Article Text

Download PDFPDF
Medical encounters at community-based physical activity events (parkrun) in the UK
  1. Charles R Pedlar1,2,
  2. Kyriaki Myrissa1,
  3. Megan Barry1,
  4. Iman G Khwaja1,
  5. Andrew J Simpkin3,
  6. John Newell3,
  7. Carl Scarrott3,
  8. Greg P Whyte4,
  9. Courtney Kipps2,
  10. Aaron L Baggish5,6
  1. 1 Faculty of Sport and Applied Performance Science, St Mary's University Twickenham, Twickenham, UK
  2. 2 Institute of Sport, Exercise and Health, University College London, London, UK
  3. 3 School of Mathematics, Statistics and Applied Mathematics, NUI Galway, Galway, Ireland
  4. 4 Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, Liverpool, UK
  5. 5 Harvard Medical School, Boston, Massachusetts, USA
  6. 6 Cardiovascular Performance Program, Massachusetts General Hospital, Boston, Massachusetts, USA
  1. Correspondence to Professor Charles R Pedlar, Faculty of Sport, Allied Health and Performance Science, St Mary's University Twickenham, Twickenham, UK; pedlarc{at}stmarys.ac.uk

Abstract

Objective To determine the incidence, clinical correlates and exposure risk of medical encounters during community-based physical activity events in the UK.

Methods An analysis of medical data from weekly, community-based physical activity events (parkrun) at 702 UK locations over a 6-year period (29 476 294 participations between 2014 and 2019) was conducted in order to define the incidence and clinical correlates of serious life-threatening, non-life-threatening and fatal medical encounters.

Results 84 serious life-threatening encounters (overall incidence rate=0.26/100 000 participations) occurred including 18 fatalities (0.056/100 000 participations). Statistical modelling revealed that the probabilities of serious life-threatening encounters were exceptionally low, however, male sex, increasing age, slower personal best parkrun time and less prior running engagement/experience (average number of runs per year and number of years as a parkrun participant) were associated with increased probability of serious life-threatening encounters. These were largely accounted for by cardiac arrest (48/84, 57%) and acute coronary syndromes (20/84, 24%). Non-life-threatening medical encounters were mainly attributed to tripping or falling, with a reported incidence of 39.2/100 000 participations.

Conclusions Serious life-threatening and fatal medical encounters associated with parkrun participation are extremely rare. In the context of a global public health crisis due to inactivity, this finding underscores the safety and corollary public health value of community running/walking events as a strategy to promote physical activity.

  • exercise
  • public health
  • death
  • running
  • cardiology

Data availability statement

Data are available on reasonable request. Following publication, data analysed in the present study will be made available to other research groups subject to approval of a proposal and a signed data access agreement.

Statistics from Altmetric.com

Introduction

The health benefits of routine physical activity (PA), including reductions in all-cause mortality,1–3 cardiovascular disease4 and cancer,5 are well established. Accordingly, PA guidelines recommending weekly target doses of PA have been promoted for disease prevention by numerous national and global organisations.6–8 However, the majority of adults in developed countries worldwide fail to meet recommended PA targets.9 Inactivity has emerged as a public health crisis with attendant clinical and financial implications,10 and historically, strategies aimed at increasing PA have largely been ineffective and inadequate.

Strategies to improve adherence to PA recommendations including public policy,11 purposeful design of the built environment,12 13 and cognitive–behavioural incentives,14 have proven effective in some situations but have far from eliminated the problem. Mass endurance sporting events provide opportunities for PA, however, the financial cost, logistical challenges and recommendations for on-site medical infrastructure associated with these events preclude most communities from supporting an adequate number of events to serve as an effective platform for routine PA. A network of free, weekly exercise events (parkrun) that began in the UK, has emerged as a unique exception (http://www.parkrun.org.uk). Parkrun enables communities to hold weekly 5 km run/walks over a measured course in public venues. In addition, ‘junior’ parkrun events of 2 km distance are operated from some locations for individuals aged 4–14 years. In aggregate, parkrun has developed into the largest community-based exercise initiative worldwide.

