Report No. 36945-BD Bangladesh Country Environmental Analysis (In Two Volumes) Volume II: Technical Annex: Health Impacts of Air and Water Pollution in Bangladesh August 23, 2006 South Asia Environment and Social Development Unit South Asia Region Document of the World Bank Table of Contents Page Executive Summary............................................................................ .i 1.Introduction ................................................................................... 1 2. Ambient Air Pollution ....................................................................... 2 3. Indoor Air Pollution ......................................................................... .5 4. Access to Clean Water and Sanitation.. ................................................... 10 5. Discussion and Conclusions. ................................................................ 13 References. ...................................................................................... .15 Annexes Annex-I Methodology and calculationo f Health Cost o f Ambient Air Pollutionin Bangladesh Annex-I1HealthBurden due to Indoor Air PollutioninBangladesh Annex-111HealthCost ofWater and Sanitation inBangladesh Acknowledgements This TechnicalAnnex was prepared for the Bangladesh Country EnvironmentalAnalysis byDr.M.Khaliquzzaman, withvaluable guidance fiomK.Lvovsky, R.B.Kaufhan, P. Martin, and B.A.Haque. HealthImpacts ofAir andWater PollutioninBangladesh Executive Summarv 1. The health impact of air and water pollution inBangladeshhas beenestimated usingthe most recently available data and widely acceptedmethodologies reported inthe literature. These estimatesrange between 1.20-3.35% of GNI. Even withthe benefit o f includingmore recent data, the results obtained still containmany uncertainties arising from the variety of assumptions made inorder to arrive at the values reported. Ambient Air Pollution 2. The most important pollutant from the healthpoint of view inDhaka i s particulate matter (PMlo andPM2.5). The levels of these parameters aremuchhigherthan the proposedrevised standards. The levels of other airborne pollutants are still within the proposed revisedstandards withthe exception ofNO2whichis now close to the standard. The number o f cases o fmortality andmorbidity that canbe avoidedper year ifthe PMlo pollutionlevel is reducedbya modest 20% of the current level, or to the proposednational standard, are found to be between 1,200- 3,500 and 80- 235 millioncases respectively. The costs involved are betweenUS$ 169-492 millionper year, which correspond to 0.34 -1 .O% o f GNIinthe WTP basedestimates. Indoor Air Pollution 3. The disease burdendue to indoor air pollution has been estimated usingWHO recommended fuel basedmethodology. Ifthe indoor air pollutioncan be reduced inrange from a modest 20% of the current levelto 80% of the current level (i.e., approximately equivalent to the reduction to proposed national ambient air quality standard), then 7,600-30,400 deaths and about 0.30- 1.20 millionDALYscan be avoided per year. The economic costs o f these health impacts are betweenU S $ 114-458 millionper year inthe human capital approach. Interms o f GNI, these correspond to 0.23 -0.92 %per year. As some new exposure data have become available recently, these also have beenusedto calculate the indoor air pollution healthburdenusingthe same methodology as for ambient air pollution. It i s shown that comparable healthburdenare obtained inboththe methods but cost estimates are muchlower inthe humancapital approach. Water Supply and Sanitation 4. A total ofbetween0.82-1.94 millionDALYscanbe savedthroughthe provisionof clean water and sanitation inBangladesh. The economic value o f water-related mortality and morbidity has beenestimated byusingthe human capital approach andalso the GNI-scaled USEPA annualized value of life approach. This latter approach gives higher values comparedto the humancapital approach. The coststhat canbe savedper year are estimated to bebetween US$313-739 millioninthe humancapital approach method. InGNIterms, these correspond to 0.63-1.43 %per year. Compared to the earlier (1997) study, the water-related healthcosts have increased somewhat. This increasereflects the impact o f increaseinthe populationandper capita GNI. Reducedaccessto cleanwater as a result ofArsenic contaminationhas not been considered. 1 HealthImpactsof Air andWater Pollutionin Bangladesh 1. Introduction 1. Analysis o f the overall burden o f diseases inthe south andeast Asia regions shows that environmental factors account for about a fifth o f the total('). The main contributors to this burden are ambient air pollution, indoor air pollution and diseases arisingfrom the lack o f access to clean water. Itis necessary to quantifythe nature and the extent o f the problems and evaluate their costs inorder to prioritize key environmental challenges. The first such work on Bangladesh was done by Brandon (2) in 1997 andit was based on the rather sparse data available at the time. However, this pioneering study provided a first estimate o f the economic cost o f environmental pollution and helpedpolicy-makers to realize the importance o f environmental factors in the national economy. The present work i s an attempt to update that work with more recent data. 2. Important developments have taken place inthe field o f ambient air quality since 1997,both interms o fmonitoring and actions to reduce or limit the emissions. Six criteria pollutants (PMlo, PM2.5, CO, S02, NO2, 0 3 ) are now being regularly measured in Dhaka since April, 2002 under the World Bank financed Air Quality Management Project (AQMP) (3). These measurements are being made using state o f the art equipment andmethods conforming to USEPA Federal Reference Methods (FRM). The data are quality assured and the station is located so that it provides the average population exposure to the pollutants. The Bangladesh Atomic Energy Commission (BAEC) has also beenmonitoring PMlo and PM2.5inDhaka and Rajshahi. They also monitor lead (Pb) levels through the chemical analysis o f P M samples and elemental or black carbon through reflectance measurements. These measurements have shown that PMlo and PM2.5 are the most important pollutants from the healthpoint o fview (4).The level o f the other criteriapollutants are still withinthe proposedrevised standards with the exception o f N02. 3. Regarding actions to reduce air pollution, one o f the major achievements was the introduction o funleaded gasoline from July, 1999. The lead level inair has now fallen by about two thirds o fthe highprevious level to well withinthe recentlyrevised standard (5,6). Inaddition, the two-stroke three-wheelers inDhaka which contributed a large proportion o fthe P Mpollution were withdrawn from lSto fJanuary, 2003. This lead to the immediate decline o f PMlo andPM2.5levels by about 30% and 40% re~pectively(~). The new data available from the measurements at CAMS andby BAEC have beenused for health impact and their cost evaluation. 