Household energy and health: where next for research and practice?

Cooking and home heating with coal and biomass fuels (wood, crop residues, animal dung, and charcoal) are ideal subjects for well intentioned epidemiology. Cooking and heating with such fuels generate large amounts of pollutants that can harm people’s health throughout the lifecourse, a risk that largely aff ects poor communities. In a simple world, epidemiology would investigate the hazardous eff ects and test the benefi ts of any interventions, and rational individuals and policy bodies would use this information to initiate positive change. The world, however, is not simple when we study something as central to daily life as household energy. In the 1970s, an Australian respiratory epidemiologist studying adult lung disease in Papua New Guinea documented the positive association between domestic woodsmoke and children’s respiratory infections. Subsequent studies documented the hazardous role of smoke from biomass and coal in the development of childhood pneumonia and other adverse clinical outcomes. Using this early evidence, the Comparative Risk Assessment Study attributed 1·6 million annual deaths to biomass and coal use in the early 2000s (attributable deaths [with inclusion of other outcomes] have since been estimated at about 3·5 million). In the late 1990s and early 2000s, two directions were advocated for epidemiological research on so-called household air pollution to help develop appropriate public health and policy responses: observational research with measurement of personal exposure to better characterise the exposure–response relationship, which would then be used to determine pollution reductions needed to achieve health benefi ts; and randomised trials free of confounding to measure the pure intervention eff ects. Over the past two decades, neither type of research has been as informative as hoped. Exposure–response studies have been limited by the diffi culties in measuring personal exposure to pollutants. Trials have so far not implemented interventions that substantially reduce exposure while functionally replacing the traditional biomass and coal stoves, and are scalable in a community setting. In The Lancet, Kevin Mortimer and colleagues report the Cooking and Pneumonia Study (CAPS) cluster randomised controlled trial done in two rural districts of Malawi. CAPS tested an alternative biomass stove, comparing it with existing cooking methods (typically open fi res). Each household in the intervention group received two stoves (both Philips HD4012LS), a solar panel, and user training, while the control group continued to use their existing cooking method. New stoves were repaired and replaced as needed, with 13 192 repairs or replacements for stoves (3·1 per intervention household) and 5259 (1·2 per intervention household) for solar panels. By the second year of the follow-up, the subset of stoves that were objectively monitored were used for only 0·34 cooking events per day. The primary outcome was WHO Integrated Management of Childhood Illness (IMCI)-defi ned pneumonia episodes diagnosed through routine visits to local health facilities. The stove intervention had no eff ect on the primary outcome (the facility-diagnosed IMCI pneumonia incidence rate in the intervention group was 15·76 [95% CI 14·89–16·63] per 100 child-years and in the control group 15·58 [14·72–16·45] per 100 child-years; intervention vs control group incidence rate ratio [IRR] of 1·01 [0·91–1·13]; p=0·80). There was a borderline signifi cant increase in the risk of severe pneumonia in the intervention group (intervention vs control group IRR for severe pneumonia episodes was 1·30 [95% CI 0·99–1·71]; p=0·06). The strength of CAPS is its large sample size with 10 750 children from 8626 households across 150 clusters enrolled, and 10 543 children from 8470 households contributing 15 991 child-years of follow-up data to the intention-to-treat analysis. CAPS also has disadvantages, such as a reliance on health facilities for identifying pneumonia cases instead of active case fi nding used in previous studies, and not reporting information about impacts on home concentrations and personal exposure to pollutants, and on pathogen-specifi c pneumonia, both of which were presented in the earlier RESPIRE trial. This information is needed to understand the reasons for null eff ect and to inform intervention choices (ie, no or insuffi cient reduction in pollution or personal exposure vs absence of an aetiological relationship between exposure and pneumonia). Published Online December 6, 2016 http://dx.doi.org/10.1016/ S0140-6736(16)32506-5