Journal Logo

Applied Sciences: Biodynamics

Kinetic Changes with Fatigue and Relationship to Injury in Female Runners


Author Information
Medicine & Science in Sports & Exercise: April 2005 - Volume 37 - Issue 4 - p 657-663
doi: 10.1249/01.MSS.0000158994.29358.71
  • Free


The promotion of fitness as part of a healthy lifestyle has led to increased numbers of participants in both competitive and recreational running (24). Unfortunately, the incidence of running-related injuries is fairly high, with estimates ranging from an annual incidence of 37 to 56% in studies involving more than 500 steadily training recreational runners (24).

Previous work has suggested that running injuries result from a complex interaction of numerous factors. The correlation of running injuries with a rapid increase in training volume or intensity, weekly running distance, and history of previous injury is well established (15,18,24,28). Impulsive forces that occur when the foot impacts the ground may also have a causative role in running injuries. This contention is based on studies that established a correlation between impact forces and injury using animal models (2). Linking impact forces to overuse injury in humans as cause and effect has proven more difficult. Previous studies have failed to establish a link between repetitive impact loading and osteoarthritis in runners (16), and although one prospective study (4) reported a significant relationship between impact peaks and loading rates and specific types of running injuries, the number of injured subjects used in this study was very low (ranging from N = 3 to 5, depending on injury type). Also, later postanalysis of this data failed to find an increased risk of injury in runners with high impact peaks and loading rates, compared with runners with low impact peaks and loading rates (20). Three recent studies of impact forces and running injuries reported conflicting results. Hreljac et al. (14) found that runners with at least one previous overuse injury had a significantly greater magnitude and rate of impact loading than runners who were injury free, suggesting a link between GRF and running injuries. In contrast, a study of noninjured runners and runners with anterior knee pain found that noninjured runners had greater peak forces and loading rates than the injured runners (11). A third study (7) reported no difference at all in GRF in male runners with and without a history of tibial stress fractures. Results from these three studies underline the current confusion surrounding the effect of GRF on running injuries, and support the need for additional work in this area.

The role of fatigue in influencing GRF, and how a change in these forces with fatigue may affect injury development, is also unclear. Several authors report that muscular fatigue caused changes in bone strain patterns, such as increases in bone strain rate and magnitude (5), as well as alterations in strain location, so that regions of normally low bone strain were subjected to much higher forces (13). Muscular fatigue has also contributed to delayed muscle activation (22) and both kinematic and kinetic changes in running stride (6,9,26,30), but much of the previous work in this area has produced inconsistent results. Some studies have reported a postfatigue increase in impact forces (10), an increase in limb acceleration (which may or may not be accompanied by a decrease in impact forces) (9,26), a shift in impact forces from the heel to the forefoot (30), or a decrease in impact forces (19). And yet another study (6) reported that the location of muscle fatigue affected impact forces. Dorsiflexor fatigue, for example, resulted in increased loading rates, whereas invertor fatigue decreased impact peaks.

Some of the inconsistency in these results may have occurred because of differences in measurement techniques or experimental protocol. For example, Derrick et al. (9) measured limb accelerations, rather than actual limb impact forces, during an exhaustive run. Acceleration measures are problematic because of difficulties associated with mounting accelerometers and discerning the actual effect of limb acceleration on impact forces (20). Limitations associated with examining a single stride (10) or single limb (9) are also causes for concern. Better insight into the problem of stride-to-stride variability and bilateral differences would be gained by considering impact-force magnitudes and loading rates for multiple strides from both legs.

No previous studies have examined how changes in impact forces with fatigue might directly correlate with prospective injury risk, especially for female runners. It is possible that runners who are ineffective at altering movement kinematics are more susceptible to injury. Impact magnitudes and loading rates may actually increase to a greater extent in these runners.

The purpose of this study was to examine how GRF were affected by an exhaustive treadmill run in female habitual runners. It was hypothesized that both impact-force magnitudes and loading rates increase in the fatigued condition, whereas active (pushoff) force magnitudes decrease. In addition, a separate retrospective and prospective analysis examined whether subjects whose impact forces increased the most with fatigue had a higher incidence of overuse running injuries. It was hypothesized that runners who exhibited the greatest increase in impact forces and lower-extremity loading rates following the fatigue protocol have a higher prevalence and incidence of running-related injuries.


