No. 13 KNOWLEDGE EXCHANGE SERIES March 2009 Improving Indoor Air in Rural Bangladesh: 48647 Results of Controlled Experiments by Susmita Dasgupta, Mainul Huq, M. Khaliquzzaman, and David Wheeler In rural Bangladesh, Indoor Air Pollution (IAP) is dangerously high for poor households dependent on biomass cooking fuels. Based on earlier World Bank research, controlled, scientifically monitored experiments were conducted in Burumdi village, Narayanganj District, to test the effects of structural arrangements and ventilation practices on IAP. Findings suggest optimal structural choices and village-level measures to reduce IAP exposure. Most of Bangladesh's rural households collect or pur- chase biomass--fuelwood, cow dung, crop residues, Traditional stove use in Burumdi village. tree leaves, and grass--to cook all or part of their studied building materials, housing configurations, meals, using traditional, fixed clay stoves. The high and construction techniques in various regions of the moisture content of biomass cooking fuels, combined country to develop a set of structural options. At each with the inefficiency of traditional stoves, results in site, local workers were hired to construct experimental incomplete combustion, producing indoor air pol- houses identical to the structures used by poor families lution (IAP) (Asaduzzaman, Barnes, and Khandker in the area, using standard local building practices. 2009).1 Recent World Bank research using the latest air-monitoring technology and results of a national Four sets of houses were built, corresponding to per- household survey show that, for respirable airborne manent walls made of thatch, mud, corrugated iron particulates (PM10),2 concentrations of 300 µg per m3 or (tin), and brick. The thatch-, mud-, and tin-walled greater are common, implying widespread exposure houses had mud flooring on rammed earth; while to a serious health hazard (Dasgupta et al. 2006).3 the brick-walled houses had conventional cement concrete flooring. Roofing materials were altered Study Rationale to produce a variety of standard combinations (Table 1). To date, most IAP studies have focused on the use The axes of the houses were aligned in a north-south of modern fuels, such as kerosene and liquefied pe- direction to capture varying wind conditions. Houses troleum gas (LPG), and improved stove designs. But were furnished to simulate life settings. structural arrangements--from building materials Within these diverse sets of cooking environments, and space configurations to cooking locations--and housewives cooked the standard midday meal for a ventilation practices are also a major issue. Testing the family of four on traditional stoves.5 Four common significance of these factors in reducing IAP exposure space configurations for kitchens were used: interior, is particularly important in rural Bangladesh, where attached, detached,6 and open air. For attached and many poor families are likely to depend on biomass detached kitchens, construction materials for walls, cooking fuels and stoves for some time.4 floors, and roofs varied.7 Kitchen configurations and Study Method Architects familiar with the climatic conditions and cul- 4 Burumdi village, located 27 km southeast of the capital city of Dhaka, was selected as the study site. Most of Burumdi's 1,600 resi- tural constraints faced by poor Bangladeshi households dents (about 290 households) are either self-employed workers in nonagricultural sectors or service providers. The nearest second- ary road and manufacturing industrial unit are each located 3 km 1 According to the World Health Organization, acute respiratory from the village; other potentially polluting manufacturing indus- infection from IAP kills an estimated 1 million children per year in tries are more than 10 km away. No villagers own motor vehicles developing countries (see WHO Global and Regional Burden of for private transport or motorized agricultural equipment. Disease Report, 2004 [www.who.int/publications/cra/en]). 5 The midday meal consists of rice, lentils, vegetable curry, and fish. 2 PM10 refers to particulate matter with a diameter of 10 microns. 