Open Access
Open Peer Review

This article has Open Peer Review reports available.

How does Open Peer Review work?

Sources of variation for indoor nitrogen dioxide in rural residences of Ethiopia

  • Abera Kumie1Email author,
  • Anders Emmelin2,
  • Sonny Wahlberg2,
  • Yemane Berhane3,
  • Ahmed Ali1,
  • Eyassu Mekonen4,
  • Alemayehu Worku1 and
  • Doris Brandstrom2
Environmental Health20098:51

DOI: 10.1186/1476-069X-8-51

Received: 27 November 2008

Accepted: 18 November 2009

Published: 18 November 2009



Unprocessed biomass fuel is the primary source of indoor air pollution (IAP) in developing countries. The use of biomass fuel has been linked with acute respiratory infections. This study assesses sources of variations associated with the level of indoor nitrogen dioxide (NO2).

Materials and methods

This study examines household factors affecting the level of indoor pollution by measuring NO2. Repeated measurements of NO2 were made using a passive diffusive sampler. A Saltzman colorimetric method using a spectrometer calibrated at 540 nm was employed to analyze the mass of NO2 on the collection filter that was then subjected to a mass transfer equation to calculate the level of NO2 for the 24 hours of sampling duration. Structured questionnaire was used to collect data on fuel use characteristics. Data entry and cleaning was done in EPI INFO version 6.04, while data was analyzed using SPSS version 15.0. Analysis of variance, multiple linear regression and linear mixed model were used to isolate determining factors contributing to the variation of NO2 concentration.


A total of 17,215 air samples were fully analyzed during the study period. Wood and crop were principal source of household energy. Biomass fuel characteristics were strongly related to indoor NO2 concentration in one-way analysis of variance. There was variation in repeated measurements of indoor NO2 over time. In a linear mixed model regression analysis, highland setting, wet season, cooking, use of fire events at least twice a day, frequency of cooked food items, and interaction between ecology and season were predictors of indoor NO2 concentration. The volume of the housing unit and the presence of kitchen showed little relevance in the level of NO2 concentration.


Agro-ecology, season, purpose of fire events, frequency of fire activities, frequency of cooking and physical conditions of housing are predictors of NO2 concentration. Improved kitchen conditions and ventilation are highly recommended.


Biomass fuel is the primary source of household energy in developing countries. Fifty two percent of the global population and more than 90% of rural homes in developing countries use solid biomass fuels for cooking, heating, and lighting purposes [1]. Biomass fuel, also designated as unprocessed or dirty solid biofuel, mainly includes firewood, animal dung, agricultural residues and plant leaves.

Indoor air pollution is considered as one of the risk factors causing high burden of diseases and of premature deaths in developing countries [25]. IAP is recognized as a silent and unprotected killer among rural women and children who spend much of their time in the kitchen [6]. The burden of disease due to IAP is high in developing nations that use biomass fuels. Smith estimated that IAP in India accounted for 4.2-6.1% of the total national burden of disease, which is considered as a major public health concern. This proportion is tantamount to about half million premature deaths annually [7].

The health risk of direct exposure to biomass combustion is evident in part due to cooking practices in poorly ventilated and crowded single roomed houses in developing countries [4, 8, 9]. Studies have shown that proxy factors related to socio-economic status like income education, and use of biomass fuel to be highly associated with the level of IAP [7, 4, 10, 11]. Despite the prevailing knowledge that exposure to biomass combustion products is high [4, 9, 1219], quantitative studies on factors affecting the level of IAP are limited in developing countries [13].

Data on the measurement of indoor NO2 is limited in developed countries. Research has shown that the mean concentration of NO2 in kitchens with a gas stove varied between 26 parts per billion (ppb) and 112 ppb [20, 21], while this was 18 ppb in kitchens equipped with electric stoves [20]. Median NO2 concentrations in living rooms were 5.8 ppb in Ashford (United Kingdom), 6.1 ppb in Minorca (Spain), 23.8 in Barcelona (Spain) [22] and 11 μg/m3 (5.8 ppb) in Umea (Sweden) [23]. In low income homes of USA, mean concentrations of NO2 were found to be 43 ppb and 36 ppb in kitchens and living rooms, respectively [24] and 30 ppb in children's bedroom [25]. Increased level of NO2 was related with type of fuel and ventilation efficiency [26]. In addition, these studies consistently demonstrated that the type of fuel, purpose of fuel use, and location of indoor activities are associated with increased NO2. The use of gas stoves and heaters were often accompanied with exceeding the current World Health organization (WHO) Air Quality Guideline (AQG) of 21 ppb for 8 hours [27].

