The untapped health and climate potential of cycling in France: a national assessment from individual travel data

physical activity; cycling; health impact assessment; active transportation; health co-benefits of climate action; urban mobility . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 2

Objectives Promoting active modes of transportation such as cycling may generate important public health, economic, and climate mitigation benefits. We aim to assess mortality and morbidity impacts of cycling in a country with relatively low levels of cycling, France, along with associated monetary benefits; we further assess the potential additional benefits of shifting a portion of short trips from cars to bikes, including projected greenhouse gas emissions savings.

Using individual data from a nationally-representative mobility survey, we described the French 2019 cycling levels by age and sex. We conducted a burden of disease analysis to assess the incidence of five chronic diseases (breast cancer, colon cancer, cardio-vascular diseases, dementia, and type-2 diabetes) and numbers of deaths prevented by cycling. We assessed the corresponding tangible costs saved based on direct medical costs and intangible costs based on the value of a statistical life year. Lastly, based on individual simulations, we assessed the likely additional benefits of shifting 25% of short (<10km) car trips were shifted to cycling.

The French adult (20-89 years) population was estimated to cycle on average of 1min 17sec pers -1.day-1 in 2019, with important heterogeneity across gender and age. This yielded benefits of 1,919 (uncertainty interval, UI: 1,101-2,736) premature deaths and 5,963 (95% UI: 3,178-8,749) chronic disease cases prevented, with males enjoying nearly 75% of these benefits. Direct medical costs prevented were estimated at €191 million (UI: 98-285) annually, while the corresponding intangible costs were nearly 25 times higher (€4.8 billion, UI: 3.0-6.5). Shifting 25% of short car trips to biking would yield approximatively a 3-fold increase in benefits, while also generating important CO2 emission reductions (0.688 MtCO2e, UI: 0.524-0.854).

In a country of low- to moderate cycling culture, cycling already generates important public health and health-related economic benefits. Further development of active transportation would increase these benefits while also contributing to climate change mitigation targets. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 3 Insufficient physical activity is responsible for a substantial global burden, with an approximatively 7% of all-cause deaths attributable. This burden is even more pronounced in Europe, with an approximate 10% of all-cause deaths and 30% of direct health-care costs of non-communicable diseases and mental health conditions attributable to physical inactivity (1,2). In high-income Western countries, the prevalence of physical activity reaches more than 40% while showing no decreasing trends over time (3). While the disquieting situation results in part from individual behaviors and choices, collective and societal decisions bear an essential responsibility, especially with regards to transport-related physical activity (4). Active travel offers a unique opportunity to boost physical activity by encouraging seamless integration in daily activities, requiring little to no additional time commitments and costs to the individual (5). Despite clear benefits of the promotion utilitarian cycling, most countries do not use it to its full potential: cycling rates for example in temperate regions such as the UK or France, the modal share of cycling remains below 3%, far from the 15 to 27% rates obtained in countries such as the Netherlands or Denmark (6,7). Collective choices regarding mobility also largely contribute to climate change, which is widely recognized as one the biggest public health issues of the upcoming decades (8). In Europe, the transportation sector represents the second largest contributor of greenhouse gas emissions (9), and the first in France where it accounts for 31% of the national emissions, among which approximately 50% are attributable to cars (10). Therefore, promoting modal shift from cars to active transportation, such as cycling, represents a relevant response to both the public health burden of physical inactivity and the urgent need to cut down emissions. While the public health benefits of cycling have already been assessed in a large variety of contexts (11), national assessments based on representative mobility data are less common (12). Moreover, few burden of disease studies have assessed its impact in terms of both mortality and morbidity, and few have estimated the associated direct medical and intangible costs avoided. Such assessments can contribute meaningfully to intense debates currently ongoing about the future of mobility in a context of ambitious decarbonisation targets and energy crises in many high-income Western countries such as France. Our aim in this study was to demonstrate how cycling can contribute to health promotion in a Western country with relatively low cycling rates. We first assessed the benefits associated with current cycling levels in France in terms of prevented mortality and morbidity, based on nationally-representative individual mobility data from 2019. Secondly, we assessed the benefits that could be associated with shifting a portion of short trips currently made by car to bike trips.

