC, MAT, KD, SSD, ABC, AFF, EBG, JH, PP, TY, and JH. This study was funded by the Evidence Synthesis Working Group (project 390), which is supported by the National Institute for Health Research School for Primary Care Research. The National Institute for Health Research Greater Manchester Patient Safety Translational Research Centre has partly funded the time of MP.

    Data Availability

    A data sharing agreement form was completed by each participant that provided individual participant data; there has not been any agreement to share the data. Therefore, authors seeking individual participant data should contact the principal investigators from the individual trials if they desire access. However, we aim to discuss with the authors the possibility of making the individual participant data database publicly available.

    Disclaimer

    This paper presents independent research funded by the National Institute for Health Research School for Primary Care Research Project 390. The views expressed are those of the authors and not necessarily those of the National Institute for Health Research or the Department of Health and Social Care. The research team members were independent from the funding agencies. The funders had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, and approval of the manuscript.

    Authors' Contributions

    MP, EK, and PB developed the initial research idea and obtained funding for this study. AH, MP, EK, HVM, and PB formulated the research questions and designed the study. AH searched for published work, selected articles, extracted and analyzed data, and drafted the protocol, statistical analysis plan, and manuscript. MP, SSZ, and CG helped the article search, data extraction, and quality checking. MP substantially contributed to designing the searches and the statistical analysis plan, writing the manuscript, and interpreting the findings. EK provided statistical support for using the ipdforest command in Stata. MP, EK, PB, HVM, BM, SSZ, CG, DT, and all other authors substantially contributed to the manuscript by providing review comments and edits. All authors have read and approved the final manuscript. The manuscript’s guarantor (AH) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported, that no important aspects of the study have been omitted, and that any discrepancies from the study as planned have been explained. The corresponding author attests that all listed authors meet the authorship criteria and that no others meeting the criteria have been omitted.

    Conflicts of Interest

    None declared.

    ‎
    Multimedia Appendix 1

    Supplementary appendix booklet.

