Introduction
There is a paucity of knowledge about risk factors for running-related injuries (RRI). The key factor associated with injury development seems to be somehow linked to training volume because errors in training volume have been suggested as the main cause of as much as 60–70% of all RRIs (16,19,28). In particular, sudden increases in weekly volume are considered injurious because the increased number of applied stress exposures because of an excessive increase in number of steps taken may overwhelm the ability for adaptive change and tissue repair (26).
To avoid injury, a principle that the running volume should be increased gradually over time is considered important. In a training program based on a progression in weekly volume of a maximum of 10%, the so called 10% rule, a gradually adaptive change may occur and the risk of overuse injury reduced (17). On the contrary, the risk of injury may be increased by following a training program with a progression above 10%. Buist et al (7) investigated the injury survival among novice runners after a 10% increase in weekly training compared with novice runners training with an increase of 24%. No difference in injury survival was found between the 2 groups. Still, participants in other studies have reported the sudden change in training volume before injury origin as the main reason for injury development (15,19). Based on this, it seems plausible to assume that progression in weekly volume is somehow linked with RRI, but no clear threshold between safe progression and injurious progression exists. A possible reason for the lacking evidence on the link between progression in running volume and RRI may be the methods used to obtain information about the training volume.
Previously, information about running volume has been collected through questionnaires or self-reported running diaries (4,7,15,21,22,36). Several studies conclude that this approach leads to training volume being estimated wrongly (18), possibly because of subject recall bias, and time-consuming self-reporting may demotivate the participant to continue in the study (13,35). Therefore, the methods used in previous studies to measure running volume by subjective methods (questionnaires, surveys, diaries) should be taken into careful consideration. New methods to quantify training volume are highly needed because valid and reliable measurements may create a foundation to investigate if excessive progression in training volume, suggested by many, is associated with development of RRI.
Previously, the GPS assessment of human speed and distance has been tested in field sports (1,9), in court-based sports (8), and during walking (27) and running (30,32,33). Townshend et al (33) concluded that, with reduced size, cost, and ease of use, the GPS offers a valid alternative to subjective self-reporting that can be used in the study of running or walking (33). The recent development of the GPS technology has made it possible for novice, recreational, and elite runners around the world to buy and use GPS watches to quantify their training volume. Hereby, a possibility exists for researchers to gather detailed information of training volume from large groups of runners. Based on the information on training volume gathered by GPS, the progression in weekly volume can be calculated.
Following the hypothesis that progression in weekly volume may contribute to injury if the progression is severe enough, we designed an explorative, prospective follow-up study using GPS to quantify training volume. The purpose of this study was to investigate the validity and reliability of running volume quantified by GPS watches and to investigate if GPS can be used to detect deleterious progression in weekly training volume among runners.
Methods
Experimental Approach to the Problem
Two studies were conducted: A prospective follow-up study including novice runners and a trial to investigate the validity and reliability of GPS watches.
The follow-up trial was designed as a prospective 10-week observational study including healthy novice runners who used GPS to register their training. The dependent variable of interest was RRI; “An injury was defined as any musculoskeletal complaint of the lower extremity or back causing a restriction of running for at least 1 week.” This definition is a modified version of the definition used by Buist et al (7). The independent variable of interest was progression in weekly training volume. Training progression was calculated in percentage, based on the training volume covered in 1 week divided by the training volume covered in the week before, multiplied by 100.
Validity trials were conducted to investigate the measurement errors between the GPS watch and a gold standard. Furthermore, the reliability between watches was investigated.
