‘This smoke will finish us’: impacts of cooking with polluting fuels on air quality, health and education in three schools in Nairobi, Kenya

Links between cooking with polluting fuels (e.g. wood and charcoal), air pollution and health in domestic settings are well-established. However, few studies have been conducted in schools that rely on such fuels for catering. This study is the first investigation of air pollution, cooking, and health in schools in Nairobi, Kenya. We carried out an in-depth mixed-methods study in three schools (two primary schools and a college) in an informal settlement using wood and/or charcoal for catering. In each school, repeated 24-hour air pollution measurements (fine particulate matter (PM2.5) and carbon monoxide (CO)) were collected to assess concentrations in the main kitchen and a nearby classroom, in addition to personal exposure for the main cook. Surveys with catering staff collected data on perspectives on air pollution and health symptoms. Focus groups were conducted with catering staff, teachers and senior management to understand perceived impacts on health and the school environment. 24 hour levels of PM2.5 were found to surpass world health organization interim target level 1 (IT-1) (35 μg m−3) in all schools—with levels three times higher (107.6 μg m−3) in classrooms, ten times higher (316.2 μg m−3) in kitchens and nearly six times higher (200.9 μg m−3) among cooks. Peak levels of pollution were closely linked to times of stove use, as concentrations doubled in classrooms and tripled in kitchens during cooking. Catering staff reported being concerned about their health, and reported experiencing wheezing, chest pains, eye irritation and headaches and attempted to avoid the smoke to reduce exposure. Disturbance to classes from cooking smoke was reported by teachers, with students reporting coughing and sneezing from exposure. Support is needed to enable clean cooking transitions in schools to create a healthy and safe learning environment.