Preliminary data emerging from the parkrun experience suggest that participants gain improvements in key determinants of health and longevity15 and parkrun has been designated a social prescribing activity in the UK by the Royal College of General Practitioners for the promotion of health and well-being.16 However, the well-established association between acute bouts of exercise and increased risk of adverse medical outcomes underscores the need to assess the risks,4 17 in parallel with benefits, of the parkrun experience. Accordingly, we performed a comprehensive assessment of medical encounters recorded during a recent 6-year period in the UK with a primary goal of determining the incidence, clinical correlates and exposure risk (including the identification of higher risk subgroups) of medical encounters occurring during parkrun participation. The term ‘encounters’ is deliberately used since it recognises that a proportion of medical problems occurring at parkrun events may not be formally recorded.18 As an a priori secondary aim, we also sought to characterise less serious medical encounters defined broadly as not representing a threat to life.

Methods

Study design

We approached these synergistic aims by collaborating with the parkrun organisation to extract and analyse data from their centralised medical encounter data repository as described below. While the first official parkrun was held in 2004, a standardised medical encounter reporting system was not implemented by the central sponsoring organisation until April 2013 and the use of this system became mandatory in 2014. As such, we confined our analyses to the entire population of participations and medical encounters during the 6-year time period spanning 2014–2019.

Medical encounter data capture

Every parkrun is coordinated by a volunteer run director, supported by a team of volunteer marshalls. During the study period, run directors (or a substitute volunteer if the run director is not present) were required to describe any incident including medical encounters or to attest to an incident-free event on the day of event completion. This process was enabled by the use of a simple web-based incident reporting tool designed for use among run directors with no formal medical training (https://volunteer.parkrun.com/principles/incidents accessed April 2021). Compliance with this mandatory reporting system was maximised by the use of an automated system that withheld the posting of participant event results on the parkrun website until receipt of the incident data.

Serious life-threatening medical encounters

The central parkrun team reviewed all medical encounter data from all parkrun events held in the UK, assigning a numeric severity score ranging from ‘1’ (least serious) to ‘5’ (most serious) as defined in online supplemental table S1. All encounters assigned a score 4 or 5 were followed up by direct correspondence with the corollary event director. The initial objective of this correspondence was to confirm the accuracy of the assigned encounter severity score. Following this confirmation, additional data (see online supplemental material S2) were collected to further characterise each serious life-threatening encounter, defined in this analysis as events with a final adjudication score of 4 or 5.

Supplemental material

Non-life-threatening medical encounters

Non-life-threatening medical encounters were characterised by performing a comprehensive independent review of the centralised parkrun data repository within the first 4 years of the study period (2014 through 2017). The primary goal of this review process was to confirm the incidence, aetiology and disposition of all medical encounters characterised by a severity score of 1–3. This was necessary as the centralised parkrun data repository did not routinely assign these attributes for low severity encounters in a fashion analogous to its handling of serious life-threatening encounters. Two trained research assistants used all available data to assign each medical encounter an aetiology or mechanism of pathology (see online supplemental material S3).

Statistical analysis

Incidence rates for the total number of serious life-threatening and fatal encounters were calculated as the simple proportion of encounters divided by the total number of parkrun participations within the 6-year period of study, with corresponding 95% CIs where appropriate. The participations data included all finishers and the non-finishing encounters, but not all the starters since we did not have information on other non-finishers. This approach leads to a conservative overestimation of risk, since the numerator is maximised and the denominator is minimised. For the purposes of this paper, junior participants are defined as those ≤14 years old in both the 2km and 5km events. All junior data were excluded from the statistical analysis since the risk factors are not comparable.

Comparisons of age (years), personal best parkrun time (minutes) and parkrun experience (completed parkruns), between those who survived serious medical incidents and those that died were made using the non-parametric Wilcoxon rank-sum test. Poisson and logistic regression were used in the exploratory data analysis to model the changes in the aggregated yearly counts and rates of medical encounters respectively to provide simple unadjusted measures of risk. A χ2 test was used to compare the proportions of deaths and serious events between males and females using a simulation based p value due to the low counts.