Mobility data Individual-level mobility data were accessed from the 2019 Enquête mobilité des personnes (“People’s mobility survey”) (13). This survey is conducted every 10 years and the most recent data represents the mobility of the metropolitan French population in 2019, before the onset of the Covid-19 pandemic. 13,825 individuals aged 5 years and older were asked face to face about their travels the day before. Distances travelled are consolidated using geographic information systems, based on reported departure and arrival location. The sampling design and sampling weights ensure that the survey is representative of travel behaviours across weekdays and weekends in France. We analyzed all cycling . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 4 trips, regardless of travel purpose and week day. Distances cycled were converted into exposure time (i.e. minutes of cycling per day) considering an average cycling speed of 14.9 km.h-1 (14). As e-bikes were not frequently used in 2019, for simplicity, we assumed the same speed and metabolic equivalent of task (MET) values for e-bikes than for classical bikes. Morbi-mortality and demographic data Based on a previous systematic review (15), we assumed a protective effect of physical activity on the following five diseases: breast cancer in females, colon cancer, cardiovascular disease, dementia, and type 2 diabetes (16–20). We also assumed a protective effect of cycling on all-cause mortality as documented by the meta-analysis by Kelly et al (21). As this meta-analysis considered all-cause mortality, potential increases in exposure to air pollution or injury are implicitly controlled. We thereafter use the term “morbi-mortality” to refer to these five chronic diseases and all-cause mortality. For each morbi-mortality event, the relative risk (RR) identified in the literature was scaled for a

volume of 11.25 MET.hours (supplementary Table 1). France-specific incident cases for cardiovascular disease, dementia, and type 2 diabetes were determined for the year 2018 from a database of new beneficiaries of long-term illness coverage (Affectations Longues Durées, ALD) by the French National health insurance system, using the most recent data available. The 2019 incidence data for breast cancer and colon cancer were taken from the French National Cancer Institute. Deaths, death rates, life-expectancy and population count estimates for 2019 were taken from the National Institute of Statistics and Economic Studie (22). (Table 1). Health impact modelling The health benefits of 2019 levels of cycling were assessed compared to a counterfactual “zero cycling” scenario in which trips did not entail any physical activity. For each individual surveyed, time exposure to cycling (if any) was converted into a percentage reduction in the age-specific risk for each morbi-mortality event. Therefore, for each event, we estimated the reduced individual risk attributable to cycling as compared to the zero cycling counterfactual scenario. Following the Health Economic Assessment Tool (HEAT) approach, we assumed linear dose-response functions (DRF) for the mortality and morbidity effects of cycling while disregarding baseline levels of physical activity, and we capped the all-cause mortality reduction at 45% (23). The age range considered for reductions in morbidity and mortality was 20-89 years. We disregarded younger and older ages because of the scarcity of evidence for the health effects of physical activity in these age ranges for these specific outcomes. For each morbi-mortality event, by summing up the reduction in individual probability of event across the study sample, we estimated the total number of events that have been averted by cycling. Survey weights were applied in the calculation in order to obtain nationally representative figures across weekdays and weekends. Burden of Disease Based on the estimated age-specific numbers of morbi-mortality events prevented, we estimated the corresponding Disability-Adjusted Life Years (DALY) prevented by adding up the years of life lost (YLL) and the years lived with disability (YLD). To calculate YLD, for each of the five morbidity event, we estimated an average disability weight by dividing disease-specific YLD-estimates by disease-specific prevalence estimates obtained from the 2019 Global Burden of Disease (GBD) (24), based on the

described in (25). Health economic evaluation . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 5 We evaluated the monetarized benefits of 2019 French cycling levels by considering two different definitions of costs. We first estimated the medical costs associated with chronic diseases included in the analyses. Disease-specific medical costs were extracted from a recent French governmental report on the assessment of health impacts of public investments and were expressed in Euros 2018 (Table) (25). However, such evaluation of the tangible medical costs prevented by cycling does not include the benefits on mortality. Therefore, we also estimated the intangible costs prevented by cycling using the standard value of statistical life year (VSLY) that is recommended in France for the socioeconomic evaluation of public investments (26). Estimates of intangible costs prevented by cycling thus rely on the VSLY of 133k€, expressed in Euros 2019. Uncertainty analysis Central values and 95% confidence intervals for cycling exposure were estimated while accounting for the survey sampling design and weights using the R package survey (27). Uncertainty