Introduction
Transitions to clean cooking fuels are a recognised global public health and development priority to promote health, well-being, and gender equity, forming a central part of the United Nations 2030 Agenda for sustainable development (Goal 7) [1] and the world health organization (WHO) guidelines on air quality [2].Domestic use of polluting fuels (e.g.wood, charcoal and kerosene) leads to more than 3 million deaths globally each year due to exposure to household air pollution [3], particularly from fine particulates (PM 2.5 ) causing respiratory and cardiovascular diseases [4,5].The burden is estimated to be greatest (with 900 000 deaths annually) in sub-Saharan Africa (SSA), where 85% of the population remains dependent on polluting fuels for cooking [6].Harvesting biomass fuels (wood, charcoal) can also negatively impact the local environment through deforestation and soil degradation, and further contribute to climate change through the emission of CO 2 and short-term climate-forcing black carbon and other greenhouse gases from incomplete combustion [4,7].Therefore, transitions to clean fuels that have no or very low emissions at the point of use, such as liquefied petroleum gas (LPG), biogas, and electricity, are urgently advocated to achieve co-benefits for health and climate [8].
In addition to household air pollution, populations are likely to be at risk from exposure to cooking-generated air pollution in public institutions (schools, hospitals, etc) where high usage of biomass fuels dominates.The world food programme estimates that 80% of school meals supported under feeding programmes in low-and middle-income countries (LMICs) are prepared on three-stone fires, with wood fuel often supplied by children and their parents [9].Despite a significant increase in research during the last two decades focusing on cooking, air pollution and health in household settings in LMICs [10], research focused on cooking, air pollution and health in non-domestic settings, such as schools, has been lacking.The recent publication of the international standard test protocol for institutional cookstoves highlights the importance of evaluating institutional cookstove emissions and safety [11].Moreover, the world bank's report, The State of Cooking Energy Access in Schools, highlights how scant data on energy use for cooking and health implications in schools has limited visibility and lack of investments for clean cooking transitions [12].
Air pollution in schools has been associated with impacts on children's health, cognitive development and learning performance in high and middle-income countries [13][14][15][16][17]. Studies that have investigated air pollution in schools in SSA have focused on the health impacts of external pollution sources with little or no consideration of the additional impacts from the combustion of cooking fuels within the school.In South Africa, a higher prevalence of asthma was found among children from schools located in industrial areas [18], and levels of sulphur dioxide in classrooms and around schools located near coal mines were found to be of concern [19].In Kenya, PM 2.5 and PM 10 levels in classrooms from five schools in an industrialised area exceeded WHO-recommended guidelines, but school cooking fuels were not detailed [20].Similarly, in Kigali, Rwanda, annual mean PM 2.5 concentrations in a school classroom were recorded to be eight times higher than WHO guideline levels-with some suggestion that the school cooking practices influenced peak levels outside school drop-off periods [21].Further research is vital to understand how cooking practices in schools contribute to air pollution and impact health and learning.This will enable the development of solutions that can help improve air quality.
Education in Kenya is overseen by the Ministry of Education (MoE) and funded through public funds allocated per student head across pre-primary, primary and secondary levels [22].However, there is also a proportion of privately funded schools and public schools that often receive support from external organisations as well as parental contributions [23].90% of public schools are reported to use firewood for cooking [24].However, a 2018 nationwide survey of educational institutions found that reliance on wood fuel was as high as 99% in primary and secondary schools, with the remaining schools relying on charcoal [25].A typical Kenyan public boarding school requires an estimated 168 metric tons of wood fuel annually, equating to 56 acres of forest [26] if not sustainably sourced.As schools generally lack a stock of wood fuel, the pressure of sourcing firewood is often transferred to families, which further contributes to local deforestation.Although the Kenyan government's school meal policy stipulates providing a 'healthy, safe, nutritious and culture-sensitive school meal' to enhance retention of learners in schools [27], no consideration is given to the provision of fuel(s) to sustain the feeding programme.Increasing fuel and food prices [28], and rising student numbers combined with limited budgets for school meals, consequently, present a significant financial challenge for schools which have limited capacity to transition to clean energy solutions [29].Significant action (e.g. through national directives/ policies) is needed to support schools to transition to clean fuels to protect people, their environments, and the planet [9].
Historically, cooking programmes in East African schools have promoted the introduction of improved cookstoves (ICS) that increase efficiency and reduce firewood consumption [30].These programmes have been driven by environmental motivations, to reduce deforestation rather than potential health benefits from clean cooking transitions.Recently, increased awareness of the health impacts alongside increased availability of other fuels has led to growing calls to move away from polluting biomass fuels in schools and switch to cleaner cooking options, such as LPG.In Kenya, the current 2022 presidential administration announced in February 2023 an objective to switch all Kenyan schools to LPG by 2025 [31] and programmes to support school clean cooking fuel transitions have been recently initiated, including through a loan scheme through a development bank [32].
To understand the links between cooking with polluting fuels and air pollution in Kenyan schools and the potential for transitions to clean cooking, pilot research was conducted in three schools (two primary schools and a college) located in Mukuru-one of Nairobi's largest informal settlements.The research aimed to (1) understand the links between cooking and air pollution in the schools, (2) investigate potential impacts on the health and well-being of staff and students from poor air quality and (3) explore perspectives on air pollution and potential to transition to clean fuels for cooking.This paper presents the results from air pollution monitoring carried out to assess levels of air pollution in the school kitchen and a classroom and exposure for the main cook, along with impacts on health and the school environment gathered through surveys and focus group discussions (FGDs).

Study setting
Nairobi is Kenya's capital and largest city, with a population of 4.4 million [33], where an estimated 60%-70% of the population live in informal settlements [34].The study was conducted in Mukuru, Nairobi, specifically the settlements of Mukuru Kwa Rueben and Mukuru Kwa Njenga.Mukuru is a 2.3 km 2 collection of informal settlements located in the southeast of Nairobi, near the international airport, home to approximately 1 million inhabitants [35].Mukuru suffers from overcrowded housing, poor infrastructure, and limited access to water and sanitation among other basic services [35].Residents experience significant environmental health risks, along with high poverty rates making them vulnerable to disease and child mortality [35].Given the high proportion of the population residing in informal settlements in Nairobi and predicted rates of rural-to-urban migration, Mukuru was selected to generate a contextual understanding of the challenges for schools in achieving clean air.

Study design
The study was designed to explore the impacts of schools' reliance on polluting cooking fuels (wood and/or charcoal) on air pollution, health and education.It employed mixed methods to investigate school cooking fuel consumption, cooking practices and how these influenced levels of air pollution in the school and the health and well-being of staff and students.Due to the scarcity of research investigating cooking, air pollution and health in schools, an exploratory case study approach was deemed appropriate for this pilot research [36].Over a 6 month period from September 2021 to February 2022, mixed-methods research was conducted in three case study schools.Initial school visits catalogued school demographics (number of staff, students and catering staff), cooking practices (including meal choice and timings), kitchen set-up and fuel use prior to the main research phase.

