A Descriptive Assessment of Household Air Pollution in Rural Kitchens in Kenya

Abstract

Efforts to ensure households transition to modern fuels are expected to reduce household air pollution. However, exposure to toxic particles and gases in fuel stacking households remains under-researched. We implemented a household survey to identify household energy sources and assess exposure to particulate matter with diameter of ≤5 microns (PM_2.5), ≤10 microns (PM₁₀) and select polluting gases (Sulfur Dioxide (SO₂), Total Volatile Organic Compounds (TVOCs), Carbon Dioxide (CO₂), Nitrogen Dioxide (NO₂), Carbon Monoxide (CO)) in a rural community. Wood was the main cooking fuel in 94.2% (1615/1703) households with fuel stacking reported in 86.1% (1462/1703) of total households. Daily time-weighted average concentrations of PM_2.5 and PM₁₀ were beyond World Health Organization (WHO) limits in wood-using households (189.53 (Standard deviation (SD) = 268.80) µg/m³ and 592.38 (SD = 623) µg/m³, respectively) and Liquid Petroleum Gas (LPG) -using households (57.2 (SD = 53.6) µg/m³ and 189.86 (SD = 168) µg/m^3, respectively). Only daily average CO and TVOC concentration in wood-using households exceeded recommended levels. Household socio-economic status, education level of the head of household, use of a separate kitchen and household size influenced household energy choices. Rural households using wood as the main cooking fuel are exposed to high levels of particulate matter, carbon monoxide and total volatile organic compounds. LPG-using households may not realize health benefits if stacking with polluting fuels is practiced.

1. Introduction

A third of the global population relies on biomass and kerosene to meet their cooking and heating energy needs [1]. In Kenya, combined estimates indicate that 74.5% of households use biomass and kerosene for cooking, with rural areas having the highest proportion (93.4%) compared to urban areas at 44.6% [2]. This is despite growing concerns on the effect of biomass fuels and kerosene on household air quality and its associated health and non-health impacts.

The responsibility for collection, transportation and use of wood fuel largely lies on women and children. They not only bear the largest burden of exposure to risks during wood collection and transportation [3], but also endure toxic kitchen emissions associated with traditional fuels, with added implications for health including adverse pregnancy outcomes [4,5,6,7,8]. Most times, women and young girls are also the caretakers of young children and spend time in kitchen environments causing early life exposure to air pollutants which may influence subsequent health outcomes [9].

The combustion of biomass fuels in rudimentary stoves such as the three-stone stove commonly found in rural African homes, has been associated with high levels of pollutant emissions. Evidence from rural Kenya show that women and girls were expose to extremely high levels of particulate matter [10,11] as well as high levels of carbon monoxide [12,13]. These studies reported peaks in pollutant concentrations during cooking episodes, indicating higher personal exposure for cooks and other household members present in the kitchen environment during cooking. Similarly high levels of kitchen emissions have been documented in other countries across Africa [14,15,16].

This study seeks to characterize the fuel-stove mix in a rural community in Eastern Kenya, and profile fine particulate matter and gaseous pollutants in kitchens.

2. Materials and Methods

2.1. Study Design

We use cross-sectional data from a baseline survey targeting 2000 households sampled using a two-stage approach. We implemented a household baseline survey to assess current household sources of energy and willingness to shift to cleaner options.

We purposively sampled 20 households from the survey participants based on primary cooking fuel to ensure representation of the range of fuels used in the community (

Table A1. Distribution of primary and secondary fuels in households recruited for air quality monitoring.

Primary Fuel	Secondary Fuel					Total
	Wood	Charcoal	LPG	Kerosene	None
Wood		5			3	8
Charcoal				3		3
LPG	1	3		1		2
Kerosene		2				5

). In addition, we considered equipment safety and ease of access to the households for deployment of the monitoring equipment.

