Development of the Health economic assessment tools (HEAT) for walking and cycling: consensus workshop: Bonn, Germany, 11-12 December 2014: meeting report

WORLD HEALTH ORGANIZATION REGIONAL OFFICE FOR EUROPE WELTGESUNDHEITSORGANISATION REGIONALBÜRO FÜR EUROPA ORGANISATION MONDIALE DE LA SANTÉ BUREAU RÉGIONAL DE L'EUROPE ВСЕМИРНАЯ ОРГАНИЗАЦИЯ ЗДРАВООХРАНЕНИЯ ЕВРОПЕЙСКОЕ РЕГИОНАЛЬНОЕ БЮРО Development of the Health economic assessment tools (HEAT) for walking and cycling Consensus workshop EUDCE1408105/5.3/61493/8 Bonn, Germany 12 May 2015 11-12 December 2014 Original: English MEETING REPORT © World Health Organization 2015 Document number: WHO/EURO:2015-6816-46582-67607 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 2 - Contents Executive summary ............................................................................................................................ 3 Acknowledgements ............................................................................................................................ 4 1 Introduction and background...................................................................................................... 5 2 Welcome, introductory presentations and core principles........................................................ 8 3 Integration of air pollution effects from walking and cycling on all-cause mortality into HEAT .................................................................................................................. 10 3.1 Introduction and overview of proposed approach ........................................................ 10 3.2 Air pollution indicator .................................................................................................. 10 3.3 Exposure assessment of air pollution ........................................................................... 11 3.4 Inhaled dose during transport and conversion factors of air pollution exposures ..................................................................................................................... 14 3.5 Risk assessment for all-cause mortality ....................................................................... 20 3.6 Relative risk for walking and cycling taking into account air pollution effects .......................................................................................................................... 22 3.7 Input data required ....................................................................................................... 26 3.8 Presentation of results of air pollution effects in HEAT .............................................. 26 4 Additional considerations and future developments ............................................................... 27 5 Conclusions, next steps and closing ........................................................................................... 28 6 References .................................................................................................................................... 30 Annex 1 Workshop programme ................................................................................................. 33 Annex 2 List of participants ....................................................................................................... 35 Annex 3 Sensitivity analysis of using PM2.5 versus elemental carbon as air pollution indicator ...................................................................................................................... 37 Annex 4 Data harmonization and derivation of conversion factors of air pollution exposures while walking or cycling compared to background concentrations ............................................................................................................................. 39 Annex 5 Summary of studies identified on conversion factors or air pollution exposure ....................................................................................................................................... 41 Annex 6 Methodology to assess air pollution effects on relative risks from cohort studies included in HEAT meta-analysis on effects of cycling or walking on all-cause mortality ................................................................................................................. 43 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 3 - Executive summary The promotion of active transport (cycling and walking) for everyday physical activity is an important approach to address the challenge of high levels of physical inactivity in most regions of the world. This requires building effective partnerships with the transport and urban planning sectors, whose policies are highly influential in providing appropriate conditions for such behavioural changes to take place and be maintained. Economic appraisal is an established practice in transport planning. However, until recently the health effects of transport interventions have seldom been taken into account in such analyses. The Health Economic Assessment Tools (HEAT) for Walking and Cycling1 provide guidance and practical, web-based tools for economically assessing the health effects of walking or cycling. It is based on the evidence on the association between walking and cycling and all-cause mortality. Coordinated by WHO, steered by a core group of multi-disciplinary experts and supported by ad-hoc invited relevant international experts, the HEAT project holds regular consensus meetings to discuss and agree upon methodological updates and new features of HEAT. The fourth consensus meeting was convened to achieve scientific consensus on 1) the possibility to integrate the influence of air pollution on mortality of cyclists and pedestrians in HEAT, based on best available evidence, feasibility and state- of-art knowledge; and 2) on concrete options for integrating air pollution considerations into HEAT. The meeting was attended by 21 experts from public health, transport and environmental sciences and four WHO staff members of the WHO Regional Office for Europe. A method used for recent health impact assessments of air pollution and transport modes [1,2] was agreed to serve as basis for inclusion of air pollution effects into HEAT for walking and cycling. This method uses PM2.5 as the air pollution measure, based on background PM2.5 concentrations, conversion factors between modes of transport, ventilation rates by mode and a dose-response function for PM2.5 and all-cause mortality. Air pollution is a mixture of substances and particles, which have been associated with different health effects. To assess the health impacts of air pollution on pedestrians and cyclists PM2.5 was agreed to be used as the air pollution indicator because there is a large body of evidence mainly from cohort studies to support quantification of the effects of long-term exposure of PM2.5 on mortality and morbidity [36, 37, 38]. In absence of PM2.5 concentration measurements an internationally accepted conversion factor of 0.6 was agreed to transform more widely available PM10 measurements into estimates of PM2.5. Consensus was achieved on how to estimate the equivalent change of air pollution intake due to cycling or walking compared to a reference scenario: mode-specific inhalation rates, duration and PM2.5 concentrations are multiplied to calculate an intake and added to the intake during the rest of the day. The option to add a scenario switching from other modes of transport was considered for later updates. 1 www.heatwalkingcycling.org Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 4 - To derive mode-specific PM2.5 concentrations, the participants agreed on using conversion factors with the background concentrations. The agreed factors were derived from a purposive review of studies that estimated PM2.5 concentrations in the different microenvironments (cycling and walking) and background concentrations and are as follows: cycling to background 2 and walking to background 1.6. Consensus was also achieved on the mode-specific ventilation rates to be use based on a method developed by the US EPA [3, 4]. The agreed rates are 1.37 m3/hr for walking and 2.55 m3/hr for cycling. A recently published meta-analysis included 14 international cohort studies and summarized the relative risk between all-cause mortality and each increment of 10 g/m3 of PM2.5 as 1.07 (1.04-1.09) [5] and was agreed by the participants for use in HEAT. Addressing a concern raised at the previous consensus workshop regarding possible double counting of health impacts of air pollution by using relative risk estimates derived from a meta-analysis of 14 walking and 7 cycling studies and all-cause mortality, the participants agreed to consider this issue as negligible with regard to developing a separate HEAT module on air pollution effects. The participants also endorsed the proposed required data input for the calculation of the air pollution effects in HEAT: annual mean concentration of PM2.5 or PM10 in the place of interest; trip duration in minutes or distance travelled in km (already part of the existing HEAT mortality modules); whether the walking/cycling is recreational or utilitarian (to be included in new question in HEAT, to define if background or traffic exposure level are used); and all-cause mortality rate in adults in the study population (already part of the existing HEAT mortality module). Participants also discussed ways of presenting the results to the HEAT users and putting them into perspective vis-à-vis the totality of health effects of air pollution in the entire population. The workshop concluded with an outlook on future HEAT-related activities and suggestions for further developments, in particular the preparations of discussing an additional module on traffic crashes. Acknowledgments The workshop was organized by the WHO Regional Office for Europe and supported by the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety. It was carried out in close collaboration with the Transport, Health and Environment Pan-European Programme (THE PEP). Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 5 - 1 Introduction and background Given the magnitude of the physical inactivity problem in most global regions, classical health promotion approaches, although an important part of the solution, will not be sufficient to eliminate the problem. Regular cycling and walking, for example, as part of trips to work and back, might facilitate the integration of physical activity into an already busy day. The promotion of active transport (cycling and walking) is a win-win approach since it not only promotes health but can also lead to positive environmental effects especially if cycling and walking replace particularly short and later possibly also medium-length car trips. There is a large potential for active travel in European urban transport systems, since many car trips are short and could be substituted, at least partly, by trips undertaken on foot or by bicycle. This requires building effective partnerships with the transport and urban planning sectors, whose policies are highly influential in providing appropriate conditions for such behavioural changes to take place and be maintained. Transport is an essential component of life, providing access to services, goods and activities. Different modes of transport are associated with specific effects on society, one being health effects. Fully appraising these effects is an important basis for evidence-informed policy-making. Economic appraisal is an established practice in transport planning. However, until recently, such analyses have seldom considered the part of the health effects of transport interventions related to physical activity. Valuing health effects is a complex undertaking, and transport planners are often not well equipped to fully address the methodological complexities involved. The Health Economic Assessment Tools (HEAT) for Walking and Cycling, launched by the WHO Regional Office for Europe through a collaborative project in 2007, provides guidance and a practical, web-based approach for economically assessing the health effects of walking and cycling (1–Fehler! Verweisquelle konnte nicht gefunden werden.). Coordinated by WHO, steered by a core group of multi-disciplinary experts and supported by ad-hoc invited relevant international experts2, the project was started in 2005, aimed at developing guidance and practical tools for economic assessments of the health effects from cycling and from walking. The main goal of the project is the development of the Health Economic Assessment Tool (HEAT) for walking and cycling, a harmonized method for economic valuation of health effects of cycling and walking, based on best available evidence and international expert consensus. HEAT calculates: if x people cycle or walk y distance on most days, what is the economic value of all-cause mortality rate changes? HEAT is primarily aimed at transport planners, traffic engineers, economists and special interest groups. Since this audience may not necessarily have ready access to epidemiological and economic expertise and modelling tools, HEAT is intended to be scientifically robust, yet easy to use. It is not intended to be a comprehensive health impact assessment tool but aims 2 See full lists at www.heatwalkingcycling.org Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 6 - at providing an estimate of the health effects of regular walking and cycling (currently mortality only) based on minimal data input for use in cost-benefit analyses in transport planning. The HEAT tool and its accompanying guidance are open to further developments with the aims of:  keeping the tool abreast of relevant scientific developments (such as in relation to the emergence of new epidemiological studies that provide improved relative risk function on relevant health effects);  expanding the functionality of the tool in response to the priority needs of users and improved scientific knowledge; and  improving the guidance offered to the users. HEAT is developed through an iterative process, consisting of the following main steps: : a) review of approaches to the inclusion of health effects into economic appraisals of transport interventions related to cycling and walking; b) critical evaluation of these approaches and indicators regarding their relevance, accuracy and feasibility; c) achievement of scientific consensus on how to apply this knowledge within the “HEAT environment”; and d) regular review and update of the approach in view of user-needs and scientific developments. A core group of eleven members and relevant international experts invited ad hoc manage the implementation of this activity. The project has been developed through systematic reviews of the published literature and a comprehensive consensus-building process, followed by a practical application based on the consensus achieved. The achievement of scientific consensus on the approach to be taken in the development of the different functionalities and components of the tools is a key element of this process. This is achieved through consensus building workshops which, under the coordination of the WHO, bring the core group together with international advisors invited on the basis of their scientific expertise in the aspects of interest, and develop consensus-based recommendations on possible ways forward for the further development of the tool. The first consensus meeting, which took place in Graz (Austria) in 2006, resulted in the publication of the first HEAT in 2007, initially only available as an Excel-based calculator for cycling [6]. In 2011, the second consensus meeting, held in Oxford (United Kingdom) resulted in the launch of an updated version of HEAT for cycling and of a new HEAT for walking as web-based tools (www.euro.who.int/HEAT) with a methods and user guide booklet [Fehler! Verweisquelle konnte nicht gefunden werden.]