Study schools
The three schools were selected to represent a range of fuel types used for cooking (table 1).School 1 (S1:Wood) was a public primary school, with over 2000 students, that used wood as its main cooking fuel.The wood was used in large modern cooking stoves with chimneys to vent as well as a traditional three stone wood stove.School 2 (S2:Charcoal) was a public primary school with sponsorship from a faith-based organisation, with over 2800 students, that used charcoal briquette produced by the school as its main cooking fuel.School 3 (S3:Wood) was a private technical college with 160 students that mainly used wood for catering in modern cooking stoves with some use of LPG used for additional cooking tasks.All schools consisted of multiple buildings, with separate kitchens and a number of classrooms with the school compound: the layouts for each school can be seen in figure 1.

Data collection 2.4.1. Air pollution and stove use monitoring (SUM)
Air pollution measurements were conducted for fine particulate matter (PM 2.5 ) and carbon monoxide (CO) using the micro-personal exposure monitor (microPEM) nephelometer and a USB-EL-CO data logger (LASCAR electronics), respectively.Real-time measurements of air pollution were taken through 1 min sampling over a 24 hour period.Real-time PM 2.5 measurements were filter-corrected by pre-and post-weighing filters that were used in the microPEM for 24 hour monitoring.For each school, three 24 hr air pollution monitoring periods were conducted.Measurements of PM 2.5 and CO were taken from the ambient environment (safe location in the school playground), the main school kitchen, a nearby classroom and the main cook's personal exposure.While the ambient monitor is representative of the local school environment it is likely impacted by cooking at the school and may not be representative of the ambient concentrations beyond the school area.Monitoring locations are shown in relation to the school layouts in figure 1.For the school kitchen, monitors were placed one meter away from the main stove at a height of one meter.For the selected classroom, monitors were placed one meter high in a safe location within each classroom.For the main school cook, monitors were worn in a bespoke apron with a centrally located pocket at the front nearthe cooks' respirable zone.School cooks were asked to wear the monitors for the 24 hour periods (removing the monitors whilst sleeping and bathing) to cover time spent in and outside school.
To associate pollution levels with periods of cooking, SUMs (Geocene thermocouples) [37] were placed on the main cooking stove to record temperature fluctuations and identify when cooking activities were carried out.The SUMs were placed approximately 10-15 cm from the center of the flame/ heat source of the main cooking stove with temperature sampling every five minutes concurrently with the air pollution monitoring.

Surveys with catering stuff and students
Surveys were conducted with ∼100 students and all catering staff from each school to collect information on knowledge of air pollution, general health and well-being, acute health symptoms and burns (linked to fuel choice and exposure to air pollution), impacts of smoke from cooking on ability to carry out daily tasks, and perceptions of menu choice and cooking facilities at the school.In S1:Wood and S2:Charcoal, students from age 10 and above were selected to ensure they were able to fully participate and answer questions.Participants were then randomly selected from classes in the final three years of primary school (i.e.students aged 10 and over).In the college (S3:Wood), respondents were selected across all courses to incorporate a variety of responses.Surveys were administered using mobile phones with secure data collection software (encrypted at source) through Mobenzi Researcher (www.mobenzi.com/),allowing for real-time quality control of the data.

Focus groups and in-depth interviews
FGDs were held separately with teachers, catering staff, and senior management in each school, except for S3:Wood, where FGDs were conducted with teachers and catering staff and separate individual interviews were conducted with senior management.The discussions explored school cooking practices, perspectives on the impacts of cooking on air pollution and links to health impacts, decision making and responsibilities for fuel procurement.Additionally, four in-depth interviews were conducted with key school decision makers (headteacher from S1:Wood, headteacher and finance coordinator from S2:Charcoal and lead coordinator from S3:Wood) and resource managers from each school, to collect more detailed information on decision making processes, financial information for cooking expenditure, and education around air pollution (two interviews in S2:Charcoal and one each in S1:Wood and S3:Wood).

