Journal of Environmental Science and Health Part A (2008) 43, 1064–1070 C Taylor & Francis Group, LLC Copyright ISSN: 1093-4529 (Print); 1532-4117 (Online) DOI: 10.1080/10934520802060035
Ambient PM10 and respiratory illnesses in Colombo City, Sri Lanka K. G. THISHAN DHARSHANA and NOWARAT COOWANITWONG Urban Environmental Management Field of Study, Asian Institute of Technology, Klong Luong, Pathumthani, Thailand
Analysis of ambient air quality data monitored at Colombo Fort monitoring unit clearly revealed that PM10 is the dominant air pollutant in the Colombo atmosphere. Further investigation showed that PM10 has strong associations with three types of respiratory illnesses, especially among children. Among these associations, the disease category which includes bronchitis, emphysema and other chronic obstructive pulmonary diseases showed a prominent association with a correlation coefficient of 0.717 at 99% confidence. In addition, an application of health impact assessment software developed by WHO revealed that nearly 20% of Asthma patients recorded at LRH (the Lady Ridgeway Hospital for Children) in 2005 could be attributed to exposure to PM10 in Colombo. It was observed that nearly 60% of the respiratory cases occurred at reasonably lower concentrations (below 80 µgm−3 ) thus, future management plans aiming toward positive health impacts should focus on shifting the entire PM10 pollution distribution towards lower ends. Keywords: Air pollution, PM10 , health impact, respiratory illnesses, guidelines.
Introduction Air pollution has become a globally crucial issue. Asia, being a region with a rapid rate of economic growth is not an exemption. Associated with high concentrations of air pollutants, adverse health effects are an increasing trend. Among the air pollutants, recent epidemiological studies have shown that particulate matter (PM) considerably influences respiratory health.[1] Inhalable particles, often known as PM10 can pass through the natural protective mechanism of the human respiratory system where as PM2.5 can go deep in to lungs.[2] Results of several studies have observed associations between ambient particle concentration and various health endpoints, ranging from increased incidences of Pneumonia/Asthma, and exacerbation of chronic obstructive pulmonary diseases (COPD) to increased rate of mortality.[3,4] Literature shows that 1,092 cases of chronic bronchitis and 4,550 cases of death in Bangkok in the year 2000 were associated with PM10 which led to an estimated cost of US$ 424, 000,000.[5] Similar to other major cities in Asia, Colombo, the commercial capital of Sri Lanka, is also suffering from PM
Address correspondence to Nowarat Coowanitwong, UEM field of study, School of Environment, Resources and Development, Asian Institute of Technology, PO Box 4, Klong Luong, Pathumthani 12120, Thailand; E-mail:
[email protected] Received December 10, 2007.
related issues. Along with other monitored air pollutants such as SO2 , NO2 , O3 and CO, PM10 frequently exceeds air quality standards and guidelines. Measured annual average concentrations in Colombo of PM10 and PM2.5 being 79 µgm−3 and 32 µgm−3 for the period of 2002 to 2004, clearly exceed the USEPA recommended values of 50 µgm−3 and 15 µgm−3 .[6] Batagoda et al.[7] reveals that coarse particles are highly contributed by windblown dust from eroded and deforested areas and dust stirred up by vehicular traffic while PM2.5 is highly contributed by soot and condensed vapor from combustion of vehicles, stationary sources such as power plants and open burning of domestic and agricultural waste. Apart from the few studies on air pollution in Colombo, some researchers focus on respiratory system related illnesses. The studies reveal that the number of patients with respiratory system related illnesses in Colombo are an increasing trend.[8] Field observations indicate that children require more frequent medical visits than in the past. Senior citizens often experience difficulties in breathing, coughing and chest tightness. School absenteeism has become common especially among children. These illnesses become prominent during the Northeast monsoon period where pollution accumulation takes place in Colombo. Though recent epidemiological studies have shown that suspended PM considerably influences respiratory health,[1] no detailed study has been conducted to identify the association of respiratory illnesses with PM10 in the Colombo atmosphere.