Subjects, experimental protocol, and instrumentation.

Ninety (N = 90) apparently healthy adult female runners aged 18–53 volunteered to participate. All runners were running a minimum of 20 miles (32 km)·wk−1 at the time of their enrollment in the study. Most runners were competitive at the local and regional level, and a few subjects were national-caliber athletes. Subjects were considered “apparently healthy” if they were asymptomatic for cardiovascular and pulmonary disease, and met the American College of Sports Medicine (ACSM) criteria for low risk stratification for coronary artery disease (1). Subjects with a current injury to the lower extremities or low back were excluded. Written informed consent was obtained from all subjects before participation. This study received approval from the human subjects institutional review board.

Before data collection, subjects were required to refrain from alcohol, caffeine, and strenuous exercise for 24 h. They were asked to arrive well hydrated, and at least 3 h after eating. Prefatigue vertical GRF data were collected from two piezoelectric force plates embedded in a force-measuring treadmill (Kistler Instrument Corporation, Amherst, NY). Subjects were allowed to warm up on the treadmill for approximately 5–10 min. At the conclusion of the warm-up, vertical GRF data were analog-to-digitally converted for a 10-s sample period at 520 Hz (12 bit), and stored for further analysis. Data were collected with the subjects running at their approximate 5-km race pace (2.7 – 4.5 m·s−1, with 4.5 m·s−1 being the maximum speed allowed by the treadmill). Each subject wore her usual training shoes.

After collection of the GRF data, subjects were “fatigued” using a modified discontinuous O2max treadmill protocol originally designed for use with high-performance runners (3). O2max was assessed with the SensorMedics Vmax 29 system (Yorba Linda, CA). Validity of the discontinuous protocol was established previously in our lab by comparing O2max measured using this protocol with O2max measured using the Bruce protocol (r = 0.90). The discontinuous protocol consisted of 3-min work intervals beginning at 2.2 m·s−1 and a 1% gradient, with successive workloads 0.6 m·s−1 faster until 4.5 m·s−1 was reached, after which workloads increased by a gradient of 2.5% until voluntary exhaustion. Each workload was separated by a 3-min rest interval. Criteria for achievement of O2max (and fatigue) included a plateau in oxygen uptake despite an increase in workload, a respiratory exchange ratio >1.10, or attainment of the age-predicted maximum heart rate. Heart rate was monitored continuously, and blood pressure was measured at the completion of each stage. Subjects were asked to rate their level of exertion (RPE) during the last minute of each stage. Immediately after completion of the exercise test, the subject’s vertical GRF data were collected in the same manner as described above.

Data on injuries that occurred in the previous year were collected by having each subject complete an initial questionnaire upon her enrollment into the study. This questionnaire asked about a) medical and menstrual history, b) training history, and c) the occurrence of any running-related injuries in the previous year. Prospective data were collected by following all subjects for 1 yr. Each subject was contacted every 3 months after the initial assessment, and asked about her training and the occurrence of any running-related injuries. A “running-related” injury was defined in both the retrospective and prospective analyses as any musculoskeletal injury to the low back or lower extremities of an overuse nature that occurred as a result of participation in running with one or more of the following consequences: reduction in the amount or level of running (including a decrease in the usual distance, frequency, or speed of training runs or races), a need for medical advice or treatment (including a visit to a physician, chiropractor, physical therapist, or other health professional and/or the use of medications), or adverse social or economic effects (such as the inability to go to work because of the injury) (25). Subjects were also asked to note the location of the injury (low back/hip/thigh, knee, calf/foot/ankle, or other), the side affected (left, right, both), and the amount of time running was disrupted. If the subject sustained a running-related injury and sought medical advice during the follow-up period, she was asked for her consent to release the medical records pertaining to that injury. All existing medical records were then obtained to confirm the injury diagnosis.

