. 2019 Mar 27;5:2055207619839883. doi: 10.1177/2055207619839883

Table 4.

Summary of main results.

Study	Key findings				Subgroup findings and sensitivity analyses^a (if applicable)	Feasibility and additional findings (including engagement, attrition and acceptability)
Study	Impact on PA/SB^a	Impact on related health/other outcomes	Impact on PA/SB	Impact on other outcomes
Brakenridge et al. 2016³⁷; Brakenridge et al. 2016³⁸	No significant between-group difference in PA or SB at 3 months (although the organisational support group only showed an increase in overall standing time, +14.6 min/day, 95% CI 2.5–26.8, P = 0.018). Significant between-group adjusted mean difference (MD) in overall daily stepping time (+20.6 min, 95% CI 3.1–38.1, P = 0.021) and number of daily steps (+846.5 steps, 95% CI 67.8–1625.2, P = 0.033) at 12 months – favouring the intervention group.	No significant within- or between-group findings for any health or work-related outcomes.	↑ PA ↑ SB (favoured control)	0	Sensitivity analysis: study completers were more likely to show larger and statistically significant changes in activity at 3 months (bias from dropout of healthy participants?).	Engagement: 70.5% of participants provided with the activity monitor used it in the first 12 weeks, with mean usage of 12.1±11.6 days. Use had ceased by 12 months in all intervention participants. Acceptability: 41/153 (26.8%) participants reported at least one adverse event (e.g. reactions to activity monitor or accelerometer wear).
Finkelstein et al. 2015³⁹; Finkelstein et al. 2016⁴⁰	No between-group difference in MVPA (P = 0.0854) or steps (P = 0.1362) between the Fitbit only and control groups at 6 months. Cash and charity incentive groups showed higher MVPA compared with control at 6 months (cash group MD = +29 MVPA bout mins/week, 95% CI 10–47, P = 0.0024; charity group MD = +21 MVPA bout mins/week, 95% CI 2–39, P = 0.0310). At 12 months, the Fitbit and charity groups showed higher MVPA than the control (Fitbit group MD = +37 MVPA bout mins/week, 95% CI 19–56, P = 0.0001; charity group MD = +32 MVPA bout mins/week, 95% CI 12–51, P = 0.0013). The cash incentive group did not differ from the control (P = 0.1363). The only significant within-group increase in MVPA at 12 months was for the Fitbit only group (+16 mins/week, 95% CI 2–30, P = 0.0301). The only significant between-group differences in mean daily steps were for the cash vs. control group at 6 months (+1050 steps, 95% CI 600–1490, P < 0.0001) and 12 months (+500 steps, 95% CI 50–960, P = 0.0289).	No evidence for improvement in health outcomes - all intervention groups showed improvement in cardiorespiratory fitness (NET-FVO₂ max) at 6 and 12 months, but control group showed improvement at 12 months.	↑ PA	0 (improved aerobic fitness in control and intervention groups)	Subgroup analysis by baseline activity level found some differences – e.g. those in the cash and charity groups who were insufficiently active at baseline showed a significant increase in MVPA from baseline to 6 months (+22 mins/week, 95% CI 5–38, P = 0.0096 and +17 mins/week, 95% CI 2–32, P = 0.0231 respectively). For those who were sufficiently active at baseline, changes in MPVA were non-significant in both groups.	Engagement: 40% abandoned the Fitbit within 6 months, and by month 12 only around 10% of all participants in the intervention groups were still wearing the device. Attrition: predictors of loss to follow-up at 12 months included gender (higher attrition in females) and ethnicity (lower attrition in Chinese). Higher adherence was seen in the cash incentive group compared with the other groups (particularly at 6 months).
