Forty percent of the world’s population, nearly 3 billion people, rely on biomass fuels for daily cooking and energy needs. Incomplete combustion of biomass fuels results in high exposures to household air pollution (HAP), a complex and toxic mixture. The most commonly measured pollutants include fine particulate matter (PM_2.5) and carbon monoxide (CO). In 2019, HAP was responsible for 2.3 million deaths and 91.5 million disability-adjusted life years (1). Together, cardiovascular disease and chronic respiratory diseases account for approximately 63% of all HAP-attributable deaths and 40% of disability-adjusted life years. Much of the burden is concentrated in low- and middle-income countries where, on average, an estimated 65% of households cook with solid fuels (2).

Despite this extraordinary burden of disease, randomized controlled trial (RCT) evidence supporting cleaner cookstove interventions to improve adult cardiopulmonary outcomes is lacking. Romieu and colleagues in 2009, and Hanna and colleagues in 2016, reported household-level chimney stove interventions versus traditional open-fire stoves, and intention-to-treat analyses found no effect on lung function decline, likely driven by low intervention uptake (3, 4). A report by Guarnieri and colleagues in 2015 similarly found no evidence that a household-level chimney stove intervention improved adult lung function, but separately published exposure-response analyses were suggestive of an effect (5, 6). Zhou and colleagues did conversely find that improved ventilation or biogas stoves for 9 years improved FEV₁ decline as compared with open fire, suggesting that a long duration of follow up may be required to see health effects (7). Although broader prospective cohorts, including work from our group, generally show a positive association between HAP exposure and blood pressure (BP) and, over the life course, cardiovascular mortality, data from RCTs is limited (8–10). For example, a secondary analysis of an ethanol intervention and a pre-/postimproved cookstove study suggests that interventions to reduce HAP exposure may improve diastolic or systolic BP, respectively (11, 12). Given these mixed results, the global health community continues to seek evidence in support of a cookstove intervention strategy to improve health.

In this issue of the Journal, Checkley and colleagues (pp. 1386–1397) examine the effect of a year-long, multifaceted liquefied petroleum gas (LPG) cookstove intervention as compared with biomass (commonly animal dung) cookstove on cardiopulmonary health in adults, specifically resting BP, peak expiratory flow (PEF), and respiratory symptoms as measured by the St. George’s Respiratory Questionnaire (13). The intervention included provision of a three-burner LPG stove, education and behavioral messaging, and biweekly LPG fuel refills and stove inspections and repairs for the duration of the study. Repeated personal exposures to PM_2.5, black carbon, and CO exposures as well as kitchen area PM_2.5, CO, and, in a subset, nitrogen dioxide (NO₂) were measured. To understand patterns of stove use, temperature loggers were placed on LPG and biomass stoves in intervention homes and biomass stoves in control homes; 24 control homes also had temperature loggers placed on LPG stoves (71% of control households already owned an LPG stove). The authors are to be commended on the strength of their intervention, exposure measurement strategy, and objective health outcome assessments.

The primary finding of this impressive study was that a year-long LPG intervention with robust measures to enable LPG stove use was not associated with improvement in BP, PEF, or respiratory symptoms in women randomized to the cleaner burning LPG stove as compared with control. Strikingly, exposure-response analyses found no association between average personal PM_2.5 exposure and systolic or diastolic BP or postbronchodilator PEF (but did suggest that higher personal PM_2.5 exposure was associated with lower prebronchodilator PEF). These findings, like the many prior RCTs, once again leave us asking why cleaner cooking interventions do not appear to improve health. Given the deep literature linking air pollution exposure to cardiopulmonary risk, we posit that these results should not be construed as evidence that smoke from polluting cookstoves is safe. Instead, two factors may explain the results in this article.

First, the latency between HAP exposure and cardiopulmonary response may be too great for a year-long exposure reduction to induce a detectable difference (8). Enrolled women were, on average, 47.9 and 48.7 years old in the control and intervention arms, respectively, and these data suggest that established cardiopulmonary health trajectories are not altered by a short-term, later-life intervention. Although the ideal approach to HAP would be to eliminate polluting stoves and fuels in perpetuity, this is not yet feasible in most low- and middle-income countries. Given the many competing demands for scarce public health dollars (or soles), a limited-duration intervention to improve cardiopulmonary health may be more impactful if deployed earlier in life during sensitive windows of exposure (14, 15) compared with in later adulthood when these trajectories may already be set.

Second, despite the 98% exclusive LPG use in the intervention arm as defined by temperature loggers, high levels of exposure (on average, 30 μg/m³ personal PM_2.5 exposure and 58 μg/m³ kitchen area PM_2.5 concentration) were still observed postintervention, as was an overlapping distribution of personal PM_2.5 exposure with women randomized to control. In other words, exclusive use of the LPG intervention stove does not appear to have sufficiently reduced average exposures or produced a large enough exposure differential to impact health. Housing density and the number of intervention households per community are not provided, but randomization at the household level may have left intervention households exposed to pollution from neighboring households’ cookstoves. As an extension of this, perhaps it is not just the average PM_2.5 exposure that is important but rather the peak exposures or the composition or mixture of pollutants that is particularly toxic. Checkley and colleagues have begun to examine this concept by measuring PM_2.5, black carbon, CO, and NO₂ exposures; banked filters and real-time exposure data should be leveraged to extend these analyses.

What is next for cleaner cookstoves research and policy? We offer three thoughts.

First, the fact that so many cookstove intervention trials register high exposures in intervention arms suggests that we must expand our focus beyond the household-level cookstove and endeavor to understand the totality of air pollution exposures in these communities. As these data suggest, a cleaner cookstove intervention alone may be insufficient to address this global public health crisis. Holistic air quality management approaches in low- and middle-income settings will entail both energy systems that deliver cleaner household energy at scale and at an affordable price and also coordinated efforts to reduce emissions from traffic, industry, and agriculture. In addition to directly improving health, many of these policies will reduce greenhouse gas emissions with global implications.

Second, this and other cleaner fuel trials throughout the Global South establish that even rural populations with limited prior exposure to modern fuels embrace cleaner cooking when it is convenient and cheap. This suggests that the central challenge going forward is to map out the regulatory and economic arrangements that will pave the way to societal transitions to cleaner household energy systems.

A final point pertains to research: as the list of cleaner cookstove RCTs accumulates, we see an opportunity to carry out pooled meta-analyses and leverage these studies’ filter libraries and biospecimen repositories to better characterize the composition and mixture of HAP exposure and biomarkers of subclinical effect that may be more sensitive than outcomes examined in the primary intention-to-treat analyses.