Data analysis.

Force records were digitally filtered with a Butterworth, fourth-order, zero-lag, low-pass filter with 100-Hz cutoff. Force records were then bias adjusted to compensate for thermal drift.

Cadence and step length were determined from the force records. Three dependent measures were calculated from the vertical force records: peak impact force, the greatest force occurring within the first 10% of the support phase; impact loading rate, the greatest slope (three-point determination) from initial foot landing to the peak impact-force value; and peak active (pushoff) force, the greatest force value during midsupport (refer to labels on Fig. 1). Data for the first six right and six left steps were averaged for each subject for the pre- and postfatigue conditions. There were no clear impact peaks for a few individual steps in three runners. In those isolated cases, the average of five steps was used in the analysis. Two subjects were forefoot strikers on one limb only; therefore, the pre- and postfatigue data for that limb were excluded in the analysis. All data were normalized to body weight (BW) for the statistical analysis.

FIGURE 1— Change in ground reaction forces with fatigue. Prefatigue values are drawn in
FIGURE 1— Change in ground reaction forces with fatigue. Prefatigue values are drawn in:
blackand postfatigue ingraywith black diamonds. A sample stride from the left foot of one subject is shown; BW, units of body weight.

Statistical analysis.

Sigma Stat 3.0 (Chicago, IL) was used to perform a two-way repeated measure, mixed factor ANOVA comparing injured and noninjured runners and pre- and postfatigue values for both the retrospective and prospective analyses. As determined a priori, runners with a history of injury in the past year, or those that developed subsequent injuries, were first compared with noninjured runners. After an exploratory analysis, higher impact rates were observed in the injured versus the noninjured leg at specific injury locations, so another two-way repeated measure, mixed factor ANOVA was used to compare pre- and postfatigue kinetics by injury location (low back/hip/thigh, knee, calf/foot/ankle) for the injured and uninjured legs of subjects with a history of, or who had developed, a subsequent injury. This additional analysis allowed us to further explore the possible relationship between higher impact forces and injury. The alpha level was set at 0.05 for all tests.


Ninety subjects were originally enrolled in the study. A total of 88 subjects met the criteria for achievement of O2max, and were used in the comparison of pre- and postfatigue kinetics. A total of 87 subjects were included in the final data analysis of retrospective and prospective injury, because one subject voluntarily withdrew during the study period. Subject characteristics were as follows (mean ± SD): age 36.5 ± 9.2 yr, height 164.1 ± 6.3 cm, weight 59.0 ± 7.1 kg, body mass index 21.9 ± 2.4 kg·m−2, body fat 19.3 ± 5.2%, O2max 45.5 ± 6.5 mL·kg−1·min−1, and miles run per week 30.0 ± 9.2 (48.3 ± 14.8 km).

There were no differences between right and left legs for active peak, impact peak, or impact loading rate for either pre- or postfatigue measures. Thus, an average of right- and left-leg values was used in the statistical analysis. Peak force at impact and peak loading rates were significantly less postfatigue (P < 0.0005 for both) (Table 1), with impact peak magnitudes declining an average of 6.6%, and impact loading rates decreasing by an average of 11.8%. Peak force at pushoff (active peak) was not affected by fatigue (P = 0.18) (Table 1). A sample force curve showing the decline in impact peak magnitude and loading rate is shown in Figure 1.

Pre- and postfatigue GRF (group mean ± SD).

Cadence decreased significantly between pre- and postfatigue (183.7 ± 11.2 vs 181.1 ± 11.1 steps per minute, average of right and left legs, P < 0.0001). This was accompanied by a small, but significant, increase in step length (∼1.5 cm) between the pre- and postfatigue conditions (prefatigue: 116.3 ± 13.8 vs postfatigue: 117.8 ± 13.8 cm, average of right and left legs, P = 0.008).

A total of 46 subjects (52.9%) reported an injury in the previous year, whereas 48 (55.2%) reported a running-related injury during the 1-yr follow-up period. Injured and noninjured subjects did not significantly differ with respect to age, miles run per week, O2max, height, weight, body mass index, or percent body fat in either the retrospective or prospective analysis. Average treadmill speed used during the exhaustive treadmill run also did not differ between injured and noninjured runners.

