You are here

Article Text

Article menu

Download PDFPDF

XML
Epidemiology
Research
The relationship between bicycle commuting and perceived stress: a cross-sectional study
  1. Ione Avila-Palencia1,2,3,
  2. Audrey de Nazelle4,
  3. Tom Cole-Hunter5,
  4. David Donaire-Gonzalez1,3,6,
  5. Michael Jerrett7,
  6. Daniel A Rodriguez8,
  7. Mark J Nieuwenhuijsen1,2,3
  1. 1 ISGlobal, Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
  2. 2 Department of Experimental and Health Sciences, Universitat Pompeu Fabra (UPF), Barcelona, Spain
  3. 3 CIBER Epidemiología y Salud Pública (CIBERESP), Barcelona, Spain
  4. 4 Centre for Environmental Policy, Imperial College of London, London, United Kingdom
  5. 5 Department of Environmental and Radiological Health Sciences, Colorado State University, Fort Collins, Colorado, USA
  6. 6 Physical Activity and Sports Sciences Department, Fundació Blanquerna, Ramon Llull University, Barcelona, Spain
  7. 7 Department of Environmental Health Sciences, University of California, Los Angeles, California, USA
  8. 8 Department of City and Regional Planning, University of California, Berkeley, California, USA
  1. Correspondence to Ione Avila-Palencia; ione.avila{at}isglobal.org

Abstract

Introduction Active commuting — walking and bicycling for travel to and/or from work or educational addresses — may facilitate daily, routine physical activity. Several studies have investigated the relationship between active commuting and commuting stress; however, there are no studies examining the relationship between solely bicycle commuting and perceived stress, or studies that account for environmental determinants of bicycle commuting and stress. The current study evaluated the relationship between bicycle commuting, among working or studying adults in a dense urban setting, and perceived stress.

Methods A cross-sectional study was performed with 788 adults who regularly travelled to work or study locations (excluding those who only commuted on foot) in Barcelona, Spain. Participants responded to a comprehensive telephone survey concerning their travel behaviour from June 2011 through to May 2012. Participants were categorised as either bicycle commuters or non-bicycle commuters, and (based on the Perceived Stress Scale, PSS-4) as either stressed or non-stressed. Multivariate Poisson regression with robust variance models of stress status based on exposures with bicycle commuting were estimated and adjusted for potential confounders.

Results Bicycle commuters had significantly lower risk of being stressed than non-bicycle commuters (Relative Risk; RR (95% CI)=0.73 (0.60 to 0.89), p=0.001). Bicycle commuters who bicycled 4 days per week (RR (95% CI)=0.42 (0.24 to 0.73), p=0.002) and those who bicycled 5 or more days per week (RR (95% CI)=0.57 (0.42 to 0.77), p<0.001) had lower risk of being stressed than those who bicycled less than 4 days. This relationship remained statistically significant after adjusting for individual and environmental confounders and when using different cut-offs of perceived stress.

Conclusions Stress reduction may be an important consequence of routine bicycle use and should be considered by decision makers as another potential benefit of its promotion.

  • Bicycling
  • Commuting
  • Physical activity
  • Stress
  • Survey

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/

https://doi.org/10.1136/bmjopen-2016-013542

Statistics from Altmetric.com

    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.

    Strengths and limitations of this study

    • The study had high internal validity, with a good representation of bicycle commuters.

    • The study was conducted in Barcelona (a dense, Mediterranean/Southern European city), adding evidence on these issues in a different context than the current literature.

    • The Transportation, Air Pollution and Physical ActivitieS (TAPAS) Travel Survey sample is representative of Barcelona’s population, taking into account home-neighbourhood deprivation, and home and work-neighbourhood population density.

    • The study used a cross-sectional design, which is not well suited to assess the direction of causation.

    • Using questionnaire data, we could have misclassification error (information bias) of bicycle commuting and physical activity due to the data being self-reported.

    Introduction

    Walking and bicycling for transport are increasingly being promoted due to their potential for increasing physical activity (PA) levels in the general population.1–3 Active commuting — walking and bicycling for travel to and/or from work or educational addresses — has been associated with multiple health benefits, from reductions of cardiovascular risk,4 5 lowering of body weight,2 5 improvement of fitness, reduction of diabetes risk,3 and increasing levels of physical and mental well-being.6 7 Specifically, bicycle commuting has been inversely associated with all-cause mortality among both men and women in all age groups,8 and it seems likely to improve the health-related quality of life in previously untrained healthy adults.9 Active commuting has been shown to have other societal benefits, such as helping reduce air pollution, greenhouse gas emissions and noise, and improving social interaction.10

    Perceived stress is a global and comprehensive stress construct that refers to the interaction between the individual and the environment in the presence of a stressor.11 The perception of an event as being stressful can result in a range of physiological, behavioural and psychological changes, and can lead to cardiovascular disease, increased negative affect, lowered self-esteem and lowered feelings of control. Hence, it is possible that mental health outcomes such as anxiety disorders and depression can be manifestations of chronic, perceived stress.12 Furthermore, others have suggested gender differences in stress-related variables. Women seem to be more likely to have daily stress, being more physiologically reactive to social rejection challenges13 and be more impacted by life events.14