A particulate matter (PM) of various aerodynamic sizes is also used to measure the level of IAP. The mean concentration of respirable suspended particles (RSP) in homes that use biomass fuels had variation depending on the type of kitchen and fuel used: 500-2000 μg/m3 in kitchens of India [28], 1200 μg/m3 in Mozambique [29], 850-1560 μg/m3 in Guatemala [16], and 1400 μg/m3 in Kenya [30]. In Bangladesh, the average concentration of particulate matter (PM) of aerodynamic diameter of 10 microns (PM10) in kitchens that use biomass fuels ranged between 237 and 291 μg/m3 [10]. Smith showed that the concentration of PM10 commonly varied between 200-5000 μg/m3 [4]. In Ghana and Nicaragua, the average concentrations of PM2.5 varied between 320-650 g/m3 in Ghana and Nicaragua [31, 32]. These studies indicated that the type of kitchen (whether inside or outside), the type of stove (traditional or improved), ventilation condition, place of measurement (kitchen or sleeping room), and type of biomass fuel (wood, dung, or residues) were major factors affecting the concentrations of indoor air pollutants. The above cited aerial concentrations related to RSP, PM10 and PM2.5 exceeded by a factor of 2 to 40 times of the standards set by US Environmental Protection Agency, (US-EPA) for 24-hr and annual standards [33] and by a factor of 10 to 80 times of the present WHO 8 hr of air quality guideline (AQG) [27].

Biomass fuel in the form of firewood, agricultural residues and animal dung is the primary source of household energy in Ethiopia [34, 35]. The majority of rural homes have only 1 room serving all types of household activities [36]. Cooking takes place in the same room using traditional unvented stoves. Such rooms do not have functional ventilation outlets [3, 37]. Pocket studies in Ethiopia have shown high level of exposure to indoor air pollution that exceeded the WHO 1-hour and 8-hours AQG [27, 3840]. However, the results of these studies could not be generalized to a larger population due to their methodological limitations.

A recent study conducted by the authors revealed increased indoor NO2 concentration levels [41]. It also has shown that ecology and season were factors that affect NO2 concentrations. The present study was undertaken with an effort to further explore other factors associated with indoor air pollution in the rural Ethiopia. Given the high level of exposure to IAP in poorly ventilated housing units, increased morbidity and mortality due to acute respiratory infection (ARI) in the general population, especially among children under-five years old, are expected [3, 42]. Therefore, this study will have great operational relevance in achieving the Millennium Development Goal 7 (MDG 7, target 9, indicators 27 & 29) by generating important information on feasible interventions for the reduction of IAP in the Ethiopian rural homes.

Materials and methods

Study setting, indoor air sampling and analysis for nitrogen dioxide

A longitudinal study was conducted to assess the level of indoor air pollution, by measuring NO2 level, in rural households over a period of 2 years (March 2000-April 2002) in a rural district (Meskan and Mareko) in mid-southern Ethiopia. The presence of a Demographic Surveillance System (DSS) that was instituted in the District since 1986 was an opportunity for the assessment of indoor NO2 concentrations. Indoor air samples for nitrogen dioxide were taken in approximately 3,300 homes with under five children.

NO2 was detected using a modified colorimetric Saltzman method. A 24 hours indoor air sampling was done by trained local enumerators at about 3 months interval to collect data on the date, start and finishing time (hh:mm) of each sample. A digital watch was used to record the time. NO2 concentration was measured using Willems Badge that was developed at the University of Wageningen in the Netherlands [4345]. The polyethylene passive sampler consists of a small cylindrical cup equipped with 2 rings, chemically impregnated fibreglass placed at the bottom of the cup and Teflon to serve as a wind barrier. The sampler was set in a central wooden post of the rural housing (locally called "tukul") after ensuring the room was used for sleeping. The samplers in a batch were sent to the field after proper assembly. Sampling in the house was started by opening up the lid of the diffusive sampler (Willems Badge) and finished by closing it back after 24 hours of sampling.

