Kathmandu Valley, Nepal, has severe air pollution, although few studies examine air pollution and health in this region. To the best of our knowledge, no previous studies in Nepal used time-activity diaries or conducted personal monitoring of individuals’ exposures. We investigated personal exposure of particulate matter (PM) with aerodynamic diameter ≤2.5 μm (PM2.5) by location, occupation, and proximity to roadways. PM2.5 monitoring, time-activity diary, respiratory health questionnaire, and spirometer testing were performed from 28 June 2009 to 7 August 2009 for 36 subjects, including traffic police (TP), indoor officer workers next to main road (IOWs_NMR) and away from main road (IOWs_AMR), in urban area (UA), urban residential area, and semi-UA (SUA). TP had the highest exposure of all the occupations (average 51.2 μg/m3, hourly maximum >500 μg/m3). TP levels were higher at the UA than other locations. IOW_NMR levels (averaged 46.9 μg/m3) were higher than those of IOW_AMR (26.2 μg/m3). Exposure was generally higher during morning rush hours (0800–1100 hours) than evening rush hours (1500–1800 hours) for all occupations and areas (78% of days for TP and 84% for urban IOW). PM2.5 personal exposures for each occupation at each location exceeded the World Health Organization ambient PM2.5 guideline (25 μg/m3). Findings suggest potential substantial health impacts of air pollution on this region, especially for TP.
Airborne particulate matter (PM) has been demonstrated to have widespread health impacts.1, 2 The World Health Organization (WHO) estimates that PM with aerodynamic diameter ≤2.5 μm (PM2.5) results in 0.8 million premature deaths and 6.4 million years of life lost worldwide annually, with Asia accounting for ∼65% of the global burden.3 However, evidence linking PM and other air pollutants to health is based mainly on studies in North America and Western Europe, with some studies in Latin America and a growing number of studies in Asia. In comparison, such studies are relatively scarce in developing countries, such as Nepal,4 hindering quantification of the impacts of air pollution.3 Differences in pollutant sources, physical and chemical characteristics of particles, technology relating to motor vehicles, and geographical and meteorological conditions in Asia may result in different health impacts from PM compared with those in the United States and Europe.4, 5, 6 Social factors, including socioeconomic status, residential location, and housing structures, can have a large role in a population's susceptibility to air-pollution exposure. For example, individuals with the same education or income may experience different health status in relation to their neighborhoods, which largely determine physical environment (e.g. distance to pollution source) and access to facilities (e.g. health clinics).7 Population's health responses to pollution may also vary because of differences in baseline health, nutritional status, health care systems, and other factors. These population differences further challenge extrapolation of results from industrialized countries to other areas.8, 9 Further research in areas such as Nepal is essential to understand the health impacts of air pollution where environment, baseline health status, and social factors may be different than from the developed world.
The first documentation of air-pollution problems in the valley came with pollution monitoring in the 1990s by several organizations, including the Kathmandu Valley Vehicular Emission Control Project, funded by the United Nations Development Programme and the Metropolitan Environment Improvement Project funded by the World Bank.10 PM was identified as the main pollutant of concern and vehicles as the main pollution source contributing to 20% of total PM10 in the emission inventories performed in 1993 and 60% in 2001.11 Today, high pollution levels result from rapid and unplanned urbanization, growing population, poor road infrastructure, traffic congestion, and fuel adulteration. Further, Nepal currently has no ambient air-pollution monitoring network, hindering enforcement of regulations.
To the best of our knowledge, no previous studies in Nepal have used time-activity diaries or personal monitoring to measure an individual's exposure. Most studies of PM in Nepal have consisted of analysis of ambient monitoring data.12, 13, 14, 15 Some studies have used personal monitor instrumentation for stationary sampling at fixed sites rather than individuals, and focused on exposures from biomass burning16 and brick kilns.17 A previous study used personal monitors to measure PM10 at three high-density traffic locations in Pokhara over 7 days, although measurements were of the location, not individuals, and the monitor was passed between traffic police (TP) at the same location as shifts changed.18 This study showed the range of exposure for the location of a TP at the street crossing for a day (0730–1930 hours) without accounting for shift changes.
