Geographic information system-based mapping of air pollution & emergency room visits of patients for acute respiratory symptoms in Delhi, India (March 2018-February 2019)

The national capital of India, Delhi, is amongst one of the most-polluted cities of the world1,2 and is facing major environmental challenge3. In Delhi, per 10 unit increase in the concentration of sulphur dioxide and particulate matter (PM)₁₀ increases respiratory disease-related hospital visits by 83 and 0.21 per cent, respectively, at a previous lag of 0-6 days4,5. Ambient air pollutant levels of PM_2.5 PM₁₀, sulphur dioxide (SO₂), nitrogen dioxide (NO₂), ozone (O₃) and carbon monoxide (CO) in Delhi have been reported to exceed Indian National Ambient Air Quality Standards6. Fine particles having diameter of ≤2.5 μm (PM_2.5) are more hazardous to human health than other pollutants and are used as a common measure for air pollution1,7. Smaller size PM is associated with higher fraction of redox-active components and is, therefore, highly toxic8. In 2017, Delhi reported having the highest annual population-weighted mean PM_2.5 concentration (209 μg/m³)9 much above the Indian (40 μg/m³)10 and WHO-recommended limit (10 μg/m³)2. The alarming levels of PM_2.5 are regional problem and are significantly contributed by vehicular (20%) and industrial emissions (11%), cooking related emissions, biomass burning, construction activities, burning of Kharif (local term for monsoon or autumn crops) crop residue, windblown dust, Diwali fireworks, etc.1,3,10-13, and meteorological factors (temperature, relative humidity and wind velocity, etc.)14. The major chemical components of PM_2.5 include secondary inorganic aerosol (16-28%), organic matter (13-20%), elemental carbon (4.6-6.3%), chloride (4.5-7.9%) and metals (14-24%). The levels of PM_2.5 are ~20-30 per cent higher in winter than in the summer months due to high total secondary aerosols and combustion-related total carbonaceous matter (elemental carbon + organic matter). However, the crustal matter is observed to be higher in summer (42%) than in winter (9%)1,13. Various time-series studies suggest that short-term exposure to PM_2.5 can induce acute respiratory symptoms such as cough and difficulty in breathing and aggravation of pre-existing condition15-19 while long-term exposure to PM_2.5 can result in the development of cardiovascular and respiratory diseases7,15. Several epidemiological studies have documented that ambient PM levels in Delhi are associated with increased respiratory morbidity and mortality5,16,20-24. Air quality index (AQI) is a single index value used to provide information on daily air quality status and its associated health effects to public15,25,26. PM_2.51,14 and AQI27 can vary depending on seasons, time of the day and locations in the same city26. Various studies from India11,14,28,29 and other parts of the world30-32 have explored geographic information system (GIS)-based tools, e.g., inverse distance weighting (IDW), kriging, etc., to estimate the spatio-temporal distribution of air pollutants and its associated heath impact. GIS is a powerful technique that can be used to accurately analyze the spatial and temporal patterns of respiratory morbidity32, chemistry of pollutants and localization of the area of potential threat33. Therefore, integration of GIS can aid in understanding environmental health modelling, infrastructure planning, transport monitoring, public transit planning, etc. A study from Delhi revealed that PM_2.5 concentrations had very high temporal and spatial variations14. However, there is no study available in Delhi that has used GIS technique to assess the spatial and temporal association of ambient air pollution with daily counts of emergency room (ER) visits of patients related to acute respiratory symptoms. Therefore, this study was aimed to determine the spatio-temporal relationship between variation of PM_2.5 concentrations and AQI with ER visits of patients having acute respiratory symptoms in Delhi, India, using GIS techniques.

Material & Methods

Study area: Delhi is geographically located in north India between 76° 50′ 24″ and 77° 20′ 37″ East longitude and 28° 24′ 17″ and 28° 53′ 00″ North latitude1,34. The study area included 22 Continuous Ambient Air Quality Monitoring Stations (CAAQMS) operated by the Delhi Pollution Control Committee (DPCC), New Delhi and four study hospitals of Delhi viz. All India Institute of Medical Sciences (AIIMS), Kalawati Saran Children Hospital & Lady Harding Medical College, National Institute of Tuberculosis & Respiratory Diseases and Vallabhbhai Patel Chest Institute, New Delhi. The study was carried out for 12 months (March 2018-February 2019) to obtain the data for all seasons.