Ganesan et al. 2016⁴¹	Significant increase pre- to post-intervention in mean daily steps (+3519 steps, 95% CI 3484–3553, P < 0.0001), exercise days/week (+0.89 days, 95% CI 0.87–0.92, P < 0.0001) and odds of exercising ≥30 min/day (1.65, 95% CI 1.61–1.68, P < 0.0001). Significant decrease pre- to post-intervention in mean sitting duration (−0.74 h/day, 95% CI –0.78 to –0.71, P < 0.0001).	Significant pre- to post-intervention reduction in weight (−1.45kg, 95% CI –1.53 to –1.38, P < 0.0001).	↑ PA ↓ SB	↓ weight	Subgroup analysis by gender, year cohort, geographic region and income group – no significant differences for any of the main outcomes. Men showed greater weight loss than women (−1.63kg, 95% CI –1.72 to –1.54 compared with –0.74kg, 95% CI –0.91 to –0.57). Predictors of weight loss included increase in step count, increase in exercise days and decrease in sitting duration.
Gilson et al. 2016⁵⁵; Gilson et al. 2017⁵⁶	Non-significant increase in mean proportions of work time spent physically active from baseline to post-intervention and follow-up (+1%, 7 min/day). Non-significant decrease in mean proportions of work time spent sedentary at post-intervention (−6%) and follow-up (−9%). Significant increase in mean proportion of workday non-work time spent sedentary baseline to follow-up (P = 0.007) and decrease in mean proportion of workday non-work time stationary+ baseline to post-intervention (P = 0.037) and follow-up (P < 0.033). 65% of participants showed positive changes in PA (and at least one dietary choice) at follow-up.	Significant increase in workday fruit (P = 0.023) and vegetable (P= 0.024) consumption by one serving/day at end of programme.	0 PA ↑ SB (workday non-work time only)	↑ diet(fruit and vegetable intake)	N/A	Engagement: 26/44 (59%) participants used the activity monitor. Use for step count monitoring remained constant but dietary logging significantly declined from baseline to study completion. Attrition: moderately high - only 19/44 (43%) participants completed the study. Acceptability: Barriers to technology use included technical issues, data usage costs and privacy concerns. From qualitative interviews, the overall intervention was perceived as feasible, acceptable and as having positive impact on PA by drivers and depot managers. The mHealth component was perceived to have a greater impact on behaviour than financial incentives.
Gremaud et al. 2018⁶⁰	Relative to control (Fitbit-only group) after the start of the intervention, the smartphone app users showed an immediate increase in mean daily steps of 2183 (95% CI 992–3344). Daily active minutes similarly increased by 12.8 (95% CI 6.3–19.3). However, participants’ steps declined during the study period overall. The mean longest bouts of sedentary time decreased by 26.6 min (95% CI –70.9 to–17.3) in the intervention group relative to control.	Not studied	↑ PA ↓ SB (but control group received mHealth intervention)	Not studied	Sensitivity analyses accounting for wear compliance, and excluding data collected following a bug in the app platform did not alter the significance of the findings.	Engagement: compliance with Fitbit wear declined over time, but app users were more likely to wear the Fitbit daily than the Fitbit-only group. Low attrition: 144/146 (99%) participants provided follow-up data (although short-term study). Acceptability: participants rated the app as easy and enjoyable to use and useful for increasing PA. The main reported barrier to technology use was Fitbit battery issues (8/48, 17% of respondents).
Jones 2016⁴²	Between-group difference in % change in mean daily steps neared significance: IG1 (activity monitor only) = +9%IG2 (activity monitor and active desk) = -17% CG = –15% (P = 0.06, favouring the activity monitor only group) Significant increase in sedentary time in intervention groups compared with control:IG1 = +255.5 min/day, 95% CI 127.5–383.5 IG2 = +353.2 min/day, 95% CI 219.1–487.3 CG = 0 min/day, 95% CI –58.6 to 58.5 (P < 0.0001)	No significant within- or between-group findings for BMI, sleep or any psychosocial outcomes.	0 PA ↑ SB (favoured control)	0	N/A	Engagement: participants reporting follow-up data wore the Fitbit activity monitor for a mean of 177 of 210 days (84% adherence). Attrition: dropouts were similar to study completers in terms of baseline activity, weight, gender and ethnicity.