In the validity trials, the measurements of distance, speed, and elevation computed from a commonly used GPS watch was compared with a Real Time Kinematic (RTK) geodetic GPS receiver (GPS1200+; Leica Geosystems AG, Heerburg, Switzerland). In case of unsimultaneous data collection, distances computed from the GPS watch were compared with an average distance measured by RTK or Electronic Distance Measurement (EDM) (Leica TPS1205+). It was hypothesized that the GPS watch, under ideal conditions, had a limit of agreement of ±20 m in 95% of the measurements compared with the gold standard over a 1,000 m distance. In comparison, on the manufacturers’ homepages it is stated that an expected positional standard deviation is a few centimeters for the RTK and less than a few centimeters for EDM over a distance of 1,000 m. Therefore, RTK and EDM were used as gold standards. The RTK and EDM equipment are tested at regular intervals at Aalborg University to ensure compliance with the specified accuracies. The reliability of 2 identical GPS watches was assessed by comparisons of distance, speed, and elevation. The authors believe that a measurement error below 10% for mean distance, speed, and elevation is not clinically relevant when the link between training characteristics and RRI has to be investigated. It was hypothesized that 95% of the measurements of distance between 2 identical watches could be assessed within ±30 m over a 1,000 m distance in ideal conditions.
Subjects
The study was granted approval from the local ethics board (M-20100272) and the Danish Data Protection Agency. All participants signed informed consent before inclusion. Healthy persons between 18 and 65 years of age, who had no injury of the lower extremity in the 3 months before inclusion, who had not been running on a regular basis (10 km total in all training sessions in the previous 12 months), had access to the internet and got an e-mail address, were eligible for inclusion in the prospective study. Participants were recruited from the municipality of Aarhus, Denmark during February 2011 via posters and mails distributed at local companies and universities. A total of 118 persons signed up for the study. Fifty-eight were excluded because of the inclusion and exclusion criteria. Finally, 60 healthy novice runners (32 men, 28 women, 39.8 ± 9.3 years, body mass index [BMI] 25.5 ± 3.9) from the Central Region, Denmark, were included. Flow chart is presented in Figure 1.
;)
Figure 1: Flow chart of the participants included, excluded, and censored in the 10-week prospective follow-up study. Initially, 118 persons requested to participate. After exclusion of 58 persons, a total of 60 persons were included in the study. Thirteen participants sustained at least one running-related injury during follow-up.
Based on the data obtained from the training characteristics, 2 participants did not upload any data and were therefore excluded from the analysis. Demographic characteristics and training-related characteristics are presented in Table 1. Thirteen participants sustained a RRI during follow-up. Medial tibial stress syndrome was the most frequently diagnosed injury (n = 4).
Table 1: Demographic characteristics and training patterns of injured and healthy participants.
Procedures
Information about the prospective study was available on the DANO-RUN homepage (25) and in material forwarded to persons who responded to the advertisements. In case of interest in participating, an online questionnaire had to be completed. After a telephone interview, an appointment for a baseline investigation was made. At baseline, participants received a GPS watch (Forerunner 110; Garmin International, Inc., Olathe, KS, USA) and were instructed to upload GPS log files to a personal online training diary. Afterwards, participants were presented to the injury definition. If the participants sustained an RRI during the follow-up period, they were instructed to contact the research group via their personal training diary. Then, the participant was contacted by telephone and an appointment for a clinical examination was made. At the examination, their injury was classified as running related or related to other causes. Only RRI was included as outcome in the study.
During follow-up, participants were able to upload log files from their GPS watch. In case data from the GPS watch was not available, participants were instructed to manually upload their training sessions to the DANO-RUN homepage. The participants ran a total distance of 4,556.5 km in 1,172 training sessions during the 10 weeks. Of the 1,172 training sessions uploaded, 1,132 were uploaded from the GPS device and 40 sessions were uploaded manually.
For the validity and reliability trial, data collection was conducted in 3 different scenarios in Aalborg and Aarhus, Denmark: First, on 2 different straight cycle paths with a clear view to the sky with RTK collecting simultaneously. Second, in an urban area with buildings up to 5 floors nearby with the RTK data collected in a different time span than data from the GPS watch. Finally, in a forest with the EDM data collected in a different time span. Measurements in all scenarios from the GPS watches were collected at different times of the day at different days to assure diverse satellite geometry. Therefore, the reliability of the GPS watches was calculated as between day-, intra-, and interwatch reliability. Sample frequency for the watches and RTK was 1 Hz. For each Hertz sampling, a three-dimensional position and a time tag were assessed. The lengths of the paths were between 1 and 1.2 km. A total of 20–40 trials were conducted in each of the scenarios with a GPS watch on each arm. Because the 2 watches were mounted on the right and left arm, they did not record the coordinates of the exact same point. In the straight path, data collections of watches and RTK were measured simultaneously. In the urban area and in the forest, separate data collections were made.
