Daily walking patterns help flag early Parkinson’s signs in large UK study

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Nov 25 2025

New UK Biobank data reveal that falling daily step counts may flag emerging Parkinson’s years before diagnosis, offering a simple digital clue to early neurodegenerative changes.

Study: Daily steps are a predictor of, but perhaps not a risk factor for Parkinson’s disease: findings from the UK Biobank. Image Credit: Mladen Mitrinovic / Shutterstock

In a recent study published in the journal npj Parkinson’s Disease, a group of researchers examined whether accelerometer-measured daily step counts predict incident Parkinson’s disease (PD) and whether observed associations persist when early follow-up windows, which are most prone to reverse causation, are excluded.

Growing Interest in Wearables and Early Parkinson’s Detection

One in three adults now tracks activity with a smartphone or wearable, yet how those numbers map to neurodegenerative risk remains unclear. PD affects millions worldwide and often develops silently for years before diagnosis, with subtle motor slowing that families may notice. While self-reported activity appears protective in cohort studies, recall bias and reverse causation complicate interpretation. Device-measured steps offer objective metrics that could differentiate true risk factors from early prodromal signs.

UK Biobank Accelerometer Cohort and Data Processing

This prospective analysis used the UK Biobank accelerometer sub-study. Participants wore Axivity AX3 wrist devices for up to seven days, and daily steps were calculated via the OxWearables “stepcount” machine-learning model. After exclusions for device errors, insufficient wear, extreme acceleration, prevalent PD, and missing covariates, 94,696 adults were eligible. Incident PD was identified through the ICD-10 code G20 from the linked hospital and death records. Researchers used Cox proportional hazards models with age as the timescale and hazard ratios (HR) with 95% confidence intervals (CI).

Covariate Controls and Reverse-Causation–Focused Analyses

Models adjusted for season, sex, ethnicity, deprivation, region, education, employment, smoking, alcohol, and coffee intake, with further analyses including BMI, depression, type 2 diabetes, constipation, bladder dysfunction, and accelerometer-derived sleep duration. To examine reverse causation, Lexis expansion separated follow-up time into 0–2, 2–4, 4–6, and ≥6 years after measurement. Sensitivity checks included re-estimation of steps using an algorithm incorporating Michael J. Fox Foundation PD gait data.

Overall Association of Step Counts With Parkinson’s Incidence

Among 94,696 participants (mean age ~62, 56% female), the average median daily steps were 9,446. Over a median follow-up of 7.9 years, 407 PD cases occurred. Across the full follow-up, higher steps were strongly inversely associated with PD. Compared with the lowest step quintile (<6,276 steps/day), the highest (≥12,369 steps/day) had a 59% lower hazard (HR 0.41, 95% CI 0.31–0.54). Every 1,000 additional steps/day corresponded to an 8% reduction in risk (HR 0.92, 95% CI 0.89–0.94), with linear associations and consistency across demographic and health subgroups.

Time-Window Analyses Show Declining Associations Over Longer Follow-Up

The strongest inverse association appeared in the first two years after step assessment (HR 0.83, 95% CI 0.70–0.90), a window marked by fewer cases but clear gradients. However, as the follow-up extended, effect sizes shrank toward the null. By ≥6 years, associations were no longer statistically significant (HR 0.96, 95% CI 0.92–1.01), suggesting that lower steps mainly reflect early PD-related decline rather than long-term causal protection. Results remained stable when excluding neurological disease or using PD-trained step algorithms.

Interpretation of Findings: Prodromal Activity Decline vs. Causal Protection

These findings align with evidence that activity decreases years before PD diagnosis due to subtle gait slowing, fear of falls, or early non-motor symptoms. They contrast with some self-report studies, in which associations persisted even after removing early follow-up, highlighting the interpretive value of objective wearables. In practice, a drop in step counts (e.g., ~10,000 to ~6,000) could eventually contribute to prodromal-risk profiling when combined with other biomarkers, but should not be used as a standalone clinical indicator.

Strengths, Limitations, and Public Health Relevance

Wearable-derived step counts are simple, scalable, device-agnostic, and easily understood, making them well-suited for integration into early-detection frameworks. Limitations include incomplete primary care PD capture, residual confounding, and limited late-window PD cases that may blunt precision. Nevertheless, the approach advances digital phenotyping for PD detection.

Core Conclusion: Steps Predict PD Mainly Through Early Disease Signals

Accelerometer-measured steps predicted incident PD when all follow-up was combined, but the association faded beyond six years, consistent with reverse causation. Lower steps appear to be early markers of emerging PD rather than true causal risk factors. Future research should extend follow-up, diversify cohorts (including PPMI and All of Us), and test gait-specific digital markers to improve early detection and risk stratification.