1065
Respiratory illnesses in Colombo City, Sri Lanka It is a timely requirement to extend current research interests to this less explored area of PM in order to identify any managerial measures that might be taken. As a vital first step of achieving this task, the current status of PM10 in Colombo was reviewed by investigating the relationships between PM10 and respiratory illnesses. Although literature reveals that O3 can cause respiratory illnesses similar to PM, a proper O3 database was not available. Also due to constraints of a comprehensive database for PM10 , two periods, March 1998 to March 2001 and June 2003 to June 2006 were considered. Because Colombo is a coastal city and experiences good ventilation throughout the year, air quality data monitored at the Fort monitoring unit is assumed to effectively represent the entire study area and, quarterly average pollution concentration is assumed to represent the entire quarter. Health related data from two major governmental hospitals, namely The National Hospital of Sri Lanka (NHSL) and the Lady Ridgeway Hospital for Children (LRH), include monthly data on outpatient department (OPD) visits for Asthma clinics and, quarterly live discharges (LD) for respiratory illnesses. This includes people who leave the hospital within a particular quarter as they recover from the respiratory system related illness which caused them to be admitted to the hospital. As such, hospital LDs represent the temporal variation of ambient pollution concentrations as recorded by hospital admissions.
Methodology Available data for this research study is shown in Table 1. Based on available 24-h average PM10 concentrations, a representative 24-h average value for the corresponding month was calculated. The highest PM10 concentration, which was recorded for a specific month was considered as the monthly maximum. The comparisons were made with appropriate Sri Lankan standards. Since there are no Sri Lankan stan-
dards for PM10 , USEPA standards (150 µgm−3 , 24-h averaging time) and WHO guideline (70 µgm−3 , 24-h averaging time) were used. It is worth highlighting the fact that responsible authorities in Sri Lanka such as the Central Environmental Authority (CEA) use USEPA 24-h average standard for the same purposes. To identify the relationships between PM10 and respiratory illnesses, a statistical tool called SPSS (Statistical Package for the Social Sciences) was used. Data obtained on monthly OPD visits to Asthma clinics at NHSL and LRH were compared with monthly averaged PM10 concentrations. Data on quarterly LDs for respiratory diseases were analyzed with quarterly averaged PM10 concentrations, as they effectively represent the entire quarter. In addition to the same quarter variations between PM10 and LDs, one quarter lag and two quarter lag effects (comparison between pollution concentration in time T with hospital admissions/discharges in time (T+t)) were also considered since there was a prevailing high possibility that patients admitted for respiratory illnesses would be discharged in few weeks time. Although lag effects in 1–5 days are common in literature[9] , studies on quarterly lag effects were not available. Since data on LDs were quarterly based, any smaller lag effects could not be considered. In addition, a health impact assessment software called “AirQ” which has been developed by WHO-European Center for Environment and Health was used to quantify the number cases for various respiratory illnesses that could be attributable to exposure to PM10 . This tool is based on an approach that is often referred to as “Risk Assessment,” which combines information on exposure-response relationships with data on population exposure to estimate the extent to which health effects are expected to result from exposure in the population. It assumes that there is a causal association between exposure and the health outcome and no major confounding effects on this association. It also assumes that Relative Risk (RR) estimate applies to the entire group of the exposed population.[10]
Table 1. Data availability for the study from 1998 to 2006. Data type
Source
Period
PM10 (24-h average)
NBRO
March 1998-March 2001 (2 samples/week) June 2003-June 2006 (2 samples/week) 1st Quarter 1998-1st Quarter 2006 (Quarterly) 1st Quarter 1998-4th Quarter 2005 (Quarterly) April 2001-July 2006 (Monthly) January 2001-June 2006 (Monthly)
CEA Respiratory illness related LDs/deaths
NHSL
LRH
Monthly OPD visits for asthma clinics
NHSL LRH
Results and discussion Reviewing the present status of PM10 in Colombo City As explained in the previous section, air pollutant concentrations were compared with appropriate standards and guidelines. Figures 1a and 1b clearly indicate that the PM10 level frequently exceeds the WHO 24-h average guideline of 70 µgm−3 . Moreover, relatively high 24-h averages throughout the year clearly indicate the seriousness of particulate pollution. This issue is becoming more important with the comparison of recently revised WHO guidelines in October 2005 (24-h average guideline of 50 µgm−3 ).[11,12] Based on Figures 1a and 1b, and the above explanations, it is evident that PM10 has become an issue in Colombo. It is
1066
Dharshana and Coowanitwong
Fig. 1. Variation of PM10 from (a) January 1998-March 2001 and (b) June 2003-June 2006 at Colombo Fort.