In the straight paths, validity and reliability comparisons of derived watch and RTK distances, speeds, and elevations measured in the same time spans were made after 300 and 1,000 m. In the urban area, reliability comparisons of derived watch distances, speeds, and elevations were compared after 300 and 1,229.5 m, whereas validity comparison between watch and RTK distance measured in different time span was made after 1,229.5 m. In the forest, reliability comparisons of derived watch distances after 300 and 998 m was made, whereas validity comparison between watch and EDM distance was made after 998 m.
For each data set, the traveled two-dimensional (2D) horizontal distance was computed for each 1-second interval. Then, these distances were added to compute the total traveled 2D distance. This was done for all data sets coming from both the RTK and the GPS watches. Likewise, the average running speed was computed for each data set by dividing the total traveled distance by the total elapsed time. For each run, the difference between the traveled distance recorded by the RTK and each of the GPS watches was computed. These differences were used for assessing the validity and reliability of the average distances derived from the watches. Similarly, the differences between the average running speed recorded by the RTK and the average running speed recorded by each of the GPS watches were computed for each run. These differences were used for assessing the validity and reliability of the running speed derived from the watches. Total distance of each run in the urban area and the forest assessed with the GPS watch was compared with a distance measured by RTK or EDM in a different time span.
Statistical Analyses
In the prospective study, descriptive data were presented as counts and percentage in case of dichotomous data and mean and range in case of continuous data. Student’s t-test with equal variance was applied to tests for differences in training progression between healthy and injured participants. Furthermore, paired t-test was applied to test for differences in weekly training progression among the injured participants in the week before injury origin compared with the average weekly progression in the weeks before this. T-tests and chi-square test were applied to test for differences in demographic characteristics. Because the prospective study was designed to explore the GPS data, no sample size calculation was made before the inclusion.
In the investigation of the validity and reliability of volume, speed, and elevation measured by GPS, the means and 95% confidence intervals (CI) were calculated based on the difference between the gold standard and the watch or two measurements from two identical watches. Bland Altman Limits of Agreement (2,3) were used to calculate the 95% prediction limits of measurement errors between two GPS watches or between the watch and the RTK in the straight path scenario. Bland-Altman plots of the differences against average of the two measurements for each of the reliability trials and validity trial in the straight path did not show any sign of the differences depending systematically on the average. In the urban area and forest, the mean difference and 95% CIs of the watch and gold standards were presented. Hypotheses were tested on a 5% alpha level.
Results
Although not significant, participants with injuries had an average increase in weekly training volume of 31.6 ± 3.1% compared with a 22.1 ± 2.1% increase among healthy participants. When the weekly progression in training volume the week before the onset of injury was compared with the weekly progressions in training volume made by the injured participants in the other weeks, the mean difference was 86% (95% CI = 12.9–159.9%, p = 0.026).
Results from the validity and reliability of GPS watches revealed the watches to measure a significantly shorter distance of 7.7 m compared with the RTK 1,000 m distance (p < 0.001) in the straight paths. The speed measured by watches was significantly different than the RTK speed with the average difference of 0.3 km·h−1 (p < 0.001). No significant difference in elevation was found between the RTK and a watch. No significant differences in distance, speed, or elevation were found between 2 watches placed in the same position. In the urban area, the watch measured a significantly shorter distance of −14 m (−18; −9) over a 1,229-m distance than the RTK. Similarly, the mean difference in distance, speed, and elevation between 2 watches was 19.3 m, 0.2 km·h−1, and 1 m, respectively.
After 998 m in the forest, a watch measured a significantly shorter average distance of −62 m (−70; −54) compared with the EDM (p < 0.001). No significant differences in distance, speed, or elevation were found between watches. Results from the measurements in straight paths, urban area, and forest are presented in Table 2. In Table 3, the measurement errors between the mean measures of distance, speed, and elevation with the watch and the gold standard are presented.
