Real-life step and activity measurement: reliability and validity

J Med Eng Technol. 2009;33(1):33-41. doi: 10.1080/03091900701682606.


Objectives: Investigation of the reliability and validity of activity monitoring using a range of methods, namely manual counting, self report and two commercially available activity monitors.

Study design: Reliability, accuracy and validity study.

Methods: Multiple convenience samples of healthy subjects were recruited to the study. Reliability of manual step count was determined using an intraclass correlation coefficient (ICC) (n = 10). Relationships between data from the Step Watch monitor (SAM) and (a) manual step counts (n = 18); (b) a second (different) activity monitor (ActivPAL); and (c) self reported activity levels (n = 22) were assessed using correlations. A Pearson's correlation and paired t-test was used to assess relations between routinely used monitors.

Results: Intra-rater reliability for manual step counts was excellent (ICC 0.99), but inter-rater reliability was poor (ICC 0.26). Indoor accuracy of the SAM was 96.06% and outdoor accuracy was 99.58%. Moderate correlations (rho = 0.423 to 0.595, p < 0.05) were identified between the SAM monitor activity levels and self report diaries. The SAM and the ActivPAL were found to be internally reliable within themselves (ICC 0.96 and 0.95 respectively), significantly correlated (r = 0.93, p < 0.001) but also significantly different (t = 2.179, p < 0.05) when used simultaneously over the same circuit.

Conclusions: Activity monitors provide information that is related to actual activity and provide accurate and reliable data when tested on functional walking circuits. Activity monitors should not be used interchangeably due to the potential for systematic differences between the measurements obtained when applied simultaneously over the same repeated circuit.

MeSH terms

  • Activities of Daily Living*
  • Adult
  • Equipment Design
  • Humans
  • Monitoring, Ambulatory / methods*
  • Reproducibility of Results
  • Statistics, Nonparametric
  • Walking