How can UK public health initiatives support each other to improve the maintenance of physical activity? Evidence from a cross-sectional survey of runners who move from Couch-to-5k to parkrun

Abstract Physical activity improves physical and mental well-being and reduces mortality risk. However, only a quarter of adults globally meet recommended physical activity levels for health. Two common initiatives in the UK are Couch-to-5k (an app-assisted 9-week walk/run programme) and parkrun (a free, weekly, timed 5-km walk/run). It is not known how these initiatives are linked, how Couch-to-5k parkrunners compare to parkrunners, and the extent to which this influences their parkrun performance. The aims were to compare the characteristics and motives and to compare physical activity levels, parkrun performance and the impact of parkrun between Couch-to-5k parkrunners and parkrunners. Three thousand two hundred and ninety six Couch-to-5k parkrunners were compared to 55,923 parkrunners to explore age, sex, ethnicity, employment status, neighbourhood deprivation, motives, physical activity levels, parkrun performance and the impact of parkrun. Couch-to-5k parkrunners were slightly older, more likely to be female and work part-time, but similar in ethnicity, and neighbourhood deprivation compared with other parkrunners. Couch-to-5k parkrunners had different motives for participation and reported high levels of physical activity at registration, which remained to the point of survey completion. This group had slower parkrun times but, when registered for a year, completed a similar number of runs (11) per year. Larger proportions of Couch-to-5k parkrunners perceived positive impacts compared with other parkrunners and 65% of Couch-to-5k parkrunners reported improvements to their lifestyle. parkrun appears to be an effective pathway for those on the Couch-to-5k programme, and the promising positive association between the two initiatives may be effective in assisting previously inactive participants to take part in weekly physical activity.


INTRODUCTION
The importance of physical activity for health is well recognized in both academic literature and approaches to public health.There is strong evidence that supports the positive dose-response association between physical activity and physical and mental well-being and mortality for adults (O'Donovan et al., 2010;Warburton et al., 2010;White et al., 2017;Ekelund et al., 2019;Strain et al., 2020) and children (Lynch, 2019).To achieve good health, the World Health Organization recommends that adults aged 18-64 years should participate in at least 2 N. Relph et al. 150 min of moderate-intensity aerobic physical activity or at least 75 min of vigorous-intensity aerobic physical activity, or an equivalent combination of moderate-and vigorous-intensity throughout the week (World Health Organization, 2020) Yet only one in four adults globally meet the recommended guidelines (Guthold et al., 2018).
The World Health Organization's Global Action Plan on Physical Activity 2018-2030(World Health Organisation, 2018) identifies the importance of community-based initiatives to support physical activity participation.In the UK, the National Health Service (NHS) promote a running/walking initiative targeting physically inactive individuals called the Couch-to-5k.The aim of this nine-week progressive programme is to increase physical activity levels using a free downloadable mobile 'app' that people can use at a time that suits them (NHS, 2020).The programme involves three runs/ walks per week, with one day of rest in between, which varies from week to week (NHS, 2020).The creator, Josh Clark, wanted to create a bridge between walking and running, gradually progressing to the final week, a 30-min continuous run (NHS, 2020).But for some, this may not result in achieving a 5-km run as alluded to in the name of the initiative.Therefore, people may look to an alternative running programme that is not time limited to continue their running participation.
parkrun is a weekly, free to enter, 5-km mass participation event delivered across 22 countries (parkrun.com) that has been taking place in the UK since 2004.The event has a strong ethos of inclusivity, social interaction and community (Hindley, 2018) and is used by General Practices in the UK and Ireland as a Public Health referral option (Fleming et al., 2020).parkrun can be completed by running, walking or a combination of both and attracts people of all ages and abilities including those with limited experience of running (Stevinson and Hickson, 2014;Haake, 2018;Quirk and Haake, 2019).Indeed, the average parkrun time is approximately 29 min, thus roughly half complete the 5km slower than 30 min and may attract Couch-to-5k participants.Furthermore, parkruns are delivered by local teams of volunteers and participants are also encouraged to volunteer to complete roles such as marshalling, timekeeping, scanning barcodes, handing out finish tokens or tail walking; these volunteers are not required to ever run/walk the event.
Mass community-based participation events including parkrun have been shown to increase physical activity levels (Heath et al., 2012;Cleland et al., 2019), and cardiorespiratory fitness levels over 12 months (Stevinson and Hickson, 2019).Furthermore, previous research has shown that group support and social interaction, which may be provided at parkrun (Grunseit et al., 2020), are crucial to physical activity adherence following a beginner running/walking programme (Wiltshire and Stevinson, 2018) such as the Couch-to-5k programme.
Presently, little is known about the characteristics and physical activity patterns of Couch-to-5k parkrunners.It would be useful for research and public health practice to understand how many progress onto parkrun and engage in the parkrun initiative in the long term.However, the extent to which Couchto-5k serves as a pathway into parkrun is currently unknown.Furthermore, evidence on physical activity in general has identified that women, people living in deprived areas and older people with chronic diseases are more likely to be inactive (World Health Organisation, 2018).There is one exploratory study to date that found parkrun does attract these underrepresented groups (Stevinson and Hickson, 2014).However, other research identified that parkrunners are still more likely to be white, have higher socio-economic status and already be active (Fullagar et al., 2020;Grunseit et al., 2020).Therefore, another important finding from this study will be to see whether there is potential for the Couch-to-5k initiative to not only offer a pathway into parkrun but to also increase the diversity of this mass participatory event and improve physical activity levels of marginalized groups with typically lower physical activity levels.Therefore, the aims of this study are as follows: 1.) To compare the socio-demographic characteristics and participation motives between a sub-group of Couch-to-5k parkrunners and parkrunners.2.) T o compare the physical activity levels, parkrun performance measures and the impact of

