YDA, RE, and JT were partially supported as consultants for Social Impact in a separate study that overlapped only at the start of the current study but Social Impact was not involved in any way in the present study. This does not alter our adherence to PLOS ONE policies on sharing data and materials. All other authors declare no potential competing interests.
Winter air pollution in Ulaanbaatar, Mongolia is among the worst in the world. The health impacts of policy decisions affecting air pollution exposures in Ulaanbaatar were modeled and evaluated under business as usual and two more-strict alternative emissions pathways through 2024. Previous studies have relied on either outdoor or indoor concentrations to assesses the health risks of air pollution, but the burden is really a function of total exposure. This study combined projections of indoor and outdoor concentrations of PM2.5 with population time-activity estimates to develop trajectories of total age-specific PM2.5 exposure for the Ulaanbaatar population. Indoor PM2.5 contributions from secondhand tobacco smoke (SHS) were estimated in order to fill out total exposures, and changes in population and background disease were modeled. The health impacts were derived using integrated exposure-response curves from the Global Burden of Disease Study.
Annual average population-weighted PM2.5 exposures at baseline (2014) were estimated at 59 μg/m3. These were dominated by exposures occurring indoors, influenced considerably by infiltrated outdoor pollution. Under current control policies, exposures increased slightly to 60 μg/m3 by 2024; under moderate emissions reductions and under a switch to clean technologies, exposures were reduced from baseline levels by 45% and 80%, respectively. The moderate improvement pathway decreased per capita annual disability-adjusted life year (DALY) and death burdens by approximately 40%. A switch to clean fuels decreased per capita annual DALY and death burdens by about 85% by 2024 with the relative SHS contribution increasing substantially.
This study demonstrates a way to combine estimated changes in total exposure, background disease and population levels, and exposure-response functions to project the health impacts of alternative policy pathways. The resulting burden analysis highlights the need for aggressive action, including the elimination of residential coal burning and the reduction of current smoking rates.
Mongolia’s capital city, Ulaanbaatar (UB), is home to nearly half of the nation’s three million residents. Driven initially by political-economic changes incurred after the fall of the Soviet Union and hastened by periodic bouts of harsh weather and famine [
The high air pollution levels in UB arise from a combination of high anthropogenic emissions, geography, and meteorology [
The Mongolian government has undertaken efforts in this decade to reduce air pollution, including the subsidization of cleaner-burning coal stoves, elimination of many institutional heat only boilers (HOB), and the promotion of energy efficiency [
Our estimates improve on previous local and regional population-wide burden assessments [
Personal PM2.5 exposures and related disease burdens were modeled for UB residents through 2024 under BAU and two alternative policy pathways. Indoor and outdoor concentration estimates were combined with time-activity data, census information, demographics projections, and estimated tobacco smoking rates to provide population-weighted total exposure estimates.
Annual average exposures were estimated for population sub-groups in 2014 and in 2024 under BAU and two alternative policy pathways from outdoor and indoor concentration models and time activity estimates using the approach summarized above. These exposures were applied to disease-specific exposure response curves to produce estimates of population attributable fraction (PAF) which were applied to background disease rates to quantify attributable disease burden. Detailed data descriptions and methods–including how interim year (2015–2023) disease burdens were calculated–are included in the manuscript and
Variations in the emissions trajectories of heating, power, and traffic sectors were considered in relation to policy pathways from baseline that followed business as usual or one of two alternative policy approaches: moderate restrictions in addition to those in place at baseline (Pathway 1) and additional aggressive restrictions (Pathway 2). The variations examined were most detailed for household heating, which has been identified as the single largest contributor to outdoor air pollution in UB [
Household Heating | Power Plants | Vehicles | |
---|---|---|---|
20,000 LPB-heated houses; semi-coke coal in gers & houses in Bayangol district; apartments heated with CIU heat including HOB units; MCA stoves in all other gers & houses | Four CHP: CHP-2, CHP-3 (two units), and CHP-4 | Nearly 100% growth over values from 2010—the most recent inventory at the time of analysis | |
All homes, except LPB and clean-heat homes, transitioned to MCA stoves | Addition of CHP-5, which meets U.S. New Source Performance Standards | 2.5%/year growth from 2014 and addition of Euro III standards | |
Transition of half of non-LPB houses to clean heat; replacement of remaining MCA stoves with "Future Tech" raw coal stove; 50% of HOB decommissioned, others retrofitted with controls | Addition of CHP-5; high-efficiency retrofits of CHP-2, CHP-3, and CHP-4. | 2.5%/year growth from 2014 and addition of Euro V standards | |
Transition of all ger and houses to clean heat; all HOB decommissioned | Addition of CHP-5; CHP-3,-4 retrofitted; CHP-2 replaced by renewables and/or imports | Increased mass-transit ridership; improved traffic flow; Euro VI standards; among others |
1.LPB = low pressure boiler, CIU = heating method with clean indoor use, HOB = heat only boiler, MCA = improved coal stove distributed by the Mongolian government and the U.S. Millennium Challenge Account, CHP = combined heat and power plant.