4. Recently, WHO has published a report on the assessment o fhealthburden o f Indoor Air Pollution(*). The methodology inthis report has been used inthe current work. Untilvery recently, there was an almost complete data vacuum regardingindoor air pollution inBangladesh. Fortunately, a recent study by DECRG o f World Bankhas producedthe first measured data (9-10) on this important problem. This work was still in progress while the present work was done and only the provisional data were available at the time. Hence, these were used here to estimate the health impact and the cost o f indoor air pollution. 5. sanitation by the Government o f Bangladesh (GOB)`13 have beenused to update Inthe caseofaccessto clean water, the recent1 publisheddata onwater and Y Brandon's 1997 estimates. Some estimates of the negativeimpact o fArsenic contamination are now available inliterature (14). However, the impact of Arsenic contamination has not been considered here and the reasons are discussed inthe Annex- 111. 2. Ambient Air Pollution 2.1 Extent of the problem 6. Economic, industrial and demographic growth are drivingurbanizationin Bangladesh as inother developing countries. The capital city Dhakanow has a population inexcessof 12million. Emergenceofsuchurbanconurbations ofextremelyhigh population density i s affectingthe quality o flife inmanydifferent ways. Uncontrolled emissions from motor vehicles and other economic activities give rise to severe air and other forms o fpollution. Highlevels of emission o f air pollutants ina small area exceed the processes o f dilution and dispersal, leadingto severe episodes of ambient air pollution, disproportionately affecting the urbanpoor. There are now three cities in Bangladesh which have populations inexcess o f one million. Inthese cities, severe episodes of air pollution can be observed with the unaided eye. Inaddition, there are eighteen other cities inthe country where population now exceeds 100,000, and air pollution i s a growing concern. The details on these can be found inAnnex-1, Fairly comprehensive air quality data are being collected only for Dhaka, bothby the AQMP andBAEC(3-5'. The summaryo f air quality data for Dhaka obtained at the CAMS is shown inTable- 1. Table 1-Average values for Criteria PollutantsMeasured at CAMS, Dhaka with standarddeviations during2003 alongwith BangladeshStandardsQ! PM2.5 24 hour --- 65 pg/m3 --- Annual --- 15 pglm3 76 k 57pg/m3 2 7. Itcanbe seenfrom Table-1 that the mainpollutant o fconcern inDhaka is particulate matter. Both PMlo and PM2.5levels are extremely high,being muchabove the proposed standard. The NO2 levels are also now close to the limit and may become a concern inthe future. Levels o f other pollutants are still low and thus are not important from health point o fview. Lead (Pb), one o f the criteria pollutants, i s not shown inthe table. The Pb level i s now sufficiently low (i.e., around 100ng/m3) that airborne leadi s no longer considered a healthissue (5). However, blood lead levels inchildren are still high(6),indicatingthat other sourcesmay exist. Basedonthis picture, the following healthimpact calculations are limited to PMlo only. 8. There i s no air quality data for the other two cities (Chittagong and Khulna) with more than one million population. For impact calculationthe pollution level inthese cities i s assumed to be same as Dhaka. As inDhaka, there are highlevels o fpublic complaint about the air quality inthese two cities. The banningo ftwo-stroke three- wheelers inDhaka has contributed to the improvement inthe air quality inDhaka, but the influx o fthese vehicles has madethe air quality worse inother cities 9. InRajshahi, one ofthe 18 cities with apopulationover 100,000, data onP M levels are available from BAEC measurements. The yearly average PMlo level is reported to be 63k25 pg/m3.This ambient level has beenused as an estimate for all the 18 cities. The detail data on the cities can be found inannex-1. 2.2 Methodologies 10. For the calculation o fhealthimpacts andtheir cost, a bottom-up approach has been adopted. Fromthe measurements, pollution levels o f PMlo have been quantified and exposure levels havebeen equated to these levels as explainedbelow. The coefficients relatin pollution and exposure levels with the epidemiological data available inthe literature("" F have been used to estimate the number o f cases o f sickness and death. Estimates o fthe number o f cases have beenrelated to the Willingness-To-Pay (WTP) data available inliterature usingscaling laws. The details o fthe procedure adopted are to be found inannex-1. Two scenarios have been usedto calculate health cost savings. The first scenario i s the reduction of pollution levels by 20%, which is considered a possible achievement level with low cost measures, and the second scenario is the reduction o f the pollution level to the proposedBangladesh standards, which i s a longterm goal. 2.3 Summary of the results 11. An important determinant o fthe healthrelevance o f air quality data is the pattern o f exposure. This varies significantly among the population, but is usually consistent in time for a given population group. Exposure can be defined as the event inwhich an individual remains incontact with a specific concentrationo f a pollutant for a certain period o ftime. Exposure assessment consists o f describing and quantifyingthe relevant conditions and characteristics o fhumanexposure. The available exposure data for Dhaka 3 have beenmeasured as 24-hour average mass-based concentrations o f PMlo at a suitable location usingfixed monitoring equipment. As these measurements were taken for regulatorypurpose, such data are usually insufficient for personal exposure assessment since information about diversity interms o f time spent at different places are not captured insuch measurements. However, inthe present case we shall take the yearly average o fPMlo level as the measure o f exposure. A significant fraction (probably as highas 20%) ofthe citypopulationlive andwork near the traffic canyons where the pollution levels maybe higher by a factor o f 2 or even more. The present calculations may therefore be considered conservative. The results ofthe calculations are shown in Tables 2 and 3. The number o f cases ofmortality and morbidity that canbe avoided if the PMlo pollution level is reduced ina range from a modest 20% o f the current level up to the proposed national standardare shown inthis table. It can be seenthat about 1,200- 3,500 deaths and about 80-235 million cases o f sickness canbe avoidedper year inthe two scenarios. The costs involved are between US$1694 9 2 million. Interms o f GNI, these correspond to 0.34 -1 .O% per year. Reductionin3 Major Cities Reduction18 Cities Total s1 HealthEffects By20% IToNational By20% IToNational By20% IToNational Reductionin3 Major Cities Reduction18 Cities Total By20% To National By20% To National By20% To National S1 HealthEffects