    DOCX File , 362 KB
    1. Laranjo L, Ding D, Heleno B, Kocaballi B, Quiroz JC, Tong HL, et al. Do smartphone applications and activity trackers increase physical activity in adults? Systematic review, meta-analysis and metaregression. Br J Sports Med 2021 Apr;55(8):422-432. [CrossRef] [Medline]
    2. Miranda JJ, NCD Risk Factor Collaboration (NCD-RisC)—Americas Working Group. Trends in cardiometabolic risk factors in the Americas between 1980 and 2014: a pooled analysis of population-based surveys. Lancet Glob Health 2020 Jan;8(1):e123-e133 [FREE Full text] [CrossRef] [Medline]
    3. GBD 2015 Risk Factors Collaborators. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016 Oct 08;388(10053):1659-1724 [FREE Full text] [CrossRef] [Medline]
    4. Physical activity. World Health Organization.   URL: https://www.who.int/news-room/fact-sheets/detail/physical-activity [accessed 2022-07-03]
    5. Physical activity. National Institute for Health and Care Excellence. 2020.   URL: https://www.nice.org.uk/guidance/lifestyle-and-wellbeing/physical-activity [accessed 2022-07-03]
    6. Global Recommendations on Physical Activity for Health. World Health Organization.   URL: https://www.who.int/publications/i/item/9789241599979 [accessed 2022-07-03]
    7. Shin G, Jarrahi MH, Fei Y, Karami A, Gafinowitz N, Byun A, et al. Wearable activity trackers, accuracy, adoption, acceptance and health impact: A systematic literature review. J Biomed Inform 2019 May;93:103153 [FREE Full text] [CrossRef] [Medline]
    8. Laranjo L, Ding D, Heleno B, Kocaballi B, Quiroz JC, Tong HL, et al. Do smartphone applications and activity trackers increase physical activity in adults? Systematic review, meta-analysis and metaregression. Br J Sports Med 2021 Apr 21;55(8):422-432. [CrossRef] [Medline]
    9. Afshin A, Babalola D, Mclean M, Yu Z, Ma W, Chen C, et al. Information Technology and Lifestyle: A Systematic Evaluation of Internet and Mobile Interventions for Improving Diet, Physical Activity, Obesity, Tobacco, and Alcohol Use. J Am Heart Assoc 2016 Aug 31;5(9):e003058 [FREE Full text] [CrossRef] [Medline]
    10. Pedometers can help people get more active as part of an exercise programme. National Institute for Health and Care Research.   URL: https:/​/evidence.​nihr.ac.uk/​alert/​pedometers-can-help-people-get-more-active-as-part-of-an-exercise-programme-/​ [accessed 2022-07-03]
    11. Tudor-Locke C, Williams JE, Reis JP, Pluto D. Utility of pedometers for assessing physical activity: convergent validity. Sports Med 2002;32(12):795-808. [CrossRef] [Medline]
    12. Bassett DR, John D. Use of pedometers and accelerometers in clinical populations: validity and reliability issues. Phys Ther Rev 2013 Jul 19;15(3):135-142. [CrossRef]
    13. Fuller D, Colwell E, Low J, Orychock K, Tobin MA, Simango B, et al. Reliability and Validity of Commercially Available Wearable Devices for Measuring Steps, Energy Expenditure, and Heart Rate: Systematic Review. JMIR Mhealth Uhealth 2020 Sep 08;8(9):e18694 [FREE Full text] [CrossRef] [Medline]
    14. Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer-wearable activity trackers. Int J Behav Nutr Phys Act 2015 Dec 18;12(1):159-151 [FREE Full text] [CrossRef] [Medline]
    15. Franssen WMA, Franssen GHLM, Spaas J, Solmi F, Eijnde BO. Can consumer wearable activity tracker-based interventions improve physical activity and cardiometabolic health in patients with chronic diseases? A systematic review and meta-analysis of randomised controlled trials. Int J Behav Nutr Phys Act 2020 May 11;17(1):57-51 [FREE Full text] [CrossRef] [Medline]
    16. Wise JM, Hongu N. Pedometer, Accelerometer, And mobile technology for Promoting Physical Activity. The University of Arizona Cooperative Extension.   URL: https://extension.arizona.edu/sites/extension.arizona.edu/files/pubs/az1491-2014.pdf [accessed 2022-07-03]
    17. Ayabe M, Ishii K, Takayama K, Aoki J, Tanaka H. Comparison of interdevice measurement difference of pedometers in younger and older adults. Br J Sports Med 2010 Feb 28;44(2):95-99. [CrossRef] [Medline]
    18. Kirk MA, Amiri M, Pirbaglou M, Ritvo P. Wearable Technology and Physical Activity Behavior Change in Adults With Chronic Cardiometabolic Disease: A Systematic Review and Meta-Analysis. Am J Health Promot 2019 Jun 26;33(5):778-791. [CrossRef] [Medline]
    19. Hodkinson A, Kontopantelis E, Adeniji C, van Marwijk H, McMillian B, Bower P, et al. Interventions Using Wearable Physical Activity Trackers Among Adults With Cardiometabolic Conditions: A Systematic Review and Meta-analysis. JAMA Netw Open 2021 Jul 01;4(7):e2116382 [FREE Full text] [CrossRef] [Medline]