92.9% | 75.2% | 0% | |
7.1% | 6.0% | 0% | |
0% | 18.8% | 0% | |
0% | 0% | 100% | |
100% | 82.9% | 0% | |
0% | 17.1% | 0% | |
0% | 0% | 100% | |
100% | 41.4% | 0% | |
0% | 17.1% | 0% | |
0% | 41.4% | 100% | |
100% | 100% | 100% |
1.Improved coal stove distributed by the Mongolian government and United States Millennium Challenge Account (MCA).
Recent efforts by the Mongolian Government, the U.S. Millennium Challenge Corporation and Millennium Challenge Account (MCA) [
Business as Usual assumed no major changes from trends underway at the time of the study (mid-2013) by 2024. This included a transition to MCA stoves of all homes not employing clean heat or LPB at baseline. Because HOB and LPB are outdated technologies, no net increases in the number of HOBs or LPB-heated homes were assumed. BAU retained the power plant emissions in 2024 and included a new 820 MW power plant (CHP-5) to be located 15km east of the UB Central Business District, as supported by recent government plans to develop a 450 MW plant and expand it to 820 MW shortly thereafter [
Pathway 1, or moderate emissions reductions, assumed all changes in BAU as well as some moderate improvements. All 20,000 LPB-heated homes from baseline remained as such under Pathway 1. Remaining non-LPB houses were assumed equally split between clean heating and an even cleaner hypothetical coal stove, called the ‘Future Tech’ stove, which improved the emissions performance of the MCA stoves by the same percentage as the MCA stoves improved upon the traditional stoves and improved indoor concentrations by 20% compared to those in MCA stove homes. Half of all HOB units from baseline were assumed decommissioned by 2024 under Pathway 1, and the other half were assumed retrofitted with cyclone control technologies. All other households were assumed to rely on clean heat from other sources. In addition to the power plant assumptions under BAU, Pathway 1 assumed high efficiency control devices, such as electrostatic precipitators, installed on units CHP-2, CHP-3, and CHP-4. This is a significant upgrade to the existing CHP infrastructure, which includes wet scrubbers or electrostatic precipitators, depending on the facility. For vehicles, the BAU rate of growth was assumed for Pathway 1, but with the implementation of Euro V emissions standards.
Pathway 2, or transition to cleaner fuels and technologies, assumed feasible but ambitious rates of change in all sectors by 2024. Solid fuel combustion was assumed eliminated in households. CHP-3 and CHP-4 were assigned high efficiency control technologies. CHP-2 was decommissioned by 2024 and replaced with renewables and/or imports (i.e. sources with negligible impacts on UB air quality). A 50% reduction in traffic emissions over Pathway 1 was assumed, opportunities for which include but are not limited to higher adoption rates for mass transit use, transportation network enhancements to improve traffic flow, and adoption of Euro VI standards, which include an additional 50% reduction in PM emission rates from heavy duty diesel vehicles compared to Euro V standards.
Throughout BAU and the alternative pathways tobacco smoking prevalence among households (not individuals) was maintained at 45% of households. This figure was based on a series of surveys [
Demographic conditions were estimated for 2006 through 2024. The methods and sources used are described in detail in the supplemental text (
(Credit: L. Drew Hill).