Studies from other regions indicate high pollution exposures and associated health effects for TP, who spend much of their workday at the center of highly polluted areas.19, 20, 21, 22, 23 We conducted a personal monitoring study of PM2.5 exposure (June 2009–Aug 2009) in Kathmandu Valley, Nepal, with a focus on TP, who we hypothesize are subject to high exposure levels, and indoor office workers (IOWs) as a comparison group. Comparison of respiratory health indicators was also performed between TP and IOWs. Ambient pollution data in this area are limited and not available for the same time period as our personal monitoring study. We present the available ambient PM data (PM10 data for 2003–2007) for context as a secondary aim to inform on the overall levels of PM in this region.
Research Location and Sampling Period
Kathmandu Valley is the main urban region in Nepal with deep bowl-shaped topography, a relatively flat area of ∼340 km2 and surrounding hills from 500 m to 1400 m. The valley experiences subtropical, temperate climate with four distinct seasons (pre-monsoon, monsoon, post-monsoon, and winter) with highest ambient temperature of 35 °C in summer and lowest temperature of −1 °C in winter. Relative humidity ranges from 50% in dry seasons to >80% in rainy seasons. Average annual rainfall is 1400 mm and wind speed is 0.5–7.5 m/s, with mostly southwesterly and northwesterly winds.14
The Ministry of Population and Environment of Nepal (now under Ministry of Environment, Science and Technology) established six ambient air-quality monitoring locations within 15 km of each other in the valley: urban area (UA), urban hospital area (UHA), urban residential area (URA), two semi-UAs (SUA1 and SUA2), and rural area (RA) (Supplementary Figure 1). Monitoring was conducted for daily PM10 from October 2002 to December 2007.12 UA, UHA, and URA are located in the inner city core. UA is mainly a commercial area with high traffic flow and narrow roads. UHA contains Patan Hospital and has high traffic flow. URA is characterized by a commercial area with tourist attractions and medium traffic flow. SUA1, SUA2, and RA are located in the outer city core. SUA1 and SUA2 have comparatively low traffic flow. The land near the semi-UAs consists mainly of open agriculture fields,12 with brick kilns in SUA1. RA has agricultural fields and negligible traffic.
Personal PM2.5 exposure analysis was conducted from 28 June 2009 to 7 August 2009 (6 weeks except on saturdays) at UA, URA, and SUA2. During the study period, 22 July and 5 and 6 August were public holidays when measurements were not performed.
Participants for the study were recruited from two occupations (TP and IOWs). Each study subject works at one of the three sites (UA, URA, and SUA2). Recruitment was first performed for TP by distributing preliminary questionnaires regarding age, body mass index (BMI), income, and smoking status. For each location, TP were selected of 20–35 years old who had worked as a TP for 4–10 years. Recruitment of TP continued until participants were identified for all locations that met the age and employment criteria. Next IOWs who were 20–35 years old and had worked at the same location for 4–10 years were recruited using the same preliminary questionnaire. Recruitment of IOWs continued until participants were identified for all locations that met the age and employment criteria. As TP in Nepal are mostly male, participants were limited to males. Each participant provided written consent.
A total of 36 study subjects participated in the study: 6 TP and 6 IOWs at each of the three sites. Recruited TP were constable or head constable, for which more hours are spent on the road compared with other positions in the traffic division. IOWs were defined as persons who worked indoors throughout the day in the first or second floor of the building and did not use office machines that potentially released high pollution levels (e.g. copier machine). The roadway system in the study area is such that for each of the three locations there is a single large road with the majority of traffic, which we refer to as the main road. Other roads in the study locations are substantially smaller with very few vehicles. IOWs next to main road (IOWs_NMR) was defined as IOWs working in a building adjacent to the main road and IOWs away from main road (IOWs_AMR) as IOWs worked in a building that was on average three buildings away from the main road.
Individual Exposure Assessment
For any given week, individual exposure assessment was performed at each site for two participants from each of the two groups: TP and IOW. Hence, for any given week, a total of 6 participants were assessed (3 locations × 2 participants at each location=6 subjects). During the 6-week study period, IOWs_NMR were considered for the first 3 weeks and IOWs_AMR for the following 3 weeks. Personal PM2.5 monitoring (1 min interval) was conducted during working hours for each study subject for one workweek (Sunday–Friday in Nepal). For TP working hours were 0800–1800 hours (10 h/day): duty at road from 0800–1100 hours and 1500–1800 hours (coinciding with rush hours) and rest at office/barracks from 1100–1500 hours. Working hours for IOW were 0900–1700 hours (8 h/day).
