Household air pollution (HAP) causes immense disease burden throughout the world. Around 3.8 million people died prematurely from illness attributed to HAP (1). Globally, “by far the most important direct health risk is the pollution caused by incomplete combustion of solid fuels for cooking, heating and lighting” (2). The adverse impacts from HAP are largely caused by energy poverty, especially in rural regions of the low-and middle-income countries (LMICs) where some residents lack access to affordable, clean energy such as electricity, biogas and gas (3). Instead, they rely on solid fuel collected from agricultural residues, hauled from kilometers away, or purchased at a low price to meet daily energy demand (3). According to the World Health Organization (3), solid fuel includes coal as well as biomass fuels (referring to renewable plant-based material such as wood, crop wastes and charcoal), providing heat and light during the process of combustion (4). Incomplete combustion of solid fuels produces high levels of HAP with a range of harmful pollutants, including particulate matter, sulfur oxides, nitrogen oxides, carbon monoxide, polycyclic aromatic hydrocarbons, formaldehyde, and dioxins, to name a few (5–9). In contrast, clean fuel mainly includes “electricity, liquefied petroleum gas (LPG), piped natural gas (PNG), biogas, solar and alcohol fuels”, which produces low levels of emissions of particulate matter, sulfur dioxide and other by-products of incomplete combustion when properly used (9). Although the past few years have witnessed a surge in technological innovation in the household energy sector, progress remains too slow to displace the polluting fuel combustion systems and thereby mitigate their health impacts. Based on the most recent global estimates, more than 2.7 billion people heavily relied on domestic solid fuels in 2015, including 450 million people in China (10).

Household air pollution from solid cooking fuel (notably coal and wood) has been categorized as a Group 2A carcinogen (11). Special attention should be placed to females who spend considerable amount of time in proximity to polluting sources due to their great involvement in daily cooking activity (4). Ambient air pollutants (e.g., particulate matter, polycyclic aromatic hydrocarbons) may cause tumor formation in breast and cervix uteri (12–14). Evidence for the relationship with household air pollutants remains scarce. Three previous studies have examined the indoor solid fuel combustion as a risk factor for breast cancer and yielded inconsistent result (15–17). Previous studies on this topic have mainly conducted in high-income countries and focused on wood burning (15, 16). However, in some coal-producing countries such as China and India, coal is considered as a domestic source of energy (11). There is a paucity of studies on the potential impact from indoor coal combustion for cooking. Furthermore, HAP from cooking and space heating are two different exposure patterns, which may have different influences on carcinogenesis. A stove might be kept going all day for heating in winter months (3). By contrast, cooking produces HAP several times per day with a shorter period (3, 18). Field measurement reported significantly lower emissions of pollutants from domestic solid fuel combustion during heating compared to those from cooking (19). Epidemiological evidence on HAP exposure from cooking and heating reported different associations with lung cancer (20). One previous CKB study on HAP from heating fuel use and breast cancer mortality did not find any evident relationship with breast cancer mortality (17). Little is known about the independent effect of cooking fuel use on female breast cancer risk. This study reported findings on the solid cooking fuel combustion with breast cancer risk among 290,396 females.

Of the 300,272 females [mean (SD) age 51.46 (10.48) years], 51.1 % always used solid cooking fuel and 18.0% always used clean fuel in all residences. Females who always used solid fuels tended to be older, more likely to live in rural region, less educated, more exposed to passive smoking, less likely to use oral contraceptive pills and had lower household income in comparison with clean fuel users (Table 1).

We documented 551 participants diagnosed with breast cancer. Compared with long-term clean cooking fuel use, long-term coal combustion was associated with a higher risk of breast cancer (fully adjusted OR:2.07, 95%CI: 1.37–3.13) (Table 2). Fully adjusted ORs of breast cancer were 1.12 (95%CI: 0.72–1.76) for those who always used wood, and 0.98 (95% CI: 0.55–1.74) for those who used mixed solid fuels to cook [mean duration of exposure: 16 years]. Long-term solid cooking fuel combustion [mean duration of exposure: 30 years] appeared to confer a higher risk of breast cancer, albeit not significant (fully adjusted OR,1.19 (95%CI: 0.84–1.67). There was no elevated cancer risk among women who had switched into clean fuels [mean duration of exposure: 18 years]. No evident relationship was observed between solid fuel use and breast cancer risk.

There was no statistical effect measure modification by environmental tobacco smoking (ETS), cooking stove ventilation, menopausal status or contraceptive use (Figure 1). The strength of observed associations remained largely unchanged after excluding the mixed fuel users. The adjusted OR was somewhat stronger in females with daily ETS exposure (OR: 3.26, 95% CI: 1.83–5.81) than in those who got exposed to ETS 1–5 days per week (OR 0.98, 95% CI: 0.38–2.48) and in those who never/occasionally got exposed to ETS (OR 0.73, 95%CI: 0.38–1.38).