Koyle 2013⁴³	The ‘control’ group showed a significant within-group reduction in mean weekly walking duration from week 1 to week 6 (−50.3 min, P < 0.001). The reduction in the intervention group was non-significant (−20.8 min/week, P = 0.99). After controlling for baseline activity level, the intervention group walked more minutes than the control but a significant difference was found only at week 6 (β = 38.21, P = 0.03).	Significant pre- to post-intervention increase in self-efficacy beliefs (walking self-efficacy scale) for both groups: Intervention 85.6±12.1 to 90.0±10.8 (P = 0.0003) and control 78.3±14.7 to 87.3±11.6 (P = 0.0288). Significant decrease in resting mean pulse rate for intervention group only: 70.0±11.0 to 63.2±10.2 (P = 0.038). No increase in likeliness to participate in other PA. No significant within- or between-group findings for weight, BMI or systolic BP.	↑ PA (but control group received mHealth intervention)	↑ self-efficacy ↓ resting pulse rate	N/A	Acceptability: Participants in both groups commented positively on their experiences of taking part and found the study motivating. Text messages were reported as helpful.Many participants felt stronger and/or reported having lost weight as a result of taking part.At the end of the trial, all but one person chose to continue using the app.
Losina et al. 2017⁵⁷	Average weekly duration of MVPA increased from 54±64 min in 2 weeks pre-intervention to 62±89 min post-intervention (statistical significance of changes not reported). 86% of participants met either their weekly PA goal or CDC PA guidelines for ≥6/24 weeks. 52% met either their PA goal or CDC guidelines for ≥12/24 weeks.	Not studied	?(statistical significance of PA changes not reported and pre-intervention measure was contaminated by use of Fitbit)	Not studied	Subgroup analysis by baseline physical ability, ethnicity and obesity found some differences. Those able to walk a mile at baseline (vs. those unable, P = 0.010), African Americans (vs. all other ethnicities, P = 0.016) and non-obese participants (vs. obese, P = 0.018) met PA guidelines more consistently throughout the programme.	Engagement: 63% of participants were classed as adherent Fitbit wearers (i.e. wearing Fitbit ≥4 days/week for ≥20 weeks). Wear declined over time (e.g. 94% adherent wearers after 1 month vs. 62% after 6 months). Acceptability: two-thirds of participants were satisfied with the programme; 79% indicated they would participate again.
Neil-Sztramko et al. 2017⁵⁸	Significant increase in mean total MVPA (+110.3 min/week, P < 0.01) and significant increase in mean daily steps (+1488.7, P < 0.01) from baseline to post-intervention. Significant reductions in objective mean sedentary time (−405.5 min/week, bouts ≥10 mins, P = 0.02) and self-reported mean sedentary time (−425.3 min/week, P < 0.01) from baseline to post-intervention.	Significant pre- to post-intervention reductions in weight (mean change –0.9kg, P = 0.03) and BMI (mean change –0.3kg/m², P = 0.04). Significant improvements in some domains of health-related quality of life: energy/fatigue (P = 0.01) and emotional wellbeing (P = 0.04). Significant improvements in sleep disturbances (P = 0.04) and day dysfunction due to sleepiness (P = 0.04).	↑ PA ↓ SB	↓ weight and BMI ↑energy/fatigue and emotional wellbeing score ↓ sleep disturbance and day dysfunction due to sleepiness	N/A	High engagement with Fitbit: all participants reported using the Fitbit. Of the 18 who provided Fitbit data, the device was worn 94.5% of the total study period. Attrition: all participants completed the study. High acceptability: 16/19 (84.2%) participants were very or somewhat satisfied with the intervention. Demand: high demand for participation. Recruitment to time and target was exceeded. Implementation: technical issues were common; 5 (25%) Fitbits were returned. A further two devices were lost.