The 2014 population of UB was estimated at 1,355,176 residents distributed among 86,246 gers, 106,353 houses, and 179,718 apartments. Projections indicated that by 2024 the population would grow nearly 40% and experience a substantial shift to apartment dwelling with over 65% of the population living in multi-family buildings (
Year | Population |
Number of Households |
Pop. per | |||
---|---|---|---|---|---|---|
- | Total Pop. | Pop. 0–4 Years | Ger | House | Apartment | Household |
2014 | 1,355,176 | 148,219 | 86,246 | 106,353 | 179,718 | 3.64 |
2015 | 1,407,196 | 155,551 | 88,547 | 109,191 | 197,539 | 3.56 |
2016 | 1,459,516 | 158,438 | 90,684 | 111,826 | 216,586 | 3.48 |
2017 | 1,511,836 | 161,325 | 92,616 | 114,209 | 236,854 | 3.41 |
2018 | 1,564,157 | 164,212 | 94,323 | 116,313 | 258,369 | 3.34 |
2019 | 1,616,477 | 167,099 | 95,781 | 118,112 | 281,152 | 3.27 |
2020 | 1,668,797 | 169,986 | 96,967 | 119,574 | 305,219 | 3.20 |
2021 | 1,715,748 | 168,427 | 96,997 | 119,611 | 330,782 | 3.13 |
2022 | 1,762,700 | 166,869 | 96,667 | 119,204 | 357,645 | 3.07 |
2023 | 1,809,651 | 165,310 | 95,954 | 118,324 | 385,792 | 3.02 |
2024 | 1,856,603 | 163,752 | 94,834 | 116,943 | 415,195 | 2.96 |
1.Interpolated from five-year “medium growth” (version 1b) projections identified in the 2010 Population and Housing Census of Mongolia Report [
2.Estimated using the techniques and sources described in
Air quality modeling was conducted to estimate outdoor PM2.5 mass concentrations. The modeling methodology followed that used by Social Impact for an impact evaluation of the Energy-Efficient Stove Subsidy Program of the Millennium Challenge Mongolia Energy and Environment Project (MCA impact evaluation), detailed in the full report [
Pathway | Vehicles | Power Plants | Heat Only Boilers | Household Stoves & LPB |
---|---|---|---|---|
2014 Baseline | 384 | 11,500 | 1,300 | 1,700 |
2024 BAU | 500 | 12,000 | 1,300 | 1,900 |
2024 Pathway 1 | 96 | 1,900 | 390 | 640 |
2024 Pathway 2 | 48 | 1,830 | 0 | 0 |
Residential heating stove emissions were assumed zero during the summer period, April through September. An MCA stove emissions profile was taken from values reported in the MCA impact evaluation as weighted by the sales-based prevalence of three variations of the MCA stove (Ulzii, Khas, and Dul) detailed in the same publication [
Baseline power plant emissions were taken from a recent JICA [
A simple scaling approach as previously described was used for motor vehicle emissions pathways, and did not account for changes in the fleet composition over time as insufficient details for the JICA 2010 inventory were available to make more sophisticated projections. All exhaust emissions were assumed to be in the PM2.5 size range. For diesel vehicles the Euro V PM emission standards are 80% -93% lower than the Euro III standards depending on vehicle class. There were no Euro standards for PM emissions from gasoline-fueled vehicles and thus 90% overall reduction would not be realized between BAU and Pathway 1. However, gasoline vehicle Total Hydrocarbon standards are 50% lower for Euro V compared to Euro III. This could result in some PM reductions for the cold wintertime conditions, which favor semi-volatile gaseous compounds entering the particle phase. Overall, Pathway 1 employed a 75% reduction in vehicle emissions in 2024 compared to those at baseline. Pathway 2 assumed a simple 50% reduction in traffic emissions over Pathway 1, as previously discussed.
Existing power plants were modeled as point sources using available geographic location and stack properties data [
Estimation of residential heating stove emissions and allocation of these emissions in space and time followed approaches developed for the Social Impact MCA impact evaluation [
Motor vehicle and HOB emissions were spatially allocated using emissions fields from 2010 with a resolution of 0.01°× 0.01° [
Year 2012 population by dwelling type (ger, house, and apartment) at the level of Khoroo, or Mongolian administrative sub-division similar to a sub-district, was also allocated to the 1 km × 1 km grids using area weighted sums. Projected changes in the peri-urban population between 2012 and the baseline and 2024 BAU and alternative pathway years were distributed across grid cells in proportion to the 2012 peri-urban population for both gers and houses. Projected changes in the population residing in apartments were allocated in proportion to the total population in each grid.