Standard Standard Standard 1 Mortality 30.21 94.2 5.3 5.5 35.5 99.7 2 IChronic Bronchltis 62.31 194.3 6.0 6.2 68.3 200.5 8 Respiratory symptoms I 42.01 131.11 4.11 4.2 46.1 135.3 lSub-Total(Morbidity) 121.81 380.11 11.81 12.2 133.6 392.3 4 3. Indoor Air Pollution 3.1 Extent of the problem 12. The issue o fhealthhazards due to air pollution inBangladesh has traditionally been linked to urbanoutdoor air quality problems. Recent studies indeveloping countries demonstrate, however, that illnesses arising from the use o ftraditional fuels in the home pose a serious healthrisk, especially to women and children inrural and urban poor households. These households inBangladesh continue to be dependent on traditional biomass fuels like dung, firewood and crop residues. Only a small fraction o f the households have access to kerosene, far less other clean fuels such as LPG. According to available information, a total o f about 70,000 tons o f LPGis sold in Bangladesh which i s mostly inurbanareas. Biomass-fuelled cook-stoves have high emissionrates o f air pollutants and the houses involved are generally inadequately vented. 13. Continued exposure to smoke from traditional fuels has been shown to cause Acute Lower Respiratory Illnesses (ALRI), Chronic Obstructive Pulmonary Diseases (COPD) and lung cancer among other diseases. ALRI includes viral and bacterial infections o fthe lungs and respiratory tract, the most severe and fatal being bacterial pneumonia. Theburden o f disease due to ALRI on a per capitabasis i s muchhigher in developing countries than indeveloped countries. Severe indoor air pollution from biomass burningfor cooking i s an important contributory environmental factor in Bangladesh. 3.2 Methodologies 14. Different methods to estimate the burden o f disease from indoor air pollution in developing countries have beenreviewed ina recent WHO publication(*).The more widely used approaches are the fuel based and the pollutant based method. Inthe fuel- based approach some fraction o f the burden due to selected diseases are attributed to indoor air pollution usinga step-wise procedure. The contribution for all the relevant diseases are summed to obtain the total burden o f diseases inDALYs ( Disability Adjusted LifeYears) andnumber o f deaths. By contrast, inthe pollutant-basedapproach, the population exposures to an indicator pollutant, generally PMlo, is estimated interms o f some measure o f concentration-time. The exposure-response relationship for the indicator pollutant i s then used to determine excess morbidity and mortality. Both the methods have advantages and disadvantages and these usually leadto widely different results. The WHO studyrecommends the fuel based approach. However, boththe approaches are usedhere for comparison andto illustrate the uncertainties involved inthe calculations. 3.3 Health Burden usingFuel-Based Methodology 15. Inthismethoda stepwise procedure is usedinorder to arrive at the estimate for the healthburden due to IAP inBangladesh. The steps involved are: 5 (i) Collection o f data relevant to calculation; (ii) Calculationofattributefunctions; (iii) Calculation o f the attributable burden; (iv) Summation o f all the burdens due to individual diseases; and (v) Estimationo funcertainty. The problem with this method i s that reasonably reliable Bangladesh specific epidemiological and solid fuel use (SFU) data are not available from the primary national sources. Forthe DALYs associated with indoor air pollution, it i s necessary to depend largely on proxy data fkom WHO SEAR Dregion and this source has also beenused for SFUdata. Suchdata do not capture the variety o flocal complexities inthe exposure to indoor air pollution. All types o f solid fuel are not equivalent interms o fpollution andin a recent study inBangladesh it has been found that cross-household variation i s strongly affected bythe structural arrangements such as cooking locations, construction materials andventilation practice^(^"^). The details o fthe methodology and calculations are given inannex-11. 3.3.1 The AttributableburdenofDALYs andDeathto SFUinFuel-BasedMethod 16. The attributable burden o f disease due to SFUinBangladesh for different diseases i s shown intable-4 below (details inAnnex-11). Table-4 Attributable burdensfrom SFUusingfuelbasedmethodology 17. The low and the highvalues intable-4 do not include the uncertainties inthe exposure estimates. The results o f R M S combination o f these uncertainties with the values intable-4 are shown intable-5. Comparison for DALYs lost and the number o f attributable deaths for Indiaand Bangladesh per 1000population is shown inTable-6. The numbers obtained are very similar which i s not surprising. The same WHO data for diseaseburden (for SEAR Dregion) havebeenusedbypopulation scaling inboth the cases. The only country specific input i s household SFUdata. InBangladesh, the percentage o f households usingbiomass i s higher whereas inIndia the percentage o f households using coal i s higher. These differences are offset by one another andresults are almost similar except for men230 years. The numbers for men2 30 years inIndia are wholly for coal smoke. 6 Table-5 Attributable burdens from SFU for Bangladeshwith exposure uncertainty s1 DALY(000) Deaths(000) Disease, sex, age group low lcentral IHigh low lcentral]high Table-6 DALYs lost and Deaths per 1000 populationinBangladesh and the corresponding values for India from literature (see annex-11) SI. Disease, sex, age group Bangladesh India DALY Death DALY Death 1. ALRI<5yrs 78.93 2.26 78.72 2.22 2. Women230yrs 7.93 0.47 6.14 0.34 3. Men230yrs 0 0 3.7 0 3.3.2 Health Cost of Indoor Air Pollution due to SFU inthe Fuel-Based Method 18. Total DALYs lost due to indoor air pollution SFU are between 1.24 -1.71 million for Bangladesh (Table-5). The economic value o f the DALYs lost can be estimated by using the humancapital approach. This implies that the statistical value o f one DALY i s equal to the annual average productivity o f a worker inBangladesh. This maybe set equal to per capita GNI for Bangladesh. A higher estimate for the cost o f a DALY i s obtained ifGNI-scaled USEPA annualized value o f life is used. This value i s 2.73 times higher thanthe GNIas reportedby Brandon(2). 