Personal PM2.5 monitoring was conducted with the Personal Data Ram (PDR) model 1200 (Monitoring Instruments for the Environment, Bedford, MA, USA) attached to GILAIR-3 (Sensidyne Industrial Health and Safety Instrumentation, Clearwater, FL, USA), a constant flow-air sampling pump. These monitors consist of particle-size-selective inlet cyclones that allow size-segregated measurements and use photometric monitor with light-scattering sensing configuration to provide real-time measurement at 1-min intervals. Standard calibration was performed as recommended by the manufacturer. Zeroing of the instrument (testing with particle-free air provided by the manufacturer) to ensure maximum accuracy of concentration measurement and span checking (internal calibration using built-in optical scattering/diffusing element) were performed before every measurement cycle. All measurements for personal PM2.5 exposure were based on the same instrumentation. Previous work found PDR measurements to provide good agreement with other methods (Beta Attenuation Monitor) at low relative humidities and follow predictable trends at higher humidity to allow correction scheme for improved accuracy.24
PDR and pump were placed in a small backpack with a tube connected to the cyclone inlet to sample at the breathing zone. TP carried the backpack with the inlet tube attached to the collar (Figure 1) during the duty hours (0800–1100 hours and 1500–1800 hours) and placed the backpack next to them with the inlet tube at 1–1.5 m above ground (breathing level) during rest hours (1100–1500 hours). IOWs placed the backpack next to their desk (Figure 1) with the inlet tube at breathing level during working hours (0900–1700 hours). The IOW position involves very little movement away from the desk, thus none of the participants carried the backpack if they moved from their working desk.
Time-activity diaries were maintained in real time. A research assistant maintained diaries for TP. For IOWs, participants maintained the diaries, which were collected at the end of the day and checked for completeness. The diaries recorded location and activity whenever change of activity occurred. For TP, activities were categorized as: morning duty, evening duty, moving (walk to/from rest area), and rest (at office/barracks). For IOWs, activities were categorized as: work, meeting, out (other rooms in the office, to the restroom, and out for lunch), cleaning, and incense burning. During peak rush hours (0800–1100 hours and 1500–1800 hours), vehicle count (number of vehicles passing a specific point) was assessed by a research assistant three times per hour for 1-min intervals each for each study location (UA, URA, and SUA2).
Individual Health Assessment
On the last day of exposure monitoring of each week, participants completed a respiratory health questionnaire (modified American Thoracic Society 1978 Adult Questionnaire) and performed a spirometer test of forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1). FVC is the total amount of air that can be forcibly exhaled after full inhalation, and FEV1 is the amount of air that can be forcibly exhaled in 1 s. The portable EasyOne spirometer (New Diagnostic Designs, Medical Technologies, Andover, MA, USA) provided grades from A (best) to E (worst). Participants were asked to stand while performing the lung test and provide at least three satisfactory blows with grades of A or B.
For each subject, we calculated the percentage of predicted normal values of FVC and FEV1, which are dependent on the person's sex, age, and height. Presence of abnormalities is indicated if FEV1/FVC <0.7 and/or percentage of predicted normal FVC and FEV1 <80%. We assessed presence of an obstructive disorder if percentage of predicted normal value of FEV1 <80%, FVC is normal or reduced, and FEV1/FVC <0.7. A restrictive disorder is indicated if percentage predicted normal values of FEV1 and FVC <80% and FEV1/FVC >0.7.25
Personal PM2.5 exposures were compared by location, participant type, activity, time of the day, and smoking status using a non-parametric Wilcoxon rank-sum test. In addition, this test was used to compare exposure with the WHO guidelines for ambient PM2.5. Nepal does not regulate PM2.5.
Ambient PM10 Monitoring
The primary objective of this paper is to estimate exposure for TP and IOWs as described above. Ambient measures of PM2.5 were not conducted, and ambient PM10 measurements ceased in January 2008. Therefore, no ambient data were available during the time frame of this study, and we cannot compare personal monitoring results with ambient data. To provide overall context, we present some information on the ambient PM data that are available, which are for PM10 from 2003 to 2007 (Supplementary Information 1. Ambient PM10 data collection).