    20. Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ 2010 Feb 05;340(feb05 1):c221-c221. [CrossRef] [Medline]
    21. Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, PRISMA-IPD Development Group. Preferred Reporting Items for Systematic Review and Meta-Analyses of individual participant data: the PRISMA-IPD Statement. JAMA 2015 Apr 28;313(16):1657-1665. [CrossRef] [Medline]
    22. Lee I, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, Lancet Physical Activity Series Working Group. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet 2012 Jul 21;380(9838):219-229 [FREE Full text] [CrossRef] [Medline]
    23. Altman DG, Altman DG. Assessing risk of bias in included studies. In: Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions. New York, NY: John Wiley & Sons; 2008.
    24. Van Buuren S. MICE: Multivariate Imputation by Chained Equations. Carnegie Mellon University.   URL: http://lib.stat.cmu.edu/R/CRAN/web/packages/mice/index.html [accessed 2022-07-03]
    25. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York, NY: John Wiley & Sons; 1987.
    26. Kontopantelis E. A comparison of one-stage vs two-stage individual patient data meta-analysis methods: A simulation study. Res Synth Methods 2018 Sep 21;9(3):417-430 [FREE Full text] [CrossRef] [Medline]
    27. Burke DL, Ensor J, Riley RD. Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ. Stat Med 2017 Feb 28;36(5):855-875 [FREE Full text] [CrossRef] [Medline]
    28. Kontopantelis E, Reeves D. A Short Guide and a Forest Plot Command (Ipdforest) for One-Stage Meta-Analysis. Stata J 2013 Sep 01;13(3):574-587 [FREE Full text] [CrossRef]
    29. Hua H, Burke DL, Crowther MJ, Ensor J, Tudur Smith C, Riley RD. One-stage individual participant data meta-analysis models: estimation of treatment-covariate interactions must avoid ecological bias by separating out within-trial and across-trial information. Stat Med 2017 Feb 28;36(5):772-789 [FREE Full text] [CrossRef] [Medline]
    30. IntHout J, Ioannidis JP, Borm GF. The Hartung-Knapp-Sidik-Jonkman method for random effects meta-analysis is straightforward and considerably outperforms the standard DerSimonian-Laird method. BMC Med Res Methodol 2014 Feb 18;14(1):25 [FREE Full text] [CrossRef] [Medline]
    31. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002 Jun 15;21(11):1539-1558. [CrossRef] [Medline]
    32. Sterne JAC, Sutton AJ, Ioannidis JPA, Terrin N, Jones DR, Lau J, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011 Jul 22;343(jul22 1):d4002-d4002. [CrossRef] [Medline]
    33. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997 Sep 13;315(7109):629-634 [FREE Full text] [CrossRef] [Medline]
    34. Ahmed I, Sutton AJ, Riley RD. Assessment of publication bias, selection bias, and unavailable data in meta-analyses using individual participant data: a database survey. BMJ 2012 Jan 03;344(1):d7762-d7762. [CrossRef] [Medline]
    35. Anderson DR. Health Beliefs, Will to Live, Hope, and Social Support in a Pedometer-Based Exercise Intervention among Cardiac Rehabilitation Patients. The Ohio State University. 2015.   URL: https:/​/etd.​ohiolink.edu/​apexprod/​rws_etd/​send_file/​send?accession=osu1434901973&disposition=inline&psig=AOvVaw1mbfdGQZAw0Jr6ZoSCUHuD&ust=1657473325229075 [accessed 2022-07-12]
    36. Dasgupta K, Rosenberg E, Joseph L, Cooke AB, Trudeau L, Bacon SL, SMARTER Trial Group. Physician step prescription and monitoring to improve ARTERial health (SMARTER): A randomized controlled trial in patients with type 2 diabetes and hypertension. Diabetes Obes Metab 2017 May 22;19(5):695-704 [FREE Full text] [CrossRef] [Medline]
    37. Houle J, Doyon O, Vadeboncoeur N, Turbide G, Diaz A, Poirier P. Innovative program to increase physical activity following an acute coronary syndrome: randomized controlled trial. Patient Educ Couns 2011 Dec;85(3):e237-e244. [CrossRef] [Medline]
    38. Houle J, Doyon O, Vadeboncoeur N, Turbide G, Diaz A, Poirier P. Effectiveness of a pedometer-based program using a socio-cognitive intervention on physical activity and quality of life in a setting of cardiac rehabilitation. Can J Cardiol 2012 Jan;28(1):27-32. [CrossRef] [Medline]
    39. Katzmarzyk PT, Champagne CM, Tudor-Locke C, Broyles ST, Harsha D, Kennedy BM, et al. A short-term physical activity randomized trial in the Lower Mississippi Delta. PLoS One 2011 Oct 25;6(10):e26667 [FREE Full text] [CrossRef] [Medline]
