Original Communication
European Journal of Clinical Nutrition (2005) 59, 561–570. doi:10.1038/sj.ejcn.1602118 Published online 16 February 2005
Reliability and validity of the combined heart rate and movement sensor Actiheart
Guarantors: S Brage and NJ Wareham.
Contributors: SB and NJW designed the study and SB was responsible for data collection. SB and NB conducted the data analysis, the results of which were discussed and interpreted by all authors. SB drafted the manuscript, which was reviewed by all authors.
S Brage1, N Brage2, P W Franks1, U Ekelund1 and N J Wareham1
- 1MRC Epidemiology Unit, Institute of Public Health, University of Cambridge, UK
- 2Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense, Denmark
Correspondence: S Brage, MRC Epidemiology Unit , Institute of Public Health, University of Cambridge, Elsie Widdowson Laboratory, Fulbourn Road, Cambridge CB1 9NL, UK. E-mail: soren.brage@mrc-epid.cam.ac.uk
Received 11 February 2004; Revised 27 August 2004; Accepted 30 November 2004; Published online 16 February 2005.
Abstract
Accurate quantification of physical activity energy expenditure is a key part of the effort to understand disorders of energy metabolism. The Actiheart, a combined heart rate (HR) and movement sensor, is designed to assess physical activity in populations.
Objective:
To examine aspects of Actiheart reliability and validity in mechanical settings and during walking and running.
Methods:
In eight Actiheart units, technical reliability (coefficients of variation, CV) and validity for movement were assessed with sinusoid accelerations (0.1–20 m/s2) and for HR by simulated R-wave impulses (25–250 bpm). Agreement between Actiheart and ECG was determined during rest and treadmill locomotion (3.2–12.1 km/h). Walking and running intensity (in J/min/kg) was assessed with indirect calorimetry in 11 men and nine women (26–50 y, 20–29 kg/m2) and modelled from movement, HR, and movement+HR by multiple linear regression, adjusting for sex.
Results:
Median intrainstrument CV was 0.5 and 0.03% for movement and HR, respectively. Corresponding interinstrument CV values were 5.7 and 0.03% with some evidence of heteroscedasticity for movement. The linear relationship between movement and acceleration was strong (R2=0.99, P<0.001). Simulated R-waves were detected within 1 bpm from 30 to 250 bpm. The 95% limits of agreement between Actiheart and ECG were -4.2 to 4.3 bpm. Correlations with intensity were generally high (R2>0.84, P<0.001) but significantly highest when combining HR and movement (SEE<1 MET).
Conclusions:
The Actiheart is technically reliable and valid. Walking and running intensity may be estimated accurately but further studies are needed to assess validity in other activities and during free-living.
Sponsorship:
The study received financial support from the Wellcome Trust and SB was supported by a scholarship from Unilever, UK.
Keywords:
physical activity, heart rate variability, accelerometer, movement, walking, running, epidemiology, energy expenditure monitoring