Personal PM2.5 Exposure
Table 1 summarizes characteristics of study participants. TP and IOWs were similar in age and years of employment. IOWs had higher BMI and income. TP were more likely to be smokers.
Table 2 shows personal PM2.5 exposure during working hours by occupation type and location. On average for all three locations, TP had the highest hourly PM2.5 exposure (51.2 μg/m3) followed by IOWs_NMR (46.9 μg/m3) and IOWs_AMR (26.2 μg/m3). Each of the three occupation groups was different from each other (P<0.05). Comparison across locations shows that the highest average personal PM2.5 exposure was at UA for TP (P<0.05). For IOWs_NMR, UA and URA exposure exceeded than that of SUA2 (P<0.05). However, for IOW_AMR, the highest levels were at URA, then SUA2, and lowest at UA but not significantly so.
Comparing across occupations within a given location, TP exposures were higher than IOWs_NMR, which exceeded IOWs_AMR, at UA and SUA2 (P<0.05). At URA, TP and IOWs_NMR had levels above that of IOWs_AMR (P<0.05). At all locations, IOWs_AMR had the lowest PM2.5 levels. Supplementary Table 1 shows personal PM2.5 exposures by occupation, location, and smoking status. Exposure was higher for smokers than nonsmokers for TP and IOWs (P<0.05).
Table 3 shows personal PM2.5 exposure by occupation type, location, and time of the day. Supplementary Figure 2 shows exposure by time of the day for each occupation and location. Exposure for TP was higher during traffic rush hours (0800–1100 hours and 1500-1800 hours) than non-rush hours (1100–1500 hours) for all locations (P<0.05). For IOWs_NMR and IOWs_AMR, rush hour levels were generally higher than non-rush hour levels, but not significantly so. The increase in exposure comparing rush hour with non-rush hours was 72.3%, 28.2%, and 67.8% for TP and 7.2%, 24.8% and −19.2% for IOW at UA, URA, and SUA2, respectively. In other words, for IOW, exposures at UA were 7.2% higher in rush hours than during non-rush hours, whereas at SAU2, exposures were 19.2% lower during rush hours than non-rush hours. In general, personal PM2.5 exposures for TP and IOW were higher during morning traffic rush hours than evening traffic rush hours, although not for every day. Rush hour exposure was higher in the morning than the evening for 78%, 84%, and 57% of days for TP and 84%, 89%, and 81% of days for IOW at UA, URA, and SUA2, respectively. Comparing by occupation and location, morning rush hour levels were higher than evening rush hour levels for IOWs_AMR and IOWs_NMR for all locations and for TP at UA and URA (P<0.05).
Although personal exposures were generally higher during morning rush hours, traffic count data revealed higher traffic flow in the evening rush hours for UA and URA, and slightly higher morning rush hour traffic flow for SUA2 (Supplementary Table 2). A much higher traffic volume was observed at UA with an average 55.7 vehicles/min during rush hours than at URA (average 16.3 vehicles/min) or SAU2 (average 12.7 vehicles/min). We did not find a relationship between personal PM2.5 exposures and traffic count (Supplementary Figures 3–5).
Although our measurements are for personal exposure, not ambient levels, we compare with ambient health-based regulations and guidelines for context. Most personal PM2.5 exposures during working hours exceeded the daily US EPA standard (35 μg/m3) and WHO guideline (25 μg/m3) for ambient 24-h PM2.5, especially for TP at UA (Figure 2). The WHO guideline was exceeded on 87.7% of days for TP, 83.0% of days for IOWs_NMR, and 38.9% of days for IOW_AMR on average across participants. PM2.5 personal exposures for each occupation at each location were higher than the WHO guideline (P<0.05). The guideline was exceeded 100%, 81.3%, and 81.8% percent of study days for TP, 93.8%, 86.7%, and 68.6% for IOWs_NMR, and 26.7%, 50.0% and 50.0% for IOWs_AMR at UA, URA, and SUA2, respectively.
Based on activity diaries, for TP, on average 57% of working time was spent on duty (27.1% during morning and 30.1% during evening), 4.3% moving, and 38.5% at rest with highest exposure at each location during morning duty hours. For IOW, on average 83.9% of time was working at the desk, 1.45% in meetings, 13.8% out (e.g., to other rooms, out for lunch, and restroom), 0.049% cleaning, and 0.57% incense burning (Supplementary Table 3).