In the sensitivity analyses, the association of solid fuel exposure and breast cancer risk was unaltered after adjusting for potential confounders and excluding regular smokers, nulliparous women and those who had ever used oral contraceptive drugs (Supplementary Table 1). When 5-year or 10-year lag period was adopted, the strength of observed associations of two cancers appeared to be increased among long-term wood users and overall long-term solid fuel users, yielding significant results (Supplementary Tables 2, 3). We did not observe excess risk of breast cancer mortality, probably due to insufficient number of deaths (Supplementary Table 4).

In this study, we observed inconsistent associations of solid cooking fuel exposure with breast cancer risk. The adjusted ORs of breast cancer were not statistically significant among persistent solid fuel users in general (OR: 1.19, 0.84–1.67). In line with our finding, a case-control study of women on Long Island demonstrated no increased risk of breast cancer incidence in females who frequently burned wood in their home (16). However, when stratifying by type of solid fuel use, we observed a higher risk of breast cancer in persistent coal users (OR 2.07, 2.37–3.13) but not in persistent wood users. Apart from that, a prospective cohort study in the United States or Puerto Rico suggested that having indoor wood-burning stove/fireplace appeared to confer higher breast cancer risk (HR=1.11, 95%CI: 1.01–1.22) (15). Reports are inconsistent on which type of wood (synthetic or wood logs) can produce more polycyclic aromatic hydrocarbon (PAH) during domestic combustion (5, 6, 28). Previous association studies and risk assessment mainly focused on household wood combustion. The present study examined both wood and coal exposure and yielded inconsistent associations with breast cancer. Further prospective evidence is needed to elucidate the relationship of individual and combined effect of wood and coal exposure with breast cancer risk. Moreover, previous CKB study on heating fuel use did not observe excess risk of breast cancer mortality in any solid fuel groups (10). In contrast, this study focused on cooking fuel use and firstly suggested a positive association of long-term coal combustion for cooking with breast cancer risk. The strength of association remained largely unchanged in sensitivity analyses (Supplementary Tables 1–3). A possible explanation is that solid fuel combustion for cooking has a longer lifetime duration and thus provides higher cumulative inhaled pollutants compared to solid fuel combustion for heating (18). HAP from heating is a seasonal exposure during winter months while HAP from cooking is a regular exposure in this study since we included long-term solid fuel users who cooked daily or weekly in each residence lived at least 1 year. Differences in study design and covariates adjustment may also lead to different findings in two CKB studies. The association of HAP from different domestic activities (e.g., cooking and heating) with breast cancer risk needs future research to elucidate.

We observed no elevated breast cancer risk among women who had ceased using solid fuels. The point estimate of risk was lower in those who had switched from solid to clean fuels than long-term solid fuel users [OR]. Those who had ceased using solid fuels may get less exposed to solid fuel burning than long-term solid fuel users [duration in years: median (IQR): 16 (9–25) vs. 30 (21–41)]. Previous CKB study has demonstrated that the excess risk of all-cause mortality decreased by more than 60% in 5 years after cessation of indoor solid fuel burning (29). The present study further reported the health impact of cession from solid fuels on breast cancer risk. On the global basis, females bear a large share of disease burden caused by HAP due to their domestic roles (3). Our findings may have unique implications on females and suggest the reduction of solid fuel use for cooking. Targeted efforts are needed to accelerate the promotion of clean fuel production facilities and distribution networks.

The association between household air pollution and breast cancer is biologically plausible. Incomplete combustion of solid fuels releases many pollutants to the indoor and outdoor air, such as carbon monoxide, particulate matter, carcinogenic polycyclic aromatic hydrocarbons (PAHs) (5–9). Of all these pollutants, PAHs have been widely investigated and classified as carcinogenic to humans (IARC Group1) (30). About 60.5% of the global total PAH emissions were from combustion of biomass fuels including wood and crop residues (31). In China, coal and biomass fuel combustion are two major emission activates of PAHs, accounting for roughly 20 and 60%, respectively (32). The field measurements showed that the total emission factors (EFs) of 28 PAHs from solid fuel combustion during a regular cooking period ranged from 20.7 to 535 mg/kg (33). EFs of PAHs varied from several mg/kg for wood fuels to about 200mg/kg for bituminous coal, a dirty fuel burned in domestic stoves in rural China due to low cost (34). Different emission profiles between coal and biomass combustion were also observed for predominant individual PAHs including benzo[a]pyrene(BaP), pyrene (PYR), perylene (PER), Benzo[e]pyrene(BeP) and dibenzo[a,l]pyrene (DBalP) (33). Experimental evidence has confirmed that PAH metabolites can react with DNA and form PAH-DNA adducts, which leads to mutations of cancer related-genes and cell death (35–37). Potential carcinogenic pathways include sister chromatid exchange, mutations in TP53 as well as DNA methylation (26, 38, 39). BaP, a marker of carcinogenic potency of PAH mixture and an endocrine-disrupting pollutant, was associated with increased risk of breast cancer in a French cohort (13, 40).