6 A space enclosed by walls and a roof located a short distance 3 In eight Italian cities with far lower annual concentrations (45­55 g from the house. per m3), Galassi et al. (2000) found that reducing PM10 concentra- 7 Variations were thatch, mud, tin, and brick (walls); cement, con- tions had substantial health benefits. crete, and mud (floors); and thatch, tin, and concrete (roofs). ESMAP Knowledge Exchange Series No. 13 1 KNOWLEDGE EXCHANGE SERIES March 2009 Table 1. Summarized Distribution of Experiments Kitchen Types Wall material Roofing material No. of experiments House Interior Brick Concrete 37 Tin 63 Tin Tin 112 Thatch 0 Open Air Mud Tin 117 Thatch 0 Thatch Tin 89 Thatch 80 Kitchen Brick Concrete 16 Tin 34 Tin Tin 134 Thatch 33 Mud Tin 46 Thatch 34 Thatch Tin 99 Thatch 53 Type No. of experiments Kitchen Attached Detached Interior 59 Attached 165 Detached 225 Monitoring air quality of kitchen types using the Airmetrics Open air 49 Minivol Portable Air Sampler. Fuel Clean 54 (pDR-1000) sampler was used to record airborne PM10 Fuelwood 186 Cow dung 100 concentrations at regular 2-minute intervals over a 24- Other 158 hour cycle,11 while the Airmetrics MiniVol Portable Air Source: Dasgupta et al. (2009). Sampler measured 24-hour average PM10 concentra- tions.12 Readings using these two types of equipment cooking arrangements were limited to those com- provided a detailed record of IAP concentrations for monly observed in rural Bangladesh. The experiments each controlled experiment. As Table 2 shows, the allowed for use of a ceiling fan in the living space. IAP readings were generally lower than those of other Typical fuels--kerosene, LPG, fuelwood, and other studies in rural Bangladesh. These likely resulted from forms of biomass--were used in the experiment with the experimental design, which had varying types of a small amount of kerosene used for initial ignition.8 living conditions. Other sources of indoor pollution were eliminated.9 Ambient PM10 concentrations were monitored 76 Monitoring PM10 Concentrations times over the experiment period using the MiniVol In these various cooking environments, indoor air air sampler. Readings revealed wide inter- and intra- quality was monitored from April 2005 to June 2006.10 seasonal variations of PM10 in the outdoor environ- This period was selected to capture high- and low- ment (Table 3). dust seasonal variations. In the high-dust season (November­March), humidity is low and rainfall is rare. Data Analysis Conversely, in the low-dust season (April­June and Regression analysis was conducted to examine the October), pre- and post-monsoon rainfall is frequent. roles of basic IAP determinants: kitchen configura- tions, building materials, and fuels. These factors Because brief exposure to highly concentrated par- were varied under fixed experimental conditions, ticulate matter may have different health effects than sustained lower-level exposure, the controlled experi- 11 The pDR-1000 sampler uses a light-scattering photometer (neph- ments used both real-time and air-sampler monitoring elometer); the operative principle is realtime measurement of equipment. The Thermo Electric personal DataRam light scattered by aerosols, integrated over as wide a range of angles as possible. For details, visit www.thermo.com/eThermo/ 8 For example, cooking with cow dung/rice husk/jute inside the CMA/PDFs/Product/productPDF_18492.pdf. house is uncommon and thus was excluded. 12 The MiniVol was programmed to draw air at 5 liters per minute 9 Smoking, lighting of candles, burning of oil or kerosene lamps, through PM10 particle-size separators (impactors) and filters; the and burning of mosquito and insect repellents were not allowed. particles were caught on filters, which were weighed pre- and post- 10 With the exception of the monsoon period (July­September, exposure with a microbalance. For details, visit www.airmetrics. 2005). com/products/minivol/index.html. 2 ESMAP Knowledge Exchange Series No. 13 Table 2. Recordings of PM10 Concentrations Experimental data from the pDR- 3 PM10 (µg per m ) concentration 1000 samplers--pollution measured Season Monitor location Mean Median Minimum Maximum at 2-minute intervals over a 24-hour Airmetrics MiniVol Portable Air Sampler (24-hour average) cycle--provided useful information High-dust Kitchen 222 213 39 473 on the relationship between exposure High-dust Living room 161 155 45 320 Low-dust Kitchen 129 125 30 311 patterns, building materials, kitchen Low-dust Living room 69 64 22 184 configurations, and cooking fuels. Thermo Electric personal DataRAM (pDR-1000) (2-minute intervals) 95th percentile To assess kitchen exposure patterns High-dust Kitchen 585 404 8 88,900 