Table 2: Results from GPS watch versus gold standard (validity) and watch versus watch measurements (reliability) in open landscape, urban area, and forest, respectively.*
Table 3: Measurement errors between a gold standard and a commonly used GPS watch.*
Discussion
In this study, the use of GPS to detect deleterious training patterns among runners is investigated. Previously, the information biases in objective self-reporting of training volume have affected the possibility to investigate the relationship between training volume and RRI (26,37). If information biases are minimized by the use of GPS, Nielsen et al (26) suggested that an interesting focus for future research would be to investigate if the sudden increase in one or more training variables, as suggested by many (10,11,28,29,37), is more strongly related to injury than the absolute volume, which is currently suggested to be the main contributor to injury (6,14,20,34). The findings in this study revealed that the “sudden” increase in weekly training volume may be associated with injury development. The average weekly progressions among healthy and injured participants were 22.1 and 31.6%, respectively. Although the difference was not significant, it shows that a progression in weekly kilometers of more than 10%, which is currently considered the threshold, may be acceptable to many. This is supported by the results reported by Buist et al (7), who found no significant difference in the time to RRI in a group of runners with a 13-week training program with a mean increase in training duration of 10% per week compared with a group of runners training an 8-week training program with a mean increase in duration of 24% per week. Based on the results from the current study, increases may become deleterious at a weekly increase above 30%. It must be emphasized that this finding should be interpreted with extreme caution because the study sample was low and the design was explorative. Still, more work has to be conducted in large-scale prospective studies to ascertain if progression in weekly volume is the main contributor to injury (31) and how the possible effect of one training variable is affected by other risk factors causing effect modification or confounding (23,24). In Table 2, an interesting finding was presented because BMI was significantly associated with RRI. Healthy participants had an average BMI of 24.8 ± 3.6, whereas injured participants had an average BMI of 27.6 ± 4.5. This finding is in agreement with the findings by Buist and Bredeweg (5) who also reported a higher risk of RRI among persons with a BMI above 25. Unfortunately, it was not possible to include BMI as an effect measure modifier on the association between progressions in weekly training volume and injury development because of the relatively small sample. An interesting focus in future research in large cohorts of novice runners would be to include BMI as an effect measure modifier on the association between progression in weekly training volume and RRI. Although the findings in this study on the link between training volume and RRI should be interpreted with caution, the importance of the topic addressed in this article must not be underestimated because training inevitable is a part in the causal chain leading to injury. Training volume has been described as a necessary cause to injury development by Shrier (31). Based on this, a runner must take part in some running to sustain an RRI. This is supported by Meeuwisse et al (24) who stated that exposure to injury is a combination of possessing one or more risk factors (like BMI or previous injury) and then to a lesser or greater degree run with the risk factor. We encourage all researchers investigating the mechanisms to injury in large cohorts of runners to register training objectively and include the training variables as main exposure to injury while BMI, previous injuries, and other risk factors should be considered as effect measure modifiers.
The greatest measurement error on training volume was found to be 6.2%. Such errors are almost identical to errors reported previously (12). Based on this, we conclude that there are no clinically relevant differences between the average distance derived from GPS watches and the gold standard measurements. The GPS seems the feasible method to obtain information on running volume in a group of runners training in different environments. However, there are some limitations that should be considered before using GPS to quantify training characteristics in clinical studies. First, participants must own a GPS watch or GPS device, such as a smart phone. A watch is rather costly, which may create selection bias in studies, because runners owning a GPS unit may have other characteristics than runners not owning one: A possible solution would be to provide a GPS watch to the included participants. Second, participants must wear the GPS watch in each training session. When the watch is out of battery or the participants forget to use it, no data are collected by the GPS. We found that 96.6% of the total training sessions uploaded to the training diaries were uploaded via the GPS log file. Although this may seem high, participants should be encouraged, on a regular basis, to register their training by GPS. Still, it stands to question if any information bias exist because of completed training sessions not updated by GPS or manually to the homepage. Third, the watch must be stopped when a running session is finished. If the watch continues to measure, it will add extra time and some distance to the training session, although the person stands still. Despite these limitations, GPS still seems like a feasible method to quantify training data among runners.
