There has been limited longitudinal assessment of the relationship between moderate-to-vigorous physical activity (MVPA) and sedentary behaviour (SB) with frailty, and no studies have explored the possibility of reverse causality. This study aimed to determine the potential bidirectionality of the relationship between accelerometer-assessed MVPA, SB, and frailty over time in older adults.

Participants were from the Toledo Study for Healthy Aging. We analysed 186 older people aged 67 to 90 (76.7 ± 3.9; 52.7% female participants) over a 4-year period. Time spent in SB and MVPA was assessed by accelerometry. Frailty Trait Scale was used to determine frailty levels. A cross-lagged panel model design was used to test the reciprocal relationships between MVPA/SB and frailty.

Frailty Trait Scale score changed from 35.4 to 43.8 points between the two times (P < 0.05). We also found a reduction of 7 min/day in the time spent on MVPA (P < 0.05), and participants tended to spend more time on SB (P = 0.076). Our analyses revealed that lower levels of initial MVPA predicted higher levels of later frailty [std. β = -0.126; confidence interval (CI) = -0.231, -0.021; P < 0.05], whereas initial spent time on SB did not predict later frailty (std. β = -0.049; CI = -0.185, 0.087; P = 0.48). Conversely, an initial increased frailty status predicted higher levels of later SB (std. β = 0.167; CI = 0.026, 0.307; P < 0.05) but not those of MVPA (std. β = 0.071; CI = -0.033, 0.175; P = 0.18).