Evaluating intervention fidelity : an example from a high-intensity interval training study

Taylor, Kathryn L. and Weston, Matthew and Batterham, Alan M. (2015) Evaluating intervention fidelity : an example from a high-intensity interval training study. PLoS ONE, 10 (4). e0125166. ISSN 1932-6203 (https://doi.org/10.1371/journal.pone.0125166)

[thumbnail of Taylor-etal-PloSO-2015-Evaluating-intervention-fidelity]
Text. Filename: Taylor_etal_PloSO_2015_Evaluating_intervention_fidelity.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo

Download (190kB)| Preview


Aim: Intervention fidelity refers to the degree to which an experimental manipulation has been implemented as intended, but simple, robust methods for quantifying fidelity have not been well documented. Therefore, we aim to illustrate a rigorous quantitative evaluation of intervention fidelity, using data collected during a high-intensity interval training intervention. Design: Single-group measurement study. Methods: Seventeen adolescents (mean age ± standard deviation [SD] 14.0 ± 0.3 years) attended a 10-week high-intensity interval training intervention, comprising two exercise sessions per week. Sessions consisted of 4-7 45-s maximal effort repetitions, interspersed with 90-s rest. We collected heart rate data at 5-s intervals and recorded the peak heart rate for each repetition. The high-intensity exercise criterion was ≥90% of individual maximal heart rate. For each participant, we calculated the proportion of total exercise repetitions exceeding this threshold. A linear mixed model was applied to properly separate the variability in peak heart rate between- and within-subjects. Results are presented both as intention to treat (including missed sessions) and per protocol (only participants with 100% attendance; n=8). Results: For intention to treat, the median (interquartile range) proportion of repetitions meeting the high-intensity criterion was 58% (42% to 68%). The mean peak heart rate was 85% of maximal, with a between-subject SD of 7.8 (95% confidence interval 5.4 to 11.3) percentage points and a within-subject SD of 15.1 (14.6 to 15.6) percentage points. For the per protocol analysis, the median proportion of high-intensity repetitions was 68% (47% to 86%). The mean peak heart rate was 91% of maximal, with between- and within-subject SDs of 3.1 (-1.3 to 4.6) and 3.4 (3.2 to 3.6) percentage points, respectively. Conclusions: Synthesising information on exercise session attendance and compliance (exercise intensity) quantifies the intervention dose and informs evaluations of treatment fidelity.