. In 2013, the third consensus meeting, held in Bonn (Germany) discussed and endorsed updated relative risks and dose-response curves for HEAT cycling and walking [Fehler! Verweisquelle konnte nicht gefunden werden.] as well as updated values of a statistical life, and considered ways to include air pollution exposure of travellers [9]. Updated versions of the online tools and the user guide booklet were launched in August 2014 [10]. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 7 - The HEAT process is particularly designed to be open to continuous updating and further developing the tools. Considering air pollution in HEAT is a relevant topic based on experiences with the HEAT target audience, which has expressed on several occasions concerns about possible negative air pollution health effects when promoting cycling and walking. The previous consensus meeting also recognized that injuries are a relevant topic to discuss further in the HEAT context. However, methodological complexities still need to be addressed before a new module to assess injuries can be integrated into HEAT. This fact and the perceived higher demand from the users’ viewpoints to address the question of air pollution supported addressing this topic first. At the third consensus meeting, an initial possible approach for including air pollution effects was presented, based on the inhaled dose of particulate matter (particulate matter with an aerodynamic diameter of 10 µm or less (PM10) or 2.5 µm or less (PM2.5)) per day in different activity modes, an available relative risk function, the fraction of health effects attributable to air pollution among those exposed and available mortality rates. Participants welcomed the proposed method as a good basis for developing a separate, optional air pollution module to calculate the effects of air pollution on cyclists and walkers. However, it was concluded that inclusion into the HEAT model would have been conditional on further insight on the following two aspects:  The assessment of possible double-counting of effects from air pollution on all-cause mortality.  The determination of the suitability of the proposed conversion factors and assumptions. In particular, further work was deemed necessary regarding the conversion factors for different modes of transport as the proposed conversion factors were derived from the few available studies that examined all modes of interest simultaneously to reduce methodological diversity, but different approaches could be considered. Based on the recommendations formulated by the third consensus meeting, additional work was undertaken to clarify these two aspects. A background document was prepared in summer 2014 that proposes an approach to the development of a new, separate HEAT module on air pollution, based on work undertaken by David Rojas-Rueda and Mark Nieuwenhuijsen (Centre for Research in Environmental Epidemiology, Barcelona) and Audrey de Nazelle and Olivier Bode (Centre for Environmental Policy, Imperial College London). It also provided further material and insights on the above two points. The document is based on two working papers, which were developed in follow-up to the recommendations formulated by the third consensus meeting on possible double-counting effects by Rojas-Rueda et al. [11] and on the conversion factors by de Nazelle et al. [12]. The background document was presented and discussed at the fourth consensus workshop on 11-12 December 2014, which was attended by 21 experts from public health, transport and environmental sciences and four staff members of the WHO Regional Office for Europe. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 8 - In order to provide a complete overview and adequate level of detail, relevant parts of this background document are included in this meeting report and its annexes. The consensus workshop was convened to achieve scientific consensus on:  the possibility to integrate the influence of air pollution on mortality in HEAT, based on best available evidence, feasibility, and state-of-art knowledge; and  concrete options for integrating air pollution considerations in HEAT. The specific objectives of the workshop were:  to discuss the main findings of a review of relative risks of all-cause mortality for air pollution exposure while walking and cycling;  to discuss and achieve consensus on the proposed approach for HEAT to incorporate air pollution considerations for walking and cycling; and  to discuss other possible future improvements that could be considered for implementation in HEAT (e.g. regarding possibilities to consider cause-specific mortality/morbidity, traffic injuries, different metrics for the economic valuation; others). Ahead of the workshop, the participants received the above mentioned background documentation prepared by the project core group, including:  a summary of the epidemiological literature on air pollution and all-cause mortality;  options for developing a module on air pollution for HEAT; and  the current HEAT for Walking and Cycling tools (www.heatwalkingcycling.org) and guidance, to familiarize with the approach taken to the valuation of the health effects of walking and cycling. 2 Welcome, introductory presentations and core principles Elizabet Paunovic, Head of Office, European Centre for Environment and Health, WHO/Europe, welcomed the participants and thanked them for their availability to support this important project step to further improve the HEAT. Michal Krzyzanowski was elected as chair and Christian Schweizer and Sonja Kahlmeier as rapporteurs of the workshop. Francesca Racioppi introduced the HEAT, its principles, motivation and aim and also highlighted the history and formal process of its development. Christian Schweizer reported on the dissemination efforts around HEAT, known applications and lessons learned from HEAT since 2008. HEAT was Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 9 - mainly disseminated through presentations at relevant meetings and conferences; in addition, HEAT has won awards or commendations. Since it was launched in May 2011, the HEAT website has been visited by more than 25.000 users. Through a call for contributions, case studies on applications have been collected. Since November 2012, web-based training sessions have also been provided; to date, about 350 experts have been trained. Based on existing experience, uptake of the tool seems strongest among advocates and planners and by academia, and Sweden and England have adopted it as part of official transport valuation toolboxes so far. Beyond that, evidence of direct impact on policy or transport investment decisions has been more limited. Participants commented that HEAT seems to be mostly perceived as tool for economic valuation and that its potential use as simple health impact assessment tool is less widely known. Additional promotion and marketing of the tools, in particular to a health audience, was noted as one of the future tasks for the project. Additionally, interpretation of the results of HEAT seems often challenging to users and future improvements of HEAT should provide more guidance on this. Sonja Kahlmeier then reminded participants of the key principles of HEAT: they are designed as practical tools for transport and urban planners to provide them with an evidence-informed, transparent, conservative, adaptable and modular approach to the economic valuation of the health benefits of cycling and walking. HEAT can be used when planning new projects, to evaluate past projects or for modelling purposes. She also explained that more sophisticated approaches to health impact assessment of cycling and walking had been developed to satisfy advanced research needs. HEAT, however, was mainly developed to facilitate the inclusion of health effects into economic transport valuation aimed at transport planners and practitioners, particularly in settings without ready access to specialized epidemiological and economic expertise, and to provide an indication of the order of magnitude of effects. She further outlined the scope of the proposed update of HEAT to include air pollution effects. Finally, Michal Krzyzanowski reminded participants to consider three main points with regard to the way of working at the workshop: (1) the content being proposed needs to withstand scientific scrutiny and any assumption made needs to be made fully transparent; 2) the way of presentation needs to follow the HEAT core principles; and (3) any changes made to the methods need to take into account user needs. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 10 - 3 Integration of air pollution effects from walking and cycling on all-cause mortality into HEAT 3 . 1 I n t rodu c t io n and o v erv iew o f p rop os ed a pp ro ac h David Rojas-Rueda presented a possible approach for including air pollution effects. It proposes to estimate the exposure to air pollution as well as a description of the risk assessment approach for all- cause mortality. This method has been used in recent health impact assessments of air pollution and transport modes [1, 2] and could serve as basis for inclusion of air pollution effects into HEAT for walking and cycling. Based on a large body of evidence, mainly from cohort studies, PM2.5 (or converted PM10 levels) was proposed as the indicator of air pollution to estimate the health effects. The change in the intake of PM2.5 related to travelling with a specific mode, i.e. cycling or walking, compared with not travelling, would be calculated as shown in Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 11 - Table 2. 3.2 Ai r pol lu t ion indica tor Air pollution is a mixture of substances and particles, which have been associated with different health effects. To assess the impacts of air pollution on pedestrians and cyclists, different air pollutants could be considered. Frequently, PM2.5 is used to estimate the health impacts of air pollution because there is a large body of evidence mainly from cohort studies to support quantification of the effects of long- term exposure on mortality and morbidity [13, 14, 15]. Measurements of PM2.5 concentrations are becoming more common, but may be still not available in all areas. In this case, an alternative approach is also available by using the annual average concentration of PM10, applying an internationally accepted conversion factor [16] to transform this into an estimate of PM2.5 from the model (Table 1). Table 1. Conversion factors for PM10 based on PM2.5 levels as proposed in the background document. Country group Conversion factors PM10/PM2.5 in developing countries 0.5 a PM10/PM2.5 in developed countries b 0.65 a PM10/PM2.5 in European high-income countries c 0.73 a a Multiply the annual average concentration of PM10 in the city by the conversion factor to obtain the estimation of concentration in PM2.5. b Suggested to be used for cities in subregions AMR A, EUR A, EUR B, EUR C, and WPR A (which included the USA, Canada, all of Europe, Japan, Singapore, Australia and New Zealand [6]. Regarding Europe, see also next footnote. c For Europe, a mean of 0.73 has been reported and is proposed as more adequate for this region [16]. Discussion The experts agreed to use PM2.5 or converted PM10-levels, using a conversion factor of 0.60 (based on the approach used by the WHO [17]), as indicator of air pollution for the HEAT applications in the European Region3. 3.3 Exposure assessment o f a i r pol lut ion The main aim of the exposure assessment of air pollution for a particular transport mode (cycling or walking) is to estimate the change in the daily inhaled dose of PM2.5 while travelling in a particular mode compared to a baseline exposure or reference scenario (e.g. staying at home). The exposure assessment of air pollution for pedestrians and cyclists is based on three factors in different 3 Subsequent to the meeting, a sensitivity analysis was carried out using elemental carbon levels as more traffic- related exposure estimate instead of PM2.5, confirming that mortality estimates that are still very comparable, given the uncertainty of the estimates used (see Annex 3). Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 12 - microenvironments: the air pollution concentrations, the minute ventilation rates (according to the intensity of physical activity) and the duration of exposure ( Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 13 - Table 2) [2]. The impact of air pollution for different transport activities compared to a baseline can then be calculated as detailed in Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 14 - Table 2. The model assumes a baseline scenario with no traffic exposure. Following the conservative approach of HEAT this is expected to lead in many cases to an underestimation of the exposure to air pollution for the baseline scenario and therefore to an overestimation of the difference in air pollution exposure compared to walking/cycling and of the health impacts of increased air pollution exposure while cycling/walking. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 15 - Table 2. General formulas to calculate the impact of air pollution for different microenvironments (ME) (i.e. sleeping, resting/staying at home, walking or cycling).) Formulas Concentration for each microenvironment ME (µg/m3) Background concentration (g/m3) * conversion factor for ME Intake for each ME (g/day) Minute ventilation (m3/h) for ME * Duration b (h/day) for ME * Concentration (g/m3) for ME Total yearly intake for cycling/walking (g/day) (Sum of all daily intakes of all MEs with cycling/walking * number of days cycling/walking per year) + (Sum of all daily intakes of all MEs without cycling/walking * (365 - number of days cycling/walking per year)) Total yearly intake without cycling/walking (g/day) a Sum of all daily intakes of all ME without cycling/walking (i.e. the reference scenario) * 365 Equivalent change (g/m3) ( 𝑇𝑜𝑡𝑎𝑙 𝑦𝑒𝑎𝑟𝑙𝑦 𝑖𝑛𝑡𝑎𝑘𝑒 𝑓𝑜𝑟 𝑐𝑦𝑐𝑙𝑖𝑛𝑔/𝑤𝑎𝑙𝑘𝑖𝑛𝑔 𝑇𝑜𝑡𝑎𝑙 𝑦𝑒𝑎𝑟𝑙𝑦 𝑖𝑛𝑡𝑎𝑘𝑒 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑐𝑦𝑐𝑙𝑖𝑛𝑔/𝑤𝑎𝑙𝑘𝑖𝑛𝑔 − 1) ∗ 𝑀𝑒𝑎𝑛 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 Relative risk Exp Ln (RR10) * Equivalent change 10 Attributable fraction among those cycling/walking 𝐴𝐹𝑒𝑥𝑝 = 𝑅𝑅−1 𝑅𝑅 Expected mortality in travellers (i.e. population exposed to the cycling/walking ME) Crude mortality in the study population (deaths per 100,000) * number of travellers / 100,000 Mortality due to extra exposure from cycling/walking Expected mortality in travellers * AFexp a Reference intake refers to staying at home, including sleeping + resting . For other scenarios, contributions from other activities such as using other modes of transport or being at work may be considered. b average daily duration of cycling/walking across an entire year RR= Relative risk; RR10= relative risk per each increment in 10 g/m3 of PM2.5; AFexp = Attributable fraction among those exposed Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 16 - Discussion The experts agreed on some clarifications of the proposed formulas for the calculation of the impact of air pollution as presented in Table 2. The experts specifically agreed on the use of ventilation rates for the approximation of air pollution intake. It was further agreed to use all-cause mortality as the health outcome, in line with the existing version of HEAT. Experts noted that this decision is taken with the clear understanding of the limitations of applying a) relative risk functions derived from spatial variability in the general population to – in some cases - specific sub-populations (such as cyclists); and b) with an expected time lag for the health effects. The use of non-linear integrated dose-response functions has been suggested to reflect indications that the relationship between air pollutants and health risk seem to become flat at the higher end of the dose- response curve, i.e. at higher pollution levels [18]. However, HEAT is being proposed predominantly for applications in a European context, where extreme exposures sometimes found in other parts of the world are rare. So the linear dose-response function proposed in Table 2 was adopted by the experts as being appropriate within the HEAT framework. Experts also noted the possible interaction of air pollution exposure and physical activity. There is a considerable body of literature indicating that air pollution exposure increases systemic inflammation and induce hypertension, two factors strongly linked ischemic heart diseases. Systemic inflammation after air pollution exposure has been observed in several experimental studies involving intra-tracheal, intranasal and ambient exposure [19, 20]. In addition, evidence based on epidemiologic studies indicates an increased risk of diabetes and diabetes-related disorders in association with exposure to ambient pollutants such as fine particulate matter [21]. Finally, air pollution exposure has been linked to adverse effects on the brain such neuro-inflammation, cognitive decline and early signs of dementia [22]. Inflammation and oxidative stress are the key factors in inducing these adverse health effects. Physical inactivity and abdominal adiposity have been associated with persistent, systemic, low-grade inflammation and exert adverse effects on mental and physical health. On the other hand, regular exercise and/or physical fitness may confer protection towards chronic diseases through minimizing excessive inflammation. One of the key mechanisms by which physical activity exerts favourable health effects appears to be due to its capacity to reduce chronic low-grade inflammation. The anti- inflammatory effects of regular exercise/activity can promote behavioural and metabolic resilience, and protect against various chronic diseases associated with systemic inflammation. Moreover, exercise may benefit the brain by enhancing growth factor expression and neural plasticity, thereby contributing to improved mood and cognition. It can be hypothesized that the positive effects of exercise on inflammation, metabolism and cognition may inhibit the negative effects of traffic-related air pollution exposure during exercise. In fact, the pathways and the mechanisms by which physical fitness promotes increased resilience and well-being and positive psychological and physical health are similar to those involved in air pollution related health effects. Such a possible positive interaction should be better assessed in human studies [23]. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 17 - Further, the participants agreed to use two alternative microenvironments for calculating the difference in health impacts resulting from a specific level of cycling or walking. The two microenvironments are: (a) leisure cycling/walking, i,e, in a park or away from roads with motorized traffic and therefore a concentration of air pollution that could be considered equal to background concentration; and (b) utilitarian cycling/walking, i.e. on a road with motorized traffic and therefore a concentration that could be considered equal to background concentration multiplied by the proposed conversion factor for cycling/walking (see chapter 3.4). The alternative activity for the baseline would in both cases be staying at home, i.e. no traffic exposure. The option to add a scenario on switching from other modes of transport was considered for later updates, provided that the necessary background information on the exact counterfactual scenario and the necessary conversion factors to use would be developed, in line with the HEAT principles. 3.4 Inhaled dose dur ing t ranspor t and convers ion factors o f a i r po l lu t ion exposures Methods As input, the model requires the annual average concentration of PM2.5 (or PM10 converted to PM2.5 levels) in the place of interest, i.e. a country, city or local community. Conversion factors from the literature and agreed upon by the experts can then be used to obtain estimates of the concentration for each relevant mode of transport (i.e. transport microenvironment). HEAT currently assumes that a certain proportion of the population changes its transport mode from an (unknown) “average (non- active) behaviour” to walking or cycling. As also assumed in epidemiological studies on health effects of air pollution, the HEAT model would be based on the assumption that this “average” transport behaviour corresponds with the background exposure level [24]. To derive conversion factors between background concentrations and concentrations while walking or cycling, a purposive review for studies that estimated PM2.5 concentrations in the different microenvironments (cycling and walking) and background concentrations was undertaken in the Science Direct and Web of Science databases [12]. Papers published in peer-review journals were selected according to the following criteria:  Monitoring studies of PM2.5 in transportation microenvironments in Europe;  At least 1 active travel mode compared to other modes or to background concentrations;  An appropriate study design for comparisons between modes (i.e. same or similar routes, concomitant or near-concomitant sampling of the 2 modes)  Monitoring studies of exposure concentrations in various transport modes are very diverse in study design and means of reporting the data. This makes quantitative syntheses challenging. To derive meaningful ratios of exposures in various modes, which make sense statistically, the following choices were made: Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 18 -  When data were provided separately within a single study for various design options such as routes, seasons and cities, the data were kept separate rather than estimating a weighted mean for each study.  When only arithmetic means and standard deviations were provided, these were transformed into geometric means and geometric standard deviation, as the distribution of air pollutant concentration X usually follows a lognormal distribution [Fehler! Verweisquelle konnte nicht gefunden werden.]. In addition, it is more meaningful to use geometric means when using normalized results (ratios to reference values), for example when estimating ratios between exposures in a transport mode X to a reference background value Y. Normalized results were calculated as the ratios of a mode A to a reference background concentration measured at fixed monitoring stations (b) by dividing the geometric mean of mode A by the geometric mean of background level b. Similarly normalized results are presented for the ratios between walking or cycling vs driving or taking the bus. For more details on the calculations, please refer to Annex 6. Some studies reported a single overall (campaign-wide) background concentrations value measured throughout the monitoring campaign, while others provided concomitant measurements at the background station with each mode separately. When background concentrations were reported only separately for each mode, an overall background concentration was estimated with a sample-weighted mean concentration. The overall background concentrations were used in the main analysis so as to include more studies (but ratios between modes and background were estimated for each type of reporting as a sensitivity analysis). Results and proposed conversion factors Eleven studies of PM2.5 measurements which compared various travel modes including at least walking or cycling in a simultaneous or quasi-simultaneous design were identified. Details of each study, including, season, route selection, mode comparison and choice of instruments are shown in Annex 4. For illustrative purposes, the table includes the arithmetic means reported in the study. A large variety of designs can be noted, making direct comparisons between studies challenging. Two studies [25, 26] were excluded from the subsequent analysis as they did not report the full PM2.5 fraction (only PM1 to PM2.5 and PM2.5 to PM10), and another study was excluded because they did not simultaneously measure in different travel modes for all of their campaign. Our studies reported measured concentrations at background fixed monitoring stations during the transport micro-environments sampling [27-30], two of which provided simultaneous mode-specific concentrations [29, 30]. Note that de Nazelle [30] reported the overall background concentration in the published article, but provided the mode-specific data for this analysis. Counting Adams et al. [27] as two campaigns (2 seasons), and Zuurbier et al. [29] as 2 campaigns for each mode including low traffic and high traffic cycling), using overall background concentrations as the reference led to 2 sets of data available for walking to background comparisons and 6 for cycling. The work paper [12] also calculated ratios for other Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 19 - transport modes (bus, car) versus background concentrations, and ratios of different modes versus walking or cycling but as HEAT does not assess the effects of switching from one mode to another, these results are not shown here. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 20 - Table 3. Calculation of ratios for walking or cycling versus background (denominator) for each study Concentration (µg/m3)^ Ratio^^ Authors Mode Comments Sample gm gsd am sd CI95.low CI95.up ratio lower bound Upper bound Walking to background de.Nazelle.et.al.(2012) [Fehler! Verweisquelle konnte nicht gefunden werden.] Walk 38 20.6 1.4 21.6 6.3 18.4 23 1.4 1.2 1.5 Kaur.et.al.(2005) [Fehler! Verweisquelle konnte nicht gefunden werden.] Walk 56 23.8 1.8 27.5 14.3 20.3 27.9 2.5 2.2 3 de.Nazelle.et.al.(2012)*w [Fehler! Verweisquelle konnte nicht gefunden werden.] Walk 38 20.6 1.4 21.6 6.3 18.4 23 1.4 1.2 1.5 Cycling to background Adams.(2001) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle Summer 40 30.7 1.7 34.5 19.7 25.9 36.4 2 1.7 2.4 Adams.(2001) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle Winter 56 20.2 1.8 23.5 15.1 17.3 23.6 1.6 1.3 1.8 de.Nazelle.et.al.(2012) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle 41 28.6 1.7 35 36.1 24.2 33.8 1.9 1.6 2.2 Kaur.et.al.