The association between commuter cycling and sickness absence

The association between commuter cycling and sickness absence

Preventive Medicine 51 (2010) 132–135 Contents lists available at ScienceDirect Preventive Medicine j o u r n a l h o m e p a g e : w w w. e l s e v...

175KB Sizes 0 Downloads 65 Views

Preventive Medicine 51 (2010) 132–135

Contents lists available at ScienceDirect

Preventive Medicine j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y p m e d

The association between commuter cycling and sickness absence Ingrid J.M. Hendriksen a,b,⁎, Monique Simons a,b,c, Francisca Galindo Garre a, Vincent H. Hildebrandt a,b a b c

TNO Quality of Life, Leiden, The Netherlands Body@Work, Research Center Physical Activity, Work and Health, TNO-VUmc, VU University Medical Center, Amsterdam, The Netherlands Department of Health Sciences and the EMGO Institute for Health and Care Research, Faculty of Earth and Life Sciences, VU University Amsterdam, The Netherlands

a r t i c l e

i n f o

Available online 24 May 2010 Keywords: Active transport Cycling Physical activity Absenteeism Dose–response relationship

a b s t r a c t Objective. To study the association between commuter cycling and all-cause sickness absence, and the possible dose–response relationship between absenteeism and the distance, frequency and speed of commuter cycling. Method. Cross-sectional data about cycling in 1236 Dutch employees were collected using a self-report questionnaire. Company absenteeism records were checked over a one-year period (May 2007–April 2008). Propensity scores were used to make groups comparable and to adjust for confounders. Zero-inflated Poisson models were used to assess differences in absenteeism between cyclists and non-cyclists. Results. The mean total duration of absenteeism over the study year was more than 1 day shorter in cyclists than in non-cyclists. This can be explained by the higher proportion of people with no absenteeism in the cycling group. A dose–response relationship was observed between the speed and distance of cycling and absenteeism. Compared to people who cycle a short distance (≤5 km) three times a week, people who cycle more often and longer distances are absent for fewer days on average. Conclusion. Cycling to work is associated with less sickness absence. The more often people cycle to work and the longer the distance travelled, the less they report sick. © 2010 Elsevier Inc. All rights reserved.

Introduction Despite warnings about the potentially negative health consequences of a sedentary lifestyle, a large proportion of employed adults are not physically active enough. Promoting physical activity (PA) that fits in well with normal daily routines is a promising way of reaching large numbers of less active people. Cycling to work is very good exercise that is relatively easy to incorporate in normal daily routines (Vuori et al., 1994; Hendriksen et al., 2000). It is also an excellent option for more frequent physical activity in a large group of employees (Oja et al., 1998). Earlier studies have shown that commuter cycling substantially reduces the risk of premature mortality (Anderson et al., 2000; Matthews et al., 2007) and cardiovascular risk (Hamer and Chida, 2008), improves health (Oja et al., 1991; de Geus et al., 2008) and physical performance (Hendriksen et al., 2000; de Geus et al., 2009), and can have a positive effect on preventing overweight (Wagner et al., 2001; Hu et al., 2002; Lindström, 2008; Wen and Rissel, 2008). It is less clear whether these health benefits of commuter cycling also lead to a measurable reduction in absenteeism. There is evidence documenting the positive effect of regular PA on sickness absence. A Dutch prospective study showed that employees who participated in sporting activities were off sick 20 days less over a 4-year period compared with non-sporting colleagues (van den Heuvel et al., 2005). Furthermore, in a study of three large databases, ⁎ Corresponding author. TNO Quality of Life, P.O. Box 2215, 2301 CE Leiden, Netherlands. Fax: +31 71 5181916. E-mail address: [email protected] (I.J.M. Hendriksen). 0091-7435/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2010.05.007

vigorous-intensity PA at least three times a week had a positive effect on sick leave (Proper et al., 2006), while PA at a less vigorous level did not seem to have this effect. However, very little research has been published that focuses specifically on active commuting and absenteeism. A recent review therefore focused on how available evidence from workplace PA promotion might be applied to walking and cycling to work (Davis and Jones, 2007). The most significant finding was that an increase in PA of more than 1 h per week, which can easily be achieved by walking or cycling to work, is expected to lead to a measurable reduction in the level of absenteeism. If it can be demonstrated that cycling to work can reduce absenteeism costs, there could be a major impact on the number of companies encouraging commuter cycling. Employers will probably be more willing to promote cycling to work if there is a demonstrable return on investment. The aim of this study was to assess the association between commuter cycling and all-cause sickness absence, and to explore the influence of distance, frequency and speed of commuter cycling on this relationship.