the dominant air pollutant as compared to others throughout the year and has been for many years, time. Since studies in other countries have revealed strong associations between PM10 and respiratory illnesses, it is of utmost important to identify such associations for Colombo as well. Identifying the relationships between PM10 and respiratory illnesses in Colombo City Paying attention to quarterly average pollution concentration and quarterly LDs at LRH, disease groups 145, 147 and 149 (assigned names for these disease categories are given in Table 2) showed strong associations with quarterly average PM10 concentration (Table 3). Disease group 149 which stands for bronchitis, emphysema and other chronic obstructive pulmonary diseases (in 1-quarter lag scenario) showed the highest degree of association with ambient PM10 with a correlation coefficient of 0.717 at 99% confidence. This indicates the impact of PM10 exposure on children and the average recovery period. Pneumonia (disease Table 2. Disease category numbers assigned for respiratory diseases at NHSL and LRH. Number∗ 142 143 144 145 146 147 148 149 150 151 152 ∗
Diseases of the respiratory system (J00–J99) Acute sinusities and acute tonsillities (J01–J03) Other acute upper respiratory infections (J00, J02, J04–J06) Influenza (J10–J11) Pneumonia (J12–J18) Acute bronchities (J20) Acute bronchiolitis (J21) Other diseases of the upper respiratory tract (J30–J39) Bronchitis, emphysema and other chronic obstructive pulmonary disease (J40–J44) Asthma (J45–J46) Bronchiectasis (J47) Other diseases of the respiratory system (J22, J60–J98)
Reference: Reports on indoor morbidity and mortality in hospitals.
group 145 in 2-quarter lag scenario) and acute bronchiolities (disease group 147 in 2-quarter lag scenario) showed associations much closer to the highest degree (0.687 and 0.672) at 95% confidence. Comparatively low correlation coefficients with relatively high lag effects might indicate that some other factors other than PM10 could trigger these two illnesses and on average it takes about two months for the patient to be fully recovered. Moreover, neither of the disease categories shows a perfect correlation basically because of confounding factors which have not been accurately quantified using the present scientific knowledge. Apart from these positive associations, some of the disease categories with different lag scenarios showed negatively moderate correlation coefficients. In other words, when the ambient PM10 concentration increases, the number of LDs has decreased and vice versa. Moreover no significant correlation was determined either at 99% or 95% confidence levels between quarterly the average PM10 concentration and quarterly LDs at NHSL, which might indicate a low influence of PM10 for young people and senior citizens. Though literature reveals that respiratory illnesses and atmospheric particle concentration depend on prevailing temperature and relative humidity (RH),[9,13] a proper meteorological data base was not accessible. Only the monthly average ambient temperatures in Celsius were available for the period of 1997–2003 which was used to calculate monthly average and quarterly average temperatures. Since Sri Lanka does not have remarkable seasonal variations as do other western, European and some other Asian countries, unavailability of corresponding monthly RH data was assumed to have no significant impact. Nevertheless, negatively significant coefficients between ambient temperature and some disease categories (Table 3) are worth noting. In addition to these findings, AirQ software was applied to quantify the number of patients affected by exposure to PM10 in 2005. Values in Table 4 were obtained considering the winter season to be December to February and the summer season to be June to August in Sri Lanka. To estimate the number of excess cases for hospital admissions
1067
Respiratory illnesses in Colombo City, Sri Lanka
Table 3. Degree of association between quarterly average pollution concentrations and quarterly LDs at LRH for the period of January 1998 to March 2001. Disease group and the corresponding time lag Temp Qtr Avg Qtr Avg PM10 Temp
143 2-Qtr lag −0.701**
145 0-Qtr lag
145 2-Qtr lag
−0.646*
0.687*
147 1-Qtr lag
147 2-Qtr lag 0.672*
0.671*
148 0-Qtr lag −0.723**
148 2-Qtr lag
149 1-Qtr lag
152 0-Qtr lag
0.717**
−0.623*
152 2-Qtr lag
Note: Degree of association has been measured in terms of value of the Pearson Correlation. N = 13 except for PM10 for which N = 12. ∗ Correlation is significant at the 0.05 level (2-tailed). ∗∗ Correlation is significant at the 0.01 level (2-tailed).