    Some literature recognises commuting as a potential source of stress15; however, active commuters have been shown to have higher levels of satisfaction, lower stress, higher relaxation and a heightened sense of freedom compared with car drivers.16–18 Recent qualitative research has suggested that commuting can be perceived as a relaxing or transitional time between home and work life, which can also be about enjoying pleasant landscape, nature and wildlife.19 Emerging literature has highlighted the relevance of a positive natural and built environment to increase bicycle commuting and to improve mental health outcomes. Bicycle lane connectivity, bikeability, separation of bicycling from other traffic, high population density, short trip distance, proximity of a cycle path, green space and also walkability have been suggested as determinants of bicycling.20–24 Green space has also been associated with better self-perceived general health and better mental health.25 26

    Several studies have examined the relationship between active commuting and commuting stress (stress directly related with the act of commuting),17 18 27 28 but none of them have studied the relationship between solely bicycle commuting and perceived stress (global and comprehensive stress construct) in adults, nor taking into account environmental determinants. Moreover, most studies of active commuting and its beneficial effects on mental health have been conducted in North America, where the urban design tends to be less dense than many parts of the world, or Northwest Europe.6 7 17 28–30 Consequently, a need exists to understand the relationship between bicycle commuting and perceived stress, particularly in dense, Mediterranean/Southern European urban environments.

    The current study aimed to evaluate the relationship between bicycle commuting and perceived stress among the working or studying adult population of a dense, Mediterranean/Southern European urban setting.

    Materials and methods

    Study population

    This cross-sectional study was based on participants from the Transportation, Air Pollution and Physical ActivitieS (TAPAS) Travel Survey. TAPAS is a relatively large study aimed at investigating the risks and benefits of active commuting. Participant recruitment was conducted by trained interviewers on the streets of Barcelona city between June 2011 and May 2012. To ensure adequate geographical coverage, a total of 40 random points (four random points within each of the ten city districts across Barcelona) were sampled. Adult bicycle commuters and non-bicycle commuters were asked in the street to answer a few screening questions, and those who fulfilled the inclusion criteria (being older than 18 years of age; living in Barcelona city since 2006 or earlier; working or going to school in Barcelona city; being healthy enough to ride a bicycle for 20 min; having a commute distance greater than a 10 min walk; and using at least one mode of transport other than walking to commute) were invited to respond to a telephone survey. Bicycle commuters were oversampled to ensure enough bicycle commuters in the study. Those solely commuting on foot were excluded as the main interest was in the contrast between motorised modes (private and public transportation) and the bicycle. Of the 18 469 participants approached across the 40 sampling random points, 6701 agreed to answer screening questions. Of these, 1508 met the inclusion criteria, and 871 participants completed the survey. After survey responses were checked by the research team, 815 still fulfilled the inclusion criteria and 789 had geocodable home address. After excluding one PA outlier (total of all walking, moderate and vigorous time variables >960 min/day), 788 participants remained. Further details on the recruitment are given elsewhere.31

    The study protocol was approved by the Clinical Research Ethical Committee of the Parc de Salut Mar (CEIC-Parc de Salut Mar), and written informed consent was obtained from all participants.

    Bicycle commuting

    The TAPAS Travel Survey assessed the regular use of transport modes,32 including bicycles.33 Participants who indicated using a bicycle (private or from a public bicycle-sharing system) to go to work or school at least once the week prior to survey administration were classified as ‘bicycle commuters’. Participants who did not commute by bicycle in the week prior to survey administration were classified as ‘non-bicycle commuters’.

    As part of the sensitivity analyses, commuting behaviour was further classified according to bicycle commuting levels and bicycle commuting propensity.24 Classification of bicycle commuting levels was based on the days of bicycle commuting in the week prior to survey administration: ‘low-level’ being 3 days or fewer, ‘medium-level’ for 4 days and ‘high-level’ for 5 or more days. This measure could be interpreted as a proxy of bicycle commuting frequency. Bicycle commuting propensity classification took into account both frequency and willingness to commute by bicycle: the ‘bicycle commuters’ were further classified as ‘frequent’ (4 or more days) or ‘infrequent’ (3 or less days), and the ‘non-bicycle commuters’ were classified as ‘willing’ or ‘unwilling’. The ‘willing’ group were those ‘non-bicycle commuters’ who indicated bicycling as ‘never or nearly never’ their general transport mode, but who also indicated that they would consider bicycle commuting in Barcelona (by answering positively to ‘considering costs, travelling time, comfort and safety, how ready would you be to use the bicycle/Bicing (public bicycle-sharing system) for your trip to work or education centre?’). The ‘unwilling’ group were those ‘non-bicycle commuters’ who indicated ‘never or nearly never’ bicycling for travel and indicated that they would not consider bicycle commuting in Barcelona by answering negatively to the above question. More details of the bicycle commuting propensity classification are given elsewhere.24 This measure was included in the analysis to assess the effect on perceived stress by being willing to commute by bicycle.

    Perceived stress

    The last four questions of the TAPAS Travel Survey were the short version of the Perceived Stress Scale (PSS-4),11 which is a well-validated psychological instrument to measure the degree to which situations in one’s life over the past month are appraised as stressful. The instrument contains four statements, which measure how unpredictable, uncontrollable and overloaded respondents feel that their lives are (online supplementary table S1). The higher the score on the PSS-4 (from 0 to 16), the greater the respondent perceives that their demands exceed their ability to cope. There are no cut-off scores. Instead, an individual’s score is compared with a normative value.34 In the TAPAS Travel Survey, the 5-point Likert scale was modified to a 4-point Likert scale, removing the midpoint option for consistency with other questions in the survey (using a 4-point Likert scale). The sample did not have high levels of perceived stress (online supplementary table S2); therefore, for an easier interpretation, participants with a PSS-4 score higher than 3 (median of the total sample) were classified as ‘stressed’, and those equal or lower than 3 were classified as ‘non-stressed’. The sensitivity of our results to this choice was examined further in sensitivity analyses by classifying the respondents with PSS-4 scores in the 75th percentile (P75) and above (a score higher than 4) and in the 90th percentile (P90) and above (a score of 6 and above) as stressed and all others as non-stressed.