during the period of maximum IAP High-dust Living room 468 370 1 382,400 exposure for each experiment, 150 Low-dust Kitchen 284 146 1 95,800 Low-dust Living room 843 132 1 194,600 regular-interval pD-RAM observations Source: Dasgupta et al. (2009). were drawn from 5 hours covering Note: Because the pDR-1000 samplers recorded shorter, more extreme exposures each time fuel was the midday meal preparation period. added, the resulting distributions had higher maximums, means (pulled upward by higher maximums), and standard deviations. MiniVols recorded flatter distributions for the 24-hour cycle. For each experiment, inferences about distribution patterns were drawn from Table 3. Recordings of Ambient PM10 Concentrations the mean, maximum, minimum, standard deviation, Using Airmetrics MiniVol Portable Air Sampler and median. Building materials did not significantly (24-hour average) affect exposure patterns. Attached kitchens promoted Ambient PM10 (µg per m3) more intense exposure, while detached configurations No. of concentration Season readings Mean Minimum Maximum promoted more sustained exposure. High-dust 41 171 82 274 Findings Highlights Low-dust 35 54 15 125 Seasonality is a key determinant of household-level with prescribed burn times for fuels, in both high- and IAP. In the high-dust season: low-dust seasons. Regression was conducted only for Outdoor air pollution significantly affects in- houses where PM10 concentrations were monitored door ambient pollution levels.14 concurrently in kitchens and living rooms. Interior kitchens have better air quality than de- In the high-dust season, the 24-hour average PM10 tached or open-air kitchens.15 concentration recorded in interior kitchens was Pollution from fuelwood, dung, and other bio- 187 µg per m3 lower than that of detached or open-air mass fuels is more severe. kitchens. Attached kitchens had PM10 concentrations Building materials significantly affect indoor 49 µg per m3 lower than that of detached or open- pollution; of the four materials studied, tin con- air kitchens. In the high-dust season, kitchen PM10 tributes the most to healthy air quality, followed concentrations were significantly higher for brick by thatch, mud, and brick.16 and mud than for thatch and tin. Both fuelwood and cow dung accounted for increments of 70 µg per m3 Use of a fan in the living room, which can be compared to cleaner-burning fuels; other biomass seriously affected by IAP from cooking smoke,17 fuels added an even greater increment (about 90 µg offers significant benefits.18 per m3).13 In the low-dust season, dung accounted for the greatest increment (109 µg per m3) over cleaner fuels, followed by fuelwood (62 µg per m3) and other 14 Regression analysis revealed a difference of 100 g per m3 be- biomass fuels (57 µg per m3). tween high-dust (155 g per m3) and low-dust (55 g per m3) sea- sons (Dasgupta et al. 2004). Diffusion of kitchen pollution into the living room was 15 This is not the case during the low-dust season; but, at that time, large during the high-dust season. All else being equal, it is difficult to cook outside. living-room pollution increased 0.5­0.6 µg per m3 for 16 In kitchens, tin or thatch walls are significantly less air-trapping each increase of 1 µg per m3 in kitchen pollution. With than mud walls, which, in turn, are less so than brick. With regard to kitchen roofs, tin offers better air quality than thatch. In living regard to building materials, inter-seasonal results for rooms, tin walls offer better air quality than mud ones, which, in living-room air were mixed. turn, are better than thatch or brick. 17 Thus, male household members' avoidance of cooking areas 13 Differences in performance between improved and traditional does not protect them from IAP. biomass stoves were not statistically significant, which may have 18 Where households have access to electricity, use of a ceiling or resulted from the small sample size of improved stove experiments. table fan is common practice. ESMAP Knowledge Exchange Series No. 13 3 Conclusion pollution as a health issue, and changing traditional The above findings suggest various structural ar- cooking patterns has often proven difficult around rangements and village-level measures that could the world. Effective public education is required to significantly mitigate IAP exposure for poor families convince village men and women that IAP poses a in rural Bangladesh. First, given