Clinical data: Daily counts of ER visits of four study hospitals of Delhi was recorded to obtain the data for acute respiratory ER visits during the study period. All children visiting the ER of AIIMS (South West) and Kalawati Saran Children Hospital (Central) and adults visiting ER of AIIMS, National Institute of Tuberculosis and Respiratory Diseases (South West) and Vallabhbhai Patel Chest Institute (North) were screened round-the-clock for enrolment. Eligible children (0-15 yr) and adults were included only if, on presentation, they reported acute onset (less than or equal to two weeks) of respiratory symptoms or an acute exacerbation of a pre-existing lung disease in the last two weeks and were currently residing in Delhi (staying continuously for at least four weeks). The patients who were not available because of investigations or procedures or did not provide informed written consent to participate were excluded from the study. Residential PIN code along with demographic and clinical data were recorded. All four participating hospitals obtained approval from their respective Institutional Ethics Committees.

Air pollution data: Daily air quality, viz. 24 h average values for PM_2.5, AQI and meteorological variables (temperature and relative humidity) were obtained from DPCC for 22 CAAQMS16,35. The description of 22 CAAQMS along with longitude, latitude and districts is presented in Table I.

AQI was used to assess the air quality status of the city. AQI given by the Central Pollution Control Board in 2014 was calculated by transforming real-time hourly concentrations of various air pollutants into single index value15,36. AQI varied from 0 to 500; values were categorized as: good (0-50), satisfactory (51-100), moderate (101-200), poor (201-300), very poor (301-400) and severe (401-500). The higher the value of AQI, the greater the level of air pollution. PM_2.5 breakpoints were categorized as: good (<30 μg/m³), satisfactory (31-60 μg/m³), moderate (61-90 μg/m³), poor (91-120 μg/m³), very poor (121-250 μg/m³) and severe (>250 μg/m³)15,36,37. The time period of the study was divided into three seasons, viz. summer (March, April, May and June), monsoon season (July, August and September) and winter (October, November, December, January and February). In order to have the overall picture of air pollution levels in Delhi, the average PM_2.5 and AQI was calculated for every location in three seasons. The total study duration of 365 days was divided according to AQI categories in each of the three seasons37.

Geographic information system mapping: GIS tools were used to study the spatio-temporal changes in PM_2.5 levels, AQI and the associated daily counts of acute respiratory ER visits for the study duration. CAAQMS at different locations of a city represents the ambient air pollution for a particular point. Therefore, inverse distance weighted (IDW) interpolation method was used to spatially predict PM_2.5 concentrations at unmeasured locations in the study area31. In order to locate spatially the number of enrolled patients coming from a particular area in a particular season, the numbers of enrolled patients were mapped by PIN code corresponding to each patient’s residential address. The maps were plotted for winter, monsoon and summer seasons on the basis of breakpoints PM_2.5 pollutant and AQI. These maps were then compared in relation to air quality and number of acute respiratory ER visits of patients during different seasons and at different locations of Delhi. The analysis was done using ArcGIS software, version 10.3.1. (USA).

Statistical analysis: Pearson’s correlation analyses of PM_2.5 levels with temperature and relative humidity were performed. Linear regression models were built for three seasons to assess the relationship between: (i) total enrolled cases and PM_2.5 levels and (ii) duration of acute respiratory symptoms and indoor air pollution indicators (such as choice of cooking fuel, smoker at home, smoker and separate kitchen). The analysis was performed using “mgcv” package in R-software version 3.6 (https://cran.r-project.org/web/packages/mgcv/mgcv.pdf).