Methods Participants and study design A cross-sectional study was performed at companies where a substantial proportion of employees cycled to work regularly. Participation was restricted to organisations with white-collar workers, a minimum of 1000 employees, and an

133

I.J.M. Hendriksen et al. / Preventive Medicine 51 (2010) 132–135

overall absenteeism of at least 4% (this is approximately average for the Netherlands; CBS, 2007). Organisations that had recently implemented major health promotion activities or a drastic reorganisation were excluded to eliminate the confounding effects of these temporary activities. In April 2008, a web-based questionnaire (Table 1) was used to collect data on cycling behaviour, perceived barriers and motivational factors of commuter cycling, meeting the PA guidelines (Haskell et al., 2007), health status and personal characteristics. Employees were invited by email (one company) or a call was made using intranet (both other companies), all were reminded using the same method once and small rewards were raffled to stimulate participation. The participating employees were divided in three groups: (1) Cyclists: one way cycling distance ≥ 3 km and a cycling frequency ≥ three times a week OR one way cycling distance ≥ 2 km and cycling frequency ≥ four times a week; (2) Non-cyclists: cycling frequency less than once a week; (3) Irregular cyclists: all other participants. The primary outcome was all-cause absenteeism. Company absenteeism records were used after permission was obtained from the respondents. Absenteeism calculations included weekends and reduced working hours due to illness; maternity leave was excluded. The total duration (in days) and frequency of absenteeism were calculated over a one-year period (1 May 2007–31 April 2008). A distinction was made between very short absence (1 to 2 days), short absence (3 to 7 days), medium-long absence (8 to 21 days), and long absence (more than 22 days). Frequency of sick leave was defined as the number of new sick-leave spells during the year studied. Statistical analysis The nature of the study precluded a randomised study since cyclists and non-cyclists will probably not be similar in terms of potential confounders. Propensity scores were used to adjust for confounders and make groups comparable (Rosenbaum and Rubin, 1983). In our analysis, the propensity score of an employee is the probability of being a cyclist conditional on the individual's covariate values. To compute propensity scores, a logistic regression model was used with the cycling variable as outcome and candidate confounders (i.e. personal, health and lifestyle factors that can influence the fact whether someone will cycle to work or not) as predictors. Variables showing significant (p b 0.10) imbalance in answers between cyclists and noncyclists were included in the propensity score model: age, gender, education level, type of job, subjective health, estimated physical fitness, Body Mass Index (BMI), smoking behaviour, chronic disease, living distance from work and perceived barriers and motivational factors of commuter cycling (with exception of good cycling arrangements at work). Estimated propensity scores were then used to divide the original group into quintiles to achieve a balance between treatment groups within each quintile, and to remove 90% of the bias due to each covariate (Rosenbaum and Rubin, 1985). The probability of being a cyclist is similar for each employee in a given quintile. To evaluate differences between cyclist and non-cyclist absenteeism, zeroinflated Poisson models were used since the proportion of people without absenteeism was expected to be very high. This model was also used to evaluate the relationship between cycling distance, speed and frequency, and absenteeism. Analyses were performed with the software packages SPSS version 14.0 and R version 2.6.0 (R Development Core Team, 2008), and p-values b 0.05 were considered significant.

Table 1 Content of web-based questionnaire on cycling behaviour, perceived barriers and motivational factors of commuter cycling, physical activity, health status and personal characteristics, used in April 2008 in the Netherlands. Subject