for respiratory diseases and respiratory mortality, baseline incidences were calculated based on NHSL and LRH data from 1998–2005. For asthma attack hospital admissions in children as well as adults, baseline incidences were calculated based on data from 2002–2005. It was noted that the calculated figure for hospital admissions for respiratory diseases (2183) is remarkably high compared to the software default figure (1260, which is automatically displayed when running the program) while calculated the figure for respiratory mortality (45) is lower than the software default figure (66). Although default RR values had to be used, confidence increases with the tailor made baseline incidences. Figure 2 shows the software interface at the data input stage, where the user has the flexibility to select the health endpoint, input the baseline incidence and RR values including both upper and lower estimates. Then the program generates the estimated AP percentage and estimated number of excess cases. Table 5 indicates that an estimated number of 671 patients were admitted to hospitals in Colombo for the treatments of respiratory diseases due to exposure to PM10 in 2005. Comparing to the total number of LDs for the same year in two hospitals, this figure rounds off to 4% (Table 6). Figure 3 is a graphical presentation of the hospital admission data in Table 5. Hospital admissions for respiratory illnesses vary with the ambient PM10 concentration. The upper RR scenario (rr = 1.0112) represents the cumulative number of excess cases as a result of increases in PM10 . Attributed cases are at the highest compared with the other two scenarios. The middle curve (rr = 1.0080) represents the variation of cumulative number of excess cases due to
PM10 at a middle level of RR estimates. Similarly, the bottom curve stands for the lower RR estimates (rr = 1.0048). Thus, the value of the middle curve is considered with the possibilities of variations between corresponding top and bottom curves. Table 6 summarizes the estimates based on application of AirQ software. It reveals that nearly 20% of Asthma patients recorded at LRH and 2% of Asthma patients recorded at NHSL could be attributed to PM10 in 2005. Asthma attacks in children due to PM10 are nearly 10 times higher than that of adults. The percentage of PM10 related hospital admissions for respiratory diseases is 4%. Nearly 50,000 of people in the CMC area lived below the poverty line in 2002.[14] This population, whose living conditions and time-activity patterns vary significantly from the rest of the population, are also highly affected by malnutrition and other adverse impacts of poor social conditions making
Table 4. Statistics of PM10 concentration in the year 2005 [Units: µgm−3 ].
Arithmetic mean Maximum
Annual
Winter
Summer
71 126
90 125
67 94
Source: Central Environmental Authority’s database, 2005.
Fig. 2. AirQ software screen for parameter selection.
1068
Dharshana and Coowanitwong
Table 5. Hospital admissions for respiratory diseases due to PM10 in Colombo. Health endpoint: Hospital admissions for respiratory disease Baseline incidence: 2183 per 100,000 per year Scientific certainty of relative risk: High RR per 10 µgm−3 Lower (1.0048) Middle(1.0080) Upper (1.0112)
Estimated AP %
Estimated # of excess cases
2.8904 4.7262 6.494
410 671 922
Note: 1) AP: Attributable Proportion-Fraction of all cases that could be attributed to a specified exposure causing health outcome. 2) Excess cases: Number of additional cases that result due to PM10 in the atmosphere. In other words, if the PM10 in the atmosphere is in safer levels, the excess cases should be equal to zero. 3) Since the exact number of cases cannot be identified based on the present scientific knowledge, a particular range is given. In other words, estimated number of cases can vary within the lowest and the upper estimate with a high possibility to be much closer to the middle estimate.
them highly vulnerable to respiratory illnesses.[9] Thus, a significantly higher proportion of the cases that have been recorded at NHSL and LRH for respiratory treatments could be reliably assigned to this population. Although there is no evidence to justify the effect of smoking on respiratory illness related to hospital admissions in the CMC area, there is a growing trend of people breaking smoking habits, possibly partly due to government policies and the strong religious and cultural boomed which started around year 2000. This trend might also have an impact on the number of patients receiving treatment for respiratory illnesses. Apart from the above arguments, it is worth noting the rationale behind the theory of associations. Thus, the absence of correlation only means that the variables (here, PM10 concentration and the number of patients receiving treatment for various respiratory illnesses) are not linearly associated. In other words, there is a possibility that these variables are associated in some other form that is much more complex than the simple linear association.