The Handheld 3016 IAQ^TM Airborne Particle Counter (Lighthouse Worldwide Solutions, Medford, MA, USA) uses a laser-diode light source and collection optics for particle size detection (particle size range 0.3–10 μm). Particulate matter was monitored using the Lighthouse Handheld 3016 IAQ Airborne Particle Counter set in mass concentration mode, and mass concentration logged every minute. Purging procedure was done weekly according to manufacturer instructions.

The particle counter was anchored on a Wolfpack^® Modular Area Monitor as an integrated system via the respective brackets. Each instrument logged data separately and left to run for at least 12 h to cover the typical three cooking periods in a day. The Wolfpack^® Modular Area Monitor had two GrayWolf DirectSense^® probes plugged in and simultaneously monitoring Sulfur Dioxide (SO₂), Total Volatile Organic Compounds (TVOCs), Carbon Dioxide (CO₂), Nitrogen Dioxide (NO₂), Carbon Monoxide (CO), temperature and Relative Humidity. After setting up the Wolfpack unit, probes were allowed to stabilize, and data logged at one-minute intervals. All monitors were placed one meter off the ground and a similar distance from the cookstove.

For each household, air quality monitoring was conducted every other day over a 7-day period to allow for an assessment of variation in emission levels between weekdays and weekends. This translated to four days of monitoring in a week for most households (

Table A2. Hours per day when particulate matter monitoring was done in different households.

Primary Fuel	HH No	Hours Monitored Per Day				Days Monitored
Wood		Day 1	Day 2	Day 3	Day 4
	4 *	24				1
	5	23	23	24	24	4
	6	24	24	24		3
	7	18	24	16	24	4
	8 *	23	11			2
	9	24	24	24		3
	11	24	24	24	22	4
	12	24	23	23	24	4
	13	24	24	24	19	4
	20	24	24	24	24	4
Charcoal	14	21	24	24	24	4
	17	24	24	24	24	4
	19	24	20	15		3
LPG	1	24	24	22	24	4
	2	22	24	24		4
	10	24	24	20		3
	15	22	24	24	20	4
	18	12	16	12	8	4
Kerosene	3	24	24	24	24	4
	16	24	24	24	24	4

* Denotes households that were not included in analysis due to reduced days of monitoring.

), totalling to 71 days or 1505 h of monitoring. Monitoring ran from February to April 2021, which coincided with the end of the dry season (in mid- March) and the long rains (late March/April). It is worth noting that 2021 long rains in the area failed and the period was relatively dry.

2.2. Kitchen Characteristics

Most of the households sampled for air quality monitoring had a separate kitchen (stand-alone house) which was mostly a one-roomed structure with earthen floors. A traditional three-stone stove (Figure 1a) or a variation of this with the stones covered with mud (Figure 1b) was the most common type of cooking stove in the area. In a few of the sampled homes, the kitchen was part of the main house/room.

2.3. Data Analysis

Survey data was analysed using Stata software version 15.1 [17] to produce descriptive tables using the svy command, after applying sample weights. Analysis of fine particulate matter was done separately from gas pollutants. We excluded from the analysis two households that were monitored for less than three days and computed time weighted daily average concentrations of particulate matter, CO, SO₂, NO₂, CO₂ and TVOCs by fuel types as well as the overall average over the monitoring period for each household. Average CO concentrations were computed at 24 h, 8 h, 1 h, and 15 min intervals. The Kruskal-Wallis rank-sum test method is suitable for testing differences in particulate matter concentrations among fuel types (4 levels) because particulate matter observations are independent in each fuel type and the distribution type of PM_2.5 and PM₁₀ data is unknown. Kruskal-Wallis test was used to check for differences in PM_2.5 and PM₁₀ concentrations between different types of fuel at a significant level of 5%.

3. Results

3.1. Household Cooking Fuels

Table 1. Adjusted distribution of primary and secondary fuels.

