Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Mar 1.
Published in final edited form as: Environ Pollut. 2019 Dec 4;258:113697. doi: 10.1016/j.envpol.2019.113697

Kitchen concentrations of fine particulate matter and particle number concentration in households using biomass cookstoves in rural Honduras

Megan L Benka-Coker a,b, Jennifer L Peel b, John Volckens c, Nicholas Good c, Kelsey R Bilsback c, Christian L’Orange c, Casey Quinn c, Bonnie N Young b, Sarah Rajkumar b, Ander Wilson d, Jessica Tryner c, Sebastian Africano e, Anibal B Osorto f, Maggie L Clark b
PMCID: PMC7068841  NIHMSID: NIHMS1547969  PMID: 31875572

Abstract

Cooking and heating with solid fuels results in high levels of household air pollutants, including particulate matter (PM); however, limited data exist for size fractions smaller than PM2.5 (diameter less than 2.5 μm). We collected 24-hour time-resolved measurements of PM2.5 (n=27) and particle number concentrations (PNC, average diameter 10–700 nm) (n=44; 24 with paired PM2.5 and PNC) in homes with wood-burning traditional and Justa (i.e., with an engineered combustion chamber and chimney) cookstoves in rural Honduras.

The median 24-hour PM2.5 concentration (n=27) was 79 μg/m3 (interquartile range [IQR]: 44 – 174 μg/m3); traditional (n=15): 130 μg/m3 (IQR: 48 – 250 μg/m3); Justa (n=12): 66 μg/m3 (IQR: 44 – 97 μg/m3). The median 24-hour PNC (n=44) was 8.5 × 104 particles (pt)/cm3 (IQR: 3.8 × 104 – 1.8 × 105 pt/cm3); traditional (n=27): 1.3 × 105 pt/cm3 (IQR: 3.3 × 104 – 2.0 × 105 pt/cm3); Justa (n=17): 6.3 × 104 pt/cm3 (IQR: 4.0 × 104 – 1.2 × 105 pt/cm3). The 24-hour average PM2.5 and particle number concentrations were correlated for the full sample of cookstoves (n=24, Spearman ρ: 0.83); correlations between PM2.5 and PNC were higher in traditional stove kitchens (n=12, ρ: 0.93) than in Justa stove kitchens (n=12, ρ: 0.67). The 24-hour average concentrations of PM2.5 and PNC were also correlated with the maximum average concentrations during shorter-term averaging windows of one-, five-, 15-, and 60-minutes, respectively (Spearman ρ: PM2.5 [0.65, 0.85, 0.82, 0.71], PNC [0.74, 0.86, 0.88, 0.86]).

Given the moderate correlations observed between 24-hour PM2.5 and PNC and between 24-hour and the shorter-term averaging windows within size fractions, investigators may need to consider cost-effectiveness and information gained by measuring both size fractions for the study objective. Further evaluations of other stove and fuel combinations are needed.

Keywords: household air pollution, solid fuel, particulate matter, ultrafine particles, real-time measurements

Graphical Abstract

graphic file with name nihms-1547969-f0001.jpg

1. Introduction

Approximately three billion people, predominantly in low- and middle-income countries, rely on solid fuels (e.g., wood, charcoal, dung) as their primary energy source for cooking (Bonjour et al., 2013). Combustion of solid fuels often results in levels of household air pollution that exceed World Health Organization (WHO) air quality guidelines (e.g., 24-hour mean PM2.5 concentrations greater than 25 μg/m3) (Thomas et al., 2015; World Health Organization, 2006a). This household air pollution is one of the top environmental risk factors for the global burden of disease and was estimated to be responsible for 1.6 million deaths and 59 million disability-adjusted life years in 2017 (Stanaway et. al, 2018).

Human exposure to particulate matter air pollution is typically assessed using gravimetric (i.e., mass-based) sampling of fine particulate matter (PM2.5). The smallest particles, especially those smaller than 0.1 μm (i.e., ultrafine particles), may have important health implications. Ultrafine particles can penetrate deep into the lungs resulting in oxidative stress and systemic inflammation (Brauer et al., 2001; Brauner et al., 2007; Brook et al., 2010; Donaldson et al., 2001; Donaldson and Stone, 2003; Sioutas et al., 2005). Ultrafine particles have a high surface area, but very little mass, thus measuring these size fractions via gravimetric sampling is not suitable. As a result, particle number concentration (PNC) is a more relevant metric for ultrafine particles than mass concentration.

Several studies have explored the size distribution of ultrafine particles emitted from different cookstove technologies in the laboratory setting (Rapp et al., 2016; Shen et al., 2017; Tiwari et al., 2014; Tryner et al., 2018). Measuring concentrations of PM2.5 and PNC in kitchens using traditional or engineered cookstoves in the field setting is logistically challenging and inhibited by monetary barriers associated with monitoring ultrafine particles. To date, only a few studies have used portable monitors to compare concentrations of PM2.5 and particle number in kitchens using three-stone fires and other biomass cookstoves (Chowdhury et al., 2012; de la Sota et al., 2018; Eilenberg et al., 2018; Wangchuk et al., 2015; Zhang et al., 2012). There is currently no standard protocol for measuring PNC in household air pollution research, and results from field studies vary substantially due to variation in the instrumentation, sample duration, stove type evaluated, and sample sizes.

Generally, it is assumed that engineered cookstoves (i.e., those designed with the intent to burn fuel more efficiently) reduce indoor particulate matter mass, but the resulting changes in particle size are less clear. Risk assessments for household air pollution, to date, are often focused on PM2.5 mass exposure and do not account for ultrafine PNC (Armendáriz-Arnez et al., 2010; Jetter et al., 2012; Smith et al., 2014). Laboratory studies suggest that certain engineered cookstoves (such as a natural draft top-lit up-draft stove) emit fewer particles in the ultrafine range (<0.1 μm) compared to traditional cookstoves (such as a three-stone fire) (Jetter et al., 2012); other engineered cookstoves (e.g., rocket and forced-draft gasifier cookstoves) demonstrate a shift to higher numbers of smaller particles (<0.03 μm) despite substantially reducing emissions of PM2.5 mass (Jetter et al., 2012; Just et al., 2013; Rapp et al., 2016). Given the inconsistencies observed in the relationship between emissions of PM2.5 mass and ultrafine particles during laboratory testing of various engineered biomass cookstove technologies, the estimated health benefits of engineered cookstoves may be misstated if field studies do not account for ultrafine particles.

The Justa cookstove is a commonly used engineered cookstove in Latin America and features an insulated, rocket-elbow combustion chamber, chimney, side compartment to remove excess ash, and plancha (Figure 1). Laboratory tests of the Justa stove show PM reductions of approximately one third compared to traditional three stone fires (Still et al., 2012). Additionally, a field study in Honduras demonstrated that the emission factor for improved stoves with chimneys was approximately 50% lower than that for traditional cookstoves (4.5 g kg vs. 8.2 g kg) (Roden et al., 2009). A study among 59 households in Honduras also showed that kitchens with Justa cookstoves had 73% lower PM2.5 concentrations compared to kitchens with traditional stoves (Clark et al., 2010). A field test of 5 kitchens in Honduras found that the Justa cookstove had lower PM2.5 emissions factors than the traditional stove and that the geometric mean particle diameter was 48 nm (Eilenberg et al., 2018). To date, no field studies have examined PNC in kitchens using Justa cookstoves.

