A new study using data from the National Institutes of Health’s All of Us Research Program shows that tracking physical activity over a longer period of time reveals stronger links to health conditions than the one-week snapshots common in wearable research. The study, published in Nature Communications, was led by researchers at the National Human Genome Research Institute (NHGRI) and Oxford Population Health.
Using about 11 million days of Fitbit data, the study found that a measure as simple as daily step count was associated with hundreds of health conditions. The team analysed Fitbit data from 29,303 All of Us participants who chose to share both their wearable data and their electronic health records.
They calculated four measures of daily activity ‒ step count, peak one-minute cadence (the tempo at which a physical exercise is performed), peak 30-minute cadence, and heart rate per step ‒ over time windows ranging from one day to one year. They then tested each against more than 1,400 existing and more than 700 later-diagnosed health conditions.
Longer windows revealed more
After adjusting for the many comparisons tested, average daily step count over one year was associated with 389 existing and 61 later-diagnosed conditions, compared with 271 and 33 for a single day. Some links appeared only with longer windows. For example, an association with lupus emerged at the six-month and one-year windows but not at shorter ones. Reliability gains leveled off at around 30 days of data, and even short windows captured meaningful signals, so the best window depends on the question being asked.
New and confirmed associations
The study confirmed many known relationships, such as those with obesity, sleep apnea, and type 2 diabetes, and identified more than 300 associations not reported in earlier studies of this kind. These include links to autoimmune conditions and chronic pain syndrome.
Signal or risk factor?
Some conditions, such as joint pain and rheumatoid arthritis, were linked to activity only among people who already had them, suggesting illness can change how much people move. Others showed stronger links to conditions diagnosed later, suggesting lower activity may come before some diseases. The findings suggest activity measures can serve as both early markers of disease and risk factors.
Evelynne Fulda, lead author and NIH Oxford-Cambridge Scholars Program scholar said:
Most wearable studies rely on about a week of data because that is what is practical. Having a year or more from the same people let us see how much that choice matters.
Aiden Doherty, Professor of Biomedical Informatics at Oxford Population Health and the Big Data Institute, and one of the senior authors of the paper said:
Consumer wearable devices offer a powerful opportunity to understand the impact of physical activity on our health. Millions of people use these devices every day, yet the valuable health information they generate is rarely incorporated into large-scale biomedical research. It's time for other major international studies to make greater use of the wearable data that people are already collecting.
Anya Topiwala, Senior Clinical Researcher at Oxford Population Health and the Big Data Institute, and one of the senior authors of the paper added:
The way we move could provide a digital fingerprint of our health, revealing subtle changes long before a disease is diagnosed. By tracking activity over months and years, rather than just a few days, wearable devices could help us recognise early signs of disease and identify who may be at greater risk.
The work reflects a longstanding effort by All of Us to expand wearable data sharing. Participants linked their own Fitbit devices, and the Wearables Enhancing All of Us Research (WEAR) study previously provided devices to participants from groups less represented among those who already owned one.
Stephanie Devaney, PhD, Acting CEO of the All of Us Research Program said:
Findings like this depend on scale: tens of thousands of participants sharing wearable and health record data over time, something All of Us is built to support. The longer we can follow people, the more clearly researchers can see how everyday behavior connects to health.
The authors note that this is an observational study. It shows associations, not cause and effect, and it does not show that taking more steps would prevent any condition. Follow-up time was relatively short for many chronic diseases, and people who share Fitbit data may differ from the broader population.