Data analysis
Survey data were cleaned and analysed using STATA v10.Descriptive statistics were used to summarize the responses.Air pollution monitoring data and SUMs data were carefully checked for errors before processing (including correction based on gravimetric calibration).Stove use 'events' were generated from SUMs temperature time series using the Geocene 'FireFinder' proprietary algorithm.The minimum stove use 'event' duration was five minutes and any events occurring within 30 min were grouped into a single event.The processed SUMs data was then linked to the real-time air pollution monitoring data.Arithmetic means were calculated for 24 hr periods, for stove use and non-stove use periods, as well as for school and non-school hours for both PM 2.5 and CO.
Qualitative interview data were analysed in Nvivo [38], following the framework method-a 7-step process of transcription, familiarisation, coding, framework development, framework application, data charting and interpretation, developed to support applied research and inform health policy and practice [39].Three researchers were involved in the analysis process and regularly convened to ensure consensus at each step.Open coding was completed to generate an initial set of codes.These key themes and related codes were refined through an iterative approach of generating researcher notes and discussions between the researchers before a coding frame was agreed upon.Coding was completed by applying the developed framework to all the transcripts and data, with 10% double coded and compared (with a high level of agreement (90%+) found) to ensure consistency between researchers.Thereafter, the data was charted by generating a matrix of cases (school case and participant groups) and codes for each of the key themes identified.Interpretation and mapping were guided by the original research objectives and carried out by reviewing the data contained in the matrices and making connections within and between schools and research participants.

Ethical considerations
Ethical approval was granted from Amref International University Ethics and Scientific Review Committees, and research permits were acquired from the National Council for Science, Technology and Innovations (NACOSTI).Upon securing local approvals, additional ethical approval was granted by the University of Liverpool Research Ethics Committee.Permission to collect information from the schools was also obtained from the MoE County Director of Education and County Government of Nairobi.Full informed consent to carry out the study was given by the three participating schools and participants involved in each aspect of data collection.Data collected were only available to the investigator and research team and were managed and analysed according to strict data management protocols (general data protection regulation compliant).

Cooking practices and kitchen characteristics
The characteristics of the school kitchen and cooking practices from each school are described in table 2. Differences in cooking practices between schools, appeared to reflect differences in school governance and financing.
School 1 (S1:Wood) used wood for almost all cooking, with LPG used occasionally when catering during board of management meetings.Fuel consumption increased in rainy seasons when wood was often wet.The school had two kitchens, the main kitchen that catered for the primary school students and a second kitchen providing for the pre-school students.The main kitchen was a permanent structure constructed from concrete blocks and a corrugated iron roof, with total floor area of 64 m 2 and with two windows and two doors that provided ventilation.Cooking took place on large industrial-sized stoves with insulated pots, and an integrated chimney system (figure 2(A)).Githeri (a mixture of maize and beans) was provided throughout the week as the main school meal at lunch times, with students catered for if they had paid the termly fee of 100Ksh to match the government funds.Catering staff started work between 07:00 and 08:00, spending around 5 h in the kitchen.The mean stove use time for each cooking event was around 2 h (118 min), as measured through the SUM.
School 2 (S2:Charcoal) used charcoal in the main kitchen for catering for students and charcoal and LPG in a second kitchen for staff and commercial catering.Charcoal briquettes were commercially produced on site through a dedicated production process, where charcoal dust was purchased and then combined with waste paper from the school to form briquettes (figure 2(C)).-Thisprovided an income source for the school, through producing roughly 1/3 more briquettes than is used by the school.The kitchen was largely an open structure constructed from open metal framing (figure 2(B)).School meals were cooked in large pots and meals varied from enriched githeri (added pumpkin to githeri), ugali and vegetables, and rice and beans.Due to external support from a faith-based donor, the school was able to provide a more varied menu for students.Catering staff started work between 07:00 and 9:00, typically spending 5-6 h in the kitchen cooking and each stove use event taking roughly 2 h.School 3 (S3:Wood) used wood for main catering needs with some supplementary use of LPG.The main kitchen was a permanent structure constructed from concrete blocks and corrugated iron roof and total floor area of 56 m 2 , with three windows and a door that provided ventilation.Cooking took place on large, insulated pots with a chimney system (similar to those shown in figure 2(A)) and meals were typically githeri.Cereals were soaked in water overnight to soften in readiness to cook the following day.Catering Staff started work at between 07:00-08:00 and spent between 5-10 h in the kitchen, with each stove use event under 2 h.