Figure 1:

Figure 1:

Open in a new tab

Left: Example of a traditional cookstove in the Honduran study homes. Right: Example of a Justa cookstove

Although time-integrated gravimetric sampling is most often used to assess household air pollution exposure, time-resolved measurements provide additional insight into PM2.5 concentrations with respect to temporal variability or intensity of exposure during and between cooking events (Carter et al., 2016; Chen et al., 2016; Ezzati et al., 2000a; Fischer and Koshland, 2007; Northcross et al., 2015; Park and Lee, 2003; Van Vliet et al., 2013). For example, Van Vliet et al. (2013) reported that a few short-term periods of elevated PM2.5 concentrations constituted a substantial portion of daily exposure. It is unclear, however, whether metrics evaluated over periods shorter than 24-hours (“shorter-term” metrics) such as one-hour average maximum concentrations, may be relevant for health models evaluating effects of household air pollution.

In this study, we used real-time instrumentation to quantify kitchen concentrations of PM2.5 and PNC in rural areas surrounding La Esperanza, Honduras where biomass (wood-fueled) cookstoves were primarily used for cooking. Our objective was to evaluate and compare wood-burning traditional and Justa cookstoves (the latter of which had an engineered combustion chamber and chimney). Our goals were to 1.) characterize real-time PM2.5 mass concentrations and real-time PNC, 2.) evaluate the correlations between 24-hour average PM2.5 mass and particle number concentrations (for all stoves and by stove type), and 3.) evaluate correlations between 24-hour average concentrations and shorter-term averaging windows for both pollutants. To our knowledge, our study is the first household air pollution study to measure paired 24-hour real-time concentrations of PM2.5 and PNC.

2. Materials and Methods

2.1. Study Site, Population, and Stove Types

This study was conducted in rural communities surrounding La Esperanza, Department of Intibucá, Honduras as part of a larger study evaluating the health effects of exposure to household air pollution. In brief, the larger study included 230 women, aged 24–59, who were non-smokers and not pregnant. We measured real-time PM2.5 and PNC in a subsample of the women’s kitchens. With only one set of monitoring equipment, we were limited to collecting data from one kitchen per day. We collected forty-seven 24-hour samples in 36 unique kitchens from August 2015 to December 2016. We used a household survey to assess physical characteristics of the kitchen. We recorded the number of walls, windows, and doors; kitchen volume (height x length x width); wall material (mud, sticks); floor material (concrete, dirt, tile); roof material (sheet metal or tile); and presence of eaves. Additionally, women self-reported their use of a secondary stove as well as the number of cooking events and the number of people they cooked for during the 24-hour monitoring period (Young et al., 2019).

Of the 47 samples collected, 30 were collected in households that used a traditional cookstove and 17 were collected in households that used a Justa cookstove. Traditional cookstoves were typically self-built adobe stoves, with a metal plancha (griddle), a non-insulated open combustion area, and sometimes a chimney (Kshirsagar and Kalamkar, 2014; Kumar et al., 2013). All Justa stoves were installed in the homes approximately six months prior to the measurements. Stove users reported burning gathered wood, including split logs and sticks, as the primary fuel in both cookstoves. Additionally, users reported burning small sticks of a local wood called ocote (a species of pine) and corncobs to start the fire.

2.2. Particle Measurements

2.2.1. Fine particulate matter (PM2.5)

PM2.5 was sampled using an aerosol nephelometer, the personal DataRam (pDR) 1200 (Thermo Fisher Scientific Inc., Waltham MA, USA), powered by a 9V lithium ion rechargeable battery. The pDR was set up in an active-flow mode (1.5 L/min) using a pump (SKC AirChek XR5000 pump) and PM2.5 cyclone inlet (Triplex Cyclone; Mesa Labs, Butler NJ, USA). A 37mm filter (Fiberfilm™, Pall Corporation, Port Washington NY, USA) was installed downstream of the pDR photometric sensing chamber. The setup enabled estimation of time-resolved (60-second averaged) PM2.5 mass concentration followed by (downstream) collection of a time-integrated gravimetric sample. We collected field blanks once a week. An external data logger (EasyLog EL-USB-2, Lascar Electronics Ltd., Erie PA, USA) recorded the one-second pDR analog voltage data (0–5 V), corresponding to PM2.5 concentrations between 0 and 4,000 μg/m3. The pDR was zeroed in ambient air and the triplex cyclone was thoroughly cleaned before each 24-hour sample. At the field house in La Esperanza, the pump flow rate was checked pre- and post-sample using a flow meter (Bios International DryCal Lite, Mesa Labs, Butler NJ, USA). Sample filters were stored in a −20°C freezer in Honduras until they were transported back to Colorado State University and stored in a −80°C freezer. All filters were pre- and post-weighed to the nearest microgram (Mettler Toledo MX5, Mettler OH, USA) at Colorado State University, USA. Filters were equilibrated for 24 hours prior to weighing. Filter mass was determined by weighing each filter twice and averaging the weights. If the weights differed by more than 5 μg, a third weight was taken and the average of all three was used.

2.2.2. Ultrafine PNC

PNC was measured with the DiSCMini (Testo AG, Germany; Fierz et al., 2011). The DiSCMini is a handheld diffusion size classifier that estimates particle number for particles with diameters between 10 and 700 nm and provides data on airborne PNC between 103 and 106 particles (pt)/cm3. Portable, direct reading instruments, such as the DiSCmini, are a relatively new technology for field and personal monitoring of particle number. In both laboratory and field tests, the DiSCMini demonstrates high correlation with other instruments that measure particle number concentration (Aerotrack 9000, P-TRAK, and scanning mobility particle sizer [SMPS]) when tested in the same settings, indicating that the DiSCMini is a useful instrument for field monitoring (Asbach et al., 2012; Bau et al., 2017; Fierz et al., 2009a; Meier et al., 2013; Mills et al., 2013; Viana et al., 2015). The instrument has been shown to report within ± 30% for mean particle size and number concentration (Asbach et al., 2012). We equipped the DiSCMini with an external rechargeable battery to ensure 24 hours of continuous monitoring (7.4V 7.8Ah custom lithium ion battery). The DiSCMini recorded and logged concentrations at 1 Hz. The impactor on the DiSCMini inlet (cutpoint = 0.7 μm) was thoroughly cleaned before each 24-hour sample to help maintain flow through the instrument. The DiSCMini instrument turns off the pump for one minute in every hour to measure the zero offset in order to account for any long-term drifts in temperature or humidity (Fierz et al., 2011).

2.2.3. Household-level Field Measurements

The DiSCMini and pDR were collocated 40 to 70 inches from the front edge of the stove, 42 to 95 inches above the ground, and 41 to 61 inches from the nearest wall in each kitchen (Figure 2). Both instruments were started manually. The pump for the active PM2.5 measurements was programmed to turn off after 24 hours; the DiSCmini was manually switched off after 24 hours. A temperature and relative humidity monitor with a 60-second resolution (EasyLog EL-USB-2, Lascar Electronics Ltd., Erie PA, USA) was also collocated with the pDR and DiSCMini (Figure 2).

Figure 2:

Figure 2:

Open in a new tab

Left: DiSCmini instrument used to measure ultrafine particle number concentration. Right: Example set-up of DiSCMini and pDR monitors.