Retrospective and prospective data on both pre- and postfatigue kinetics and injured/noninjured runners are presented in Tables 2 and 3 (pre- and postfatigue groupings have been eliminated from these tables because these data are contained in Table 1). A trend toward a significant interaction between retrospective injury and pre- and postfatigue measurements of impact loading rate (P = 0.07) was seen. No interaction was seen between previous injury and pre- or postfatigue impact peak magnitude (P = 0.85), nor for injury during follow-up and pre- or postfatigue kinetics (either impact peak (P = 0.53) or loading rate (P = 0.96)).

Means and SEM for retrospective injury × pre- and postfatigue kinetics.
Means and SEM for prospective injury × pre- and postfatigue kinetics.

We also analyzed data from only injured runners to investigate whether injury location had a specific effect on GRF (Tables 4 and 5). Values from the noninjured leg were compared with those from the injured leg (subjects with bilateral injuries were excluded). Impact peak magnitudes did not differ by injury location. Impact loading rates tended to be higher for the injured leg of subjects with knee and calf/foot/ankle injuries in both the retrospective (P = 0.09) and prospective analysis (P = 0.08), but these differences did not reach statistical significance.

Mean and SEM for retrospective injury location and injured/noninjured lower extremity.


Our hypothesis that both impact-force magnitudes and loading rates in habitual female runners increase after a fatiguing run, whereas active (pushoff) force magnitudes decrease, was not supported. The magnitude and loading rate of the impact forces both significantly decreased with fatigue, whereas the active peak magnitude did not change. The finding of decreased impact forces with fatigue is in agreement with Willson and Kernozek (30), who measured these forces directly, but disagrees with another study in which an increase in impact force with fatigue was reported (19). And yet another study (6) reported that the location of muscle fatigue affected impact forces. Dorsiflexor fatigue, for example, resulted in increased loading rates, whereas invertor fatigue decreased impact peaks.

The observed decrease in cadence and increase in step length of the runners in the fatigued condition may partially explain the decrease in impact forces. Our results are similar to previous work by Derrick et al. (9), who reported a 3-cm increase in stride length with fatigue induced by treadmill running, and to those of Verbitsky et al. (26), who found a decrease in cadence with fatigue induced by a 30-min run at the anaerobic threshold. The decrease in cadence in that study was associated with an increase in tibial acceleration at impact, which Verbitsky et al. suggest results in an increased magnitude of the heel-strike shock wave (26). However, as discussed further in the following paragraphs, the work by Derrick et al. (9) found that an increase in acceleration may actually be associated with a decrease in impact forces, such as we observed.

Changes in joint kinematics at heel strike, such as knee- and ankle-flexion angles, would accompany any change in stride length and cadence, and may be partially responsible for the changes seen in GRF in our study. Such changes in joint kinematics that accompany fatigue have been reported by several others (6,9,12,20). In a study utilizing a direct dynamics simulation technique, Gerritsen et al. (12) reported that impact peak forces were greatly influenced by the amount of ankle plantarflexion (85 N·deg−1 of change in foot angle) and knee flexion (68 N·deg−1 of change in knee angle). A flatter foot angle and a more extended knee at heel strike resulted in higher impact forces. Applying these observations to our study, runners may have adopted a running style characterized by an increase in knee flexion and/or ankle dorsiflexion at heel contact, such as described by Gerritsen et al. (12). These changes may have acted to decrease the impact forces, perhaps by decreasing the effective mass (8,23).

