Tibial acceleration measures have received much attention in relation to impacts during running and risk for tibial stress fracture. Initial work identified higher peak axial tibial impact acceleration in runners with a history of tibial stress fracture compared with controls (1). Since then, studies have sought to understand the effects of running speed (2,3), fatigue (4–6), running surface (7–9), stride length (10), footwear (11,12), and gait retraining (13–15) on tibial acceleration. Most of these studies have used a traditional overground laboratory gait analysis of runners. However, several more recent studies have measured tibial acceleration outside the laboratory using various types of accelerometer. In particular, fatigue during a run (4,16), running speed (2), and running surface (2,8,9) have been explored during such field testing outside the laboratory. In some cases, field-based studies of tibial acceleration refer to tibial shock magnitudes from laboratory studies. However, it cannot be assumed that the magnitude of tibial shock in a runner is comparable between the laboratory and the field.
Initial explorations of tibial acceleration in relation to injury in runners focused on the peak positive tibial axial acceleration (1). Higher peak axial acceleration was found in runners with a history of tibial stress fracture compared with control runners with no history of bony injury (1). It should be noted that the field-based studies (4,16) reported peak axial acceleration in the control conditions ranging from 6.1g to 24.6g, markedly higher than the 5.8g to 11.3g reported in laboratory studies (1,5). However, it is unclear whether these differences in peak axial acceleration magnitude are due directly to the field versus laboratory environment. Thus, there is a need to determine whether peak axial acceleration differs between the laboratory and field-testing conditions.
With advances in accelerometer technology, triaxial accelerometers have become widely available and are now used more frequently in studies of tibial acceleration during running. In addition, the resultant tibial acceleration is becoming more common as an outcome measure. This is likely a result of its increased consistency due to being insensitive to alignment of the accelerometer with the tibial axis. In particular, peak resultant acceleration has been reported in a field-based study of downhill running (17) and a laboratory study of gait retraining (18) in runners. Thus, there is a need to determine whether peak resultant acceleration differs between laboratory and field environments.
There is widespread interest in tibial impact acceleration in the running injury and gait retraining literature. However, most previous studies reported peak axial tibial acceleration magnitude during laboratory gait analysis, which may differ from the magnitude during field-based running. Furthermore, peak resultant tibial acceleration is also being reported, but its relationship to peak axial acceleration has not been determined. Therefore, the purpose of this study was to determine whether laboratory and field measures of tibial acceleration are comparable, and whether peak axial and peak resultant tibial acceleration are interchangeable. We hypothesized that laboratory and field measures of tibial acceleration would differ and that axial and resultant tibial acceleration in these conditions would also differ from each other. A secondary aim was to determine whether peak resultant acceleration predicted peak axial acceleration.
This study was approved by the institution’s review board before commencing the study. Nineteen healthy rearfoot striking runners (10 female, 9 male) between 18 and 45 yr of age provided written informed consent and participated in this study (age, 31 ± 6 yr; height, 1.70 ± 0.08 m; weight, 68.6 ± 11.6 kg). Inclusion criteria were as follows: a history of running at least 10 miles per week for a year or more, being healthy enough for physical activity according to the Physical Activity Readiness Questionnaire (19), and currently free of injury. Participants who reported a history of major lower extremity injury or surgery were excluded. Sample size for a one-factor repeated-measures ANOVA with four levels was estimated using G*Power (20). A Cohen’s d effect size of 0.74 was used to determine the number of participants needed to detect differences that are large enough to affect risk of injury (1). With an α of 0.05 and 80% power, a sample size of 12 was indicated. Thus, our sample of 19 was adequate to power the study.
After enrollment into the study, participants were screened for footstrike pattern as the final inclusion criterion. Participants wore athletic shorts and their own running footwear. Anatomical markers were attached over the medial and lateral malleoli, and head of the first and fifth metatarsals. The markers on the fifth metatarsal heads were also tracking markers. Additional tracking markers were placed on the superior calcaneus, medial calcaneus, and lateral calcaneus. Markers were also placed on the shoe tip, above the tip of the second toe, and on the posterior shoe heel for the calculation of strike index. A standing calibration trial was recorded by an eight-camera motion capture system (Vicon T40S, Oxford, United Kingdom) while participants stood in a standard position (21). Anatomical markers were then removed. Participants ran at 3.0 m·s−1 ± 5% for five good trials. Running velocity was monitored by two photocells place 3 m apart and connected to a timer (Brower, Draper, UT). Three-dimensional marker position data were recorded by the camera system at 200 Hz, and ground reaction forces were recorded by a synchronized force platform at 1000 Hz (AMTI, Inc., Watertown, MA). Foot strike pattern was determined using Visual3D software (C-Motion, Germantown, MD) according to established norms (22). Only those participants with a rearfoot strike pattern, indicated by a strike index of 33% or less, participated in the study. We enrolled and screened 27 runners to find 19 rearfoot strikers.