19. The results o fthe calculations for cost savings by suitable interventions intwo scenarios are shown inTable-7. The first scenario involves 20% reduction inDALYs which can probablybe achieved at low cost through actions such as kitchen ventilation improvement. The second scenario involves reduction o fthe exposure level to proposed national ambient standards.As the measured exposure has been estimated to be about 250 pg/m3,itwould require about 80% reductioninthe exposure levelto reachthe standard o f 50 pg/m3ifa linear relationship betweenDALYs saved and exposure i s assumed. It can be seenthat about 8-40 thousand deaths and about 0.30- 1.20 DALYs (central estimate) can be avoided per year inthe two scenarios (Tables-7). The costs involved range from about US$114 to US$458 million per year. These savings correspond to 0.22-0.92 % of GNIper year. 7 Table-7 Cost for possible savings inDALYS (million) and their Cost Savings inDALYs Cost per Estimates (million) DALY Savings ( US$ Million) DALYs Scenario- Lost Scenario- 1 I1 (million) (20%) (80%) U S $ Scenario-1 Scenario-I1 Low 1.246 0.2492 0.9968 380 94.7 378.8 Central 1.505 0.301 1.204 380 114.4 457.5 3.4 Pollutant-basedMethod 20. Inthe pollutant-based method, data onthe exposurepatterns for different populations are needed. Such patterns are even more important indetermining the health relevance o f indoor air quality than ambient air. Inthe case o f indoor air pollution, exposure levels are significantly different for men, women andinfants. M e ninthe rural areas mostly spend their time outside the household. Women, however, spend significant amounts o f time inthe kitchen andinfants(0-4 years) spend most o fthe time inside the household. Indealingwith indoor air pollution, one should be aware o f the major influence o f lifestyle and cultural factors indetermininghealthimpacts. Inthe absence o f reliable data on these factors, exposure to indoor environmental factors i s inadequately characterized and this leads to large uncertainties inhealth risk assessments for indoor air pollution. 21. Some provisional data are available on indoor air pollution from rural andperi- urbanareas inBangladesh from the DECRG, WB The measurements were taken infive different kitchen types and for five different fuels which included firewood, cow dung, crop residues (rice husWstraw), bagasse (including jute sticWsawdust) and cleaner fuel like kerosene. The un-weighted 24 hour averaged PMlo levels inthe kitchen and livingroomhavebeen found to be 256 k 108and 226+ 114 pg/m3respectively. The ambient levels o fPMlO inthe rural areas are reportedto be below the national standard. 22. Inorder to maintainconsistency withthe ambient air methodology, the same approach as inthe ambient air case has been adopted. Some o f the difficulties inthis method(8)are reduced by the availability o fnewer data. For example, coefficients relating exposure to mortality andmorbidity which are applicable to developing countries are now available(">12). The level o f exposure inthe case o f indoor air pollutioni s also not all that different from highlypollution cities inthe developing countries, so it is a reasonable assumption to use these for the indoor air pollution also. Thus, the pollution (i.e., direct exposure) based methodology i s progressivelybecoming less uncertain. 23. As data from survey were unavailable at the time o fthese calculations, the following contingent estimates havebeenmade inorder to derive exposure levels for different population groups: 8 Men spend 8 hours indoors (living room). Women spend 16hours indoor out o fwhich 4 hours is inthe kitchen. Infantsspend20 hours indoors ( livingroom) New data have become recently available("), however, and use o fthese would change some o f the details o fthe pollution based calculations for the health impact o f SFU, and their costs are given inannex-11. 3.4.1 SummaryofResultsfromPollutant-BasedMethod 24. As inthe case o f ambient air pollution, two scenarios havebeenused. The first scenario is the reduction o f exposure level by 20% which i s a probable achievement level with low cost measures andthe second scenario is the reduction o fthe pollution level to proposed Bangladesh standards which is a long term goal. Under these two scenarios the exposure levels reductions that can be achieved are shown intable 8. Table 8: IndoorAir exposurelevelfor differentpopulationgroupsin@m3for PMlo. Exposure inScenerio-2 2. Women 0.0 0.0 The results o fthe calculations are shown inTable-9. Table 9: Casesavoidedandannualhealthcost savings (US$ million), two scenarios Health Effects Scenerio-1 Scenario-2 Cases Cost Million $ Cases ' Cost Million $ Mortality 5718 161.5 21,031 594.2 Chronic Bronchitis 77,926 265.0 305,139 1,037.7 Resp Hospital adm. 31.075 2.3 121.493 9.0 Asthma attack 4,220,967 4.6 16,502,854 18.1 Emergency room visits 609,580 1.3 2,3 83,296 5.2 Restrictedday activities 72,577,889 67.1 284.198.084 262.7 Lower respiratory illness 595,725 0.5 2,322,632 1.8 Respiratory SPPtoms 233.013.223 178.8 912.425.428 700.1 Sub-Total(Morbidity) 311,126,383 519.7 1,218,258,926 2,034.6 Total 681.2 2,628.7 9 25. The number o f cases o f mortality and morbidity that canbe avoided ifthe PMlO pollution level i s reduced ina range from a modest 20% o f the current level up to the proposed national standard are shown inthis table. It canbe seen that about 5,700- 21,000 deaths and about 311 1,2 18 million cases o fmorbidity canbe avoidedper year in - the two scenarios. The costs involved are between US$681 -2,629 million. As pointed out earlier, the present estimates are quite conservative and may be considered underestimates. Interms GNI, these correspond to 1.4-5.2% per year. 3.5 Comparison of the Results for Two Different Methods for Indoor Air Pollution 26. The central estimates for different parameters calculated inthe fuel and pollution basedmethodologies are shown intable-10. Table 10: Values for different parameters obtained inthe fuel and pollutant based approach inindoor air pollution Item Scenario-I Scenario-I1 Fuel PollutantCost(US$) Fuel Pollutant Cost(US$) Based Based Million Based Based 1.Mortality (Avoided Deaths) 7,600 5,718 - 30,400 21,03 1 - 2. Morbidity DALYs Saved (million) 0.301 114 1.204 - 458 Reduced number o f Cases I I I I I I I (million ) - 311 520 1218 2,629 3. DALY Lost/Case 0.001 - -- - Itcanbe seenthat the mortality results obtained are remarkably close for boththe methods. Inthe case o f morbidity, the results from the two method are obtained intwo different units as such they can not be directly compared. The morbidity i s found inunits o f DALY inthe fuel based method andincases for the pollutant based method. However, ifwe lookat the lifeyears lost andcomparethemwiththe numberofcases, we seethat it amounts to about 0.35 days per case, which appears quite reasonable. So, it i s probably the methods o fvaluation rather thanhealth effects which are at the root o f the large difference (a factor o f about 5) inhealth costs obtained inthe two methods. These are discussed inmore detail inAnnex-11. 