Accuracy of a smartphone pedometer application according to different speeds and mobile phone locations in a laboratory context

Background The purpose of this study was to compare the accuracy of a smartphone application and a mechanical pedometer for step counting at different walking speeds and mobile phone locations in a laboratory context. Methods Seventeen adults wore an iPphone6© with Runtastic Pedometer© application (RUN), at 3 different locations (belt, arm, jacket) and a pedometer (YAM) at the waist. They were asked to walk on an instrumented treadmill (reference) at various speeds (2, 4 and 6 km/h). Results RUN was more accurate than YAM at 2 km/h (p < 0.05) and at 4 km/h (p = 0.03). At 6 km/h the two devices were equally accurate. The precision of YAM increased with speed (p < 0.05), while for RUN, the results were not significant but showed a trend (p = 0.051). Surprisingly, YAM underestimates the number of step by 60.5% at 2 km/h. The best accurate step counting (0.7% mean error) was observed when RUN is attached to the arm and at the highest speed. Conclusions RUN pedometer application could be recommended mainly for walking sessions even for low walking speed. Moreover, our results confirm that the smartphone should be strapped close to the body to discriminate steps from noise by the accelerometers (particularly at low speed).