Although this pilot study cannot directly link air pollution to health outcomes, we were able to collect respiratory health data and compare by occupation type. The frequency of respiratory symptoms and spirometer readings for TP and IOW are provided in Table 4. In all, 2 out of 18 TP had restrictive lung function. Similarly, 2 out of the 18 IOWs had restrictive lung function, and one IOW had moderately obstructive lung function.
Ambient PM10 Monitoring
As a secondary aim, we examined ambient PM10 concentrations in the region for the available data (2003–2007). Daily PM10 levels were similar by day of the week for all six monitoring stations with a slight increase in Thursday and Friday and slight decrease on Saturday for inner-city core locations (Supplementary Figure 6). On average, PM10 levels decreased from workweek to weekend days by 8.2% for UA and UHA, 2.3% for URA, and 1.0% for outer-city core sites.
Average concentrations at inner-city core sites exceeded Nepal's daily PM10 standard (120 μg/m3), and all sites exceeded the overall standard (120 μg/m3 not to be exceeded>4.9% of days/calendar year and not on two consecutive days) (Supplementary Figure 7). The average percent of days exceeding the standard ranged from 3.4% for RA to 80.5% for UA.
PM10 levels exhibited seasonal differences, with the lowest values for all sites in August during the monsoon season (June–September) (range 17.9–116.8 μg/m3), compared with other times of year. Highest levels occurred in January during winter (December–February) (range 115.8–312.0 μg/m3) for all sites, except for RA with highest level in April during pre-monsoon (March–May). The difference between ambient PM10 levels in winter season and in monsoon season was higher for outer-city core sites (80.4%), especially for SUA1, and lower for inner-city core sites (64.5%).
We investigated PM exposure in Kathmandu Valley, an under-studied region, and conducted the first use of time-activity diaries and personal monitoring to measure exposure of individuals in Nepal. Personal PM2.5 exposure varied by occupation, location, activity, and time of the day. Exposure for TP was much higher than that for IOW, especially for the urban location. IOW exposures are likely lowered by their indoor working location, especially for those working on a building's second or higher floor. Studies have shown that PM2.5 exposure decreases with height, which could lead to lower exposure for IOW. At Macao, China, PM2.5 concentrations decreased with height (2–79 m) showing a significant vertical profile.26
Higher personal exposure was observed during traffic rush hours (0800–1100 hours and 1500–1800 hours) than other hours (1100–1500 hours) for TP in all locations. In addition to lowered traffic during non-rush hours, during the afternoons, strong westerly winds enter the valley helping to disperse pollutants.15 In general, most studies observe higher air-pollution levels during evening rush hours than morning because of the accumulation effect,27, 28 but a reverse case is observed for the valley, with significantly higher personal exposures during morning rush hours (0800–1100 hours) than evening rush hours (1500–1800 hours) for IOWs_NMR and IOWs_AMR for all locations and for TP at UA and URA. Lowered ambient temperature and wind speed in the morning hours may decrease the rate of ambient ventilation and increase the atmospheric stability in the morning,12, 13 contributing to higher morning PM2.5. Additionally in the evening, polluted air rises as colder, cleaner air flows underneath, whereas in the morning, the elevated pollutants descend towards the breathing zone.15 Our result is consistent with a previous finding of higher morning-peak ambient PM2.5 than evening-peak concentration in the valley during December 2006.13
Analysis of vehicle count did not yield a strong association with PM2.5 exposures, indicating that other factors have a role, such as meteorology, road infrastructure (e.g. roadway material) activities near roadways (pedestrian movement, road-side vendors), vehicle idling, vehicle type and speed, and construction. For TP, fairly similar PM2.5 levels were observed between SUA2 and URA despite URA having higher vehicle count, which may be due to presence of a public transportation stand with vehicles idling, poor road conditions and a high number of pedestrians at SUA2 in comparison with URA. A weak and insignificant association between road traffic and PM was found in Basel, Switzerland, suggesting that factors other than traffic influence PM levels.29
