Household air pollution from cooking is a widespread exposure affecting pregnant women and children living in low- and middle-income countries (LMICs) (1). In utero and postnatal exposures may detrimentally influence lung growth and the development of children directly through inflammation and indirectly through mediators such as pneumonia (2). Research on the association of household air pollution exposure and lung health in children would benefit from lung function assessment.

Spirometry is a cornerstone tool to evaluate lung health. It can detect impairment before clinically apparent disease (3); however, it is not feasible in preschool-aged children given the effort required to conduct the test (4). Options for lung function testing in young children have been limited until recently and either have been too expensive or require highly controlled environments and technical expertise often not available in low-resource settings. These limitations have hampered our understanding of lung function early in life.

The landscape is changing. Tests obtained during passive tidal breathing, such as respiratory oscillometry, do not require the same degree of coordination as spirometry and can be successfully used in preschool-aged children (5). In a study reported in this issue of the Journal, Agyapong and colleagues (pp. 716–726) used oscillometry in four-year-old children participating in GRAPHS (Ghana Randomized Air Pollution and Health Study), a cluster-randomized trial of a cleaner cooking stove intervention among 1,414 pregnant women assigned to one of three arms: a liquefied petroleum gas (LPG) stove and fuel, an improved biomass stove (BioLite; BioLite Inc.), or continued cooking with biomass (6, 7). The LPG and BioLite stove interventions were implemented at an average gestational age of 17–18 weeks and continued through the first year of life (7). The LPG stove reduced personal exposures to carbon monoxide by 47% compared with control subjects. There was no difference in personal carbon monoxide or fine particulate matter when comparing BioLite with control children (7). In intention-to-treat analysis, there were no differences between arms in birth weight or severe pneumonia (7).

In this follow-up study, a subset of 4-year-old trial participants born to women randomized to LPG had oscillometry indices consistent with improved lung function compared with control subjects. This is the first study to demonstrate a difference between fetal cookstove interventions and traditional stoves on the lung function of preschool-aged children 3–5 years of age in LMICs. Previous research using the interrupter technique (Whistler LFMi; MediSpirit BV) with 1-month-old infants from this same Ghanian cohort and respiratory oscillometry (tremoflo C-100; Thorasys) from 3-year-olds in Nigeria reported an exposure–response association between household air pollutants and lung function (8, 9); however, intention-to-treat analyses showed no lung function benefit from the cleaner cookstove interventions. Agyapong and colleagues (6) also report associations between worse oscillometry measures and increasing carbon monoxide concentrations in the prenatal period but not postnatally. These exposure–response results, which are likely generalizable to other settings and pollutant sources, held when adjusted for particulate matter concentrations at 4 years of age. Further analyses showed positive associations between prenatal carbon monoxide concentrations and several oscillometry parameters during gestational age windows (when airway branching and growth predominate) that may be especially pollutant sensitive and therefore strategic for a short-term, discrete intervention given that access to fuels such as LPG remains limited. Unexpectedly, 4-year-olds randomized to the BioLite stove had worse lung function than control subjects. In utero and postnatal exposures to carbon monoxide and fine particulate matter were similar between BioLite and control children, which may explain why lung function was not any better in the BioLite arm; other unmeasured pollutants may explain the suggestion of worse lung function among the BioLite arm.

Although oscillometry was first described in the 1950s, it is not widely used and has only recently become commercially available. As a result, most healthcare providers and researchers are unfamiliar with it. The oscillometry findings from the LPG intervention are from one time point (at four years of age), lack uniformity across relevant frequency ranges, and have uncertain clinical significance. Although the present findings are an important, these results would likely be enhanced by incorporating additional approaches, including longitudinal lung growth and developmental trajectories in children and potentially sensitive bronchodilator reversibility and intrabreath oscillometry measurements to enrich the functional evaluation (6, 10). The authors also report a lower difference between resistance at 5 Hz (R5) and at 20 Hz in LPG children compared with control subjects but not in R5. The lack of an effect on R5 between the LPG and control arms could reflect greater variation in this oscillometry measure from high pediatric respiratory rates, which may distort oscillometry measures at lower frequency ranges (11). This concern could be addressed by using higher frequencies (7 or 8 Hz). The authors also stated that the difference between R5 and resistance at 20 Hz captures small-airway resistance, but the specificity of this parameter for the small airways remains debatable and should be interpreted cautiously, especially as this study lacks a clinical correlate.

Oscillometry continues to evolve, and device advances and refinements in application, data processing, and measurements continue. Oscillometry measurements vary by device and depend on signal-processing and data-cleaning practices (11–13). To better assess and replicate this research, we recommend more fully describing all oscillometry procedures, including reporting on units and reference values used, in line with international guidance (11).

We also note that these results were observed even though the LPG stove and fuel intervention did not reduce all pollutants to low concentrations. Although carbon monoxide was reduced by 47% in the intervention arm, and most values were below the World Health Organization’s 24-hour average air quality guideline, the postintervention concentrations of fine particulate matter were relatively high and mostly above the World Health Organization’s annual and daily air quality guidelines, demonstrating an opportunity to reduce exposure even more. Furthermore, no households from the former LPG intervention arm reported continued use of LPG, and pollution concentrations in the LPG arm at four years of age were higher than during the intervention and nearly back to those measured at baseline and at four years of age in the control arm. These aspects highlight the potential importance of exposure, and interventions, in the prenatal and early infancy periods regardless of what happens after this period, and they also highlight the continued reliance on biomass and continued lack of access to cleaner options, which continue in many areas of the world.

Overall, these results provide valuable evidence that prenatal use of LPG cookstoves during pregnancy and the first year of life may positively benefit early-life lung development among biomass users. The results also emphasize the importance of evaluating longer term effects of cookstove interventions even when no discernable immediate impact is detected. If replicated, these results could have important implications for future lung health potential and the prevention of respiratory diseases later in life. These findings also highlight the challenges in achieving equitable access to cleaner energy sources, particularly in low-resource settings in LMICs. As we learn about the potential benefits of interventions, we must also address access for the science to be relevant.