    Supplementary Material

    Supplementary material 1

    Other explanatory measures

    Individual determinants of bicycle commuting and perceived stress such as PA levels,35 sociodemographic variables and work or school addresses were also derived from the TAPAS Travel Survey to be used as potential confounders. In addition, the MEDEA Index (Mortalidad en áreas pequeñas Españolas y Desigualdades socioEconómicas y Ambientales, in Spanish; Environmental and socioEconomic Inequalities in Mortality in small Spanish areas, translated to English) was used as an area deprivation indicator assigned to each participant’s address. MEDEA measures deprivation at the census tract level based on five domains, namely percentage of manual workers, temporary workers, total population with low education, young population with low education and unemployment.36

    Environmental determinants of bicycle commuting and perceived stress within a 400 m buffer surrounding home and work/study addresses, and a route-by-area surrounding predicted commute routes, were calculated to be used as potential confounders too. The number of public bicycle stations within a 400 m buffer surrounding home and work/study addresses was calculated based on information from the Ajuntament de Barcelona – Informació de Base i Cartografia (Barcelona City Council – Basic information and mapping). Greenness was calculated as the interquartile range of the mean of Normalised Difference Vegetation Index via satellite imagery (LANDSAT 4 and 5, NASA). The mean NO2 levels were estimated using a land-use regression model developed for a previous project.37 Noise was calculated as the proportion of street length above a 55 dB(A) threshold.38 A bikeability index was calculated taking into account five factors shown to influence bicycling: bicycle facility availability, bicycle facility quality, street connectivity, topography and land use.39 Commute distance did not use buffers and it was calculated in kilometres following the street network of the shortest route from home address to work address. Further details of the environmental determinants calculation are given elsewhere.24

    Statistical analyses

    Descriptive univariate analyses were done for all study variables. Descriptive bivariate analyses were done using χ2 and Mann-Whitney U tests to assess the relationship between determinants and bicycle commuting variables (bicycle commuting status, bicycle commuting levels and bicycle commuting propensity), and using Poisson regression with robust variance models to assess the relationship between determinants and perceived stress. A generalised additive model was used to test linearity between perceived stress and total physical activity (total PA), moderate-to-vigorous physical activity (MVPA), vigorous physical activity (VPA) and age.40 As there was no statistical evidence to reject linearity between perceived stress and total PA (p=0.382), MVPA (p=0.503), VPA (p=0.163) and age (p=0.228), these variables were included as continuous variables in the models assuming a linear relationship. Multivariate Poisson regression with robust variance models were used to assess the relationship between bicycle commuting and perceived stress. Four regression models were done: (1) unadjusted; (2) adjusted by individual determinants that showed a p value <0.05 in the model; (3) adjusted by the individual determinants of model 2, as well as those found to be statistically significant within previous literature; and (4) adjusted by the individual determinants of model 3 and environmental determinants that showed a p value <0.05 in the bivariate analyses. All multivariate regression models were conducted with a complete case analysis. Possible mediation by different levels of PA between bicycle commuting and perceived stress, and any interaction between gender and bicycle commuting, were also tested with Poisson regression with robust variance models. The first descriptive statistical analyses (univariate, χ2 and Mann-Whitney U tests) were conducted in Stata SE V.12, while Poisson regression with robust variance models were conducted in Stata SE V.14.

    Results

    The included sample had an equal distribution of genders and the median age (P25–P75) was 36 (29–43) years (table 1). The majority of participants were non-stressed (having a PSS-4 score of 3 or lower), were Spanish, had completed university studies or equivalent-level education, were living with their family or partner, with at least two employed people and were not with children (64.34%). Among those living with children, ~8% had children younger than 3 years of age. The sample had positive self-perception of health (with only <1% of participants self-perceiving bad or very bad health), healthy weight (body mass index of 18.5–24.9; 71.12%) and generally no chronic disease (92.26%). The majority of participants considered that they could release stress when riding a bicycle and that they enjoyed their trip more if they used a bicycle. Compared with non-bicycle commuters, bicycle commuters were statistically significantly (p<0.05) more likely to be non-stressed; younger (35 years); men; have higher levels of PA; possess a university or equivalent-level education; live alone and/or with flat mates; live with one or less employed people; live with no children; and have better self-perception of health, and healthy weight, but more chronic diseases. Bicycle commuters had shorter commutes compared with non-bicycle commuters, and we observed a gradient between commute distance and bicycle commuting levels; shorter distances were travelled for those who bicycle commuted more frequently. This tendency was also seen with bicycle commuting propensity; shorter distances were travelled for those more willing to bicycle commute (online supplementary table S3). Bicycle commuters also had more public bicycle stations around the home and work/study addresses, less greenness around the home address, and higher levels of bikeability at home, work/study address and on the commute route compared with non-bicycle commuters (table 1). These environmental determinants stayed statistically significant for bicycle commuting propensity, but not between bicycle commuting levels (online supplementary table S3).