Air quality modeling was conducted at hourly resolution using meteorology data from April 2012 through March 2013 provided to us by the National Agency for Meteorology, Hydrology, and Environmental Monitoring of Mongolia–data available to other users upon written request to the Environmental Monitoring Department at what is now the National Agency for Meteorology and Environmental Monitoring of Mongolia. Un-modeled emission sources were assumed to have a spatially and temporally constant contribution of 10 μg/m3 across the city and over the ten-year assessment period. The model underestimated outdoor PM2.5 measurements conducted during the 2012–2013 winter heating season and these measurement data [
The vast majority of gers and houses in peri-urban areas heat with raw coal lit by small amounts of wood in small chimney stoves, while apartment households almost exclusively employ CIU heat that creates no indoor emissions. These differences combined with variations in outdoor particle infiltration between building types likely result in substantially different indoor concentrations between gers, houses, and apartments. Indoor concentrations of PM2.5 were thus estimated by home type, household heating source, presence of secondhand tobacco smoke (SHS), and season. Estimates were made for 2014 (baseline) and 2024 under BAU and the two alternative policy pathways.
Indoor air concentrations in homes with heating stoves were estimated by applying linear modeling techniques to data collected during the 2012–2013 winter season discussed in [
Indoor concentrations in homes that employ CIU heating sources like district heating, HOB, electric heat, or gas-based heat were estimated differently from those with heating stoves. Such concentrations were assumed to be governed primarily by SHS and by the penetration of outdoor PM2.5 into the indoor environment. Infiltration efficiencies were estimated at 64% in the summer and 53% in the winter for houses and apartments, and 100% in the summer and 70% in the winter for gers based on blower door tests and relevant literature detailed in
Smoking rates in Mongolia are among the highest in the world [
For simplicity and due to limited information on Mongolian workplace environments, the concentration profiles of the indoor environments in which the population spends their time away from home were assumed the same as those of their home indoor environments.
Time activity information was informed by a recent survey of UB households [
The UB population was divided into sub-groups based on the major exposure-related features of the indoor model and time activity estimates: home type, heating type, presence of SHS at home, and age. More specifically, sub-groups were made for children (< 5 years old), caretakers (≥ 5 years, assumed 1 per child), and all others (≥ 5 years) in smoking and non-smoking households representing each of the following home-heating combinations: gers with MCA stoves or semi-coke coal stoves, future tech stoves, or clean heat; houses with MCA stoves or semi-coke coal stoves or low pressure boilers, future tech stoves, or clean heat; and apartments with clean heat. For exposure estimation, children, caretakers, and non-children were distributed evenly to each household, and exposures were not distinguished by gender. Baganuur and Bagakhangai—the two districts for which ambient air quality estimates were handled outside of the outdoor models—were assigned the same distribution of population sub-groups as the overall population, with the exception that none of the LPB homes were included in these excluded districts as previously discussed. Household proportions in Baganuur and Bagakhangai were identified in the 2012 city census [
Average annual PM2.5 exposure concentrations for each population sub-group “i” were estimated at baseline and in 2024 under each pathway “j” (BAU, Pathway 1, or Pathway 2) by averaging seasonal exposure values (S; winter as April–September, summer as October–March) calculated from indoor (in) and outdoor (out) concentrations (C) at night (N; 18:00–8:00) and during the day (D; 8:00–18:00) as weighted by the fraction of time (t) during a typical 24-hour period spent in each environment during the specified time period (
Citywide population-weighted average exposures were calculated by aggregating the exposure concentrations of each sub-group from i = 1 to i =“n”, where n is the total number of sub-groups in each pathway-year j, as weighted by their representative fraction (λ) of the total population (
Burden attributable to PM2.5 exposures was calculated for lung cancer, ischemic heart disease (IHD), stroke, and chronic obstructive pulmonary disorder (COPD) in all UB residents as well as acute lower respiratory tract infection (ALRI) in children (ages 0–4 years) for 2014–2024 using a version of the Household Air Pollution Intervention Tool (HAPIT) [
Average annual exposures were used to calculate disease-specific relative risks (RR) of mortality due to PM2.5 exposure in each population sub-group. Mean, lower bound, and upper bound RR were taken from the integrated exposure-response functions produced by Burnett et al [
Disease-specific PAF estimates for 2015–2023 were linearly interpolated from baseline and 2024 PAF values under BAU and each alternative pathway. Finally, PAF values were applied to estimated background mortality estimates to produce disease-specific estimates of total attributable death in each year. Morbidity was calculated in the form of disability adjusted life years (DALYs). DALYs are widely used to take into account both the age distribution of premature mortality and the severity of non-fatal diseases. Disease-specific DALY estimates were calculated using the national disease-specific Death: DALY ratio produced during the 2010 GBD [
Whiskers represent 10th and 90th percentile concentrations.