Persistent coal users had a higher risk of breast cancer than persistent wood users. We cannot directly compare our estimate for coal exposure with prior studies. To our knowledge, this is the first study which reports the association of breast cancer with coal combustion for cooking. Our results should be interpreted with caution due to relatively small number of cases. Given the sample sizes in the subgroups, we have sufficient power for coal combustion analysis (approximately 100%) but not for wood combustion (<50%). Although the play of chance cannot be ruled out, our analysis may suggest that pollutants from coal combustion could have more hazardous effect on breast cancer development than those from wood combustion. Different fuel properties and environmental condition contributes to the different formation and changes of trace organics emitted from combustion which may have adverse effects on breast carcinogenesis (41). Results from a previous field emission test study revealed that there was a statistically positive relationship between PAH derivatives and corresponding parent PAHs in emissions from coal combustion, but insignificant relationships for those from wood burning (41). PAHs exposure could be ubiquitous and concurrent multiple indoor sources of PAHs were associated with a 30–50% increase in breast cancer risk (28). Similarly, PAH profiles from inhalation and digestion could be modifiable risk factors (28). Further studies are warranted to monitor the multiple sources of PAH emissions between coal and wood combustion for cooking and elucidate their association with breast cancer.

The chief strengths of this study include the large number of cooking fuel users (in particular for coal users), geographical diversity and completeness of data collection. Moreover, to discount the exposure that is thought irrelevant to the outcome, we conducted sensitivity analyses and selected 5-year or 10-year lag period. Our study has several limitations as well. First, the cross-sectional design of this study precludes a causal inference between solid fuel exposure and risk of breast cancer. Further prospective studies are needed to confirm the causal relationship. Second, like other CKB studies, recall-bias is possible because of the self-reported nature of the baseline survey. Nevertheless, about 78% of the participants in the resurvey reported the same type of cooking fuel as in the baseline survey, and the kappa value for cooking information was acceptable (0.6) (42). The physician-diagnosed cancer history was also confirmed by hospital admission information in the resurvey. Third, although self-reported primary cooking fuel has been adopted as a practical proxy of HAP in many studies, it remains an inherently limited indicator (3). It is possible that secondary fuel exposure and pollutants from neighborhood also contribute to the HAP. Primary fuel use represents a compromise which balances imperative of capturing detailed information on HAP with the pragmatic considerations such as feasibility of conducting surveys and eliciting reliable information from participants. Fourth, we did not account for ambient air pollution that might contribute to breast cancer risk. Since CKB public database did not disclose home address due to privacy protection, GIS method (grid-based method) cannot be used to locate and control for ambient air pollutants. However, CKB disclosed the province where study participants were resided in, and in each province the study participants were all located in the same community or village. Although we could not obtain ambient air pollution data, we adjusted for study region in all models and assumed a similar pattern of ambient air pollution exposure from the same region (29). We expected this strategy could somehow account for residual confounding from ambient air pollution (29). Finally, CKB project does not include histotype or genetic information.

Household air pollution from solid cooking fuel combustion may elevate the risk of female breast cancer. The strength of the association is higher in long-term coal users than in long-term wood users. This study may have global implications as many countries are in the transition to clean energy. Efforts to disseminate clean and affordable alternatives (electricity and gas) are gaining momentum in LMICs (3). Adoption of sustainable clean energy solutions hinges on improved understanding of gender dynamics of household energy use and sex-specific health impacts (3). Gender-responsive interventions which taking into account the gender roles in household energy acquisition and uses are required. More evidence on health impacts on females is needed for implementation of policies to promote health, as females are often the primary cooking fuel users and the ones who benefit most from transition to clean cooking fuels (3).

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethical Review Committee of the Chinese Center for Disease Control and Prevention and the Oxford Tropical Research Ethics Committee. The patients/participants provided their written informed consent to participate in this study.

TL and SW designed the study. SW acquired the data. TL analyzed and drafted the manuscript. All authors contributed to the interpretation of data and revised the article. All authors read and approved the final article.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.