BIOMECHANICAL DIFFERENCES OF FOOT STRIKE PATTERNS DURING RUNNING: A SYSTEMATIC REVIEW WITH META-ANALYSIS

Transcription

1 1 BIOMECHANICAL DIFFERENCES OF FOOT STRIKE PATTERNS DURING RUNNING: A SYSTEMATIC REVIEW WITH META-ANALYSIS Matheus O. Almeida, PT, PhD 1 Irene S. Davis, PT, PhD 2 Alexandre D. Lopes, PT, PhD 1 1 Masters and Doctoral Program in Physiotherapy, Universidade Cidade de São Paulo (UNICID), SP, Brasil 2 Department of Physical Medicine and Rehabilitation, Harvard Medical School, MA, United States The authors certify that they have no affiliations with or financial involvement in any organization or entity with a direct financial interest in the subject matter or materials discussed in the article. Address correspondence to Matheus Oliveira de Almeida, Universidade Cidade de São Paulo, Rua Cesário Galeno 448, Tatuapé, São Paulo SP, CEP , Brazil, mathewsalmeida@hotmail.com.

2 2

3 ABSTRACT Study design: Systematic review with meta-analysis. Objective: To determine the biomechanical differences between foot strike patterns using when running. Background: Strike patterns during running has received attention in the recent literature due to the mechanical differences and associated injury risks between them. Methods: Electronic databases (Medline, Embase, Lilacs, Scielo, and SPORTDiscus) were searched through July Studies (cross-sectional, case control, prospective, and retrospective) comparing biomechanical characteristics between foot strike patterns during running of distance runners with at least 18 years of age were included in this review. Two independent reviewers evaluated the risk of bias. A meta-analysis with a random-effects model was used to combine the data from the included studies. Results: Sixteen studies were included in the final analysis. In the metaanalyses of kinematic variables, significant differences between forefoot and rearfoot strikers were found for foot and knee angle at initial contact and knee flexion range of motion. A forefoot strike pattern resulted in a plantar flexed ankle position and a more flexed knee position, compared to a dorsiflexed ankle position and a more extended knee position for the rearfoot strikers, at initial contact with ground. In the comparison of rearfoot and midfoot strikers, midfoot strikers demonstrated greater ankle dorsiflexion range of motion and decreased knee flexion range of motion compared to rearfoot strikers.. For kinetic variables, the meta-analysis revealed that rearfoot strikers had higher vertical loading rates compared to forefoot strikers.

4 Conclusion: There are differences in kinematic and kinetic characteristics between foot strike patterns when running. Clinicians should be aware of these characteristics to help in the management of running injuries and advice on training. Keywords: biomechanics, jogging, landing, runners

5 INTRODUCTION Running is one of the most popular types of physical activity in the world, with more than 30 million people running regularly in United States. 1 In Europe, it is estimated that 36% of the population are recreational runners. 4 Despite being considered a simple activity, running involves the complex integration of movements in all joints and body segments. Unfortunately, running injuries are common with rates ranging between 18 and 92%. 42, 43 This large variation may be explained in part due to the lack of consensus on the definition of a running injury. 46, 47 The most frequent running-related injuries are medial tibial stress syndrome, achilles tendinopathy, plantar fasciitis, and patellar tendinopathy. 24 Special attention has been given to the foot strike pattern using during running because it has been suggested that higher magnitude and rates of change of vertical impact forces transmitted to the lower limbs during running may contribute to running-related injuries. 6, 23, 28, 33, 36, 48 Adopting different foot strike patterns has been shown to modify the characteristics of these vertical impact forces. However, investigation of the relationship between foot strike patterns and running related injuries is still scarce in the literature and only 8, 13 based on retrospective studies. These studies found significantly higher rates of musculoskeletal injuries in rearfoot strikers compared to the midfoot and forefoot strikers. Three primary foot strike patterns for running have been described: rearfoot, midfoot, and forefoot. During rearfoot striking, initial contact with the ground occurs at the heel or posterior part of the foot. striking is used to describe a striking pattern in which the posterior and anterior portions of the foot simultaneously contact the ground, and forefoot striking is characterized by a

6 pattern in which the anterior region of the foot strikes the ground first. 6, 23 It has been reported that up to 89% of runners are rearfoot strikers. 16, 18, 21 This may be attributed to the use of running shoes with increased thickness and cushioning on the posterior part of the sole, making landing on the heel comfortable. running has been adopted by many coaches and individuals as the correct way to run in a recreational setting, 16 although there is no evidence to support or refute this argument. Foot strike pattern modification, as part of gait retraining of runners, is plausible and has shown early promising results as a potential intervention for injured runners. 38 In a case-series, runners with anterior knee pain were trained to run with a forefoot strike pattern using an instrumented insole. As a result, runners experienced a reduction of pain and improvement of function which persisted at a 3 month follow-up. 7 In another study, runners with anterior compartment syndrome were also instructed to adopt a forefoot strike pattern. After 6 weeks of training, there was a reduction in post-running anterior compartmental pressure, and improvement of pain and disability. 9 Because foot strike pattern during running can be modified, potentially with the goal to prevent or treat injuries, understanding the biomechanical differences between patterns would be helpful in recommending that runners conform to a pattern that attempts to minimize specific biomechanical loads. There are specific biomechanical measures, due their possible relationship with running related injuries, that should be considered when comparing different foot strike patterns. Ground reaction force is an important kinetic variable because it is an approximate measure of the loading of the lower-extremity musculoskeletal system, 48 and increased vertical loading rates