Although personal PM2.5 exposures differed by location, the order differed by participant category. At all locations, personal PM2.5 exposure for IOWs_NMR was higher than that for IOWs_AMR, suggesting that exposure is related to proximity to main roads. PM2.5 concentration at ground level showed no significant decrease with increasing distance from 0 to 228 m at Macao, China,26 unlike other traffic pollutants (e.g. NO2 and black smoke) where rapidly decreasing concentration was observed with increasing distance from roads.30 In Brisbane, Australian concentrations of small particles (0.016–0.626 mm) around a building about 15 m from an arterial road were as high as levels closer to the road.31
Differences between air pollution in Nepal and other regions are demonstrated by the similar ambient PM10 levels by day of the week in Nepal, indicating little change in human activities from weekends to weekdays in the valley. This contrasts with studies in other regions identifying higher daily PM10 levels on weekdays than on weekends. In Basel, Switzerland, PM10 was 17% higher on weekdays than weekends, suggesting increased human activities during weekdays such as traffic, construction, and industrial processes.29 Similarly, significantly higher daily PM10 concentrations during weekdays than weekends were observed in Kolkata, India,32 Taiwan,33 Milan,34 and London.35
The high PM2.5 exposure observed in this study suggests substantial potential health impacts for this population, especially for TP, although health impacts of PM2.5 were not directly studied in this research. Other studies found evidence of respiratory-related health problems in TP. In Jalgoan City, India (n=60), TP had higher frequency of coughing (40%) and shortness of breath (10%) than other healthy young adults.19 Similar studies in India showed high rate of respiratory, eye irritation, and skin problems among TP with significant number having lung disorders.19 In Bangkok, Thailand, TP (n=530) had higher prevalence of frequent coughing (18.3%), frequent phlegm (30.95%), wheezing (27.7%), and breathlessness (4.3%) than their wives.20 In Thonburi district, Bangkok, TP (n=242) showed more coughing (18.6%) and significantly lower FEV1 and FVC than the general Thai population (n=129).23 In Jaipur, India, a significant decline in percentage of the predicted normal values of FEV1 for TP (n=300) was observed compared with a healthy control group (n=164).21 For TP (n=1603) in Hong Kong, prevalence of nonspecific respiratory disease in polluted, moderately polluted, and suburban areas was 13.0%, 10.9%, and 9.4%, respectively, indicating an association between prevalence of respiratory symptoms and urban traffic air pollution.22
Although the sample size is limited, results on respiratory symptoms and spirometry do not show strong differences across occupations. In fact, some spirometry readings were slightly higher for TP than IOW. Future epidemiological studies of health effects from air pollution in these populations should consider the potential of a healthy worker effect.
Our study suggests that air pollution in the valley likely has substantial health impacts, with high personal PM2.5 exposures (summer 2009) and ambient PM10 levels (2003–2007) far above guidelines and regulations, and exceeding levels observed in other studies that directly linked PM and health endpoints.36, 37, 38 For TP, hourly personal PM2.5 levels reached >500 μg/m3 and averaged 51.2 μg/m3. Exposures for IOW were also high, averaging 36.9 μg/m3, well above the WHO guideline and EPA standard. Although personal exposures were measured for the working hours and WHO and EPA values are for 24-h periods of ambient levels, results suggest the potential for sizeable health risks from PM in this area.
Limitations of this study include the small sample size. IOWs maintained their time-activity diaries, whereas for TP, research assistants maintained the diaries, which may be a source of bias. Research in this region is hindered by a lack of data. For example, all the six government monitoring sites of ambient pollution used in this study are no longer in operation. The health impacts of air pollution on Nepal may differ from cities in developed countries because of differences in exposures patterns (e.g. day of the week exposure pattern), underlying population characteristics (e.g. shorter life span), and pollutant sources. As another example, additional data are needed to investigate whether variation among individuals’ exposures relates more to location, occupation, activity patterns, distance from roadways, etc. Thus, future research requiring a larger dataset to evaluate the link between air pollution and health in Nepal is needed.
We thank Sunil Kumar Joshi, Brian Leaderer, Judith A. Sparer, and Bill Galdenzi. This work was supported by the Jubitz Family Endowment for Research Internship, Yale School of Forestry and Environmental Studies Summer 2009 Globalization Internship Fund, and the Carpenter/Sperry Internship and Research Fund.
About this article
Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website (http://www.nature.com/jes)