(2005) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle 48 30.6 1.5 33.5 14.5 27.2 34.4 3.3 2.9 3.7 Zuurbier.et.al.(2010) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle HT 16 53 2.2 72.3 67 34.8 80.7 1.8 1.2 2.8 Zuurbier.et.al.(2010) [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle LT 16 52.9 2.2 71.7 65.5 34.8 80.5 1.8 1.2 2.8 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 21 - de.Nazelle.et.al.(2012)*c [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle 41 28.6 1.7 35 36.1 24.2 33.8 1.9 1.6 2.2 Zuurbier.et.al.(2010)*c [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle HT 16 53 2.2 72.3 67 34.8 80.7 2.1 1.4 3.2 Zuurbier.et.al.(2010)*c [Fehler! Verweisquelle konnte nicht gefunden werden.] Cycle LT 16 52.9 2.2 71.7 65.5 34.8 80.5 2.1 1.4 3.1 HT: high traffic, LT: Low traffic ^ Concentration of the numerator mode: geometric mean (gm), geometric standard deviation (gsd), arithmetic mean (am), standard deviation (sd), lower limit of 95.5% confidence interval (CI95.low), upper limit of 95.5% confidence interval (CI95.up) ^^ The ratio is the normalized quotient of the geometric means for the mode considered (walking or cycling) to the background level. Ratio lower (upper) bound is the lower (upper) limit of 95.5% confidence interval for the normalized ratio. * Studies reporting background concentrations during mode-specific sampling campaigns, with *w, *c, denoting respectively sampling measurements taken during walking, cycling. Note we have obtained both mode-specific and study average background concentrations from de Nazelle et al. 2012, and for Zuurbier et al. (2010) we estimated overall background concentrations based on sample-weighted mean concentrations across the study campaign. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 22 - As shown in Table 4, on average the various modes all had higher concentrations than the background concentrations, with mean ratios of 1.9 for walking to background and 2 for cycling. The higher concentrations during cycling were also found in another recent review of studies on personal exposure to air pollution during commuting, which however did not compare the exposure with the background concentrations and did not include walking studies [Fehler! Verweisquelle konnte nicht gefunden werden.]. Results for other modes are also shown in the Table. Four campaigns assessed cycling to walking concentrations and vice versa and found an average ratio of 1.3, and 0.8, respectively. Considering only campaigns that simultaneously measured a fixed monitoring site and the modes, there was only one available dataset for the walk to background comparisons and three for the bikes, respectively. Ratios were slightly lower in this case for walking vs background (1.4 vs 1.9; only one study campaign), and the same for bike vs. background (2; 2 study campaigns). To base the approach on the highest number of studies available, it was suggested to use the ratio of modes to overall background concentrations (walking: 1.9, cycling 2) rather than mode-specific background concentrations for the proposed HEAT tool approach. Car to walk and car to bike ratios were found to be 1.4 and 1.2 respectively Table 4. Summary of ratios across studies for all modes N study campaigns ratio CI95.ratio.l CI95.ratio.u 25th.perc 75th.perc Bike/walk 4 1.3 1.2 1.4 1.1 1.6 Bus/walk 4 1.5 1.4 1.6 1.3 1.7 car/walk 5 1.4 1.3 1.5 1.2 1.7 walk/bike 4 0.8 0.7 0.9 0.7 1 Bus/bike 8 1.1 1 1.2 1 1.2 car/bike 22 1.2 1.2 1.2 1.1 1.3 walk/background* 2 1.9 1.7 2.1 1.5 2.4 bike/background* 6 2 1.9 2.1 1.7 2.4 walk/background** 1 1.4 1.3 1.6 1.1 1.8 bike/background** 3 2 1.8 2.3 1.5 2.6 * Studies with overall background concentrations; ** studies with mode-specific background concentrations As agreed previously in principle during the HEAT consensus workshop in 2013, the simplest approach for users would be to request (optionally) data input solely on background PM2.5 concentrations in their study area (or alternatively to derive those based on PM10 values from international databases). The HEAT would then calculate automatically the difference in air pollution exposure for cycling or walking compared to background concentrations. Given that air pollution measurement studies in transport microenvironment are designed primarily to assess contrasts between modes (e.g. bike vs. walking or car vs. bike) rather than between background levels and modes (e.g. bike vs. background), it would be more appropriate to rely on between-mode ratios rather than using the walk-to-background and cycling-to-background ratios directly for HEAT walking and HEAT cycling, respectively. Therefore two steps are proposed to derive the conversion factors for the HEAT walking and cycling tools: 1) convert the background concentrations into one mode-specific concentration Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 23 - (walking (RWalkvBkgd = 1.9) or cycling (RbikevBkgd = 2); and 2) convert the walk or bike concentration to the other mode. However, different contrasts and concentrations are obtained depending on whether the comparison is made with walking or cycling modes in the first place. Specifically, taking an example with a background concentration of 10 µg/m3, if the cycling contrast is used as a basis, the cycling concentrations would be 20µg/m3 (10µg/m3 x 2) and the walking concentration when compared to walking would be16µg/m3 (20µg/m3 x 0.8, i.e. walk-bike ratio, see Table 4). If the walking contrast were to be used in the first place, the walking concentration would be 19µg/m3 (10µg/m3 x 1.9) and the bike concentrations as compared to walking would then be 24.7 µg/m3 (19µg/m3 x 1.3, i.e. bike to walk ratio, see Table 4), respectively. While this is a faithful representation of findings from the literature, it leads to slightly different results. Both approaches could be used depending on the contrast of interest for the application (cycling or walking). However, it is suggested to always use the bike to background concentration first, and estimate the walking concentration, if needed, in contrast to the bike concentration as this is based on a larger number of studies (6 campaigns vs 2, see Table 4). The annual average concentration of PM2.5 in the city (see also chapter 3.2) would be multiplied by the conversion factor (see Table 5) to estimate the concentration in the corresponding transport mode. The model also requires travel duration in minutes, or alternatively the distance in kilometres. This is part of the data input of the mortality module of HEAT. The current model transforms this input into minutes travelled based on average speeds for each mode of transport (4.8 km/h for walking or 14 km/h for cycling). Discussion The experts agreed to use the conversion factor derived from the literature as indicated in Table 5, noting that the conversion factor for walking is estimated by combining the conversion factors of cycling-to-background (2) and walking-to-cycling (0.8). Furthermore, it was concluded that while the studies used for these estimates are heterogeneous in their methods and provide results that are highly dependent on context, the experts expect an overestimation of the concentrations for the walking and cycling microenvironments with these recommended conversion factors. This is in line with the conservative approach of HEAT as it would lead to an overestimation of the health impacts of air pollution while walking/cycling. Table 5. Conversion factors based on average ratio for cycling and walking Concentration of PM2.5 Conversion factors Cycling / background concentration of PM2.5 2 Walking / background concentration of PM2.5 * 1.6 * The conversion factor for background-to-walking is achieved by combining the conversion factors from the literature of cycling-to-background (2.0) and walking-to-cycling - (0.8) and rounding to one decimal place. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 24 - Ventilation rates The estimation of the total daily intake is obtained by adding the intake during the trip (walking or cycling), and the intake of the rest of the day (intake during sleep and intake for resting time for reference scenario “staying at home”). It is proposed to calculate the minute ventilation based on a random population distribution and algorithms developed by the EPA [31, 32] from average Metabolic Equivalent of Task (METs) measured for cycling, walking, resting and sleeping as 6.8, 4, 1 and 1, respectively (see Table 6). The METs for cycling and walking are the same as the ones used in the current version of HEAT. Other ventilation rates have been suggested as well (e.g. [33-35]), but the proposed method seems to represent the most appropriate approach in a HEAT framework as it is based on a random population distribution rather than a study population. Table 6. Minute ventilation for different activities [31, 32] Minute ventilation (m3/hr) Sleepinga 0.27 Restinga 0.61 Walking 1.37 Cycling 2.55 a Reference scenario “staying at home”, including sleeping and resting. Discussion: The experts agreed on the proposed minute ventilation rates as listed in Table 6, pointing out that the underlying METs [36] are the same as those used in the current version of HEAT. 3.5 R isk assessment for a l l -cause mor ta l i ty The intake is used to estimate the change in relative risk for travelling in an active mode of transport compared to the reference scenario “staying at home”. To estimate this risk, it is necessary to use a relative risk and a dose-response function for exposure to PM2.5 and its impact on health. Using long-term epidemiological studies for this methodology is based on the following considerations: The basic assumption in the health impact assessment (HIA) of air pollution is that the target population of the HEAT assessment matches (with respect to the exposure range as well as demographic, health characteristics and susceptibility to the exposure) that of the underlying epidemiological studies providing the dose-response function. The question relevant for the HEAT is: do the people who cycle or walk comply with this assumption? Can we apply a dose-response function from a meta-analysis to them? In this regard, the following considerations can be made:  It seems likely that younger and healthier people are more likely to choose active modes of transport and might at the same time be less susceptible to detrimental effects of air pollution. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 25 -  On the other hand, people with pre-existing cardio-vascular or respiratory disease i.e. those responding more rapidly to air pollution exposure during physical activity than non-symptomatic people (e.g. [Fehler! Verweisquelle konnte nicht gefunden werden.,Fehler! Verweisquelle konnte nicht gefunden werden.]) may be more reluctant to cycle. However, as long as no specific air-pollution related RR for such more “active” groups (walkers or cyclists) are available; the use of a RR from long-term studies in the general population, including both more and less active people, was agreed to be a reasonable approximation. It is likely, however, that the expected health effects of air pollution might be over-estimated under this assumption. At the same time, traffic particles may be more toxic than the general background air pollution [Fehler! Verweisquelle konnte nicht gefunden werden., Fehler! Verweisquelle konnte nicht gefunden werden.], which would lead to an underestimation of the effects [Fehler! Verweisquelle konnte nicht gefunden werden.]. In addition, the air pollution dose is also dependent on the specific road chosen and distance to the main traffic flow [Fehler! Verweisquelle konnte nicht gefunden werden.,34]. However, HEAT is based on an average amount of walking or cycling which assumes a mix of everyday cycling and walking behaviour on different routes throughout a city or a country. The most recently published meta-analysis included 14 international cohort studies and summarized the relative risk between all-cause mortality and each increment of 10 µg/m3 of PM2.5 as 1.07 (1.04-1.09) [18] (Figure 1). Figure 1. Meta-analysis of the association between PM2.5 and all-cause mortality (relative risk per 10 µg/m3) [18] Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 26 - To estimate the number of deaths in the population, the relative risk associated with the inhalation of PM2.5 in different transport modes (pedestrian or cyclist) requires the calculation of the attributable fraction among the exposed ( Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 27 - Table 2) and mortality rate in the overall population (already provided in HEAT). Discussion: The experts agreed to use 1.07 as relative risk between all-cause mortality and each increment of 10 µg/m3 of PM2.5 in HEAT. 3.6 Rela t ive r isk for walk ing and cyc l ing tak ing in to account a i r pol lu t ion e f fec ts One open question from the previous HEAT consensus meeting regarding developing a separate HEAT air pollution model concerned possible double counting of the health impacts of air pollution by using the relative risk (RR) estimates derived from a meta-analysis of 14 walking and 7 cycling studies and all-cause mortality [Fehler! Verweisquelle konnte nicht gefunden werden.,Fehler! Verweisquelle konnte nicht gefunden werden.]. To study this possible effect, the effect of air pollution on the RRs of the walking or cycling studies included in the meta-analysis was calculated. The calculation of the effect on the RRs was performed based on the following steps: a) estimation of air pollution (PM2.5) exposure in each of the cohort studies and risk group (by intake during the cycling or walking period) and b) the assessment of the effect by applying the approach of multiplicative synergy of multiple exposures for epidemiological studies [Fehler! Verweisquelle konnte nicht gefunden werden.], calculating the RR from the air pollution exposure during walking or cycling and adjusting the original RR by this effect. Further details on the methodology can be found in Annex 5. The results of these calculations are shown in the Table 7 (cycling) and Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 28 - Table 8 (walking). Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 29 - Table 7. Relative risk of all-cause mortality and cycling and calculated risks adjusted by particulate matter <2.5 μm exposure during cycling Study author Reported exposure Exposure levels in MET/ hr/week RR for PA LCI UCI RR of PM2.5 Inhalationa RR for PA adjusted by PM2.5 LCI UCI Sahlqvist (2013) [Fehler! Verweisquelle konnte nicht gefunden werden.] 0 min/wk 0 (ref) 1 1-59 min/wk 3.4 0.96 0.78 1.17 1,0012 0,9588 0,7790 1,1685 60+ min/wk 10.2 0.91 0.84 0.99 1,0025 0,9077 0,8379 0,9875 Johnsen (2013) [Fehler! Verweisquelle konnte nicht gefunden werden.] Non-active 0 (ref) 1 Active – 2hrs/wk (women) 13.6 0.79 0.73 0.85 1,0055 0,7857 0,7260 0,8453 Active – 2hrs/wk (men) 13.6 0.91 0.85 0.97 1,0055 0,9050 0,8453 0,9647 Andersen (2011) [Fehler! Verweisquelle konnte nicht gefunden werden.] No cycling 0 (ref) 1 0-3 hr/wk 10.2 0.78 0.69 0.88 1,0041 0,7768 0,6900 0,8800 3-7 hr/wk 34 0.76 0.68 0.85 1,0134 0,7500 0,6800 0,8500 7+ hr/wk 61.2 0.7 0.62 0.78 1,0184 0,6873 0,6200 0,7800 Schnohr (2011) [Fehler! Verweisquelle konnte nicht gefunden werden.] 0-0.5 hr/day Slow 7.0 (ref) 1 0-0.5 hr/day Average 11.9 0.67 0.49 0.92 1,0005 0,6696 0,4626 0,8226 0-0.5 hr/day Fast 17.5 0.54 0.31 0.94 1,0011 0,5394 0,2927 0,8405 0.5-1.0 hr/day Slow 21 0.87 0.57 1.33 1,0000 0,8690 0,5381 1,1892 0.5-1.0 Average 35.7 0.7 0.51 0.95 1,0016 0,6992 0,4815 0,8494 0.5-1.0 Fast 52.5 0.44 0.28 0.69 1,0033 0,4395 0,2644 0,6170 1.0+ hr/day Slow 35 0.85 0.53 1.35 1,0000 0,8490 0,5004 1,2071 1.0+ hr/day Average 59.5 0.71 0.52 0.97 1,0021 0,7092 0,4909 0,8673 1.0+ hr/day Fast 87.5 0.68 0.46 1.01 1,0043 0,6792 0,4343 0,9031 Besson (2008) [Fehler! Verweisquelle konnte nicht gefunden werden.] Non-cyclist 0 (ref) 1 Cycling up to 30 min per week 1.7 1.02 0.77 1.35 1,0006 1,0193 0,7695 1,3491 Cycling over 30 min per week 5.1 1.01 0.76 1.36 1,0012 1,0087 0,7590 1,3583 Matthews (2007) [Fehler! Verweisquelle konnte nicht gefunden werden.] 0 MET-hours/day 0 (ref) 1 0.1-3.4 MET-hours/day 1.8 0.79 0.61 1.01 1,0094 0,7826 0,6043 1,0005 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 30 - 3.5+ MET-hours/day 5.2 0.66 0.4 1.07 1,0191 0,6476 0,3925 1,0499 Andersen (2000) [Fehler! Verweisquelle konnte nicht gefunden werden.] 0 min per week 0 (ref) 1 Ave 3 hours per week 20.4 0.72 0.57 0.91 1,0083 0,7141 0,5653 0,9025 MET: Metabolic equivalent of task; RR: Relative Risk; LCI: Low confident interval; UCI: Upper confident interval; PA: Physical activity; a PM2.5 inhalation according to the physical activity of each exposure group. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 31 - Table 8. Relative risk of all-cause mortality and walking and calculated risks adjusted by particulate matter <2.5 μm exposure during walking Study author Reported exposure Exposure levels in MET/hr/week RR for PA LCI UCI RR of PM2.5 Inhalationa RR for PA adjusted by PM2.5 LCI UCI Johnsen (men) (2013) [Fehler! Verweisquelle konnte nicht gefunden werden.] non-active 0 1 active = 3 hrs/wk 12 1.01 0.91 1.13 1,0027 1,0073 0,9075 1,1269 Johnsen (women) (2013) non-active 0 1 active = 3 hrs/wk 12 0.92 0.8 1.06 1,0027 0,9175 0,7978 1,0571 Nagai (men) (2013) [Fehler! Verweisquelle konnte nicht gefunden werden.] <1 hr/day 0 1 >1 hr/day 28 0.9 0.82 0.98 1,0051 0,8954 0,8158 0,9750 Nagai (women) (2013) <1 hr/day 0 1 >1 hr/day 28 0.95 0.82 1.1 1,0051 0,9452 0,8158 1,0944 Wang (2013) [Fehler! Verweisquelle konnte nicht gefunden werden.] none 0 1 5-6 hrs/wk @ 3.3 METs 18.15 0.79 0.69 0.86 1,0138 0,7793 0,6806 0,8483 Sabia (2012) [Fehler! Verweisquelle konnte nicht gefunden werden.] <3.5 hrs/wk 0 1 3.5-5.9 hrs/wk 12 0.83 0.63 1.08 1,0027 0,8278 0,6283 1,0771 6+ hrs/wk 22 0.81 0.61 1.07 1,0034 0,8073 0,6079 1,0664 Stamatakis (men) [Fehler! Verweisquell e konnte nicht gefunden werden.] (2009) None 0 1 Med (2 sessions/wk) 4 1.1 0.86 1.14 1,0013 1,0985 0,8589 1,1385 High (7 sessions/wk) 14 0.91 0.73 1.13 1,0045 0,906 0,7268 1,125 Stamatakis (women) (2009) [Fehler! Verweisquell e konnte nicht gefunden werden.] None 0 1 Med (2 sessions/wk) 4 0.89 0.69 1.14 1,0013 0,8888 0,6891 1,1385 High (7 sessions/wk) 14 0.92 0.72 1.18 1,0045 0,9159 0,7168 1,1747 Besson (2008) [Fehler! Verweisquelle konnte nicht gefunden werden.] Non walker 0 1 Walk 0-90 min/wk 3 0.96 0.8 1.15 1,0006 0,9594 0,7995 1,1493 Walk 90min+/wk 9 0.89 0.73 1.09 1,0012 0,8889 0,7291 1,0886 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 32 - Matthews (2007) [Fehler! Verweisquell e konnte nicht gefunden werden.] 0-3.4 MET.hrs/day 0 1 3.5-7 MET.hrs/day 24.85 0.94 0.89 1.09 1,0182 0,9232 0,8741 1,0705 7.1-10 MET.hrs/day 47.6 0.83 0.69 1 1,0279 0,8075 0,6713 0,9729 10+ MET.hrs/day 69.3 0.86 0.71 1.05 1,0318 0,8335 0,6881 1,0177 Schnohr (men) (2007) [Fehler! Verweisquelle konnte nicht gefunden werden.] <0.5 hrs/day 0 1 0.5-1.0 hrs/day 14 0.87 0.68 1.1 1,0024 0,8679 0,6784 1,0974 1.0-2.0 hrs/day 35 0.95 0.75 1.21 1,0045 0,9457 0,7466 1,2045 2.0+ hrs/day 63 0.89 0.69 1.14 1,0059 0,8848 0,6859 1,1333 Schnohr (women) (2007) [Fehler! Verweisquelle konnte nicht gefunden werden.] <0.5 hrs/day 0 1 0.5-1.0 hrs/day 14 1 0.77 1.3 1,0024 0,9976 0,7682 1,2969 1.0-2.0 hrs/day 35 1.04 0.8 1.36 1,0045 1,0353 0,7964 1,3538 2.0+ hrs/day 63 0.8 0.59 1.1 1,0059 0,7953 0,5865 1,0935 Smith (2007) [Fehler! Verweisquelle konnte nicht gefunden werden.] None 0 1 <1 mile/day 4.3 0.98 0.76 1.25 1,0016 0,9784 0,7588 1,248 >1 mile/day 12.9 0.89 0.67 1.18 1,0046 0,8859 0,6669 1,1746 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 33 - Table 8. continued Study author Reported exposure Exposure levels in MET/hr/week RR for PA LCI UCI RR of PM2.5 inhalationa RR for PA adjusted by PM2.5 LCI UCI Lee (2000) [Fehler! Verweisquelle konnte nicht gefunden werden.] <5 km/wk 0 1 5 to 10 km/wk 3.9 0.91 0.82 1.02 1,0007 0,9094 0,8194 1,0193 10 to 20 km/wk 9.6 0.92 0.83 1.01 1,0016 0,9185 0,8287 1,0084 >20 km/wk 17.3 0.84 0.75 0.94 1,0028 0,8376 0,7479 0,9373 Bath (1998) [Fehler! Verweisquelle konnte nicht gefunden werden.] Less than 10 min 0 1 More than 10 min 4.7 0.75 0.63 0.82 1,0004 0,7497 0,6297 0,8197 Hakim (1998) [Fehler! Verweisquelle konnte nicht gefunden werden.] Less than 0.5 hr 0 1 0.5-1.0 hr 9.3 0.66 0.47 0.91 1,0011 0,6593 0,4695 0,909 more than 1 hr 42 0.55 0.37 0.83 1,0014 0,5492 0,3695 0,8288 Wannamethee (1998) [Fehler! Verweisquelle konnte nicht gefunden werden.] 0 min/day 0 1 0-20 min/day 4.7 1.15 0.73 1.79 1,0007 1,1492 0,7295 1,7888 21-40 min/day 14 1.06 0.75 1.5 1,002 1,0578 0,7485 1,4969 41-60 min/day 23.3 0.97 0.65 1.46 1,0033 0,9668 0,6478 1,4552 60+ min/day 32.7 0.62 0.37 1.05 1,004 0,6175 0,3685 1,0458 LaCroix (1996) [Fehler! Verweisquelle konnte nicht gefunden werden.] Less than 1 hr 0 1 1-4 hr 8 0.83 0.53 1.29 1,0008 0,8293 0,5296 1,2889 more than 4 hr 20 0.91 0.58 1.42 1,0016 0,9085 0,5791 1,4177 MET: Metabolic equivalent of task; RR: Relative Risk; LCI: Low confident interval; UCI: Upper confident interval; PA: Physical activity; a PM2.5 inhalation according to the physical activity of each exposure group. The change in the RR of all-cause mortality and physical activity related to PM2.5 during the physical activity reported by each exposure group was small in all the cohort studies (percentage of change of 0.04% -3%). The magnitude of the change was low because only the incremental risk due to increased inhalation dose of air pollution during the periods of cycling or walking was estimated, and this increased inhalation dose is small compared to inhalation of air pollution during the rest of the day. Two studies were identified where the impact of the air pollution was greater than in others studies (percentage change of 0.9%-3%) [Fehler! Verweisquelle konnte nicht gefunden werden.,Fehler! Verweisquelle konnte nicht gefunden werden.]. These studies were conducted in the city of Shanghai, which has considerably higher concentrations of air pollution than other cities (PM2.5 annual concentrations of 40.5 μm/m3). Nevertheless, even in this highly polluted environment, the air pollution effects remain minor Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 34 - compared to those of physical activity, as also concluded by another recent review on air pollution exposure while commuting [Fehler! Verweisquelle konnte nicht gefunden werden.]. These findings are also confirmed by a forthcoming systematic review of health impact assessments of active transport, which found that the health impacts of the air pollution in the traveler were considerably smaller than the health impacts of the physical activity [Fehler! Verweisquelle konnte nicht gefunden werden.]. Discussion: The experts accepted the recommendation to consider the issue of double-counting as negligible with regard to adding the agreed calculations on health effects from air pollution to the current version of HEAT. 3.7 Input da ta required The following input data would be required to use the air pollution model:  Annual mean concentration of PM2.5 or PM10 in the place of interest o This could be provided similarly like the crude mortality rate is in the current version of HEAT (as available through the WHO databases), with the possibility of modification by the user.  Trip duration in minutes or distance travelled in km (already part of the existing HEAT mortality modules).  All-cause mortality rate in adults in the study population (already part of the existing HEAT mortality modules).  Environment of cycling/walking (is the cycling/walking mainly leisure or utilitarian, i.e. is it done in background or traffic environments? o This would need to be asked in a new question in HEAT. 3.7.1 Possible data sources for users The main source for air pollution data are the national or local environmental protection and air pollution authorities. In many cases, annual average values are available from their websites and/or annual reports There are two possible data source that could be integrated into HEAT to facilitate use of the air pollution module by users that are not air pollution experts:  Annual mean concentration of PM2.5 and PM10 in 1100 cities in 91 countries, reported by World Health Organization (2003 – 2010), http://www.who.int/phe/health_topics/outdoorair/databases/en/  Annual mean concentration PM10 in European cities, reported by European Environmental Agency (2011), http://www.eea.europa.eu/themes/air/interactive/pm10 Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 35 - 3.8 Presentat ion of resu l ts o f a i r po l lu t ion e f fec ts in HE AT The experts agreed to make the assessment of air pollution effects on all-cause mortality in HEAT optional for users and to present the results in a graphical manner together with the existing results of HEAT. The tool shall calculate and present the change in number of deaths per year due to the difference in exposure to air pollution in cyclists/walkers in absolute figures. However, to provide an appropriate perspective to the user, an estimate of the total number of attributable deaths due to the background air pollution in the population used for the HEAT assessment shall also be provided. Details on how to graphically display the combined results are to be further decided by the core group. Furthermore, the participants suggested providing the following caveats and information to accompany the presentation of the results:  The results are likely to overestimate the negative impact of air pollution on health while walking/cycling.  Mention that modal shifts impacts the health of the entire population and not only the cyclists/pedestrians (as estimated in HEAT). For example, larger shifts in mode use from car to cycling would lead to a reduction in air pollution exposure for the entire population.  Mention other co-benefits of modal shifts (e.g. climate change)  Include a recommendation to avoid heavily polluted (such as by vehicular emissions) areas in planning cycling and walking routes. 4 Additional considerations and future developments The participants of the consensus workshop further discussed additional aspects of HEAT and possible future developments. The following issues and suggestions were raised for consideration by the core group:  Review of the default discount rate of 5% currently used in HEAT in view of changed economic conditions in Europe.  Costs of morbidity: o Existing approaches to morbidity suggest that mortality costs make up between 50 to 85% of total costs. o Harry Rutter and Frank George updated the participants on the outcome of a recent expert meeting on the health economics of air pollution. This ongoing work, also as part of collaboration between WHO and the OECD, could serve as a basis for the inclusion of morbidity in HEAT.  David Rojas-Rueda presented an outlook on HEAT-related activities foreseen under an ongoing four- year research project on Physical Activity through Sustainable Transport Approaches (PASTA). The study includes a literature reviews on determinants of cycling and walking and on successful promotion approaches. A longitudinal study to better understand correlates of active mobility and their effects on overall physical activity, injury risk and air pollution will be carried out as well. The project Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 36 - will develop a detailed HIA model. The work will inform further updates of HEAT, for example regarding some of its assumptions, exposure measures and additional health effects.  