7 may be associated with an increased risk of tibial stress fractures. 48 In addition, foot and ankle kinematics, such as peak rearfoot eversion and rearfoot eversion at foot strike, which have beeninvestigated as possible risk factors for anterior knee pain 11, 26 and medial tibial stress syndrome, 27, 34, 35 may also be altered with changes in foot strike patterns. A number of factors, such as the type of shoes worn, 23, 40 need to be considered when comparing footstrike patterns. Running with versus without shoes as well as the familiarity with the foot strike pattern, 45 will likely influence biomechanical characteristics. Therefore, an analysis of the biomechanical differences between foot strike patterns should control for these confounding factors. Despite the recent increase in interest in the scientific and clinical communities, no systematic review investigating biomechanical differences between foot strike patterns in running have been conducted. Synthesis of the literature through the use of rigorous methods and a comprehensive and up-todate search may help clinicians, physical therapists, and the running community to make informed decisions on which pattern is more appropriate for various distance runners. Therefore, the purpose of this systematic review and metaanalysis is to determine the biomechanical differences (spatio-temporal, kinematic, and kinetic variables) between footstrike patterns (rearfoot, midfoot, and forefoot patterns) in distance runners. METHODS Identification and selection of studies

8 Medline via PubMed, Embase, Lilacs, Scielo, and SPORTDiscus databases were searched through July 31, There were no restrictions on publication dates or article languages. In an attempt to improve search results, the search terms were adapted and explored for each database (APPENDIX A). Two independent reviewers (M.O.A., A.D.L.) initially accessed and selected potential studies for inclusion based on titles and abstract evaluation. Disagreements were solved by judgment from a third reviewer (I.S.D.). Full texts of selected articles were then collected and evaluated in the same manner. Study inclusion was limited to cross-sectional, case control, prospective, and retrospective analyses. Clinical trials were not included because the objective of this review was not to evaluate the effect of any particular intervention. All studies included in the review had to compare biomechanical characteristics between foot strike patterns during running. Articles were only included if the participants were distance runners at least 18 years of age. Assessment of the characteristics of the studies Risk of bias assessment Two independent evaluators (M.O.A, A.D.L.) assessed the risk of bias for all included articles. In instances where there was no consensus on risk of bias, a third reviewer (I.S.D.) was used to resolve the disagreement. Because a scale to assess risk of bias does not exist for evaluating different study designs and biomechanical studies, a modified version of the Downs and Black Quality Index was used. 10 The scale used in this study was used in a previous systematic 132 review (APPENDIX B). 14 As the original scale also evaluates clinical trials, the

9 items relative to this study design were disregarded. The scale was composed of 20 items related to information reporting (items 1 to 9), external validity (items 10 and 11), internal validity (items 12 to 15), and selection bias (items 16 to 20). Each item was scored 0 to represent high risk of bias, or 1 to represent low risk of bias, or No. Studies that scored from 0 to 6 were classified as High risk of bias ; from 7 to 13 as Moderate risk of bias ; and from 14 to 20 as Low risk of bias. Data extraction Data extraction of the selected articles was performed by one investigator (M.O.A.). The study design, sample size, participant characteristics, natural foot strike pattern, evaluation methodology, and outcome variables were recorded for each study. Authors of the included studies were contacted to request missing data when necessary. Data were expressed as mean difference (MD) with 95% confidence interval (CI). The biomechanical variables from each study were then classified into 1 of 3 categories: spatial-temporal, kinematic, or kinetic. Data analysis For all included biomechanical variables, a meta-analysis with a randomeffects model was planned when data from at least 3 studies could be combined. Data synthesis from shod conditions were planned to be performed separately from barefoot conditions. To avoid a potential bias, it was decided to not pool data from studies that evaluated participants using their natural foot strike pattern with studies in which participants were evaluated using a foot

10 strike pattern for which they had to adjust and become familiar with. For example, data from a study that evaluated natural rearfoot strikers using either a forefoot or midfoot strike pattern were not combined with data from a study that evaluated natural rearfoot strikers and natural forefoot strikers. A randomeffects model was chosen because of methodological heterogeneity between studies, which by themselves could generate substantial statistical heterogeneity. RESULTS Flow of studies through the review The search results identified a total of 2110 citations, of which 226 were duplicates. After application of the inclusion criteria and the selection process, 16 articles were included in this systematic review. The flow diagram of the full selection process and the inclusion of studies is presented in FIGURE 1. Characteristics of included articles Risk of bias assessment The average score for risk of bias assessment was 12.4 (0 20 scale). Thirteen studies were classified as having moderate risk of bias and 3 as low risk of bias. Most common items that resulted in a higher risk of bias were related to lack of reporting adverse events; participants not representative of the running population; lack of examiners blinding; inadequate adjustment for confounding in the analysis, and lack of reporting sample size calculation (TABLE 1).