Description

• Are you cycling to work? (no, never; yes, sometimes; yes, regular; yes, almost every time I go to work) • For how long are you cycling to work? (years) • How often did you cycle to work in the last 4 weeks? (b1 time a week, 1–5 times a week) • What distance do you cycle to work? (km) • How long does it take to cycle this distance to work? (min) • At what speed do you normally cycle to work? (slowly (b 16 km/h); average (16–20 km/h), fast (N20 km/h) • Did you cycle to work in the past? [this question was only for the participants who did not cycle to work currently] Cycling for other purposes Questions about other purposes for cycling (e.g. leisure, doing groceries) and frequency of cycling for other reasons then commuting Barriers and motivational Questions about to what degree (never, rarely, factors of commuter cycling sometimes, often, almost always) the following factors hindered or motivated people to cycle to work: • Barriers (lack of interest or enjoyment in cycling, bad weather, traffic jam, picking up the children, cycling is taking more/too much time, many external business appointments) • Motivational factors (good cycling arrangements at work, good for health, good for the environment, financial reasons, getting some fresh air/ being outside) Physical activity Looking back at the last month and thinking of a normal week: • On how many days of the week were you physical active at a moderate intensity for at least 30 min? (0–7 days a week) • How many times were you physical active at a high intensity for at least a continuous 20 min? (0–10 times a week) [this is a validated Dutch questionnaire on the current physical activity guidelines (Douwes and Hildebrandt, 2000)] • Did your physical activity behaviour of the last three months changed compared to the rest of the year? General health • How would you describe you health? (excellent, good, moderate, mediocre, poor) • How would you describe your physical fitness? (extreme high, high, average, low, extreme low) Chronic disease Questions about possible chronic diseases or handicaps that influence work capability or physical activity behaviour Smoking behaviour • Do you smoke? (No, I never smoked; no, but I did in the past; yes, b 21 cigarettes a day or b 7 cigars a day; yes, ≥21 cigarettes or ≥7 cigars a day) Personal characteristics Date of birth, gender, stature, weight, highest education followed, type of job, hours a week working, days a week working, distance from home to work

Commuter cycling behaviour

Results Characteristics of the study population Three government organisations participated in this study (n=5410). Valid responses were obtained from 1878 employees (response rate of 35%). After excluding respondents with incomplete or missing absenteeism data, and people with more than 90 days sick leave (making categorisation according to cycling habits impossible), 1621 subjects with complete data remained. Irregular cyclists (n=209) were excluded from the analysis and 176 respondents were omitted because they were in the non-overlapping tail areas when the propensity score method was applied. Ultimately, the study population consisted of 1236 respondents

with a mean age of 43 years (Table 2). Cyclists were significantly more frequently male, met the PA guidelines more often, had a better selfreported physical fitness, better subjective health, lower self-reported BMI and smoked less.

Results on absenteeism The mean total duration of absenteeism during the study year was significantly shorter in cyclists compared to non-cyclists: 7.4 days and 8.7 days, respectively (Table 3).

134

I.J.M. Hendriksen et al. / Preventive Medicine 51 (2010) 132–135

Table 2 Baseline characteristics of the study population (mean (SD) or %), collected in April 2008 in the Netherlands. Characteristics Mean age (years) Gender (women) (%)* Advanced education (%) Meeting PA guidelinesa (%)** Low estimated physical fitness (%)** Good subjective health (%)* Mean self-reported BMI (kg·m− 2)** Smoking behaviour (%)* Chronic health complaints (%)

Cyclists (n = 785)

Non-cyclists (n = 451)

Total (n = 1236)

43.9 (9.6) 48 77 81 11 87 24.1 (3.1) 13 12

43.1 (10.4) 55 72 63 23 80 24.7 (3.9) 19 16

43.3 (9.8) 51 75 72 15 84 24.4 (3.6) 15 13

Difference between cyclist and non-cyclists *p b 0.05, **p b 0.01. a Physical activity (PA) of at least moderate intensity for at least 30 min on at least 5 days a week and/or intensive physical activity of at least 20 min at least three times a week.

The predicted mean number of days absent from work for cyclists is 0.03 lower than for non-cyclists, which is not significant. However, when the zero-inflated Poisson model (adjusted for confounders using propensity scores) was fitted to the data, the predicted probability of observing cyclists without absenteeism was 0.29 higher (p=0.03) than for non-cyclists. This means that the lower absence rate among cyclists can be explained by the higher proportion of people with zero days of absenteeism. Although the non-corrected results for absenteeism frequency show that very short (1 to 2 days) and short absenteeism (3 to 7 days) was less frequent in cyclists, this was no longer significant when adjusting for confounders using the propensity score (p = 0.24). When including the covariates mentioned in Table 2 in the relationship between cycling and absenteeism, only the variable “subjective health” was significant: people with fair or bad subjective health were absent on more days (0.37 and 0.68, respectively, p=0.00) than people with good or excellent subjective health. When this variable was included in the model, the effect of cycling was no longer significant. Finally, the interaction between subjective health and cycling was included in the model. Taking non-cyclists with good or excellent health as a reference category, the predicted mean number of days absent from work for cyclists with good or excellent subjective health is −0.08 lower (p=0.00) than for non-cyclists with good or excellent subjective health. Cyclists and non-cyclists with fair or bad subjective health are all absent more than non-cyclists with good or excellent subjective health.

