Measurement and modeling of short- and long-term commuter exposure to traffic-related air pollution

Background. Many epidemiological studies have reported associations between traffic-related air pollu¬tion exposure and acute and chronic health problems. Exposure assignment in those stud¬ies has typically relied on home outdoor locations and ignored exposure during commuting and at non-residential locations. However, because of high concentrations of harmful air pollutants in proximity to traffic, time spent in transport may contribute considerably to a person’s total daily exposure to traffic-related air pollution. An understanding of how activity patterns affect exposure to traffic-related air pollution in space and time is important for im¬proved exposure assessments.
Concentration levels and individuals’ exposures to harmful traffic-related air pollutants in the various transport microenvironments are not well understood. Recently, exposure to ultrafine particles (UFP, particles smaller than 100 nm) has attracted particular interest. UFP are considered harmful to human health in view of their small size and the probability to penetrate deeply into the respiratory tract. Little is known about the variability in UFP concentrations and most notably the average particle size in various transport environ¬ments. This is largely due to the lack of a robust portable device to measure UFP charac¬teristics.
Objectives. The aim of this thesis was to characterize exposure to both UFP concentration and average particle size distribution diameters in commonly used transport environments in Basel. In addition, a simulation of commuter exposure to traffic-related air pollution of a general pop-ulation was carried out to estimate the contribution of commute (i.e., the time spent in traffic traveling between home and work or school) to total exposure and inhalation dose as well as its relevance in epidemiological studies on long-term health effects of traffic-related air pollution.
Methods. Three sub-studies were performed to characterize personal exposure to UFP concentration and average particle size distribution diameters in frequently traveled commuter microenvi-ronments. The personal monitoring campaign was carried out in the city of Basel and sur-rounding area between December 2010 and September 2011 using a newly developed portable device, the miniature Diffusion Size Classifier (miniDiSC), which measures particles in the size range of 10 to 300 nm. First, the spatial variation of sidewalk UFP exposures within urban areas and transport-specific microenvironments was explored. Measurements were conducted along four predefined walks once per month. Second, exposure to UFP concentration and average particle size were quantified for five modes of transportation (walking, bicycle, bus, tram, car) during different times of the day and week, along the same route. Finally, the contribution of bicycle commuting along two different routes (along main roads, away from main roads) to total daily exposures was assessed by 24-hour personal measurements. Measurements were equally distributed over weekdays (Monday to Friday) across three seasons – winter, spring and summer.
The simulation of commuter exposure to traffic-related air pollution was conducted based on spatially and temporally resolved data on commuter trips of residents working (or attending a school) within the Basel area (Cantons Basel-City and Basel-Country). The information on commuter routes, transportation modes and home, work and school locations were ex¬tracted from the year 2010 Swiss Mobility and Transport Microcensus survey. An approach to simulate travel routes based on the transportation mode and origin/destination location of the legs (pieces of the trips with the same transportation mode) was developed and vali¬dated. Individuals’ exposures to NO2 during commuting and at home, work and school loca¬tions were computed by overlapping the locations and travel routes with annual mean maps of NO2 in a geographic information system (GIS). Three air pollution models (a land use regression model (LUR), a high and a low resolution dispersion model) were evaluated for estimating commuter exposures to NO2 as a marker of long-term exposure to traffic-related air pollution. Finally, the bias in health effect estimates resulting from using home outdoor exposures only and ignoring other non-residential exposures including commuter exposure was quantified.
This thesis is part of the Europe-wide project, Transportation Air Pollution and Physical ActivitieS (TAPAS), which is an integrated health risk assessment program on climate change and urban policies.
Results. In general, smaller average particle sizes and higher UFP concentration levels were meas-ured at places and for transportation modes in close proximity to traffic. Average trip UFP concentrations were highest in car (31,800 particles cm-3) followed by bicycle (22,700 parti-cles cm-3), walking (19,500 particles cm-3) and public transportation (14,100-18,800 particles cm-3). Concentrations were highest for all transportation modes during weekday morning rush hours, compared to other time periods. UFP concentration was lowest in bus, regard¬less of time period. Average particle diameters followed an opposite trend than UFP con¬centration, showing larger average particle sizes for transportation modes and sampling times with lower UFP number concentrations and vice versa. Bicycle travel along main streets between home and work place (24 min on average) contributed 21% and 5% to total daily UFP exposure in winter and summer, respectively. Contribution of bicycle commutes to total daily UFP exposure could be reduced by half if main roads were avoided.
Within Basel-City, estimated average time-weighted NO2 population exposure during com-muting was similar among all air pollution models (around 39-41 µg m–3). The spatial varia-bility in NO2 concentrations, as typically encountered in urban street environments, was best reflected by the dispersion model with the highest resolution (grid size of 25 m). By com-parison, both the LUR model (applied to a 50x50 m grid) and the dispersion model with a lower resolution (100x100 m) underestimated the NO2 concentrations on the higher end, and overestimated the values on the lower end.
The population working (>= 50% work load) or attending a school within the region of Basel spent on average 49 minutes for daily commutes. Work or school occupied 22% of the subjects’ time on average. Median contribution of commuting to total weekly NO2 exposure was 2.7% (range 0.1-13.5%). With regard to inhalation dose, the commute contributed slightly more when assuming moderate (3.5%, range: 0.2-16.8%) or high (4.2%, range: 0.2-33.0%) breathing rates during active transportation. The median contribution of commute to the total NO2 exposure was highest for subjects using mainly public transportation (4.7%, range: 1.3-13.5%) who also spent the longest time in traffic (more than an hour). The com¬parison between the transportation modes based on the legs of the trips, however, revealed the highest NO2 exposures for motorized transportation.
The failure to differentiate between outdoor NO2 exposure at work/school and at home could result in a 12% (95%-CI: 11-14%) underestimation of related health effects. This bias was stronger for the subjects commuting between Basel-City and the rural to suburban sur-rounding areas of Basel-Country (33% underestimation) than for the subjects commuting within those areas. For the same population sub-group, potentially significant underestima¬tion of health effects (5%, 95%-CI: 4-5%) attributable to including outdoor exposures at home and at work/school but omitting exposure during the commute was found.
Conclusions and Outlook. This thesis provides important insights in the spatial and temporal variability of UFP within an urban area and provides an approach for modeling commuter exposures to traffic-related air pollution in epidemiological studies. Results confirmed the expectation that people are exposed to potentially high exposures during their daily travels and that ignoring time-activity patterns in epidemiological studies results in exposure misclassification and bias associated health effects.
The benefit of incorporating non-residential locations and daily commute patterns in expo¬sure assignments of future epidemiological studies should carefully be evaluated based on (1) spatial and temporal variability of the pollutants of interest, and (2) the spatial spread of home and work/school locations and subjects’ level of mobility. Improved exposure estima¬tion thus requires information on subjects’ travel duration, distance, transportation modes, trip timings, route choices and work load.
Future exposure assessments of large cohorts will need to more frequently combine mod¬eling approaches with actual personal exposure measurements of pollutants of interest to refine and validate exposure estimates spatially as well as temporally.