11 Description of studies Characteristics of the studies are shown in TABLE 2. All studies had a cross-sectional design, providing data on a total of 1260 runners. Reported mean age across studies, when reported, ranged between 18 and 45 years. Most studies were conducted in laboratory settings running overground, with only 2 studies evaluating runners on a treadmill. Only 2 studies evaluated the participants in both shod and barefoot conditions. From the 16 included articles, 12 compared rearfoot to forefoot strike pattern, 3 compared rearfoot to midfoot strike pattern and only 1 study compared the 3 foot strike patterns. Eleven articles investigated kinetic variables, 11 explored kinematic variables, and 5 evaluated spatial-temporal variables. Six studies compared natural rearfoot strikers with unnatural forefoot strikers, and 1 study compared natural rearfoot strikers with unnatural midfoot strikers. The remaining studies compared natural rearfoot strikers with natural forefoot or midfoot strikers. Based on the criteria stated in the data analysis section, only 5 studies 6, 13, 19, 30, 45 that evaluated participants in shod condition and using their natural strike pattern could be included in the meta-analysis. Spatial-temporal variables It was not possible to conduct meta-analyses for any of the spatiotemporal variables, but some important findings were observed. 204 Shih et al 39 evaluated cadence among natural rearfoot strikers and unnatural forefoot strikers during running and did not find statistically significant difference (MD -2.4 steps/min, 95% CI 10.6, 5.8; P =.57). Kulmala et al, 19 who compared natural rearfoot strikers and natural forefoot strikers in shod

12 condition, similarly did not find statistically significant differences in cadence (MD steps/min, 95% CI -22.8, 0.8; P =.07). Ardigo et al 2 found no statistically significant difference, when running in shod condition, when comparing step length (MD 0.1 meters, 95% CI 0.0, 0.2; P =.05) between natural rearfoot strikers (mean of 1.0 meter, SD = 0.1) and unnatural forefoot strikers (mean of 0.9 meters, SD = 0.1), and for cadence (MD -3.0 steps/min, 95% CI -12.6, 6.6; P =.54) between natural rearfoot strikers (mean steps/min, SD = 0.1) and unnatural forefoot strikers (mean of steps/min, SD = 0.2). Arendse et al 3 did not find difference in stride length when comparing natural rearfoot pattern (mean of 1.09 meters, SD = 0.3) to unnatural midfoot strikers in shod condition (mean of 1.09 meters, SD = 0.4). Goss and Gross 12 comparing natural rearfoot strikers (mean of steps/min, SD = 0.1) with natural midfoot strikers (mean of steps/min, SD = 0.2) in shod condition, also did not find a statistically significant difference (MD -5.1 steps/min, 95% CI -11.7, 1.5, P =.13). Kinematic variables Meta-Analysis: versus The kinematic variables that could be combined in a meta-analysis of the comparison of natural rearfoot and forefoot strike patterns were foot angle at initial contact and knee angle at initial contact. Forest plots of these analyses for runners in shod condition are provided in FIGURE 2. A statistically significant difference was found for foot angle with rearfoot strikers contacting the ground in a dorsiflexed position compared to a plantar flexed position for the forefoot

13 strikers. Knee angle at initial contact was also statistically different with greater knee flexion being seen for those using the forefoot strike pattern Data from individual studies Nunns et al 30 and Williams et al 45 evaluated rearfoot eversion range of motion in shod condition, and statistically significant greater values were exhibited by natural forefoot strikers when compared to natural rearfoot strikers (MD -1.6º, 95% CI -2.7, -0.6; P =.001; MD -6.0º, 95% CI -9.9, -2.1; P =.002; respectively). Ankle dorsiflexion range of motion was also statistically greater for the natural forefoot strikers (MD -2.2º, 95% CI -3.3, -1.2; P <.0001; MD -7.9º 30, 45 95% CI -12.9, -2.9; P =.002; respectively). Nunns et al 30 found that natural rearfoot strikers demonstrated statistically greater knee flexion excursion values compared to natural forefoot strikers when running in shod condition (MD 3.6º, 95% CI 2.9, 4.4; P <.00001). Similarly, Kulmala et al, 20 who evaluated peak knee flexion angle, found greater values for the natural rearfoot strikers compared to natural forefoot strikers running in shod condition (MD 4.0º, 95% CI 1.5, 6.5; P =.001). Details of all kinematic variables in shod conditions are presented in TABLE 3, including the variables that could not be pooled in a meta-analysis. While, due to lack of sufficient data, we did not conduct a meta-analysis for any kinematic variable for running in barefoot conditions, an important 254 finding was noted. Shih et al, 39 studying natural rearfoot strikers, found a significant difference for foot angle at initial contact (MD of 13.4º, 95% CI 10.6, 16.2; P <.0001) between when using their natural rearfoot strike pattern

14 (dorsiflexion position) and an unnatural forefoot strike pattern (plantar flexion position) Meta-Analysis: versus It was not possible to conduct a meta-analysis for any of the kinematic variables comparing rearfoot midfoot strike patterns. Arendse et al 3 compared natural rearfoot strikers with unnatural midfoot strikers during shod running and found significant difference in foot angle at initial contact with the use of the rearfoot strike pattern leading to a dorsiflexed position at landing (MD of 24.7, CI 21.5, 27.9; P <.0001). Nunns et al 30 also evaluated the foot angle at initial contact in shod condition, but between natural rearfoot strikers and natural midfoot strikers and found significant differences (MD of 6.2º, CI 5.6, 6.8, P <.0001) with a more dorsiflexion position in rearfoot strikers. When looking at the amount of ankle dorsiflexion range of motion, Nunns et al 30 reported statistically significant greater values for natural midfoot strikers compared to natural rearfoot strikers (MD -2.4º, 95% CI -3.1, -1.5; P <.0001). Nunns et al 30 and Goss et al 13 found statistically lower values of knee flexion range of motion in natural midfoot strikers compared to natural rearfoot strikers (MD 2.1º, 95% CI 1.4, 2.9; P <.0001; MD 4.7º, 95% CI 1.2, 8.3; P =.009; respectively). Details of all kinematic variables of rearfoot and midfoot pattern comparison in shod conditions are presented in TABLE 4. Kinetic variables Meta-Analysis: versus