Table 4 Characteristics of the cyclists (mean (SD) or %) based on data collected in April 2008 in the Netherlands. Characteristics

(n = 785)

Distance

Mean cycling distance one way (km) Short distance (≤5 km) (%) Medium distance (6–10 km) (%) Long distance (N10 km) (%) Mean cycling time one way (min) Low subjective speed (≤ 16 km·h− 1) (%) Medium subjective speed (17–20 km·h− 1) (%) High subjective speed (N20 km·h− 1) (%) Mean frequency of cycling (times a week)

Intensity

Frequency

6.8 (4.0) 47.5 38.8 13.7 21.6 (10.8) 7.4 73.0 19.7 4.1 (0.7)

In the model with only main effects (Model 1), it was found that, compared to cycling three times a week, higher cycling frequencies result in a significantly lower number of days of absenteeism. When looking at the model with two-order interactions (Model 2), it can be concluded that, compared with people who cycle a short distance three times a week, people who cycle more often and longer distances are absent on fewer days. The only exception was the group of people who cycled a medium distance three times a week. The number of days absent was significantly higher in this group. Finally, high cycling speed seems to be associated with increased absenteeism. Discussion The results of the current study indicate that cycling to work is associated with less absenteeism. Subjective health was the only covariate that significantly interfered with the relationship between cycling and absenteeism. However, when the interaction between subjective health and cycling was included in the model, the predicted mean number of days of absenteeism for cyclists with good or excellent subjective health was still significantly lower than for noncyclists with good or excellent subjective health. Cycling to work does not therefore only contribute to employee health; the reduction in absenteeism may also result in a financial benefit for the employer. This may be a strong argument for employers to implement commuter cycling programmes. So far, literature in peer-reviewed journals has not examined the link between commuter cycling and absenteeism. Although several documents indicate that regular cycling (to work) reduces absenteeism (Saelensminde, 2004; Macdonald, 2007), these statements are based on assumptions and do not use actual absence data for cyclists. One crosssectional study was recently published in a Dutch national journal

Dose–response relationship Table 4 shows the characteristics of the cyclists. Distance was categorised as short (≤5 km), medium (6–10 km), and long distance (N10 km). The mean cycling distance was 6.8 km one way and 73% of the cyclists reported cycling at medium speed (17–20 km·h− 1). The relationship between cycling distance, subjective cycling speed, frequency of cycling and absenteeism is shown in Table 5. Cycling frequency was defined as freq3, freq4 and freq5 (cycling on average three, four or five times a week, respectively). The low- and high-speed groups were too small to test interactions with distance and frequency.

Table 3 Absenteeism in cyclists and non-cyclists (mean (SD) and estimates (p-value)), based on data collected from May 2007 to April 2008 in the Netherlands. Absenteeism

Cyclists (n = 785)

Non-cyclists (n = 451)

Count model estimates

Zero-inflated estimates

Absence rate (days) Absence frequency (times)

7.4 (12.1) 1.4 (1.6)

8.7 (12.5) 1.7 (1.7)

− 0.03 (0.34) − 0.04 (0.59)

0.29 (0.03) 0.25 (0.24)

Table 5 Dose–response relationship between cycling and absenteeism (estimates, Standard Error (SE)) based on data collected from May 2007 to April 2008 in the Netherlands.

Short distance (≤ 5 km) Medium distance (6–10 km) Long distance (N 10 km) Low speed (≤16 km·h−1) Medium speed (17–20 km·h−1) High speed (N 20 km·h− 1) Freq3 (three times·week−1) Freq4 (four times·week−1) Freq5 (five times·week−1) Freq4*medium distance Freq5*medium distance Freq4*long distance Freq5*long distance

Model 1a Parameter estimates (SE)

Model 2b Parameter estimates (SE)

Reference 0.15 (0.27)** 0.01 (0.04) Reference − 0.05 (0.05) 0.12 (0.05)* Reference − 0.17 (0.03)** − 0.13 (0.04)**

Reference 0.41 (0.06)** 0.15 (0.08) Reference − 0.04 (0.05) 0.12 (0.05)* reference 0.06 (0.06) 0.01 (0.06) − 0.46 (0.07)** − 0.15 (0.08)* − 0.10 (0.10) − 0.33 (0.13)**

*p b 0.05, **p b 0.01. a Model 1 includes only main effects, corrected for subjective health. b Model 2 includes a two-order interaction between frequency and distance, corrected for subjective health.