Valiant (23) describes effective mass as “that part of the mass of the human body that has an effect on the forces that are developed passively within the knee joint during the impact phase” and states that a smaller effective mass is related to greater shock absorption and, thus, lower impact forces. One way of achieving a smaller effective mass is through greater knee flexion (8), or a change in rearfoot position (23) at impact. In fact, a decrease in effective mass may cause an increase in peak acceleration, as the smaller mass will be more easily accelerated. This theory is supported by Derrick et al. (9), who ran recreational runners at their average 3200-m running velocities until fatigue, and also by Verbitsky et al. (26), who reported an increase in peak limb acceleration with fatigue. Derrick et al. (9) suggested that the increase in acceleration may lead to decreased impact forces due to the subsequent decrease in effective mass caused by the increased knee flexion at heel contact and, therefore, might not be associated with increased injury risk.

Our hypothesis that the active peak magnitude decreases in the fatigued condition was not supported by our results, as no change was seen. This hypothesis was based on a study by Christina et al. (6), in which localized fatigue of the foot invertor muscles resulted in significant changes in vertical GRF during running, including a decrease in active peak. This decrease was not seen in this study, perhaps because our fatigue protocol did not result in the same type of localized muscle fatigue as the Christina et al. study (6).

The observed injury incidence among female runners of 52.9% and 55.2% in the retrospective and prospective analyses, respectively, is similar to the findings of others who studied competitive runners with similar weekly mileages (15,18,28). Unlike other studies (15,18,24,28), this study did not find a relationship between weekly mileage and injury. This is possibly due to the narrow range of weekly running mileages reported by our subjects—more than two thirds ran between 25 and 40 miles·wk−1 (40 and 65 km·wk−1), and all subjects averaged at least 20 miles (32 km)·wk−1. Other studies that reported a relationship between weekly mileage and injury compared female runners averaging 2 miles (3 km)·wk−1 with those running 70 miles (112 km)·wk−1 (18), which may explain the divergent findings.

The hypothesis that subjects with the greatest increase in ground reaction forces with fatigue would have the highest injury incidence was not supported. This hypothesis was based on the results of previous studies that showed an increase in ground reaction forces and/or limb acceleration with fatigue (6,9,10,26), as well as work by Hreljac et al. (14), which found that runners with a history of overuse running injuries had higher impact forces than runners who had never experienced an overuse running injury. The analysis of retrospective data revealed an interaction between previous injury and alterations in impact loading rates with fatigue. The runners who had previously been injured differed from the noninjured runners in that the decrease in impact loading rates seen with fatigue was less than the decrease seen in the noninjured runners. These findings suggest a possible link between previous injury and an altered response to fatigue, exposing runners to higher impact rates as they fatigue. However, we did not find this relationship in the prospective study, so the inability of injured runners to deal with fatigue in the same manner as the noninjured runners was likely an effect of the original injury, rather than its cause. The resulting exposure to relatively higher impact loading rates could contribute to subsequent reinjury, however, and might be the subject of future investigations examining what factors determine when it is safe for injured runners to return to full training.

The results from both the retrospective and prospective analyses of the injured runners showed higher impact rates in the injured limb compared with the noninjured limb for certain injury locations. We noticed a trend toward significant interaction between impact loading rates and injury at the calf/foot/ankle. It is reasonable to assume that the calf/foot/ankle and knee are more susceptible to injury because of their role in shock wave attenuation. Previous work by Voloshin et al. (27) has demonstrated that not only does fatigue interfere with the ability of the musculoskeletal system to attenuate shock waves, but also that this capacity is related to both fatigue level and orientation of skeletal structures.

Other studies (4,11,14) have looked for correlations between baseline GRF forces and injury with mixed results. As stated previously, Hreljac et al. (14) found that runners with a history of overuse running injuries had higher average impact forces than runners who had never experienced an overuse running injury. However, a study by Duffey et al. (11) reported that runners with anterior knee pain had lower average impact forces than noninjured runners. A third study on the topic failed to find any correlation between prospective injury risk and higher impact rates (4). The findings of the present study agree with those of Hreljac et al. (14) in that higher impact forces may contribute to injury development; however, our evidence for this was not overwhelmingly strong in the prospective study, as it was evident only for those runners who had injuries at the knee. Because injured runners, as an entire group, did not differ from noninjured runners in baseline GRF, it may be concluded that impact force is neither the sole nor a major contributor to development of injury.