    View this table:
    Table 1

    Descriptive analyses of participant perceived stress and its determinants as a total sample and according to bicycle commuting status

    Participants more likely to be stressed (p<0.05) were typically women, non-Spanish, living with one or less people employed and had a chronic disease (table 2). For environmental determinants, participants who had more public bicycle stations around their work/study area and higher levels of bikeability in the work/study address area, as well as on the commute route, were less likely to be stressed (p<0.05). There was no statistically significant relationship between commute distance, greenness, NO2 and noise, and perceived stress. The possible mediation of PA was not further explored as there was no statistically significant relationship between levels of PA (total PA, MVPA and VPA) and perceived stress for the three different classifications of perceived stress (P50, P75, P90) (RR, Relative Risk; RR (95% CI): 1.00 (0.99 to 1.00), all p>0.10) (table 2, online supplementary table S4).

    View this table:
    Table 2

    Bivariate analyses of the relationship between participant determinants and perceived stress

    View this table:
    Table 3

    Multivariate models assessing the relationship between bicycle commuting and participant perceived stress (median)

    Multivariate Poisson regression with robust variance analyses showed a statistically significant inverse relationship between bicycle commuting and perceived stress. Considering the total sample, bicycle commuters had a lower risk of being stressed compared with non-bicycle commuters (model 1: RR (95% CI)=0.73 (0.60 to 0.89), p=0.001). This relationship remained statistically significant in the adjusted models (model 2: RR (95% CI)=0.75 (0.62 to 0.91), p=0.003; model 3: RR (95% CI)=0.77 (0.63 to 0.94), p=0.009; model 4: RR (95% CI)=0.80 (0.66 to 0.99), p=0.036) (table 3) and when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5). Regarding bicycle commuting levels in the total sample, those who bicycle commuted 4 days per week (considered ‘medium-level’ of bicycle commuters) and those who bicycled 5 or more days per week (‘high-level’) had lower risk of being stressed than non-bicycle commuters (‘Medium-level’ — model 1: RR (95% CI)=0.46 (0.28 to 0.78), p=0.004; ‘High-level’ — model 1: RR (95% CI)=0.63 (0.49 to 0.81), p<0.001). These relationships remained statistically significant in the adjusted models (‘Medium-level’ — model 2: RR (95% CI)=0.45 (0.27 to 0.74), p=0.002; model 3: RR (95% CI)=0.45 (0.27 to 0.75), p=0.002; model 4: RR (95% CI)=0.48 (0.29 to 0.80), p=0.005; ‘High-level’ — model 2: RR (95% CI)=0.66 (0.51 to 0.85), p=0.001; model 3: RR (95% CI)=0.68 (0.52 to 0.88), p=0.003; model 4: RR (95% CI)=0.71 (0.54 to 0.92), p=0.010) (table 3) and in the majority of perceived stress sensitivity analyses (using cut-offs of P75 and P90), with the exception of the unadjusted and fully adjusted models (models 1 and 4) for ‘medium-level’ bicycle commuters using P90 as a perceived stress cut-off (‘Medium-level’ — model 1: RR (95% CI)=0.15 (0.02 to 1.05), p=0.056; model 4: RR (95% CI)=0.15 (0.02 to 1.04), p=0.054) (online supplementary table S5). Regarding bicycle commuting propensity in the total sample, ‘frequent’ bicycle commuters had lower risk of being stressed than ‘unwilling’ non-bicycle commuters (model 1: RR (95% CI)=0.53 (0.41 to 0.67), p<0.001). This relationship remained statistically significant in the adjusted models (model 2: RR (95% CI)=0.55 (0.43 to 0.70), p<0.001; model 3: RR (95% CI)=0.56 (0.43 to 0.72), p<0.001; model 4: RR (95% CI)=0.58 (0.45 to 0.76), p<0.001) (table 3) and when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5). Also, ‘willing’ non-bicycle commuters had lower risk of being stressed than ‘unwilling’ non-bicycle commuters (model 1: RR (95% CI)=0.72 (0.56 to 0.94), p=0.014). This relationship remained statistically significant in the adjusted models (model 2: RR (95% CI)=0.75 (0.58 to 0.97), p=0.029; model 3: RR (95% CI)=0.74 (0.57 to 0.96), p=0.022; model 4: RR (95% CI)=0.75 (0.58 to 0.97), p=0.031) (table 3), but not when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5).

    When considering bicycle commuting levels within the bicycle commuters only sample, ‘medium-level’ and ‘high-level’ bicycle commuters had lower risk of being stressed than ‘low-level’ bicycle commuters (‘Medium-level’ — model 1: RR (95% CI)=0.42 (0.24 to 0.73), p=0.002; ‘High-level’ — model 1: RR (95% CI)=0.57 (0.42 to 0.77), p<0.001). These relationships remained statistically significant in the adjusted models ((‘Medium-level’ — model 2: RR (95% CI)=0.39 (0.23 to 0.67), p=0.001; model 3: RR (95% CI)=0.39 (0.23 to 0.65), p<0.001; model 4: RR (95% CI)=0.38 (0.23 to 0.65), p<0.001; ‘High-level’ — model 2: RR (95% CI)=0.59 (0.44 to 0.80), p=0.001; model 3: RR (95% CI)=0.59 (0.44 to 0.80), p=0.001; model 4: RR (95% CI)=0.59 (0.44 to 0.80), p=0.001) (table 3) and when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5). Regarding bicycle commuting propensity, ‘frequent’ bicycle commuters had lower risk of being stressed than ‘infrequent’ bicycle commuters (model 1: RR (95% CI)=0.54 (0.40 to 0.72), p<0.001). This relationship remained statistically significant in the adjusted models (model 2: RR (95% CI)=0.55 (0.41 to 0.73), p<0.001; model 3: RR (95% CI)=0.54 (0.41 to 0.72), p<0.001; model 4: RR (95% CI)=0.54 (0.41 to 0.72), p<0.001) (table 3) and when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5).