A detailed multiple logistic regression encompassing data from all participants was used to quantify the impact of the risk factors of sex, age, personal best parkrun time (as a surrogate for running ability), average number of parkruns per year (as a surrogate for running consistency) and number of years as a parkrun participant (as a surrogate for running experience) on the risk of serious event occurrence. The binomial response data are the counts of each athlete’s complete record of parkruns and any serious encounters during the exposure period of 2014–2019. ORs are reported for each factor except age, which was included as a smooth function (using thin plate splines) due to its nonlinear relationship with the outcome. The 95% CIs for the OR were calculated by exponentiation of the standard Wald CI for estimated coefficient from the logistic regression. The OR was approximately equal to the relative risk, since the encounters were rare events. The data were cleaned prior to modelling by removing all junior parkrun data (n=5,115,301)) and parkrun times where no or unreliable data were available (n=1 940 322 participations; see online supplemental material S4). All analyses were performed using R (V.4.0.3). A p<0.05 was considered significant. The logistic modelling of this large dataset was made feasible using a highly memory efficient implementation of generalised additive models, incorporating efficient smoothing parameter estimation and generalised cross-validation to choosing the smoother complexity.19

Results

Parkrun demographics

Weekly 5 km parkrun events were held in 702 distinct locations within the UK during the 6-year study period (figure 1A). Junior parkrun events were held in 341 of these locations and were also typically held on a weekly basis. There were consistent increases in the number of male and female participants with a cumulative 6 years total of 2 090 129 participants (figure 1B, table 1) accounting for a total of 29 476 294 participations. There were 2 829 827 participations in junior 2 km parkruns (figure 1C) which constitute 8.8% of the 32 306 118 total participations across 2 km and 5 km parkruns. The mean (SD) number of 5 km parkrun participations per participant each year was 7.9 (10.1) for males and 6.4 (8.6) for females. This trend was paralleled by similar growth among all global parkrun events (global 6-year total: 50 385 925 recorded parkrun participations from 3 755 478 participants). Excluding junior data, the UK-based participations were 57% male, but since females tended to do less runs per athlete on average, the proportion of unique male athletes was lower at 48%. The average age per participation was 41 (12.5) years but, since the older runners tended to do more runs, the average age per unique athlete was lower at 36 (12.4) years. The median parkrun completion times among males and females were 25:20 (minute:second; IQR: 22:34, 28:46) and 31:00 (27:45,35:00), respectively (figure 1D; table 1).

Figure 1

The parkrun story (6-year time period, 2014–2019; juniors and volunteers removed as appropriate). (A) Map of the 702 UK parkrun events; (B) number of parkrun individual participants by year and sex; (C) number of parkrun participations by year; (D) distribution of finish times by sex; (E) Finish time distributions by year and sex.

Table 1

Descriptive information on overall UK 5 km parkrun data (702 locations) and as stratified by participant sex during the 6-year time period (2014–2019)

The 5 km parkrun events conducted early in the study sample period were dominated by faster club runners but gradual diversification occurred over time, characterised by increasing participation among recreational and older joggers and walkers, reflected in a gradual slowing of mean 5 km completion times which is most notable in the extending upper tail of the finish time distributions (figure 1E). In addition, 2 688 473 million volunteer instances accounted for the marshalling staff at events over the 6-year study period from 280 163 registered volunteers. The mean number of volunteer event participations per volunteer participant was 9.5 (21.2).

Serious life-threatening encounters

There were 84 serious life-threatening encounters, leading to an estimate of the probability per participation of all 2 km and 5 km parkrun events of 0.26/100 000 participations (95% CI 0.20 to 0.32 per 100 000) during the 6-year study period. The aetiological breakdown of all serious life-threatening encounters is shown in figure 2. Fifty-seven per cent (48/84) of these encounters were attributable to cardiac arrest among which 45/48 (93.8%) met criteria for defibrillation (electric shock therapy) from an automated external defibrillator. The survival rate among all cardiac arrest victims and among those known to have received defibrillation were 65% (31/48) and 69% (31/45), respectively. Twenty-four per cent (20/84) of serious encounters were attributed to acute coronary syndrome and 6% (5/84) of serious encounters were attributed to life-threatening arrythmia that was not attributed to myocardial ischemia. Cerebrovascular accident, inclusive of both haemorrhagic and ischaemic stroke, accounted for 8/84 (10%) of cases. The remaining cases comprised: traumatic C-spine injury (n=1) and primary respiratory failure (n=2).

