Open AccessValidity of six consumer-level activity monitors for measuring steps in patients with chronic heart failure
Author(s) -
Tomáš Větrovský,
Michal Širanec,
Jitka Marenčáková,
James J. Tufano,
Václav Čapek,
Václav Bunc,
Jan Bělohlávek
Publication year - 2019
Publication title -
plos one
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.99
H-Index - 332
ISSN - 1932-6203
DOI - 10.1371/journal.pone.0222569
Subject(s) - activity tracker , treadmill , medicine , statistics , mean absolute percentage error , concordance , heart failure , physical activity , activity monitor , physical therapy , mathematics , mean squared error
Although numerous activity trackers have been validated in healthy populations, validation is lacking in chronic heart failure patients who normally walk at a slower pace, making it difficult for researchers and clinicians to implement activity monitors during physical activity interventions. Methods Six consumer-level activity monitors were validated in a 3-day field study in patients with chronic heart failure and healthy individuals under free living conditions. Furthermore, the same devices were evaluated in a lab-based study during treadmill walking at speeds of 2.4, 3.0, 3.6, and 4.2 km·h -1 . Concordance correlation coefficients (CCC) were used to evaluate the agreement between the activity monitors and the criterion, and mean absolute percentage errors (MAPE) were calculated to assess differences between each device and the criterion (MAPE <10% was considered as a threshold for validity). Results In the field study of healthy individuals, all but one of the activity monitors showed a substantial correlation (CCC ≥0.95) with the criterion device and MAPE <10%. In patients with heart failure, the correlation of only two activity monitors (Garmin vívofit 3 and Withings Go) was classified as at least moderate (CCC ≥0.90) and none of the devices had MAPE <10%. In the lab-based study at speeds 4.2 and 3.6 km·h -1 , all activity monitors showed substantial to almost perfect correlations (CCC ≥0.95) with the criterion and MAPE in the range 1%-3%. However, at slower speeds of 3.0 and 2.4 km·h -1 , the accuracy of all devices substantially deteriorated: their correlation with the criterion decreased below 90% and their MAPE increased to 4–8% and 10–45%, respectively. Conclusions Even though none of the tested activity monitors fall within arbitrary thresholds for validity, most of them perform reasonably well enough to be useful tools that clinicians can use to simply motivate chronic heart failure patients to walk more.