    Considering bicycle commuting propensity within the non-bicycle commuters only sample, ‘willing’ non-bicycle commuters had lower risk of being stressed than ‘unwilling’ non-bicycle commuters (model 1: RR (95% CI)=0.72 (0.56 to 0.94), p=0.015). This relationship remained statistically significant in the adjusted models (model 2: RR (95% CI)=0.73 (0.57 to 0.95), p=0.020; model 3: RR (95% CI)=0.72 (0.56 to 0.93), p=0.013; model 4: RR (95% CI)=0.74 (0.57 to 0.95), p=0.020) (table 3), but not when using perceived stress cut-offs of either P75 or P90 (online supplementary table S5).

    In the fully adjusted models (model 4), we found no statistically significant interactions between gender and being a bicycle commuter (p=0.165), between gender and bicycle commuting levels (p=0.226, p=0.266, p=0.431) or between gender and bicycle commuting propensity (p=0.982, p=0.197, p=0.277) (results not shown).

    Discussion

    Summary of results

    We evaluated relationships between bicycle commuting and perceived stress while adjusting for several confounders in a representative sample of adults in Barcelona, Spain. We found statistically significant inverse relationships between several measures of bicycle commuting and perceived stress. Bicycle commuters who bicycled 4 or more days per week had lower risk of being stressed compared with those who bicycled less or did not bicycle commute at all. This relationship remained statistically significant in all sensitivity analyses and after controlling for individual and environmental confounders.

    Comparison with previous studies

    To our knowledge, this study is the first to assess whether a relationship exists between solely bicycle commuting and perceived stress. A few studies have focused on the relationship between active commuting (as a combination of both walking and bicycling) and mental health,6 7 29 but the relationship is still unclear. One study found a positive association of active commuting with well-being in adults,6 and another with better mental health in only men.29 Meanwhile, Humphreys and colleagues7 found a positive relationship between time spent actively commuting and levels of physical well-being, but not with mental well-being. The relationship between PA and mental health has been studied more. It has been suggested that PA can reduce stress and anxiety on a daily basis while improving positive self-perception and mood,41–43 and it has been associated with lower depressive symptomatology and greater emotional well-being.44 These findings suggest that the PA gained during bicycle commuting31 may act as a mediator in the relationship between bicycle commuting and perceived stress. Our results are consistent with the general idea that active commuting is associated with better mental health, but in our case PA did not act as a mediator in this relationship. Our sample was composed of young, healthy and active participants with relatively low PSS-4 levels of perceived stress, which might have led to an underestimation of the relationship between PA and perceived stress.

    Qualitative research has suggested that choice of travel mode may affect well-being.19 The quantity of public bicycle (Bicing) stations and the amount of greenness have been related to bicycle commuting propensity,24 which could imply that commuting by bicycle provides people with more opportunities to ‘enjoy’ or ‘experience’ greenness than commuting in public transport or a car. At the same time, the availability of green space close to one’s home has been shown to be related to better self-perceived general health and better mental health.25 26 45 Therefore, it seems that perceptual and environmental factors related to bicycle commuting could affect perceived stress, in the way that the more pleasant an environment to bicycle commuting is, the lower the perceived stress we will get. This general idea is consistent with our results, which show an inverse relationship between perceived stress and bicycle-friendly environments (public bicycle stations and bikeability levels) in work/study address area and the commute route. Importantly, the relationship between bicycle commuting and perceived stress was unchanged after controlling for environmental confounders. Our results also showed that general attitude might have a role in this relationship, as we have seen that non-bicycle commuters willing to start bicycle commuting, compared with those unwilling, had lower risk of being stressed. This becomes unclear, however, as the relationship approaches statistical non-significance in sensitivity analyses.

    Limitations and strengths

    Our study has some limitations. First, our study used a cross-sectional design, which is not well suited to assess the direction of causation, and we cannot exclude reverse causality or residual confounding. It has been suggested that stressed people can engage in unhealthy behaviours, such as poor dietary practices or a lack of PA.46 This reasoning could be applied to a behaviour like bicycle commuting, where those individuals who are more stressed would bicycle less. Second, our measurement method may be prone to information bias. With the questionnaire data we could have random misclassification error of bicycle commuting and PA due to the data being self-reported. Therefore, the risk estimate and also the potential mediation by PA could be an underestimation.47 The TAPAS Travel Survey only measured levels of PA without differentiating between travel PA (being most accurate for commute studies) and other types of PA (work, recreational). Furthermore, the modification of the 5-point PSS-4 Likert scale into a 4-point Likert scale could incorrectly estimate the perceived stress.