surrounding the RR relating cycling and morbi-mortality events were characterized by a log- normal distribution constructed for each outcome based on the RR central values and the lower and upper bounds of the 95% confidence interval provided in the literature. We then combined the uncertainties using a Monte Carlo approach where we independently sampled a RR value for each morbi-mortality event. We randomly sampled 1000 combinations of RR values. Based on a given set of randomly-sampled RR values, the mean value and the standard-error for each outcome were estimated and results were combined across sets of RR combinations using the Rubin’s rule (28). The distributions corresponding to each of the 100 replications were then combined together to generate a posterior distribution for each outcome, from which we computed the 2.5% and 97.5% percentiles to obtain 95% uncertainty intervals (UI). Modal shift scenario Lastly, based on individual-level simulations, we assessed the potential additional public health benefits associated with a modal shift scenario in which 25% of car trips less than 10km would be shifted to cycling. To do so, we randomly selected 25% of respondents reporting such car trips, while also reporting no cycling trips, and estimated the corresponding exposure time to cycling had these trips been cycled. We then re-conducted the steps of the HIA described above to assess the health and health-related economic benefits associated, while assuming no substitution effect of cycling on other types of physical activity. We performed 100 random draws of car trips to be re-allocated to bike trips. For each of these random selections, we computed the DALY prevented together with the 95% UI following the steps previously described. Results of each replication were combined to generate a posterior distribution for each outcome and compute 95% UI. Lastly, we estimated the CO 2 emissions averted in this modal shift scenario, assuming that the 2019 French car fleet emitted on average 124 gCO2.km-1 (29). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 6 Table 1: Data sources, medical costs, disability weights and dose-response functions used in the analysis to assess the health and health-related economic impact of physical activity. Morbi- mortality event Incidence data Related Medical costs (per case) Disability weight RR, for 100 min cycling, ie 11.25 MET.hours (95% CI) RR

Breast Cancer (females) National Cancer Institute 46,968 € 0.068 0.90 (0.87-0.95) Monninkhof et al., 2007 Colon Cancer National Cancer Institute 26,716 € 0.093 Males: 0.93 (0.88-0.99) Females: 0.94 (0.91-0.99) Harris et al., 2009 Cardiovascular Disease Long-term condition coverage 20,938 € 0.052 0.92 (0.90-0.95) Hamer & Chida, 2008 Dementia Long-term condition coverage 22,748 € 0.152 0.90 (0.86-0.95) Hamer & Chida, 2009 Type 2 Diabetes Long-term condition coverage 36,514 € 0.068 0.81 (0.72-0.89) Jeon et al., 2007 All-cause mortality INSEE - 1 0.90 (0.87,.94) Kelly et al., 2014 RR: Relative risk, CI: confidence interval. The relative risks for chronic diseases were those identify in a previous systematic review (15). See Supplementary Table 1 for the initial values of RR, the references used for physical activity and the scaling for RR for 100min cycling. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 7

In 2019, among the French population of 20-89 year-olds (population size, 49.0 million), the cumulative kilometers biked in France added up to 4.64 (95% CI: 3.28-6.00) billion kilometers, of which 6.23% were cycled on e-bikes. This represented a per-capita average distance biked of 0.32 (95% CI: 0.27-0.37) km.pers-1.day-1, corresponding to an average exposure time of 1min 17sec pers-1.day-1. The proportion of the population reporting any cycling trip on a given day, accounting for differences in weekends and weekdays, was 3.12% (95% CI: 2.64-3.61%), with variations according to age and sex (Figure 1). In all age groups, the proportion of cyclist was higher in males vs. females, and the average distance cycled among cyclists was higher in male vs. female (Fig 1). This resulted in 72.2% (95% CI: 60.7%-82.6%) of all cycling distances being biked by males. Figure 1: Proportion of the French adult population reporting any bike trip any cycle trip a day, accounting for differences in weekends and weekdays (top), and mean distance cycled (km) in the past day among those reporting any bike trip (bottom) according to sex and age. Enquête mobilité des personnes, France, 2019. Black lines represent 95% confidence intervals. Health impact assessment Based on the observed cycling levels, we estimated that 1,919 (UI: 1,101-2,736) premature deaths and 5,963 (UI: 3,178-8,749) chronic diseases were prevented in 2019 in France due the protective effect of . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 8 biking physical activity. The chronic disease with the greatest number of cases prevented was type-2 diabetes (n=3,743, UI: 1,576-5,912), 3743.838 followed by CVD (n=1,578, UI: 778-2,378). Of all morbi- mortality events prevented, 74.9% benefited males (Figure 2). On average, cycling reduced the yearly mortality for the total adult population by 0.68% ( UI: 0.47% - 0.87%). Figure 2: Chronic diseases and mortality prevented by the physical activity due to cycling in France among adults aged 20-89 years, 2019. Black lines represent 95% uncertainty intervals. We further estimated that overall, 35,135 DALYs ( UI: 22,693 – 48,791) were prevented through cycling- related physical activity, of which 28,416 (80.9%) were driven by mortality (YLL) and 6,719 (19.1%) by morbidity (YLD). Based on the average values of annual medical costs associated with each chronic disease considered here, we estimated that the 2019 levels of cycling prevented €191 (UI: 98-285) million annual medical (tangible) costs. Based on the value of a statistical life year, combining intangible costs of mortality and morbidity, we estimated that cycling prevented €4.75 (UI: 3.02-6.49) billion, roughly 25 times the annual medical (tangible) costs. Table 2 summarizes the burden and costs prevented for each morbi-mortality event. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 9 Table 2: Burden, tangible costs and intangible costs prevented by the physical activity due to cycling in France, 2019, for several morbi-mortality event, among adults aged 20-89 years. Morbi-mortality event Cases prevented in 2019 (Uncertainty interval, UI) Medical (tangible) costs prevented in 2019, M€ (UI) Intangible costs prevented in 2019,M€ (UI) Breast cancer 254 (107-402) 11.9 (5.0-18.9) 58.2 (34.1-82.3) Colon cancer 205 (51-359) 5.5 (1.4-9.6) 35.5 (14.3-56.6) Cardiovascular disease 1578 (778-2378) 33.0 (16.3-49.8) 175.8 (113.6-238.0) Dementia 182 (68-295) 4.1 (1.6-6.76) 29.1 (15.9-42.2) Diabetes Type 2 3,744 (1,576-5,912) 136.76 (57.56-215.9) 723.3 (382.1-1064.6) Mortality 1,919 (1,101-2,736) Not Applicable 3731.8 (2808.2-4655.4) Intangible costs are estimated based based on a value of a statistical life year (VSLY) of 133k€, expressed in Euros 2019. Figure 3: Disability-adjusted life years (DALY) prevented by physical activity due to cycling in France, 2019, according to age group. DALY associated with chronic diseases correspond to years of life with disability (YLD) and DALY associated with mortality correspond to years of life lost (YLL, in blue). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 10 Modal shift scenarios The proportion of survey participants reporting any short (<10km) car trip on a given day was 41.2% (95% CI: 37.5-44.8) overall and was quite homogeneously distributed across age and sex, as was the mean length of these short car trips (3.69 km, 95% CI: 3.57-3.81) (Figure 4). Figure 4: Proportion of the French adult population reporting any short (<10km) car trip in the past day (top), and mean distance driven (km) in the past day among those reporting any car trip (down) according to sex and age. Enquête mobilité des personnes, France, 2019. Black lines represent 95% confidence intervals. Shifting 25% of short car trip to cycling would generate an additional 5.550 billion km cycled (UI: 4.222- 6.884), ie an approximate 20% higher cycling exposure compared to the 2019 baseline. Due to the age distribution of drivers, this would translate into approximately 2.6 times more deaths prevented compared to the 2019 cycling levels (4,707 deaths prevented, UI: 2,865-7,199) (Table 3). This modal shift scenario would prevent €7.476 billions in terms of intangible costs (UI: 4.970-10.71), while also reducing CO2 emissions by 0.688 mega-tons (UI: 0.524-0.854). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 11 Table 3: Climate, health and health-related economic benefits of cycling in France, 2019, and estimated impact of a modal shift scenario. Outcome Baseline estimates (Uncertainty interval, UI) Incremental effect of shifting 25% of short (<10km) car trips to cycling (in addition to the baseline estimates) (UI) Yearly km cycled (billion) 4.640 (3.284-5.996) 5.550 (4.222-6.884) CO2 emissions prevented (Mto) 0.575 (0.4070.743) 1 0.688 (0.524-0.854) # of deaths prevented 1919 (1101-2736) 4,704 (2,689-6,721) # of chronic diseases prevented 5,963 (3,178-8,749) 8,509 (5,205-11,813) # DALYS prevented 35,135 (22,693 – 48,791) 57,4650 (34,983-78,733) Medical (tangible) costs prevented (million €) 191 (98-285) 267 (178-393) Intangible costs prevented (billion €) 4.75 (3.02-6.49) 7.56 (4.65-14.47) Intangible costs are estimated based on the value of a statistical life year (VSLY). 1 As compared to a counterfactual where individual would have done the same trips driving instead of cycling

In this study relying on a nationally-representative mobility survey, we show that, despite relatively low levels of cycling in France in 2019, physical activity due to cycling generated important public health benefits in terms of morbidity and mortality alleviated. These benefits were unequally distributed across sex, reflecting the cycling distribution, with males benefiting nearly 75% of the morbi-mortality events prevented. Of the 5 different chronic diseases considered, cycling levels mostly prevented cases of diabetes and CVD, and in all prevented €187 million in direct medical costs in 2019. Most of the public health benefits of cycling, however, resided in the mortality alleviated, which corresponded to approximately 80% of all DALY associated with cycling. When accounting for the VSLY, we estimate that the physical activity linked to the 2019 levels of cycling prevented nearly €5 billion of intangible costs annually. We thus estimated that for 1€ of direct medical (tangible) costs prevented by cycling, an additional 25€, approximatively, were prevented when considering intangible costs. Furthermore, we showed that more than twice as many deaths could be prevented, with a modest modal shift of 25% short car trips to cycling, which would also lead to sizeable CO2 emissions reduction These results confirm the relevance of promoting cycling both for planetary and public health.

