Indoor smoke from cooking on an open fire has long been recognized as a major cause of ill health , especially for women and young children (those either most vulnerable or most likely to be exposed). Improved cooking stoves represent the hopes of development professionals in that their efficient design and vented smoke should improve health, lower mortality, and reduce fuel use. In the past few years a number of evaluations of improved cooking stove programs have been released. And the conclusions from reviewing these studies are… well it’s difficult to draw any conclusions.
Perhaps the two most widely cited recent studies took place in Guatemala and India. An intensely monitored (and intensely promoted) relatively small scale efficacy trial in the Guatemalan highlands  found widespread adoption and significant reduction in exposure to pollutants. A larger randomized trial from Eastern India  found that usage of improved cooking stoves initially increased but then declined as the stoves were not maintained. There was also very little reduction in markers of smoke inhalation.
The Eastern India study was conducted by Hanna, Duflo, and Greenstone. Prominent researchers with counter-intuitive findings on a widely recognized topic typically get press, and they indeed got coverage on various  blogs  as well as in the Washington Post . The authors conducted a careful long-term follow up of a program that introduced improved cooking stoves in 44 poor villages and followed 2650 households (both treated and control) multiple times over a 4 year period. Their paper explores numerous findings. Among them:
- A 7.5 percent reduction in carbon monoxide concentration in exhaled breath (a measure of smoke exposure) among cooks in the first year of usage. This is a fairly limited effect on a key marker of smoke inhalation, even among the most highly exposed segment of the household. The effect did not persist in subsequent years.
- No impact on any health measure such as measured lung capacity or self-reported symptoms of poor lung function.
- No reduction on the quantity of wood used as fuel even though the improved cook stoves are more efficient under laboratory conditions.
- About three additional meals per week were cooked on the improved cooking stove in the first year of usage, falling to less than 2 meals per week at two years after stove introduction.
- The need for repair (and time spent on repair) was far greater for the improved cooking stoves than the traditional stoves, which contributed to high rates of discontinued use.
A long term intensive study in Guatemala, known by its acronym RESPIRE, paints a different picture. Here the cooking stove, made from concrete, was extended on a randomized basis to half of the 500 households in the study. The authors measured significant reductions in indoor pollution and fuel use as well as moderate but significant reductions in child pneumonia . However RESPIRE found little reduction in adult female health , although they did find a significant reduction in self-reported symptoms of wheezing.
Let’s contrast some key features of these two studies.
Which stove is more cost effective? In India, a relatively inexpensive improved stove was offered to households at a highly subsidized price. But in reality, was this stove a good deal? A $12.50 stove that often breaks, needs regular maintenance, and is easily mis-used may be quite expensive in terms of the price per smoke-free-meal produced. On the other hand the RESPIRE stove cost is approximately $125 and, unsurprisingly, was found to be far more durable. Even with the intensive household monitoring and encouragement in the RESPIRE study, the cost per smoke-free-meal may be much lower with this stove. So it is not clear which is the “cheaper” stove in the long run, although the high initial cost of the RESPIRE stove presents obvious challenges of financing and also invites questions of whether other more effective environmental health programs can be financed with the same resources.
How comparable are the two settings? The cooking behavior appears to be radically different in the two settings. In the very poor rural setting in Eastern India only one meal per week is cooked in an enclosed space, five meals in a semi-enclosed space, and eight meals cooked outside. Even if the population perceives benefits from improved cooking stoves, the relatively small proportion of meals cooked inside will dampen widespread adoption and use as perceived benefits will not be as dramatic. Contrast this setting with the rural highlands of Guatemala, a colder climate at a higher altitude where presumably the majority of meals are cooked indoors. Women in these households spend an average of 5 hours per day in a room with a lit fire. Even if the two stoves were equally effective in pollutant reduction and equally adopted, the higher exposure in Guatemala will translate into greater measured health effects.
How much does the stove contribute to pollutant exposure? One more important fact contributes to a diluted effect size. Exposure to combustion related air pollutants is only partially due to in-home stove usage. In the RESPIRE study, mean concentrations of carbon monoxide in the kitchen were lowered by a factor of 10 as a result of the improved cooking stove, but child exposure was lower by a factor of two. Cooking smoke that is vented outside can still be inhaled, and people are exposed to smoke from disparate sources outside and in other homes.
So what do we learn from these studies as well as mixed results from Ghana  and positive results from Senegal ? Let me agree with Hanna and co-authors who conclude that the mixed evidence “suggests that, in the context of evolving stove technologies, the new generation of stoves need to be evaluated in field settings to understand if they will actually provide the benefits measured in the laboratory before valuable resources are devoted to their deployment.” I’d add that we need to understand the characteristics of settings that facilitate adoption and those that present additional constraints, such as the degree of female autonomy in the home .
I also wonder if additional large scale evaluations of cooking stove distribution programs are the best use of research funds at this point in time. Given the uncertainty over both the effectiveness of various cooking stove models in uncontrolled settings as well as the take-up rate of cooking stoves, and how the take-up rate is affected by local factors, multiple-site small-scale studies that focus on these questions are likely a much more efficient use of resources. These studies would be the cooking stove equivalent of mechanism experiments  advocated by Ludwig, Kling, and Mullainathan. As opposed to another full-scale policy evaluation of a cooking stove programs, many smaller scale studies in diverse settings can more quickly help to establish the relative performance of competing cooking stove designs, including behavioral and contextual determinants of take-up.