Impact of the Wood-Burning Justa Cookstove on Glycated Hemoglobin (HbA1c): A Stepped-Wedge Randomized Trial in Rural Honduras

Abstract

Background: Type 2 diabetes is a rapidly growing global health challenge in low- and middle-income countries (LMICs), and evidence suggests that air pollution exposure contributes. Household air pollution from burning solid fuels for cooking is a major burden in LMICs, but studies demonstrating associations between reductions in household air pollution and improvements in HbA1c, a biomarker of diabetes risk, are lacking. We previously reported substantial reductions in fine particulate matter with an aerodynamic diameter $\le 2.5\mu m$ (${\mathrm{PM}}_{2.5}$) and black carbon concentrations following an intervention in rural Honduras with the Justa cookstove, a wood-burning stove with an engineered combustion chamber and chimney.

Objective: In a stepped-wedge randomized controlled trial among 230 Honduran women using traditional wood-burning stoves at baseline, we evaluated the effect of the Justa intervention on HbA1c and characterized the longitudinal associations between air pollution exposures and HbA1c.

Methods: At each of six visits over 3 y, we measured 24-h ${\mathrm{PM}}_{2.5}$ and black carbon concentrations, and finger-stick HbA1c levels. We used linear mixed models in intent-to-treat (condition by assigned stove type), exposure-response (using 24-h measures and modeled estimates of long-term exposures), and "per protocol" self-reported stove use analyses.

Results: HbA1c was reduced for the Justa condition in comparison with the traditional stove condition, but estimates were small and not statistically significant [$-0.03$ percentage points, 95% confidence interval (CI): $-0.13$, 0.07, $n=\mathrm{1,208}$ observations]. A slightly stronger effect was observed when using self-reported stove use in per protocol analyses. Exposure-response analyses demonstrated positive associations between HbA1c and air pollution [e.g., HbA1c was 0.22 percentage points higher (95% CI: 0.13, 0.30) per log-unit higher long-term average personal ${\mathrm{PM}}_{2.5}$].

Discussion: Our study provides novel evidence of exposure-response associations between household air pollution and HbA1c within a randomized cookstove trial, contributing to the evidence base necessary to support clean cooking policy initiatives. https://doi.org/10.1289/EHP15095.

Figure 1.

Examples of traditional solid fuel cookstoves in an outdoor setting (A) and indoors (B). Photo credits: Bonnie Young.

Figure 2.

Examples of Justa cookstoves showing an external view preparing coffee (A) and an internal view with the griddle removed (B). Photo credits: Bonnie Young.

Figure 3.

Box plots of 24-h household air pollution exposures of personal ${\mathrm{PM}}_{2.5}$ (A; traditional: $n=605$; Justa: $n=563$), kitchen ${\mathrm{PM}}_{2.5}$ (B; traditional: $n=612$, Justa: $n=555$), personal BC (C; traditional: $n=597$; Justa: $n=559$), and kitchen BC (D; traditional: $n=595$, Justa: $n=556$). For reference, the blue and red dashed lines for the ${\mathrm{PM}}_{2.5}$ plots represent the WHO 24-h interim-3 air quality target ($37.5\mathrm{\mu }\mathrm{g}/{\mathrm{m}}^3$) and air quality guideline ($15\mathrm{\mu }\mathrm{g}/{\mathrm{m}}^3$), respectively. The boxes extend from lower to upper quartiles of the data, the horizontal bar is the median; whiskers extend from minimum and maximum values within 1.5 times the interquartile range; outliers are plotted as individual points. The numeric values for this figure are also described in the main text. Note: BC, black carbon; ${\mathrm{PM}}_{2.5}$, particulate matter with an aerodynamic diameter $\le 2.5\mathrm{\mu }\mathrm{m}$; WHO, World Health Organization.

Figure 4.

HbA1c by study arm, assigned stove type, and visit number ($n=\mathrm{1,208}$ HbA1c observations among 227 unique participants). HbA1c distribution by randomized study arm (arm 1 or arm 2), assigned stove type (traditional or Justa), and study visit number. The boxes extend from lower to upper quartiles of the data, the horizontal bar is the median; whiskers extend from minimum and maximum values within 1.5 times the interquartile range; outliers are plotted as individual points. For reference, the red dashed line represents the standard cutoff for diabetes (HbA1c $\ge 6.5\%$) and the gray shaded area represents the range for prediabetes (HbA1c 5.7% to $<6.5\%$). The corresponding numeric data for this figure can be found in Supplemental Materials (Supplemental Table 3).