Journal Logo


Injured Runners Do Not Replace Lost Running Time with Other Physical Activity


Author Information
Medicine & Science in Sports & Exercise: May 2020 - Volume 52 - Issue 5 - p 1163-1168
doi: 10.1249/MSS.0000000000002227


Runners enjoy a number of long-term health benefits, ranging from a longer lifespan to a lower risk of cardiovascular disease, disability, and frailty in older age (1,2). Running requires little in the way of equipment or facilities; its accessibility may explain why one in eight American adults participate in running or jogging for exercise during a typical month (3). Though running is a popular and accessible form of physical activity, the annual incidence of injury among recreational runners may approach or exceed 50%, and median time to recovery following an injury is upwards of 10 wk (4,5).

In sports injury research, running-related injuries are traditionally defined based on time lost from running—that is, modifications made to planned training sessions due to injury-related pain or discomfort (6). Although this research definition is often used when a medical diagnosis for injury is not available, time-loss injury definitions may not align well with definitions of injury based on pain, which is often cited as the primary symptom of running-related injury (7,8). When injuries are defined based on time lost from training, running volume will necessarily decrease during weeks in which a runner is suffering from an injury. If this lost running time is not fully replaced with other forms of physical activity (e.g., walking, cycling, using an elliptical machine), a runner’s overall physical activity level will drop substantially while injured.

Notably, the presence of running-related pain is not necessarily synonymous with injury under a definition requiring a change to planned running. Indeed, the prevalence of overuse-related pain among athletes can be very high, even when time lost from sport is relatively rare (8). If running-related pain limits a runner’s participation in nonrunning physical activity, his or her overall physical activity level could decrease, even in the absence of “injury” when defined by reductions or cancellations of planned running sessions. Thus, both running-related injury and running-related pain could independently lead to reductions in overall physical activity levels during the period of injury.

The potential for reductions in overall physical activity level during periods of injury are particularly concerning in light of recent findings suggesting that runners spend a substantial proportion of their nonrunning time being inactive. Whitfield et al. (9) report that marathon and half-marathon participants reported spending 480 to 645 min·d−1 sitting, and Rantalainen et al. (10) report that 32% of healthy joggers (defined as individuals who ran 20 to 40 km·wk−1 for at least the past 5 yr) do not meet World Health Organization guidelines for moderate to vigorous physical activity (MVPA), even while participating in their usual run training. Given the high incidence and long recovery time associated with injury (5), any reductions in physical activity associated with running-related injury or pain could pose a threat to long-term maintenance of an active lifestyle. Although the overall physical activity levels of runners have been characterized before, their behavior while experiencing running-related injury or running-related pain has not.

As such, the purpose of this study was to investigate whether recreational runners enrolled in a prospective cohort study reduce their physical activity levels when they suffer running-related injury or running-related pain. We hypothesized that runners would engage in less MVPA during weeks in which they reported a running-related injury or running-related pain, compared with weeks without any reported injury or pain.


Healthy recreational runners age 18 to 65 yr were recruited for a prospective cohort study with a maximum follow-up of 1 yr. We recruited a convenience sample of subjects via paper flyers posted on university property and the surrounding areas, advertisement on the School of Public Health website for research study recruitment, postings on a university-wide listserv, and word-of-mouth. Subjects were required to have participated in running training of at least 16 km·wk−1 for a minimum of 2 yr and no history of injury in the previous 3 months. Subjects who had ever experienced injury were required to have returned to their typical running mileage postinjury. Due to a separate portion of the study involving gait analysis, subjects were also required to have no history of surgery to the back or lower extremities. Upon enrollment, subjects underwent a three-dimensional kinetic and kinematic gait analysis (data presented elsewhere (11)) and were issued a Fitbit Charge (Fitbit, San Francisco, CA), a wrist-worn physical activity monitor that has been validated for measuring physical activity levels (12). Subjects were free to wear the activity monitor on either their dominant or nondominant wrist and each participant’s choice was recorded in his or her FitBit online profile.