Practical Applications
No clinically relevant measurement errors of the GPS devices were found for training volume. Based on this, runners and coaches can use the GPS because the technology has a potential to detect errors in training volume that may be associated with the development of RRI. Still, more studies are needed to document a relationship between training and RRI in a large sample of runners who register their training by GPS. Based on the results from the current study, increases in weekly training progression may become deleterious at a weekly increase above 30%, which is way more than the 10% rule currently used as a guideline for correct progression in weekly volume by runners and coaches. Thus, no clear evidence for safe progression of weekly volume exists. But it seems like some individuals may tolerate weekly progressions around 20–25%, at least for a short period of time. We hypothesize that persons with a BMI above 25 are more prone to sustain injuries and, because of their size, may tolerate less progression in weekly training volume than individuals with a BMI between 20 and 25. If this is true, coaches should design individual training programs taking into account the BMI, and other potential risk factors like previous injury, when evaluating which weekly progression may be tolerated by the athlete/runner.
Acknowledgments
The study was financed by Orthopaedic Research Unit, Science and Innovation Center, Aalborg Hospital, Aarhus University. Garmin International, Inc. sponsored 10 GPS watches to the trial. None of the authors have financial interests related to the content of the article. None of the authors are owners or employees of Garmin who produces the GPS watches or Elkjaer IT who developed the homepage used in this study. However, Garmin International, Inc. has donated 10 GPS watches to the Division of Geomatics, Aalborg University where co-author Peter Cederholm is affiliated. Thus, there are no professional relationships with companies or manufacturers who will benefit from the results of the present study. No funding was received from National Institutes of Health (NIH); Wellcome Trust; Howard Hughes Medical Institute (HHMI); or others. The results of the present study do not constitute endorsement of the product by the authors or the National Strength and Conditioning Association.
References
1. Aughey RJ. Applications of GPS technologies to field sports. Int J Sports Physiol Perform 6: 295–310, 2011.
2. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307–310, 1986.
3. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res 8: 135–160, 1999.
4. Bovens AM, Janssen GM, Vermeer HG, Hoeberigs JH, Janssen MP, Verstappen FT. Occurrence of running injuries in adults following a supervised training program. Int J Sports Med 10(Suppl. 3): S186–S190, 1989.
5. Buist I, Bredeweg SW. Higher risk of injury in overweight novice runners. Br J Sports Med 45: 338, 2011.
6. Buist I, Bredeweg SW, Lemmink KA, van Mechelen W, Diercks RL. Predictors of
running-related injuries in novice runners enrolled in a systematic training program: A prospective cohort study. Am J Sports Med 38: 273–280, 2010.
7. Buist I, Bredeweg SW, van Mechelen W, Lemmink KA, Pepping GJ, Diercks RL. No effect of a graded training program on the number of
running-related injuries in novice runners: A randomized controlled trial. Am J Sports Med 36: 33–39, 2008.
8. Duffield R, Reid M, Baker J, Spratford W. Accuracy and reliability of GPS devices for measurement of movement patterns in confined spaces for court-based sports. J Sci Med Sport 13: 523–525, 2010.
9. Dwyer DB, Gabbett TJ. Global positioning system data analysis: Velocity ranges and a new definition of sprinting for field sport athletes. J Strength Cond Res 26: 818–824, 2012.
10. Fredericson M. Common injuries in runners. Diagnosis, rehabilitation and prevention. Sports Med 21: 49–72, 1996.
11. Galloway MT, Jokl P, Dayton OW. Achilles tendon overuse injuries. Clin Sports Med 11: 771–782, 1992.
12. Gray AJ, Jenkins D, Andrews MH, Taaffe DR, Glover ML. Validity and reliability of GPS for measuring distance travelled in field-based team sports. J Sports Sci 28: 1319–1325, 2010.
13. Hootman JM, Macera CA, Ainsworth BE, Martin M, Addy CL, Blair SN. Predictors of lower extremity injury among recreationally active adults. Clin J Sport Med 12: 99–106, 2002.