Pathway | Summer | Winter | |||
---|---|---|---|---|---|
Total Pop. | Total Pop. | Ger Pop. | House Pop. | Apt. Pop. | |
16 | 141 | 140 | 148 | 137 | |
19 | 156 | 154 | 163 | 154 | |
11 | 55 | 55 | 58 | 55 | |
11 | 11 | 11 | 11 | 12 |
Average wintertime indoor concentrations for non-smoking homes heating with MCA stoves, LPB (houses only), and stoves using semi-coke coal were modeled at 107.0 μg/m3 for gers and 118.3 μg/m3 for houses. Wintertime indoor concentrations for homes with Future Tech heating stoves were assigned at 20% lower than homes with MCA stoves. Population-weighted indoor wintertime concentrations for non-smoking homes with clean heating at baseline and in 2024 are shown in
2014 | 2024 BAU | 2024 Pathway 1 | 2024 Pathway 2 | |
---|---|---|---|---|
98 | 108 | 39 | 8 | |
79 | 86 | 31 | 6 | |
73 | 82 | 29 | 6 |
2014 | 2024 BAU | 2024 Pathway 1 | 2024 Pathway 2 | |
---|---|---|---|---|
15 | 17 | 11 | 11 | |
10 | 11 | 7 | 7 | |
11 | 13 | 7 | 7 |
Indoor exposures are stratified by SHS and non-SHS environments. The difference between indoor and outdoor contribution to total exposure is primarily from the disproportionately high fraction of time spent indoors.
Annual | Winter | Summer | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2014 | 2024 |
2024 Path 1 | 2024 Path 2 | 2014 | 2024 BAU | 2024 Path 1 | 2024 Path 2 | 2014 | 2024 BAU | 2024 Path 1 | 2024 Path 2 | ||
66 | 68 | 52 | 14 | 113 | 114 | 87 | 13 | 20 | 22 | 16 | 15 | ||
70 | 71 | 44 | 12 | 125 | 126 | 76 | 12 | 15 | 17 | 13 | 12 | ||
50 | 56 | 25 | 12 | 83 | 93 | 36 | 12 | 17 | 19 | 13 | 13 | ||
We estimated exposure to PM2.5 in UB in 2014 to be responsible for 33% (lower: 23%, upper: 42%) of all ALRI deaths in children, 19% (lower: 9%, upper: 28%) of all COPD deaths, 27% (lower: 19%, upper: 42%) of all IHD deaths, 24% (lower: 8%, upper: 34%) of all lung cancer deaths, and 42% (lower: 14%, upper: 54%) of all stroke deaths, for a total of 1,400 attributable deaths (lower: 710, upper: 1,900) and 40,000 attributable DALYs (lower: 22,000, upper: 55,000) (
Note the higher importance for ALRI in the DALY distribution because it affects young children.
Burden of measure is annual DALYs, or DALYs/year. Baseline value (2014) is marked as a dashed line. Pathway 1 averts about 110,000 total DALYs from BAU policies. The stronger reduction measures of Pathway 2 avert about 240,000 total DALYs.
Pathway 2 would save more lives than Pathway 1 by more than a factor of 2. Lower and upper bounds on total values are shown as whiskers. Note the greater importance of child ALRI in averted DALYs as compared to averted deaths.