The Fitbit tracker categorizes minute-to-minute activity as “sedentary activity,” “light activity,” “moderate activity,” and “very active activity.” According to previous validation studies of Fitbit devices (13,14), as well as information available on the Fitbit website (15), Fitbit uses metabolic equivalents-based cutpoints to determine thresholds for each category of activity. Although specific information on how these cutpoints are calculated is not available, levels of moderate to vigorous activity estimated by the Fitbit Charge are comparable to estimates from a research-grade activity monitor (12). Subjects were instructed to wear their physical activity monitor for all waking hours. Activity monitor data were regularly synchronized to a remote server via the Fitbit consumer smartphone app and were accessed by the research team using Fitabase software (Small Steps Lab, LLC, San Diego, CA). All subjects gave written informed consent before participating, and the study was approved by the university’s institutional review board.


The subjects recorded their running training, nonrunning exercise, and other physical activity in a survey delivered via email every week (Qualtrics, Provo, UT). Subjects received no guidance on training habits or physical activity from the research team and were free to modify or alter their planned running and exercise habits as desired.

Running-related injuries and pain were also recorded with the same weekly survey. The survey instructed subjects that they would be asked questions “about each painful or injured area that you had throughout the week.” The survey asked subjects to report “any pain, injury, or problem with your legs, feet, joints, pelvis, and/or back.” If subjects responded affirmatively to this question, they were asked to “Please rate the pain you are experiencing in this area on a scale from 0 (no pain) to 10 (worst pain imaginable)” and were provided with an 11-point scale ranging from 0 to 10. In the survey, the 0–10 scale was annotated as “no pain,” 0–1; “mild pain,” 2–3; “moderate pain,” 4–5; “severe,” 6–7; “very severe,” 8–9; and “worst pain imaginable,” 10 per standard pain scale markings (16,17). If a subject reported multiple injured or painful areas, the pain level recorded for that week was the maximum rating across the painful areas.

All subjects reporting running-related pain, injury, or problems were asked whether this issue had caused them to cancel or reduce any planned running sessions that week. Subjects were given four possible responses:

  1. No change in running habits and no leg, joint, or back problems
  2. No change in running habits, but experienced leg, joint, or back problems
  3. Reduction in running habits due to leg, joint, or back problems
  4. Did not participate in running due to leg, joint, or back problems

Subjects were also asked how many planned running sessions they reduced or canceled due to the problem. Following the consensus-based recommendations of Yamato et al. (6), subjects reporting a reduction or cancellation of three or more planned running sessions were categorized as injured for that week.

Thus, each week of follow-up, each subject recorded a pain rating from 0 to 10, and the number of days the pain caused a modification in planned running (if any). Because the injury status—as determined by the investigators following the definition of Yamato et al. (6)—and pain level of a subject were updated separately each week, the same subject could contribute observation time to many different pain levels and injury status combinations throughout his or her follow-up time.

Data analysis

Following the wear-day criteria of Chu et al. (18), days with less than 1500 steps recorded on the activity monitor were considered “nonwear days.” Only weeks with at least 4 d meeting the wear-day criteria were included in the data analysis, which follows protocols established previously (19). The influence of these wear-day criteria on the results were explored later in a sensitivity analysis. Moderate to vigorous physical activity was calculated as the number of total minutes spent together in “moderate” and “very active” activity levels, as assessed by the activity monitor, averaged across all wear days for the week.

To mitigate the effects of intermittent follow-up, such as long periods of survey nonresponse punctuated by occasional responses, we identified subjects with less than 75% survey completion compliance during the period between their first completed and last completed weekly survey. The final survey submitted by each of these subjects was iteratively excluded (right-censored) until that subject’s compliance during their follow-up period (i.e., proportion of surveys received from first survey until right-censoring) was 75% or greater. All data were processed in MATLAB (version R2018a; The Mathworks, Inc., Natick, MA).

Statistical analysis

Because the primary purpose of this observational cohort study was to examine the emergence of injury in healthy runners, a sample size of ≥38 and follow-up time of ≥6 months were necessary so that at least 10 injuries were likely to emerge over the course of the study, which simulation studies have demonstrated is an acceptable minimum number of events for univariate analysis (20,21).

