Treadmill running has historically been utilized in the laboratory to mimic outdoor running. Recent developments in portable technology, such as Inertial Measurement Units (IMUs) allow researchers to assess runners in their natural environment.

PURPOSE:​ The primary purpose of the study was to compare peak tibial acceleration, stance time, stride frequency, and rearfoot eversion velocity between treadmill and overground running (asphalt, track, and grass). The secondary purpose was to test the reliability of IMU-based estimations of maximum rearfoot eversion velocity during treadmill running.

METHODS:​ Twenty subjects (Age:​ 22.1 ± 2.0 yrs, Mass:​ 70.8 ± 11.9 kg, Height:​ 174.5 ± 10.0 cm, 8F/12M) participated. After consent and a warm-up period, IMUs were placed on the anteromedial aspect of the right distal tibia (1600Hz) and the posterior heel cap of the right shoe (1125 Hz) to record 3D linear accelerations and angular velocities. Subjects then ran three 30-meter trials on each overground surface (grass, track, and asphalt) at their self-selected speed. A timing system was used to ensure the same running speed was used between conditions. Subsequently, all subjects ran on a treadmill while their rearfoot motion was recorded through high-speed videography (240 Hz). Only one continuous trial was performed for each subject on the treadmill. In each of these trials, the variables of stride frequency, tibial acceleration, maximum rearfoot eversion velocity, and stance time were collected from the IMUs. Maximum rearfoot eversion velocity was the only variable processed from the rearfoot motion video data. A total of 30 steps from each condition were extracted and analyzed (10 steps from each trial for overground surfaces). In our statistical analysis, separate repeated measures ANOVAs or Friedman tests were performed on the mean and variability of each variable, depending on data normality, to examine differences between surface conditions. Post-hoc analysis was performed when appropriate through either Fisher’s LSD (α = 0.05) or Wilcoxon signed-rank tests (α = 0.008).

RESULTS:​ The means of stride frequency, peak tibial acceleration, and maximum rearfoot eversion velocity were significantly different (p < 0.001) between surfaces. Specifically, stride frequency was the fastest during treadmill running (1.39 (0.09) strides/sec) and slowest while running on grass (1.35 (0.07) strides/sec). Peak tibial acceleration was not different between the outdoor running conditions (asphalt:​ 11.0 (2.7) G, grass:​ 10.7 (3.4) G, and track:​ 10.6 (2.3) G), but significantly less during treadmill running (7.9 (1.6) G). Maximum rearfoot eversion velocity was lowest on grass (394.9 (256.5) deg/s) and greatest on track (623.5 (299.3) deg/s) and asphalt (620.7 (289.1) deg/sec). There was no difference in stance time between surfaces (p = 0.231). The variabilities of stance time, stride frequency, maximum rearfoot eversion velocity, and peak tibial acceleration were all found to be significantly different (p < 0.001) between running surface. Treadmill running presented with the lowest levels of variability in stride frequency (0.016 (0.005) strides/sec), maximum rearfoot eversion velocity (70.10 (35.58) deg/s), and peak tibial acceleration (0.79 (0.32) G) across all conditions. Due to this, all variables other than stance time were significantly less variable during treadmill running than any of the overground conditions. In contrast, running on grass displayed significantly larger variabilities across all conditions in stance time (0.014 (0.009) sec), stride frequency (0.029 (0.010) strides/sec), maximum rearfoot eversion velocity (123.71 (36.16) deg/s), and peak tibial acceleration (2.41 (1.21) G). Lastly, maximum rearfoot eversion velocities acquired from the heel mounted IMU obtained a moderate reliability with the high-speed videography (ICC = 0.739, CI:​ 0.322 – 0.899).

CONCLUSION:​ A single IMU appears to only be moderately reliable when recording eversion velocities during running, so caution may be warranted when using a single IMU to estimate rearfoot motion during running. With the use of IMUs, treadmill running has been shown to be different than overground in terms of both mean differences and variability. As a result, laboratory-based studies on running biomechanics may not truly reflect the “natural” gait used on outdoor running surfaces.