Objective: To examine the accuracy of measuring step counts using a pedometer in participants with neurological conditions and healthy volunteers in relation to a manual step count tally.
Setting: Oxford Centre for Enablement, Nuffield Orthopaedic Centre NHS Trust, Oxford, UK.
Subjects: Healthy adults (n = 13, age: mean 29, SD = 12) and adults with neurological conditions (n=20 stroke, n=16 multiple sclerosis, n=5 muscular dystrophy, n=1 spinal cord injury, n=1 traumatic brain injury; age: mean 54, SD=13).
Main measures: Individuals walked for 2 minutes at self-selected walking speeds (SSWS) wearing a pedometer. Healthy individuals were then asked to walk at slow walking speeds (SWS). Step counts were recorded manually and using a pedometer.
Results: In healthy individuals there was no difference between manually measured and pedometer counts during walking (P>0.05). In adults with neurological conditions the pedometers undercounted (P = 0.003); bias (random error): 27 (111); percentage variability 30% and intraclass correlation coefficient (ICC) 0.66. In neurological adults, from regression analysis the relationship between error and walking speed was cubic, with walking speed accounting for 29% of pedometer error. Healthy individuals showed greater variability and undercounting at SWS bias (random error): 10 (31), percentage variability 8% and ICC 0.73, compared with SSWS bias (random error): -3 (13), percentage variability 3% and ICC 0.84.
Conclusions: Pedometers may undercount when used for people with neurological conditions. There may be variability in pedometer accuracy but this was not strongly related to walking speed. The suitability of pedometer use for exercise monitoring should be individually determined.