Air pollution concentration in school micro-environment and cook exposure
Across all schools, 24 hr PM 2.5 concentrations exceeded the WHO-interim target level 1 (WHO IT-1) of 35 µg m −3 .Levels were three times higher than the WHO IT-1 in classrooms with a mean of 107.6 µg m 3 (SD 166.9) recorded, 10 times higher (mean 316.2 µg m −3 , SD 316.2) in kitchens and nearly six times higher for cooks with a mean exposure of 200.9 µg m 3 (SD 969.3) (table 3).Mean ambient levels (78.4 µg m 3 ), collected in school playground areas, were twice the WHO IT-1.CO 24 hr concentrations exceeded WHO recommended levels (6.11 ppm, ∼7 mg m −3 ) in kitchen locations only, with a mean concentration of 7.8 ppm across the three schools.
Across all schools, PM 2.5 concentrations were higher in kitchens and classrooms and for the cook's exposure during school opening times (daytime hours) and during times of stove use.In classrooms, during school opening hours, mean PM 2.5 levels were 134.5 µg m 3 (SD 249.8) compared to 87.8 µg m −3 (SD 38.2) when closed, and when the stove was in use, mean levels rose to 224.4 µg m −3 (SD 249.8) compared to 86.3 µg m −3 (SD 54.9).In kitchens, when the stove was in use-mean PM 2.5 concentrations were 676.2 µg m −3 (SD 1296.9)compared to 262.1 µg m −3 (SD 1068) when the stove was not in use; similarly, during school hours mean concentrations were 520.1 µg m −3 (SD 1525.1)compared to 147.6 µg m −3 (SD 511.4) outside school hours.The main cook PM 2.5 exposure followed a similar pattern to the kitchen location, with a mean exposure of 682.1 µg m −3 (SD 2184.2) during stove use and 122.8 µg m −3 (SD 522.2) when no stove use, and a mean exposure of 359.2 µg m −3 (SD 1432.1)four-fold higher during school hours compared to outside school hours (mean 79.4 µg m −3 (SD 229.1)).Ambient levels were highest during school opening hours but not during stove use (102.7 µg m 3 SD 41.9), highlighting the likely contribution of other sources of pollution, or a delay in the impact from stove use as the particulates dissipate in the environment.
There were similar trends for CO concentrations, with levels higher during stove use and school opening hours.CO levels in classrooms were nearly two times higher during stove use (3.91 ppm, SD 8.5) compared to periods of no stove use (2.13 ppm, SD 10.42) and in kitchens, levels were nearly three times higher during stove use (22.6 ppm, SD 40.5) than non-stove use times (8.14 ppm, SD 36.01).
The contribution of stove use to air pollution concentrations in the school can be seen in figure 3(A), which depicts the 24 hr variation from selected monitoring days in the school micro-environments and the cook's exposure.Figure 3(A) reveals how peak PM 2.5 and CO levels correspond with stove use events, with levels relatively stable until fluctuating substantially during stove use.This is further highlighted in figure 3(B), where concentrations from the selected days are split between stove use and non-stove use time, revealing large differences in PM 2.5 and CO concentrations, especially for wood using schools (S1:Wood & S3:Wood).
Air pollution levels differed between schools, with S1:Wood (primary fuel wood) generally having higher PM2.5 and CO concentrations in the kitchen and classroom and for the cook's exposure compared to concentrations in S2:Charcoal and S3:Wood.S2:Charcoal, the charcoal using school, was found to have the lowest 24 hr PM2.5 concentrations in classrooms and the kitchen, 93.5 µg m 3 and 155.2 µg m 3 , respectively, with the cooks average mean exposure 90.1 µg m 3 (226.8),with levels similar to background ambient concentrations.In S2:Charcoal, the impact of stove use also was found to be less dramatic with cook exposure to PM2.5 during cooking ∼50 µg m 3 more than during no stove use periods.This is contrasted with S1:Wood and S3:Wood where the cook exposure was ∼600 and ∼1500 µg m 3 higher during periods of stove use compared with periods of no stove use respectively.However, CO levels were fairly similar between S2:Charcoal and S3:Wood.