Samplers from the field sites were transported to Addis Ababa where a centrally located laboratory was used for analysis. The absorption filter from the sampler was extracted with an acid solution of sulphanilamide and N-(1-naphtyl) ethylenediammonium dichloride (NEDA) which converts the absorbed nitrogen dioxide into NO2 - ions. The presence of these ions develops a red coloured solution whose absorbance was measured at 540 nm (Beckman DUR .64 Spectrophotometer). A standard curve NO2solution was developed to calculate the NO2 concentrations using the following equation:


'C' is indoor air concentration of NO2 in micrograms per cubic meter, 'mass' is mass of NO2 on the glass fibre used for air sampling after subtracting mass of NO2 of the blank glass fibre, 'expmin' is indoor air sampling duration in minutes, and '40' is a constant of air sampling rate for the diffusive sampler, 40 mlmn-1 [45].

The laboratory data quality was maintained using various methods. Laboratory technicians were properly trained and supervised on each day of laboratory analysis. Laboratory protocols were structured and monitored by standard practices. Internal validity were checked by the analysis of standard solution, blank absorbance, and control chart for NO2. The inter-laboratory variation was controlled by comparing the variations of NO2 concentrations of duplicate samples and exposed samples that were taken at different conditions.

Detailed description of the study area, sampling procedures, air sample location, and the analytical method is available elsewhere [41].

Assessment of determinants of indoor nitrogen dioxide

A structured questionnaire was administered at household level immediately after the completion of the air sampling to collect fuel use related data and events that occurred at the time of NO2 sampling. Type of household fuel, purpose of having fire, type of cooked food and its timing were main variables collected during data collection. Ecological and seasonal factors were also considered due to their importance to affect indoor NO2 [41]. The physical dimensions (radius, axis, and wall height), the presence of window and separate kitchen in the study homes were extracted from the 1999 census data of the study setting.

Data management and analysis

Data were entered and cleaned using the EPI INFO (version 6.04; Center for Diseases Control and Prevention, Atlanta, GA, USA and World Health Organization, Geneva, Switzerland). Consistency and completeness of each questionnaire was checked during data collection, entry and analysis. The data set was then exported first to data base file (DBF) and then to Statistical Package for Social Sciences (SPSS) (version 15.0; SPSS Inc., IL, USA) for advanced statistical analysis.

After data exploration, the original data set of indoor air NO2 concentration was transformed into logarithmic base 10 (log10) to meet the assumptions of normality for the analysis of variance (ANOVA) and linear mixed model regression analysis. In addition, box plots and stem plots were extensively used to observe the existence of outliers when comparing mean values of NO2 by categorical variables. Variables describing firing events were categorized in such a way to avoid multiple responses. These variables were type of fuel, purpose and time of fire events, and type of cooked food. One way ANOVA was employed for the detection of any differences and changes in the dependent variable represented by average indoor NO2concentration in the presence of categorical biomass fuel variables.

A mixed linear model was used to find out the relative importance of household characteristics on NO2 after ensuring assumption of normality in the dependent variable, linearity between dependent and independent variables, and collinearity between variables. A unique identifier for each household and the time variable attached to each NO2 measurement were created for this analysis. Ecology, season, type of biomass fuel, purpose and the time of the fire vents, and the frequency of food items were used for a fixed effect, while the quarterly measurements of NO2 were considered for the repeated effect. The intercept model was only used for the random effect as all households with under-five children were involved in the study. The use of unstructured covariance structure was found to be the best fit for the linear mixed model.

One way repeated measures of ANOVA was employed to analyze the presence of any difference in NO2 across time periods after structuring the data lay out into a long data format. The effect of housing characteristics (calculated indoor volume of each home, presence of window and kitchen) on indoor NO2 level was assessed using hierarchical model for a multiple linear regression analysis. Descriptive statistics, tables and inferential statistics were mainly used to present the findings. Further detailed management and data quality control are presented elsewhere [41].


