Wearable Step Counts and Chronic Disease Risk: What They Show

Your daily step count may tell you more about future health than your wearer's celebratory confetti suggests. In a large EHR-linked Fitbit study, people with higher long-term step counts had lower subsequent rates of several diagnoses, including hypertension, obesity, sleep apnea, depression, GERD, and diabetes complications. The finding makes step count a useful health signal. It does not prove that hitting a particular target prevents disease.

The distinction matters. A person may walk less because early disease, pain, poor sleep, depression, or obesity has already made movement harder—months or years before a clinician enters a diagnosis. Step count can therefore function as both a behavior measure and an early warning light.

The broadest chronic-disease evidence comes from All of Us

The strongest direct study is a 2022 Nature Medicine analysis from the All of Us Research Program. Researchers linked longitudinal Fitbit data with electronic health records for 6,042 adults. Participants contributed 5,991,662 person-days and roughly 50.6 billion steps over a median four years of monitoring.

The investigators examined 1,711 diagnosis groupings, known as phecodes, and counted diagnoses that appeared only after participants had logged at least six months of step data. They excluded days with fewer than 100 steps. For Cox-model analyses, they also excluded months with fewer than 15 valid Fitbit days.

Each additional 1,000 daily steps was associated with lower odds of obstructive sleep apnea, obesity, hypertension, gastroesophageal reflux disease, major depressive disorder, and type 2 diabetes with neurological manifestations. The reported odds ratios ranged from 0.92 for several common outcomes to 0.69 for the diabetes-complication phecode. The publication establishes associations, not an individual patient's absolute risk reduction; the supplied results do not report confidence intervals for those estimates.

The pattern also argues against treating 10,000 steps as biology's official password. In this cohort, associations for diabetes and hypertension flattened around 8,000 to 9,000 daily steps. About 8,200 steps per day was associated with lower risk of obesity, sleep apnea, GERD, and major depressive disorder. Those plateaus describe population patterns for particular outcomes, not a prescription that fits every age, condition, or baseline activity level.

More steps tracked with lower future disease labeling across several categories. That is useful forecasting information, but it is not proof that adding steps produces the same change.

Why step count remains a credible health marker

The All of Us result does not sit alone, although its phenome-wide design is unusually broad. A separate 2023 All of Us analysis of activity and diabetes risk found that more favorable longitudinal activity measures were associated with lower diabetes incidence. That supports the diabetes-specific signal, one of the more biologically plausible findings in this literature.

Two 2024 umbrella reviews—one in BMJ Open and another in the Journal of Evidence-Based Medicine—also concluded that objectively measured daily steps were associated with favorable outcomes across multiple domains. Their evidence base was largely observational and leaned heavily toward mortality, cardiovascular disease, and metabolic outcomes. These reviews support the general proposition that steps carry health information; they do not validate a consumer wearable as a standalone chronic-disease prediction tool.

Simple measures may be more useful than they look. A 2025 Journal of Internal Medicine study found that simple step counting captured health information comparable to more complex accelerometer measurements. That does not make steps a complete measure of fitness, strength, sedentary time, or mobility quality. It does make them a scalable summary of habitual movement—especially when millions of people already collect them.

Mortality evidence adds context, not proof of prevention

Step-count research is more mature for mortality than for incident diagnoses. A 2022 meta-analysis of 15 international cohorts found that higher daily steps were associated with lower all-cause mortality, with lower apparent benefit plateaus in older adults than in younger adults. A 2023 JAMA Network Open study found that daily step patterns, not merely total counts, were associated with mortality among U.S. adults.

In a separate cohort of 1,629 U.S. adults with hypertension, followed for an average of 12.57 years, higher accelerometer-measured daily steps were associated with lower all-cause and cardiovascular mortality; 370 participants died, including 177 from cardiovascular causes. The 2025 NHANES analysis reported nonlinear associations. These results matter because mortality is a patient-important outcome. They still cannot establish that increasing one person's step count will cause the observed reduction in mortality.

A 2025 medRxiv preprint tested whether wearable steps and walking cadence improved five-year mortality prediction in older adults, using UK Biobank model development and external validation in U.S. NHANES. It is relevant to risk modeling, but it studied mortality rather than broad incident chronic disease and has not yet completed peer review.

The main problem is reverse causation

Observational step-count studies face a stubborn question: do more active people stay healthier, or do healthier people have the ability and opportunity to move more? The answer is probably some of both.

Reverse causation is especially plausible for sleep apnea, depression, obesity, GERD, and hypertension. Sleep disruption can drain daytime energy. Depression can reduce activity. Pain and metabolic disease can limit walking. If those processes begin before a formal diagnosis, low steps may predict the diagnosis partly because the disease has already begun.

Confounding adds another layer. Higher step counts may travel with lower adiposity, higher income, safer places to walk, more flexible work, better access to health care, and greater health awareness. Statistical adjustment can reduce these problems, but it cannot fully measure every relevant difference.

Wearable data have selection and measurement limits

The Fitbit-linked All of Us participants differed from participants with EHR data alone. They were more often White, female, and college educated; less often Black or Hispanic; and had lower baseline burdens of coronary artery disease and cancer. Wearable ownership and sustained use may therefore capture social advantage and health engagement along with physical activity.

Missingness also matters. Before monitoring, BMI was missing for 43.8% of included participants. In addition, 15.4% to 16.0% of months were excluded from Cox analyses because they had fewer than 15 valid Fitbit days. People stop wearing devices for reasons, and those reasons may relate to illness, routines, or access. The analysis cannot assume that missing data are random.

Nor are wrist-based steps a flawless measurement. A 2024 UK Biobank methods study noted that free-living wrist-worn step estimates can routinely exceed 20% error against camera-annotated ground truth. Its machine-learning approach improved step characterization, but the underlying issue remains: different devices, wrist placements, and algorithms do not necessarily count the same walk the same way. A 2022 systematic review of wearable validation studies likewise found substantial variation in methods and quality.

What to do with your own number

A low long-term step count is worth taking seriously as a health-risk signal. It can prompt a useful inventory: Has activity fallen because of fatigue, joint pain, poor sleep, low mood, breathlessness, medication effects, or a life that leaves little room to walk? That question can be more valuable than arguing over whether 7,500 or 10,000 is the superior integer.

For most people, increasing comfortable daily movement is a sensible behavior change. But wearable data do not justify telling someone that 8,000 to 9,000 steps will, by itself, prevent diabetes, hypertension, depression, sleep apnea, obesity, or GERD. The evidence supports step count as a practical behavioral marker and a plausible input for population-level risk stratification. It does not yet support a universal causal target or a clinical-grade forecast for broad chronic disease.

Use the trend, not one unusually sedentary Tuesday. If your usual step count drops and stays low, treat it as information worth acting on—through gradual movement changes, attention to symptoms, and medical evaluation when the pattern fits a health concern.