Accrued, 2014–2024 (per avg. 1000 Capita) | Lower, Upper Bounds | Incurred in 2014 (per 1000 Capita) | Lower, Upper Bounds | Incurred in Final Year of BAU or Pathway, 2024 (per 1000 Capita) | Lower, Upper Bounds | |
---|---|---|---|---|---|---|
18,000 (11) | 9,300–25,000 | 1,800 (0.99) | 980–2,600 | |||
14,000 (9.0) | 7,500–20,000 | 1,400 (1.0) | 710–1,900 | 1,200 (0.63) | 630–1,700 | |
9,800 (6.4) | 5,200–14,000 | 310 (0.16) | 180–450 | |||
530,000 (330) | 290,000–720,000 | 55,000 (30) | 31,000–77,000 | |||
420,000 (260) | 230,000–590,000 | 40,000 (30) | 22,000–55,000 | 34,000 (18) | 18,000–49,000 | |
290,000 (190) | 160,000–400,000 | 8,100 (4.4) | 4,400–12,000 |
The prolonged reduction period resulted in about 9,800 (lower: 5,200, upper: 14,000) unavoidable deaths and 290,000 (lower: 160,000, upper: 400,000) unavoidable DALYs under the most rigorous reduction pathway between 2014–2024 (
Relative projected urban population is also shown. Note that 2014 values are set at 100%. Pathway 2 would reduce impacts to near-counterfactual levels by 2024.
Modeled average annual exposures in Ulaanbaatar (estimated at 59 μg/m3 in 2014) remained high, despite a wide range of pollution reduction measures recently enacted by the Mongolian government, including ambient air quality standards [
We estimated that PM2.5 exposures in 2014 were responsible for 33% (lower: 23%, upper: 42%) of all deaths from ALRI in children and 19% (lower: 9%, upper: 28%), 27% (lower: 19%, upper: 42%), 24% (lower: 8%, upper: 34%), and 42% (lower: 14%, upper: 54%) of all deaths from COPD, IHD, lung cancer, and stroke, respectively. PM2.5 related mortality at baseline and as accrued under all pathways was driven by cardiovascular disease, while attributable morbidity was more evenly distributed between IHD, Stroke, and ALRI in children (< 5 years). These estimates and trends were consistent with global and national estimates from the GBD [
A business as usual approach to energy policies in UB will have little impact on citywide PM2.5 exposures by 2024 yet may result in a substantial increase in total health burden because of large increases in projected urban population. A package of policies targeting reductions in both indoor and outdoor emissions from household coal stoves alongside aggressive improvements in power and traffic sectors could reduce annual average population-weighted PM2.5 exposures by nearly 80% and annual per-capita attributable health burden by about 85% by 2024. A package of more moderate emissions control policies, including cleaner-burning coal stoves and modest improvements to the city’s power plants and vehicle fleets, may reduce PM2.5 exposures by 45% but would have less effect on health burden due in part to the non-linearity of the relationship between PM2.5 exposure and risk for many diseases [
Our investigation builds upon a small but growing body of air quality research in Mongolia [
Most inferences about the population health impacts of PM2.5 in UB and greater Mongolia have relied on outdoor concentrations modeled from emissions, chemical transport estimates, and/or measurements taken from a small number of outdoor, fixed-site monitors [
A 2011 study by the World Bank [
While few measurements exist with which to compare our values, a study [
There are several limitations to our study. Concerning the outdoor air quality modeling, Gaussian dispersion models are overly simplistic to capture all of the transport, dispersion, and terrain characteristics for UB wintertime conditions. Air quality modeling errors from the use of a Gaussian dispersion model are lumped together with emission inventory errors when calibrating the model to air quality observations. It is not clear how these errors propagate through to 2024. Un-modeled emission sources were assumed to have an impact of 10 μg/m3 and this simplification influences the exposure estimates, especially for Pathway 2 where modeled emission source contributions are low.