Olsen et al. 2018⁶¹	No significant changes in sedentary time from pre- to post-intervention: Mean change in accelerometer-assessed sedentary min/day +0.08 (95% CI –30 to +30). Mean change in total self-reported sitting time in office –56 mins/day (95% CI –128.5 to +17.0) and when working at home +20.5 (95% CI –64.5 to 105.5). No significant changes in accelerometer-assessed or self-reported PA.	Not studied	0 PA 0 SB	Not studied	N/A	Moderate attrition: only 30 of the 49 (61%) recruited participants provided some post-intervention data. Acceptability: overall acceptability of the intervention was high. 90% of participants were satisfied or very satisfied with the programme and 83% agreed that the tracker was a useful tool for behaviour change.
Patel et al. 2018⁶²	Compared with control during the intervention period, the (unadjusted) mean proportion achieving the 7000 step goal was significantly higher for the combined lottery incentive group only (0.38 vs. 0.26). The adjusted odds ratio for achieving the goal (combined vs. control) was 3.00 (95% CI 1.28–7.02, P = 0.012). No significant differences were sustained at follow-up, after incentives were removed. No significant between-group differences in mean daily steps (within-group changes not reported).	Not studied	↑ PA (but control group received mHealth intervention)	Not studied	Sensitivity analysis: adjusting for device and missing data did not change the significance of the findings.	Low attrition: only 6% of participants did not complete the study.
Patel et al. 2016⁴⁴	No significant between-group differences in mean daily steps (within-group changes not reported). Proportion of participant-days 7000 step goal was achieved was significantly higher for loss incentive group compared with control for the 13 week intervention period (MD = +0.16, 95% CI 0.06–0.26, P = 0.001). This effect was not sustained at follow-up after incentives were removed.	Not studied	↑ PA(but control group received mHealth intervention)	Not studied	Sensitivity analysis: adjusting for device and different methods of accounting for missing data did not affect the significance of the findings.	Low attrition: only 5% of participants did not complete the study.
Patel et al. 2016⁴⁵	Compared with control during the intervention period, the mean proportion achieving the 7000 step goal was significantly higher for the combined incentive group only (MD = +0.17, 95% CI 0.07–0.28, P < 0.001). The combined incentive group also had higher mean daily steps than the control at the end of the intervention period (MD = +1446 steps, 95% CI 448–2444, P = 0.005). No significant differences were sustained at follow-up.	Not studied	↑ PA (but control group received mHealth intervention)	Not studied	Sensitivity analysis: adjusting for device and missing data did not affect the significance of the findings.	Low attrition: only 4% of participants failed to complete the study.
Poirier et al. 2016⁴⁶	Mean daily steps pre- to post-intervention increased for the intervention group (+309 steps/day, ±1874) and decreased for the control group (−661 steps/day, ±1824). MD = 970 steps/day, P < 0.001. The proportion of participants achieving an increase of 1000 steps/day was significantly greater in the intervention group (29.9%) than the control (16.4%), P = 0.018.	Not studied	↑ PA	Not studied	Subgroup analysis by baseline activity level: Sedentary group ( < 5000 steps/day): Mean change +594±1558 steps/day in intervention group vs. +47±1299 steps/day in control group, P = 0.04. Low-to-somewhat active group (5000–9999 steps/day): Mean change of –110±2106 steps/day in intervention group vs. –1286±1783 steps/day in control group, P = 0.004. Sensitivity analysis: included some study non-completers – statistically significant between-group difference in mean daily steps remained (P < 0.001).	High engagement with intervention: Participants wore the activity monitor on 78.6% of days (33/42) on average; e-mails were opened on 21.9% of days (9.2/42); website visits occurred every 3.6 days (11.8/42). 130/133 (97.7%) intervention participants still wore the activity monitor, opened e-mails and/or visited the website after 6 weeks. Attrition: moderately low (around 82% provided complete follow-up data) although short term study. Participants with complete outcome data were similar to those without in terms of baseline PA level, ethnicity, income and education.