The participant’s height and weight were then measured using a stadiometer and scale. The accelerometer (Model 356A45; PCB Piezotronics, Depew, NY) was firmly attached to a thermoplastic base. It was then vertically aligned on the right distal tibia and attached just proximal to the curve of the distal tibia via double-sided skin safe tape. The area was tightly overwrapped with self-adherent wrap. The accelerometer was placed by the same investigator (J.L.H.) for all participants. Data were collected in the laboratory first. The accelerometer was hard-wired to the motion capture system via a long cord extending from the waist of the participant. Participants completed 10 good trials at 3.0 m·s−1 ± 5% making contact with the force platform with the right foot, to replicate a typical gait analysis. For the conditions outside the laboratory, the accelerometer was connected via a short cord to the portable data logging unit worn by the participant. In addition, a global positioning system (GPS) watch with heart rate monitor (Garmin 735XT, Olathe, KS) was worn by the participant to record velocity and cadence during field test conditions. For the treadmill running condition, participants completed a self-selected warm-up and then ran for 1 min at 3.0 m·s−1 on the treadmill while data were recorded. Participants then completed a brief self-selected cool-down on the treadmill. The order of the two field test conditions was randomized. All data were collected during dry conditions at 3.0 m·s−1 ± 5%, with velocity monitored via the GPS watch. Participants completed one practice trial before each field test condition. Two trials of at least a minute each were recorded during running on concrete sidewalk and on grass. Participants were free to rest at any time between trials as needed.
Data from the accelerometer and the GPS watch were downloaded to a computer for analysis. Accelerometer data were processed using custom MATLAB code (MathWorks, Natick, MA). Raw data were low pass filtered at 70 Hz using a fourth-order recursive Butterworth filter. Residual analysis was used to determine the filter cutoff frequency (23). Data were normalized to acceleration due to gravity and are reported as multiples of gravity (g). All variables of interest were determined for the first 40% of each stride. Foot contacts were determined via a custom algorithm (24). The primary variables peak positive axial and peak resultant acceleration were extracted for statistical analysis. In addition, peak medial, peak lateral, peak anterior, and peak posterior acceleration, angles of inclination of the peak resultant vector, and percent stride to peaks were extracted to aid the interpretation of the primary findings. For all variables, the mean of 10 trials of data in the laboratory and the mean of 10 consecutive steps from midtrial for the other conditions were calculated and used for further analysis.
Descriptive statistics were calculated for the variables of interest in each condition. Data were checked for normality using the Kolmogorov–Smirnov test. The primary variables were normally distributed. Therefore, repeated-measures ANOVA was used to determine significant differences among conditions for peak positive axial acceleration and peak resultant acceleration. Post hoc analyses were performed using least significant difference tests. Differences among conditions were also compared with minimal detectable differences (MDD). The MDD values for laboratory (peak positive axial acceleration, 0.6g; peak resultant acceleration, 2.0g) and treadmill (peak positive axial acceleration, 1.3g; peak resultant acceleration, 1.6g) conditions were reported previously (25). MDD values for sidewalk and grass were determined by comparing the mean of 10 trials from each of the first and second trials using intraclass correlation coefficient (3,10). The MDD was calculated as follows: standard error of measurement × 1.96/√2. The MDD values for peak positive axial acceleration were 3.7g for the grass condition and 2.5g for the sidewalk condition. For peak resultant acceleration, the MDD values were 4.3g for the grass condition and 3.5g for the sidewalk condition. Lastly, the data were reviewed and met assumptions for regression analyses. Regression analyses determined whether laboratory values of peak positive axial acceleration and peak resultant acceleration could be used to predict the magnitude during other conditions, and whether adding cadence to the analysis improved the prediction. In addition, Bland–Altman plots (26) were used to explore the relationship between peak positive and peak axial acceleration within each condition.