    This study has several strengths, too. The study has high internal validity, with a good representation of bicycle commuters. Related to participants’ characteristics (sociodemographics), the TAPAS Travel Survey sample is representative of Barcelona’s population. It was compared with data from the Catalan government’s Barcelona Active Population Survey (Statistics and information service, Catalan government 2011), and no statistically significant differences between participants’ neighbourhood deprivation index and home and work population density in both surveys were found.24 31 Finally, our study in a southern European city has added evidence on these issues in a different context from the current literature.

    Future research

    Our findings underscored the need for future research. There is a need to obtain a clear understanding of the relationship between bicycle commuting and perceived stress in longitudinal studies. The role of PA in this relationship seems unclear, and it is likely that other factors could affect the relationship between these two variables, especially those related to environmental determinants and personal attitudes. Further work related to determinants and mediators of bicycle commuting and perceived stress is needed.

    Conclusions

    We found that healthy, adult bicycle commuters had lower risk of being stressed than commuters of other transport modes. Also, bicycle commuters who bicycled 4 or more days per week had a lower risk of being stressed than those who bicycled less than that. Environmental determinants such as the number of public bicycle stations and bikeability, and also personal attitudes, seem to have an influence on this relationship. Further research is needed in order to disentangle the relationship between bicycle commuting and perceived stress, and its determinants (individual and environmental) and potential mediators. Our findings suggest that decision makers may promote bicycle commuting as a daily routine to reduce stress levels and improve public health and well-being.

    Supplementary Material

    Supplementary material 2

    Acknowledgments

    ISGlobal is a member of the CERCA Programme, Generalitat de Catalunya. The authors are grateful to the participants of TAPAS Travel Survey and the ISGlobal (CREAL) technicians who recruited them and created geographical variables (MEDEA and environmental determinants). We would like to acknowledge the ESCAPE project and its contributors for the air quality data of Barcelona, as well as the Ajuntament de Barcelona for the noise model data and street map information of Barcelona.