Impacts on catering staff
Fourteen out of 18 catering staff acknowledged the need to step out of the kitchen and all surveyed catering staff reported limiting time around the stove to minimize their exposure to air pollution (table 4).Almost half (8 out of 18) of the catering staff perceived the air quality in the school environment to be dirty or very dirty, with just 4 reporting the air quality as clean.Three-quarters (12/18) of the participants believed that their health issues were a direct result of smoke exposure.
In the FGDs, catering staff discussed concerns about the impacts of air pollution from cooking.Staff using wood for cooking reported experiencing discomfort from the smoke on a daily basis, with impacts more severe in morning hours when the firewood was first lit.The level of smoke would vary seasonally and be particularly intense in rainy seasons, if the wood had not dried fully or had been rained on:   For S1:Wood, due to issues with data quality, it is not possible to report ambient levels.Additionally, mean PM2.5 levels are representative of one day for the kitchen and of two days for the cook and classroom and mean CO levels for the cook represent two days only.
b For S3:Wood, mean PM2.5 levels are representative of two days for ambient, cook and kitchen (the combination of days is different for each micro-environment)."I have challenges with cooking using firewood due to smoke produced, sometimes it worsens especially when the firewood is not well dried, and also lighting it up is very stressful… is on a daily basis, especially in the morning when you are lighting up fire, there must be smoke before the firewood stabilizes and is well lit… and it is worse during rainy season because of the nature of our [wood] store."FGD S3:Wood -Catering Staff Smoke levels had caused participants to take measures to reduce exposure and alleviate the adverse health impacts.Staff explained how they regularly stepped outside the kitchen, sat on the floor below the smoke level and drank milk to ease the effects of smoke inhalation: 'when we leave here like me, you feel like the chest is burning; it forces me to look for milk so that I can calm it' (FGD S1:Wood Catering Staff).
Catering staff from the charcoal school (S2:Charcoal) felt that the fuel did not contribute to poor air quality and that chimneys dissipated any smoke.Although one respondent acknowledged that they could not be certain of the air quality as they did not measure emissions levels: "…we don't have those devices for measuring the air but depending on how our kitchen is built, there is…we are using charcoal, yes, but that charcoal is made in such a way the smoke goes up.So, for us who are inside, we don't feel it.We have a chimney that takes the smoke up and we also have that thing of fanning the smoke…" FGD S2:Charcoal Catering Staff.
Of the 18 cooks taking part in the survey, nearly all reported common respiratory symptoms known to be associated with cooking smoke (table 4).Over half (10/18) reported wheezing, half (9/18) chest tightness, two-thirds (12/18) eye irritation, and nearly all (16/18) headaches more than once a week.Additionally, half of the respondents complained of waking up at night with cough.
Catering staff from S1:Wood discussed how the smoke from firewood triggered allergies and caused itchy eyes, which led to sticky eyes when waking up: 'I wake up and it is [the eyelids] have stuck together' (FDG S1:Wood Catering Staff).They also reported long-term difficulty with breathing, experiencing frequent cough and congested chest, which affected their sleep.One catering staff member complained of producing black sputum when coughing: 'When I cough, I just cough black sputum' (FDG S1:Wood Catering Staff).
Staff explained that their health had deteriorated since coming to work at the school and expressed concern that it was continuing to be affected: 'Our health continues to deteriorate day by day' (FDG S1:Wood Catering Staff), with some raising very serious concerns about their fate: "When I started working here am not like the way I was in the beginning, so sometimes I feel my body has complications, sometimes I feel the chest, sometimes when I cough I breath out… I cough black sputum.We don't know if our chests are rotting or what is it?You know now we are concerned about our health we just say oh God please open for us doors somewhere else… Because this smoke will finish us….".FGD S1:Wood Catering Staff As well as being less concerned about cooking smoke, cooks using charcoal (S2:Charcoal) appeared to report less health effects.On probing specific symptoms reported in the survey, one participant expressed that eye irritation could have been a result of other causes, such as old age: 'Some of us feel our eyes are weakening, and we say that it is old age' (FGD S2:Charcoal Catering Staff).Another participant narrated that chest pains were due to the cooking pots that are heavy and strenuous to carry, which caused 'what I have felt in the chest is because those pots are usually heavy….sowhen we wash then put them on the stove, it usually affects the chest a lot, you find that between the chest and stomach, there is a problem, pain' (FGD S2:Charcoal Catering Staff).