Peak positive axial acceleration differed among conditions (F(1.94,34.86) = 21.49, P < 0.001; Table 1; Fig. 1; see Figure, Supplemental Digital Content 1, dot density plot of peak axial tibial acceleration across conditions, https://links.lww.com/MSS/B885). Magnitudes were similar in the laboratory and treadmill conditions (mean difference, 0.2g; P = 0.595) and in the grass and sidewalk conditions (mean difference, 0.5g; P = 0.470). Peak positive axial acceleration was significantly higher in the grass (mean difference, 4.1g; P < 0.001) and sidewalk conditions (mean difference, 4.6g; P < 0.001) compared with the laboratory condition. It was also significantly higher in the grass (mean difference, 3.9g; P = 0.001) and sidewalk conditions (mean difference, 4.3g; P < 0.001) compared with the treadmill condition. These differences exceeded the MDD for both grass and sidewalk conditions. Overall, peak positive axial acceleration was higher in the field test conditions, grass and sidewalk, than in the indoor conditions, laboratory and treadmill.
Similarly, peak resultant acceleration differed among conditions (F(3,54) = 27.03, P < 0.001; Fig. 2; see Figure, Supplemental Digital Content 2, dot density plot of peak resultant tibial acceleration across conditions, https://links.lww.com/MSS/B886). Magnitudes were similar in the laboratory and treadmill conditions (mean difference, 0.5g; P = 0.404) and in the grass and sidewalk conditions (mean difference, 1.5g; P = 0.060). Peak resultant acceleration in the grass condition (mean difference, 5.3g; P < 0.001) and the sidewalk condition (mean difference, 6.9g; P < 0.001) were both significantly higher than the laboratory condition. It was also significantly higher in the grass (mean difference, 4.8g; P < 0.001) and sidewalk (mean difference, 6.3g; P < 0.001) conditions compared with the treadmill condition. These differences also exceeded the MDD for grass and sidewalk, respectively. Overall, peak resultant acceleration was also higher in the field test conditions than in the indoor conditions.
Peak resultant acceleration was consistently higher than peak axial acceleration within each condition (bias: laboratory, 2.1g; treadmill, 2.4g; grass, 3.3g; sidewalk, 4.4g). The 95% limits of agreement for peak resultant compared with peak axial acceleration were −1.5g to 5.6g for laboratory, −2.0g to 6.8g for treadmill, −1.3g to 7.8g for grass, and 0.8g to 8.0g for sidewalk. However, the timing of the peaks (% stride) was similar. The peak resultant acceleration vector was posterior to the vertical axis of the tibia in the sagittal plane for all conditions. In addition, the frontal plane vector of the resultant acceleration was medial to the vertical axis of the tibia in all conditions. By definition, the peak axial acceleration vector has a zero inclination from the vertical axis of the tibia. Finally, accelerations in the secondary axes, which contribute to the resultant acceleration, were quite large.
Laboratory measures of peak positive axial and peak resultant acceleration were able to predict the magnitude of treadmill and sidewalk conditions to varying degrees, but not the grass condition (Table 2). Peak positive axial and resultant accelerations in the laboratory explained 50% and 38% of the variance in magnitudes, respectively, for the treadmill condition. The addition of treadmill cadence as a predictor did not improve the variance explained. Laboratory measures of peak positive axial and peak resultant acceleration did not predict magnitudes in the grass condition, even with the addition of grass cadence. The laboratory peak positive axial and resultant acceleration were significant predictors of the sidewalk acceleration magnitudes, explaining 48% and 45% of the variance, respectively. Adding sidewalk cadence as a predictor provided a marginal improvement. Overall, laboratory acceleration magnitudes explained about half of the variance of treadmill and sidewalk accelerations, but did not explain the variance for grass.
The purpose of this study was to determine whether laboratory and field measures of peak tibial accelerations are comparable, and whether peak axial and peak resultant tibial acceleration are interchangeable. Although laboratory and treadmill measures of tibial acceleration were similar, both were lower than magnitudes in the field test conditions. However, the field test conditions, grass and sidewalk, were similar to each other. Therefore, the magnitude of tibial impact acceleration in runners is different during indoor compared with field testing. In addition, peak axial acceleration was consistently lower than peak resultant acceleration within the same condition.