Potential limitations of this study include the use of a treadmill to investigate changes in GRF that may differ from changes occurring during overground running. Nevertheless, the use of a treadmill for data collection in studies of GRF has numerous advantages over the force platforms used in overground running studies. These advantages include the abilities to control running speed and to measure GRF over successive strides. At least one study has reported that the use of a force-measuring treadmill resulted in accurate collection of GRF data as compared with a force platform (17). Two other studies, however, suggest that both kinematics (21) and kinetics (29) differ between treadmill and overground locomotion, so the generalizability of some of our findings is in question. For example, Nigg et al. (21) reported that subject kinematic response varied greatly between treadmill and overground running in some subjects, but not in others, depending on individual running style, shoe type, and treadmill type. However, subjects in that study systematically landed in a more plantarflexed foot position on the treadmill compared with overground running, which would alter impact forces (21). Although performed on walking strides instead of running, White et al. (29) found small kinetic differences in the force magnitudes of overground and treadmill locomotion during mid- and late stance; a similar response in running may or may not be present. Other potential limiting factors included testing subjects in their own running shoes and subjects changing shoes throughout the year. Although the use of different shoe types may have increased variability, it also may have more accurately reflected a real-life situation for each runner.

In summary, this study found that fatigue induced by an exhaustive treadmill run resulted in decreased impact peaks and loading rates. The changes in GRF can be attributed to an altered running style that includes changes in cadence and stride length, and possibly alterations in lower-extremity joint kinematics, all of which would act to reduce impact forces. It is not known if the above changes represent a strategy to minimize impact forces and protect the fatigued runner from injury, or if they represent a fatigue-induced loss of optimal performance capability.

Although impact magnitudes and loading rates declined with fatigue in both injured and noninjured runners, the impact loading rates of subjects with a history of a previous injury declined significantly less than the noninjured runners. Because this decline was not observed in the prospective study, it may be that the inability of previously injured runners to compensate for fatigue in the same manner as the noninjured runners was an effect of, rather than a contributing cause, to the original injury. The relationship between impact forces and injury location was not strong, leading us to question whether higher impact forces contribute to running related injuries. Future studies may further elucidate the contribution of impact forces to injury development in runners.