2.3. Data Processing

2.3.1. Fine particulate matter (PM2.5)

The real-time pDR measurements below the limit of detection (LOD) of 5 μg (Wallace et al., 2011) were substituted with the LOD/(√2) (Hewett and Ganser, 2007). Real-time pDR measurements were then corrected for relative humidity using Equation 1 (Chakrabarti et al., 2004):

PM2.5,60−s,dry=PM2.5,60−s,wet1+0.25RH2/(1−RH) (1)

where PM2.5,60 − s,dry was the dry (i.e., RH-corrected) 60-second average PM2.5 concentration, PM2.5,60 − s,wet was the 60-second average PM2.5 concentration recorded by the pDR, and RH was the relative humidity. In addition, we normalized real-time pDR concentrations to gravimetric measurements as shown in Equation 2:

PM2.5,60−s, corr=PM2.5,60−s,dry(PM2.5,pDR/PM2.5,filter)24−hour (2)

where PM2.5,60 − s, corr was the LOD-, RH-, and filter-corrected 60-s average PM2.5 concentration, PM2.5,60 − s,dry was the LOD- and RH-corrected 60-s average PM2.5 concentration, and the denominator is the LOD- and RH- corrected 24-hour average PM2.5 concentration measured using the pDR (PM2.5, pDR) divided by the 24-hour average PM2.5 concentration measured using the filter (PM2.5,filter). The value ofPM2.5,filter, in μg/m3, was calculated from the mass accumulated on the filter (corrected for 25 filter blanks and the LOD), the sample duration, and average of the pre- and post-test flow rates. The PM2.5 mass LOD was calculated by adding the average mass of the field blanks to three times the standard deviation of field blank masses (MacDougall et al., 1980). Filter weights below the LOD were substituted with the LOD/√2 (Hewett and Ganser, 2007).

2.3.2. Particle Number Concentration

The data from the DiSCmini were preprocessed using the DiSCmini data conversion tool (Matter Aerosol 2011, version 2.0), which assumes the number median diameter on the diffusion and filter stages was 30 nm and 300 nm, respectively. All additional analyses were performed in R, version 3.4.1 (R Core Team, Vienna, Austria). Given that the DiSCMini monitor was prone to overloading due to high emissions from the cookstoves and poor ventilation in the kitchen, we checked the DiSCmini data log for each household measurement for various error codes for each second sampled. For example, the DiSCmini electrometer amplifiers can detect currents between zero and 4096fA. At very high particle concentrations, the electrometer amplifiers will reach their maximal level and produce an error code. Additional errors can occur due to large temperature variations, high relative humidity, dirt on the charger’s corona wire, or flow of the instrument falling below 0.95 liters per minute. Rapid changes in particle concentrations can also result in negative diffusion and filter stages (Fierz et al., 2009b). Of the 3,801,600 total seconds in our dataset, <1% of the data had at least of one of the following error codes: filter stage below zero, diffusion stage below zero, filter stage or diffusion stage over 4096 fA (total current). All seconds flagged with an error code were excluded from the data analyses. Following the removal of seconds flagged with errors, we aggregated data to one-minute intervals using the mean PNC of each sampling minute.

2.4. Data Analysis

2.4.1. Goal 1: Characterize PM2.5 and PNC

We calculated descriptive statistics for the samples from the one-minute averages for both the PM2.5 (LOD-, RH- and filter-corrected) and PNC data sets. We calculated the 24-hour minimum, maximum, mean, median, standard deviation, 25th and 75th percentiles, as well as maximum concentrations in one-minute, five-minute, 15-minute, and 60-minute moving windows within each kitchen. We also created descriptive plots of the 24-hour real-time concentrations of PM2.5 and PNC for each kitchen. We used the pDR data (PM2.5,60 – s, corr) to compute the number of minutes that each sample’s PM2.5 concentration was above 100 μg/m3 (the equivalent of four times the WHO 24-hour air quality guideline) (World Health Organization, 2006a); a metric previously observed to be associated with increased incidence of acute lower respiratory infections among children (Chen et al., 2016; Gurley et al., 2013). A nonparametric alternative to the t-test, Wilcoxon rank sum test, was used to test for differences in 24-hour average PM2.5, 24-hour average PNC, and number of hours spent above 100 μg/m3 by stove type. Finally, for each sample, we removed 60-s average PM2.5 concentrations above the sample’s 95th percentile and then re-calculated the 24-hour average sample PM2.5 concentration (without the top 5th percentile) to evaluate the contribution of these high-concentration periods on the 24-hour kitchen concentration.

2.4.2. Goal 2: Correlation between PNC and PM2.5

We calculated Spearman correlation coefficients (a non-parametric test used due to non-normally distributed data) between the following data: 1.) 24-hour PM2.5 and PNC, 2.) maximum one-hour average PM2.5 and PNC, and 3.) maximum one-minute PM2.5 and PNC for all households and by stove type. Since the number of air changes within the houses may affected the correlation between 24-hour PM2.5 and PNC, we conducted a sensitivity analysis to characterize correlations between PM2.5 and PNC for samples above and for those below the median number of air exchanges per hour for the sample. We used real-time PM2.5 data from the pDR to calculate the number of air changes per hour for 23 households. For each household, we selected a single decay event where the PM2.5 concentration reached a peak and then fell continuously to a lower concentration. We made sure to select a decay event that consisted of at least 15 minutes of data. We then fit the data to a linear model using ordinary least squares regression. The time since the maximum concentration occurred, t, was the independent variable and ln(c/ci) was the dependent variable (where c was the PM2.5 maximum concentration and ci was the concentration at time in hours since the maximum concentration occurred). The absolute value of the model slope was described as the number of air exchanges per hour (AEPH) (Burgess et al., 2004).

2.4.3. Goal 3: Shorter-Term Concentrations

We calculated Spearman correlation coefficients between all averaging windows (24-hour, maximum one-minute, maximum five-minute, maximum 15-minute, and maximum 60-minute) within both PM2.5 and PNC.

3. Results

Of the forty-seven 24-hour samples performed, we excluded 20 PM2.5 samples (16 samples with external data logger failures, three with missing temperature and humidity data, one with negative gravimetric data). We excluded three PNC samples (all from homes using traditional stoves) because the DiSCMini turned off prior to completing at least 80% of the 24-hour sampling period. Our final sample size was 27 PM2.5 samples (collected in 24 unique kitchens) and 44 PNC samples (collected in 36 unique kitchens).

Kitchen characteristics of the sample population are described in Table 1. The majority of kitchens were constructed of mud or stuccoed adobe walls (60%), with dirt floors (70%) and sheet metal roofs (70%). Approximately 30% of all households reported having a traditional secondary stove occasionally used for cooking. In Honduras, secondary stoves are typically used outside the home for cooking large pots of beans or corn. During the 24-hour monitoring period, women reported cooking a mean of 3.1 times (SD: 0.88 times) for a mean of 5.5 people (SD: 2.5 people). The median number of air exchanges per hour was 9.7 (mean: 10.5; range: 3.6 to 18.2).

Table 1:

Kitchen characteristics of study homes in rural Honduras

n* Mean SD
Kitchen volume (m3) 36 35.4 15.4
Number of walls 36 4.0 0.0
Number of windows 36 1.1 0.7
Number of doors 36 1.6 0.7
Number of people cooked for in past 24 hours 47 5.5 2.5
Number of times cooked in past 24 hours 47 3.1 0.88
n %
Stove type 47
 Traditional 30 64%
 Justa (Improved) 17 36%
Wall material 36
 Mud (adobe) 13 36%
 Stuccoed adobe 16 45%
 Wood/sticks 4 11%
 Concrete 3 8%
Floor material 36
 Dirt 24 65%
 Concrete 10 27%
 Ceramic tile 3 8%
Roof material 36
 Sheet metal 25 70%
 Tiles 12 30%
Use of secondary stove 47
 Yes 12 30%
Number of Air Changes Per Hour** 23 10.47 3.92
Open in a new tab
*

A total of 47 samples were collected in 36 unique households; 11 houses had repeated measurements and household characteristics remained the same

**

One of the real-time pDR samples could not be calculated for an air exchange per hour due to short decay rates less than 15 minutes

3.1. Goal 1: Characterize PM2.5 and PNC

The median gravimetrically-determined 24-hour average PM2.5 concentration for all samples (n=27) was 79 μg/m3 (IQR: 44–174 μg/m3) (Table 2). On average, households using traditional primary stoves had higher PM2.5 concentrations (median: 130 μg/m3; IQR: 48–250 μg/m3; n=15) compared to households using Justa stoves (median: 66 μg/m3; IQR: 44–97 μg/m3; n=12) (Wilcoxon rank sum test, p = 0.11) (Table 2). The average ratio of the PM2.5 concentrations measured using the pDR and the filter [(PM2.5,pDR/PM2.5,filter)24−hour; the “response factor”] was 0.56 (median: 0.57), indicating that the nephelometer tended to underestimate PM2.5 concentrations relative to the time-integrated filter measurements (IQR: 0.39–0.66; n=27). The median pDR response factors by stove type were 0.60 (IQR: 0.45–0.74; n=15) for traditional stoves and 0.50 (IQR: 0.37–0.58; n=12) for Justa cookstoves.