from the 2019 wave of the French transport survey show that France remains within the countries with a low cycling culture (30). The French adult population reported cycling on average less than 10 minutes per week, which is low compared to the more than 70 minutes per week reported by the Dutch adult population (12). Despite repeated calls to decrease motorized trips to help tackle climate change and physical inactivity, cycling mode share for France remained at about 2.7% of all trips in France between 2008, the year of the previous wave of travel survey, and 2019. The bike modal . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 12 remained the same (2.7%) between the two waves (7). This explains the relatively low levels of the health benefits of cycling in France as compared to other countries. In France, we estimate that cycling reduces the mortality risk by 0.6%, compared to 7.4% in the Netherlands (12). We showed that currently males are the main beneficiaries of cycling health benefits in France because they represent a large fraction of the cyclists. However, previous surveys have shown that females are equally represented in countries where cycling is normalized, providing an additional motivation for cycling promotion (31). Our burden of disease approach allowed us to estimate the relative contribution of mortality and morbidity alleviated in the total health benefits of cycling. We found that cycling would prevent nearly 3 times more chronic diseases than deaths (approximatively 5,900 diseases and 1,800 deaths). This ratio was similar to the one reported in a previous HIA in Barcelona (15,32). However, the chronic conditions prevented are associated with relatively low disability weights (ranging between 0.05 and 0.15), which explains the large contribution of prevented mortality in the DALY. In terms of morbidity, we showed that the disease most frequently prevented were diabetes and cardio-vascular diseases. Previous studies reported that cycling also contributes to reducing mortality among adults with diabetes (33), making it a relevant intervention for both primary and secondary prevention. We further estimated the public and climate mitigation benefits of a modal shift scenario. We estimated that shifting a quarter of short car trips to cycling would further prevent approximately 4,700 annual deaths and avoid €7.5 billion intangible costs. Table 4 further presents a broad comparison of elements to put into context the results of the modal shifts scenarios we assessed here. As a rough comparison, this is triple the number of deaths avoided due to efforts in road safety made in France over the past 10 years, which represented a public investment of more than €3.5 billion a year (34). From a climate change perspective, shifting 25% of short car trips to cycling would yield roughly half of CO 2 emission reductions of a measure that has been publicly debated in France in the context of the 2022 energy crisis: reducing the maximal speed on highway from 130 to 110 km/h (35). However, it is also important to note that investments to foster active mobility may possibly result in large additional indirect cuts in CO2 emissions, for instance through encouraging multi-modal trips and/or re-location of activities which would reduce global distances travelled. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 13 Table 4: The public health, health-related economic and climate mitigation benefits of cycling in context, France Dimension Outcome Value Reference Public health Expected deaths yearly prevented by shifting 25% of short (<10km) car trips to bike Approx.. 4,700 deaths yearly Present study Deaths prevented by recent efforts for road safety over the past 10 years Approx.. 1,500 deaths yearly (34) Deaths prevented yearly by reducing alcohol consumption by 20% Approx.. 1,500 deaths yearly (36) Economics Expected medical (tangible) costs prevented by shifting 25% of short (<10km) car trips to bike Approx. €270 million yearly Present study Intangible costs prevented by shifting 25% of short (<10km) car trips to bike Approx. €7,500 million yearly Present study Yearly budget of the French National Cancer Institute (INCa) €118.5 million in 2023 https://www.e- cancer.fr/Institut-national- du-cancer/Qui-sommes- nous/Budget Climate change mitigation CO2 emissions prevented by shifting 25% of short (<10km) car trips to bike Approx. 0.7 Mto yearly Present study CO2 emissions prevented by reducing the maximal speed on highway from 130 to 110 km/h Approx. 1.45 Mto yearly (35) Energy efficiency tax credit for households’ investment in home thermal renovation Approx. 0.12 Mto yearly in 2015 and 2016 (37) Our study suffers from several limitations. First, we disregarded the baseline levels of physical activity when assessing the health benefits of existing or projected levels of cycling. However, the DRF we used came from meta-analyses of studies conducted in samples of participants with heterogeneous levels of physical activity. Second, we did not account for the reduced physical activity that e-bikes may represent as compared to regular bikes. However, e-bikes only contributed to 6% of the distances we considered, suggesting a modest over-estimate of the benefits we document. Similarly, the modal shift scenarios we assessed did not account for a possible compensation during non-transport physical activity which could reduce the benefits we assessed. However, the existing evidence suggests that such substitution effect is unlikely or limited, in healthy adults at least (38,39). On the other hand, the dose-response relationship we used for all-cause mortality may be considered as conservative, as a more recent meta- analysis suggested a substantially more beneficial dose-response relationship (40). In a previous assessment, the choice of this dose-response relationship was identified as the main source of sensitivity in the estimates (41). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 14 This study is one of the few to assess the benefits of cycling at the country level based on detailed transportation data and, to our knowledge, the first to do so for France. One of its main strengths lies in the fact that it documents both the medical and the social costs prevented by cycling. Although they disregard a major part of the estimated health benefits (those related to the mortality alleviated), medical costs represent tangible costs effectively saved for collective benefit. On the other hand, social costs based on VSLY are intangible, because they represent the propensity of society to pay for the corresponding health benefits, but capture much more comprehensively the benefits expected from specific policies. The ratio of 1:25 we document here for these tangible and intangible costs may be useful to make sense of these figures in similar assessments. The present study contributes to highlight the public health and climate mitigation benefits expected from the development of active transportation (41,42). Our results suggest that public investments to encourage modal shift toward cycling may translate into important climate, health and health-related economic benefits, which are likely to exceed the costs implied (43). Recently, due to the Covid-19 impact on public transportation, some local authorities have rapidly incentivized cycling by rolling out pop-up bike lanes. This resulted in a large short-term increase in cycling, including in France (44). Commitment of national and local authorities is critical to sustaining these changes and the contribution of cycling to public and planetary health. Competing interests The authors declare that they have no competing interests. Authors’ contribution ES, PQ, and KJ conceived the original idea. ML extracted and compiled input data. ES and KJ conducted the analysis and produced output figures and tables. ES, MS, AND, PQ, and KJ interpreted the results. ES and KJ wrote the first draft of the article. All authors provided critical feedback and helped shape the research, analysis and manuscript.