1. American College of Sports Medicine. ACSM‘s Guidelines for Exercise Testing and Prescription. Philadelphia: Lippincott Williams & Wilkins, 2000, pp. 24–27.
2. Anderson, D. D., T. D. Brown, K. H. Yang, and E. L. Radin. A dynamic finite element analysis of impulsive loading of the extension-splinted rabbit knee. J. Biomech. Eng. 112:119–128, 1990.
3. Australian Sports Commission. Physiological Tests for Elite Athletes. Champaign, IL: Human Kinetics, 2000, pp. 337–338.
4. Bahlsen, A. The etiology of running injuries: a longitudinal, prospective study. Doctoral Dissertation. University of Calgary, Calgary, Alberta, Canada, 1989.
5. Burr, D. B. Bone, exercise, and stress fractures. Exerc. Sport Sci. Rev. 25:171–194, 1997.
6. Christina, K. A., S. C. White, and L. A. Gilchrist. Effect of localized muscle fatigue on vertical ground reaction forces and ankle joint motion during running. Hum. Mov. Sci. 20:257–276, 2001.
7. Crossley, K., K. L. Bennell, T. Wrigley, and B. W. Oakes. Ground reaction forces, bone characteristics, and tibial stress fracture in male runners. Med. Sci. Sports Exerc. 31:1088–1093, 1999.
8. Denoth, J. Load on the locomotor system and modelling. In: Biomechanics of Running Shoes, B. M. Nigg (Ed.). Champaign, IL: Human Kinetics, 1986, pp. 63–116.
9. Derrick, T. R., D. Dereu, and S. P. Mclean. Impacts and kinematic adjustments during an exhaustive run. Med. Sci. Sports Exerc. 34:998–1002, 2002.
10. Dickinson, J. A., S. D. Cook, and T. M. Leinhardt. The measurement of shock waves following heel strike while running. J. Biomech. 18:415–422, 1984.
11. Duffey, M. J., D. F. Martin, D. W. Cannon, T. Craven, and S. P. Messier. Etiologic factors associated with anterior knee pain in distance runners. Med. Sci. Sports Exerc. 32:1825–1832, 2000.
12. Gerritsen, K. G. M., A. J. van den Bogert, and B. M. Nigg. Direct dynamics simulation of the impact phase in heel-toe running. J. Biomech. 28:661–668, 1995.
13. Grimston, S. K., and R. F. Zernicke. Exercise-related stress responses in bone. J. Appl. Biomech. 9:2–14, 1993.
14. Hreljac, A., R. N. Marshall, and P. A. Hume. Evaluation of lower extremity overuse injury potential in runners. Med. Sci. Sports Exerc. 32:1635–1641, 2000.
15. Jacobs, S. J., and B. L. Berson. Injuries to runners: a study of entrants to a 10,000 meter race. Am. J. Sports Med. 14:151–155, 1986.
16. Konradsen, L., E. M. Berg-Hanson, and L. Sondergaard. Long distance running and osteoarthritis. Am. J. Sports Med. 18:379–381, 1990.
17. Kram, R., T. M. Griffin, J. M. Donelan, and Y. H. Chang. Force treadmill for measuring vertical and horizontal ground reaction forces. J. Appl. Physiol. 85:764–769, 1998.
18. Macera, C. A., R. R. Pate, K. E. Powell, K. L. Jackson, J. S. Kendrick, and T. E. Craven. Predicting lower-extremity injuries among habitual runners. Arch. Intern. Med. 149:2565–2568, 1989.
19. Nicol, C., P. V. Komi, and P. Marconnet. Fatigue effects of marathon running on neuromuscular performance. Scand. J. Med. Sci. Sports 1:10–17, 1991.
20. Nigg, B. M., G. K. Cole, and G. P. Bruggemann. Impact forces during heel-toe running. J. Appl. Biomech. 11:407–432, 1995.
21. Nigg, B. M., R. W. De Boer, and V. Fisher. A kinematic comparison of overground and treadmill running. Med. Sci. Sports Exerc. 27:98–105, 1995.
22. Nyland, J. A., R. Shapiro, R. L. Stine, T. S. Horn, and M. L. Ireland. Relationship of fatigued run and rapid stop to ground reaction forces, lower extremity kinematics, and muscle activation. J. Orthop. Sports Phys. Ther. 20:132–137, 1994.
23. Valiant, G. A. Transmission and attenuation of heelstrike accelerations. In: Biomechanics of Distance Running, P. R. Cavanagh (Ed.). Champaign, IL: Human Kinetics Books, 1990, pp. 225–247.
24. van Mechelen, W. Running injuries: a review of the epidemiological literature. Sports Med. 14:320–335, 1992.
25. van Mechelen, W., H. Hlobil, and H. C. G. Kemper. Incidence, severity, aetiology and prevention of sports injuries: a review of concepts. Sports Med. 14:82–99, 1992.
26. Verbitsky, O., J. Mizrahi, A. Voloshin, J. Treiger, and E. Isakov. Shock transmission and fatigue in human running. J. Appl. Biomech. 14:300–311, 1998.
27. Voloshin, A. S., J. Mizrahi, O. Verbitsky, and E. Isakov. Dynamic loading on the human musculoskeletal system: effect of fatigue. Clin. Biomech. 13:515–520, 1998.
28. Walter, S. D., L. E. Hart, J. M. McIntosh, and J. R. Sutton. The Ontario cohort study of running-related injuries. Arch. Intern. Med. 149:2561–2564, 1989.
29. White, S. C., H. J. Yack, C. A. Tucker, and H. Y. Lin. Comparison of vertical ground reaction forces during overground and treadmill walking. Med. Sci. Sports Exerc. 30:1537–1542, 1998.
30. Willson, J. D., and T. W. Kernozek. Plantar loading and cadence alterations with fatigue. Med. Sci. Sports Exerc. 31:1828–1833, 1999.


©2005The American College of Sports Medicine