Reijonsaari et al. 2009⁴⁷; Reijonsaari et al. 2012⁴⁸	No significant within- or between-group differences in PA: 6-month between-group adjusted MD = –365 weekly MET-minutes, 95% CI –733 to 3; 12-month between-group adjusted MD = –207 weekly MET-minutes, 95% CI –531 to 116 (negative values favour control).	No significant between-group difference in productivity (adjusted MD in QQ score at 6 months = 1.3, 95% CI –2.0 to 4.7 and adjusted MD at 12 months = –1.1, 95% CI –4.9 to 2.8). Between-group difference for change in weight and % body fat favoured control (adjusted MD for weight change at 12 months = –0.5kg, 95% CI –1.0 to 0.0; adjusted MD for change in % body fat = –0.6%, 95% CI –1.0 to –0.2).	0 PA	↑ weight (favoured control) ↑ % body fat (favoured control)	Subgroup analyses by gender, job characteristics, age, self-rated baseline PA level and sickness absence days in the past year did not modify the results. Adherence to the intervention did not mediate sickness absence (MD between adhering and non-adhering subgroups was 0.0 days, 95% CI –1.2 to 0.9). Sensitivity analysis: adjusting for missing data did not affect the results.	Engagement: decline in engagement (use of website, communication with coaches) over time, particularly in the last 6 months.
Reed et al. 2018⁶³	Initial increase in MVPA but significant decline from week 2 to week 6 (i.e. post-intervention), P < 0.05. Significant decline in daily steps from baseline to week 6 (P < 0.05). There were no significant between-group differences in either MVPA (P = 0.292) or steps (P = 0.333).	Within-group significant reductions in % body fat (−0.8±4.8, P = 0.015) and resting systolic BP (−2.6±8.8 mm Hg, P = 0.019). No significant within- or between-group changes in body mass, BMI or waist circumference.	↓ PA(steps only, and no control group)	↓ % body fat ↓ systolic BP	N/A	Engagement: Participants wore the activity monitor for at least 10 hours/day for 31/42 intervention days on average (overall compliance rate 74%). Wear declined over time (e.g. average of 6.0±1.9 days per week at baseline compared with 3.5±3.0 days in week 6). Low attrition: 72/75 (96%) participants completed the study.
Rowe-Roberts et al. 2014⁴⁹	Findings only reported descriptively. Average daily steps reported by AUSDRISK score at beginning and end of trial; ‘high’ baseline score participants that moved to ‘low’ risk at the end of the study had the highest average daily steps at the end of the study (12,294). Average overall daily steps were: High risk group: 8588 Medium risk group: 7836 Low risk group: 7878	23% of participants reduced their AUSDRISK score over 7 months.	?(pre- to post- change in PA unclear and no control group)	↓ AUSDRISK score	Results stratified by AUSDRISK score – device seemed more effective for those at high risk of diabetes (see left).	Engagement: overall low engagement with the activity monitor – average monthly dropout rate was 15% and only 36% of participants were still using the device at the end of the study. High baseline diabetes risk participants showed the highest level of engagement: Mean number of months engaged with the activity monitor = 5.7 for high risk; 4.4 for medium risk; 4.2 for low risk Acceptability: low engagement was predominantly driven by device issues, e.g. broken, lost, forgotten devices. Individual differences in preferred motivational strategies, e.g. inactive/unengaged participants preferred games whereas active/engaged participants preferred ‘goal-oriented functionalities’, e.g. smart reminders and normative information about appropriate PA levels.