Results

During the study period, a total of 70,594 patients attending ER were screened from ER of the participating hospitals. Of these, 18,063 were found eligible of having acute respiratory symptoms (less than or equal to two weeks) and residing in Delhi for the past four weeks. Table II presents the characteristics of the enrolled patients. The average age (mean±SD) of enrolled children was 2.0±3.4 yr and that of enrolled adults was 47.9±16.4 yr, of whom 30.8 per cent (n=2427) were elderly (>60 yr). Adult patients experienced acute respiratory symptoms for 4.6±3.0 days. Among the enrolled children (n=10,186), 10,031 (98%) had cough, 7,439 (73%) had noisy breathing, 8,837 (87%) had difficulty in breathing and 8,901 (87%) reported nasal symptoms. Eighty six per cent (n=8,755) of the children were given ambulatory treatment, 1,394 (13%) were advised admission, and 30 (0.3%) were referred to other departments and seven (0.1%) expired. Among the enrolled adults (n=7877) visiting ER, 7,074 (90%) had cough, 3,953 (50%) had noisy breathing, 7,305 (93%) had difficulty in breathing and 1,970 (25%) had nasal symptoms. Of the enrolled adults who visited ER, 6,220 (79%) were given ambulatory treatment, 1,680 (17%) were advised admission, 40 (0.5%) were referred to other departments and eight (0.1%) expired. The mean number of household members was five living in the average of two-room houses of which about 82 per cent (n=6,462) had a separate kitchen. Around 28 per cent (n=1,631) of adult patients were smokers and 21 per cent (n=2,129) of children lived in households where family members/relatives regularly smoked. Seven hundred and sixty one patients (children: n=692; adults: n=69) had repeat visits during the study period. Of these, 277 patients (children=271; adults=6) visited ER in summer and visited again in winter.

Table III summarizes the total number of enrolled patients, mean of PM_2.5 concentrations, AQI and meteorological data of Delhi during the three seasons. PM_2.5 concentrations of Delhi were high during the study period, with an annual 24-hourly average of PM_2.5 being 120.6±87.0 µg/m³. The maximum concentrations of PM_2.5 with worst AQI were observed in winter season, while relatively lower concentrations of PM_2.5 with moderate AQI were noticed during summer, and minimum concentrations of PM_2.5 with satisfactory AQI were observed in monsoon. The daily ER visits of patients having acute respiratory symptoms mirrored this seasonality in pollution (most in winter, followed by summer and monsoon season). Maximum relative humidity was recorded in monsoon while it was at its minimum during summer. The PM_2.5 concentrations had significant negative correlations with temperature (r=−0.593, P≤0.001) and relative humidity (r=−0.249, P≤0.001). Of 365 days, 64 out of 151 days in winter were reported to have ‘very poor’ AQI. In summer, 54 out of 122 days were observed under ‘moderate’ category, while in monsoon, 48 out of 92 days were observed under ‘satisfactory’ AQI. None of the days during the study period was found to have ‘good’ air quality (AQI<50). The number of enrolled cases was relatively higher in winter months (41.2%) in comparison to summer (27.9%) and monsoon months (30.8%).

Figure 1 shows the distribution of days (March 2018-February 2019) according to AQI range of categories for defining air quality of Delhi. The air quality and daily patient count attending ER due to acute respiratory symptoms varied in different regions and during different seasons of Delhi. Table IV summarizes the seasonal and annual mean concentrations of PM_2.5 concentrations and AQI recorded from 22 CAAQMS in Delhi. During the study period, the lowest annual mean PM_2.5 concentration was observed at Sri Aurobindo Marg, whereas the highest PM_2.5 concentration was observed at Anand Vihar. The lowest annual mean AQI was in ‘poor’ category observed at Dr Karni Singh Shooting range station, whereas the highest AQI was in ‘very poor’ category reported at Anand Vihar.

Close

Fig. 1

Air quality index status of Delhi during the three seasons (March 2018-2019).

