Self-Report and Contemporaneously Recorded Running Agreement in Recreational Athletes

Leech RD, Grunseit A, and Edwards KL. Self-report and contemporaneously recorded running agreement in recreational athletes. J Strength Cond Res 36(7): 1922-1929, 2022-Accurate assessment of running exposure in research or clinical practice relies on the ability to measure the behavior validly and reliably. Self-reported physical activity (PA) (including running), although commonly used, is subject to bias. User-owned wearable technological devices provide a potential contemporaneously collected data source for validating retrospective running-specific questionnaires. This study assesses agreement between self-reported running and contemporaneously collected running data. Self-reported running and corresponding contemporaneously recorded for current (at the time of questionnaire completion) and historical (6 months prior) was collected. Concordance Correlations and Bland-Altman Limits of Agreement measured extent of agreement. Categories for Metabolic Equivalent Task (MET) hours per week were calculated from total running distance. Unweighted and weighted Kappas were used to compare classification of levels of running by the 2 data sources. Running data were collected from 139 subjects (mean age 51.1 years). Self-report data were higher than contemporaneous data. Average estimated pace from both sources were similar, with high variance in some individuals. Differences of 43.5-57.1% were reported for current mean weekly running distances for 7-day, 4-week, and 12-week running periods. The level of agreement in classification between self-report and contemporaneous running data MET/HR categories was fair (unweighted K = 0.24-0.33) to moderate (weighted K = 0.46-0.63). Compared with contemporaneous running data, runners over-estimate the frequency, distance, and duration of running in self-report. Runners tend to discount nonrunning periods and unusual runs when estimating current and historical "average" running. The opportunities and limitations of different contemporaneous running data collection methods as a PA comparator must be acknowledged.