Changes in MVPA were modeled using a mixed-effects linear model, which accounts for the differential follow-up and repeated-measure nature of the data (22). Injury status (yes/no) and weekly pain level (0 to 10) were modeled as fixed effects with random slopes and random intercepts, with daily average MVPA as the dependent variable. All statistical analyses were conducted in R (version 3.5.1, R Core Team, Vienna, Austria) using the “lme4” package with restricted maximum likelihood estimation (23,24). The 95% confidence intervals (95% CI) for all model parameters were estimated using parametric bootstrapping with 10,000 replications. Changes to specific levels of physical activity were assessed using four separate models identical to the primary model described above, but setting the dependent variable to daily average minutes spent in sedentary, light, moderate, or vigorous activity.

Sensitivity analysis

To test whether the findings were unduly influenced by any particular subject, a leave-one-subject-out strategy was used. For N subjects, the primary analysis was repeated on a subset of the data including only the N − 1 other subjects. This procedure was repeated N times to generate N estimates for the fixed effects. The mean and standard deviation of these fixed effects were measured to assess whether any individual subjects had a disproportionate effect on the results of the primary analysis. To test the influence of the wear-day criteria on the primary results, the primary analysis was replicated on activity monitor data processed following the protocol of Paul et al. (25), which used no wear-time validation. To test whether nonwear days were more common during periods of injury, a logistic mixed-effects model was used to assess whether suffering injury or reporting pain increased the odds of recording less than seven wear days.


Fifty-five subjects qualified for enrollment. Subjects were excluded for completing zero weekly surveys (n = 1), having an unreported preexisting injury that was disclosed after enrollment (n = 2), or uploading no activity monitor data for any weeks of follow-up (n = 3). The iterative right-censoring procedure to ensure ≥75% compliance from all subjects resulted in the exclusion of a total of 32 weekly surveys from 10 subjects (approximately 2.1% of all surveys received). After applying all exclusion criteria, 49 subjects (30 women, 19 men) contributed 1460 person-weeks to the analysis. Mean follow-up time was 35.6 wk (SD, ±19.7) across all 49 subjects. Subject demographics are summarized in Table 1.

Subject characteristics at time of enrollment.

Follow-up and outcomes

Across all days of follow-up (including those in which subjects reported injury or pain), subjects averaged 58.6 min of MVPA (SD ±27.2). Twenty-three (46.9%) subjects reported at least 1 wk of injury. Four (8.16%) subjects did not meet American College of Sports Medicine recommendations of 150 min of MVPA per week on average. The prevalence of injury and the prevalence of pain were nonzero for 51 of 52 wk of follow-up (Fig. 1). That is, at least one subject reported injury and at least one subject reported pain for 51 of the 52 wk of follow-up.

Left: Week-by-week prevalence of injury, defined as a cancellation or reduction in at least three planned training sessions during a given week (6). Right: Week-by-week prevalence of pain rated on a 0–10 scale, categorized for display purposes here as mild (2–3/10), moderate (4–5/10) and severe (≥6/10) pain.

Injury, pain, and MVPA

After controlling for pain level, runners engaged in 14.1 fewer minutes of MVPA per day when they reported injury, compared with uninjured weeks (Table 2). There was no significant association between reported pain level and MVPA, independent of injury status (P = 0.300). Pain was prevalent during weeks in which subjects did and did not meet the criteria for being injured; reporting an injury often coincided with reporting little or no pain and vice versa (Fig. 2). Decreases in MVPA during injury weeks were primarily the result of substituting moderate and vigorous activity with sedentary activity (Fig. 3).

Mixed-effect model coefficients.
The distribution of weekly survey reports of pain levels (0–10 scale) during “injured weeks” (left) and “uninjured weeks” (right). Count (vertical axis) refers to the total number of person weeks accumulated during the follow-up period at each level of pain, for weeks classified as “injured weeks” and weeks classified as “uninjured weeks.” Injury status (as determined by the investigators, using the definition of Yamato et al. (6)) and pain level were updated weekly for each subject. Injury status was determined based on whether any running-related issue led to a cancellation or reduction in at least three planned training sessions during a given week (6).
Changes in minutes spent in sedentary, light, moderate, and vigorous activity during injured weeks, compared with weeks that had no pain and no injury reported. Positive values represent an increase in activity and negative values represent a decrease in activity relative to weeks with no pain or injury. The decrease in MVPA is mostly attributable to an increase in time spent in sedentary activity. Error bars represent 95% CI.