Figure 2

Aetiological breakdown of all serious life-threatening medical encounters (n=84) over 6 years (2014–2019) including junior (2 km) and 5 km parkrun events.

Excluding cases among junior parkrun participants (n=2) the rate of serious life-threatening encounters was higher at 0.32 per 100 000 participations (95% CI 0.25 to 0.40 per 100 000), confirming the resulting risk estimates are conservative. Among the serious life-threatening encounters, the majority occurred in men (67/82=82%; 95% CI 71% to 89%; figure 3A) and in participants ≥45 years of age (68/82=83%, 95% CI 75% to 91%; figure 3B). A total of 18/82 serious life-threatening encounters resulted in death (16 males, 2 females) yielding a rate of 0.071 deaths/100 000 participations (95% CI 0.038 to 0.104 per 100 000). Thus, the mean fatality rate associated with serious life-threatening encounters was 22% (annual range 14%–60%) and did not change significantly over time (figure 3C, p=0.25).

Figure 3

Serious life-threatening medical encounters and survival status at 5 km parkrun events across the 2014–2019 time period (A) by sex; (B) by age; (C) absolute count; (D) /100 000 participations; (E) by personal best parkrun time; (F) by parkrun experience (number of parkruns completed per participant). Box plots represent median and IQR.

The change in the annual absolute number of deaths over time was not significant (p=0.17) neither was the rate of fatalities per parkrun participation over time (p=0.83). The rate of serious life-threatening medical encounters per participation did increase over time (in figure 3D, p=0.02) and there was a significant increase in the annual number of serious encounters (p<0.001).

Parkrun personal best completion times were significantly slower in the fatality group compared with the survivors of serious encounters (p=0.007; figure 3E). There were no significant differences in age between survivors of serious life-threatening encounters and fatal life-threatening encounters (p=0.70; figure 3B). The participation distribution, as measured by the number of parkruns completed, was notably lower for those that had a fatal life-threatening encounter just reaching statistical significance (p=0.02; figure 3F).

The results of the advanced logistic regression model to evaluate impacts of the risk factors (excludes junior parkrun serious incidents; n=2) are presented in figure 4 and table 2. Participant age category, sex, personal best (as a surrogate for running ability and fitness level), average number of runs (as a surrogate for regular exercise) and number of years as a parkrun participant during 2014–2019 (as a surrogate for running experience) were identified as having a significant effect on the risk of a serious medical encounter. The risk was higher for males compared with females (OR 5.02; 95% CI 2.73 to 9.20; p<0.0001) and for slower athletes (OR 1.07 per minute extra personal best; 95% CI 1.03 to 1.11; p<0.0001) and decreased with increasing parkrun participation (OR 0.97; 95% CI 0.95 to 0.99; p=0.02) and number of years participating (OR 0.82; 95% CI 0.70 to 0.95; p=0.01). Insufficient evidence was found for non-linear effects of the personal best, average number of runs or number of years as a parkrun participant on the risk. There was insufficient evidence of any interaction between these risk factors.

Figure 4

Example plots of the athlete specific estimated risk per participation at UK parkruns by personal best 5 km run time (PB), age, sex and parkrun participation experience (number of runs per year and number of years of parkrun participation).