    References

    1. 1.↵
      2. Faulkner GE ,
      3. Buliung RN ,
      4. Flora PK , et al
      . Active school transport, physical activity levels and body weight of children and youth: a systematic review. Prev Med 2009;48:3–8.doi:10.1016/j.ypmed.2008.10.017
      OpenUrlCrossRefPubMedWeb of Science
    2. 2.↵
      2. Wanner M ,
      3. Götschi T ,
      4. Martin-Diener E , et al
      . Active transport, physical activity, and body weight in adults: a systematic review. Am J Prev Med 2012;42:493–502.doi:10.1016/j.amepre.2012.01.030
      OpenUrlCrossRefPubMed
    3. 3.↵
      2. Saunders LE ,
      3. Green JM ,
      4. Petticrew MP , et al
      . What are the health benefits of active travel? A systematic review of trials and cohort studies. PLoS One 2013;8:e69912.doi:10.1371/journal.pone.0069912
    4. 4.↵
      2. Hamer M ,
      3. Chida Y
      . Active commuting and cardiovascular risk: a meta-analytic review. Prev Med 2008;46:9–13.doi:10.1016/j.ypmed.2007.03.006
      OpenUrlCrossRefPubMedWeb of Science
    5. 5.↵
      2. Xu H ,
      3. Wen LM ,
      4. Rissel C
      . The relationships between active transport to work or school and cardiovascular health or body weight: a systematic review. Asia Pac J Public Health 2013;25:298–315.doi:10.1177/1010539513482965
      OpenUrlCrossRefPubMed
    6. 6.↵
      2. Martin A ,
      3. Goryakin Y ,
      4. Suhrcke M
      . Does active commuting improve psychological wellbeing? longitudinal evidence from eighteen waves of the british Household Panel Survey. Prev Med 2014;69:296–303.doi:10.1016/j.ypmed.2014.08.023
      OpenUrlCrossRefPubMed
    7. 7.↵
      2. Humphreys DK ,
      3. Goodman A ,
      4. Ogilvie D
      . Associations between active commuting and physical and mental wellbeing. Prev Med 2013;57:135–9.doi:10.1016/j.ypmed.2013.04.008
      OpenUrl
    8. 8.↵
      2. Andersen LB ,
      3. Schnohr P ,
      4. Schroll M , et al
      . All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch Intern Med 2000;160:1621.doi:10.1001/archinte.160.11.1621
      OpenUrlCrossRefPubMedWeb of Science
    9. 9.↵
      2. de Geus B ,
      3. Van Hoof E ,
      4. Aerts I , et al
      . Cycling to work: influence on indexes of health in untrained men and women in Flanders. coronary heart disease and quality of life. Scand J Med Sci Sports 2008;18:498–510.doi:10.1111/j.1600-0838.2007.00729.x
      OpenUrlPubMed
    10. 10.↵
      2. de Nazelle A ,
      3. Nieuwenhuijsen MJ ,
      4. Antó JM , et al
      . Improving health through policies that promote active travel: a review of evidence to support integrated health impact assessment. Environ Int 2011;37:766–77.doi:10.1016/j.envint.2011.02.003
      OpenUrlCrossRefPubMedWeb of Science
    11. 11.↵
      2. Cohen S ,
      3. Kamarck T ,
      4. Mermelstein R
      . A global measure of perceived stress. J Health Soc Behav 1983;24:385–96.doi:10.2307/2136404
      OpenUrlCrossRefPubMedWeb of Science
    12. 12.↵
      2. Katsarou AL ,
      3. Triposkiadis F ,
      4. Panagiotakos D
      . Perceived stress and vascular disease: where are we now? Angiology 2013;64:529–34.doi:10.1177/0003319712458963
      OpenUrlCrossRefPubMed
    13. 13.↵
      2. Stroud LR ,
      3. Salovey P ,
      4. Epel ES
      . Sex differences in stress responses: social rejection versus achievement stress. Biol Psychiatry 2002;52:318–27.doi:10.1016/S0006-3223(02)01333-1
      OpenUrlCrossRefPubMedWeb of Science
    14. 14.↵
      2. Matud MP
      . Gender differences in stress and coping styles. Pers Individ Dif 2004;37:1401–15.doi:10.1016/j.paid.2004.01.010
      OpenUrlCrossRef
    15. 15.↵
      2. Koslowsky M
      . Commuting stress: problems of definition and variable identification. Appl Psychol 1997;46:153–73.doi:10.1111/j.1464-0597.1997.tb01222.x
      OpenUrl
    16. 16.↵
      2. St-Louis E ,
      3. Manaugh K ,
      4. van Lierop D , et al
      . The happy commuter: a comparison of commuter satisfaction across modes. Transp Res Part F Traffic Psychol Behav 2014;26:160–70.doi:10.1016/j.trf.2014.07.004
      OpenUrl
    17. 17.↵
      2. LaJeunesse S ,
      3. Rodríguez DA ,
      4. a RD
      . Mindfulness, time affluence, and journey-based affect: exploring relationships. Transp Res Part F Traffic Psychol Behav 2012;15:196–205.doi:10.1016/j.trf.2011.12.010
      OpenUrlCrossRef
    18. 18.↵
      2. Anable J ,
      3. Gatersleben B
      . All work and no play? the role of instrumental and affective factors in work and leisure journeys by different travel modes. Transportation Research Part A: Policy and Practice 2005;39:163–81.doi:10.1016/j.tra.2004.09.008
      OpenUrlCrossRefWeb of Science
    19. 19.↵
      2. Guell C ,
      3. Ogilvie D
      . Picturing commuting: photovoice and seeking well-being in everyday travel. Qual Res 2015;15:201–18.doi:10.1177/1468794112468472
      OpenUrlCrossRef
    20. 20.↵
      2. Titze S ,
      3. Stronegger WJ ,
      4. Janschitz S , et al
      . Association of built-environment, social-environment and personal factors with bicycling as a mode of transportation among austrian city dwellers. Prev Med 2008;47:252–9.doi:10.1016/j.ypmed.2008.02.019
      OpenUrlCrossRefPubMedWeb of Science
    21. 21.↵
      2. Brown BB ,
      3. Smith KR ,
      4. Hanson H , et al
      . Neighborhood design for walking and biking: physical activity and body mass index. Am J Prev Med 2013;44:231–8.doi:10.1016/j.amepre.2012.10.024
      OpenUrlCrossRefPubMed
    22. 22.↵
      2. Fraser SD ,
      3. Lock K
      . Cycling for transport and public health: a systematic review of the effect of the environment on cycling. Eur J Public Health 2011;21:738–43.doi:10.1093/eurpub/ckq145
      OpenUrlCrossRefPubMed
    23. 23.↵
      2. Grasser G ,
      3. Van Dyck D ,
      4. Titze S , et al
      . Objectively measured walkability and active transport and weight-related outcomes in adults: a systematic review. Int J Public Health 2013;58:615–25.doi:10.1007/s00038-012-0435-0
      OpenUrl
    24. 24.↵
      2. Cole-Hunter T ,
      3. Donaire-Gonzalez D ,
      4. Curto A , et al
      . Objective correlates and determinants of bicycle commuting propensity in an urban environment. Transp Res D Transp Environ 2015;40:132–43.doi:10.1016/j.trd.2015.07.004