I.J.M. Hendriksen et al. / Preventive Medicine 51 (2010) 132–135

(Koenders and van Deursen, 2008), where results from a lifestyle questionnaire (including some questions about travel to work) were linked to the absenteeism data of the company. They concluded that active commuting (walking and cycling) was related to lower absenteeism compared with commuting by car or public transport. It should be noted that they did not define cycling to work and correction for confounders was only applied to age, gender and level of education. Dose–response results show that the more often people cycle to work and the longer the distance, the lower the absenteeism, with the exception of those who cycle a medium distance three times a week. There is no clear explanation for this peculiar finding. The only remarkable feature of this group was a relatively high percentage of cyclists who were absent from work for long periods. The same holds for high cycling speed; the association with higher absenteeism may be due to a small group of participants with a very long absenteeism (more than 70 days). Almost three-quarter of the cyclists in this study said they cycled to work at an average speed (17–20 km·h− 1). This concurs with previous, objectively measured, results from a Dutch study (Hendriksen et al., 2000), in which cycling speed to work was on average 18 km·h− 1 for women and 20 km·h− 1 for men. Cycling at this speed corresponds to 6 METs (Ainsworth et al., 2000), which is on the borderline between moderate-intensity activity (3–6 MET) and vigorous-intensity activity (N6 MET) in the current PA guideline (Haskell et al., 2007). The results of the current study show that regular commuter cycling, at an intensity on the borderline between moderate and vigorous, can reduce absenteeism.

Study limitations and strengths The most important limitation of this study is the cross-sectional design resulting from the impossibility of using a randomised controlled trial. The drawbacks of this design were reduced by using a propensity score analysis, which is an attempt to reconstruct a situation similar to randomisation by adjusting for relevant confounders. However, a longitudinal study is still needed to demonstrate a causal relationship between cycling to work and absenteeism.

Conclusions Cycling to work is associated with less all-cause sickness absence. The more often people cycle to work and the longer the distance travelled, the lower the absenteeism. Mean absenteeism in cyclists is significantly lower than in non-cyclists, even after controlling for subjective health, so cycling to work not only contributes to employee health, it may also result in a financial benefit for the employer.

Conflict of interest statement The authors declare that there are no conflicts of interest.

Acknowledgments The authors wish to thank the funding organisations, the Dutch Ministry of Health, Welfare and Sport and the Dutch Ministry of Transport, Public Works and Water Management. We also thank the subjects and the companies for their cooperation.