Table 2:

24-hour average fine particulate matter (PM2.5) and particle number concentration (PNC) among kitchens in rural Honduras

Mean Standard Deviation Min 25th Percentile Median 75th Percentile Max
PM2.5 μg/m3* All samples (n=27) 180 301 11 44 79 174 1467
Justa (n=12) 76 51 11 44 66 97 173
Traditional (n= 15) 263 386 23 48 130 250 1468
PNC (pt/cm3) All samples (n=44) 1.20E+05 1.00E+05 4.40E+02 3.80E+04 8.50E+04 1.80E+05 4.10E+05
Justa (n=17) 9.10E+04 6.90E+04 2.50E+04 4.00E+04 6.30E+04 1.20E+05 2.40E+05
Traditional (n=27) 1.30E+05 1.10E+05 4.40E+02 3.30E+04 1.30E+05 2.00E+05 4.10E+05
Reduced dataset for 24-hour samples that had both PM2.5 and PNC
Mean Standard Deviation Min 25th Percentile Median 75th Percentile Max
PM2.5 μg/m3* All samples (n=24) 124 164 11 41 70 142 805
Justa (n=12) 76 51 11 44 66 97 173
Traditional (n= 12) 172 220 23 44 99 201 804
PNC (pt/cm3) All Stoves (n=24) 9.50E+04 7.90E+04 1.90E+04 3.80E+04 6.20E+04 1.40E+05 2.60E+05
Justa (n=12) 8.10E+04 7.10E+04 2.50E+04 3.80E+04 5.50E+04 8.50E+04 2.40E+05
Traditional (n=12) 1.10E+05 8.60E+04 1.90E+04 3.80E+04 8.00E+04 1.60E+05 2.60E+05
Open in a new tab
*

Corrected to gravimetric samples

During the one-hour averaging windows, maximum PM2.5 concentrations ranged from 51 to 4026 μg/m3 for all 27 samples, 141 to 4026 μg/m3 for traditional stoves, and 51 to 2098 μg/m3 for Justa stoves (Figure 3). On average, one-hour maximum concentrations were higher for traditional stoves (mean: 1469 μg/m3; SD: 1141 μg/m3; n=15), compared to Justa stoves (mean: 957 μg/m3; SD: 719 μg/m3; n=12; Wilxcon rank sum, p= 0.32; Figure 3). The average number of hours a kitchen PM2.5 concentration exceeded 100 μg/m3 was 4.0 hours (SD: 3.7; n=27) and ranged from less than 1 hour to over 15 hours. Kitchen PM2.5 concentrations exceeded 100 μg/m3 for a mean of 5.5 hours (SD: 4.4; n=15) in kitchens with traditional cookstoves and a mean of 2.3 hours (SD: 1.3; n=12) in kitchens with Justa stoves (Wilcoxon p=0.08) (Table 3).

Figure 3:

Figure 3:

Open in a new tab

Maximum 1-hour average PM2.5 concentrations measured in kitchens using traditional and cleaner-burning Justa stoves (N = 27) in rural Honduras. Black dots represent the observed concentrations. The lower boundary of the box represents the 25th percentile; the line within the box is the median; the upper boundary represents the 75th percentile. Bars indicate the 10th and 90th percentiles and the “◊” represents the mean.

Table 3:

Number of Hours Spent Above 100 μg/m3

Mean Standard Deviation Min 25th Percentile Median 75th Percentile Max
All samples (n=27) 4.0 3.7 0.43 1.6 2.3 5.2 15.2
Justa (n=12) 2.3 1.3 0.43 1.4 2.1 3.1 4.7
Traditional (n= 15) 5.5 4.4 1.1 1.8 4.7 8.3 15.6
Open in a new tab
*

Corrected to gravimetric samples

When corrected 60-s average PM2.5 concentrations (PM2.5,60 – s,corr) above the 95th percentile were removed, the median 24-hour PM2.5 concentration in kitchens was 25 μg/m3 (IQR: 15–62 μg/m3; n=27) (Traditional stoves: 50 μg/m3; IQR: 16–127 μg/m3; n=15; Justa stoves: 19 μg/m3; IQR: 15–30 μg/m3; n=12; Wilcoxon rank sum, p=0.14). On average, the 60-second concentrations above the 95th percentile exposure values accounted for 42% of the 24-hour average concentration (46% among traditional stoves and 37% among Justa stove). The Spearman rho correlation between the full 24-hour average PM2.5 concentration dataset and the dataset with the top 5% removed was 0.90. The Spearman rho correlation between the full 24-hour average PM2.5 concentration and the dataset with only concentrations above the 95th percentile was 0.96.

The 24-hour average mean and median PNC for 44 samples are shown in Figure 4; the median concentration was 8.5×105 pt/cm3 (IQR: 3.8×104- 1.8×105 pt/cm3; n=44). PNC was lower among the households with Justa cookstoves (median: 6.3×104 pt/cm3; IQR: 4.0×104-1.2×105 pt/cm3; n=17) compared to traditional cookstoves (median: 1.3×105 pt/cm3; IQR: 3.3×104-2.0×105 pt/cm3; n=27); however, the pollutant distributions largely overlapped for the two stove types (Wilcoxon rank sum, p = 0.76). Descriptive plots of the 24-hour concentrations for PM2.5 mass and PNC for individual kitchens demonstrate similar patterns in PNC and PM2.5 emissions throughout the day (example presented in Figure 5).

Figure 4:

Figure 4:

Open in a new tab

A: 24-hour average PM2.5 (n=27), B: 24-hour particle number concentration (n=44). Measurements were in kitchens using traditional and cleaner-burning Justa cookstoves in rural Honduras. Dots represent the observed concentrations. The lower boundary of the box represents the 25th percentile; the line within the box is the median; the upper boundary represents the 75th percentile. Bars indicate the 10th and 90th percentiles and the “◊” represents the mean.

Figure 5:

Figure 5:

Open in a new tab

Example of real-time minute-level kitchen concentrations of PNC and PM2.5 mass over a 24-hour time period. (A-1: Justa stove PNC, A-2: Justa stove PM2.5, B-1: Traditional stove PNC, B-2: Traditional stove PM2.5)

3.2. Goal 2: Correlation between PM2.5 and PNC

The Spearman correlation coefficient between filter-corrected 24-hour average PM2.5 and 24-hour average PNC was 0.83 (n=24). Correlations between 24-hour average PM2.5 and PNC were higher in traditional stove households (ρ=0.93) than in Justa stove households (ρ=0.67). The Spearman correlation between the maximum one-hour-average PM2.5 and the maximum one-hour-average ultrafine PNC was 0.54 (traditional stoves: ρ=0.62; Justa stoves: ρ=0.43, while the correlation between maximum one-minute concentrations of these two pollutants was 0.43 (traditional stoves: ρ=0.83; Justa stoves: ρ=0.36). The Spearman correlations between the number of air exchanges per hour and the 24-hour average pollutant concentrations were low (PM2.5: ρ = 0.02; PNC: ρ = −0.08). Similarly low correlations between number of AEPH and 24-hour average concentrations were observed when each stove type was considered individually (traditional: PM2.5: ρ = 0.17; PNC: ρ = 0.01; Justa: PM2.5 ρ = −0.12; PNC ρ = −0.16). Spearman correlations between PM2.5 and PNC above and below the median AEPH were high and similar between the two metrics (above the median: ρ = 0.82; below the median: ρ = 0.79).

