SEARCH

SEARCH BY CITATION

Keywords:

  • Actigraph;
  • Agreement;
  • Measurement error;
  • Paediatrics

Abstract

Aim: To compare Actigraph-defined moderate-to-vigorous physical activity (MVPA) cutpoints among children, combining statistical and biobehavioural analyses.

Methods: One hundred and thirteen children aged 10.0 ± 0.8 years wore accelerometer for three days. The time they spent in MVPA was estimated using 10 thresholds ranged from 3000 to 3900 cpm. A statistical construct including 45 Bland and Altman pairwise analyses was used to compare the 10 estimates of MVPA. A regression was performed to develop an equation relating mean differences to the between-cutpoint gaps.

Results: Mean differences in the MVPA estimates ranged from 1.6 to 12.8 min as a function of increment. Raw estimates of MVPA decreased according to an arithmetic sequence with a common difference of 200 cpm. This difference translates into a drop of 12% in MVPA and a misclassification of up to 5% of children. Mean differences (Y) could be predicted from increments (X) using: Y= 0.02 X (R2= 0.99, SEE = 0.72, p < 0.0001).

Conclusion: When a lack of agreement should be assumed as the between-cutpoint gap exceeds 200 cpm, statistical differences may occur earlier at 90 cpm. Yet, the current equation makes it possible to compare and adjust results from studies/interventions using diverse cutpoints for MVPA among children.