Impacts on students
Concerns were expressed by management, teachers and catering staff from schools (S1:Wood and S3:Wood) that air pollution was having a negative impact on their students, impacting their health, well-being and education, both directly through exposure to poor air quality as well as indirectly through disruption of teaching.
Teachers reported direct impacts on student health, associating air pollution with different symptoms, including chest pains, coughing and flu-like symptoms: 'Students' most of the time complain of chest pains'.(FGD S1:Wood Senior Staff).Teachers recognised that the impacts were often not visible initially, as health effects were longer-term, making the impacts difficult to identify "Children complain a lot of flu, not sure if connected to air pollution.A lot of them [the students] say their chest is blocked, and complain of cough, it like an allergy.[It's] very difficult to explain the impacts of air pollution, as a "some type of slow killer" where you get diseases."FGD S3:Wood Teachers Participants explained how smoke generated from the cooking diffused into teaching and shared spaces in the school: 'Depending on the intensity, sometimes it goes beyond the kitchen area, we have bordering classrooms that also get affected, and sometimes it even crosses to the pathways'.(FGD S3:Wood Catering Staff).Teachers described how, at times, the smoke caused disruption to teaching activities.One teacher described that they moved the class outside due to smoke emitted from firewood (weather permitting): "If there is a class, they must go out like 1 h (due to the smoke).It depends on the firewood that is being used.There are some types of firewood they are using that are okay, the smoke is not much.But when it is raining, we must feel the smoke".FGD S3:Wood Teachers Concerns about smoke produced from cooking and its effects on students had been raised by the surrounding community: 'Even the people outside usually think that the school is burning, and most of the time they usually ask, today the school is burning?And it is just the way that smoke comes out.' (FDG S1:Wood Catering Staff) and had been the subject of meetings with parents: "If you have walked around you have seen that our kitchens are just next to our classrooms and imagine those teachers and students and the quantity of smoke from preparing 1000kgs of maize every day, that is a lot of air pollution in those classrooms and compound where they are so that's a big concern to me and we have kept discussing with parents meetings over the same to see if we can one day obtain an alternative.It is a great concern" IDI S1:Wood Headteacher charcoal will likely result in adverse health impacts and other fuels that are classified as clean for health (e.g.electric cooking, LPG) should be prioritised.
Our study provides a snapshot of air quality for selected schools in a Kenyan informal settlement.Whilst there is limited published evidence on air quality from schools in sub-Saharan Africa, levels of PM 2.5 observed in the classrooms from our study appeared nearly two and a half times higher than the annual 24 h mean reported (43.34 µg m −3 ) in a study of air quality in a school in Kigali, Rwanda (also in East Africa and used wood in the school kitchen) [21].Unsurprisingly levels are substantially higher than those reported in studies of schools in high-income countries [21].The current study adds to previous research, for example, a comprehensive review on air quality in schools reports that PM exposure is largely driven by pollutants in the outdoor air (from sources such as traffic) [41] but studies have seldom considered the influence of cooking fuel in the school on air pollution.While we cannot different sources to the air pollution concentrations or confirm the true ambient levels around the school environment, our study underscores the importance of considering fuels used within the school environment, particularly in LMIC settings that tend to rely on polluting solid fuels for cooking.
Pollution generated from the cooking fuels constitutes a major occupational hazard for school staff, with the catering staff bearing the brunt of exposure to smoke emitted during stove use.PM 2.5 exposure for the main cook was four-fold higher during school hours than outside school hours, illustrating the contribution of the school environment to their overall exposure.Catering staff attempted to minimize the time they spent around the cookstove, often leaving the kitchen when they were able for short periods to escape the smoke.Catering staff experienced a high occurrence of self-reported health symptoms, with over half reporting wheezing, tightness in the chest and waking up at night with cough.Additionally, nearly all staff experienced mild headaches and moderate eye irritation most days.These findings align with those of other cross-sectional studies, which have linked exposure to household air pollution from cooking with polluting fuels to acute respiratory health symptoms such as wheezing, frequent coughing, and shortness of breath [42,43].Catering staff from both wood-using schools expressed serious concerns about the health impacts from cooking in school and recognised their health had deteriorated since working there.They worried about their future health due to continuous exposure to smoke.However, catering staff from the charcoal-using school did not connect their health symptoms to cooking activities in the school and were less concerned about levels of air pollution in the school, this is possibly due to relatively lower levels of PM 2.5 compared to firewood.Transitions to clean fuels (such as LPG and electricity) in schools are likely to benefit health; however messaging on the benefits to health may be more difficult to use in schools currently using charcoal-as although air pollution levels were still concerning, charcoal was seen as relatively clean compared to wood fuels by school staff members-and other benefits (climate, environment and costs) be more effective messages.
For children attending school, exposure to air pollution from cooking will limit the benefits of using clean fuels at home for cooking (with the majority of students (64%) surveyed reporting LPG as the main cooking fuel at home, see supplementary material -table S1).Despite the high level of LPG use at home, the majority of students (62%) regarded the school air quality to be better than compared to their home environment.This might be due to proximity to other sources and local surroundings at home, which are perceived to be polluted.However, school teaching staff were aware of the adverse health and social effects of air pollution from cooking on student health and reported substantial disruption to learning environments.Teachers from wood-using schools reported various health symptoms in students, such as chest pains, coughing, and flu-like symptoms and smoke generated from cooking caused substantial disruption to teaching activities.With 36% of Kenyans (∼22 million people) of pre-primary, primary and secondary school age (4-17 years old) [22] and almost all (99%) schools currently using wood for cooking [25], young Kenyans are likely to be exposed to high levels of particulate matter at school and at home, impacting their health and educational development.
Despite the small-scale study and limited air pollution monitoring conducted in each school, evidence from our research highlights the impacts of polluting fuels on school air quality and health and reveals notable concerns from teaching staff about the potential disruption to educational activities from poor air quality.Monitoring was only conducted for three 24 hr periods and in limited locations (main kitchen and nearest classroom) and therefore represents the worst-case scenario, rather than mean concentrations across classrooms in each school.Additionally, SUM was only conducted on the main stove and therefore may not fully capture stove use periods.The survey data was self-reported responses, and although was anonymous, may have reporting bias: Particularly, for catering staff as the survey was conducted in the school environment, and they may not have wanted to highlight certain issues or behaviors (such as leaving the kitchen due to smoke) in their place of employment.Cross-sectional and longitudinal research in additional schools, and conducted over longer periods with repeated measurements and a larger representation of classrooms, would allow better understanding of how air pollution levels vary across classrooms, across schools in different locations in Kenya and between seasons.A larger sample size would be needed to identify any statistically significant differences in pollution concentrations between schools using different fuel for cooking.School absenteeism and performance data and objective measurements for health (such as hospital attendances for respiratory illness) could help provide better evidence on the impacts of air pollution in the school environment.
It is increasingly understood that household transitions to cleaner cooking options such as gas and electricity will likely have multiple benefits for health, the environment and the climate [8].This study highlights the importance of clean cooking transitions for educational institutions that can have additional benefits for health and the environment (including deforestation) but also for education.With already limited budgets and resources [29], Kenyan public schools ideally need support to make switches and require the close engagement of stakeholders such as the MoE and Ministry of Energy, school management, finance and clean fuel technology providers among others.Where wood is purchased (often in urban locations), it could well be cheaper to use LPG.A recently launched programme by a development bank in Kenya, provides finance mechanisms for schools to purchase LPG equipment and fuel for clean cooking, with fuel costs savings enabling loan repayment [32].Whilst promising, it is important that the scheme is effectively evaluated to achieve scale, particularly for the more resource poor schools.Clean cooking conversion programs should ensure that no schools are left behind-especially those schools with less financial means and where the school population is likely to be more reliant on polluting fuels at home.Despite the MoE's efforts to ensure a healthy meal for all Kenyan students via the national school feeding programme, focus must be also given to providing a healthy and safe learning environment.