15 Separate meta-analyses, comparing natural rearfoot and forefoot strikers in shod condition, could be performed for the following 3 variables: vertical loading 283 rate (VLR), 2 nd peak of vertical ground reaction force (VGRF2), and ankle plantar flexion moment, with forest plots provided in FIGURE 3. Natural rearfoot strikers had statistically higher VLR compared to natural forefoot strikers. There were no statistically significant differences for VGRF2 and ankle plantar flexion moment between the groups. Data from individual studies Kulmala et al 20 evaluated the first peak of vertical ground reaction force (VGRF1) and found greater values for natural rearfoot strikers compared to natural forefoot strikers when running in shod condition (MD of 0.7 BW, 95% CI 0.5, 0.9; P <.0001). Details of all kinetic variables for rearfoot and forefoot comparison in shod conditions are presented in TABLE 5. When looking at barefoot running, Lieberman et al 23 found greater VGRF1 values in natural rearfoot strikers compared to natural forefoot strikers (MD of 1.3 BW, 95% CI 0.8, 1.8; P <.0001). This is consistent with the data from Oakley and Pratt 31 who reported 1.5 times higher VGRF1 values in rearfoot strikers. Despite the similarity of the studies, the data could not be combined in a meta-analysis, because the mean and standard deviation values for both groups were not reported in this latter study and the data could not be obtained despite contacting the authors. 304 Shih et al 39 compared VLR between natural rearfoot strikers and 305 unnatural forefoot strikers running in a barefoot condition. Significant greater

16 values were found with the rearfoot strike pattern (MD of 17.5 BW/s, 95% CI 11.6, 23.3; P <.0001) Meta-Analysis: versus In the comparison of natural rearfoot and midfoot patterns, only the data for the variable VGRF2 could be pooled in a meta-analysis (FIGURE 4). There was no statistically significant difference between foot strike patterns when running in shod condition. Data of individual studies Cavanagh et al 6 reported no discernible VGRF1 in natural midfoot strikers, consistent with this finding, Arendse et al 3 reported significantly higher VGRF1 in natural rearfoot strikers compared to unnatural midfoot strikers (MD of 0.3 BW, 95% CI 0.2, 0.5; P <.0001). Details of all kinetic variables for rearfoot and midfoot comparisons when running in shod conditions are presented in TABLE 6. DISCUSSION Results from the 16 studies were used for the systematic review (1260 runners aged between 18 and 45 years). These studies demonstrated differences in kinematic and kinetic variables between strike patterns. No differences in spatiotemporal variables were noted. Of these 16 articles, the data from 5 studies could be pooled for meta-analysis for some of the kinematics and kinetics variables.

17 The meta-analysis revealed that in the shod condition natural rearfoot strikers made initial contact with the ground with the foot in a dorsiflexed position while natural forefoot strikers landed in a plantar flexed position. These findings were expected because it partially define the foot strike positions. Kinematic differences were also seen for the frontal plane with 2 studies reporting greater rearfoot eversion excursion with natural forefoot strikers compared to natural rearfoot striker when running in shod condition. Greater rearfoot eversion excursion could result in abnormal loading of the foot and lower extremity. However, there is only limited evidence in the literature that 5, 25, 29, 35 rearfoot eversion is a risk factor for running-related injuries. Natural forefoot strikers, running in shod condition, also have a greater knee flexion angle at initial contact compared to natural rearfoot strikers. This is likely due to the shorter stride length associated with this strike pattern. Shorter stride lengths have recently been associated with reduced loads to the hip and knee thereby potentially reducing injury risk to these areas. 17 Ground reaction forces, in particular excessive vertical load rates, have been linked to running injury. 48 The meta-analysis revealed that natural rearfoot strikers exhibited significantly higher VLRs compared to natural forefoot strikers when running in a shod condition. This difference may be attributed to the dampening effect of the triceps surae immediately after initial contact. With higher load rates, rearfoot strikers may be more susceptible to injuries such as tibial stress fractures 48 and plantar fasciitis. 33 The meta-analysis revealed no statistically significant difference between natural rearfoot and midfoot strikers running in shod condition for the VGRF2, which represents the force generated by muscular activity during the propulsion

18 phase of gait. The VGRF1, the impact transient, did not meet the criteria to be included in the meta analysis. However, when looking at 2 individual studies, VGRF1 was significantly higher in natural rearfoot strikers compared to unnatural midfoot strikers in shod condition 3 as well as natural forefoot strikers in barefoot condition. 23 It is also notable that the VGRF1, also known as the transient impact, was found to be absent in one study looking at shod natural midfoot strikers. 6 The transient impact is characterized by a high magnitude abrupt force that is transmitted to the lower extremities when initial contact with the ground is made with the heel, during which little energy is dissipated. 23 It is sometimes present in midfoot strikers, but seldom in natural forefoot strikers. runners, who contact the ground in a plantar flexed position, transform the vertical forces into rotational kinetic energy through eccentric control of the triceps surae. This mechanism may reduce the transient impact and subsequently reduce the axial force transmitted through the lower extremities during running. 23 In contrast, higher eccentric activity of the calf musculature is required for forefoot strikers to control the speed of ankle dorsiflexion after foot contact with the ground, as demonstrated by the greater negative work values at the ankle in forefoot strikers. 45 Consequently, forefoot strikers may be more susceptible to develop Achilles tendinopathy and calf muscle strains if their soft tissues are not adequately prepared for the load imposed. When assessing the risk of bias across studies, no study was classified as high risk. All studies had methodological features that limited the strength of the reported evidence. For example, in many studies, the extent to which the participants represented the normal running population was unknown, because