    40. Araiza, Hewes H, Gashetewa C, Vella CA, Burge MR. Efficacy of a pedometer-based physical activity program on parameters of diabetes control in type 2 diabetes mellitus. Metabolism 2006 Oct;55(10):1382-1387. [CrossRef] [Medline]
    41. Cupples M, Dean D, Tully MA, Taggart M, McCorkell G, O’Neill S, et al. Using Pedometer Step-Count Goals to Promote Physical Activity in Cardiac Rehabilitation: A Feasibility Study of a Controlled Trial. Int J Phys Med Rehabil 2013;04(07):1 [FREE Full text] [CrossRef]
    42. Grey EB, Thompson D, Gillison FB. Effects of a Web-Based, Evolutionary Mismatch-Framed Intervention Targeting Physical Activity and Diet: a Randomised Controlled Trial. Int J Behav Med 2019 Dec 25;26(6):645-657 [FREE Full text] [CrossRef] [Medline]
    43. Yates T, Edwardson CL, Henson J, Gray LJ, Ashra NB, Troughton J, et al. Walking Away from Type 2 diabetes: a cluster randomized controlled trial. Diabet Med 2017 May 19;34(5):698-707. [CrossRef] [Medline]
    44. Fayehun AF, Olowookere OO, Ogunbode AM, Adetunji AA, Esan A. Walking prescription of 10 000 steps per day in patients with type 2 diabetes mellitus: a randomised trial in Nigerian general practice. Br J Gen Pract 2018 Feb;68(667):e139-e145 [FREE Full text] [CrossRef] [Medline]
    45. Greaney ML, Askew S, Wallington SF, Foley PB, Quintiliani LM, Bennett GG. The effect of a weight gain prevention intervention on moderate-vigorous physical activity among black women: the Shape Program. Int J Behav Nutr Phys Act 2017 Oct 16;14(1):139 [FREE Full text] [CrossRef] [Medline]
    46. Baskerville R, Ricci-Cabello I, Roberts N, Farmer A. Impact of accelerometer and pedometer use on physical activity and glycaemic control in people with Type 2 diabetes: a systematic review and meta-analysis. Diabet Med 2017 May 19;34(5):612-620. [CrossRef] [Medline]
    47. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized controlled trials. BMC Med 2014 Feb 27;12(1):36 [FREE Full text] [CrossRef] [Medline]
    48. Physical activity; walking and cycling: Public health guideline [PH41]. National Institute for Health and Clinical Excellence.   URL: https://www.nice.org.uk/guidance/ph41 [accessed 2018-09-10]
    49. School Health Guidelines to Promote Healthy Eating and Physical Activity. Centers for Disease Control and Prevention.   URL: https://www.cdc.gov/mmwr/pdf/rr/rr6005.pdf [accessed 2022-07-03]
    50. Tudor-Locke C, Burkett L, Reis JP, Ainsworth BE, Macera CA, Wilson DK. How many days of pedometer monitoring predict weekly physical activity in adults? Prev Med 2005 Mar;40(3):293-298. [CrossRef] [Medline]
    51. Hall KS, Hyde ET, Bassett DR, Carlson SA, Carnethon MR, Ekelund U, et al. Systematic review of the prospective association of daily step counts with risk of mortality, cardiovascular disease, and dysglycemia. Int J Behav Nutr Phys Act 2020 Jun 20;17(1):78 [FREE Full text] [CrossRef] [Medline]
    52. Rowlands A, Davies M, Dempsey P, Edwardson C, Razieh C, Yates T. Wrist-worn accelerometers: recommending ~1.0 m as the minimum clinically important difference (MCID) in daily average acceleration for inactive adults. Br J Sports Med 2021 Jul 14;55(14):814-815. [CrossRef] [Medline]
    53. Jayedi A, Gohari A, Shab-Bidar S. Daily Step Count and All-Cause Mortality: A Dose-Response Meta-analysis of Prospective Cohort Studies. Sports Med 2022 Jan 21;52(1):89-99. [CrossRef] [Medline]
    54. The Lancet Public Health. Time to tackle the physical activity gender gap. Lancet Public Health 2019 Aug;4(8):e360 [FREE Full text] [CrossRef] [Medline]
    55. Barker J, Smith Byrne K, Doherty A, Foster C, Rahimi K, Ramakrishnan R, et al. Physical activity of UK adults with chronic disease: cross sectional analysis of accelerometer measured physical activity in 96 706 UK Biobank participants. Int J Epidemiol 2019 Aug 01;48(4):1386 [FREE Full text] [CrossRef] [Medline]
    56. Manjoo P, Joseph L, Pilote L, Dasgupta K. Sex differences in step count-blood pressure association: a preliminary study in type 2 diabetes. PLoS One 2010 Nov 22;5(11):e14086 [FREE Full text] [CrossRef] [Medline]
    57. Hallal PC, Andersen LB, Bull FC, Guthold R, Haskell W, Ekelund U, Lancet Physical Activity Series Working Group. Global physical activity levels: surveillance progress, pitfalls, and prospects. Lancet 2012 Jul 21;380(9838):247-257. [CrossRef] [Medline]
    58. Vaccarezza M, Papa V, Milani D, Gonelli A, Secchiero P, Zauli G, et al. Sex/Gender-Specific Imbalance in CVD: Could Physical Activity Help to Improve Clinical Outcome Targeting CVD Molecular Mechanisms in Women? Int J Mol Sci 2020 Feb 21;21(4):1477 [FREE Full text] [CrossRef] [Medline]