Characteristics of fuel use

The study was conducted for 2 years period involving 17215 indoor air samples in 3300 households with no refusal of participation. About 98% of air samples were taken at times of fire events in households. Biomass fuel in a form of wood, crop residues, and cow dung were largely used in 71%, 65%, and 32% of samples, respectively. In 2.6% of households, other type of fuels that were seasonally used including eucalyptus dry leaves, corncobs, and leaves of false banana. The use of mixed type of fuel was a common pattern (Figure 1). All firing events, whether for cooking or not, took place mainly indoors.
Figure 1

Type of fuel and its use pattern that was observed during 24 hours of indoor NO 2 sampling, Butajira, Ethiopia, 2000-2002 (n = 16899). For each type of fuel, the number in brackets indicates the proportion of type fuel used singly and in combination with others in reference to respondent's judgment comparing with the usual days.

Cooking, lighting, heating, and insect repellent were reported as main reasons for having fire events in households in the last 24 hours during the time of air sampling (Table 1). Cooking foods and heating the space, (in 98% and 34% of the samples, respectively), were the major activities for fuel use. The use of biomass smoke for insect repellent was observed in approximately 13% of samples. Cooking and heating activities simultaneously took place in one third of the samples, while other activities in combination were rarely practiced representing less than 2% of the samples. Households had the practice of fire use three times a day. There were fewer activities at night that required the use of biomass fuel. Respondents in 73% of the samples perceived that firing at home took place relatively longer in the evenings than other times.
Table 1

Distribution of purpose of having fire and its timing that were observed during the 24 hours indoor NO2 sampling, Butajira, Ethiopia, 2000-2002 (n = 16899)

Purpose of

Having fire


# (%)*


# (%)*


# (%)*


# (%)*


# (%)*


15670 (98.9)

14583 (99.1)

16325 (98.8)


16622 (98.4)


668 (4.2)

615 (4.2)

684 (4.1)

95 (9.7)

703 (4.2)


5529 (34.9)

5185 (35.2)

5718 (34.6)

331 (33.6)

5838 (34.5)

Insect repellent

1945 (12.3)

1527 (10.4)

2009 (12.2)

212 (21.5)

2122 (12.6)

Total use of fire

15836 (93.7)


16520 (97.8)

984 (5.8)


*Percentages did not add up 100% due to multiple responses

With regard to cooked food items, kale cooking, traditional coffee ceremony, bread and local staple diet (locally called "kocho") baking were the usual type of traditional foods that were prepared during the 24 hours of indoor NO2 sampling. Traditional flat bread (locally called "injera") and its accompanying sausage (locally called "wa t") were rarely cooked, and only observed in less than 10% of the samples. Over 90% of cooking activities took place in the mornings and evenings, and rarely nights. Commonly cooked food items were kale ("gomen"), traditional coffee ceremony, and bread in 82%, 81% and 68% of samples, respectively. Other cooking activities were reported in 7.5% of the samples. These include roasting and boiling of peas and beans (locally called "kolo" and "nifro"), boiling of milk, cooking of cucumber and boiling of maize.

The association between fuel characteristics and ecological setting is presented in Table 2. Overall, there was a difference in the type of biomass fuel and its purpose of use. The use of wood predominated in the highland, while crop residues prevailed in lowlands. Heating of the housing space was more frequent in highlands (40%) than lowlands (28%). Cooking of any three food items and having three fire events per indoor air sampling day were commonly practiced in 73% and 80% of samples, respectively. Ecology was strongly related to all fuel use characteristics (p < 0.05).
Table 2

The characteristics of fuel use by ecological setting, Butajira, Ethiopia, 2000-2002 (n = 16899)


Ecology type




# (%)*


# (%)*

P value for X2


# (%)*

Type of fuel


   Only wood

6209 (73.5)

2183 (8.5)

p < 0.0001

8392 (52.1)

   Only Crop residue

2180 (25.8)

5231 (68.4)


7411 (46.0)

   Only dung

63 (0.7)

234 (3.1)


297 (1.8)


8452 (52.5)

7648 (47.5)



Purpose of having fire



8905 (98.6)

7717 (98.1)

p < 0.0001

16622 (98.4)


304 (3.4)

399 (5.1)


703 (4.2)


3609 (40.0)

2229 (28.3)


5838 (34.5)