Simons et al. 2018⁶⁴; Simons et al. 2018⁶⁵	No significant between-group differences in any of the objective or self-reported PA outcomes at post-intervention or follow-up (PA decreased over time in the intervention and control groups).	No significant impact on any self-reported psychosocial variables.	0 PA	0 (improved knowledge of PA guidelines in control and intervention groups)	N/A	Engagement: Decline in engagement over time, with significant reductions in Fitbit wear, viewing graphs in the app and reading notifications. Attrition: 110/130 (85%) participants provided primary outcome data at follow-up. Acceptability: The majority of participants rated the Fitbit and app as self-explanatory (36/51, 71%), user friendly (40/51, 78%), and interesting (34/51, 67%), but few found the tips and facts motivating (10/41, 24%), used them to be physically active (8/41, 20%) and believed they were tailored to their lifestyle (7/41, 17%). Barriers to technology use included technical problems and forgetting to wear or charge the Fitbit.
Schrager et al. 2017⁵⁰	No significant overall change in PA level. Self-reported median (IQR) days/week of ≥30 min PA: Baseline 2.5 (1.9) 1 month 2.8 (1.5) 6 months 3.0 (2.0) (P = 0.67 for change baseline to 1 month; P = 0.36 for change baseline to 6 months) PA monitor-measured median (IQR) days/week ≥10,000 steps or ≥30 min PA: Baseline 2.5 (1.9) 1 month 2.5 (2.7)(P = 0.69 for change baseline to 1 month)	18/30 (60%) participants described a positive impact on their wellness after one month of activity monitor use.	↑ PA (low baseline activity level only)	↑ wellness (qualitative report only)	Subgroup analysis by baseline activity level and device use. Significant increase in self-reported median (IQR) days/week of ≥30 min PA for the most inactive (n = 10): Baseline 1.5 (0.9) 1 month 2.4 (1.2) 6 months 2.0 (2.0) (P = 0.04 for change baseline to 1 month; P = 0.04 change baseline to 6 months) No significant between- or within-group differences in PA level for those who used the activity monitor for 6 months and those who discontinued use prior to the study end.	Engagement: decline in engagement over time – 67% continued to use the device after one month, but only 33% still used their device after 6 months. Acceptability: barriers to use included forgetfulness, not wanting to wear the device, boredom, beliefs it was not accurate, technical issues and fashion.
Skogstad et al. 2016⁵¹	Significant increase in self-reported frequency of PA from baseline to follow-up (P = 0.001). % exercising at baseline:37% ≤1 day/week 47% 2–3 times/week 15% ≥4 times/week % exercising at follow-up: 13% ≤1 day/week 58% 2–3 times/week 28% ≥4 times/week Half of participants reported increased PA frequency at follow-up. Mean increase in daily low intensity PA (e.g. walking): 13.7±29.4 min for men 13.7±17.2 min for women. Mean increase in daily high intensity PA (e.g. jogging: 8.3±18.2 min for men 8.6±14.6 min for women	Significant improvement in maximal oxygen uptake (+2.8 ml/kg/min, 95% CI 1.4–4.3, P = 0.00022). Significant reduction in total cholesterol (−0.12 mmol/L, 95% CI –0.22 to –0.01, P = 0.032) and LDL cholesterol (−0.13 mmol/L, 95% CI –0.22 to –0.04, P = 0.0034). Significant increase in diastolic BP (+1.67 mm Hg, 95% CI 0.23–3.12, P = 0.024). Despite the significant improvements in health outcomes (aerobic fitness, total and LDL cholesterol), further analysis could not attribute these to the change in individual PA levels.	↑ PA	↑ aerobic fitness ↓ total cholesterol ↓ LDL cholesterol ↑ diastolic BP	Subgroup analysis by education. Education was an effect modifier for diastolic BP, total and LDL cholesterol: Only workers with low education showed a significant increase in diastolic BP (+4.4 mm Hg, 95% CI 2.03–6.86, P = 0.0004). Only workers with high education showed a significant decrease in total cholesterol (−0.21 mmol/L, 95% CI –0.08 to –0.34, P = 0.0015) and LDL cholesterol (−0.22 mmol/L, 95% CI –0.12 to –0.32, P = 0.0001).	Attrition: Participants lost to follow-up differed from study completers – younger, lower HDL and higher CRP (and mostly men and blue collar workers). Acceptability: perceived impact on other outcomes – 12 participants reported improved nutritional habits at follow-up, 3 increased quality of sleeping, 4 reduced or quit smoking and 2 reduced alcohol intake.