19 important information about their running routine, such as previous injury history, volume of running, and running experience were not reported. It should also be noted that none of the studies reported sample size calculations, potentially reducing the statistical power of the available data. Despite the findings of the meta-analysis in this review, the available evidence is not robust. One limitation of this review was that we found a large variation in methodology and outcome variables, highlighting the lack of methodological standardization in this field of study. For example, some included studies evaluated participants with their natural foot strike pattern, where others forced the participants to adopt specific foot strike patterns, unnatural pattern. We chose not to combine data from natural patterns with unnatural patterns in the meta-analysis, thereby limiting the number of included studies. Another limitation was that running speed was not controlled in the analysis. Given the effect running speed has on biomechanical variables, the differences found in this review may not be generalized due to strike pattern across running speeds. We propose that future work should consider standardization of evaluation and metrics, which will enhance our ability to draw meaningful conclusions from the literature. Knowledge of the biomechanical characteristics of foot strike patterns is essential for professionals who work with runners, including athletic trainers, coaches, and physical therapists. While the current analysis did not assess the effect of running patterns on injury risk, examining biomechanical differences is an important first step in characterizing the mechanical effects of different movement strategies. The results found in this systematic review indicate some biomechanical characteristics of rearfoot strike pattern that may be associated

20 with an increased risk of some specific types of injury in distance runners. But, while forefoot and midfoot strikers appear not be exposed to the high loading rates experienced by rearfoot strikers, they must produce high eccentric activity of their calf musculature, potentially leading to other types of injury. Clinicians should be aware of these characteristics to guide the runners about which foot strike pattern maybe more appropriate based on their physical characteristics and training goals. CONCLUSION There is a limited amount of studies in the literature that document significant kinematic and kinetic differences between foot strike patterns when running. KEY POINTS Findings: There are biomechanical differences between foot strike patterns when running, primarily at the moment of initial foot contact. In contrast to rearfoot striker, those using a forefoot strike pattern make contact with the ground with the ankle in a more plantar flexed position and more knee flexion. In addition, rearfoot strikers demonstrated a higher vertical loading rate at initial contact with the ground. Implications: Clinicians should be aware of foot strike patterns characteristics (pros and cons) and differences so they can provide the best care to their patients and guide the runners about which foot strike pattern is more appropriate.

21 Caution: There is a lack of standardization of the methods and the reported variables in the literature, making pooling of data difficult

22 REFERENCES 1. USA R State of the Sport Part II: Running Industry Report. 2013; 2. Ardigo LP, Lafortuna C, Minetti AE, Mognoni P, Saibene F. Metabolic and mechanical aspects of foot landing type, forefoot and rearfoot strike, in human running. Acta Physiol Scand. 1995;155: Arendse RE, Noakes TD, Azevedo LB, Romanov N, Schwellnus MP, Fletcher G. Reduced eccentric loading of the knee with the pose running method. Med Sci Sports Exerc. 2004;36: /01.MSS B0 [doi] 4. Asics. Reasons to run. Available at: Accessed June 24, 2014, Barton CJ, Levinger P, Menz HB, Webster KE. Kinematic gait characteristics associated with patellofemoral pain syndrome: a systematic review. Gait Post. 2009;30: /j.gaitpost Cavanagh PR, Lafortune MA. Ground reaction forces in distance running. J Biomech. 1980;13: (80) [pii] 7. Cheung RT, Davis IS. Landing pattern modification to improve patellofemoral pain in runners: a case series. J Orthop Sports Phys Ther. 2011;41: /jospt Daoud AI, Geissler GJ, Wang F, Saretsky J, Daoud YA, Lieberman DE. Foot strike and injury rates in endurance runners: a retrospective study. Med Sci Sports Exerc. 2012;44: /MSS.0b013e Diebal AR, Gregory R, Alitz C, Gerber JP. running improves pain and disability associated with chronic exertional compartment

23 syndrome. Am J Sports Med. 2012;40: / Downs SH, Black N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and nonrandomised studies of health care interventions. J Epidemiol Community Health. 1998;52: Duffey MJ, Martin DF, Cannon DW, Craven T, Messier SP. Etiologic factors associated with anterior knee pain in distance runners. Med Sci Sports Exerc. 2000;32: Goss DL, Gross MT. A comparison of negative joint work and vertical ground reaction force loading rates in Chi runners and rearfoot-striking runners. J Orthop Sports Phys Ther. 2013;43: /jospt Goss DL, Gross MT. Relationships among self-reported shoe type, footstrike pattern, and injury incidence. US Army Med Dep J. 2012; Hall JP, Barton C, Jones PR, Morrissey D. The Biomechanical Differences Between Barefoot and Shod Distance Running: A Systematic Review and Preliminary Meta-Analysis. Sports Med. 2013; /s Hamill J, Derrick TR, McClay I. Joint stiffness during running with different footfall patterns. Arch Physiol Bioch. 2000;108: Hasegawa H, Yamauchi T, Kraemer WJ. Foot strike patterns of runners at the 15-km point during an elite-level half marathon. J Strength Cond Res. 2007;21: R [pii] /R [doi]