Sensitivity analysis

Using no wear-time validation resulted in only minor changes in the model results. In the alternative model, sustaining injury was associated with 11.6 fewer minutes of MVPA per day, versus 14.1 fewer minutes in the primary analysis. Each one-unit increase in pain was associated with a nonsignificant 0.168 additional minutes of MVPA per day in the alternative model, compared with a nonsignificant 0.703 additional minutes in the primary analysis (see Table, Supplemental Digital Content 1: sensitivity analysis results, Leave-one-subject-out analysis indicated that the estimate for the fixed effect of injury and pain varied (±1 SD) by 0.807 and 0.090 min·d−1 per unit change, respectively, across 49 models. Based on this result, we concluded that our findings could not be attributed to the effects of any unduly influential subjects. Failing to achieve seven full wear days was more likely during injury weeks, but not to a statistically significant extent (odds ratio, 1.811; 95% CI, 0.648 to 3.747; see Table, Supplemental Digital Content 2: nonwear day analysis results,


The purpose of this study was to investigate whether injured runners replace lost running time with other forms of MVPA, and to investigate whether running-related pain, independent of the existence of a formally defined injury, would influence daily MVPA. We found that runners who were currently injured—as defined by a reduction or cancellation of at least three planned training sessions in a week (6)—did not fully replace their lost running time with other forms of MVPA. Pain level, however, did not have a significant association with MVPA, possibly because runners elected to run through pain in order to maintain their typical training schedule.

Our results suggest that running-related pain alone is not a good indicator of reductions in a runner’s overall physical activity level. We found that low to moderate levels of pain (ratings of 2–6/10) occurred frequently in weeks that did not meet the formal definition of injury (Fig. 2). Given our definition of injury, and given that measured MVPA must have included both running and nonrunning activity, a reduction in MVPA in weeks with pain but no injury could have been achieved by reducing nonrunning activity but maintaining planned running behavior. However, MVPA was unchanged in these uninjured but painful weeks, indicating that runners chose to continue with both planned running sessions and nonrunning exercise, despite experiencing pain. Even during weeks that were not categorized as “injury weeks,” low to moderate levels of pain (ratings of 2–6/10) occurred frequently (Fig. 2). Maintaining both nonrunning and running activity while experiencing pain could potentially lead to altered running mechanics, more severe pain, or clinical injury, but future work is needed to confirm these relationships. Though self-reported pain level is often used to assess the severity of sports injuries (8), this metric is not without limitations. Our subjects reported their pain level on a weekly basis, but running-related pain may fluctuate on a day-to-day or hour-to-hour basis. Though we found no association between pain level and physical activity, assessing pain more often (e.g., multiple times per day) might uncover a relationship between acute pain levels and physical activity behavior.

The exchange of moderate and vigorous activity time for sedentary time during injured weeks in this study suggests that running represents the primary source of MVPA for most runners. Despite the injury-related reduction in MVPA observed among our subjects, absolute levels of physical activity were still high among our subjects (i.e., greater than 30 min·d−1 on average), even while injured. This observation may be explained in part by the definition of injury used in this study and others (6), that an “injured” week did not necessarily correspond to a complete cessation of all running. Even so, replacing relatively small amounts of MVPA with sedentary behavior is still associated with worse long-term health outcomes—von Rosen et al. (26) estimate that substituting 20 min of MVPA per day with 20 min of sedentary behavior is associated with a 26% increase in hazard for all-cause mortality over 15 yr. Given that novice and recreational runners exhibit high levels of sedentary activity to begin with (9,10), injury-related reductions in MVPA may represent an obstacle to maintaining an active lifestyle, particularly if the injury is prolonged or if these sedentary habits continue after the injury has resolved.