3.3. Goal 3: Shorter-Term Concentrations

The maximum PM2.5 concentrations for the one-minute, five-minute, 15-minute and 60-minute averaging periods were highly correlated with the 24-hour average for the household (ρ ranging from 0.65–0.85). Correlations between maximum values in one-minute, five-minute, and 60-minutes ranged from 0.58–0.95. Maximum PNC concentrations for the one-minute, five-minute, 15-minute, and 60-minute averaging windows were also highly correlated with the 24-hour average for the household (ρ ranging from 0.74–0.88).

4. Discussion

4.1. Goal 1: Characterize PM2.5 and PNC

4.1.1. PM2.5

Kitchen concentrations of PM2.5 were higher among households with traditional stoves compared to kitchen concentrations where Justa stoves were used; however, there were substantial overlaps in average concentrations. Additionally, the mean 24-hour average PM2.5 concentrations exceeded the World Health Organization (WHO) air quality guideline of 25 μg/m3 for both cookstove types (World Health Organization, 2006b). Although our measures were for kitchen (area) and not personal concentrations, only 3 of 27 (22%) kitchens had 24-hour average concentrations below 25 μg/m3 (one traditional, two Justa).

We observed variation over each 24-hour sampling period. Indoor PM2.5 mass and ultrafine particle concentrations peaked in the morning, likely due to cookstove startup (generally between 4am-5am), and were lowest overnight when the stove was likely off. Similar studies using temporally-resolved emissions monitoring in Kenya and China have also observed elevated concentrations of PM2.5 coinciding with diurnal patterns and phases of cooking (i.e., startup) (Carter et al., 2016; Ezzati et al., 2000a; Kaur et al., 2017; Park and Lee, 2003). The substantial variation in kitchen concentrations of PM2.5 (within households) suggests that peaks of exposure occur during cooking (especially in scenarios without other primary sources of pollution, as with our study population) and highlight the importance of using personal monitoring to capture a better estimate of exposure (Clark et al. 2013).

4.1.2. PNC

In general, our results show lower PNC among the kitchens with a Justa cookstove, compared to traditional cookstove kitchens, despite substantial overlap in the concentrations. This result is similar to results reported in previous studies conducted in Senegal and China. De la Sota et al. (2018) used the DiSCmini to monitor PNC during cooking periods in three households using a traditional stove and three households using an improved rocket stove in Senegal. de la Sota et al. (2018) observed lower PNC during cooking events in the households with improved rocket stoves (median PNC 1.5×106 pt/cm3) compared to the households with traditional stoves (median PNC 2.2×106 pt/cm3). Similarly, a field study of 15 households using coal or wood for heating and cooking in China measured PNC with an AEROTRAK 9000 and reported that cookstoves with chimneys reduced kitchen concentrations of ultrafine particle by a factor of four during cooking periods (Hosgood et al., 2012). We did not calculate cooking event concentrations, which limits our ability to compare our results to both of these prior studies.

4.2. Goal 2: Correlation between PNC and PM2.5

Correlation between 24-hour average PM2.5 and PNC concentrations was high among traditional stoves (ρ=0.93) and moderate among improved stoves (ρ=0.67). The mechanism influencing the lower correlation of PM2.5 mass and PNC in households with the Justa stove, compared to households with traditional stoves, is unclear and may be driven by differences in particle formation and oxidation processes resulting from differences between the two stove designs. The lower correlation could also be due to measurement error or influenced by a small number of Justa households sampled. Several Justa cookstove kitchens had high PM2.5 concentrations with relatively low PNC; these samples, in particular, may have been subject to measurement error. The 0.7 μm impactor on the DiSCmini inlet was occasionally clogged by large particles in the high-concentration kitchen setting, resulting in reduced flow through the instrument.

The correlations between PM2.5 and PNC above and below the median number of air exchanges per hour were similar. We observed low correlations between the 24-hour average concentrations and the number of air exchanges per hour. One might expect 24-hour average pollutant concentrations to be negatively correlated with the number of air changes per hour, since higher air exchange rates would help remove cookstove emissions from the home; however, we observed both low positive and low negative correlations. One may also expect the correlations between the number of air exchanges per hour and concentrations to be higher, however we did not ask participants to keep a log of activities and it could be possible that changes in behavior such as opening or closing windows and doors.

For both stove types combined, correlation between maximum one-hour average PM2.5 mass and PNC was moderate (ρ=0.64) and correlation between one-minute maximum PM2.5 mass and PNC was lower (ρ=0.41). These results indicate possible differences in exposure to PM2.5 and ultrafine PNC over short time periods, perhaps during transient operating conditions that occur when the fire is started or refueled. Evidence from both lab and field studies support the hypothesis that variations in particle size and PNC are related to certain cooking activities and phases of cooking (Arora et al., 2013; Carter et al., 2016; Ezzati et al., 2000b; Park and Lee, 2003; Tryner et al., 2018; Zhang et al., 2012, 2010).

4.3. Goal 3: Shorter-Term Concentrations

Our one-hour average maximum PM2.5 concentrations for individual samples ranged from 44 to 3929 μg/m3 and were slightly lower than one-hour average concentrations reported by Fischer (159 to 6200 μg/m3; n=43) among households using a variety of stoves in China (Fischer and Koshland, 2007). The highest PM2.5 concentrations in one-hour averaging windows among traditional and Justa stove samples (3929 and 1682 μg/m3, respectively) were 15 and 22 times higher than the 24-hour averages. Park and Lee observed peak kitchen concentrations (defined as at least 7 minutes before pollutant decay was observed) between 32 and 39 times higher than 24-hour averages among traditional and improved biomass stove users in Costa Rica (Park and Lee, 2003). The mean sample one-hour average maximum values were 207 and 87 times larger than baseline levels measured at night for traditional and Justa stove groups, respectively. Other studies suggest that high short-term particulate matter concentrations likely result from cooking activities such as tending or adding fuel to the fire (Bartington et al., 2017; Ezzati et al., 2000a; Just et al., 2013). Additionally, Van Vliet et al. observed that mitigating the overall highest 1–5% of the 60-second PM2.5 concentrations during a 24-hour sampling period in a Ghana field study could reduce mean personal 24-hour exposure by 49–75% (Van Vliet et al., 2013). We observed that removing 60-second PM2.5 real-time concentrations above the 95th percentile decreased the overall 24-hour average concentrations by 42%, 46% among traditional stoves and 37% among Justa stoves. Given the high contribution of the most polluted concentrations to the overall average, it may be important to understand how reductions in start-up and cooking-specific emissions from different stove designs could reduce overall average concentrations.

The mean number of hours spent over 100 μg/m3 (4.0 hours; IQR: 1.5–5.2) in our study kitchens was similar to the mean number of hours spent over 100 μg/m3 (mean: 5.3, IQR: 4.0–6.9) among both homes using clean fuels and homes using biomass fuels in Bangladesh (Gurley et al., 2013). The health implications of short-term high concentrations of PM mass or ultrafine PNC are unclear in the field of household air pollution. Ambient air pollution studies have observed the association of one-hour average maximum ambient PM2.5 on hospital admission and mortality but this association has not been studied for household air pollution (Burgan et al., 2010). Although our measured short-term kitchen concentrations of PM2.5 are highly correlated with 24-hour concentrations, we do not know if this relationship is similar for personal exposure. It may be useful to measure short-term intensity of personal exposure when studying exposure-response relationships, especially for cardiovascular endpoints.