24 Heiderscheit BC, Chumanov ES, Michalski MP, Wille CM, Ryan MB. Effects of step rate manipulation on joint mechanics during running. Med Sci Sports Exerc. 2011;43: /MSS.0b013e3181ebedf4 18. Kasmer ME, Liu XC, Roberts KG, Valadao JM. Foot-strike pattern and performance in a marathon. Int J Sports Physiol Perform. 2013;8: Kulmala J-P, Avela J, Pasanen K, Parkkari J. Strikers Exhibit Lower Running-Induced Knee Loading than Strikers. Med Sci Sports Exer. 2013;45: Kulmala JP, Avela J, Pasanen K, Parkkari J. strikers exhibit lower running-induced knee loading than rearfoot strikers. Med Sci Sports Exerc. 2013;45: /MSS.0b013e31829efcf7 21. Larson P, Higgins E, Kaminski J, et al. Foot strike patterns of recreational and sub-elite runners in a long-distance road race. J Sports Sci. 2011;29: / Laughton CA, McClay Davis I, Hamill J. Effect of strike pattern and orthotic intervention on tibial shock during running. J Applied Biomech. 2003;19: Lieberman DE, Venkadesan M, Werbel WA, et al. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature. 2010;463: nature08723 [pii] /nature08723 [doi] Lopes AD, Hespanhol Junior LC, Yeung SS, Costa LO. What are the main running-related musculoskeletal injuries? A Systematic Review. Sports Med. 2012;42: /

25 Lun V, Meeuwisse WH, Stergiou P, Stefanyshyn D. Relation between running injury and static lower limb alignment in recreational runners. Br J Sports Med. 2004;38: /bjsm Messier SP, Davis SE, Curl WW, Lowery RB, Pack RJ. Etiologic factors associated with patellofemoral pain in runners. Med Sci Sports Exerc. 1991;23: Messier SP, Pittala KA. Etiologic factors associated with selected running injuries. Med Sci Sports Exerc. 1988;20: Milner CE, Ferber R, Pollard CD, Hamill J, Davis IS. Biomechanical factors associated with tibial stress fracture in female runners. Med Sci Sports Exerc. 2006;38: /01.mss Munteanu SE, Barton CJ. Lower limb biomechanics during running in individuals with achilles tendinopathy: a systematic review. J Foot Ankle Res. 2011;4: / Nunns M, House C, Fallowfield J, Allsopp A, Dixon S. Biomechanical characteristics of barefoot footstrike modalities. J Biomech. 2013;46: /j.jbiomech Oakley T, Pratt DJ. Skeletal transients during heel and toe strike running and the effectiveness of some materials in their attenuation. Clin Biomech (Bristol, Avon). 1988;3: Pohl MB, Buckley JG. Changes in foot and shank coupling due to alterations in foot strike pattern during running. Clin Biomech (Bristol, Avon). 2008;23: S (07) [pii] /j.clinbiomech [doi]

26 Pohl MB, Hamill J, Davis IS. Biomechanical and anatomic factors associated with a history of plantar fasciitis in female runners. Clin J Sport Med. 2009;19: /JSM.0b013e3181b8c270 [doi] [pii] Pohl MB, Mullineaux DR, Milner CE, Hamill J, Davis IS. Biomechanical predictors of retrospective tibial stress fractures in runners. J Biomech. 2008;41: /j.jbiomech Raissi GR, Cherati AD, Mansoori KD, Razi MD. The relationship between lower extremity alignment and Medial Tibial Stress Syndrome among non-professional athletes. Sports Med Arthrosc Rehab Ther Technol. 2009;1: / Robbins SE, Hanna AM. Running-related injury prevention through barefoot adaptations. Med Sci Sports Exerc. 1987;19: Rooney BD, Derrick TR. Joint contact loading in forefoot and rearfoot strike patterns during running. J Biomech. 2013;46: /j.jbiomech Samaan C, Rainbow M, Davis I. Reduction in ground reaction force variables with instructed barefoot running. J Sport Health Sci. 2014;3: Shih Y, Lin KL, Shiang TY. Is the foot striking pattern more important than barefoot or shod conditions in running? Gait Post. 2013; /j.gaitpost Squadrone R, Gallozzi C. Biomechanical and physiological comparison of barefoot and two shod conditions in experienced barefoot runners. J Sports Med Phys Fitness. 2009;49:6-13.

27 Stackhouse CL, Davis IM, Hamill J. Orthotic intervention in forefoot and rearfoot strike running patterns. Clin Biomech (Bristol, Avon). 2004;19: S [pii] 42. van Gent RN, Siem D, van Middelkoop M, van Os AG, Bierma-Zeinstra SM, Koes BW. Incidence and determinants of lower extremity running injuries in long distance runners: a systematic review. Br J Sports Med. 2007;41: ; discussion /bjsm Van Middelkoop M, Kolkman J, Van Ochten J, Bierma-Zeinstra SM, Koes B. Prevalence and incidence of lower extremity injuries in male marathon runners. Scand J Med Sci Sports. 2008;18: /j x 44. Williams DS, Green DH, Wurzinger B. Changes in lower extremity movement and power absorption during forefoot striking and barefoot running. Int J Sports Phys Ther. 2012;7: Williams DS, McClay IS, Manal KT. Lower extremity mechanics in runners with a converted forefoot strike pattern.. J App Biomech. 2000;16: Yamato TP, Saragiotto BT, Hespanhol Junior LC, Yeung SS, Lopes AD. Descriptors used to define running-related musculoskeletal injury: a systematic review. J Orthop Sports Phys Ther. 2015;45: /jospt Yamato TP, Saragiotto BT, Lopes AD. A consensus definition of runningrelated injury in recreational runners: a modified delphi approach. J Orthop Sports Phys Ther. 2015;45: /jospt

