Endotoxin concentrations reported in this study are high and much higher than those found in a recent study measuring airborne endotoxin in 10 homes in northern California (Chen and Hildemann 2009
), where mean concentrations were generally < 1 EU/m3
, and in a study of homes of rural asthmatic children, where the GM inhalable endotoxin was 5.8 EU/m3
= 326) (Thorne and Duchaine 2007
). They were also considerably higher than those measured from a large study of the homes of 332 children in Canada (Dales et al. 2006
). The mean airborne endotoxin concentration in the Canadian study was 0.49 EU/m3
, almost 100 times less than the 24-hr average levels measured in this study for charcoal-burning homes and close to 1,000 times lower than the average level during cooking with dried dung in homes in Nepal. However, results from the Canadian study showed that even at the relatively low levels of exposure experienced by the Canadian study population, there was a statistically significant relationship between airborne endotoxin and respiratory illness in the first 2 years of life.
The only previous study of endotoxin concentrations in biomass-burning homes was carried out in two homes in the Ladakh region of India (Rosati et al. 2005
), where endotoxin levels of 24 and 190 EU/m3
were found, broadly in line with our data. The Indian homes were small, portable tentlike structures with little in the way of ventilation or extraction of smoke generated from burning dung and crop residues.
A health-based guidance limit of 50 EU/m3
has been recommended for occupational settings in the Netherlands (Heederik and Douwes 1997
) for an 8-hr time-weighted average exposure. The median value of 24-hr samples collected from charcoal-burning homes (using respirable dust size selection and hence conservative compared with the total inhalable dust sampler used for the limits proposed in the Netherlands) was approximately 20 EU/m3
. Scaling this to an 8-hr time-weighted average would produce levels of around 60 EU/m3
, exceeding the concentration deemed to be acceptable for a healthy workforce. From our results, we would anticipate much higher 8-hr time-weighted average values from wood- and dung-burning homes, and it seems likely that many of these would approach or exceed the health-based guidance limit value.
The health effects of exposure to the endotoxin concentrations measured in the homes in this study may be considerable, particularly because exposure is sustained and occurs from birth in most homes. Personal exposures of women who carry out cooking and fire lighting have the potential to be even higher than the static or area measurements made in this study because of regular close proximity to the smoke plume. There is a need for personal exposure data in these settings.
We acknowledge that this study has several important weaknesses. We did not design the study to collect samples for analysis of endotoxin, but rather “piggy-backed” it onto two studies that set out to characterize PM concentrations in homes in Malawi and Nepal. As a consequence, our results present data from both short cooking periods and longer 24-hr samples and also a mixture of total inhalable and respirable PM size selection. In addition, there was an extended period between the collection of the filters and analysis for endotoxin, and we believe that this led to the high levels of contamination of some of the field blanks that we have reported. This is particularly evident for the Malawi samples, which were stored for the longest duration. We report our data separately by size fraction, sampling duration, country, and fuel type and used appropriate methods for blank correction to overcome these weaknesses where possible.
Further work should use a standard protocol for endotoxin measurement and should seek to standardize durations of sample collection. Optimally, personal exposure measurements should be considered, especially in the context of health-related exposure measurement. Our study design collected only two samples from homes burning crop residues, and any future study should seek to address this data gap.
Controlling and reducing exposure to biomass fuel smoke in homes in the developing world are complex and difficult areas with such options as modifications of behavior, introduction of better and more efficient stoves, and improved household ventilation (Zhang and Smith 2007
). Methods of reducing airborne endotoxin concentrations will be broadly similar, but there may also be opportunities to reduce bacterial and endotoxin content of the source fuel via harvesting and/or production methods and changes to how fuel is stored. Higher cooking temperatures are likely to degrade endotoxin, and more efficient cooking using improved stove technologies can also reduce the generation of PM-bound endotoxin. A recent study has also suggested that outdoor storage of wood chips increased endotoxin content (Sebastian et al. 2006
), so dry, indoor storage areas for fuel may reduce the airborne endotoxin levels when burning eventually takes place.
Our study raises the possibility of an important new risk factor, and preventive strategies, for respiratory morbidity and mortality in the developing world. The mechanism for the association between biomass smoke exposure and infections of the lower respiratory tract in children remains unclear but is likely to be multifactorial and influenced by housing conditions, nutritional status, and other coexposures. It is possible that inhaled endotoxin, being proinflammatory, may be one contributory factor in this mechanistic pathway. Pneumonia remains one of the largest contributors to under-five mortality, and exposure to high concentrations of airborne endotoxin may be an important risk factor for the severity of illness (Dales et al. 2006
). From a public health perspective, interventions to reduce PM and endotoxin exposures generated from household combustion of solid fuels should be implemented as a matter of urgency.