Although injured runners are, by definition, unable to engage in all planned running sessions because of pain, further research is needed to explain why injured runners do not replace lost running time with other forms of physical activity. Injured runners may find other forms of physical activity less enjoyable than running, or they may not have access to equipment or facilities required for other forms of exercise. Determining these or other possible barriers to physical activity for injured runners, as well as other recreationally active people prone to exercise-related injury, could lead to interventions to help maintain high levels of overall physical activity.

Some factors could have introduced bias or limit the generalizability of the findings of this study. Our study population was a convenience sample of recreational runners. Although we were able to recruit runners across a range of age, body mass index, and years of running experience, our study population was not a random sample of the broader pool of recreational runners. The longitudinal design of our study and the use of email-distributed surveys and wearable activity monitors allowed us to assess physical activity habits at a high level of detail, but runners who are willing to participate in this type of longitudinal study may not be representative of runners as a whole. Thus, our findings may not generalize to all segments of the running population. However, this sampling method is consistent with other prospective cohort studies of recreational runners, which have also used convenience sampling (27,28).

Most subjects in this study met published guidelines for physical activity, and all had at least 2 yr of running experience. As such, these findings may not apply equally to novice runners. Because not all runners were followed up for the entire year of observation, bias could have been introduced if physical activity behaviors in response to injury influenced the likelihood that a subject would drop out of the study early. It is also possible that subjects in our study engaged in other forms of MVPA that was not registered on the activity monitor, either because the monitor was not worn, or because the activity (e.g., riding an indoor training bicycle) did not generate sufficient movement to register as moderate or vigorous activity on the activity monitor. Subjects may have removed their activity monitor to participate in different types of nonrunning physical activity, such as swimming or deep-water running, so these activities would not be recorded in the activity monitor data. However, the results of our sensitivity analyses provide evidence that wear time was not significantly influenced by injury, and that individual subjects did not unduly influence the overall results, suggesting that potential limitations of follow-up time or activity detection did not threaten the validity of the study.

Using an activity monitor to objectively assess sedentary, light, moderate, and vigorous physical activity allowed us to circumvent the limitations of self-report assessments of physical activity, which include low reliability and poor agreement with objective measures of activity level (29). However, the activity monitor used in this study cannot determine the types of activities that the user is engaging in. As a result, we cannot report what kinds of sedentary or nonrunning exercise behavior our subjects were engaged in while injured or uninjured. Future research using more advanced activity monitors, or a combination of objective and self-report data, will be needed to identify how runners alter the specific activities they engage in while injured.


Healthy runners engage in more sedentary activity after suffering an injury, instead of replacing their lost running time with other forms of moderate or vigorous activity. Pain by itself is not associated with reductions in MVPA. The findings of this study suggest that researchers should not conflate injury status and injury-related pain, as both are not equally indicative of reductions in physical activity level.

The authors would like to thank Shane P. Murphy, Jacob E. Vollmar, Ashley B. Nguyen, and Emily G. Wagoner for their assistance collecting and organizing the survey data. The authors are grateful to Dr. Paul DeVita and Dr. Andrew W. Brown for providing feedback on the manuscript prior to submission. This study was supported by the Indiana University School of Public Health Bloomington Faculty Research Grant Program.

This study was supported by the Indiana University School of Public Health Bloomington Faculty Research Grant Program. The authors declare no conflicts of interest. The results of this study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results and views of this study do not constitute endorsement by the American College of Sports Medicine.