Peak axial acceleration was lower while running in the laboratory than during field tests on grass or sidewalk. The magnitude was comparable with previous studies in the laboratory (5.8g–7.7g), grass (11.1g–17.1g), and concrete sidewalk conditions (9.4g–12.4g) (1,2,8,27). However, our treadmill magnitude was lower than reported previously (11.6g–15.3g) (8,16). Peak axial acceleration measured in the laboratory has been associated with tibial stress fracture both retrospectively (1) and prospectively (28). Given the consistently higher magnitudes recorded during field conditions, thresholds for increased likelihood of injury developed from laboratory data are inappropriate for field data. Thresholds for increased association with injury during running in the field are likely much higher than laboratory measures and should be determined in future studies.
Furthermore, laboratory measures of peak axial acceleration account for, at best, only half of the variance in field measures, including during treadmill running. This indicates that additional factors contribute to the magnitude of peak axial acceleration during running outside the laboratory. The addition of running cadence did not provide much improvement in the variance explained. A major difference between traditional laboratory gait analysis and field testing is the very short running distances in the laboratory. To test the importance of this difference, a post hoc regression analysis was conducted to determine whether peak axial acceleration measured during treadmill running was a better predictor of grass and sidewalk peak axial acceleration than the laboratory measure. Peak axial acceleration during treadmill running was no better at predicting grass (R2 = 0.22, P = 0.044) or sidewalk (R2 = 0.49, P = 0.001) peak axial acceleration than the laboratory measure. Further work is needed to determine additional factors that influence the magnitude of tibial acceleration during running.
The magnitude of peak resultant acceleration followed a similar pattern to peak axial acceleration. With the advent of easily available triaxial accelerometers, peak resultant acceleration is increasingly being reported as an outcome measure. There are some advantages to reporting the resultant acceleration, such as it being insensitive to accelerometer alignment with respect to the tibia (29). Although this reduces negative effects of poorly placed accelerometers, it is important to note that peak resultant acceleration was consistently higher than peak axial acceleration within each condition. The difference is explained by the magnitudes of secondary axis acceleration, which may be as large as the axial acceleration. All three axes are combined into the resultant acceleration, so it will always be larger than the axial acceleration during impact. The variation in the secondary axes also explains the wide limits of agreement between the two variables. The differing orientations of the two acceleration vectors and the magnitude of acceleration in the secondary axes account for the different magnitudes during the same step.
Because these variables are not equivalent, they cannot be used interchangeably. Furthermore, interpreting field measures of peak resultant tibial acceleration based on magnitudes of peak axial acceleration measured in the laboratory is strongly advised against. Such interpretations would likely result in many participants appearing to have excessively high tibial acceleration values. In addition, given the low regression coefficients, the patterns may not be present if the same participants were measured in laboratory conditions. New studies to determine associations of peak resultant tibial acceleration magnitude with different groups of runners and running conditions are needed.
As is the nature of field-based studies, ambient weather conditions were varied for data collections on different days. We were unable to determine any effects of weather condition on outcome measures with this study design. However, data were not collected when rain had fallen within the previous 24 h, when the ground was wet, when “feels like” temperatures were greater than 100°F or less than 40°F, or when heat or cold weather advisories were active. In addition, it should be noted that there are many environmental differences between laboratory and field testing conditions. We were unable to determine which specific differences account for the differences in magnitude of tibial acceleration techniques with this study design.
In summary, laboratory and treadmill measures of both peak axial and peak resultant tibial acceleration were lower than grass and sidewalk measures. Peak axial acceleration was consistently lower than peak resultant acceleration with wide limits of agreement. Furthermore, laboratory measures accounted for no more than half of the variance in the magnitude of the variable of interest in other conditions. Thus, tibial impact acceleration magnitude is influenced by test conditions in runners. These findings support measuring tibial impact acceleration in the field to determine new metrics associated with injury.
This study was supported by a Drexel University College of Nursing and Health Professions Seed Grant. The results of this study do not constitute endorsement by the American College of Sports Medicine. The authors declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.
The authors have no conflicts of interest to disclose.