      OpenUrl
    25. 25.↵
      2. Triguero-Mas M ,
      3. Dadvand P ,
      4. Cirach M , et al
      . Natural outdoor environments and mental and physical health: relationships and mechanisms. Environ Int 2015;77:35–41.doi:10.1016/j.envint.2015.01.012
      OpenUrlCrossRefPubMed
    26. 26.↵
      2. Dadvand P ,
      3. Bartoll X ,
      4. Basagaña X , et al
      . Green spaces and General Health: roles of mental health status, social support, and physical activity. Environ Int 2016;91:161–7.doi:10.1016/j.envint.2016.02.029
      OpenUrl
    27. 27.↵
      2. Olsson LE ,
      3. Gärling T ,
      4. Ettema D , et al
      . Happiness and satisfaction with work commute. Soc Indic Res 2013;111:255–63.doi:10.1007/s11205-012-0003-2
      OpenUrlCrossRefPubMed
    28. 28.↵
      2. Gottholmseder G ,
      3. Nowotny K ,
      4. Pruckner GJ , et al
      . Stress perception and commuting. Health Econ 2009;18:559–76.doi:10.1002/hec.1389
      OpenUrlCrossRefPubMedWeb of Science
    29. 29.↵
      2. Ohta M ,
      3. Mizoue T ,
      4. Mishima N , et al
      . Effect of the physical activities in leisure time and commuting to work on mental health. J Occup Health 2007;49:46–52.doi:10.1539/joh.49.46
      OpenUrlCrossRefPubMedWeb of Science
    30. 30.↵
      2. Hansson E ,
      3. Mattisson K ,
      4. Björk J , et al
      . Relationship between commuting and health outcomes in a cross-sectional population survey in Southern Sweden. BMC Public Health 2011;11:834.doi:10.1186/1471-2458-11-834
    31. 31.↵
      2. Donaire-Gonzalez D ,
      3. de Nazelle A ,
      4. Cole-Hunter T , et al
      . The Added Benefit of Bicycle commuting on the regular amount of physical activity performed. Am J Prev Med 2015;49:842–9.doi:10.1016/j.amepre.2015.03.036
      OpenUrl
    32. 32.↵
      Institut d’Estudis Regionals i Metropolitans de Barcelona La Mobilitat quotidiana a Catalunya. .. Regió Metropolitana de Barcelona 2008.
    33. 33.↵
      2. Forsyth A ,
      3. Krizek KJ ,
      4. Agrawal AW , et al
      . Reliability testing of the pedestrian and bicycling survey (PABS) method. J Phys Act Health 2012;9:677–88.doi:10.1123/jpah.9.5.677
      OpenUrlPubMed
    34. 34.↵
      2. Warttig SL ,
      3. Forshaw MJ ,
      4. South J , et al
      . New, normative, English-sample data for the short form perceived stress Scale (PSS-4). J Health Psychol 2013;18:1617–28.doi:10.1177/1359105313508346
      OpenUrlCrossRefPubMed
    35. 35.↵
      2. Craig CL ,
      3. Marshall AL ,
      4. Sjöström M ,
      5. Sj??Str??m M , et al
      . International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003;35:1381–95.doi:10.1249/01.MSS.0000078924.61453.FB
      OpenUrlCrossRefPubMedWeb of Science
    36. 36.↵
      2. Domínguez-Berjón MF ,
      3. Borrell C ,
      4. Cano-Serral G , et al
      . [Constructing a deprivation index based on census data in large spanish cities(the MEDEA project)]. Gac Sanit 2008;22:179–87.
      OpenUrlPubMedWeb of Science
    37. 37.↵
      2. Beelen R ,
      3. Hoek G ,
      4. Vienneau D , et al
      . Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe – The ESCAPE project. Atmos Environ 2013;72:10–23.doi:10.1016/j.atmosenv.2013.02.037
      OpenUrlCrossRefWeb of Science
    38. 38.↵
      World Health Organisation. Guideline for Community Noise, 2011. http://www.who.int/docstore/peh/noise/Commnoise4.htm. (accessed 10 April 2015).
    39. 39.↵
      2. Winters M ,
      3. Brauer M ,
      4. Setton EM , et al
      . Mapping bikeability: a spatial tool to support sustainable travel. Environment and Planning B: Planning and Design 2013;40:865–83.doi:10.1068/b38185
      OpenUrl
    40. 40.↵
      2. Hastie T ,
      3. Tibshirani R
      . Generalized Additive models. Statistical Science 1986;1:297–310.doi:10.1214/ss/1177013604
      OpenUrlCrossRef
    41. 41.↵
      2. Fox KR
      . The influence of physical activity on mental well-being. Public Health Nutr 1999;2:411–8.doi:10.1017/S1368980099000567
      OpenUrlCrossRefPubMedWeb of Science
    42. 42.↵
      2. Sexton H ,
      3. Søgaard AJ ,
      4. Olstad R
      . How are mood and exercise related? results from the Finnmark study. Soc Psychiatry Psychiatr Epidemiol 2001;36:348–53.doi:10.1007/s001270170040
      OpenUrlCrossRefPubMedWeb of Science
    43. 43.↵
      2. Peluso MA ,
      3. Guerra de Andrade LH
      . Physical activity and mental health: the association between exercise and mood. Clinics 2005;60:61–70.doi:10.1590/S1807-59322005000100012
      OpenUrlCrossRefPubMed
    44. 44.↵
      2. Galper DI ,
      3. Trivedi MH ,
      4. Barlow CE , et al
      . Inverse association between physical inactivity and mental health in men and women. Med Sci Sports Exerc 2006;38:173–8.doi:10.1249/01.mss.0000180883.32116.28
      OpenUrlCrossRefPubMedWeb of Science
    45. 45.↵
      2. Maas J ,
      3. Verheij RA ,
      4. Groenewegen PP , et al
      . Green space, urbanity, and health: how strong is the relation? J Epidemiol Community Health 2006;60:587–92.doi:10.1136/jech.2005.043125
      OpenUrlAbstract/FREE Full Text
    46. 46.↵
      2. Stults-Kolehmainen MA ,
      3. Sinha R
      . The effects of stress on physical activity and exercise. Sports Med 2014;44:81–121.doi:10.1007/s40279-013-0090-5
      OpenUrlCrossRefPubMed
    47. 47.↵
      2. Baron RM ,
      3. Kenny DA ,
      4. a KD
      . The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173–82.doi:10.1037/0022-3514.51.6.1173
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Contributors MJN and AdN obtained the funding and designed the study. IAP conducted the analyses and drafted this version of the paper and received input from all the authors. All authors read and commented on the paper and agreed with the final version.

    • Competing interests None declared.

    • Patient consent Obtained.

    • Ethics approval Clinical Research Ethical Committee of the Parc de Salut Mar (CEIC-Parc de Salut Mar).

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

    • Data sharing statement Extra data are available by emailing the corresponding author (IAP: ione.avila@isglobal.org).