Slootmaker et al. 2009⁵²	No overall significant between-group difference in sedentary time or PA (mins/week) at 3 months or 8 months follow-up. 3 month between-group difference (adjusted for gender, age, education and BMI), β and 95% CI: Sedentary time: 10 (−435 to 455) Light intensity PA: –129 (−337 to 79) Moderate intensity PA: -13 (−89 to 63) Vigorous intensity PA: –6 (−75 to 62)	No overall significant intervention effect on any secondary outcomes (aerobic fitness, determinants of PA and body composition) at 3 or 8 months.	↓ PA (light-intensity, highly educated only) 0 SB	↑ awareness of PA level (overweight only) ↓ body weight (lower educated only)	Subgroup analysis by education, adherence to programme and BMI. Education was an effect modifier for PA; higher educated intervention participants showed a significant reduction in light intensity PA at 3 months compared with control: adjusted difference in mins/week, β and 95% CI = –349 (−632 to –66), P = 0.02. The proportion of participants aware of their adherence to PA recommendations increased among overweight participants in the intervention group compared with control at 3 months (adjusted OR = 16.4, 95% CI 1.3–214, P = 0.02). There was a reduction in body weight among the lower educated intervention participants compared with control at 3 months (adjusted difference, β = –1.6kg, 95% CI –2.8 to –0.4, P = 0.01). Higher engagement/adherence to the programme did not result in increased PA.	Engagement: majority of participants engaged with the intervention; 73% wore the activity monitor regularly and the website was used almost once a week. Acceptability: barriers to technology use included lack of interest and difficulty finding items on the website.74% of activity monitor users read the tailored advice, of whom 39% found it unappealing.
Thorndike et al. 2014⁵³	At the end of phase 1, there was no significant difference between the intervention and control groups in median daily steps overall (intervention 6369 vs. control 6063, P = 0.16) or mean daily steps on days the monitor was worn (intervention 7886±3622 vs. control 7600±3492, P = 0.63). Mean daily steps were significantly higher in phase 2 (team competition) than phase 1 (individual monitoring) for those assigned to the control group in phase 1 (7971 vs. 7567, P = 0.002) but not for those in the intervention group (7832 vs. 7739, P = 0.13).	Significant overall reduction in systolic BP from baseline to end of study: 121±15.4 mm Hg to 117±12.6 mm Hg (P = 0.004). Significant overall increase in HDL cholesterol from baseline to end of study: 57±14.7 mg/dL to 61±15.7 mg/dL (P < 0.001). No significant change in diastolic BP, weight, BMI, waist circumference, total or LDL cholesterol.	↑ PA(team-based competition only)	↓ systolic BP ↑ HDL cholesterol	The authors compared mean daily steps during inpatient rotations with outpatient rotations during the whole study; physicians were significantly more active during outpatient rotations (difference of 648 steps, P < 0.001).	Engagement: compliance with wearing the activity monitor was significantly higher in phase 2 than phase 1: 77% vs. 60%, P < 0.001.