Steps forward: Some of the policy implications As shown in the above sections, PM10 considerably influences the respiratory health of Colombo citizens, especially among children. More importantly, it is a vexing challenge beyond shortening life and contributing to respiratory illnesses. Although any threshold limit for PM10 below which adverse health effects could be considered zero, a limit at which people can considered to be completely safe has not been identified,[15] there are few feasible management methods to attain the optimal levels. For any method for improvement to be successful, strong and effective policies and reliance on scientific proof and effective instrumentation with careful consideration of stakeholders’ interests are essential. Because human and environmental health is strongly related, both should be considered in the context of a temporal scale from short-term to long-term (Figure 4). More importantly, proper balance is required between both short-term and long-term policies.[16] Considering the present situation in Colombo, improper economic and planning policies have become the root cause of the problem. Use of low quality fuels, incoherent traffic management policies, poor land use planning, use of reconditioned vehicles and engines are some of the examples that contribute to the problem. Although since 1992 Sri Lanka has taken mixed measures (a combination of economic instruments, regulatory instruments and suasive instruments) such as the Clean Air 2000 Action Plan, revised standards for fuel quality and vehicle importation and banning of leaded petrol, some attempts including control of importing two stroke vehicles and the introduction of vehicle testing programs have failed due to socio-economic and political issues.[17] Based on these past experiences, it is clearly evident that effective implementation of any policy aimed on reduce respiratory illnesses associated with PM10 will depend upon the consideration of all stakeholders’ interests. Although proper source apportionment studies are not available, previous studies suggest that vehicular emissions yield the highest contribution of PM10 to the ambient atmosphere in Colombo followed by powerplant emissions.[7] Thus, encouraging technological innovations to achieve lower levels of emissions and by encouraging public
Table 6. Summary of the excess cases which could be attributed to PM10 in 2005. Health endpoint Hospital admissions for respiratory diseases Respiratory mortality Asthma attacks for children (LD+OPD) Asthma attacks for adults(LD+OPD) Asthma attacks for children (LD) Asthma attacks for adults(LD) ∗
Total # of cases recorded
Estimated # of excess cases attributable to PM10
% contribution
17,770∗ 438 1234 2281 1099 1442
671 20 275 39 216 22
4 5 22 2 20 2
Since the records on hospital admissions for respiratory diseases were not available, this figure shows the total LDs for respiratory diseases.
1069
Respiratory illnesses in Colombo City, Sri Lanka
Fig. 3. Excess number of hospital admissions for respiratory diseases due to PM10 .
transportation, and forcing larger organizations in Colombo to introduce carpooling and/or official buses for their employees, are a few future possibilities. More importantly, care should be taken to avoid any adverse impacts associated with the implementation of policies on particulate pollution. As an example, if the implementation of strict pollution abatement regulation required shutting down a plant with limited profits basically because of its inability to afford the cost of new abatement equipment, it would directly impact its employees and indirectly the municipal income, whose economic base significantly depends on its tax revenue.[18] In such a situation, government assistance in subsidizing the purchase of pollution control equipment would create a win-win situation. The fact that a number of thermal powerplants and refineries are located outside the boundaries of the CMC area and are close enough to have direct impacts from their plume dispersions; “pollution transport” should also be reflected in any policy considerations. Therefore, health impacts due to exposure to ambient PM10 should become more prominent in policy discussions and more likely to stimulate all parties to come up with collective efforts. To
address these issues, the areas of policy, enforcement and technology innovation require much more research and future investigations.
Conclusions Analysis of major air pollutants monitored at the Colombo Fort monitoring unit revealed that the presence of PM10 in the atmosphere requires a great attention. Further analysis based on this identification led to a significant conclusion that PM10 concentration has stronger relationships with three respiratory disease categories among children, namely, pneumonia (disease group 145), acute bronchiolitis (disease group 147) and the category which includes bronchitis, emphysema and other chronic obstructive pulmonary diseases (disease group 149). The application of health impact assessment software developed by WHO revealed that nearly 20% of Asthma cases recorded at LRH and 2% of Asthma cases recorded at NHSL could be attributed to PM10 in 2005. By considering the software outputs, 4% of total cases for hospital admissions for respiratory diseases and respiratory mortality in general could be attributed to PM10 pollution in Colombo. In addition to PM10 as the highest contributing air pollutant to respiratory related hospital admissions, the conclusion may be drawn that there are contributing factors, other than air pollution which have a great impact on the respiratory system related hospital admissions/deaths of Colombo citizens. A relatively high percentage of the population of Colombo is living below the poverty line with its associated poor living conditions and malnutrition. There are no doubt other confounding, non-linear factors that should also be considered.
Acknowledgments Fig. 4. Relationship between policy, ambient PM10 and respiratory illnesses.
The authors wish to thank the National Building Research Organization (NBRO), the Lady Ridgeway Hospital for
1070 Children (LRH) and the National Hospital of Sri Lanka (NHSL) for providing data free of charge and the Central Environmental Authority (CEA) for providing air quality data at a discounted rate. Moreover, the Norwegian Agency for Development Corporation (NORAD) is acknowledged for providing the research grant.
References [1] Gupta, A.K., Nag, S.; Mukhopadhyay, U.K. Measurements of inhalable particles