4.4. Limitations and Lessons Learned

Our study is limited by a small sample size. Use of the DiSCmini in settings with very high particle concentrations over a long period was also challenging. The inability of the DiSCMini and analog function on the pDR to report concentrations above 1,000,000 particles/cm3 and 4,000 μg/m3, respectively, could have led to measurement errors that would affect the correlation between the one-minute maximum PM2.5 and PNC values. True concentrations may be higher than reported; however only <1% of all PNC data was above the upper limit of detection. The pDR instrument is suitable for fieldwork and areas of high concentrations; however, we experienced frequent loss of pDR data. We suspect that data loss resulted from the connection between the pDR and external logger coming loose during transportation, setup, or operation. Although humidity levels in our sample were high, 82% of minutes had humidity levels of 62%, we believe humidity corrections for PM2.5 and the internal corrections for the DiSCMini were adequate in addressing humidity concerns for the measurements. The correlation between minute-level humidity and corrected concentrations were low (PM2.5 = 0.05; PNC = −0.18 for PNC).

For each household, we estimated the number of air changes per hour using pDR data recorded during a single decay event. More robust estimates might have been obtained by using data from multiple decay events. Our ability to identify multiple decay events for each household was limited because (a) we did not ask occupants to record the times of cooking events and (b) multiple short-term concentration peaks were observed during many of the presumed cooking events.

Placement of the collocated instruments in the kitchens varied (due to logistical challenges of placing the instrument away from the stove and windows or door) and this variation could have affected the measurements. There are currently no standards for measuring ultrafine particles in the household setting, and it is unclear as to how the distance between the stove and the instruments may affect individual kitchen concentrations (He et al., 2005). Given the lack of standards for consistent spacing between monitors and stoves for measuring ultrafine PNC, it is possible our 24-hour concentrations and correlations between PM2.5 mass and PNC would not be generalizable to other populations with different cooking environments (i.e., kitchen layout, ventilation), fuels, and stove types. Finally, our results for kitchen concentrations of PM2.5 and PNC may not translate to measurements of personal PM2.5 and PNC.

5. Conclusions

This study is the first to characterize 24-hour time-resolved PM2.5 and ultrafine particle number concentrations in kitchens. Our study reveals variability in PM2.5 and PNC within and between samples and indicates that the highest exposure periods account for almost one half of the 24-hour average concentration. Correlations between PM2.5 and PNC differed between traditional and Justa stoves, indicating that additional research may be needed to understand how the correlation between PM2.5 mass and ultrafine particle concentrations differs by stove type. This information would provide insight regarding whether measurements of fine particulate matter are sufficient for characterizing exposure to household air pollution, particularly for studies evaluating multiple stove types. High correlations between 24-hour averages and sub-daily concentrations of PM2.5 and PNC indicate that monitoring 24-hour average concentrations in similar rural settings may be a cost-effective method (i.e., without incorporating real-time instrumentation) to evaluating household-level concentrations of PM2.5 and ultrafine particulate matter.

Highlights.

  • Burning solid fuels for cooking produces high concentrations of particulate matter

  • PM2.5 and particle number concentration (PNC) were moderately correlated

  • Shorter-term averaging windows were moderately correlated with the 24-hour average

  • Peak concentrations accounted for a substantial portion of 24-hour concentration

Acknowledgements

We would like to acknowledge our field team members Gloria Bautista, Jonathan Stack, Quinn Olson, Annalise Wille, and Rebecca Hermann.

Funding

Research was funded by the National Institute of Environmental Health Sciences of The National Institutes of Health [grant number: ES022269]. The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Main Findings: Kitchen concentrations of fine particulate matter (PM2.5) and particle number concentration were moderately correlated between traditional and improved biomass cookstoves.