Primary Fuels	Wood	LPG	Charcoal	Kerosene	Total
Percent (%)	94.2	2.4	3	0.4		100
Frequency	1615	36	46	6		1703
Secondary fuels *	Wood	LPG	Charcoal	Kerosene	Electricity	Ethanol
Percent (%)	3.1	39.5	84.8	17.6	0.1	0.1
Frequency	45	576	1239	256	1	2
	Households using secondary fuels			Households not using secondary fuels
Percent	86.1			13.9
Frequency	1462/1703			241

* Total percent is more than 100% due to multiple responses.

presents the range of primary and secondary fuels used in the study community.

Secondary fuels were used alongside the primary fuel in 1462/1703 (86.1%) households, reportedly for their ability to cook fast in 927/1703 (63.6%) households.

We present stacking behaviour in households and coin terminologies for different fuel combinations as described and shown below (Figure 2). In this paper, we use the terms “higher rung” and “lower rung” fuels to refer, respectively, to fuels that fall higher on the fuel ladder such as LPG, and those falling lower on the ladder such as firewood. Results show up-stacking (use of a “lower rung” primary fuel and a “higher rung” secondary fuel) accounting for 8.2% while down-stacking (use of a “higher rung” primary fuel and a “lower rung” secondary fuel) accounting for 2.3%. In addition, 58.7% of households used “lower rung” fuels as both primary and secondary fuels- in what we term as horizontal stacking while 30.8% of households used both “higher rung” and “lower rung” secondary fuels (we coin the term partial up-stacking for such households).

3.2. Kitchen Emissions Levels

The levels of particulate matter are presented in

Table 2. Comparison of mean PM concentrations by fuel types.

Fuel	Particle Size	Mean Concentration (µg/m3)				Four Days’ Mean (SD)	p Value * (Wilcoxon Test)
		Day 1	Day 2	Day 3	Day 4
LPG	PM2.5	82.77	49.47	48.10	45.03	57.18 (53.6)
	PM10	234.10	182.91	155.47	186.62	189.86 (168)
Charcoal	PM2.5	15.47	25.35	15.95	15.74	18.41 (39.2)
	PM10	106.29	496.58	857.78	499.73	470.62 (392)	(p < 0.001)
Kerosene	PM2.5	97.30	79.20	108.91	115.61	100.25 (15.7)
	PM10	254.37	221.37	483.32	351.20	327.56 (155)	(p < 0.001)
Wood	PM2.5	200.00	189.42	246.53	106.67	189.53 (268.8)
	PM10	617.78	922.20	436.44	326.93	592.38 (623)	(p < 0.001)

* LPG is the reference group.

below across the different primary cooking fuels reported. We present hourly and daily variations of PM_2.5 and PM₁₀ by fuel type. Emission trends of SO₂, NO₂, CO and TVOC are also shown making comparison with WHO guidelines for short-term exposure (24 h).

Time-weighted average (TWA) daily concentrations of PM_2.5 and PM₁₀ was highest in wood-using households and varied significantly (χ² (2) = 22,978, df = 3, p < 0.001 and χ² (2) =8677.4, df = 3, p < 0.001, respectively) between different fuel types. We compare PM_2.5 and PM₁₀ levels in households that use LPG as their primary fuel with households using other primary cooking fuels.

PM_2.5 hourly mean concentrations were high in wood- and kerosene-using households with consistent peaks in the morning and evening hours, typical cooking time among households in the study area (Figure 3). Households using charcoal and wood had high PM₁₀ hourly mean concentrations with peak concentrations coinciding with morning and evening cooking times (Figure 4). PM₁₀ concentration in LPG-using households was low and stable. PM_2.5 and PM₁₀ concentration varied by day of the week and was stable in kerosene- and LPG-using households (Figure 5 and Figure 6). In some of the days, PM_2.5 and PM₁₀ concentration spikes were observed in charcoal- and wood-using households. This could point to households switching between the two fuels with lower use of charcoal accompanied by increased use of wood over the weekend. This potentially points to availability of children at the end of the school week, who can accompany older females to collect more firewood for use over the weekend.

Daily mean SO₂ and NO₂ concentrations were below 1µg/m³, which is indicative of minimal to zero traffic-related sources (

Table 3. (a): Mean 24 h concentrations (µg/m³) of SO₂, and NO₂. (b): Mean CO concentrations against WHO limits.