Injuries: o Thomas Götschi provided an overview of possible approaches to including specific considerations of road traffic crashes as an additional module in the HEAT model. The main issue to tackle is the measurement of exposure as there is a high variability in crash risks regarding location, age, gender and other factors. A possible approach could be the use of carefully selected existing scenarios with published crash risks. It was agreed to initiate preparations for a more detailed discussion of such a module as next step of the HEAT process. o It was also suggested to consider possibly including in any additional risk calculations in HEAT some recommendations on risk reduction. For example, combine crash risk estimates with suggestions on how to improve road safety and air pollution risk estimates with the recommendation to separate cycling and walking from motorized traffic to reduce exposure.  Target audience and main messages: o Participants suggested to possibly consider other target audiences for future versions of HEAT, including in particular the medical professions and health sector in general. For this target groups, however, other economic measures than VSL might be needed. o Connected to the target audience, future HEAT versions should also consider specifying the main message and aligning it with other public health messages (e.g. current public health advice during air pollution episodes as to avoid physical activity). o Alternative health-relevant outcomes might also be considered in the future to make results more applicable to other target audiences. 5 Conclusions, next steps and closing The workshop concluded with Michal Krzyzanowski thanking all participants for their input and fruitful discussions. Christian Schweizer outlined the next steps, including:  comments on the background paper in terms of minor corrections or clarifications;  a brief follow-up meeting of the HEAT core group taking place after the closing of the consensus workshop to review conclusions and to decide on the specific next steps;  the core group to decide on how exactly to display and integrate the results of the air pollution estimations in the online HEAT website;  a timeline for the production of the updated version of HEAT including the estimates of air pollution effects; Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 37 -  conducting a survey among current HEAT users to better understand positive sides as well as challenges and barriers from using HEAT and on requested future developments;  further exploring the issue of possible substitution of leisure time physical activity by walking and cycling;  further exploring alternative approaches to costing of health effects;  further exploring possibilities to include changes in injury risks depending on the walking and cycling levels into HEAT; and  further exploring possibilities to include the health effects from reduced morbidity related to regular cycling and walking into HEAT. Francesca Racioppi closed the workshop with warm thanks to the chair for excellent guidance through the workshop and expressing appreciation on behalf of the WHO to all participants for the valuable support and inputs provided to the HEAT process. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 38 - 6 References 1. Johan de Hartog J, Boogaard H, Nijland H, Hoek G. Do the health benefits of cycling outweigh the risks? Environ Health Perspect. 2010;118:1109–16. 2. Rojas-Rueda D, de Nazelle A, Tainio M, Nieuwenhuijsen MJ. The health risks and benefits of cycling in urban environments compared with car use: health impact assessment study. BMJ. 2011;343:d4521. 3. Johnson T (2002): A guide to selected algorithms, distributions, and databases used in exposure models developed by the office of air quality planning and standards. US Environmental Protection Agency. 4. de Nazelle A, Rodríguez DA, Crawford-Brown D (2009): The built environment and health: impacts of pedestrian-friendly designs on air pollution exposure. The Science of the total environment 407(8):2525- 35. 5. WHO Regional Office for Europe (2014): WHO Expert Meeting: Methods and tools for assessing the health risks of air pollution at local, national and international level. Meeting report, Bonn, Germany, 12-13 May 2014. Copenhagen, WHO Regional Office for Europe. 6. Rutter H et al. Economic impact of reduced mortality due to increased cycling. Am J Prev Med. 2013;44:89–92. 7. Kahlmeier S., Cavill N., Dinsdale H., et al. (2011): Health economic assessment tools (HEAT) for walking and for cycling. Methodology and user guide. WHO Regional Office for Europe, Copenhagen, 2011. 8. Kelly P., Kahlmeier S., Götschi T., et al. (2014): Systematic review and meta-analysis of reduction in all-cause mortality from walking and cycling and shape of dose response relationship. International Journal of Behavioral Nutrition and Physical Activity. 11(132). 9. WHO Regional Office for Europe (2013): Development of the health economic assessment tools (HEAT) for walking and cycling. Meeting report of the consensus workshop in Bonn, Germany, 1–2 October 2013, WHO Regional Office for Europe: Copenhagen. 10. Kahlmeier S., Kelly P., Foster C., et al. (2014): Health economic assessment tools (HEAT) for walking and for cycling. Methodology and user guide – updated reprint 2014. WHO Regional Office for Europe, Copenhagen, 2014. 11. Rojas-Rueda D., Nieuwenhuijsen M. (2014): Adjustment of risk estimates of physical activity and mortality by the impact of air pollution (particulate matter of less than 2.5µm, CREAL Centre for Research in Environmental Epidemiology: Barcelona. 12. De Nazelle A., Bode O. (2014): Proposal for an approach to quantify air pollution exposure contrasts between modes to be used in the HEAT tool. Report to the WHO., Centre for Environmental Policy, Imperial College London: London. 13. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, Amann M, et al.A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012 Dec 15;380(9859):2224-60. doi: 10.1016/S0140-6736(12)61766-8. 14. WHO Regional Office for Europe. Review of evidence on health aspects of air pollution – REVIHAAP Project: Final technical report. Copenhagen. 2013 15. Ostro B (2004) Outdoor air pollution - Assessing the environmental burden of disease at national and local levels. World Health Organization, Geneva. Environmental Burden of Disease Series No.5. 16. Development of the Health Economic Assessment Tools (HEAT) for walking and cycling: consensus workshop. Meeting background document. Copenhagen: WHO Regional Office for Europe; 2013. 17. PM2.5, an indicator for fine particles. In: Air quality guidelines. Global update 2005. Particulate matter, ozone, nitrogen dioxide and sulfur dioxide. Copenhagen: WHO Regional Office for Europe, 2006: 40– 1 (http://www.who.int/phe/health_topics/outdoorair/outdoorair_aqg/en, accessed 18 October 2017). Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 39 - 18. WHO Regional Office for Europe (2014): WHO Expert Meeting: Methods and tools for assessing the health risks of air pollution at local, national and international level. Meeting report, Bonn, Germany, 12-13 May 2014. Copenhagen, WHO Regional Office for Europe. 19. Brook RD, Rajagopalan S. Particulate matter, air pollution, and blood pressure. J Am Soc Hypertens 2009; 3: 332-50. 20. Brook RD, Rajagopalan S, Pope CA 3rd, Brook JR, Bhatnagar A, Diez-Roux AV, Holguin F, Hong Y, Luepker RV, Mittleman MA, Peters A, Siscovick D, Smith SC Jr, Whitsel L, Kaufman JD; American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism. Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association. Circulation. 2010 Jun 1;121(21):2331-78. 21. Rao X, Montresor-Lopez J, Puett R, Rajagopalan S, Brook RD. Ambient air pollution: an emerging risk factor for diabetes mellitus. Curr. Diab. Rep. 2015;15:603. 22. Wilker EH, Preis SR, Beiser AS, Wolf PA, Au R, Kloog I, Li W, Schwartz J,Koutrakis P, DeCarli C, Seshadri S, Mittleman MA. Long-term exposure to fine particulate matter, residential proximity to major roads and measures of brain structure. Stroke. 2015 May;46(5):1161-6. 23. Andersen ZJ, de Nazelle A, Mendez MA, Garcia-Aymerich J, Hertel O, Tjønneland A, Overvad K, Raaschou-Nielsen O, Nieuwenhuijsen MJ. A Study of the Combined Effects of Physical Activity and Air Pollution on Mortality in Elderly UrbanResidents: The Danish Diet, Cancer, and Health Cohort. Environ Health Perspect. 2015 Jan 27. [Epub ahead of print]. 24. Karanasiou A, Viana M, Querol X, et al. (2014): Assessment of personal exposure to particulate air pollution during commuting in European cities--recommendations and policy implications. Sci Total Environ. Aug 15(490):785-97. doi: 10.1016/j.scitotenv.2014.05.036. Epub 2014 Jun 5. 25. Briggs, D.J., de Hoogh, K., Morris, C., Gulliver, J., 2008. Effects of travel mode on exposures to particulate air pollution. Environment international 34, 12-22. 26. Gulliver, J., Briggs, D.J., 2007. Journey-time exposure to particulate air pollution. Atmospheric Environment 41, 7195-7207. 27. Adams HS, Nieuwenhuijsen, MJ, Colvile RN et al. (2001): Fine particle (PM2.5) personal exposure levels in transport microenvironments, London, UK. The Science of the total environment 279, 29-44. 28. Kaur, S., Nieuwenhuijsen, M., Colvile, R., 2005. Personal exposure of street canyon intersection users to PM2.5, uhrafine particle counts and carbon monoxide in Central. London, UK. Atmospheric Environment 39, 3629-3641. 29. Zuurbier M, Hoek G, Oldenwening M et al. (2010): Commuters' exposure to particulate matter air pollution is affected by mode of transport, fuel type, and route. Environmental health perspectives 118, 783(787). 30. de Nazelle A, Fruin S, Westerdahl D et al. (2012): A travel mode comparison of commuters' exposures to air pollutants in Barcelona. Atmospheric Environment 59:151-159. 31. Johnson T (2002): A guide to selected algorithms, distributions, and databases used in exposure models developed by the office of air quality planning and standards. US Environmental Protection Agency. 32. de Nazelle A, Rodríguez DA, Crawford-Brown D (2009): The built environment and health: impacts of pedestrian-friendly designs on air pollution exposure. The Science of the total environment 407(8):2525- 35. 33. Karanasiou A, Viana M, Querol, Moreno T et al. (2014): Assessment of personal exposure to particulate air pollution during commuting in European cities - Recommendations and policy implications. Science of the Total Environment, http://dx.doi.org/10.1016/j.scitotenv.2014.05.036. 34. Int Panis L, de Geus B, Vandenbulcke G et al. (2010): Exposure to particulate matter in traffic: a comparison of cyclists and car passengers. Atmos Environ 2010;44(19):2263–70. Meeting report Consensus workshop, 11-12 December 2014, Bonn, Germany - 40 - 35. Int Panis L, Meusen R, Thomas I, De Geus B et al. (2011): Systematic analysis of health risks and physical activity associated with cycling policies «SHAPES». Final Report. Brussels : Belgian Science Policy (Research Programme Science for a Sustainable Development). 36. Ainsworth, B., Haskell, W., Herrmann, S., Meckes, N., Bassett, J.D., Tudor-Locke, C., Greer, J., Vezina, J., Whitt-Glover, M., Leon, A., 2011. 2011 Compendium of Physical Activities: a second update of codes and MET values. Medicine and Science in Sports and Exercise, 43, 1575-1581. Meeting report Annexes - 41 - Annex 1 Workshop programme Thursday, 11 December 2014 09:00 – 09:30 Registration 09:30 – 09:45 Welcome, introduction and election of the chair and rapporteur of the meeting Elizabet Paunovic, Head of Office, European Centre for Environment and Health, WHO/Europe Francesca Racioppi, WHO/Europe 09:45 – 10:00 Introduction to HEAT for walking and cycling, recent update Francesca Racioppi, WHO/Europe 10:00 – 10:30 Lessons learnt from HEAT since 2008: scope of the proposed update Christian Schweizer, WHO/Europe and Sonja Kahlmeier, University of Zurich 10:30 – 10:45 Methodology and proposed way of working for the consensus meeting Chair 10:45 – 11:15 Coffee/tea break 11:15 – 11:45 New evidence on the role of air pollution effects on all-cause mortality while walking and cycling and deriving mode-specific conversion factors for exposure David Rojas, Centre for Research and Environmental Epidemiology (CREAL) 11:45 – 12:30 Discussion of the review of epidemiological literature on air pollution while walking and cycling and all-cause mortality Chair 12:30 – 13:30 Lunch 13:30 – 14:30 Discussion of the review of epidemiological literature on air pollution while walking and cycling and all-cause mortality (continued) Chair 14:30 – 14:45 Suggestions for integration of air pollution exposure into HEAT David Rojas, Centre for Research and Environmental Epidemiology (CREAL) 14:45 – 15:15 Discussion and consensus on integrating air pollution exposure into HEAT Chair 15:15 – 15:45 Coffee/tea break 15:45 – 17:00 Discussion and consensus on integrating air pollution exposure into HEAT (continued) Chair 17:00 Closure day one Francesca Racioppi, WHO/Europe Meeting report Annexes - 42 - Friday, 12 December 2014 09:00 – 09:15 Welcome and summary of day one Francesca Racioppi, WHO/Europe 09:15 – 10:15 Discussion and consensus on integrating air pollution exposure into HEAT (continued) Chair 10:15 – 11:00 Discussion of possible future directions of developments for HEAT Chair Possible approaches to economic assessment of morbidity based on OECD’s approach developed for air pollution Harry Rutter, London School of Hygiene and Tropical Medicine Update on relevant developments in the PASTA project:  The PASTA HIA model and HEAT David Rojas, Centre for Research and Environmental Epidemiology (CREAL)  Possible approaches to including injuries into HEAT Thomas Götschi, University of Zurich 11:00 – 11:30 Coffee/tea break 11:30 – 12:15 Discussion of possible future directions of developments for HEAT (continued) Chair 12:15 – 12:25 Next steps and other items 12:25 Closure Francesca Racioppi, WHO/Europe 12:30 – 13:30 Lunch Meeting report Annexes - 43 - Annex 2 List of participants Karim Abu-Omar FA University Erlangen Nuremberg Institute for Sport Science and Sport Erlangen Germany Professor Hugh Ross Anderson St George’s University of London Population Health Research Institute London United Kingdom Tegan Boehmer Centers for Disease Control and Prevention National Center for Environmental