Where should new parkrun events be located? Modelling the potential impact of 200 new events on socio-economic inequalities in access and participation.

Backgroundparkrun, an international movement which organises free weekly 5km running events, has been widely praised for encouraging inactive individuals to participate in physical activity. Recently, parkrun received funding to establish 200 new events across England, specifically targeted at deprived communities. This study aims to investigate the relationships between geographic access, deprivation, and participation in parkrun, and to inform the planned expansion by proposing future event locations. MethodsWe conducted an ecological spatial analysis, using data on 455 parkrun events, 2,842 public green spaces, and 32,844 English census areas. Poisson regression was applied to investigate the relationships between the distances to events, deprivation, and parkrun participation rates. Model estimates were incorporated into a location-allocation analysis, to identify locations for future events that maximise deprivation-weighted parkrun participation. ResultsThe distance to the nearest event (in km) and the Index of Multiple Deprivation (score) were both independently negatively associated with local parkrun participation rates. Rate ratios were 0.921 (95%CI = 0.921-0.922) and 0.959 (0.959-0.959), respectively. The recommended 200 new event locations were estimated to increase weekly runs by 6.9% (from 82,824 to 88,506). Of the additional runs, 4.1% (n=231) were expected to come from the 10% most deprived communities. ConclusionParticipation in parkrun is wide spread across England. We provide recommendations for new parkrun event location, in order to increase participation from deprived communities. However, the creation of new events alone is unlikely to be an effective strategy. Further research is needed to study how barriers to participation can be reduced. Online Map, data, and source codeAn interactive online map is available here, and the annotated R source code and all data that were used to generate the results of this study are provided on a repository.