14. Hreljac A.
Etiology, prevention, and early intervention of overuse injuries in runners: A biomechanical perspective. Phys Med Rehabil Clin N Am 16: 651–67, 2005.
15. Jacobs SJ, Berson BL. Injuries to runners: A study of entrants to a 10,000 meter race. Am J Sports Med 14: 151–155, 1986.
16. Johnson R. Common running injuries of the leg and foot. Minn Med 66: 441–444, 1983.
17. Johnston CA, Taunton JE, Lloyd-Smith DR, McKenzie DC. Preventing running injuries. practical approach for family doctors. Can Fam Physician 49: 1101–1109, 2003.
18. Knobloch K, Yoon U, Vogt PM. Acute and overuse injuries correlated to hours of training in master running athletes. Foot Ankle Int 29: 671–676, 2008.
19. Lysholm J, Wiklander J. Injuries in runners. Am J Sports Med 15: 168–171, 1987.
20. Macera CA. Lower extremity injuries in runners. advances in prediction. Sports Med 13: 50–57, 1992.
21. Macera CA, Pate RR, Powell KE, Jackson KL, Kendrick JS, Craven TE. Predicting lower-extremity injuries among habitual runners. Arch Intern Med 149: 2565–2568, 1989.
22. Marti B, Vader JP, Minder CE, Abelin T. On the epidemiology of running injuries. The 1984 bern grand-prix study. Am J Sports Med 16: 285–294, 1988.
23. Meeuwisse WH. Athletic injury
etiology: Distinguishing between interaction and confounding. Clin J Sport Med 4: 171–175, 1994.
24. Meeuwisse WH, Tyreman H, Hagel B, Emery C. A dynamic model of
etiology in sport injury: The recursive nature of risk and causation. Clin J Sport Med 17: 215–219, 2007.
25. Nielsen R. DANO-RUN Homepage [Internet]. Aarhus, DK: DANO-RUN, 2011. Available at:
http://www.vilober.dk/ and
http://www.webcitation.org/62wa184il. Accessed January 27, 2012.
26. Nielsen RO, Buist I, Sorensen H, Lind M, Rasmussen S. Training errors and running related injuries: A systematic review. Int J Sports Phys Ther 7: 58–75, 2012.
27. Noury-Desvaux B, Abraham P, Mahe G, Sauvaget T, Leftheriotis G, Le Faucheur A. The accuracy of a simple, low-cost GPS data logger/receiver to study outdoor human walking in view of health and clinical studies. PLoS One 6: e23027, 2011.
28. Renstrom AF. Mechanism, diagnosis, and treatment of running injuries. Instr Course Lect 42: 225–234, 1993.
29. Rzonca EC, Baylis WJ. Common sports injuries to the foot and leg. Clin Podiatr Med Surg 5: 591–612, 1988.
30. Schutz Y, Herren R. Assessment of speed of human locomotion using a differential satellite global positioning system. Med Sci Sports Exerc 32: 642–646, 2000.
31. Shrier I. Understanding causal inference: The future direction in sports injury prevention. Clin J Sport Med 17: 220–224, 2007.
32. Terrier P, Schutz Y. How useful is satellite positioning system (GPS) to track gait parameters? A review. J Neuroeng Rehabil 2: 28, 2005.
33. Townshend AD, Worringham CJ, Stewart IB. Assessment of speed and position during human locomotion using nondifferential GPS. Med Sci Sports Exerc 40: 124–132, 2008.
34. 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 41: 469–480, 2007.
35. Van Middelkoop M, Kolkman J, Van Ochten J, Bierma-Zeinstra SM, Koes BW. Risk factors for lower extremity injuries among male marathon runners. Scand J Med Sci Sports 18: 691–697, 2008.
36. Walter SD, Hart LE, McIntosh JM, Sutton JR. The ontario cohort study of
running-related injuries. Arch Intern Med 149: 2561–2564, 1989.
37. Wen DY. Risk factors for overuse injuries in runners. Curr Sports Med Rep 6: 307–313, 2007.