1. Milner CE, Ferber R, Pollard CD, Hamill J, Davis IS. Biomechanical factors associated with tibial stress fracture in female runners. Med Sci Sports Exerc
2. Boey H, Aeles J, Schutte K, Vanwanseele B. The effect of three surface conditions, speed and running experience on vertical acceleration of the tibia during running. Sports Biomech
3. Dufek JS, Mercer JA, Griffin JR. The effects of speed and surface compliance on shock attenuation characteristics for male and female runners. J Appl Biomech
4. Giandolini M, Gimenez P, Temesi J, et al. Effect of the fatigue induced by a 110-km ultramarathon on tibial impact acceleration and lower leg kinematics. PLoS One
5. Clansey AC, Hanlon M, Wallace ES, Lake MJ. Effects of fatigue on running mechanics associated with tibial stress fracture risk. Med Sci Sports Exerc
6. Abt JP, Sell TC, Chu Y, Lovalekar M, Burdett RG, Lephart SM. Running kinematics and shock absorption do not change after brief exhaustive running. J Strength Cond Res
7. Montgomery G, Abt G, Dobson C, Smith T, Ditroilo M. Tibial impacts and muscle activation during walking, jogging and running when performed overground, and on motorised and non-motorised treadmills. Gait Posture
8. Fu W, Fang Y, Liu DMS, Wang L, Ren S, Liu Y. Surface effects on in-shoe plantar pressure and tibial impact during running. J Sport Health Sci
9. Greenhalgh A, Sinclair J, Leat A, Chockalingam N. Influence of footwear choice, velocity and surfaces on tibial accelerations experienced by field hockey participants during running. Footwear Sci
10. Derrick TR. The effects of knee contact angle on impact forces and accelerations. Med Sci Sports Exerc
11. Sinclair J, Greenhalgh A, Brooks D, Edmundson CJ, Hobbs SJ. The influence of barefoot and barefoot-inspired footwear on the kinetics and kinematics of running in comparison to conventional running shoes. Footwear Sci
12. Sinclair J, Fau-Goodwin J, Richards J, Shore H. The influence of minimalist and maximalist footwear on the kinetics and kinematics of running. Footwear Sci
13. Clansey AC, Hanlon M, Wallace ES, Nevill A, Lake MJ. Influence of tibial shock feedback training on impact loading and running economy. Med Sci Sports Exerc
14. Creaby MW, Franettovich Smith MM. Retraining running gait to reduce tibial loads with clinician or accelerometry guided feedback. J Sci Med Sport
15. Crowell HP, Davis IS. Gait retraining to reduce lower extremity loading in runners. Clin Biomech (Bristol, Avon)
16. Garcia-Perez JA, Perez-Soriano P, Llana Belloch S, Lucas-Cuevas AG, Sanchez-Zuriaga D. Effects of treadmill running and fatigue on impact acceleration in distance running. Sports Biomech
17. Giandolini M, Horvais N, Rossi J, Millet GY, Samozino P, Morin JB. Foot strike pattern differently affects the axial and transverse components of shock acceleration and attenuation in downhill trail running. J Biomech
18. Wood CM, Kipp K. Use of audio biofeedback to reduce tibial impact accelerations during running. J Biomech
19. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sport Sci
20. Faul F, Erdfelder E, Lang AG, Buchner A. G*power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods
21. McIlroy WE, Maki BE. Preferred placement of the feet during quiet stance: development of a standardized foot placement for balance testing. Clin Biomech (Bristol, Avon)
22. Cavanagh PR, Lafortune MA. Ground reaction forces in distance running. J Biomech
23. Winter DA. Biomechanics and Motor Control of Human Movement
. 4th ed. Hoboken (NJ): John Wiley & Sons, Inc.; 2009. 384 p.
24. Aubol K, Hawkins JL, Milner CE. Foot contact identification using a tibial mounted accelerometer during running. Med Sci Sports Exerc
. 2019;51(6 Suppl):775.
25. Milner CE, Hawkins JL, Aubol K. Reliability and minimum detectable difference of tibial acceleration during running. Med Sci Sports Exerc
. 2019;51(6 Suppl):690–1.
26. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet
27. Kim W, Voloshin AS. Dynamic loading during running on various surfaces. Hum Mov Sci
28. Davis I, Milner CE, Hamill J. Does increased loading during running lead to tibial stress fractures? A prospective study. Med Sci Sports Exerc
. 2004;36(5 Suppl):S585.
29. Sheerin KR, Besier TF, Reid D, Hume PA. The one-week and six-month reliability and variability of three-dimensional tibial acceleration in runners. Sports Biomech