Osaka study of 20,950 app users: predicted healthcare costs level off near 11,000 daily steps

Associate Professor Haruka Kato of the Graduate School of Human Life and Ecology at Osaka Metropolitan University analyzed how objectively measured daily steps relate to later healthcare expenditures, ResearchSEA reports. The study covered 20,950 long-term app users aged 40 and over, 338,706 person-months in total from 2020 to 2023. It was published in SSM - Population Health on 3 August 2026 asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.

Data and method

Step counts came from "A-Smile," the official mobile health app of Osaka Prefecture, and were linked to administrative health insurance claims asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗. The material notes that earlier studies mostly relied on self-reported questionnaires and linear models, so the actual shape of the relationship and its change across the lifespan remained largely hidden asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.

The analysis used generalized additive models, a statistical technique that estimates flexible curves rather than assuming straight lines asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗. The result is a surface of predicted 6-month healthcare expenditures in JPY across age and daily steps. The surface describes associations, not causal effects.

Results by age

Among adults aged 40–64 the predicted-cost surface was nearly flat. Costs in this group appeared to be driven by factors other than walking asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗. Among those aged 65 and over, predicted costs peaked in the low-step region and fell steeply as step counts increased. The gradient linking steps to costs therefore steepens with age asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.

The point where the decline ends was similar in both groups. Predicted costs stopped declining and leveled off at around 11,000 steps a day: at 11,317 steps for middle-aged adults and 10,612 steps for older adults asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.

Kato said that predicted costs are far more sensitive to daily steps in older adults, and that promoting walking in this group could contribute to lower healthcare expenditures. She also noted that the model treats costs as a flexible function of age and steps. This makes it possible to estimate the expected change in expenditures for any given increase in steps at any given age asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.

Kato has released an interactive version of the fitted model. It displays the predicted expenditure and its confidence interval for any combination of age and daily steps within the range of the study asiaresearchnews.comResearchSEAasiaresearchnews.comOpen source ↗.