(a)
Pollutant	Fuel Type	Mean	SD	Min	Max	WHO AQGs *
SO2	Wood	0.17812	0.49526	0.00	8.750
	Charcoal	0.01044	0.02028	0.00	0.450	40 µg/m3
	LPG	0.03435	0.05445	0.00	0.550
	Kerosene	0.21717	0.77785	0.00	8.130
NO2	Wood	0.02295	0.12354	0.00	2.710
	Charcoal	0.00107	0.00633	0.00	0.110	25 µg/m3
	LPG	0.00250	0.01273	0.00	0.260
	Kerosene	0.01216	0.02146	0.00	0.240
(b)
Fuel Type	Mean CO Concentration (µg/m3)
	24 h	8 h	1 h	15 min
Wood	24.48	24.7	24.7	24.7
Charcoal	3.70	3.29	3.29	3.30
LPG	2.27	1.40	1.40	1.40
Kerosene	6.42	7.02	6.73	6.71
WHO limit (µg/m3)	4	10	35	100

* AQGs denotes Air Quality Guidelines.

). CO emissions across different cooking fuels was low and within recommended levels (Table 3).

4. Discussion

Our study has demonstrated the dominance of wood (94.2%) as a primary source of cooking energy for households in this rural community, while also showing the high prevalence of stacking with charcoal being the most preferred secondary fuel. Recent findings from a cooking sector study showed that 86% of rural households relied on firewood for cooking [18].

We present the levels of gaseous emissions and particulate matter in households using wood, charcoal, LPG or kerosene as their primary cooking fuel. Daily average concentrations of PM_2.5 were high and beyond WHO limits in households using wood fuel, kerosene and LPG. Studies have found wood using homes to have extremely high particulate pollution, in some cases exceeding guideline levels in the order of 100 or higher [10,11,19]. This underscores the potential risk to health in an area where 94% of households rely on wood as their primary fuel. Although LPG is a more processed fuel compared to traditional cooking fuels like wood, PM_2.5 levels in LPG using households were considerably high in this setting. This is likely due to fuel stacking practices common in rural households where emissions from traditional cooking fuels compound exposure to fine particles. Similar findings were reported for urban slum households where wood, charcoal and kerosene using homes had the highest PM_2.5 levels while LPG and electricity users also had high levels [20]. In addition, most kitchens in the study community have earthen floors which when swept contribute to particulate pollution from dust.

Gaseous pollutants were below guideline levels. The low prevalence of LPG as well as the rural nature of the study community where traffic influence on indoor levels of SO₂ and NO₂ was very low explain these findings.

The spatial orientation and location of rural kitchens is a key factor to consider in characterizing collective household exposure to fine particles and toxic gases. Rural households in this community have a separate kitchen a few yards from the main house as a ‘corrective strategy’ to reduce exposure to other household members during cooking times, with the cook having the highest exposure. Therefore, these emission profiles provide crude estimates and are not an accurate measure of personal exposure to fine particles and toxic gases.

Limitations

We acknowledge the following limitations in our study:

• Emissions were not analysed separately when households switched between fuels to characterize confounding secondary fuels.
• We present uncorrected values since we did not collect data to compute a correction factor.
• Kitchen HAP levels estimated in this study may not reflect actual exposure to individual household members. Monitoring personal exposure to fine particles and gaseous pollutants would give more accurate exposure assessment and shed light into the effectiveness of building kitchens a few yards from the main house.
• The study did not assess the emissions associated with the use of LPG

We are, however, confident that the findings of this study present critical evidence needed for action to improve air quality in rural kitchens and in effect protect the health of the most exposed household members i.e., women and children.

5. Conclusions

Rural households in this study community primarily depend on wood fuel for cooking, exposing household members to fine particulate matter and toxic gases. To reduce HAP in rural households, interventions that encourage the transition to affordable cleaner fuels need to be scaled up.