Background:​ Dynamic high-intensity physical activity is thought to be beneficial for older adults' bone health. Traditional volume-based processing of accelerometer-measured physical activity data, quantified on a minute-per-minute basis, may average out sporadic high impact activity, whereas accelerometer data processing approaches based on identifying impacts can capture also these potentially beneficial short activity bursts. We investigated the associations between habitual physical activity and proximal femur bone traits among sedentary older adults utilizing three different numerical treatments of accelerometer-data to examine, if impact-based processing approaches are more suitable to assess bone loading than volume-based processing of physical activity data among older adults.

Methods:​ This cross-sectional study utilized the baseline data from the PASSWORD-study (n = 284, mean ± SD age 74 ± 4 years, 57% women). Total femur bone mineral content (BMC) and bone mineral density (BMD), femoral neck BMC, BMD, section modulus and minimal width (MNW) were measured with dual energy x-ray absorptiometry. Physical activity was measured for seven consecutive days with a tri-axial accelerometer. Raw acceleration data was processed in three different ways and quantified as i) mean daily minutes in sedentary, light and moderate-to-vigorous-intensity activity, ii) mean daily number of acceleration peaks divided into low (1.5 g to 2.0 g), medium (2.0 g to 2.5 g) and high (>2.5 g) impacts, and iii) mean daily osteogenic index, which is a summary score calculated from log-transformed number of impact peaks in 32 intensity bands (≥1.3 g). Associations between physical activity measures and each bone trait were estimated with multiple linear regression adjusted with covariates (age, sex, weight, height, smoking, physical function, medication).

Results:​ Participants recorded on average 10 h sedentary, 2.5 h light and 33 min moderate-to-vigorous activity, and 3937 low, 494 medium and 157 high impacts per day. Mean osteogenic index score was 173. Light physical activity was positively associated with all bone traits (beta = 0.147 to 0.182, p < 0.001 to p = 0.005) except MNW. Sedentary or moderate-to-vigorous activity, low, medium or high impacts or osteogenic index were not associated with any bone parameter.

Conclusions:​ Light physical activity may decelerate the age-related bone loss in older adults who do not meet the physical activity recommendations. In this population, the amount of high impact activity may be insufficient to stimulate bone remodelling.