Association between COVID-19 emergency declarations and physical activity among community-dwelling older adults enrolled in a physical activity measurement program: Evidence from a retrospective observational study using the regression discontinuity design

Abstract Background The current study examines the negative impact of the coronavirus disease 2019 (COVID-19) emergency declarations on physical activity among the community-dwelling older adults, the participants of a physical activity measurement program, in Japan. Methods This retrospective observational study included 1,773 community-dwelling older adults (aged 74.6 ± 6.3 years, 53.9% women) who had participated in the physical activity measurement project from February 2020 to July 2021. We measured physical activity using a tri-axial accelerometer during 547 consecutive days. Three emergency declarations, requesting people to avoid going outside, occurred during the observational period. We multiply-imputed missing values for daily physical activity, such as steps, light physical activity (LPA), and moderate-to-vigorous physical activity (MVPA) for several patterns of datasets according to the maximum missing rates on a person level. We mainly report the results based on less than 50% of the maximum missing rate (n = 1,056). Other results are reported in the supplemental file. Changes in physical activity before and after the start of each emergency declaration were examined by the regression discontinuity design (RDD) within 14-, 28-, and 56-day bandwidths. Results For all the participants in the multiply-imputed data with the 14-day bandwidth, steps (coefficients [ $${\widehat{\gamma }}_{1}$$ γ ^ 1 ] $$,$$ , 964.3 steps), LPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , 5.5 min), and MVPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , 4.9 min) increased after the first emergency declaration. However, the effects were attenuated as the RDD bandwidths were widened. No consistent negative impact was observed after the second and third declarations. After the second declaration, steps ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -609.7 steps), LPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -4.6 min), and MVPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -2.8 min) decreased with the 14-day bandwidth. On the other hand, steps ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , 143.8 steps) and MVPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , 1.3 min) increased with the 56-day bandwidth. For the third declaration, LPA consistently decreased with all the bandwidths ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -2.1, -3.0, -0.8 min for the 14, 28, 56-day bandwidth), whereas steps ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -529 steps) and MVPA ( $${\widehat{\gamma }}_{1},$$ γ ^ 1 , -2.6 min) decreased only with the 28-day bandwidth. Conclusions For the community-dwelling older adults who regularly self-monitor their physical activity, the current study concludes that there is no evidence of consistently negative impacts of the emergency declarations by the COVID-19 pandemic.