28 Zadpoor AA, Nikooyan AA. The relationship between lower-extremity stress fractures and the ground reaction force: a systematic review. Clin Biomech (Bristol, Avon). 2011;26: /j.clinbiomech

29 TABLES TABLE 1. Results of risk of bias assessment of included studies 1) Aim clearly described 2) Outcomes described 3) Subjects clearly described 4) Interventions clearly described 5) Distribution of confounders described 6) Main findings clearly described 7) Estimates of random variability in data 8) All important adverse events reported 9) Actual probability values reported 10) Subjects asked representative of population 11) Subjects used representative of population 12) Examiners blinded 13) Data dredging 14) Appropriate statistical tests 15) Valid and reliable main outcome measures 16) Subjects recruited from same population Cavanagh et al Oakley et al Ardigo et al Hamil et al Williams et al Laughton et al Arendse et al Stackhouse et al Pohl & Buckley Lieberman et al Williams et al Shih et al Goss & Gross Kulmala et al Nunns et al 17) Subjects recruited over same time period 18) Intervention order randomised 19) Adequate adjustment for confounding 20) Sufficient power Total (out of 20) Rooney & Derrick = Yes; 0= No. Score: 0 to 6 (high risk of bias); 7 to 13 (moderate risk of bias); 14 to 20 (low risk of bias).

30 Downloaded from at UNIVERSIDADE FEDERAL DE SO PAULO on August 30, For personal use only. No other uses w TABLE 2. Description of characteristics of included studies. Abbreviations: FFS, forefoot striker; MFS, midfoot striker; RFS, rearfoot striker. Study Sample size Participants Characteristics Cavanagh, 1980 Oakley, 1988 Ardigo, 1995 Hamill, 2000 Williams, 2000 Laughton, 2003 Arendse, 2004 Stackhous e, 2004 Pohl, 2008 Lieberman, 2010 Williams, (10 males and 7 females) 18 (10 males and 8 females) 8 (males) Varsity and recreational runner with mean age of 24 years Active runners with mean age of 22 years and free of injury Healthy active students with mean age of 24 years Natural foot strike pattern 12 5 Foot strike comparison Local of evaluation Evaluation condition RFS x MFS Overground Shod Variables Kinematic Spatial-temporal Not reported RFS x MFS Overground Barefoot Kinetic RFS x FFS Treadmill Shod Spatial-temporal 5 (males) Young runners and free of injury Not reported RFS x FFS Overground Shod Kinetic 18 (12 males and 6 females) 15 (genders not reported) 20 (10males and 10 females) 15 (genders not reported) 12 (6 males and 6 females) 16 (13 males and 3 females) 20 (10 males and 10 females) Age between 18 and 45 years 9 9 RFS x FFS Overground Shod Mean age of 22 years RFS x FFS Overground Shod Recreational runners with mean age of 33 years Active runners with age between 18 and 45 years and free of injury RFS x MFS Track Shod Kinetic Kinematic Kinetic Kinematic Kinetic Kinematic Spatialtemporal RFS x FFS Overground Shod Kinematic Active runners with mean age of 21 years and free of injury Not reported RFS x FFS Overground Barefoot Kinematic Age between 18 and 45 years and free of injury in last 6 months Recreational runners with age between 20 and 30 years and free of injury 10 6 RFS x FFS Overground / Track Barefoot / Shod Kinetic RFS x FFS Overground Shod Kinematic 30

31 Downloaded from at UNIVERSIDADE FEDERAL DE SO PAULO on August 30, For personal use only. No other uses w 604 Shih, 2013 Goss, 2013 Kulmala, 2013 Nunns, 2013 Rooney, (males) 34 (18 males, 16 females) 38 females 982 males 30 Mean age of 24 years and free of injury Experienced runners with age between 18 and 45 years and free of injury Athletes with mean age of 19 years and free of injury Active runners with mean age of 21 years and free of injury Competitive runners free of injury RFS x FFS Overground Barefoot / Shod RFS x MFS Treadmill Shod RFS x FFS Overground Shod RFS x MFS x FFS Overground Shod Kinetic Kinematic Kinetic Kinematic Spatial-temporal Kinetic Kinematic Spatial-temporal Kinematic Kinetic RFS x FFS Overground Shod Kinetic 31

32 Downloaded from at UNIVERSIDADE FEDERAL DE SO PAULO on August 30, For personal use only. No other uses w TABLE 3. Description of kinematic variables comparison between rearfoot and forefoot patterns in shod conditions. Variables Foot angle at initial contact ( ) angle at initial contact ( ) eversion peak angle ( ) eversion range ( ) eversion velocity ( /s) Ankle dorsiflexion peak angle ( ) Ankle dorsiflexion range ( ) Foot strike pattern Williams 2000 Laughton 2003 Stackhouse 2004 Williams 2012 Shih 2013 Kulmala 2013 Nunns ± ± (8.0, 19.6)* 11.5 ± ± (-9.9, -2.1)* ± ± (-183.7, -52.7)* 22.0 ± ± (-12.9, -2.9)* ± ± (-5.6, 0.1) ± ± (-15.6, -9.1)* 10.5# 8.8# 13.6# 16.4# 190.9# 270.6# ± ± (-15.6, -9.1)* ± ± (23.3, 31.0)* ± ± (17.8, 23.8)* 24.8 ± ± (17.9, 27.1)* ± ± (10.7, 13.9)* 1.79 ± ± (-0.9, 0.4) 4.07 ± ± (-1.9, -0.3)* 5.89 ± ± (-2.7, -0.6)* ± ± (0.3, 2.5)* ± ± (-3.3, -1.2)*