that outdoor pollu- severe risk to themselves and their children, that the tion significantly affects indoor air quality for much of sources of IAP have been correctly identified, and that the year, simply venting cooking smoke to the outside their actions could significantly improve their health. would likely worsen indoor air quality for many vil- Villagers might accept centralized cooking arrange- lage households, who typically live in clusters. Alter- ments if such activities were organized and financed natively, cooking smoke could be ventilated through a effectively. Currently, the World Bank and the Gov- stack tall enough to disperse smoke over a broad area, ernment of Bangladesh are providing financial and thereby reducing particulate concentration in village technical assistance on a small scale via a collaborative households. Over the longer term, the solution is to pilot program. A follow-on impact evaluation could switch to cleaner-burning stoves and modern fuels; quantify the benefits associated with indoor air qual- to increase their affordability, villagers could make ity, IAP exposure, and health outcomes in a variety negotiated bulk purchases.19 Second, more permeable of rural household settings. construction materials, such as tin, could be used to References construct kitchens.20 Finally, where extended family Asaduzzaman, M., Douglas F. Barnes, and Shahidur R. members live in household clusters, women could Khandker. 2009. Restoring Balance: Bangladesh's Rural En- rotate cooking roles to reduce their exposure. ergy Realities. ESMAP Special Report 006/09. World Bank, Washington, DC. Are Bangladeshi villagers likely to adopt such collec- Dasgupta, S., M. Huq, M. Khaliquzzaman, K. Pandey, and D. tive innovations?21 Most residents do not recognize Wheeler. 2004. Who Suffers from Indoor Air Pollution?: Evi- dence from Bangladesh. World Bank Policy Research Working 19 Although not the focus of this work, high-quality, improved Paper 3428. World Bank, Washington, DC. stoves with better combustion might also alleviate some of the conditions of both indoor and outdoor air pollution. ------. 2006. "Indoor Air Quality for Poor Families: New Evi- 20 A nationwide household survey conducted by the World Bank dence from Bangladesh." Indoor Air 16(6): 426­44. highlighted the popularity of tin as a construction material. Of the ------. 2009. "Improving Indoor Air Quality for Poor Families: 1,550 households surveyed, 22 percent used tin to construct the A Controlled Experiment in Bangladesh." Indoor Air 19(1): walls of their living spaces, and 75 percent had tin roofs; 41 per- 22­32. cent of kitchens had tin walls or roofs. In flood-prone areas, tin walls and roofs could be dismantled and reused. Some families Galassi, C., B. Ostro, F. Forastiere, S. Cattani, M. Martuzzi, and R. noted that the heat-trapping property of tin walls and roofs kept Bertollini. 2000. "Exposure to PM10 in the Eight Major Ital- poisonous snakes away from attached and detached kitchens. ian Cities and Quantification of the Health Effects." Report 21 Community-based sanitation approaches have proven suc- presented to the International Society for Environmental cessful in Bangladesh and other developing countries. Epidemiology (ISEE), Buffalo, New York, August 19­23. The Energy Sector Management Assistance Program (ESMAP) Knowledge Exchange Series (KES) shares results of recent energy sector work to catalyze discussion and learning among energy policy makers and practitioners. The KES is produced by ESMAP, a global knowledge and techni- cal assistance partnership administered by the World Bank and sponsored by bilateral official donors since 1983. ESMAP's mission is to assist clients--low- and middle-income countries--to provide clean, reliable, and affordable energy services required for poverty reduction and environ- mentally sustainable economic development. About the authors: Susmita Dasgupta is Lead Environmental Economist, Research Department, World Bank. Mainul Huq is CEO, Development Policy Group, Bangladesh. M. Khaliquzzaman is a World Bank consultant, Bangladesh. David Wheeler is Senior Fellow, Center for Global Development, Washington, DC. Douglas F. Barnes, Technical Editor · Norma Adams, Editor · Marjorie K. Araya, Production Editor Photos: Susmita Dasgupta, p.1; Subrata Ghosh and Prabir Mallik, p. 2. Copyright © 2009 Printed on recycled paper. Visit us online at www.esmap.org