Indoor concentrations in ‘clean’ heating homes, which were calculated by applying infiltration factors to outdoor concentrations, may have propagated any error incurred by the outdoor model. These methods, which employed climate-relevant but non-local infiltration rates, could further be improved by future work characterizing local building infiltration rates. In addition, the linear model used to estimate wintertime indoor concentrations in stove heated homes, as described in
The use of linear models to project background disease rates through 2024 may not reflect future trends in areas like rural to urban migration, economic development, regulatory shifts, and healthcare improvements which may have non-linear impacts on disease-influencing factors (e.g. introduction of pneumococcal conjugate vaccines). This general limitation is highlighted by the weak fit of the linear background disease model to historical data for several diseases, as demonstrated in
Our BAU and alternative policy pathways were limited by gaps in the literature, too. For example, while recent government anti-smoking campaigns [
Attributable burden estimates in UB may, in general, be underestimated. Evidence suggests that cold-air exposures may increase sensitivity to risk factors for cardiovascular diseases [
Air pollution in Ulaanbaatar has reached a critical level, and immediate measures must be taken to reduce its health impacts on the city’s growing population. Current exposures are projected to produce unprecedented levels of respiratory illness, especially in children, and cardiovascular disease. Using some of the latest available exposure-response techniques and novel data on local emissions and indoor concentrations, this analysis is the most holistic view of population-wide air quality exposures in Ulaanbaatar to date. The results highlight the need for aggressive actions, including the elimination of residential coal burning and the reduction of current smoking rates, if the health burden of air pollution is to be reduced. Our conclusions support recent findings that PM2.5 emissions, especially from household heating, contribute substantially to mortality and morbidity from cardiovascular and pulmonary disease in the city. In addition, without efforts to moderate indoor concentrations, the full benefits of pollution reductions in UB will not be realized.
S1 Text delivers more detail on the methods summarized in the primary publication. The details of the demographic, indoor air quality, and outdoor air quality models included in this analysis are beyond the scope and narrative of the primary publication, but are integral to the analysis and produce ancillary results novel in their own right. This supplement is intended to fill in gaps in the primary publication in order to provide a fully transparent account of the methods and sources used.
(DOCX)
Identified by the National Statistics Office of Mongolia for 2000–2010, and estimated using extrapolation and assumptions of the Total Fertility Rate for 2011–2030.
(TIF)
(TIF)
Projection models, model fits (α = 0.05), and model 95% Confidence Intervals are shown.
(EPS)
(TIF)
10 μg/m3 was subtracted from each of the observed concentration values to adjust for sources not included in the modeling.
(TIF)
S1 Table provides projection estimates for population (total and child) and household number (by household type) for 2013 through 2030 as applied in the manuscript.
(XLSX)
S2 Table provides aggregated mortality data according to disease-specific ICD codes (see
(XLSX)
S3 Table provides mean, lower, and upper estimates of disease-specific population attributable (Risk) fraction, background disease burden, and attributable risk (deaths and Disability-Adjusted Life Years [DALYs]) by year and policy pathway (BAU, Pathway 1, Pathway 2) as analyzed in the main text and
(CSV)
S4 Table provides exposure and disease-specific risk data for each population subgroup as analyzed in the main text and
(CSV)
S5 Table provides PM2.5 mass concentration data for the period January 22—March 2, 2013 from the MCA-funded ambient fine particulate matter speciation study conducted by Ecography/Ecoworld.
(XLSX)
S6 Table provides modeled PM2.5 mass concentration data for modeled districts of Ulaanbaatar, Mongolia by grid cell.
(XLSX)
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The authors are grateful to the National Statistics Office of Mongolia and the Statistics Department of Ulaanbaatar, Health Development Center of the Ministry of Health and Sports which provided access to various demographics and health databases. We are grateful to Social Impact and the Millennium Challenge Corporation for their open access databases of household measurements conducted as part of the impact evaluation of the Energy and Environment Projects. We thank Maria Hernandez, Ajay Pillarisetti, Paul Chung, and Alan Hubbard of the University of California, Berkeley and Nick Lam from the University of Illinois for their advice and generous assistance during the project. We acknowledge Boldkhuu Nanzad of the Ministry of Energy of Mongolia for advocating for our research, and appreciate the Berkeley Air Monitoring Group for facilitating financial arrangements. We also acknowledge that the final analysis benefits from comments made by many participants at a workshop presenting preliminary results conducted as part of the impact evaluation of the Energy and Environment Projects. During the peer-review process, a modified version of this paper was included as a chapter in a doctoral dissertation at the University of California, Berkeley. The full dissertation, which includes five total chapters and is titled “A breath of fresher air: improving methods for PM2.5 exposure assessment from Mongolia to California” by Lawson Andrew Hill, will eventually be available at ProQuest/UMI or the University of California’s online repository, “eScholarship.”