CONTRIBUTION TO HEALTH PROMOTION
• Two popular public health initiatives are Couch-to-5k (9 weeks of progressive running) and parkrun (a free, weekly, 5-km walk/ run in local communities).• Runners who state Couch-to-5k as a reason for parkrun participation complete the same number of runs as other parkrunners.• parkrun is an effective pathway for those on the Couch-to-5k programme, and the positive association may increase the number of older females who take part in weekly physical activity.• Time-limited physical activity programmes should establish a link to regular community-based activities and may have the potential to attract groups who are typically less active in community-based activities.

Research design, procedures and participants
The current study was comparative cross-sectional.
The parkrun Health and Wellbeing Survey 2018: UK (Haake et al., 2018) was distributed online to all registered parkrunners in the UK aged 16 or over (2.2 million) between October and December 2018.The survey included a maximum of 47 questions, all were optional apart from identification of role in parkrun; either runners/walkers, runners/walkers who also volunteered at parkrun or volunteers only, one current health condition, disability or illness question and two life satisfaction questions.Full details of survey development and data handling processes are reported elsewhere (Quirk et al., 2021).A total of 100,864 individuals initially responded to this survey (4.5% participation rate), however once incomplete responses and volunteers only were removed the sample size was 59,999 parkrunners.For the current study, parkrunners who reported 'it was part of a Couch-to-5k programme' as one of their top three motives for participation in parkrun as a runner/ walker were included in a sub-group analysis (herein referred to as 'Couch-to-5k parkrunners').This subgroup comprised 3,296 people (5.5% of those who responded to the original survey).

Outcome measures
Supplementary File 1 details all outcome measures, the full survey and a copy of the participant information sheet provided to all participants.The outcome measures used in the current study are detailed below.

Socio-demographic characteristics
Participants reported their date of birth and hence age, sex, ethnicity and employment status.Neighbourhood deprivation was calculated from participant-reported postcodes provided at parkrun registration using English Indices of Multiple Deprivation (IMD) for lower-layer super output areas (Ministry of Housing, 2019).These scores were then collapsed into quartiles ranging from the most (level 1) to the least (level 4) deprived.

Motives
Participants were asked What motivated you to first participate at parkrun as a runner or walker?Respondents were asked to select a maximum of three answers out of a possible 21 motives.The answer choices were displayed in randomized order to help reduce response bias.The final choice was 'other', and respondents were asked to specify their motive.Participants were placed in the Couch-to-5k sub-group if they selected 'it was part of a Couch-to-5k programme' as one of their three motives.

Physical activity levels
Self-reported physical activity level at parkrun registration was collected using the following question: Over the last 4 weeks, how often have you done at least 30 min of moderate exercise (enough to raise your breathing rate)?Response options were as follows: (i) less than once per week; (ii) about once per week; (iii) about twice per week; (iv) about three times per week; (v) four or more times per week; and (vi) rather not say/do not know.Participants were asked this question again at the time of the survey to calculate the change in physical activity since registration.
parkrun performance Participants provided their parkrun ID number (a unique ID number provided to all parkrun registrants to identify them on the parkrun database and enable the collation of all their parkrun participation data).This ID (or their name, DOB, home parkrun-if their ID was not provided) was then matched to their parkrun profile and provided mean parkrun time, the number of years registered, total number of parkruns completed since registration and parkruns completed per year (if registered more than 1 year).

Data analysis
Frequency and percentage were used as descriptive statistics for categorical variables and median with interquartile range were used to summarize the continuous variables.Median was chosen because variables were highly skewed.Comparisons between Couch-to-5k parkrunners and the remaining parkrunners were analysed using Mann-Whitney U and Pearson's chi-squared tests with accompanying effect sizes.Significance was accepted at p < 0.05 level.If p was calculated as <0.001, we have reported this as such.

Socio-demographic characteristics
Couch-to-5k parkrunners were older than other parkrunners (median 50.5 years compared with 48.8 years; p < 0.001, effect size = 0.03) and more likely to be female than other parkrunners (72.5% vs 50.5%, p < 0.001, effect size = 0.10).Other parkrunners were predominantly white with 3.0% from BAME backgrounds and Couch-to-5k parkrunners were not significantly different to this (p > 0.05).However, Couch-to-5k parkrunners were more likely to be in part-time employment (19.7% vs 13.4%) and less likely to be in full-time paid employment (51.5% vs 54.8%) or self-employed (7.4% vs 9.4%) compared with other parkrunners (p < 0.001, effect size = 0.05); the proportion who were retired were similar for both (12.0%vs 12.2%, p > 0.05).A total of 9.5% of the other parkrunners came from the most deprived neighbourhoods and 40.1% from the least deprived: there was no difference between Couchto-5k parkrunners or other parkrunners (p > 0.05).

Motives for parkrun
Overall, 5.5% of total survey respondents chose Couch-to-5k as a motive.The top motives were fitness, physical health and sense of personal achievement (see Table 1).Motives for Couch-to-5k parkrunners tended to be ranked in the same order as other parkrunners but with lower proportions; this is probably due to the limit of three motives per person (i.e.only two additional motives to choose for Couch-to-5k parkrunners).Despite this, Couch-to-5k parkrunners may have been less motivated to first participate because of fitness (35.3% vs 57.4%; χ 2 = 627.1,p < 0.001, effect size = 0.10) to feel part of a community (3.9% vs 11.4%, Χ 2 = 182.5,p < 0.001, effect size = 0.06) and to spend time outdoors (3.1% vs 10.8%; Χ 2 = 201.9,p < 0.001, effect size = 0.06).Ranked third as a motive, Couch-to-5k parkrunners were equally likely to select a sense of personal achievement when compared with the rest of the sample (27.7% vs 26.8%; p = 0.270).

Physical activity levels
Physical activity at registration for Couch-to-5k parkrunners was significantly different to other parkrunners (p < 0.001, effect size = 0.10).Those who did up to 1 day of activity per week at registration represented 10.8% of the Couch-to-5k parkrunners compared with 16.9% for other parkrunners; additionally, 52.1% of Couch-to-5k parkrunners did about 3 days of activity at registration, compared with 32.6% other parkrunners.
Table 2 shows the change in physical activity from registration to the point of the survey.Just over a third (33.7%) of Couch-to-5k parkrunners increased their activity category level, while 22.5% decreased it; 43.8% stayed the same.Thus, there were 1.5 times as many Couch-to-5k parkrunners who increased their activity as decreased it.In comparison, 41.7% of the other parkrunners increased their activity while 16.5% decreased it, a ratio of 2.5.The distribution for the Couch-to-5k parkrunners and the rest of the sample was different at p < 0.001 (see Table 2 for corresponding effect sizes).

parkrun performance
The mean completion time of Couch-to-5k parkrunners was significantly longer at 34 min 36 s (SD = 5.7 min) compared with 29 min 52 s (SD = 6.2 min) for the rest of the sample (Figure 1: F = 15.4,p < 0.001, effect size d = 0.74).Couch-to-5k parkrunners were registered for a median of 1.21 years compared with 2.74 years for the rest of the sample (p < 0.001, effect size = 0.13) and hence had completed less parkruns (a median of 11 vs 22; p < 0.001, effect size = 0.10).However, when only considering participants who had been registered for at least a year (Couch-to5k parkrunners, n = 1429, rest of the sample, n = 32 782), the Couch-to-5k parkrunners had completed the same number of parkruns per year as the rest of the sample at approximately 11 per year (p > 0.05).

The impact of parkrun
The impact of running/walking for Couch-to-5k parkrunners compared with the rest of the sample is shown in Table 3, showing the proportions who indicated better and much better for each measure.All measures were significantly different at p < 0.001 between Couch-to-5k parkrunners and the remaining sample.

DISCUSSION
The aims of the study were to compare the socio-demographic characteristics and participation motives between a sub-group of Couch-to-5k parkrunners and parkrunners and to compare the physical activity levels, parkrun performance measures and the impact of parkrun between Couch-to-5k parkrunners and parkrunners.We did this in order to establish whether parkrun provides an effective pathway for people who have previously been inactive to continue physical activity following the Couch-to-5k programme.
This study identified that Couch-to-5k parkrunners were slightly older, more likely to be female and be in part-time employment, but with similar ethnicity (mainly white) and neighbour deprivation levels compared with other parkrunners.Ages in the current study are similar to previous parkrun research, between 35 and 54 years (Cleland et al., 2019;Stevinson and Hickson, 2019;Fullagar et al., 2020).Hence, it appears both initiatives are likely to attract middle-to older-aged groups, which is important as it is known that these groups have higher levels of inactivity (Pinto Pereira et al., 2018).
Couch-to-5k parkrunners were more likely to be female compared with other parkrunners.Similarly, a recent study on the Couch-to-5k programme reported more female participants than male participants (Relph   (Stevinson and Hickson, 2014).Thus, the route from the Couch-to-5k programme into parkrun may be an effective avenue to increase physical activity among females.Stride et al. (2020) suggest inclusivity may help females participate in parkrun, for example the flexibility, low time commitment and family atmosphere of parkrun events.The perception that the tail walker volunteer role (the last person to cross the finish line, ensuring that everyone is accounted for) is always female may also help women feel more welcome and  confident to in parkrun (Stride et al., 2020).This is a positive finding that suggests a potentially important association between the two running interventions to attract more females to sustain physical activity after completion of the time-limited Couchto-5k programme.Employment levels were in the current study, which corresponds to other research, with most participants across the whole sample employed (Cleland et al., 2019;Fullagar et al., 2020).White British was the main ethnicity in this study and other UK parkrun samples (Grunseit et al., 2020), and participants were mainly from neighbourhoods with low levels of deprivation, although areas of higher deprivation were represented.Smith et al. (2020) reported that areas in England of higher ethnic diversity and IMD have lower levels of parkrun participation even when controlling for population density, distance to the nearest parkrun event and age of population.The parkrun organizations are aware of this lack of diversity and keen to address this through initiatives such as their Outreach Ambassador Programme (Fullagar et al., 2020).