4. Access to Clean Water and Sanitation 4.1 Extent of the problem 27. Availability o f water inadequate quality and sufficient quantityi s a necessity for humansurvival, healthprotection, andsocial andeconomic development. Bangladesh is one o f the most densely populated countries inthe world, with a range o f competing demands for water. Microbiological contamination o f surface water i s so pervasive that there is now no potable surface water available inthe country except for a few natural springs inthe hilly areas. Infectious diseases from microbiological contaminants such as 10 viruses, protozoa andbacteria are likely to pose increasingchallenges inthe future due to increasingwater pollution. Urban, agricultural and industrial systems interact with their immediate environment andmay involve the release o f treated or untreated effluents, fertilizers and other agricultural chemicals into water bodies. All these substances may potentially alter the quality o f natural water making it less suitable or, indeed, unsuitable for consumption or even for recreationaluse. Discharges containing organic matter and especially municipal wastewater near large urbanareas may cause deterioration inriver water quality sufficient to make existing water treatment plantsinoperative. Apart fiom anthropogenic contamination o fwater, naturally occurring problems, i.e., ground water contamination with highlevels o f arsenic affect a large section o f the populationin Bangladesh. 28. The incidence o fwater related mortality andmorbidity depends ina complex way on a range o fparameters andtheir interactions, including water quality, water quantity, sanitation, and the status o f community hygiene practice, among others. The health impact o f these parameters and their various interactions are not yet clearly understood andthus manyassumptions areneeded inorder to estimate the effect o f any specific parameter andits potential cost implications. This necessarily gives rise to uncertainties inthe calculations andinmost cases nouniquevalue canbe arrived at. Consequently, it i s more appropriate to present a range o fplausible values under various assumptions. 29. The primary epidemiological data on morbidity and mortality for water and sanitation related diseases are analyzed and reportedinterms o f DALYs. Fromthe most recent data available, about 3.62 millions o f DALYs are lost each year inBangladesh (annex-3) due to water and sanitationrelated diseases. The challenge i s to quantify the achievable reduction inDALYsdue to provision o f access to clean drinkingwater and to compute the savings inhealth costs that canbe achieved. 4.2 Methodology 30. The same methodology reportedby Brandon(2)has beenused here, which was itselfbased on previous work within the World Bank group. The details o f the methodology are given inannex-111. The methodology uses a stepwise procedureto calculate the savings inthe DALYs due to improved access to clean water and sanitation. These savings are calculated as a function o fparameters which define plausible limits o f interventions from the composite value for DALYs obtained from field data through epidemiological surveillance with current level o f access to clean water and sanitation. Available data andthe analytical framework do not allow unique apportionment o f DALYssaved due to access to clean drinkingwater. Two assumptions are neededto . estimate the limits o f such savings. One o fthese relates to the relative importance o f water and sanitationfor adequate safety from waterborne diseases and the other i s a hygiene factor which defines the level o fhygiene among two groups o fpeople with and without adequate access to water and sanitation services. O fnecessity, these two variables have to be considered as parameters and limits are defined from qualitative consideration o fplausibility. As a result, we obtain multiple scenarios for health impacts andrelated costs. The lowest limit o fthe range probably gives the indication o fthe benefits from access to clean drinkingwater. 11 4.3 SummaryofResults 31. The results o fthe calculations are summarized intable-11. Total DALYs between 0.82 -1.94 million canbe saved through the provision o f cleanwater and sanitation. The economic value o f water-related mortality andmorbidity canbe estimated byusing the humancapital approach. This impliesthat the statistical value o fone DALY is equal to the annual average productivity o f a worker inBangladesh. This may be set equal to per capita GNI for Bangladesh. Higher estimates for the DALYs are obtained ifGNI-scaled USEPA annualized value o f life are used. This value i s 2.73 times higher thanthe GNI as reportedby Brandon. The cost savings involved inboth the approaches are shown in the tables above. The savings are between US$313-739 million inthe humancapital approach method. InGNIterms, these correspond to 0.63-1.48 % per year. Comparedto Bandon's study (1997) the water-related health impacts interms o f DALYs have decreasedto some extent but the costs still remain similar. This reflects the impact o f increase inthe per capita GNI. Reduced access to clean water as a result o f Arsenic contamination has not been considered. 32. Some discussion o fthe numbers obtained for DALYs saved inthese calculations i s necessary, as it i s not clear what quantities impact these savings. The impacts are not due to water pollution per se but a combination o f access to water, sanitation andtheir interactions. However, through parameterizationo fthe relative importance o f water and sanitation for adequate safety from waterborne diseases (x) and the hygiene factor which defines the level o fhygiene among two groups o fpeople with andwithout adequate access to water and sanitation services (H), the impacts range from access to cleanwater to the maximumlevel o fbenefits obtainable from access to clean water and sanitation. The lowest value obtainedprobably corresponds to access to cleanwater. The range o f values obtained i s rather large, refinements inmethodology since Brandon's work may not reduce the uncertainties to any great extent. The same methodology as Brandon's original paper has been followed as this allows a comparison for the situations then and now. Value o f Reduced DALYs DALYs Reduced (Million $) Saving inDALYs (Millions) This work (Variable values) Present Human USEPA Brandon work Brandon Capital AVOL Lowest Savings Range (X=O.9, H=0.7) 0.82 1.31 313 856 289-788 Highest Savings X=0.5. H=0.5 1.94 2.3 739 2017 506-1.380 Average Savings o f 15 Scenarios 1.39 1.82 528 1441 401-1,093 12 5. Discussion and Conclusions 33. The three environmental factors considered here account for about 1.2-3.35 % o f GNIinterms o fhealthimpact. The values o fdiseaseburden obtained inthe present calculations are rather conservative estimates compared with the regional figures available inliterature(') for the Asia region inTable 12. It maybe noted that the calculations reported here are based on the economic indicators for the base year 2002. These indicators change with time (e.g., GNI), so absolute cost figures are likely to vary substantially with time. However, the costs interms percentage o f GNI are likely to be fairly stable inthe short and even inthe medium term. Table 12: Environmental Factors inthe Burden of Disease in Bangladesh and Asia Total 18.5 18.5 17 - 15.7 Environmental 34. Itshould againbe emphasized that there are a lot o funcertainties inthe economic valuation o f environmental impacts as discussed earlier andthe figures here shouldbe considered as indicative only. The costs calculated here are not the maximumhealth costs imposed by air andwater pollution. Inthe case o f ambient air pollution, the lower limit correspond to health cost savings for 20% reduction inpollution level andthe higher limit corresponds to reduction o fthe pollution level to proposednational standards. Infact, health cost savings due to air pollution do not stop at the standardsandbenefits can be derived byreducing the pollution levels below the standards. Inthe case o f health cost savings from improved water supply and sanitation, the average o f 15 scenarios has been taken inview o f the complexities introduced by coupling o f water and sanitation impacts andthe hygiene factor. 