1. Lee DC, Pate RR, Lavie CJ, Sui X, Church TS, Blair SN. Leisure-time running reduces all-cause and cardiovascular mortality risk. J Am Coll Cardiol. 2014;64(5):472–81.
2. Chakravarty EF, Hubert HB, Lingala VB, Fries JF. Reduced disability and mortality among aging runners: a 21-year longitudinal study. Arch Intern Med. 2008;168(15):1638–46.
3. Ham SA, Kruger J, Tudor-Locke C. Participation by US adults in sports, exercise, and recreational physical activities. J Phys Act Health. 2009;6(1):6–14.
4. Walter SD, Hart LE, McIntosh JM, Sutton JR. The Ontario cohort study of running-related injuries. Arch Intern Med. 1989;149(11):2561–4.
5. Nielsen RO, Rønnow L, Rasmussen S, Lind M. A prospective study on time to recovery in 254 injured novice runners. PLoS One. 2014;9(6):e99877.
6. Yamato TP, Saragiotto BT, Lopes AD. A consensus definition of running-related injury in recreational runners: a modified Delphi approach. J Orthop Sports Phys Ther. 2015;45(5):375–80.
7. Ballas M, Tytko J, Cookson D. Common overuse running injuries: diagnosis and management. Am Fam Physician. 1997;55(7):2473–84.
8. Bahr R. No injuries, but plenty of pain? On the methodology for recording overuse symptoms in sports. Br J Sports Med. 2009;43(13):966–72.
9. Whitfield G, Gabriel KKP, Kohl HW III. Sedentary and active: self-reported sitting time among marathon and half-marathon participants. J Phys Act Health. 2014;11(1):165–72.
10. Rantalainen T, Pesola AJ, Quittner M, Ridgers ND, Belavy DL. Are habitual runners physically inactive? J Sports Sci. 2018;36(16):1793–800.
11. Gruber AH, Murphy SP, Vollmar JE, Kennedy-Armbruster C, Chomistek AK. Does non-running physical activity contribute to the risk of developing a running related overuse injury? Med Sci Sports Exerc. 2016;5S(181):48.
12. Hargens TA, Deyarmin KN, Snyder KM, Mihalik AG, Sharpe LE. Comparison of wrist-worn and hip-worn activity monitors under free-living conditions. J Med Eng Technol. 2017;41(3):200–7.
13. Van Blarigan EL, Kenfield SA, Tantum L, Cadmus-Bertram LA, Carroll PR, Chan JM. The Fitbit One physical activity tracker in men with prostate cancer: validation study. JMIR Cancer. 2017;3(1):e5.
14. Alharbi M, Bauman A, Neubeck L. Validation of Fitbit-Flex as a measure of free-living physical activity in a community-based phase III cardiac rehabilitation population. Eur J Prev Cardiol. 2016;23(14):1476–85.
15. Fitbit [Internet]. What are active minutes?. 2019. [cited 2019 Oct 3]. Available from:
16. Balille L. A review of pain assessment tools. Nurs Stand. 1993;2(7):25–9.
17. Ho K, Spence J, Murphy MF. Review of pain-measurement tools. Ann Emerg Med. 1996;27(4):427–32.
18. Chu AH, Ng SH, Paknezhad M, et al. Comparison of wrist-worn Fitbit flex and waist-worn ActiGraph for measuring steps in free-living adults. PLoS One. 2017;12(2):e0172535.
19. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181.
20. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373–9.
21. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and cox regression. Am J Epidemiol. 2007;165(6):710–8.
22. Pinheiro J, Bates D. Mixed-Effects Models in S and S-plus: Springer Science & Business Media. 2006;25–7.
23. Bates D, Maechler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67(1):1–48.
24. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2018.
25. Paul SS, Tiedemann A, Hassett LM, et al. Validity of the Fitbit activity tracker for measuring steps in community-dwelling older adults. BMJ Open Sport Exerc Med. 2015;1(1):e000013.
26. von Rosen P, Dohrn I, Hagströmer M. Association between physical activity and all-cause mortality: a 15-year follow-up using a compositional data analysis. Scand J Med Sci Sports. 2020;30(1):100–7.
27. Messier SP, Martin DF, Mihalko SL, et al. A 2-year prospective cohort study of overuse running injuries. The runners and injury longitudinal study (TRAILS). Am J Sports Med. 2018;46(9):2211–21.
28. Damsted C, Parner ET, Sørensen H, Malisoux L, Nielsen RO. Design of ProjectRun21: a 14-week prospective cohort study of the influence of running experience and running pace on running-related injury in half-marathoners. Inj Epidemiol. 2017;4(1):30.
29. Shephard RJ. Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med. 2003;37(3):197–206.


Supplemental Digital Content

Copyright © 2019 by the American College of Sports Medicine