References

  1. Armendáriz-Arnez C, Edwards RD, Johnson M, Rosas IA, Espinosa F, Masera OR, 2010. Indoor particle size distributions in homes with open fires and improved Patsari cook stoves. Atmos. Environ 44, 2881–2886. 10.1016/j.atmosenv.2010.04.049 [DOI] [Google Scholar]
  2. Arora P, Jain S, Sachdeva K, 2013. Physical characterization of particulate matter emitted from wood combustion in improved and traditional cookstoves. Energy Sustain. Dev 17, 497–503. 10.1016/j.esd.2013.06.003 [DOI] [Google Scholar]
  3. Asbach C, Kaminski H, Von Barany D, Kuhlbusch TAJ, Monz C, Dziurowitz N, Pelzer J, Vossen K, Berlin K, Dietrich S, Götz U, Kiesling HJ, Schierl R, Dahmann D, 2012. Comparability of portable nanoparticle exposure monitors. Ann. Occup. Hyg 56, 606–621. 10.1093/anhyg/mes033 [DOI] [PubMed] [Google Scholar]
  4. Bartington SE, Bakolis I, Devakumar D, Kurmi OP, Gulliver J, Chaube G, Manandhar DS, Saville NM, Costello A, Osrin D, Hansell AL, Ayres JG, 2017. Patterns of domestic exposure to carbon monoxide and particulate matter in households using biomass fuel in Janakpur, Nepal. Environ. Pollut 220, 38–45. 10.1016/j.envpol.2016.08.074 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bau S, Payet R, Witschger O, Jankowska E, 2017. Performance study of portable devices for the real-time measurement of airborne particle number concentration and size (distribution) Performance study of portable devices for the real-time measurement of airborne particle number concentration and size. J. Phys. Conf. Ser 838. [Google Scholar]
  6. Bonjour S, Adair-Rohani H, Wolf J, Bruce NG, Mehta S, Prüss-Ustün A, Lahiff M, Rehfuess EA, Mishra V, Smith KR, 2013. Solid fuel use for household cooking: Country and regional estimates for 1980–2010. Environ. Health Perspect 121, 784–790. 10.1289/ehp.1205987 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brauer M, Avila-Casado C, Fortoul TI, Vedal S, Stevens B, Churg A, 2001. Air pollution and retained particles in the lung. Environ. Health Perspect 109, 1039–1043. 10.1289/ehp.011091039 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brauner EV, Forchhammer L, Mller P, Simonsen J, Glasius M, W??hlin P, Raaschou-Nielsen O, Loft S, 2007. Exposure to ultrafine particles from ambient air and oxidative stress-induced DNA damage. Environ. Health Perspect. 115, 1177–1182. 10.1289/ehp.9984 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brook RD, Rajagopalan S, Pope CA, Brook JR, Bhatnagar A, Diez-Roux AV, Holguin F, Hong Y, Luepker RV, Mittleman MA, Peters A, Siscovick D, Smith SC, Whitsel L, Kaufman JD, 2010. Particulate Matter Air Pollution and Cardiovascular Disease: An Update to the Scientific Statement From the American Heart Association. Circulation 121, 2331–2378. 10.1161/CIR.0b013e3181dbece1 [DOI] [PubMed] [Google Scholar]
  10. Burgan O, Smargiassi A, Perron S, Kosatsky T, 2010. Cardiovascular effects of sub-daily levels of ambient fine particles: A systematic review. Environ. Heal. A Glob. Access Sci. Source 9 10.1186/1476-069X-9-26 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Burgess WA, Ellenbecker MJ, Treitman RD, 2004. Ventilation for Contol of the Work Environment, 2nd ed. [Google Scholar]
  12. Carter E, Archer-Nicholls S, Ni K, Lai AM, Niu H, Secrest MH, Sauer SM, Schauer JJ, Ezzati M, Wiedinmyer C, Yang X, Baumgartner J, 2016. Seasonal and Diurnal Air Pollution from Residential Cooking and Space Heating in the Eastern Tibetan Plateau. Environ. Sci. Technol 50, 8353–8361. 10.1021/acs.est.6b00082 [DOI] [PubMed] [Google Scholar]
  13. Chakrabarti B, Fine PM, Delfino R, Sioutas C, 2004. Performance evaluation of the active-flow personal DataRAM PM2.5 mass monitor (Thermo Anderson pDR-1200) designed for continuous personal exposure measurements. Atmos. Environ 38, 3329–3340. 10.1016/j.atmosenv.2004.03.007 [DOI] [Google Scholar]
  14. Chen C, Zeger S, Breysse P, Katz J, Checkley W, Curriero FC, Tielsch JM, 2016. Estimating indoor PM2.5 and CO concentrations in households in southern Nepal: The Nepal cookstove intervention trials. PLoS One 11, 1–17. 10.1371/journal.pone.0157984 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Chowdhury Z, Le LT, Masud A. Al, Chang KC, Alauddin M, Hossain M, Zakaria ABM, Hopke PK, 2012. Quantification of indoor air pollution from using cookstoves and estimation of its health effects on adult women in Northwest Bangladesh. Aerosol Air Qual. Res 12, 463–475. 10.4209/aaqr.2011.10.0161 [DOI] [Google Scholar]
  16. Clark ML, Peel JL, Balakrishnan K, Breysse P, Chillrud SN, Naeher LP, Rodes CE, Vette AF, Balbus JM, Maggie Clark ML, 2012. Household Air Pollution Related to Solid Fuel Use and Health: The Need for Improved Exposure Assessment 1120, 42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Clark ML, Reynolds SJ, Burch JB, Conway S, Bachand AM, Peel JL, 2010. Indoor air pollution, cookstove quality, and housing characteristics in two Honduran communities. Environ. Res 110, 12–18. 10.1016/j.envres.2009.10.008 [DOI] [PubMed] [Google Scholar]
  18. de la Sota C, Lumbreras J, Pérez N, Ealo M, Kane M, Youm I, Viana M, 2018. Indoor air pollution from biomass cookstoves in rural Senegal. Energy Sustain. Dev 43, 224–234. [Google Scholar]
  19. Donaldson K, Stone V, 2003. Current hypotheses on the mechanisms of toxicity of ultrafine particles. Ann. Ist. Super. Sanita 39, 405–410. 10.1016/B978-0-444-62619-6.00027-6 [DOI] [PubMed] [Google Scholar]
  20. Donaldson K, Stone V, Clouter A, Renwick L, MacNee W, 2001. Ultrafine particles. Br. Med. J 58, 211 10.1136/oem.58.3.211 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Eilenberg SR, Bilsback KR, Johnson M, Kodros JK, Lipsky EM, Naluwagga A, Fedak KM, Benka-Coker M, Reynolds B, Peel J, Clark M, Shan M, Sambandam S, L’Orange C, Pierce JR, Subramanian R, Volckens J, Robinson AL, 2018. Field measurements of solid-fuel cookstove emissions from uncontrolled cooking in China, Honduras, Uganda, and India. Atmos. Environ 190 10.1016/j.atmosenv.2018.06.041 [DOI] [Google Scholar]
  22. Ezzati M, Mbinda BM, Kammen DM, 2000a. Comparison of emissions and residential exposure from traditional and improved cookstoves in Kenya. Environ. Sci. Technol 34, 578–583. 10.1021/es9905795 [DOI] [Google Scholar]
  23. Ezzati M, Saleh H, Kammen DM, 2000b. The contributions of emissions and spatial microenvironments to exposure to indoor air pollution from biomass combustion in Kenya. Environ. Health Perspect 108, 833–839. 10.1289/ehp.00108833 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Fierz M, Houle C, Steigmeier P, Burtscher H, 2011. Design, Calibration, and Field Performance of a Miniature Diffusion Size Classifier. Aerosol Sci. Technol 45, 1–10. 10.1080/02786826.2010.516283 [DOI] [Google Scholar]
  25. Fierz M, Keller A, Burtscher H, 2009a. Charge-based personal aerosol samplers. Inhal. Toxicol. 21 Suppl 1, 30–34. 10.1080/08958370902942632 [DOI] [PubMed] [Google Scholar]
  26. Fierz M, Weimer S, Burtscher H, 2009b. Design and performance of an optimized electrical diffusion battery. J. Aerosol Sci 40, 152–163. 10.1016/j.jaerosci.2008.09.007 [DOI] [Google Scholar]
  27. Fischer SL, Koshland CP, 2007. Daily and Peak 1 h Indoor Air Pollution and Driving Factors in a Rural Chinese Village 41, 3121–3126. [DOI] [PubMed] [Google Scholar]
  28. Gurley ES, Homaira N, Salje H, Ram PK, Haque R, Petri W, Bresee J, Moss WJ, Breysse P, Luby SP, Azziz-Baumgartner E, 2013. Indoor exposure to particulate matter and the incidence of acute lower respiratory infections among children: A birth cohort study in urban Bangladesh. Indoor Air 23, 379–386. 10.1111/ina.12038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. He G, Ying B, Liu J, Gao S, Shen S, Balakrishnan K, Jin Y, Liu F, Tang N, Shi K, Baris E, Ezzati M, 2005. Patterns of household concentrations of multiple indoor air pollutants in China. Environ. Sci. Technol 39, 991–998. 10.1021/es049731f [DOI] [PubMed] [Google Scholar]
  30. Hewett P, Ganser GH, 2007. A comparison of several methods for analyzing censored data. Ann. Occup. Hyg 51, 611–632. 10.1093/annhyg/mem045 [DOI] [PubMed] [Google Scholar]