13 35. For comparison, the monetary value for water and sanitation inBangladeshhas been normalized to IndianDALY percentage. With such normalization all the different units are converted to same units (Le,, percentage DALYs). Inview o f the uncertainties in the estimates, the level o f the burden o f disease due to environmental causes canbe considered as comparable inthe Asia region. As such, similar policy measures for the remediation o fpollution maybe applicable regionally with customization to suit the local needs. 14 References 1. World Bank2000, Healthand Environment. Backgroundpaper for the World Bank Environment Strategy, Washington, DC. 2. Carter Brandon, Economic Valuation o fAir and Water Pollution inBangladesh, World Bank Report (1997) 3. Monthly Report by AQMP (2002-04), DOE, GOB 4. Investigationof Sources o f Atmospheric ParticulateMatter (APM) at anUrban area inBangladesh, S.K. Biswas, S.A. Tarafdar, A. Islam and M.Khaliquzzaman, Report, AECD/CH/55 (2001) 5. S. K.Biswas, S.A. Tarafdar, A. Islam, M.Khaliquzzaman, H. Tervahattu, K. Kupiainerl, Impact o f Unleaded Gasoline Introductionon the Concentration o f LeadinDhaka Air. - Journal o f Air & Waste management Association, 53 (2003)1355-1362. 6. R. Kaiser, A.K. Henderson, W. R Daley, M.Naughton, M.H.K.Khan, M, Rahman, S. Kieszak, C. H.Rubin, .BloodLead Levels o fPrimary School Children inDhaka, Bangladesh. Environmental Health Perspectives. 2001;109(6): 563-566. 7. M.Khalisuzzaman, M.Billah, M.A. Rab, S. M.A. Bari, M.Kojima, P. Martin? and J. Shah, Environment Policy Implementation: Case OfTwo Stroke Three Wheeler Ban InDhaka, Bangladesh, Presented at the Better Air Quality Workshop (2003),Manila, Philippines (December, 17-19,2003). 8. M.A. Desai, S. Mehta ,K.R. Smith. Indoor smoke from solid fuels: Assessing the environmental burden of diseaseat national and local levels. Geneva, World Health Organization, 2004 (WHO Environmental Burden o f Disease Series, No. 4. 9. S. Dasgupta, M.Huq,M.Khaliquzzaman, K.Pandey, D.Wheeler, Indoor Air Quality for Poor Families: New Evidence from Bangladesh, Policy Research Working Paper- 3393(2004), The World Bank Development Research Group, Washington. 10. S. Dasgupta, M.Huq,M.Khaliquzzaman, K.Pandey, D.Wheeler, Who Suffers from Indoor Air Pollution: Evidence from Bangladesh, PolicyResearch Working Paper- 3428(2004), The World Bank Development Research Group, Washington. 11.K.Lvovsky, G. Hughes, D.Maddison, B. Ostro, D.Pearce, Environmental Cost o fFossil Fuels: A Rapid Assessment Method with Application to Six Cities, Environment Department Paper No. 78 (2000), The World Bank, Washington. 12. M.Cropper, N.Simon, A. Alberini ,and P. K. Sharma, 1997, The Health Effects ofAir Pollution inDelhi, India, PolicyResearch Working Paper 1860, DECRG, World Bank, Washington D.C. 13. Statistical Yearbook, 2001, Pub:BBS, Dhaka(2003), ISBN-984-508-507-5 14. M.F.Ahmed, Alternative Water supply Options for Arsenic Affected Areas o f Bangladesh, Int.Workshop on Arsenic MitigationinBangladesh, Dhaka, January 14-16,2002, 15 ANNEX-I METHODOLOGY AND CALCULATION OF HEALTH COSTS OF AMBIENT AIR POLLUTIONIN BANGLADESH 1. Introduction The methods for calculating the health effects o f air pollution are quite well developed and are routinely used. With appropriate scaling these methods can be used for Bangladesh with a reasonable degree o f confidence. The present calculations are based on the methodology in the World Bank study by Lvovsky et a1 (2000)'". The models used in this work can be scaled for population and income levels for applicationinBangladesh, as was done inthe work for six major cities insix different developing countries. Only the health effects o f PMlo have been considered here as data are available only for this component o f air pollution inBangladesh. The health impacts shown here are, therefore, underestimates and actual health burden o f the total effects o f air pollution are likely to be higher. This aspect and other issues are taken up inmore details inthe latter sections. 2. Estimatefor HealthImpacts Healthimpact are calculated from the relation AHi=b,, AAj P .......................................... 'J* * (1) Where: A stands for change; Hii s the health impact o ftype iper year; bij stands for the slope o f the dose response function o fhealth effect o ftype ifor exposure to pollutantj per year; Aj i s the ambient concentration o fpollutant j and P i s the population exposed to the pollutants. The quantityAAj i s defined as AAj =max [ 0, Aj1-max( Aj0 ,Sj)] ....................................... (2) Where: A,' i s the observed concentration; Ajoi s the background concentration and Sj i s the relevant threshold or air quality standard. The coefficients bij are available inliterature (1,2)from healthresearch work and are given intable1. The changes inthe ambient concentrations are obtained from measurements. 1 Table 1:Air pollution dose response function slope (bij ) per pg/m3change inthe mean annual level 1. Mortality Percentage change 0.046*, 0.084 2. Chronic Bronchitis Per 100,000 adults 6.12 3. Respiratory Hospital Per 100,000 population 1.2 admission 4. Asthma attack Per 100.000 asthmatics 3.260 - 3. Methodology for the Estimation of cost of health effects The values for the cost o fhealth effects including VOSL (Value o f Statistical Life) havebeen obtained inU S through extensive research. These values are related to economic indicators andthese have to be scaled byper capita GNI (Gross national Income) for variation intime and space. Ithas been shown that Country specific variation can be scaled by using the relation('): Where: V i s the valuation parameter for given country k and Y i s the per capita GNI. Various value o fr inthe range o f 0.4 to 1.2 havebeen reported but it has been observed that r=1provides conservative estimates('). With this approximation, the relation for Bangladesh (indicated by subscript B) becomes linearizedto: VB = vu, * (YB / Yu,)....................................................... (4) This simplified relation is usedinthe calculations here and are given intable 2. The VOSL value obtained i s considered highand so i s the value obtained inthe case for chronic bronchitis. 