Conclusion
This pilot study highlights the potential detrimental effects from reliance on polluting solid fuels for cooking in schools.The presence of high levels of PM 2.5 and CO above WHO recommended air quality guideline levels for health poses serious health risks to catering staff, students and teaching and administrative staff, in addition to detriments to education through poor air quality.Our findings underscore the urgency of transitioning schools to modern and cleaner cooking options to improve air quality, safeguard health, and enhance the overall learning environment.Such a shift would likely have multiple benefits, positively impacting both the students' well-being and the delivery of educational activities in addition to positively impacting the environment through reductions in deforestation from reliance on fuel wood.Urgent support from government and international donors is needed to facilitate transitions to clean cooking technologies, and in particular to ensure poorer schools are able to access clean cooking options.

E Nix et al Figure 1 .
Figure 1.Location of kitchen, classrooms and monitoring for each school.

Figure 3 .
Figure 3. (A) −24 h variation in PM2.5 and CO levels (ambient, classroom, kitchen concentrations and cook exposure) during Day 3 of monitoring for each school).(B) -Mean concentrations during stove use and no stove use periods (horizontal lines depict median levels and error bars represent the standard deviation).

Table 1 .
Overview of the study schools.

Table 2 .
Cooking practices, fuel consumption & procurement and kitchen/stove characteristics.

Table 3 .
Mean PM2.5 and CO concentrations in school micro-environments.

Table 4 .
Catering staff health symptoms and air pollution perceptions.