Wearables in Cardiac Rehab Work—If You Keep Wearing Them

A tracker only works while it is on the wrist

Wearable devices can increase walking in people with cardiovascular disease, including those in cardiac rehabilitation. The catch is less glamorous than the hardware: people use the devices very differently. In a 2025 secondary analysis of the RESILIENT trial, older adults in mHealth cardiac rehabilitation fell into at least three wear-pattern groups—consistently high use, declining use, and persistently low use. Participants in the high-adherence group had greater improvement in the six-minute walk test. But the analysis cannot yet tell clinicians, before treatment starts, who will end up in which group.

That distinction changes how to read the broader evidence. A wearable is not an exercise program in a tiny rectangle. It is a feedback tool that can support goals, self-monitoring, and reinforcement. When someone stops wearing it, the feedback loop stops too.

Step counts rise on average, with wide variation between programs

The best pooled estimate comes from a 2023 systematic review and meta-analysis by Heizmann et al. of 14 randomized controlled trials involving about 1,270 cardiovascular disease and cardiac rehabilitation participants. Wearable-based interventions with feedback increased daily steps over three months or less, with a standardized mean difference (SMD) of 0.54 (95% CI, 0.36 to 0.72; p<0.001). At more than three months, the estimated effect remained positive at SMD 0.47 (95% CI, 0.15 to 0.79).

Those are statistically significant average effects, not a promise of a specific number of additional steps for any one patient. The trials differed substantially: heterogeneity was high at I²=73% for short-term step counts and 79% for medium-term results. Programs varied in participants, devices, counseling, goals, and feedback. Pooling them gives a useful direction of travel, but not a personalized forecast.

A 2021 cardiac-rehabilitation-specific meta-analysis by Ashur et al. examined nine randomized trials and also found significant increases in steps. This supports the overall finding, although it should not be treated as a wholly independent replication: formal cardiac rehabilitation is a subset of the broader cardiovascular population, so some trial overlap with the 2023 review is likely.

The interventions with larger effects commonly combined goal-setting, self-monitoring feedback, and positive reinforcement. That pattern makes practical sense, but it also creates an attribution problem. The research cannot cleanly separate the effect of the device from the effect of coaching and behavioral support wrapped around it.

Fitness signals are real, but their clinical meaning remains unsettled

In the Heizmann meta-analysis, wearable interventions were associated with statistically significant improvements in functional outcomes: six-minute walk test performance improved by SMD 0.43 (95% CI, 0.18 to 0.69), and peak oxygen uptake (VO2peak) improved by SMD 0.31 (95% CI, 0.04 to 0.57). These results make a benefit plausible, but they do not establish a clinically meaningful improvement for an individual patient.

Clinicians often use roughly 30 meters for the six-minute walk test and 1 mL/kg/min for VO2peak as reference points for meaningful change in cardiac and pulmonary populations. The pooled standardized effects cannot reliably be converted into patient-level clinical gains without making assumptions about baseline variability across trials. Their confidence intervals also span a range that could fall below or exceed those practical thresholds. Track steps for accountability, not because the tracker alone has proved it can transform cardiorespiratory fitness.

AI phenotyping identifies a useful research question, not a screening test

The RESILIENT secondary analysis by Barua et al. followed 298 adults aged 65 or older with ischemic heart disease during a 90-day Fitbit Inspire mHealth cardiac rehabilitation program at five U.S. academic centers. Researchers divided device use into six 15-day windows and defined a valid wear day as at least 100 steps. K-means clustering identified adherence trajectories that included sustained high use, declining use, and persistently low use; the sustained-use group improved more on the six-minute walk test.

This is a sensible explanation for why pooled effects can look modest: a group that rarely wears the device cannot receive much device-based feedback. Still, the study is exploratory. The investigators chose cluster counts after examining silhouette candidates of two, five, and eight clusters, and they did not test the phenotypes in a holdout or external validation sample. The participants were also older adults treated at academic centers, with dementia and substantial mobility impairment excluded. The findings justify prospective validation; they do not justify assigning patients to high- or low-adherence categories before rehabilitation begins.

Remote care may work, but coaching and setting still need direct tests

A 2025 randomized trial by Nishio et al. directly compared wearable-supported remote cardiac rehabilitation with and without weekly online coaching in patients with coronary artery disease. It is the only head-to-head coaching comparison in this evidence base. The available report establishes that the design addresses a useful question, but it does not provide full between-group effect estimates here. It therefore cannot settle whether weekly human coaching adds enough benefit to justify the added effort.

Evidence from peripheral artery disease offers context, not a substitute for cardiac trials. An individual participant data meta-analysis found that home-based walking improved six-minute walk distance as much as, or more than, supervised treadmill programs in peripheral artery disease. Peripheral artery disease and coronary disease share vascular risk, but they are different populations with different rehabilitation pathways. Remote cardiac rehabilitation still needs direct comparisons with center-based care.

Population-specific programs can also disappoint. In the open-label OPTICARE XL randomized trial, 102 patients assigned to a one-year obesity-focused cardiac rehabilitation program were compared with 99 patients receiving standard rehabilitation. The specialized program included exercise, behavioral coaching, aftercare, and physical-activity monitoring, yet it did not improve the primary outcome of health-related quality of life beyond standard care. More structure does not guarantee a better patient-reported outcome.

The harder problem is keeping behavior after the program ends

Wearable-supported behavior changes often fade when active support ends. In the 12-week SIT LESS randomized trial among patients with coronary artery disease, a cardiac-rehabilitation-integrated behavioral intervention reduced sedentary time by 0.2 to 0.4 hours per day during the active period. At two-month follow-up, the reduction was about 0.3 hours per day versus baseline. The trial shows some persistence, but less than the active-phase improvement, with sedentary time still high.

A separate 2024 prospective cohort study by Lönn et al. found that accelerometer-measured physical activity declined progressively throughout the first year after myocardial infarction, regardless of cardiac rehabilitation participation. Neither study randomized people to continued versus discontinued device support, so neither can prove that removing a tracker caused the decline. Together, they show that maintenance without ongoing structure is not the default.

What to do with a wearable in cardiac rehabilitation

For people with established cardiovascular disease or recovering from myocardial infarction, a wearable can be a reasonable adjunct to standard cardiac rehabilitation. The evidence supports a moderate average increase in steps when tracking is paired with structured feedback, while leaving substantial uncertainty about the size and durability of the benefit for an individual.

• Use the device to support a specific walking target, regular review, and reinforcement.
• Plan maintenance before the formal program ends: scheduled check-ins, coaching, or a social commitment can preserve the accountability that the device alone does not provide.
• Treat sedentary time and walking as related but distinct targets. A higher step count does not automatically erase long periods of sitting.
• Do not use current AI adherence clusters to decide who deserves rehabilitation support. The prediction tool has not been validated.

An umbrella review by Alam et al. found methodological flaws in all six systematic reviews and meta-analyses it assessed, including critical flaws in several. That does not erase the consistent positive direction of the step-count evidence. It does mean the exact pooled number deserves less confidence than the basic conclusion: tracking helps some cardiac rehabilitation patients walk more, especially when a real behavioral program keeps the tracker from becoming expensive drawer décor.