2 Table 2: WTP based health effect costs per case inBangladesh Obtainedbyper capita GNIscaling from values inUS(GNI(1990)=$21,790). (Reference-1: Table 4.4.). (Bangladeshper capita GNI (2002) (3):$380. (1$=Tk58)) s1 Health Effects WTP BD(2002) BD(2002) in US(1990) US$ Taka 1. Mortality (VOSL) 1,620,000 28,251 1,638,587 2. Chronic Bronchitis 195,000 3,40 1 197,23 7 3. Respiratory Hospital 4,225 73.68 4. Some data and assumptions inthe health impact and cost estimates The following datahavebeenusedinobtaining the numbersintable-2. The percentageof asthmaticsinthe Bangladeshpopulation hasbeentaken from unpublishedinformal literature andthus can only be considered as a contingent estimate. i. CrudeMortalityRateper1000is4.8 (4) ii. Populationunder15years38.2% (5) iii. Adultpopulation(61.8%f5) iv. PercentageofAsthmatic population is assumedto be 5%. 5. Urbanization and city population There are 3 cities inBangladeshwherepopulation exceedsone million andthere are 18 more cities where populationexceed 100,000. The population statistics for these cities are shown intable 3 & 4. 3 Table 3: CitiesinBangladeshwith more than one million p~pulation'~). Per capita income has been assumed to same as GNIwhich i s taken as US$ 380. Population (2003) SI. City inMillion Per capita Income($) 1 Dhaka 13.07 380. 2 Chittagong 3.70 380. 3 Khulna 1.43 380. Total 18.20 Table 4: CitiesinBangladeshwith more than 100,000 population(5! Per capita income has been assumed to same as GNIwhich i s taken as US$380. 7 Nawabganj 0.171 380. 8 Bogra 0.174 380. 9 Comilla 0.167 380. 6. Air QualityData For the first group o f three cities, fairly comprehensive air quality data are being collected only for Dhaka bythe Air Quality Management Project o fthe Department o f Environment (AQMPf6). The P M data (both PMlo and PM2.J are also collected by the Bangladesh Atomic Energy Commission (BAECf7'. The summary o f air quality data for Dhaka obtained at the Continuous Air Monitoring Station (CAMS) o fAQMP are shown intable-5. 4 Table 5: Average values for Criteria Pollutants Measured at CAMS, Dhaka with standard deviations during2003 along with proposed revised Bangladesh Standards(6). Itcanbe seenfrom table-5 that the mainpollutant o fconcern inDhakais particulate matter. BothPMlo and PM2.5levels are extremely highbeing much above the proposed standard.The NO2 levels are also now close to the standardandmaybecome a concern in the future. The level o f other pollutants is still low and thus not important from a health point o fview. One o f the criteria pollutant i.e., Pb (Lead) i s not shown inthe table. The lead level i s now sufficiently low (i.e., around 100ng/m3), so that air borne lead i s no longer considered as a health issue (8).However, blood leadlevels inchildren are still highwhich come from other sources('). So, the healthimpact calculations willbelimited to PMlo only. There i s no air quality data for other two cities (Chittagong and Khulna) with more thanone millionpopulation. For impact calculation, the pollution level in these cities i s assumed to be same as Dhaka. Inthese cities also, there are highlevels o f public complaints about the air quality. The banningo ftwo stroke three wheelers in Dhaka has contributed to the improvement inthe air quality inDhaka but the influx o f these vehicle has made the air quality worse inChittagongand Khulna. So, it is a plausible assumption that the decrease inthe pollution levels inDhaka and increase inthe levels inChittagong and Khulna have made the levels comparable inall the three cities. For one o f the 18 cities i.e., Rajshahi data on P M levels are available from BAEC measurement^(^) which are given intable 6. The yearly average PMlo level is reportedto be 63k25 pg/m3.This i s the ambient level which has been used inthe case o f all the 18 cities with population more than 100,000. 5 Table 6: PMlo data for Dhaka and Rajshahi ('Iin pg/m3 PMlO level in PMlO level in Comments Dhaka Rajshahi Month (2003) January 271 89 Monthly data for Rajshahi generated February 244 112 from measurements during2001-03 March 202 88 bymonthwise averaging as data are April 104 58 not available for all the months inany May 97 58 given year June 68 36 I Julv I 46 I 37 ' I August September October November 183 I December I 173 I 76 I Average 133 63 Stdev 78 25 7. HealthImpact and cost calculation for Ambient Air Pollution An important aspect determiningthe healthrelevance o fair quality is the exposure patterns. These vary significantly among the population but are usually consistent intime for a given population group. Exposure can be defined as the event inwhich an individual remains incontact with a specific concentration o f a pollutant for a given period o f time. Exposure assessment consists o f describing and quantifying the relevant conditions and characteristics o fhuman exposure. Air quality data have beenmeasured as 24 hour average mass-based concentrations o f PMlo at a suitable location inDhaka usingfixed monitoring equipment. These measurements were done for regulatory purposes and such data are usually insufficient for personal exposure assessment, as information about diversity interms o f time spent at different places are not captured insuch measurements. However, inthe present case we shall take yearly average o f PMlo level as the measure o f exposure. A significant fraction (probably as highas 20%) o fthe city population live and work near the traffic canyons where the pollution level maybe higher bya factor o f2 or even more. So, the present calculationsmaybe considered conservative. The results o fthe calculations are shown intables 7 and 8. The number o f cases o f mortality and morbiditythat can be avoided ifPMlo pollution level i s reduced ina range from a modest 20% o f the current level up to the proposednational standard are shown in this table 7. Itcanbeseenthat about 1,200-3,500 deaths and about 80- 235 millioncases o fmorbidity canbe avoidedper year inthe two scenarios (table-7). The costs involved range from about US$169 to US$492 million as shown intable-8. 6 Table-7 Number of case reduction for mortality and morbidity per year for two different scenarios 6 Restricted day activities 15.8 49.2 1.5 1.6 17.3 50.8 7 Lower respiratory illness 0.2 0.7 0.0 0.0 0.3 0.8 8 Respiratory symptoms 42.0 131.1 4.1 4.2 46.1 135.3 Sub-Total(M0rbidity) 121.8 380.1 11.8 12.2 133.6 392.3 Total 152.0 474.4 17.2 17.7 169.2 492.1 8. Discussion As pointed out above the estimates interms ofhealth impacts (mortality andmorbidity) are underestimates inthe current calculation because only one pollutant impact has been considered and elevated exposures at hot spots have not beentaken into account. However, inthe case of cost estimates this may not necessarily so. As discussed inmore detail inthe case o f indoor air pollution (annex-2), the costs obtained per case for chronic bronchitis inthe GNI scaling method used here i s much higher compared to contingent market based estimate( treatment cost+ income loss). The cost of mortality using GNI scaled VOSL are higher compared to human capital approach which i s also discussed in 7 annex-2. Hence, the cost estimates reportedhere are still probably underestimatesbut may not be grossly so. References 1. K.Lvovsky, G. Hughes, D.Maddison, B.Ostro, D.Pearce, Environmental Cost of Fossil Fuels:A Rapid Assessment Methodwith Application to Six Cities, Environment Department PaperNo. 