Health Policy and Planning Advance Access originally published online on October 9, 2006
Health Policy and Planning 2006 21(6):444-458; doi:10.1093/heapol/czl027
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Who suffers from indoor air pollution? Evidence from Bangladesh
Development Research Group, World Bank, Washington DC, USA
Correspondence: Susmita Dasgupta, Development Research Group, World Bank, 1818 H Street, NW, Washington DC 20433, USA. E-mail: sdasgupta{at}worldbank.org
In this paper, we investigate individuals’ exposure to indoor air pollution. Using new survey data from Bangladesh, average hours spent by members of households in the cooking area, living area and outdoors in a typical day are combined with the estimates of pollution concentration in different locations in order to estimate exposure. We analyse exposure at two levels: differences within households attributable to family roles, and differences across households attributable to income and education. Within households, we relate individuals’ exposure to pollution in different locations during their daily round of activities. We find high levels of exposure for children and adolescents of both sexes, with particularly serious exposure for children under 5 years. Among prime-age adults, we find that men have half the exposure of women (whose exposure is similar to that of children and adolescents). We also find that elderly men have significantly lower exposure than elderly women. Across households, we draw on results from a previous paper, which relate pollution variation across households to choices of cooking fuel, cooking locations, construction materials and ventilation practices. We find that these choices are significantly affected by family income and adult education levels (particularly for women). Overall, we find that the poorest, least-educated households have twice the pollution levels of relatively high-income households with highly educated adults.
Our findings further suggest that young children and poorly educated women in poor households face pollution exposures that are four times those for men in higher income households organized by more highly educated women. Since infants and young children suffer the worst mortality and morbidity from indoor air pollution, in this paper we consider measures for reducing their exposure. Our recommendations for reducing the exposure of infants and young children are based on a few simple, robust findings. Hourly pollution levels in cooking and living areas are quite similar because cooking smoke diffuses rapidly and nearly completely into living areas. However, outdoor pollution is far lower. At present, young children are only outside for an average of 3 hours per day. For children in a typical household, pollution exposure can be halved by adopting two simple measures: increasing their outdoor time from 3 to 5 or 6 hours per day, and concentrating outdoor time during peak cooking periods.
Key Words: indoor air pollution, human exposure, household, Bangladesh
1The ‘Regions’ are defined as in the World Health Organization's Global and Regional Burden of Disease Report 2004, online at [http://www.who.int/publications/cra/en], and do not correspond to continent/region boundaries.
2Although we use the term ‘peri-urban’ to describe areas proximate to Dhaka, our sample includes many rural farm households.
3Our stratification had been designed for cell values large enough to test fuel and ventilation effects on IAP, and was not intended to represent all Bangladeshi households.
4While the MiniVol is not a reference method sampler, it produces results that closely approximate data from US Federal Reference Method samplers.
5In each district, uniformly rich/poor neighbourhoods in urban and peri-urban areas were left out after discussion with the local experts. One neighbourhood was selected randomly from the heterogeneous neighbourhoods. Interviewers then approached each consecutive house, starting from one end, until 25 surveys were completed.
6In each district, on two consecutive days, the survey team travelled exactly the same number of kilometres away from the main town boundary in an East, West/North or South direction (direction was decided randomly). From the first village then encountered, starting from one end, each consecutive house was approached until 25 surveys were completed.
7Although this region is near Dhaka, many of the sampled households are in rural settings.
8PM10 concentration was monitored in 236 cooking areas and 244 living areas continuously, without intervention, for 24-hours, with MiniVol.
9Ambient readings were undertaken, with MiniVol, for 4 days per location.
10The pDR-1000 monitors need calibration, being secondary instruments. Calibration is usually done by colocated experiments. In our case, we conducted colocated experiments with MiniVols. The calibration factors used were as follows:
- PDRAM Unit – 05217: average (165 ug/m3) and Minivol reading (259 ug/m3), Calibration factor = 1.57
- PDRAM Unit – 5190: average (170 ug/m3) and Minivol reading (259 ug/m3), Calibration factor = 1.52
Zeroing of the instruments was done by putting the pDR-1000s in the zero air pouch and adjusting the reading to zero.