1. Katzmarzyk PT, Friedenreich C, Shiroma EJ, Lee IM. Physical inactivity and non-communicable disease burden in low-income, middle-income and high-income countries. Br J Sports Med. 2021 Mar 29;bjsports-2020-103640. 2. Santos AC, Willumsen J, Meheus F, Ilbawi A, Bull FC. The cost of inaction on physical inactivity to public health-care systems: a population-attributable fraction analysis. Lancet Glob Health. 2023 Jan 1;11(1):e32–9. 3. Guthold R, Stevens GA, Riley LM, Bull FC. Worldwide trends in insufficient physical activity from 2001 to 2016: a pooled analysis of 358 population-based surveys with 1·9 million participants. Lancet Glob Health. 2018 Oct;6(10):e1077–86. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 15 4. Koohsari MJ, Badland H, Giles-Corti B. (Re)Designing the built environment to support physical activity: Bringing public health back into urban design and planning. Cities. 2013 Dec 1;35:294–8. 5. WHO Regional Office for Europe. Walking and cycling: latest evidence to support policy-making and practice [Internet]. Copenhagen, Denmark; 2022 [cited 2023 Apr 4]. Available from: https://www.who.int/europe/publications/i/item/9789289057882 6. Nielsen TAS, Mulalic I, Christiansen H. Drivers of Cycling Mode-share: Analysis of Danes Travel Behavior 1996-2013. Transp Res Procedia. 2016 Jan 1;14:2284–8. 7. Goel R, Goodman A, Aldred R, Nakamura R, Tatah L, Garcia LMT, et al. Cycling behaviour in 17 countries across 6 continents: levels of cycling, who cycles, for what purpose, and how far? Transp Rev. 2021 May 9;0(0):1–24. 8. Atwoli L, Baqui AH, Benfield T, Bosurgi R, Godlee F, Hancocks S, et al. Call for emergency action to limit global temperature increases, restore biodiversity, and protect health. BMJ. 2021 Sep 6;374:n1734. 9. European Environment Agency. Annual European Union greenhouse gas inventory 1990–2018 and inventory report 2020 [Internet]. Copenhagen, Denmark: European Environment Agency; 2020 [cited 2022 Jan 14]. Available from: https://www.eea.europa.eu//publications/european-union- greenhouse-gas-inventory-2020 10. The High Council on Climate (HCC). Renforcer l’atténuation, engager l’adaptation [Reinforce mitigation, engage adaptation] [Internet]. Paris, France; 2021 [cited 2022 Jan 14]. Available from: https://www.hautconseilclimat.fr/publications/rapport-annuel-2021-renforcer-lattenuation-engager- ladaptation/ 11. Mueller N, Rojas-Rueda D, Cole-Hunter T, de Nazelle A, Dons E, Gerike R, et al. Health impact assessment of active transportation: A systematic review. Prev Med. 2015 Jul 1;76:103–14. 12. Fishman E, Schepers P, Kamphuis CBM. Dutch Cycling: Quantifying the Health and Related Economic Benefits. Am J Public Health. 2015 Aug;105(8):e13-15. 13. Service de la donnée et des études statistiques (SDES). Résultats détaillés de l’enquête mobilité des personnes 2019 [Internet]. Données et études statistiques pour le changement climatique, l’énergie, l’environnement, le logement, et les transports. [cited 2022 Mar 17]. Available from: https://www.statistiques.developpement-durable.gouv.fr/resultats-detailles-de-lenquete-mobilite- des-personnes-de-2019 14. Egiguren J, Nieuwenhuijsen MJ, Rojas-Rueda D. Premature Mortality of 2050 High Bike Use Scenarios in 17 Countries. Environ Health Perspect. 2021 Dec;129(12):127002. 15. Rojas-Rueda D, de Nazelle A, Teixidó O, Nieuwenhuijsen MJ. Health impact assessment of increasing public transport and cycling use in Barcelona: a morbidity and burden of disease approach. Prev Med. 2013 Nov;57(5):573–9. 16. Monninkhof EM, Elias SG, Vlems FA, van der Tweel I, Schuit AJ, Voskuil DW, et al. Physical Activity and Breast Cancer: A Systematic Review. Epidemiology. 2007 Jan;18(1):137. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 16 17. Harriss DJ, Atkinson G, Batterham A, George K, Tim Cable N, Reilly T, et al. Lifestyle factors and colorectal cancer risk (2): a systematic review and meta-analysis of associations with leisure-time physical activity. Colorectal Dis. 2009;11(7):689–701. 18. Hamer M, Chida Y. Walking and primary prevention: a meta-analysis of prospective cohort studies. Br J Sports Med. 2008 Apr 1;42(4):238–43. 19. Hamer M, Chida Y. Physical activity and risk of neurodegenerative disease: a systematic review of prospective evidence. Psychol Med. 2009 Jan;39(1):3–11. 20. Jeon CY, Lokken RP, Hu FB, van Dam RM. Physical Activity of Moderate Intensity and Risk of Type 2 Diabetes: A systematic review. Diabetes Care. 2007 Mar 1;30(3):744–52. 21. Kelly P, Kahlmeier S, Götschi T, Orsini N, Richards J, Roberts N, et al. Systematic review and meta- analysis of reduction in all-cause mortality from walking and cycling and shape of dose response relationship. Int J Behav Nutr Phys Act. 2014 Oct 24;11:132. 22. Projections de population 2005-2050 pour la France métropolitaine. Insee Résultats [Internet]. 2006; Available from: https://www.insee.fr/fr/statistiques/2530035?sommaire=2530059 23. Götschi T, Kahlmeier S, Castro A, Brand C, Cavill N, Kelly P, et al. Integrated Impact Assessment of Active Travel: Expanding the Scope of the Health Economic Assessment Tool (HEAT) for Walking and Cycling. Int J Environ Res Public Health. 2020 Jan;17(20):7361. 24. Murray CJL, Aravkin AY, Zheng P, Abbafati C, Abbas KM, Abbasi-Kangevari M, et al. Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet. 2020 Oct 17;396(10258):1223–49. 25. Dervaux B, Rochaix L, Meurisse B. L’évaluation socioéconomique des effets de santé des projets d’investissement public [Internet]. France: CGDD, France Stratégie, SGPI; 2022 [cited 2022 May 2]. Available from: https://www.strategie.gouv.fr/publications/levaluation-socioeconomique-effets-de- sante-projets-dinvestissement-public-0 26. Quinet E. Cost benefit assessment of public investments. Final Rep Summ Recomm Paris Fr CGSP Policy Plan Comm [Internet]. 2013; Available from: https://www.strategie.gouv.fr/sites/strategie.gouv.fr/files/archives/CGSP_Evaluation_socioeconomiq ue_17092013.pdf 27. Lumley T. survey: analysis of complex survey samples. 2020. 28. Rubin DB. Multiple imputation for nonresponse in surveys. Vol. 81. John Wiley & Sons; 2004. 29. Leroutier M, Quirion P. Tackling car emissions in urban areas: Shift, Avoid, Improve. SocArXiv [Internet]. 2022; Available from: https://osf.io/preprints/socarxiv/f5kmd/ 30. Castro A, Kahlmeier S, Götschi T. Exposure-Adjusted Road Fatality Rates for Cycling and Walking in European Countries [Internet]. ITF. 2018 [cited 2022 Dec 13]. Available from: https://www.itf- oecd.org/exposure-adjusted-road-fatality-rates-cycling-and-walking-european-countries . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 17 31. Center for Transport Analytics, Transport DTU. The Danish National Travel Survey (TU) - Latest dataset [Internet]. [cited 2022 Jan 14]. Available from: https://www.cta.man.dtu.dk/english/national- travel-survey/latest_dataset 32. Rojas-Rueda D, de Nazelle A, Teixidó O, Nieuwenhuijsen MJ. Replacing car trips by increasing bike and public transport in the greater Barcelona metropolitan area: a health impact assessment study. Environ Int. 2012 Nov 15;49:100–9. 33. Ried-Larsen M, Rasmussen MG, Blond K, Overvad TF, Overvad K, Steindorf K, et al. Association of Cycling With All-Cause and Cardiovascular Disease Mortality Among Persons With Diabetes: The European Prospective Investigation Into Cancer and Nutrition (EPIC) Study. JAMA Intern Med. 2021 Sep 1;181(9):1196–205. 34. Observatoire national interministériel de la sécurité routière (ONISR). Bilan 2020 de la sécurité routière [Internet]. 