Export to PPT

Figure 2 shows the spatio-temporal relationship between seasonal variation of PM_2.5 concentrations obtained from 22 CAAQMS of Delhi versus enrolled cases [i.e., having acute respiratory symptoms (less than or equal to two weeks) and residing in Delhi for the past 4 wks] in three seasons, viz. monsoon, summer and winter from March 2018 to February 2019. The seasonal mean values of PM_2.5 concentrations at each station and number of enrolled cases from various regions of Delhi were relatively higher in winter months (Fig. 2C) compared to summer (Fig. 2B) and monsoon months (Fig. 2A). During the winter season, it was observed that the number of enrolled children had a negative correlation (β=−0.02, P=0.01) while number of enrolled adults had a positive association (β=0.01, P=0.03) with PM_2.5 concentrations. The duration of acute respiratory symptoms was not found to be associated with indoor air pollution indicators such as cooking fuel (β=−0.30, p=0.49) and separate kitchen (β=−0.17, P=0.06) in households, smoking status (β=−0.17, P=0.15) and smoker at house (β=0.04, P=0.56) during winters. As shown in Figure 2C, during the winter season, the number of enrolled cases was high in ‘very poorly polluted’ north-west region (n=1886, PIN code 110 086, Begumpur) of Delhi exposed to the highest mean levels of PM_2.5 (244.8±131.8 and 232.3±109.5 µg/m³) observed at Wazirpur and Jahangirpuri monitoring station, respectively, situated in northwest of Delhi. The number of enrolled cases reporting to ER was lower from the ‘moderately polluted’ southwest region (n<5, PIN code 110 080, Sangam Vihar; 110 060, Rajinder Nagar; 110 066, RK Puram; 110 072, Jharoda Kalan; 110 097, Kapashera etc.) of Delhi exposed to relatively lower mean levels of PM_2.5 (144.8±58.9 and 146.7± 69.0 µg/m³) recorded at Najafgarh and Sri Aurobindo Marg station, respectively, during the same season.

Close

Fig. 2

Spatio-temporal relationship between the seasonal variation of PM_2.5 concentrations obtained from 22 Continuous Ambient Air Quality Monitoring Station of Delhi versus enrolled cases in three seasons from March 2018-February 2019: (A) Monsoon, (B) Summer, and (C) Winter.

Export to PPT

Figure 3 shows the spatio-temporal relationship between seasonal variations of AQI obtained from 22 CAAQMS of Delhi vs. enrolled cases in three seasons vs. monsoon, summer and winter. The seasonal mean AQI of 22 continuous monitoring stations ranges from ‘poor’ to ‘severe’ and the number of enrolled cases was relatively higher in winter months (Fig. 3C) contrasting with the relatively better AQI observed in summer (Fig. 3B) and monsoon months (Fig. 3A). As shown in Figure 3C, during winter season, the number of enrolled cases was high in the ‘severely polluted’ northwest region (n=1886, PIN code 110 086) of Delhi having high AQI (433) observed at Wazirpur station. The number of enrolled cases reporting at ER was low from the ‘poorly polluted’ southwest region (n<5, PIN code 110 038, Rajokari; 110 060, 66, 72, 97, etc.) of Delhi having comparatively low AQI (290 and 291) observed at Sri Aurobindo Marg and Najafgarh station, respectively.

Close

Fig. 3

Distribution of average air quality index obtained from 22 Continuous Ambient Air Quality Monitoring Stations of Delhi versus enrolled cases in three seasons from March 2018-February 2019: (A) Monsoon, (B) Summer, and (C) Winter.