Health Atlanta, Georgia United States of America Nick Cavill Cavill Associates Stockport Cheshire United Kingdom Hywell Dinsdale Consultant Manchester United Kingdom Francesco Forastiere Azienda Sanitaria Locale RME Department of Epidemiology Rome Italy Eszter Füzeki Johann Wolfgang Goethe-Universität Department of Sports Medicine Frankfurt am Main Germany Thomas Götschi University of Zurich Institute of Social and Preventive Medicine Zurich Switzerland Gerard Hoek University of Utrecht IRAS Utrecht The Netherlands Luc Int Panis VITO PHARE Mol Belgium Sonja Kahlmeier University of Zurich Institute of Social and Preventive Medicine Zurich Switzerland Paul Kelly University of Edinburgh Institute for Sport, Physical Education and Health Sciences Edinburgh United Kingdom Michal Krzyzanowski King's College London Environmental Research Group London United Kingdom Ian Mudway King’s College London Centre for Environment and Health London, United Kingdom Mark J Nieuwenhuijsen Centre for Research in Environmental Epidemiology (CREAL) Barcelona Spain Pekka Oja UKK Institute for Health Promotion Research Tampere Finland Laura Perez Swiss Tropical and Public Health Institute Basel Switzerland David Rojas Rueda Centre for Research in Environmental Epidemiology (CREAL) Barcelona, Spain Meeting report Annexes - 44 - Harry Rutter London School of Hygiene and Tropical Medicine London United Kingdom Marko Tainio University of Cambridge Centre for Diet and Activity Research (CEDAR) Cambridge United Kingdom James Woodcock University of Cambridge Centre for Diet and Activity Research (CEDAR) Cambridge United Kingdom World Health Organization Regional Office for Europe Frank George Division of Communicable Diseases, Health Security and Environment WHO European Centre for Environment and Health Bonn Germany Marie-Eve Heroux Division of Communicable Diseases, Health Security and Environment WHO European Centre for Environment and Health Bonn Germany Francesca Racioppi Division of Communicable Diseases, Health Security and Environment Copenhagen Denmark Christian Schweizer Division of Communicable Diseases, Health Security and Environment Copenhagen Denmark Meeting report Annexes - 45 - Annex 3 Sensitivity analysis of using PM2.5 versus elemental carbon as air pollution indicator (Provided by Mark Nieuwenhuijsen and Gerard Hoek, 2015) In the HEAT air pollution tool, mortality effects of air pollution are characterized with PM2.5. This choice is consistent with many air pollution health impact assessments. Many studies have documented that PM2.5 is less affected by traffic emissions than the elemental or black carbon content of PM or ultrafine particles [1]. PM2.5 may therefore represent commuting exposures less. To evaluate the sensitivity of the main HEAT tool to the choice of pollutant, we performed a sensitivity analysis with the EC concentration instead. We quantitatively evaluated elemental carbon (EC) as the alternative, because: 1. EC is substantially more traffic related than PM2.5; 2. EC or close proxies of EC (such as Black carbon, absorbance of PM) measurements have been performed extensively in commuting studies; 3. 3. A fairly well-established exposure response relationship exists for long-term exposure effects on mortality. For ultrafine particles the third condition is not well fulfilled. Current insights suggest that the main health-relevant component of the complex mixture of air pollution is likely in the particle, which is why we did not select NO2 or other gaseous pollutants. The input variables that were changed were:  Ratio of commuting versus background of 2.2 (larger than the 1.6 used in HEAT for PM2.5), based on a study in Arnhem, the Netherlands [2];  A relative risk of 1.06 per 1 µg/m3 EC taken from a review [1]  Background concentration. When a background EC concentration of 1.5 µg/m3 is used, we calculate an increase in mortality of 8.6 cases, starting from a background mortality of 1000 cases. This agrees very well with the calculations performed for PM2.5, the HEAT parameters and otherwise the same assumptions and a background concentration of 20 µg/m3 PM2.5, resulting in 9.2 extra cases. The results are sensitive to the selected background concentrations. The selected values were taken from the 2009-2010 ESCAPE measurements in the Netherlands [3]. In Barcelona, the background concentration is higher (~2.5 µg/m3) than in the Netherlands while PM2.5 is similar [3]. Using EC in the Barcelona setting would result in 14.3 extra mortality cases instead of 9.2 with PM2.5, estimates that are still very comparable, given the uncertainty of the estimates. Meeting report Annexes - 46 - References 1. Janssen NA, Hoek G, Simic-Lawson M, Fischer P, van Bree L, ten Brink H, Keuken M, Atkinson RW, Anderson HR, Brunekreef B, Cassee FR. Black carbon as an additional indicator of the adverse health effects of airborne particles compared with PM10 and PM2.5. Environ Health Perspect. 2011 Dec;119(12):1691-9. doi: 10.1289/ehp.1003369. Epub 2011 Aug 2. PMID: 21810552; PMCID: PMC3261976. 2. Zuurbier M, Hoek G, Oldenwening M, Lenters V, Meliefste K, van den Hazel P, Brunekreef B. Commuters' exposure to particulate matter air pollution is affected by mode of transport, fuel type, and route. Environ Health Perspect. 2010 Jun;118(6):783-9. doi: 10.1289/ehp.0901622. Epub 2010 Feb 25. PMID: 20185385; PMCID: PMC2898854. 3. Eeftens M et al. Spatial variation of PM2.5, PM10, PM2.5 absorbance and PMcoarse concentrations between and within 20 European study areas and the relationship with NO2 – Results of the ESCAPE project. Atmospheric Environment (62): December 2012, Pages 303-317. Meeting report Annexes - 47 - Annex 4 Data harmonization and derivation of conversion factors of air pollution exposures while walking or cycling compared to background concentrations (Extract from work paper by de Nazelle et al., 2014) i. When only arithmetic means (𝜇𝑥) and standard deviations (𝜎𝑥) were provided these were transformed into geometric means (?̃?𝑥) and geometric standard deviations (σ̃x). The reason for this is the distribution of air pollutant concentration X habitually follows a lognormal distribution [8]. ii. 𝑋~ 𝑙𝑜𝑔𝑁 (𝜇∗, 𝜎∗) (1) There are many advantages in representing the data with geometric moments. For example, and contrary to arithmetic means, the geometric mean of a ratio is the ratio of geometric means. Then it is much more meaningful to use geometric means when using normalized results (ratios to reference values), for example when estimating ratios between exposure in a transport mode X to a reference mode Y: ?̃? ( 𝑋 𝑌 ) = ?̃?𝑥 ?̃?𝑦 (2) In addition, the geometric moments are also robust estimates for the location and scale parameters of lognormal distributions: 𝑋~ 𝑙𝑜𝑔𝑁 (?̃?𝑥, σ̃x) (3) iii. The following equations were applied to estimate the geometric mean ?̃?𝑥 and the geometric standard deviation σ̃x of the distribution of air pollutant concentration from the given arithmetic mean 𝜇𝑥 and arithmetic standard deviation 𝜎𝑥: First, the standard deviation 𝜎𝑦 of the log-transformed data 𝑌 = 𝑙𝑛(𝑋) ~ 𝑁(𝜇𝑦, 𝜎𝑦), is calculated as: 𝜎𝑦 = √ln(1 + 𝜎𝑥 2 𝜇𝑥 2) (4) Then, the geometric mean is estimated by: ?̃?𝑥 = 𝜇𝑥 𝑒 (− 𝜎𝑦 2 2 ) (5) And finally, the geometric standard deviation is computed with: ω = 1 + ( 𝜎𝑥 𝜇𝑥⁄ ) 2 (6) σ̃x = 𝑒 𝜎𝑦 = 𝑒√ln (𝜔) (7) iv. Normalized results were calculated as the ratios of a mode A to a background concentrations measured at fixed monitoring stations (b) by dividing the geometric mean of mode A (?̃?𝐴) by the geometric mean of the background b (?̃?𝑏): ?̃?𝐴/𝑏 = ?̃?𝐴 ?̃?𝑏 (8) This ratio is considered as a ratio to a constant ?̃?𝑏 rather than the ratio of two log-normally distributed random variables (given the likely correlation between the background concentrations and the various modes, it was deemed best not to consider them to be two independent random variables, hence the choice to consider the denominator to be a constant). Meeting report Annexes - 48 - Hence, the result is a change in the scale parameter (?̃?𝐴) to ?̃?𝐴/𝑏 but σ̃A the shape parameter remains unchanged: 𝐴 𝑏⁄ ~ 𝑙𝑜𝑔𝑁 (?̃?𝐴/𝑏 , σ̃A) (9) v. Comparison of ratios across studies were made by considering the ratios of different studies as n multiple independent log-normally distributed random variables, Xi. We consider a new random variable M log-normally distributed, defined as the geometric mean of normalized ratios across studies: M = (∏ 𝑋𝑖 𝑛 𝑖 ) 1 𝑛⁄ Where: Xi = A𝑖 b𝑖 ; i = 1, 2, … n; n being the number of different studies/cities. A𝑖 is the transport mode A (cycling or walking) for study “i” and b𝑖 the corresponding background concentration in study “i”. The geometric mean of ratios across studies is now: ?̃?𝑀 = (∏ ?̃?𝑖 𝑛 𝑖 ) 1 𝑛⁄ Meeting report Annexes - 49 - Annex 5 Summary of studies identified on conversion factors or air pollution exposure Study City Measurem ent period Study design Routes Instrument Mode Concentration arithmetic mean (PM2.5 μg/m 3)* Fixed Monitor arithmetic mean (μg/m3)* Roadside ¦ U Bkgd Int Panis (2010) 3 towns, Belgium Jun-09 Drive followed by cycle trip 1 loop/city mix of traffic and cycle lanes TSI DustTrak (light scattering) car 23.2 cyclist 27.2 Booga ard (2009) 11 cities, Netherla nds August - October 2006, 12h- 19h. Simultaneous bike/car on 12 routes in each city same origin and destination but not always same route/mode TSI DustTrak (light scattering) car 49.4 cyclist 44.5 Briggs (2008) London, UK May-June 2007 Simultaneous walk/car 48 routes, mix of urban, suburban, city centre OSIRIS (light scattering) car 4.8 21 13 pedestri ans 10.0 Kaur (2005 b) London, UK April-May 2003. Morning, lunch, afternoon 4 volunteers at a time with simultaneous modes, 5 modes 2 Routes 1)heavy traffic; 2) congested + for ped and bike mixed with back streets Adam's sampler (gravimetric) car 38.0 22.6 9.9 bus 34.5 cyclist 33.5 pedestri ans 27.5 Zuurbi er (2010) Arnhem, Netherla nds June 2007 - June 2008. 8- 10am simultaneous: diesel & petrol car/ electric& diesel bus/ Hi- &lo-traffic cycling same high traffic route for car, bus and high traffic cycle; low traffic cycling route close by DataRAM (pDR)1200 (light scattering) car 80.7 22.8 bus 66.4 cyclist 68.7 McNa bola (2008) Dublin, Ireland January 2005 - June 2006; peak morning + evening 2 simultaneous modes on same route. 2 routes: 1) congested, with footpaths, bus lane and cycle lane. 2) less congested, with bus lane and cycle lane Adam's sampler (gravimetric) car 85.8 bus 115.8 cyclist 80.5 pedestri ans 64.8 de Nazell e (2012) Barcelon a May-June 2009. throughout the day 2 simultaneous modes on same route. 2 routes, both mix of traffic & presence of bike lane DustTrak (light scattering) car 35.5 15 bus 25.9 cyclist 35.0 pedestri ans 21.6 Gulliv er & Briggs (2004) Northam pton, UK winter 1999- 2000, morning & afternoon peak Simultaneous car-walk. 2 routes: 1) Hi-traffic, w/ some minor rds; 2) suburban less traffic, some congestion OSIRIS (light scattering) car 15.5 26.5 pedestri ans 15.1 Gulliv er & Briggs (2007) Leiceste r, UK Winter 2004-2005 Simultaneous car-walk. 2 routes of mix traffic and urban design OSIRIS and DUSTMATE (light scattering) car 11.2 11.8 (rural) pedestri ans 15.7 Adam s (2001) London, UK July 1999 and February 2000, morning & afternoon 4 simultaneous sampling, multiple comparisons 3 routes:1)highly congested + canyon effect, 2)mix, 3) some congestion but lower density Adam's sampler (gravimetric) car 36.8 29.1 14.2 bus 39.0 cyclist 28.1 * Arithmetic mean concentrations are provided here for illustrative purposes but were not used as such in the current analysis. Where several campaigns (seasons, routes, cities) were presented in the study, the weighted mean arithmetic concentrations are shown here. Meeting report Annexes - 50 - References Adams, H.S., Nieuwenhuijsen, M.J., Colvile, R.N., McMullen, M.A., Khandelwal, P., 2001. Fine particle (PM2.5) personal exposure levels in transport microenvironments, London, UK. The Science of the total environment 279, 29-44. Boogaard, H., Borgman, F., Kamminga, J., Hoek, G., 2009. Exposure to ultrafine and fine particles and noise during cycling and driving in 11 Dutch cities. Atmospheric Environment 43, 4234-4242. Briggs, D.J., de Hoogh, K., Morris, C., Gulliver, J., 2008. Effects of travel mode on exposures to particulate air pollution. Environment international 34, 12-22. de Nazelle, A., Fruin, S., Westerdahl, D., Martinez, D., Ripoll, A., Kubesch, N., Nieuwenhuijsen, M., 2012. A travel mode comparison of commuters' exposures to air pollutants in Barcelona. Atmospheric Environment 59, 151-159. Gulliver, J., Briggs, D.J., 2007. Journey-time exposure to particulate air pollution. Atmospheric Environment 41, 7195-7207. Int Panis, L., de Geus, B., Vandenbulcke, G., Willems, H., Degraeuwe, B., Bleux, N., Mishra, V., Thomas, I., Meeusen, R., 2010. Exposure to particulate matter in traffic: A comparison of cyclists and car passengers. Atmospheric Environment 44, 2263-2270. Kaur, S., Nieuwenhuijsen, M., Colvile, R., 2005. Personal exposure of street canyon intersection users to PM2.5, ulrafine particle counts and carbon monoxide in Central. London, UK. Atmospheric Environment 39, 3629-3641. McNabola, A., Broderick, B.M., Gill, L.W., 2008. Relative exposure to fine particulate matter and VOCs between transport microenvironments in Dublin: Personal exposure and uptake. Atmospheric Environment 42, 6496-6512. Zuurbier, M., Hoek, G., Oldenwening, M., Lenters, V., Meliefste, K., van den Hazel, P., Brunekreef, B., 2010. Commuters' exposure to particulate matter air pollution is affected by mode of transport, fuel type, and route. Environmental health perspectives 118, 783(787). Meeting report Annexes - 51 - Annex 6 Methodology to assess air pollution effects on relative risks from cohort studies included in HEAT meta- analysis on effects of cycling or walking on all-cause mortality (Extract from Rojas-Rueda D., Nieuwenhuijsen M. (2014): Adjustment of risk estimates of physical activity and mortality by the impact of air pollution (particulate matter of