  31. Hosgood HD, Lan Q, Vermeulen R, Wei H, Reiss B, Coble J, Wei F, Jun X, Wu G, Rothman N, 2012. Combustion-derived nanoparticle exposure and household solid fuel use in Xuanwei and Fuyuan, China. Int. J. Environ. Health Res 22, 571–581. 10.1080/09603123.2012.684147 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Jetter J, Zhao Y, Smith KR, Khan B, Yelverton T, Decarlo P, Hays MD, 2012. Pollutant emissions and energy efficiency under controlled conditions for household biomass cookstoves and implications for metrics useful in setting international test standards. Environ. Sci. Technol 46, 10827–10834. 10.1021/es301693f [DOI] [PubMed] [Google Scholar]
  33. Just B, Rogak S, Kandlikar M, 2013. Characterization of ultrafine particulate matter from traditional and improved biomass cookstoves. Environ. Sci. Technol 47, 3506–3512. 10.1021/es304351p [DOI] [PubMed] [Google Scholar]
  34. Kaur M, Ravindra K, Mor S, John S, 2017. Household air pollution from various types of rural kitchens and its exposure assessment. Sci. Total Environ 1–11. 10.1016/j.scitotenv.2017.01.051 [DOI] [PubMed] [Google Scholar]
  35. Kshirsagar MP, Kalamkar VR, 2014. A comprehensive review on biomass cookstoves and a systematic approach for modern cookstove design. Renew. Sustain. Energy Rev 30, 580–603. 10.1016/j.rser.2013.10.039 [DOI] [Google Scholar]
  36. Kumar M, Kumar S, Tyagi SK, 2013. Design, development and technological advancement in the biomass cookstoves: A review. Renew. Sustain. Energy Rev 26, 265–285. 10.1016/j.rser.2013.05.010 [DOI] [Google Scholar]
  37. MacDougall D, Crummet W, Al E, 1980. Guidelines For Data Acquisition And Data Quality Evaluation In Environmental Chemistry. Anal. Chem 52, 2242–2249. 10.1021/ac50064a004 [DOI] [Google Scholar]
  38. Meier R, Clark K, Riediker M, 2013. Comparative Testing of a Miniature Diffusion Size Classifier to Assess Airborne Ultrafine Particles Under Field Conditions. Aerosol Sci. Technol 47, 22–28. 10.1080/02786826.2012.720397 [DOI] [Google Scholar]
  39. Mills JB, Hong Park J, Peters TM, 2013. Comparison of the DiSCmini aerosol monitor to a handheld consenation particle counter and a scanning mobility particle sizer for submicrometer sodium chloride and metal aerosols. J Occup Env. Hyg 33, 395–401. 10.1038/nbt.3121.ChIP-nexus [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Northcross AL, Hwang N, Balakrishnan K, Mehta S, 2015. Assessing Exposures to Household Air Pollution in Public Health Research and Program Evaluation. Ecohealth 12, 57–67. 10.1007/s10393-014-0990-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Park E, Lee K, 2003. Particulate exposure and size distribution from wood burning stoves in Costa Rica. Indoor Air 13, 253–259. 10.1034/j.1600-0668.2003.00194.x [DOI] [PubMed] [Google Scholar]
  42. Rapp VH, Caubel JJ, Wilson DL, Gadgil AJ, 2016. Reducing Ultrafine Particle Emissions Using Air Injection in Wood-Burning Cookstoves. Environ. Sci. Technol 50, 8368–8374. 10.1021/acs.est.6b01333 [DOI] [PubMed] [Google Scholar]
  43. Roden C. a., Bond TC, Conway S, Osorto Pinel AB, MacCarty N, Still D, 2009. Laboratory and field investigations of particulate and carbon monoxide emissions from traditional and improved cookstoves. Atmos. Environ 43, 1170–1181. 10.1016/j.atmosenv.2008.05.041 [DOI] [Google Scholar]
  44. Shen G, Gaddam CK, Ebersviller SM, Wal R.L. Vander, Williams C, Faircloth JW, Jetter JJ, Hays MD, 2017. A Laboratory Comparison of Emission Factors, Number Size Distributions, and Morphology of Ultrafine Particles from 11 Different Household Cookstove-Fuel Systems. Environ. Sci. Technol 51, 6522–6532. 10.1021/acs.est.6b05928 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sioutas C, Delfino RJ, Singh M, 2005. Exposure assessment for atmospheric Ultrafine Particles (UFPs) and implications in epidemiologic research. Environ. Health Perspect 113, 947–955. 10.1289/ehp.7939 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sioutas C, Kim S, Chang M, Terrell LL, Gong H, 2000. Field evaluation of a modified DataRAM MIE scattering monitor for real-time PM2.5 mass concentration measurements. Atmos. Environ 34, 4829–4838. 10.1016/S1352-2310(00)00244-2 [DOI] [Google Scholar]
  47. Smith KR, Bruce N, Balakrishnan K, Adair-Rohani H, Balmes J, Chafe Z, Dherani M, Hosgood HD, Mehta S, Pope D, Rehfuess E, 2014. Millions dead: how do we know and what does it mean? Methods used in the comparative risk assessment of household air pollution. Annu. Rev. Public Health 35, 185–206. 10.1146/annurev-publhealth-032013-182356 [DOI] [PubMed] [Google Scholar]
  48. Stanaway JD, 2018. Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990–2017: A systematic analysis for the Global Burden of Disease 201. Lancet 1923–1994. 10.1016/S0140-6736(18)32225-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Still D, MacCarty N, Ogle D, Bond T, Bryden M, 2012. Test results of cook stove performance, Partnership for Clean Indoor Air, Portland, OR. [Google Scholar]
  50. Thomas E, Wickramasinghe K, Mendis S, Roberts N, Foster C, 2015. Improved stove interventions to reduce household air pollution in low and middle income countries: a descriptive systematic review. BMC Public Health 15, 650 10.1186/s12889-015-2024-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Tiwari M, Sahu SK, Bhangare RC, Yousaf A, Pandit GG, 2014. Particle size distributions of ultrafine combustion aerosols generated from household fuels. Atmos. Pollut. Res 5, 145–150. 10.5094/APR.2014.018 [DOI] [Google Scholar]
  52. Tryner J, Volckens J, Marchese AJ, 2018. Effects of operational mode on particle size and number emissions from a biomass gasifier cookstove. Aerosol Sci. Technol 52, 87–97. 10.1080/02786826.2017.1380779 [DOI] [Google Scholar]
  53. Van Vliet EDS, Asante K, Jack DW, Kinney PL, Whyatt RM, Chillrud SN, Abokyi L, Zandoh C, Owusu-Agyei S, 2013. Personal exposures to fine particulate matter and black carbon in households cooking with biomass fuels in rural Ghana. Environ. Res 127, 40–48. 10.1016/j.envres.2013.08.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Viana M, Rivas I, Reche C, Fonseca AS, Perez N, Querol X, Alastuey A, Álvarez-Pedrerol M, Sunyer J, 2015. Field comparison of portable and stationary instruments for outdoor urban air exposure assessments. Atmos. Environ 123, 220–228. 10.1016/j.atmosenv.2015.10.076 [DOI] [Google Scholar]
  55. Wallace L. a, Wheeler a J., Kearney J, Van Ryswyk K, You H, Kulka RH, Rasmussen PE, Brook JR, Xu X, 2011. Validation of continuous particle monitors for personal, indoor, and outdoor exposures. J. Expo. Sci. Environ. Epidemiol 21, 49–64. 10.1038/jes.2010.15 [DOI] [PubMed] [Google Scholar]
  56. Wangchuk T, Mazaheri M, Clifford S, Dudzinska MR, He C, Buonanno G, Morawska L, 2015. Children’s personal exposure to air pollution in rural villages in Bhutan. Environ. Res 140, 691–8. 10.1016/j.envres.2015.06.006 [DOI] [PubMed] [Google Scholar]
  57. World Health Organization, 2006a. Air Quality Guidelines: Global Update 2005: Particulate Matter, Ozone, Nitrogen Dioxide, and Sulfur Dioxide, World Health Organization; Geneva, Switzerland: 10.1007/BF02986808 [DOI] [Google Scholar]
  58. World Health Organization, 2006b. WHO Air quality guidelines for particulate matter, ozone, nitrogen dioxide and sulfur dioxide: global update 2005: summary of risk assessment. Geneva World Heal. Organ 1–22. 10.1016/0004-6981(88)90109-6 [DOI] [Google Scholar]
  59. Young B, Peel J, Benka-Coker M, Rajkumar S, Walker E, Broo R, Nelson T, Volckens J, L’Orange C, Good N, Quinn C, Keller J, Weller Z, Africano S, Osorto Pinel A, Clark M, 2019. Study protocol for a stepped-wedge randomized cookstove intervention in rural Honduras: household air pollution and cardiometabolic health. BMC Public Health 19, 1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Zhang H, Wang S, Hao J, Wan L, Jiang J, Zhang M, Mestl HES, Alnes LWH, Aunan K, Mellouki AW, 2012. Chemical and size characterization of particles emitted from the burning of coal and wood in rural households in Guizhou, China. Atmos. Environ 51, 94–99. 10.1016/j.atmosenv.2012.01.042 [DOI] [Google Scholar]
  61. Zhang Q, Gangupomu RH, Ramirez D, Zhu Y, 2010. Measurement of ultrafine particles and other air pollutants emitted by cooking activities. Int. J. Environ. Res. Public Health 7, 1744–1759. 10.3390/ijerph7041744 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES