Changes in physical activity after joining a bikeshare program: a cohort of new bikeshare users

Background There are hundreds of bikeshare programs worldwide, yet few health-related evaluations have been conducted. We enrolled a cohort of new bikeshare members in Philadelphia (Pennsylvania, USA) to assess whether within-person moderate and vigorous physical activity (MVPA) increased with higher use of the program and whether effects differed for vulnerable sub-groups. Methods During 2015–2018, 1031 new members completed baseline and one-year follow-up online surveys regarding their personal characteristics and past 7-day MVPA minutes per week (minutes per week with- and without walking). Participants were linked to their bikeshare trips to objectively assess program use. Negative binomial (for continuous outcomes) and multinomial (for categorical outcomes) regression adjusted for person characteristics (socio-demographics, health), weather, biking-infrastructure, and baseline biking. Results Participant median age was 30, 25% were of Black or Latino race/ethnicity, and 30% were socioeconomically disadvantaged. By follow-up, personal bike ownership increased and 75% used bikeshare, although most used it infrequently. Per 10 day change in past year (PY) bikeshare use, non-walking MVPA min/wk increased 3% (roughly 6 min/wk, P < 0.014). More days of bikeshare was also associated with change from inactive to more active (odds ratio for ≥ 15 days in PY vs. no bikeshare use 1.80, CI 1.05–3.09, P < 0.03). Results were consistent across vulnerable sub-groups. In general, impacts on MVPA were similar when exposure was personal bike or bikeshare. Conclusions Bikeshare facilitated increases in cycling, slightly increased non-walking MVPA, and showed potential for activating inactive adults; however, for larger program impact, members will need to use it more frequently. Supplementary Information The online version contains supplementary material available at 10.1186/s12966-022-01353-6.


Enrollment.
We did not have direct access to the count of new members who joined the bikeshare each day. However, we used Google Analytics [2] to count the number of people who went to the survey webpage. Approximate 38% of people who went to the survey webpage, moved forward and completed the 'eligibility' screener (N= 3366 persons); among those, 36% (N=1206) were eligible and completed the main survey. Eligibles and ineligibles had roughly similar demographics and baseline value on a single physical activity question that was embedded in the enrollment screening tool (data not shown). The cohort's race/ethnicity profile was a good representation of residents in the bikeshare service area, however, the cohort had higher income/education (determined from Census [3]).
Retention over the follow-up period. Participants were invited to participate in the follow-up survey approximately 365 days after their baseline survey (mean days between baseline and completed follow-up survey = 368 [STD=10 days]). For the present analyses, from the 1206 participants who completed the baseline survey, we excluded an additional 106 participants (N=5 who did not agree to release their bikeshare trip data and N=101 who moved outside of Philadelphia during follow-up).
Validation checks and missing data. The evaluation's online questionnaire collected detailed socio-demographic data, bikeshare membership status, personal bike ownership and use, main mode of transport, car ownership, and physical activity and health-related factors, among other questions. The online survey had multiple validation checks (including summarizing their physical activity days/minutes and asking them if it 'sounded about right' or whether they wanted to revise their response). For the present analysis, there were no missing data. Demographic responses missing from the baseline survey were collected at the 12-month follow-up survey and/or study staff follow-up with participants via phone and email if answers to key questions were missing. Figure 1. Flow diagram of recruitment and retention.

SUPPLEMENT METHODS TEXT
A. Assessment of total physical activity levels Total physical activity levels were assessed using a modified version of the International Physical Activity Questionnaire (IPAQ-L) [4,5]. This instrument has been widely used [6] and found to have acceptable measurement properties, at least as good as other established self-reports [4,6]. More information about data collection is published separately. [7] Modifications were made following recommendations by Rzewnicki et al. (2003) [8], for low-numeracy populations, and to accommodate online administration/response.
Total physical activity was represented by past 7-day self-reported activity along 3 domains, work, leisure, and transportation. What follows are details on collection of each of these domains. For work and leisure activity, participants were asked 'how many days' in past 7 days, 'how many minutes (or hours) on a typical day', and to only report activity that 'made you breathe harder than normal for at least 10 minutes at a time'. Additionally, for work activity, participants were asked to write-in the type of activity and job title. For leisure activity, participants used a pulldown menu to select from 63 activities (plus a write-in 'Other' option if not on the list). Job title and leisure activity selections were subsequently used to ascertain metabolic equivalent intensity level [9 , 10] in order to ascertain whether the activity required moderate or vigorous effort. For transportation activity, data were collected separately for transport bicycling and walking. Participants were asked days, trips, minutes each trip ('how many days' in past 7 days, on a typical day how many one-way trips and how many minutes each trip).
Additionally, for work activity, participants were asked to write-in the type of activity and job title. For leisure activity, participants used a pull-down menu to select from 63 activities (plus a write-in 'Other' option if not on the list). Job title and leisure activity selections were subsequently used to ascertain metabolic equivalent intensity level [9]. For transportation activity, data were collected separately for transport bicycling and walking. Participants were asked days, trips, minutes each trip ('how many days' in past 7 days, on a typical day how many one-way trips and how many minutes each trip).

B. Disadvantaged status
In order to control for socio-economic disadvantage (hereafter referred to as 'disadvantage') and also assess whether program use affected physical activity differently by disadvantaged status, we derived an indicator for disadvantage using educational attainment, employment/ occupational status, income and number of persons supported by the reported income. Details on this indicator are in Supplement Text A. Persons were classified as disadvantaged if they had any of the following: 1. lower education (defined as adults aged >= 30 with less than 4-year college), 2. underemployed (participant selected "unemployed and seeking work" regardless of other employment/occupation categories selected), 3. lower income (reported <$35000 income per capita, approximately 200% of the federal poverty level). We took a conservative approach by excluding students aged <30 from criteria numbers 2 and 3. The rationale was that many students acquire little income but have access to family resources (allowances, housing costs covered, other living expenses paid by family). Our decision was supported by local research that found less than 30% of students from the three largest colleges in the bikeshare area come from families with lower-incomes [11].

Secondary data: Neighborhood bikeshare stations, roadway bikeability
We linked participant survey responses to bikeshare use data (described above) as well as other spatial data. Participant residential addresses were geocoded and spatially linked to bikeshare station locations and roadway bikeability.
Bike share station locations came from the bikeshare program. Circular buffers were calculated around each participant's residence. Density of bikeshare stations per land area around each participant's residence were derived for a 400m area; this distance represents a convenient walking distance to bikeshare [12] and other transit infrastructure [13,14].
A roadway bikeability index was derived using an approach developed by others [15] and adapted to the Philadelphia context, using data from the city's offices of transportation and planning. We used the index to determine whether street segments had a high level of traffic-related road stress (AKA 'low bikeability') and then calculated the fraction of road segments around each participant's residence that had low bikeability. Low bikeability was for a 800m buffer area in order to correspond with perceptions of neighborhood bikeability. [16,17] C. Supplemental text for unadjusted change in physical activity Average change in physical activity across the cohort Across the cohort, there was a slight increase in minutes of MVPA without walking. At baseline, median minutes was 180 minutes, IQR 0-450 and 26% had no weekly minutes. At follow-up, median minutes of MVPA without walking was 225 (IQR 40-450) and 22% had no weekly minutes of MVPA.

Within-person change in physical activity
However, when considering within-person change in MVPA, most participants (73%) did not change their activity status ( Table 2) and unadjusted median change in minutes was 0 (IQR -120, 180, not shown in tables). Older and disadvantaged members were more likely to become active; we found no other major differences across demographic groups (Table 1) Below shows results where participants were excluded if they decreased their walking minutes at follow-up. The intention of creating this subset was to remove those who could have substituted bikeshare for walking. Results shown for adjusted within-person differences in non-walking moderate or vigorous physical activity (MVPA) minutes and change in activity status (became active, became inactive), according to number of days used the program and use any type of bike. N=503. Below shows results after excluding those with zero days of bikeshare during the past 12 months, results were complimentary to results for the full sample but, in general, the magnitude of the effect was stronger. Adjusted within-person differences in non-walking moderate or vigorous physical activity (MVPA) minutes and change in activity status (became active, became inactive), according to number of days used the program and use any type of bike.