78 (2000), The World Bank, Washington. 2. M.Cropper, N.Simon,A. Alberini ,andP. K.Sharma, 1997, The HealthEffects of Air PollutioninDelhi, India,PolicyResearchWorkingPaper 1860, DECRG, World Bank, WashingtonD.C. 3. World Bank, Bangladeshat a glance (8/29/03) 4. Statistical Yearbook, 2001, Pub:BBS, Dhaka(2003), ISBN-984-508-507-5 5. NationalReport (Provisional), Population Census (2001), Pub: BBS, Dhaka (2003) 6. Monthly Reportby AQMP (2002-04), DOE, GOB 7. Private Communications (2003): P M data from BAEC/ESMAP monitoring. 8. S.K.Biswas, S.A. Tarafdar, A. Islam, M.Khaliquzzaman, H.Tervahattu, K. Kupiainen, Impact of Unleaded Gasoline Introduction on the Concentration o f Lead inDhakaAir.- Journal ofAir & WastemanagementAssociation, 53 (2003)1355- 1362. 9. R.Kaiser, A.K. Henderson, W. RDaley, M.Naughton, M.H.K.Khan, M,Rahman, S. Kieszak, C. H.Rubin, Blood Lead Levels of Primary School Children inDhaka, . Bangladesh. Environmental HealthPerspectives. 109(6): (2001) 563-566. 8 Annex-I1 Health Burden due to IAP inBangladesh 1.Introduction Different methods to estimate the burden o f disease from indoor air pollution in developing countries have been reviewed ina recent WHO (2004) publication"'. The more widely used approaches are the fuel-based and the pollutant-based methods. Inthe fuel-based approach some fraction o fthe burden due to selected diseases are attributed to indoor air pollution usinga step-wise procedure. The contribution o f all relevant diseases i s summed to obtain the total burden o f diseases inDALYs and number o f deaths. By contrast, inthe pollutant-based approach, the population exposures to an indicator pollutant, generally PMlo, i s estimated interms o f some measure o f concentration-time. The exposure-response relationship for the indicator pollutant i s then usedto determine excess morbidity andmortality. Boththe methods have advantages and disadvantages andthese usually leadto significantly different results. The WHO studyrecommends the fuel-based approach, however, boththe approaches are usedhere for comparison and to illustrate the uncertainties involved inthe calculations. 2. HealthBurden usingFuel-Based Methodology 2.1 The Methodology The stepwise procedure prescribed inWHO document(') are usedhere inorder to arrive at the estimate for healthburden due to IAP inBangladesh. The steps involved are the following: (vi) Collection o f data relevant to calculation; (vii) Calculation o f attribute functions; (viii) Calculation o fthe attributableburden; (ix) Summation o f all the burdens due to individual diseases; and (x) Estimation o funcertainty. 2.2 Data relevant to Bangladesh 2.2.1 Diseases relevant to IAP s1 Relative Confidence Evidence Healthoutcome Group Risk Interval 1. Strong ALRI Child <5y 2.3 1.9-2.7 2. COPD Women 230v 3.2 2.3-4.8 4. Moderate-I COPD Men 230y 1.8 1.O-3.2 5. Lungcancer(Coa1 Smoke) Men230y 1.5 1.O-3.2 1 Abbreviations: ALRI= Acute lower respiratory infection, COPD= Chronic Obstructive pulmonary disease. 2.2.2 HouseholdSolid FuelUse inBangladesh Fuel Type Household% Comments SFU 96 Probably highestimate Biomass 96 Coal 0 Fraction ofthe population exposed(F) to air pollutiondue to SFU canbe found usingthe following equation: F= P, / P = (H*Vl + (1-H)"V'). .....................................................( 1) Where: F= Fraction ofthe population exposed P =Population ofthe country Pe=Population exposed H=PercentageofhouseholdusingSF V1=Ventilation coefficient for householdusingSFU (Vl=l) V2=Ventilation coefficient for householdusingimprovedstove or outside cooking (Vf =0.25) As improvedcook stoves are rarelyusedinBangladesh andno reliable statistics are available, all the householdswill be assumedto be exposedwhere SF i s used(i.e., F=0.96). The uncertainty i s taken to be 5% as suggested inref-1(table-5.1 inAnnex-5). 2.2.3 BangladeshPopulation Thepopulation distribution data available fiom 2001 census(') havebeennormalized to projected total for 2004. The data are givenintable-3. Age Population inMillion Population Fraction (Years) Male Female All Male Female All 0-4 8.65 8.15 16.80 0.06 0.06 0.12 5-14 17.43 16.72 34.15 0.13 0.12 0.25 >15 21.67 20.82 42.49 0.16 0.15 0.31 >3 0 23.06 21.45 44.5 1 0.17 0.16 0.32 Total 70.81 67.14 137.95 0.51 0.49 1.oo 2 2.2.4 Burdenof DiseaseinBangladesh Data specific to Bangladesh for the burdeno f diseases (Bi) relevant to indoor air pollution (shown intable-1) are not available from national health surveillance sources. So, followingthe example inWHO report (l), data have been obtained from W H O the SEARDregiondata. This regiono fWHO encompasseseight countries including Bangladesh. Bangladesh population i s 11.03% o fthe region, so the same percentage o f burdeno f diseasehas been assumed for the country as for the region, and are shown in table-4. The numbers inrow 3 and 5 for men2 30 years' age are due to coal smoke. Hence, these are set to zero inthe subsequent steps as coal smoke i s not important in Bangladesh. s1 DALYs Deaths Disease, sex, age group Lost (000) (000) Comments 1. ALRK5y-s 2,389 67 2. COPD. Women230m 251 14 3. As the numbers are due to coal smoke, this COPD, men23Oyrs 255 14 willbe set to zero during calculations. 4. Lungcancer, Women23Oy 28 2 5. Lungcancer, men23Oy 103 10 Y Y All causes 3.027 108 2.3 The Attributable fraction of DALYs and Deathto SFU The attributable fraction (Ai) o f a given disease (i) to SFU can be calculated using the equation (ref-1): Ai= ((F* Ri (l-F))-l)/ (F" Ri+ (1-F)). ... . . .. . + . . . . . ..(2) Where: F= Fraction o fthe populationexposed as explainedinthe context o f equation-1 Ri=Relativerisk or odds ratio for a diseaseias given intable-1 for specific diseases. The calculated attributable fractions for different diseases inBangladesh are shown in table-5. Table-5 Attributable fraction from SFUinBangladesh s1. Attributable fraction Comments Disease, sex, age group Low Central High 1. ALRI<5yrs 0.46 0.56 0.62 2. COPD, Women230yrs 0.56 0.68 0.78 3. COPD, men23Ows 0 0 0 Cole smoke effect-set to zero 4. Lungcancer, WomeQ3Oy 0 0.32 0.51 5. Lungcancer, men>=3Oy 0 0 0 Cole smoke effect-set to zero 3 2.4 The Attributableburdenof DALYs andDeathto S F U The attributable burden o f diseases due to SFU can be found bymultiplying burdeno f diseases intable-4 by the attributable fractions intable-5. The attributable diseaseburden (Di) for a given disease can be written as (ref-1): Di=Ai* Bi ............,............................................................... (3) The values obtained for different diseases are shown intable-6 below. Table-6 AttributableburdensfromSFUfor Bangladesh DALYs (000) Deaths (000) SI Disease, sex, age group low central High low central high 1. ALRI