Total
Couch-to-5k parkrunners were less motivated by fitness, feeling part of a community, spending time outdoors, spending time with family or spending time with friends compared with other parkrunners.This may be because of their prior training and community created by the Couch-to-5k programme.However, future work could explore these findings using more qualitative methods.
The majority of Couch-to-5k parkrunners did approximately 3 days per week of activity at registration; this was likely because of the Couch-to-5k programme, which has participants doing this frequency of activity by the end of the nine weeks.As a consequence, Couch-to-5k parkrunners were less likely to increase activity after parkrun participation.This is a positive finding for the Couch-to-5k programme.The other parkrunners also reported good levels of physical activity levels, with more participants increasing activity levels in the time from registration to survey completion than Couch-to-5k parkrunners.Research supports this finding; UK parkrunners self-reported on average 350 min per week of moderate-to vigorous-intensity activity with only 8.8% reporting below recommended physical activity thresholds for health maintenance (Stevinson and Hickson, 2019).Stevinson and Hickson (2014) reported that most parkrunners in the UK classified themselves as regular runners (48%).Hence, the association between the two initiatives appears to help maintain regular physical activity levels.
The current study suggests Couch-to-5k parkrunners may not be considered inactive participants, a global target public health population (World Health Organisation, 2018).This may be explained in part as the Couch-to-5k programme involves running three times a week prior to participation in parkrun.Fullagar et al. (2020) note that perceptions of the 'run' in the name 'parkrun' may be a barrier to inactive participants who may be unaware that it is possible to walk the full 5-km route at each event.However, parkrun does attract smaller proportions of inactive people (Quirk et al., 2021).Future research should consider how less active people could be attracted to take part in both physical activity initiatives.
Couch-to-5k parkrunners on average took longer to complete the parkrun but did the same number of parkruns per year (at just less than one per month) when compared with other parkrunners.Therefore, it appears that Couch-to-5k parkrunners are similarly integrated into parkrun as other parkrunners.This is an important finding as it demonstrates that one, time-restricted physical activity intervention, can be successfully linked to another physical activity intervention to maintain activity levels.
Almost all impact measures of parkrun relating to health and well-being were greater for Couch-to-5k parkrunners compared with the rest of the sample, with sense of personal achievement, fitness and physical health being the top three improvements.The Couchto-5k programme is designed to attract those who are inactive, and hence may explain why almost all the sub-group deemed these to improve.Furthermore, the sub-group did not view competition as an important impact, again, likely due to the less performance-orientated nature of this group.

Limitations
The findings should be considered in light of the following methodological limitations.The study is a cross-sectional design, which limits the ability to report cause and effect and there may have been a selection bias effect as recruitment was not random.Furthermore, the survey relies on some self-reported measures, which can introduce recall error and response bias.However, it is important to note that the outcome measure of number of parkruns completed was measured using parkrun ID, which is recorded at each event.Finally, the sub-group of Couch-to-5k parkrunners was generated based on participants listing the Couch-to-5k as one of their top three motives for doing parkrun.Therefore, this sub-group may have missed parkrunners who came from the Couch-to-5k but did not list it in their top three motives.

CONCLUSION
Around 5% of our parkrun sample identified Couchto-5k as a motive for first participating as a runner or walker at parkrun.This group appears to be largely female, around years of age and similar in ethnicity, employment status and neighbourhood deprivation levels to other parkrunners.The majority registered with around 3 days of activity per week similar to the Couch-to-5k programme.Couch-to-5k parkrunners appear to go on to become integrated parkrunners, doing same number of parkruns per year as others (a median of 11 per year).They were less motivated by fitness and improving social connections, but almost all impact measures relating to health and well-being were greater for Couch-to-5k parkrunners including fitness, physical health, mental health, ability to manage their weight and lifestyle.parkrun appears to be an effective pathway for those on the Couch-to-5k programme and the association between the two running initiatives may be effective in increasing the number of females who take part in weekly physical activity.
The findings of this study have important implications for future public health initiatives that aim to increase and sustain physical activity levels.First, programmes that are time limited, such as Couch-to-5k, should establish a link with a local parkrun to provide regular opportunities for participation in activity and continued opportunities to experience the multiple health benefits of a mass participatory event.If a parkrun is not available, regular community-based activity should be embedded at the end of a shortterm programme.Second, beginner, time-limited programmes like Couch-to-5k have the potential to attract groups who are typically less active, such as women, to community-based events like parkrun which provide opportunities to sustain levels of physical activity.

Table 1 :
Comparison of motives for first participating in parkrun using Pearson's chi-squared tests.n = number of participants, χ 2 = chisquared statistic, p = p-value, ϕ = effect size

Table 2 :
Comparison in change in physical activity in categories using Pearson's chi-squared tests.χ 2 = chi-squared statistic, p = p-value, ϕ = effect size

Table 3 :
Impact of parkrun.Comparison of those reporting better and much better using Pearson's chi-squared tests.χ 2 = chi-squared statistic, p = p-value, ϕ = effect size Couch-to-