less than 2.5µm, CREAL centre for research in environmental epidemiology: Barcelona) Methods The studies included in this analysis were derived from the meta-analysis of HEAT for walking and cycling. Using an extraction tool, we identified the main data from each study (cohort, population description, year, city, region and country). The annual average concentrations of PM2.5 for each cohort (city or region) were searched in two databases (WHO Outdoor Air Pollution database, 2011; EEA Airbase Europe, 2011). If the concentration of PM2.5 were not available, the particulate matter <10 μm (PM10) concentrations were used and transformed to PM2.5 using a conversion factor reported by Ostro [15]. That was the case of five cohort studies (Besson, 2008; Metthews, 2007; Bath, 1998; Nagai, 2013; Wang, 2013). One study included in the meta-analysis of HEAT for walking was based in the city of Ohsaki, Japan (Nagai, 2003), for which did jno concentrations of PM2.5 or PM10 were available from the databases. In this case the annual average concentration of PM10 of Tokio was used, because Ohsaki is part of the metropolitan area of Tokyo, and was transformed to PM2.5 concentrations. Another study (Johnsen, 2013) was performed in two different cities (Copenhagen and Aarhus), but given the lack of data to differentiate the RR of cycling and mortality for each city, was assumed the city of Copenhagen as a reference city and its PM2.5 concentrations was applied for both cities. Two studies were conducted in multiple locations, i.e. the British Regional Heart Study in 24 locations in the United Kingdom (Wannamerhee, 1998) and the Scottish Health Survey, which included all of Scotland (Stamatakis, 2009). In both cases, PM2.5 levels were derived from the average annual concentration in the United Kingdom. The inhaled doses of PM2.5 were estimated for each reported exposure group and each study according to the inhalation rate (it was assumed that the only change was the inhalation rate during the duration of walking or cycling and inhalation during other periods stayed the same) (see table 4 below), duration of the exposure and PM2.5 concentration. Using the dose-response function reported by Hoek, et al (2013) for each increment of 10g/m3 of PM2.5 (RR= 1.06, CI 1.04-1.08), were estimated the relative risks (RR) of all-cause mortality and exposure to PM2.5 in each exposure group and study. Meeting report Annexes - 52 - Finally the relative risks of all-cause mortality and physical activity reported by each exposure group and cohort study were adjusted with the exposure of PM2.5 during the physical activity period. In this adjustment we applied the approach of multiplicative synergy of multiple exposures for epidemiological studies (Dubin, 1999). Results Table 1 and table 2 include a description of the studies evaluated and the levels of air pollution (annual average concentration of PM2.5) in the city or area where the cohort studies were performed. Annex - 53 - Table 1. Description of the population/air pollution concentrations of the studies included in the meta-analysis used in HEAT for cycling. Study author Cohort study (population) City (Country) Year Annual concentration of PM2.5 (g/m3) Year of annual concentration Air pollution reference Sahlqvist (2013) European Prospective Investigation into Cancer (EPIC)- Norfolk Norwich (United Kingdom) 1993-2000 13.8 2012 EEA (22,450 women and men / 40-79 years) Johnsen (2013) Diet, Cancer and Health Study Copenhagen/Aarhus (Denmark) 1993-2010 14.8 2008 (Copenhagen) WHO (29,129 women and 26,576 men / 50-64 years) Andersen (2011) Copenhagen City Heart Study Copenhagen (Denmark) 1964-1994 14.8 2008 WHO (14,976 women and men / 20-93 years) Schnohr (2011) Copenhagen City Heart Study Copenhagen (Denmark) 1991-1994 14.8 2008 WHO (19,698 women and men / 20-100 years) Besson (2008) European Prospective Investigation into Cancer (EPIC)- Norfolk Norwich (United Kingdom) 1993-2006 13.8 2012 EEA (14,903 women and men / 45-79 years) Matthews (2007) Shanghai Women's Health Study Shanghai (China) 1997-2004 40.5* 2009 WHO (67,143 women / 40-70 years) Andersen (2000) Copenhagen City Heart Study /Glostrup Population Study / Copenhagen Male Study Copenhagen (Denmark) 1964-1994 14.8 2008 WHO (13,375 women and 17,265 men / 20-93 years) * Air pollution data come from PM10, and was transformed to PM2.5 by convention factors according to the country, reported by Ostro B, 2004. MET: Metabolic equivalent of task; RR: Relative Risk; LCI: Low confident interval; UCI: Upper confident interval; PM2.5: Particulate matter less than 2.5 micrometers; EEA: European Environmental Agency; WHO: World Health Organization. Meeting report Annexes - 54 - Table 2. Description of the population/air pollution concentrations of the studies included in the meta-analysis used in HEAT for walking. Study author Cohort study (population) City (Country) Year Annual concentration of PM2.5 (g/m3) Year of annual concentration Air pollution reference Bath (1998) Nottingham Logitudinal Study of Activity and Ageing (NLSAA) Nottingham (United Kingdom) 1985-1993 11.6 2012 EEA (1,042 women and men / >65 years) Hakim (1998) Honolulu Heart Program Honolulu (USA) 1980-1992 6.9 2009 WHO (707 men / 61-81 years) LaCroix (1996) Group Health Cooperative of Puget Sound (GHC) Western Washington (USA) 1994 9.8 2009 WHO (3,823 women and men / 65 and older) Besson (2008) European Prospective Investigation into Cancer (EPIC)- Norfolk Norwich (United Kingdom) 1993-2006 13.8 2012 EEA (14,903 women and men / 45-79 years) Wannamethee (1998) British Regional Heart Study 24 towns (United Kingdom) 1995-2000 13.5 2008 (UK)** WHO (4,311 men / 40-59 years) Stamatakis (2009) Scotish Health Survey Scotland (United Kingdom) 1995 - 2003 13.5 2008 (UK)** WHO (7,624 women and 6,102 men / > 35 years) Lee (2000) Harvard Alumni Health Study Boston (USA) 1977-1992 10.2 2009 WHO (13,485 men / 57.5 years) * Air pollution data come from PM10, and was transformed to PM2.5 by convention factors according to the country, reported by Ostro B, 2004. ** Concentrations from the city not available/ or multiple cities study, in this case we use tha annual average concentration of the country or region. MET: Metabolic equivalent of task; RR: Relative Risk; LCI: Low confident interval; UCI: Upper confident interval; PM2.5: Particulate matter less than 2.5 micrometers; EEA: European Environmental Agency; WHO: World Health Organization. Meeting report Annexes - 55 - Continuation - Table 2. Description of the population/air pollution concentrations of the studies included in the meta-analysis used in HEAT for walking. Study author Cohort study (population) City (Country) Year Annual concentration of PM2.5 (g/m3) Year of annual concentration Air pollution reference Matthews (2007) Shanghai Women's Health Study Shanghai (China) 1997-2004 40.5* 2009 WHO (67,143 women / 40-70 years) Sabia (2012) Whitehall II Study London (United Kingdom) 1997-2009 13.5 2008 WHO (7,456 women and men / 55.9 years) Schnohr (2007) Copenhagen City Heart Study Copenhagen (Denmark) 1991-1994 14.8 2008 WHO (19,329 women and men / 29-93 years) Smith (2007) Study Heart Disease Risk Factors Rancho Bernardo (USA) 1984-1994 15.0 2009 WHO (1,664 women and men / 50-90 years) Nagai (2011) Ohsaki National Health Insurance (NHI) Cohort Study Ohsaki (Japan) 1994-2007 24 Tokio 2005 Ref. (12,217 women and 15,521 men / 40-79 years) Johnsen (2013) Diet, Cancer and Health Study Copenhagen/Aarhus (Denmark) 1993-2010 14.8 2008 (Copenhagen) WHO (29,129 women and 26,576 men / 50-64 years) Wang (2013) Shanghai Men's Health Study Shanghai (China) 2002-2009 40.5* 2009 WHO (61,477 men / 40-74 years) * Air pollution data come from PM10, and was transformed to PM2.5 by convention factors according to the country, reported by Ostro B, 2004. MET: Metabolic equivalent of task; RR: Relative Risk; LCI: Low confident interval; UCI: Upper confident interval; PM2.5: Particulate matter less than 2.5 micrometers; EEA: European Environmental Agency; WHO: World Health Organization. Annex - 56 - Air pollution model The air pollution assessment model considered the exposure to particulate matter less than 2.5 m (PM2.5), which has shown strong associations with all-cause mortality. We compared exposure concentrations and inhaled dose between the exposure groups. We estimated yearly inhaled dose of PM2.5 accounting for specific inhalation rates (Annex - Table 1 and 2), exposures, and activity duration. To estimate the RR of mortality associated with the change of pollutant intake for each exposure group compared to reference group, we applied the ratio (equivalent change) between the estimated inhaled doses for each active group (Annex Table 2). Table 4. Minute ventilation used for each activity Minute ventilation (m3/hr) a Sleeping 0.27 Resting 0.61 Walking 1.37 Cycling 2.55 Minute ventilation in bike is calculated using a random population distribution and algorithms developed by the EPA (Johnson 2002; de Nazelle et al. 2009) from average METs measured for [Cycling, walking, sleeping, resting] = [6.8, 4, 1, 1]. Table 5. Air pollution model formulas Inhaled dose (g/day) Minute ventilation(m3/h) * Duration(h/day) * Concentration(µg/m3) Total dose (g/day) Inhaled dose during Sleep + Rest + Activity (exposure group) Equivalent change (g/m3) Total dose in the exposure group -1 * Mean concentration of pollutant Total dose in the reference group Relative risk for air pollution Exp[ Ln (RR10) * Equivalent change ] a 10 a RR10: Dose response function used in this case RR= 1.06 CI 1.04-1.08, reported by Hoek et al. Meeting report Annexes - 57 - References Andersen, L.B., et al., All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch Intern Med, 2000. 160(11): p. 1621-8. Andersen, L.B. and A.R. Cooper, Commuter cycling and health, in Transport and Health Issues 2011: Studies on mobility and transport research, W. Gronau, K. Reiter, and R. Pressl, Editors. 2011, Verlag MetaGISInfosysteme: Mannheim, Tyskland. p. 9-19. Bath PA, Morgan K: Customary physical activity and physical health outcomes in later life. Age & Ageing 1998, 27 Suppl 3:29-34. Besson H, Ekelund U, Brage S, Luben R, Bingham S, Khaw KT, Wareham NJ: Relationship between subdomains of total physical activity and mortality. Medicine and Science in Sports and Exercise 2008, 40:1909-1915. de Nazelle,A., Rodriguez,D.A., and Crawford-Brown,D. The built environment and health: impacts of pedestrian-friendly designs on air pollution exposure. Sci. Total Environ. 2009, 407: 2525-2535 Dubin J. Statistical Analysis of the Additive and Multiplicative Hypotheses of Multiple Exposure Synergy for Cohort and Case-Control Studies. Social Science Working Paper 1066, California Institute of Technology. 1999. EEA. Airbase Europe 2011. http://www.eea.europa.eu/themes/air/interactive/pm10 Hakim AA, Petrovitch H, Burchfiel CM, Ross GW, Rodriguez BL, White LR, Yano K, Curb JD, Abbott RD: Effects of walking on mortality among nonsmoking retired men. New England Journal of Medicine 1998, 338:94-99. Hoek G, Krishnan RM, Beelen R, Peters A, Ostro B, Brunekreef B, Kaufman JD. Long-term air pollution exposure and cardio- respiratory mortality: a review. Environ Health. 2013 May 28;12(1):43. Johnsen, N., et al., Leisure time physical activity and mortality. Epidemiology, 2013. 24(5): p. 717-25. LaCroix AZ, Leveille SG, Hecht JA, Grothaus LC, Wagner EH: Does walking decrease the risk of cardiovascular disease hospitalizations and death in older adults? Journal of the American Geriatrics Society 1996, 44:113-120. Lee IM, Paffenbarger RS, Jr.: Associations of light, moderate, and vigorous intensity physical activity with longevity. The Harvard Alumni Health Study. Am J Epidemiol 2000, 151:293-299. Matthews CE, Jurj AL, Shu XO, Li HL, Yang G, Li Q, Gao YT, Zheng W: Influence of exercise, walking, cycling, and overall nonexercise physical activity on mortality in Chinese women. Am J Epidemiol 2007, 165:1343-1350. Mueller N, Rojas-Rueda D, Cole-Hunter T, de Nazelle A, Dons E, Gerike R, Götschi T, Int Panis L, Kahlmeier S, Nieuwenhuijsen MJ. Health impact assessment of active transport policies: a systematic review. Submited. Nagai M, Kuriyama S, Kakizaki M, Ohmori-Matsuda K, Sone T, Hozawa A, Kawado M, Hashimoto S, Tsuji I: Impact of walking on life expectancy and lifetime medical expenditure: the Ohsaki Cohort Study. BMJ Open 2011, 1:e000240. Oja, P., et al., Health benefits of cycling: a systematic review. Scand J Med Sci Sports, 2011. 21(4): p. 496- 509. Ostro B (2004) Outdoor air pollution - Assessing the environmental burden of disease at national and local levels. Environmental Burden of Disease Series No.5. World Health Organization, Geneva. Sabia S, Dugravot A, Kivimaki M, Brunner E, Shipley MJ, Singh-Manoux A: Effect of intensity and type of physical activity on mortality: results from the Whitehall II cohorte study. American Journal of Public Health 2012, 102:698-704. Schnohr P, Scharling H, Jensen JS: Intensity versus duration of walking, impact on mortality: the Copenhagen City Heart Study. European Journal of Cardiovascular Prevention & Rehabilitation 2007, 14:72-78. Smith TC, Wingard DL, Smith B, Kritz-Silverstein D, Barrett-Connor E: Walking decreased risk of cardiovascular disease mortality in older adults with diabetes. Journal of Clinical Epidemiology 2007, 60:309-317. Stamatakis E, Hamer M, Lawlor DA: Physical activity, mortality, and cardiovascular disease: is domestic physical activity beneficial? The Scottish Health Survey -- 1995, 1998, and 2003. Am J Epidemiol 2009, 169:1191-1200. Wang N, Zhang X, Xiang YB, Li H, Yang G, Gao J, Zheng W, Shu XO: Associations of tai chi, walking, and jogging with mortality in chinese men. Am J Epidemiol 2013, 178:791-796. Wannamethee SG, Shaper AG, Walker M: Changes in physical activity, mortality, and incidence of coronary heart disease in older men. Lancet 1998, 351:1603-1608. WHO. Urban outdoor air pollution database, 2011, Geneva. http://www.who.int/phe/health_topics/outdoorair/databases/en/