135

References Ainsworth, B.E., Haskell, W.L., Whitt, M.C., Irwin, M.L., Swartz, A.M., Strath, S.J., et al., 2000. Compendium of physical activities: an update of activity codes and MET intensities. Med. Sci. Sports Exerc. 32, S498–S504. Andersen, L.B., Schnohr, P., Schroll, M., Hein, H.O., 2000. All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch. Intern. Med. 160, 1621–1628. Centraal Bureau voor de Statistiek (CBS), 2007. National statistics on sickness absence. [Nationale Verzuim Statistiek]. Statline, Statistics Netherlands. Available at: http:// statline.cbs.nl/StatWeb/publication/?VW=T&DM=SLNL&PA=70812ned&D1=a&D2=0&D3=a&HD=081223-1209&HDR=T&STB=G1,G2 Davis, A., Jones, M., 2007. Physical activity, absenteeism and productivity: an evidence review. Project report UPR T/102/07. JMP Consulting. Available at: http://www.tfl. gov.uk/assets/downloads/corporate/Physical-activity-absenteeism-and-productivity-evidence-review.pdf. de Geus, B., van Hoof, E., Aerts, I., Meeusen, R., 2008. Cycling to work: influence on indexes of health in untrained men and women in Flanders. Coronary heart disease and Quality of life. Scan. J. Med. Sci. Sports 18, 498–510. de Geus, B., Joncheere, J., Meeusen, R., 2009. Commuter cycling: effect on physical performance in untrained men and women in Flanders: minimum dose to improve indexes of fitness. Scand. J. Med. Sci. Sports 19, 179–187. Douwes, M., Hildebrandt, V.H., 2000. Vragen naar de mate van lichamelijke activiteit (Questionnaire for Physical Activity). Geneeskunde en Sport 33, 9–16. Hamer, M., Chida, Y., 2008. Active commuting and cardiovascular risk: a meta-analytic review. Prev. Med. 46, 9–13. Haskell, W.L., Lee, I.M., Pate, R.R., Powell, K.E., Blair, S.N., Franklin, B.A., et al., 2007. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Med. Sci. Sports Exerc. 39, 1423–1434. Hendriksen, I.J., Zuiderveld, B., Kemper, H.C., Bezemer, P.D., 2000. Effect of commuter cycling on physical performance of male and female employees. Med. Sci. Sports Exerc. 32, 504–510. Hu, G., Pekkarinen, H., Hanninen, O., Tian, H., Jin, R., 2002. Comparison of dietary and non-dietary risk factors in overweight and normal-weight Chinese adults. Br. J. Nutr. 88, 91–97. Koenders, P.G., van Deursen, C.G.L., 2008. Reizen naar en voor het werk en verzuim in de banksector [Traveling to and for work and sickness absence in the banking sector]. TBV 16, 143–148. Lindström, M., 2008. Means of transportation to work and overweight and obesity: a population-based study in southern Sweden. Prev. Med. 46, 22–28. Macdonald, B., 2007. Valuing the benefits of cycling: A report to Cycling England. Cambridge, SQW Limited. Available at: http://www.networks.nhs.uk/uploads/ 07/09/final_executive_summary.pdf. Matthews, C.E., Jurj, A.L., Shu, X.O., Li, H.L., Yang, G., Li, Q., Gao, Y.T., Zheng, W., 2007. Influence of exercise, walking, cycling, and overall nonexercise physical activity on mortality in Chinese women. Am. J. Epidemiol. 165, 1343–1350. Oja, P., Manttari, A., Heinonen, A., Kukkonen-Harjula, K., Laukkanen, R., Pasanen, M., et al., 1991. Physiological effects of walking and cycling to work. Scand. J. Med. Sci. Sports 1, 151–157. Oja, P., Vuori, I., Paronen, O., 1998. Daily walking and cycling to work: their utility as health-enhancing physical activity. Patient Educ. Couns. 33, S87–S94. Proper, K.I., van den Heuvel, S.G., de Vroome, E.M., Hildebrandt, V.H., van der Beek, A.J., 2006. Dose–response relation between physical activity and sick leave. Br. J. Sports Med. 40, 173–178. R Development Core Team, 2008. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. 3-900051-003. Available at: http://www.R-project.org. Rosenbaum, P.R., Rubin, D.B., 1983. The central role of the propensity score in observational studies for causal effects. Biometrika 70, 41–55. Rosenbaum, P.R., Rubin, D.B., 1985. Constructing a control group using multivariate matched sampling methods that incorporate the propensity scores. AmerStat 39, 33–38. Saelensminde, K., 2004. Cost-benefit analyses of walking and cycling track networks taking into account insecurity, health effects and external costs of motorized traffic. Transportation Research part A 38, 593–606. van den Heuvel, S.G., Boshuizen, H.C., Hildebrandt, V.H., Blatter, B.M., Ariëns, G.A., Bongers, P.M., 2005. Effect of sporting activity on absenteeism in a working population. Br. J. Sports Med. 39, e15. Vuori, I.M., Oja, P., Paronen, O., 1994. Physically active commuting to work: testing its potential for exercise promotion. Med. Sci. Sports Exerc. 26, 844–850. Wagner, A., Simon, C., Ducimetiere, P., Montaye, M., Bongard, V., Yarnell, J., et al., 2001. Leisure-time physical activity and regular walking or cycling to work are associated with adiposity and 5 y weight gain in middle-aged men: the PRIME Study. Int. J. Obes. Relat. Metab. Disord. 25, 940–948. Wen, L.M., Rissel, C., 2008. Inverse associations between cycling to work, public transport, and overweight and obesity: Findings from a population based study in Australia. Prev. Med. 46, 29–32.