Torquati et al. 2018⁶⁶	Significant reduction in daily MVPA from baseline to 3 months, median (IQR): 19.1 (24.6) to 13.3 (13.9) min/day (P = 0.01). Near significant reduction in daily MVPA from 3 months to 6 months: median (IQR): 13.3 (13.9) to 12.5 (13.4) min/day (P = 0.07). Significant reduction in mean daily steps from baseline to 3 months (8496±2528 to 8136±2395, P = 0.04). No significant changes in sedentary time from baseline to 3 months (P = 0.17) or from 3 months to 6 months (P = 0.64).	Significant increase in daily fruit and vegetable intake from baseline to 3 months (P = 0.04). Significant improvement in self-rated health from month 3 to month 6 (P < 0.05). No significant changes in weight, BMI, waist circumference, blood pressure, PA or diet self-efficacy or social support.	↓ PA 0 SB	↑ diet(fruit and vegetable intake) ↑ self-rated health	Subgroup analysis with participants with complete data only did not change the significance of the findings (although MVPA increased at month 6 following an initial reduction).	Engagement: Low engagement with the smartphone app, with 68.4% using it less than once per month or never. PA goals were set infrequently, and social components were not used. Attrition was high, with only 12/47 (26%) attending the 6 month follow-up. Acceptability: Participants reported that changing both PA and diet at the same time was challenging. Interviews revealed low perceived usefulness of the smartphone app. Overall reach was poor (13% of potential participants were reached and 9.4% were willing to take part). Implementation: Participants required more frequent contacts with researchers or a workplace champion.
van Dantzig et al. 2013⁵⁴	A significantly higher reduction in computer activity (mean minutes of activity, 30 min before and after receiving (virtual) text message) was observed in the intervention group compared with control: Intervention group reduction of 10 min vs. control group reduction of 5.9 min, P < 0.001. Non-significant within- or between-group change in PA (average value during 5 min interval before and after a message) (P-value not reported).	Not studied	0 PA ↓ SB (but control group received mHealth intervention)	Not studied	Engagement with the intervention was explored with subgroup analysis. There was no significant impact of proportion of text messages read on computer activity (P>0.10) and no significant interaction between proportion read and time (P>0.10), i.e. receiving messages led to breaks but content was not important.	Engagement: an average of 46% (SD = 34.6) of the total number of text messages sent were read.
Yeung et al. 2017⁵⁹	Significant increase in median daily steps from blinded to unblinded intervention period (P = 0.001): Median (IQR) Weeks 1–4 (blinded): 7260 (2410) steps/day Weeks 5–8 (unblinded): 8266 (3306) steps/day Significant increase in number of participants achieving an average of ≥10,000 steps/day (P = 0.04): Weeks 1–4 (blinded): n = 9 (12%) Weeks 5–8 (unblinded): n = 17 (23%)	Not studied	↑ PA	Not studied	Subgroup analysis by occupation. Surgical residents had significantly higher steps than non-surgical residents (P = 0.018), however differential impact of the intervention was unclear. Participants who elected to join the activity tracking group had higher median daily steps than those who did not (7938 vs. 7442, P = 0.042), however the direction of the effect is unclear (may be due to prior increased motivation).	Moderate engagement with Fitbit: device wear ranged from 91% in those volunteering for the activity tracking group to 51% in those not volunteering (unblinded period). A decline in engagement was seen over the 12 weeks in the latter group only. Attrition: high adherence with 74/86 (86%) participants completing the study. Acceptability: The most common barrier to participation was loss of the Fitbit (7/86, 8%).

IG: intervention group; CG: control/comparison group; PA: physical activity; SB: sedentary behaviour; MVPA: moderate to vigorous physical activity; MD: mean difference; BMI: body mass index; BP: blood pressure; ±: standard deviation; N/A: non-applicable; AUSDRISK: Australian Type 2 Diabetes Risk Assessment Tool; IQR: interquartile range; LDL: low-density lipoprotein; HDL: high-density lipoprotein; CRP: C-reactive protein; OR: odds ratio; CDC: Centers for Disease Control.

↑ = significant increase in outcome ↓ = significant decrease in outcome 0 = no significant change in outcome (between-group or within-group)

^aSignificant P-values (where reported in included papers) are in bold