Table 2

Estimated coefficients and ORs for the risk of a serious life-threatening medical encounter at UK parkrun events

The magnitude of the reported ORs must be interpreted with care given the very low risk of serious events in general. A more informative summary is given by the plots of the (smooth) effect of age, sex, personal best, average number runs per year and years participating in parkrun on the probability of a serious event (figure 4) which translates the ORs into athlete specific estimated risk per participation. For example, the risk of experiencing a serious medical encounter, for a male runner aged 60–64 years, with a personal best of 40 min, who participated in an average of 1 parkrun event per year, in his first year of parkrun participation translates to 8.806 per 100 000 participations (95% CI 5.483 to 14.144) which is higher than for the equivalent female runner at 1.756 per 100 000 participations (95% CI 0.953 to 3.235). A personal best parkrun time of 40 min would place such a runner in the slowest 8% of this parkrun population. Notice that the ratio of the male:female probabilities is consistent with the OR of 5.02, since this risk ratio is approximately equal to the OR due to the events being rare.

By contrast, the same runners with more experience (6 years of parkrun) and participation (30 participations on average per year) have a lower risk of 1.457 (95% CI 0.699 to 3.038) for males and 0.290 (95% CI 0.133 to 0.634) for females per 100 000 participations. Further, similarly experienced and high participation runners but in the lower age group of 20–24 years old have a much lower risk of 0.054 (95% CI 0.014 to 0.209) for males and 0.011 (95% CI 0.003 to 0.040) for females per 100 000 participations. So even though the male runners are at 5.02 times the risk compared with females, this risk is very low for younger adult runners.

Non-life-threatening medical encounters

Among 19 638 709 parkrun total participations that occurred during the first 4 years of study, we identified 7492 non-life-threatening medical encounters which included 1563 (20.1%) encounters occurring at junior parkrun events and 145 (1.9%) encounters impacting volunteers rather than participants. Among runners and walkers only, the rate of encounters was 1 per 2552 parkrun participations or 0.392 encounters per 1000 participations (95% CI 0.381 to 0.398). The majority of all non-life-threatening encounters (94.2%) were minor (see table 3), however, 698 encounters resulted in a hospital visit (1 per 28 136 parkrun participations or 0.036 per 100 000 participations; 95% CI 0.033 to 0.038). 4965 (66.3%) of the non-life-threatening encounters were associated with a trip or a fall and 305 (4.1%) encounters were associated with a collision.

Table 3

Non-serious medical encounters (n=7492 during 19 638 709 parkrun participations over 4 years (2014–2018), including runners and volunteers

Discussion

Parkrun, a PA initiative that originated in the UK, now enables millions of individuals worldwide to participate in routine community-based exercise. The primary objective of this study was to determine the incidence, clinical correlates and outcomes of all reported medical encounters attributable to parkrun. The overall rates of serious life-threatening and fatal medical encounters (1 per 384 597 parkrun participations and 1 per 1 794 784 parkrun participations, respectively) are substantially lower than those documented among other exercising cohorts including male recreational habitual joggers (1 fatality/396 000 hours of jogging),20 competitive triathlon participants (1 fatality/39 151 male or female finishers)21 and recreational marathon runners (1 fatality/259 000 male or female finishers).22 Accordingly, the parkrun safety profile appears to be substantially better than those associated with other forms of organised sports and exercise. Further data for comparison were recently comprehensively reviewed and reported by Breslow et al.17

Routine moderate to vigorous PA reduces all-cause mortality, and PA recommendations have been published by numerous governing bodies.6–8 However, inadequate PA is a major public health crisis with attendant clinical and economic implications.10 Novel strategies that promote PA are a global health imperative,23 however, the development and implementation of exercise adherence initiatives requires consideration of both the benefits and the risks. Accurate determination of the net benefit of PA interventions like parkrun necessitate data delineating both their benefit and harm. It is well established that acute bouts of vigorous PA transiently increase the risk of myocardial infarction,24 and sudden cardiac death.25 26 As summarised above, the incidence of fatal medical encounters during organised sports and athletics has been documented in numerous settings. Findings from the current study, the first large scale initiative designed to quantify the risks associated with a community-based weekly 5 km running and walking programme, indicate that parkrun is associated with markedly lower encounter rates than previously reported in any prior cohorts.