2021. Available from: https://www.onisr.securite-routiere.gouv.fr/etat-de-l- insecurite-routiere/bilans-annuels-de-la-securite-routiere/bilan-2020-de-la-securite-routiere 35. Fragnol L. Réduction des vitesses sur les routes - Analyse coûts bénéfices. THEMA Anal - Ministère Transit Écologique [Internet]. 2018; Available from: https://www.ecologique- solidaire.gouv.fr/sites/default/files/Th%C3%A9ma%20- %20R%C3%A9duction%20des%20vitesses%20sur%20les%20routes.pdf 36. Bonaldi C, Hill C. Alcohol-attributable mortality in France in 2015. Bull Épidémiologique Hebd. 2019;(5/6):97–108. 37. Domergue S, Vermont B. Étude d’impact du crédit d’impôt pour la transition énergétique (CITE). Commissariat général au développement durable; 2018. 38. Donaire-Gonzalez D, Nazelle A de, Cole-Hunter T, Curto A, Rodriguez DA, Mendez MA, et al. The Added Benefit of Bicycle Commuting on the Regular Amount of Physical Activity Performed. Am J Prev Med. 2015 Dec 1;49(6):842–9. 39. Brondeel R, Wasfi R, Perchoux C, Chaix B, Gerber P, Gauvin L, et al. Is older adults’ physical activity during transport compensated during other activities? Comparing 4 study cohorts using GPS and accelerometer data. J Transp Health. 2019 Mar 1;12:229–36. 40. Zhao Y, Hu F, Feng Y, Yang X, Li Y, Guo C, et al. Association of Cycling with Risk of All-Cause and Cardiovascular Disease Mortality: A Systematic Review and Dose–Response Meta-analysis of Prospective Cohort Studies. Sports Med. 2021 Jul 1;51(7):1439–48. 41. Barban P, De Nazelle A, Chatelin S, Quirion P, Jean K. Assessing the Health Benefits of Physical Activity Due to Active Commuting in a French Energy Transition Scenario. Int J Public Health [Internet]. 2022 [cited 2022 Sep 2];0. Available from: https://www.ssph- journal.org/articles/10.3389/ijph.2022.1605012/full 42. Jarrett J, Woodcock J, Griffiths UK, Chalabi Z, Edwards P, Roberts I, et al. Effect of increasing active travel in urban England and Wales on costs to the National Health Service. Lancet Lond Engl. 2012 Jun 9;379(9832):2198–205. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 18 43. Bouscasse H, Gabet S, Kerneis G, Provent A, Rieux C, Ben Salem N, et al. Designing local air pollution policies focusing on mobility and heating to avoid a targeted number of pollution-related deaths: Forward and backward approaches combining air pollution modeling, health impact assessment and cost-benefit analysis. Environ Int. 2021 Dec 7;159:107030. 44. Kraus S, Koch N. Provisional COVID-19 infrastructure induces large, rapid increases in cycling. Proc Natl Acad Sci U S A. 2021 Apr 13;118(15):e2024399118. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint 19 Supplementary material Supplementary table 1: Relative risks relating physical activity and several chronic diseases as identified in a previous systematic review (Rojas-Rueda et al, 2013). Morbi- mortality event

RR (95% CI) Unit RR scaled for 100 min cycling, ie 11.25 MET.hours (95% CI) Breast Cancer (women) Monninkhof et al., 2007 0.94 (0.92, 0.97) For each additional hour per week 0.90 (0.87-0.95) Colon Cancer Harris et al., 2009 Males: 0.80 (0.67, 0.96) Females: 0.86 (0.76, 0.98) Men: Per 30,1 METs per week Women: Per 30,9 METs per week Males: 0.93 (0.88-0.99) Females: 0.94 (0.91- 0.99) Cardiovascular Disease Hamer & Chida, 2008 0.84 (0.79,0.90) 3 h per week of physical activity of moderate intensity 0.92 (0.90-0.95) Dementia Hamer & Chida, 2009 0.72 (0.60, 0.86) 33 METs per week 0.90 (0.86-0.95) Type 2 Diabetes Jeon et al., 2007 0.83 (0.75, 0.91) Per 10 METs per week 0.81 (0.72-0.89) RR: relative risk; CI: confidence interval. The scaled RR was obtained using the formula: RRscaled = 1-(1-RRref)*(11.25*Unitref). . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 11, 2023. ; https://doi.org/10.1101/2023.04.04.23288155doi: medRxiv preprint