11Of course, households differ significantly in the timing of daily peaks; some have three rather than two, and there are also significant differences in peak levels and change gradients. Figure 2 represents the central tendency in the observed patterns for 27 households.
12This cycle is clearly different from the cycle in Figure 2, for two apparent reasons. First, the outdoor cycle reflects the combined effect of fuel burning in many households, whose daily cycles reach peaks at different times, and at different PM10 emissions intensities. However, this does not explain the large difference in the relative size of the two outdoor peaks, as compared with rough parity for the indoor peaks. Since the outdoor pattern is reproduced across several widely separated monitoring points, we hypothesize that different atmospheric conditions lead to more pronounced accumulation and duration of suspended particulates in the evening. Alternatively, preparation of evening meals may cluster more tightly in time, leading to more concentrated loading of PM10 in the evening. Future research may shed more light on this phenomenon.
13Source: Central Pollution Control Board, Delhi, undated. National Ambient Air Quality Standards, online at [http://www.cpcb.nic.in/as.htm].
14With mean values of cooking and living area concentrations from continuous 24-hour monitoring of 27 households in Narshingdi.
15Studies of short exposures to outdoor particulate concentrations suggest some impact on heart rate variability and the rate of heart attacks. However, a recent study in Palm Springs, California suggests that the short-period effect disappears when 24-hour average exposure is controlled for. Similarly, average exposures seem to dominate day-to-day variations in daily time-series studies. Our thanks to Dr Bart Ostro, Chief, California Office of Environmental Health Hazard Assessment, for his insights.
16We recognize that our illustration achieves such large reductions by assuming that infants are outside during the least desirable time in the status quo situation (i.e. the mid-afternoon period when indoor pollution is also relatively low). However, we believe that our essential point is generally valid – optimally chosen outside time for infants has the potential for considerable reduction of health damage.
17As we note in Dasgupta et al. (2004), average PM10 concentrations are smaller in our study.
18The correlation coefficient was 0.93 for median PM10 concentrations in cooking and living areas for households where indoor air was monitored with pDR-1000.
19High pollution indoors has been documented in rural households in Tamil Nadu (Balakrishnan et al. 2002; Parikh et al. 2003) and Andhra Pradesh, India (Balakrishnan et al. 2004b).
20Education, in the questionnaire, was a categorical variable: no schooling, less than or equal to 5 years of schooling, 6–10 years of schooling, 11–12 years of schooling, more than 12 years of schooling.
21The indices are constructed from the following equations: Fuel Choice = 258.563 – 45.23 3 * Jute – 106.729 * Kerosene – 112.523 * Lpg/Lng – 155.285 * Piped Natural Gas; Construction Materials/Cooking Location =258.563 + 253.896 * Mud Walls – 124.058 * Mud Walls, Detached Kitchen – 9.887 * Open-Air Kitchen – 37.599 * Detached Kitchen. The fuel choice index (mean 237.3, s.d. 48.2) and construction materials/cooking location index (mean 266.3, s.d. 87.4) are predicted indoor particulate concentrations, measured in micrograms of PM10 per cubic metre. The constant term in each equation is the average indoor concentration in sample households for all fuels and structural characteristics that are not explicitly controlled for in the regression analysis. The fuel choice equation incorporates the pollution-reducing effects of clean fuels (jute, kerosene, lpg/lng, piped natural gas). The construction materials/cooking location equation incorporates the pollution-increasing effect of mud-wall construction, with a downward adjustment for the mud-wall case when the kitchen is detached, along with pollution-reducing adjustments for open-air and detached kitchens. All explanatory variables in this regression are coded as 0–1 dummy variables.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  O. P. Kurmi, S. Semple, M. Steiner, G. D. Henderson, and J. G. Ayres Particulate Matter Exposure during Domestic Work in Nepal Ann. Hyg., June 3, 2008; (2008) men026v1. [Abstract] [Full Text] [PDF]
|
|