Serious life-threatening medical encounters

Although extremely rare, several observations about serious life-threatening medical encounters during parkrun participation are noteworthy. First, male sex and increasing age were associated with increased risk of a serious medical encounter, and the majority of life-threatening encounters during parkrun participation were attributable to a cardiovascular aetiology. These observations are in line with prior studies.22 27 The number of, and rate of, serious encounters significantly increased over time and we speculate that this is due to the widening of the demographics of the participants, particularly the increasing proportion of older runners. Second, we observed a cardiac arrest survival rate of 65% in the context of near uniform utilisation of automated external defibrillators (mandatory at all UK parkrun events). This survival rate approximates prior reports among other sporting cohorts and is substantially better than reported out-of-hospital cardiac arrest survival rates worldwide (approximately 10%)28 underscoring the importance of defibrillator availability at community-based exercise events,. Third, less prior running experience, both in terms of the average number of parkrun participations per year (within the 6-year sample) and the number of years participating (within the 6-year sample) were associated with an increased risk of a serious medical encounter. Finally, we observed a low rate of serious life-threatening medical encounters attributed to heat illness. Compared with US data examining a popular summertime 10 kilometre run (1.6 heat illness encounters per 1000 finishers29 and Isreali data characterising Tel Aviv-based races of 10 km or greater distance (21 life-threatening heat illness cases among 137 580 runners,30 we observed a markedly lower incidence (<0.01 per 1000 parkrun participations) of serious life-threatening medical encounters related to heat. We speculate this discrepancy is attributable both to the comparatively short parkrun distance and the comparatively temperate UK climate.

Non-life-threatening medical encounters

This study also examined the incidence of non-life-threatening medical encounters finding an estimated incidence of 0.4 per 1000 parkrun participations. This event rate compares favourably with data derived from prior studies of 10 km races (6.2 per 1000 finishers),29 half marathon races (2.2 per 100 000 registered participants)27 and marathon races (18.9 per 1000 entrants).31 The majority of non-life-threatening medical encounters during parkrun were accounted for by soft tissue (ie, cuts, scrapes, abrasions, etc) and musculoskeletal injury (ie, sprains, joint dislocations and fractures). This aetiological profile is consistent with a prior report of medical encounters occurring during a half marathon distance event.27 While the incidence of injury during parkrun compares favourably to longer distance running events, our data suggest a valuable role for basic first aid services at these community running events.

Study implications

There are direct public health and clinical implications of the findings generated in this study. Emerging data suggest positive impacts of parkrun participation on both mental and physical health. Stevinson and Hickson) reported a net increase in weekly PA of 39 min and reduced body mass coupled with increased happiness and reduced psychosocial stress after 12 months of parkrun participation.15 32 Importantly, the beneficial effects of parkrun participation appear to be achievable by former non-runners and people with obesity and physical disability.

Preliminary data demonstrating the health benefits of parkrun participation coupled with the favourable safety profile documented in this study support the widespread use of parkrun as a safe and effective way to promote PA with its attendant health benefits. From a clinical perspective, findings from this study may inform the optimal provision of medical services at future parkrun and similar community-based events. At present, parkrun medical encounter recording and reporting requirements differ from recently published recommendations.18 In addition, there is no requirement for medically trained personnel to be present for a parkrun event to go ahead, or for those reporting incidents in parkrun’s incident database to be medically trained, which would be required to enable a greater degree of medical encounter reporting, closer to that described in a recent consensus statement.18 While our data underscore the importance of access to defibrillators and basic first aid, the relative scarcity of serious medical encounters at parkrun suggests that investment in a more comprehensive medical infrastructure is unnecessary. Future potential mandates for comprehensive onsite medical services at parkrun,18 or a requirement for preparticipation screening, may translate into a marked decline in the number of communities capable of hosting parkrun or deter individuals from participating. The important finding that a higher risk of a serious medical encounter exists in the older (notably over 60 years) age groups, and among slower, novice runners, warrants consideration of interventions to further reduce the rate of serious medical encounters at parkrun. One previous study demonstrated that intervening with targeted educational materials, and preparticipation clearance of higher risk runners from a medical practitioner, resulted in a reduction of all race-day medical encounters (by 29%) at a 21.1 km and 56 km running events in South Africa33 and a 64% reduction in serious life-threatening medical encounters or death. A similar strategy may be applicable at parkrun. However, it should be noted that a far higher rate of serious life-threatening medical encounters or death was reported for these longer events (56 per 100 000 starters prior to the intervention, reducing to 21 after the intervention, compared with 0.32 per 100 000 in the present study). Regardless, the creation and implementation of specific strategies designed to reduce the risk of adverse events among parkrun participants represents a logical and important area of future work.