33 Downloaded from at UNIVERSIDADE FEDERAL DE SO PAULO on August 30, For personal use only. No other uses w Ankle dorsiflexion velocity ( /s) Knee angle at initial contact ( ) Knee flexion peak angle ( ) Knee flexion range ( ) Knee flexion velocity ( /s) Knee internal rotation velocity ( /s) Hip flexion angle at initial contact ( ) ± ± (-168.9, -39.7)* 14.0 ± ± (-12.8, 0.6) 28.9 ± ± (-1.5, 8.9) ± ± (-119.6, -5.2)* ± ± (1.7, 6.6)* ± ± (-155.9, )* ± ± (-2.5, 5.2) ± ± (1.1, 7.2)* ± ± (2.3, 71.9)* Foot angle at initial contact: positive values indicate dorsiflexion; negative values indicate plantar flexion ± ± (-4.3, 4.0) 26.0 ± ± (-6.5, 6.1) ± ± (-13.3, -5.6)* 28.6 ± ± (-1.4, 4.8) Values are mean ± SD; MD (CI): Mean difference (95% confidence interval), positive values indicate RFS > FFS; negative values indicate FFS > RFS 21.0 ± ± (-4.6, 1.4) 50.9 ± ± (1.5, 6.5)* 46.0 ± ± (-0.5, 6.0) ± ± (-4.8, -1.8)* ± ± (-0.8, 2.1) ± ± (2.9, 4.4)* # Standard deviation was not reported (authors were contacted via ); * P <

34 34 TABLE 4. Description of kinematic variables comparison between rearfoot and midfoot patterns in shod conditions. Variables Foot strike pattern Arendse 2004 Goss 2003 Nunns 2013 Foot angle at initial contact ( ) 13.2 ± ± (21.5, 27.9)* 8.92 ± ± (5.6, 6.8)* angle at initial contact ( ) eversion peak angle ( ) eversion range ( ) Ankle dorsfilexion peak angle ( ) Ankle dorsiflexion range ( ) Ankle angle in terminal swing phase ( ) Knee angle at initial contact ( ) Knee flexion peak angle ( ) Knee flexion range ( ) Knee angle in midstance ( ) Knee angle in terminal swing ( ) Shank external rotation range ( ) ± ± (-34.3, )* 27.3 ± ± (-3.5, 3.7) 85.6 ± ± (-2.1, 3.3) 49.4 ± ± (-1.6, 5.2) 23.3 ± ± (-4.2, 3.6) ± ± (-3.9, 0.9) ± ± (1.2, 8.3)* 1.79 ± ± (-1.4, -0.5)* 4.07 ± ± (-1.4, 0.1) 5.89 ± ± (-2.7, -0.8)* ± ± (-0.8, 0.8) ± ± (-3.1, -1.5)* ± ± (-1.9, 0.4) ± ± (0.3, 2.5)* ± ± (1.4, 2.9)* Foot angle at initial contact and ankle angle in terminal swing phase: positive values indicate dorsiflexion; negative values indicate plantar flexion Values are mean ± SD; MD (CI): Mean difference (95% confidence interval), positive values indicate RFS > MFS; negative values indicate MFS > RFS * P <.05

35 Downloaded from at UNIVERSIDADE FEDERAL DE SO PAULO on August 30, For personal use only. No other uses w TABLE 5. Description of kinetic variables comparison between rearfoot and forefoot patterns in shod conditions. Variables VGRF 1 st peak (BW) VGRF 2nd peak (BW) Vertical loading rate (BW/s) Antero-posterior loading rate (BW/s) HGRF peak (BW) Peak knee extension moment (N.m/kg) Ankle plantar flexion moment (N.m/kg) Ankle negative work (J) Ankle power absorption (W/kg) Foot strike pattern Hamill 2000 Williams 2000 Laugthon 2003 Williams 2012 Shih 2013 Kulmala 2013 Rooney 2013 Nunns ± ± (-0.4, -0.1)* 64.2 ± ± (13.6, 43.5)* 0.31 ± ± (-0.2, -0.1)* 1.63 ± ± (-0.1, 0.4) 1.51 ± ± (-0.5, -0.1)* ± ± (0.5, 38.1)* ± ± (1.1, 3.4)* 2.48 ± ± (-0.2, -0.1)* ± ± (-3.4, 14.8) 9.46 ± ± (-21.4, )* 0.36 ± ± (-0.2, -0.1)* ± ± (2.4, 5.3)* 64.0 ± ± (11.6, 23.3)* 1.93 ± ± (0.5, 0.9)* 2.49 ± ± (-0.4, 0.1) 98.5 ± ± (34.5, 58.7)* 2.54 ± ± (-0.9, -0.3)* ± ± (-0.1, 0.2) 0.18 ± ± (-0.1, 0.1)