    59. Such E, Salway S, Copeland R, Haake S, Domone S, Mann S. A formative review of physical activity interventions for minority ethnic populations in England. J Public Health (Oxf) 2017 Dec 01;39(4):e265-e274 [FREE Full text] [CrossRef] [Medline]
    60. Khanom A, Evans BA, Lynch R, Marchant E, Hill RA, Morgan K, et al. Parent recommendations to support physical activity for families with young children: Results of interviews in deprived and affluent communities in South Wales (United Kingdom). Health Expect 2020 Apr;23(2):284-295 [FREE Full text] [CrossRef] [Medline]
    61. Watson KB, Carlson SA, Gunn JP, Galuska DA, O'Connor A, Greenlund KJ, et al. Physical Inactivity Among Adults Aged 50 Years and Older - United States, 2014. MMWR Morb Mortal Wkly Rep 2016 Sep 16;65(36):954-958 [FREE Full text] [CrossRef] [Medline]
    62. Kivimäki M, Kuosma E, Ferrie JE, Luukkonen R, Nyberg ST, Alfredsson L, et al. Overweight, obesity, and risk of cardiometabolic multimorbidity: pooled analysis of individual-level data for 120 813 adults from 16 cohort studies from the USA and Europe. Lancet Public Health 2017 Jun;2(6):e277-e285 [FREE Full text] [CrossRef] [Medline]
    63. Knaeps S, Bourgois JG, Charlier R, Mertens E, Lefevre J, Wijndaele K. Ten-year change in sedentary behaviour, moderate-to-vigorous physical activity, cardiorespiratory fitness and cardiometabolic risk: independent associations and mediation analysis. Br J Sports Med 2018 Aug 04;52(16):1063-1068 [FREE Full text] [CrossRef] [Medline]
    64. Carson V, Rinaldi RL, Torrance B, Maximova K, Ball GDC, Majumdar SR, et al. Vigorous physical activity and longitudinal associations with cardiometabolic risk factors in youth. Int J Obes (Lond) 2014 Jan 26;38(1):16-21. [CrossRef] [Medline]
    65. Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, et al. Using pedometers to increase physical activity and improve health: a systematic review. JAMA 2007 Nov 21;298(19):2296-2304. [CrossRef] [Medline]
    66. Vickers AJ, Altman DG. Statistics notes: Analysing controlled trials with baseline and follow up measurements. BMJ 2001 Nov 10;323(7321):1123-1124 [FREE Full text] [CrossRef] [Medline]
    67. Paluch AE, Gabriel KP, Fulton JE, Lewis CE, Schreiner PJ, Sternfeld B, et al. Steps per Day and All-Cause Mortality in Middle-aged Adults in the Coronary Artery Risk Development in Young Adults Study. JAMA Netw Open 2021 Sep 01;4(9):e2124516 [FREE Full text] [CrossRef] [Medline]
    68. Khunti K, Griffin S, Brennan A, Dallosso H, Davies MJ, Eborall HC, et al. Promoting physical activity in a multi-ethnic population at high risk of diabetes: the 48-month PROPELS randomised controlled trial. BMC Med 2021 Jun 03;19(1):130-131 [FREE Full text] [CrossRef] [Medline]
    69. Crouter S, Schneider PL, Bassett DR. Spring-levered versus piezo-electric pedometer accuracy in overweight and obese adults. Med Sci Sports Exerc 2005 Oct;37(10):1673-1679. [CrossRef] [Medline]
    70. Feito Y, Bassett DR, Tyo B, Thompson DL. Effects of body mass index and tilt angle on output of two wearable activity monitors. Med Sci Sports Exerc 2011 May;43(5):861-866. [CrossRef] [Medline]
    71. Bassett DR, Toth LP, LaMunion SR, Crouter SE. Step Counting: A Review of Measurement Considerations and Health-Related Applications. Sports Med 2017 Jul 22;47(7):1303-1315 [FREE Full text] [CrossRef] [Medline]
    72. De Vries SI, Van Hirtum HWJEM, Bakker I, Hopman-Rock M, Hirasing RA, Van Mechelen W. Validity and reproducibility of motion sensors in youth: a systematic update. Med Sci Sports Exerc 2009 Apr;41(4):818-827. [CrossRef] [Medline]

    ‎
    ANCOVA: analysis of covariance
    CVD: cardiovascular disease
    HbA1c: hemoglobin A1c
    IPD: individual participant data
    MD: mean difference
    RCT: randomized controlled trial
    RoB: risk of bias

    Edited by R Kukafka; submitted 11.01.22; peer-reviewed by C Brakenridge, A Martinko; comments to author 23.01.22; revised version received 14.02.22; accepted 04.04.22; published 30.08.22

    Copyright

    ©Alexander Hodkinson, Evangelos Kontopantelis, Salwa S Zghebi, Christos Grigoroglou, Brian McMillan, Harm van Marwijk, Peter Bower, Dialechti Tsimpida, Charles F Emery, Mark R Burge, Hunter Esmiol, Margaret E Cupples, Mark A Tully, Kaberi Dasgupta, Stella S Daskalopoulou, Alexandra B Cooke, Ayorinde F Fayehun, Julie Houle, Paul Poirier, Thomas Yates, Joseph Henson, Derek R Anderson, Elisabeth B Grey, Maria Panagioti. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 30.08.2022.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.