Limitations

While this study represents a unique large-scale analysis of a community-based exercise programme with capture of both medical encounters and total participation numbers, several potential limitations are noteworthy. First, we acknowledge that the parkrun database, a resource that requires medical encounter reporting by volunteer event organisers, may have failed to capture all relevant medical encounters. However, the impact of under reporting is anticipated to be minimal based on the mandatory reporting system implemented by parkrun, and the risk assessments and incident reporting typically required in public parks or recreational areas that host these events in the UK. Next, we did not have access to any pre-existing health conditions data collected at parkrun registration that would potentially have provided further insights into the risk of a serious medical encounter. Similarly, we did not have access to hospital records documenting the care of participants that experienced life-threatening events thereby introducing the possibility of aetiology misclassification. However, the parkrun system requires detailed follow-up of all life-threatening events thereby reducing the likelihood of inaccurate event causality classification. The risk estimates result from an observational study, so care must be taken if generalising to a wider population. Finally, data presented in this study characterise the medical risk of parkrun participation prior to the emergence of the COVID-19 pandemic. Continued acquisition and analyses of medical event data in the wake of this viral pandemic represents an important area of future work.

Conclusion

Participation in 5 km community running and walking (parkrun) is associated with an exceptionally low incidence of serious life-threatening and fatal medical events. Accordingly, this rapidly growing mass-participation community-based exercise programme appears to be a safe and viable means of promoting PA on a large scale. The higher incidence among male, older, slower and less experienced parkrun participants warrants consideration for preventative strategies. Future work will be required to fully delineate the overall positive health impacts of parkrun in the context of a global physical inactivity health crisis.

What are the findings?

  • The present study advances our understanding of the medical risk of exercise. The study is unique in several ways: (1) the nature and location of the exercise involved, (2) the wide age range and near equal sex/gender distribution of the participants and (3) the size of the cohort studied.

  • The study demonstrates that the risk of serious life-threatening medical encounters at 5 km and 2 km community running/walking events is very rare and the risk of death is extremely rare.

  • A higher probability of a serious life-threatening medical encounter exists in males vs females, with increasing age, slower parkrun finish time, with less parkrun engagement/experience.

How might it impact on clinical practice in the future?

  • Results from this study underscore the important role of on-site Automated External Defibrillators at mass-participation community-based 5 km physical activity events such as parkrun, but suggest that a medical infrastructure beyond basic first aid provision appears to be of limited value.

Data availability statement

Data are available on reasonable request. Following publication, data analysed in the present study will be made available to other research groups subject to approval of a proposal and a signed data access agreement.

Ethics statements

Patient consent for publication

Ethics approval

St Mary's University Twickenham ethics committee. Reference number: SMEC_2018-19_059.

Acknowledgments

Thanks to Tom Williams, Mike Graney, Mark Hetherington, Professor Steve Haake and Ian Rutson for facilitating access to parkrun data.

References

Footnotes

  • Twitter @pedlarcr, @DrMyrissa

  • Contributors CP, AB, CK and GPW conceptualised the research study aims and methodology; CP, KM, IK, MB, CK and AB curated, reviewed and categorised all data provided by parkrun; JN, CS and AS conducted the formal statistical analysis; CP, AB, JN and CS wrote the manuscript; All authors reviewed and approved the final manuscript.

  • Funding A grant was awarded by parkrun (UK-based charity number 1175062) to support effort of the lead author and two research assistants. However, the study design, analyses and conclusions presented in this manuscript were conducted independently of the parkrun organisation.

  • Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.