Export to PPT

Discussion

The study examined the spatio-temporal relationship between seasonal variation of PM_2.5 concentrations, AQI and related ER visits of patients having acute respiratory symptoms in Delhi. During the study period, the annual average of PM_2.5 concentration (120.6±87.0 µg/m³) in Delhi exceeded the Indian-recommended limits10. Such finding has been consistently reported by previous Delhi-based studies3,6,9,14,37. There was not a single day to register ‘good’ AQI37. AQI was ‘poor’ for most of the winter days compared to summer and monsoon reporting ‘moderate’ and ‘satisfactory’ AQI, respectively. The minimum annual mean PM_2.5 concentration was noticed at Sri Aurobindo Marg situated in south-west Delhi district5. The maximum PM_2.5 concentration was observed at Anand Vihar37, which is one of the most polluted areas in Delhi due to high traffic congestion and is affected by emissions from road dust, industries, commercial activities of hotels, etc.27,37. The variation observed in the PM_2.5 levels in the present study could be due to the type of emission sources1,10-12 and meteorological factors14. Several Indian reports5,12,14,27-29 have recognized that there is significantly high seasonal and regional variation in ambient air quality and prevalence of respiratory symptoms. Low temperature and high relative humidity play a vital role in the formation and rise in PM_2.5 levels, thereby resulting in evident seasonal variation in air quality3. In the current study, a weak and negative correlation was found between PM_2.5 versus temperature and relative humidity38. GIS analysis showed that spatio-temporal association between air pollution and the number of ER visits for acute respiratory symptoms varied across different geographic areas of Delhi during three different seasons. In winter, ER visits of patients having acute respiratory symptoms were higher from the north-west region of Delhi exposed to very high air pollution recorded at Wazirpur and Jahangirpuri station. Furthermore, enrolled children had negative, whereas adults had a positive association with PM_2.5 levels. Adults are more likely to be exposed to outdoor environmental pollution and even for longer duration, for example, commuting to work place; hence, they might have different intensity of cumulative exposure than children. In contrast, the southwest region of Delhi was less polluted (Najafgarh and Sri Aurobindo Marg station) corresponding to the low ER visits of patients. These results were in line with a previous study from Kanpur, India, which concluded that individuals with respiratory disease were at greater risk of hospital visits than those residing in low polluted area28.

The study had some limitations. First, the present study, we used a wide network of air quality monitoring stations spread around the city; however, the hospitals from each region of Delhi could not be included for collecting daily counts of acute respiratory ER visits. Second, adjustment for differential lifetime exposure to environmental pollutants of children and adults was not possible. Third, assessment of personal exposure at home and workplace to air pollution was practically not feasible in large sample, which could better illustrate the health impact of ambient air pollution32. Fourth, we could not study the sources of regional emission that could have helped us to know the reason for observed pattern. Fifth, we could not take into account the levels of indoor air pollutants and individual exposures to the pollutants that might have affected the quality of associations. Despite these limitations, the present study had several strengths. First, this multisite study had covered a large sample size in Delhi, India. Second, it was possible to study the role of seasonal variation in spatio-temporal association between ambient air pollution and acute respiratory ER visits in Delhi using novel GIS tools. Regional health impact was estimated based on PM_2.5 levels and AQI. Although AQI is not a refined tool, it is an easily understandable generic information tool and help drafting advisories issued to the public. Third, we obtained 24 hourly real-time air quality data of Delhi from 22 newly installed DPCC CAAQMS on the daily basis, which contributed to high spatial resolution and robust results.

To strengthen the findings of the present study, systematic investigations are needed in Delhi to: (i) establish adequate monitoring system for air quality and health outcomes, (ii) demonstrate the causal relationship between air pollution and associated health outcomes and (iii) identify emission sources and their contribution to air pollution and economic evaluation of health impact of air pollution1,29. Severity of air pollution starts from post-monsoon and continues through winter. We acknowledge that indoor air pollution has a significant role to play in the type of investigation we conducted. In the present study, poor air quality was observed for majority of the days throughout the year in Delhi. Therefore to gain potential health benefits, effective measures to control air pollution should be executed throughout the year instead of focusing only during highly polluted winter season. In India, air pollution is one of the causes for producing damaging health effects and not the only cause as there are many aspects such as socio-economic issues, living conditions, location of emission sources, land use pattern, occupational exposure, food habit, other health ailments, etc. Growing evidence suggests that specific toxic compounds of PMs produce harmful effects on lungs and are carcinogenic and genotoxic. The exposure to PM may induce inflammatory response, oxidative stress, hormone dysregulation39 as well as placental dysfunction40. Therefore, geo-spatial distribution of respiratory diseases associated ER visits may serve as a good tool for location-based prevention and control of outdoor air pollution. The present study findings provide reference for decision-makers to improve air quality and related health outcomes at specific locations in Delhi. The study also provides, relevant information for public to modify their outdoor behaviour according to exposure to varying ambient air pollution.

Overall, the acute respiratory symptoms related ER visits of patients were associated with PM_2.5 concentrations and AQI that varied by regions and seasons in Delhi. The study provides GIS-based scientific evidence for policy-makers to make adequate regional monitoring and emphasize localized improvement strategies for management of air pollution and associated respiratory health outcomes in Delhi.