Measuring step count: why it is important not to assume measures are reliable

Abstract

I read the recent article with great interest entitled “Measuring step count: why it is important not to assume measures are reliable” by Minns Lowe et al. [[1]]. The authors of the study purposed to demonstrate the validity and reliability of pedometer software Apps versus the previously investigated spring lever and piezoelectric pedometers. In order to count the step, two different smartphone applications (Walk Star and Accupedo) on two different devices (iPhone 5 portable media player and Samsung S3 mini) were preferred. In addition, a spring lever and a piezoelectric pedometer were also analyzed. While I believe the publication provides essential contribution to the literature, there are some methodological concerns that I would like to address which may affect the results of the study.

First, the authors of the study stated that they included 18 volunteer participants due to the sample size of the previous similar studies and depending on available resources and time. This sample size is suitable for some statistical analyzes, and it has revealed precise results in line with the study purposes. However, a statistical study concluded that the sample size should not be <33 for the Bland–Altman agreement analysis, even at high standardized agreement limits (δ/σ) [[2]]. Also, there are no reference values regarding the acceptable agreement value limit for measurement with a wearable technology device for step count. In a previous similar study conducted with 40 healthy participants, accelerometer-based measuring portable devices were found to be reliable (e.g. The Fitbit Surge for inter-device and test-retest reliability ICC 3,1 = 0.76–0.90; Garmin Vivosmart HR + for Bland-Altman plot LoA (steps) for Walk = −56 to 46) in the analysis performed with similar statistical methods [[3]]. Detailed information on power analysis will further prove the results of this study.

Second, Minns Lowe et al. stated that one of the aims of the study was to check the validity and reliability of two pedometer software applications. However, both the sensitivity of the accelerometer and the integrated assistive functioning pressure sensor were analyzed together with the software [[4]]. Therefore, using the same device for both Walk Star and Accupedo and comparing the results would be more effective for the purpose of demonstrating software application's reliability and validity. I would welcome the comments of the authors to address these issues, which will further validate the findings of the study.

Ethical approval: n/a.Conflict of interest: Nothing to disclose.