   Insect repellent

360 (4.0)

1762 (22.4)


2122 (12.6)


9031 (53.4)

7868 (46.7)



Food frequency


   Any one food item

318 (3.6)

331 (4.5)

p < 0.0001

649 (4.0)

   Any two food items

1022 (11.6)

1423 919.2)


2445 (15.1)

   Any three food items

6393 (72.9)

5433 (73.3)


11826 (73.1)

   Any 4 & + food items

1040 (11.9)

224 (3.0)


1264 (7.8)


8773 (54.2)

7411 (45.8)



Frequency of having fire


   Any one time

147 (1.6)

175 (2.2)

p < 0.0001

322 (1.9)

   Any two times

1088 (12.0)

1333 (16.9)


2421 (14.3)

   Any three times

7386 (81.8)

6164 (78.3)


13550 (80.2)

   Any four times

410 (4.5)

196 (2.5)


606 (3.6)


9031 (53.4)

7868 (46.6)



*Percentages did not add up 100% due to multiple responses

Consistency of fire use events

There was not any difference in the time and frequency of fire use, type of fuel and type of cooked food items that occurred between the sampling time and 1 week recall period prior to that. It was only possible to identify that a religious holiday related to "Romodan" (the Moslem fasting month) was implicated to be a factor for additional cooking food items such as vegetable and meat soup, which took place relatively longer than the usual days of cooking.

The level of NO2 by the characteristics of fuel use

The relative difference in NO2 concentration by proxy fuel factors is indicated in Table 3. The concentration of NO2 was found to significantly differ by type of fuel. On the average, households that used wood had geometric mean (GM) and geometric standard deviation (GSD) of 71.2 (2.8) μg/m3. These values for cow dung and crop residues were 67.5 (2.9) μg/m3 and 56.1 (2.7) μg/m3, respectively. Any combination of biomass fuel use did not significantly affect NO2 concentration.
Table 3

24 hr indoor NO2 concentrations related to household characteristics in a bivariate analysis, Butajira, Ethiopia, 2000-2002 (n = 16899)



log NO2

mean(SD), μg/m3




X2 linear trend*





1.89 (0.45)

77.2 (2.82)

p < 0.0001




1.70 (0.43)

50.5 (2.67)




1.80 (0.45)

63.3 (2.81)


Biomass fuel type


   Only wood


1.85 (0.45)

71.2 (2.8)

p < 0.0001

p < 0.0001

   Only cow dung


1.83 (0.47)

67.5 (2.9)


   Only Crop residues


1.75 (0.44)

56.1 (2.7)





63.7 (2.8)


Purpose of having fire


   Any one activity


1.84 (0.44)

69.2 (2.7)

p < 0.0001

p < 0.0001

   Any two activity


1.76 (0.46)

57.1 (2.9)


   Any three and above


1.75 (0.45)





1.80 (0.45)

63.4 (2.8)


Type of foods cooked


   Any one food item


1.76 (0.46)

58.1 (2.9)

p < 0.0001

p < 0.0001

   Any two food item


1.74 (0.44)

54.7 (2.7)


   Any three food item


1.81 (0.45)

65.3 (2.8)


   Any 4 food & above


1.90 (0.43)

80.0 (2.7)




1.81 (0.45)

64.3 (2.8)


Time of having fire


   Any one time


1.50 (0.52)

30.8 (3.3)

p < 0.0001

p < 0.0001

   Any two time


1.75 (0.42)

56.2 (2.6)


   Any three time


1.82 (0.45)

65.7 (2.8)


   Any four times


1.82 (0.42)

66.2 (2.6)




1.80 (0.45)

63.3 (2.8)




Dry season


1.78 (0.45)

59.64 (2.8)

p < 0.0001


Wet season


1.83 (0.45)

70.0 (2.8)




1.80 (0.45)

63.3 (2.8)


*Linear trend was using EPI INFO version 6.04 STATCALC calculator Chi for linear trend. Those who used and not used the respective fuel type were considered as cases and controls.

The GM (GSD) concentration of NO2 representing a single purpose of having a fire in a household was 69.2 (2.7) μg/m3, any two purposes was 57.1 (2.9) μg/m3 and any three or more purposes was 55.6 (2.8) μg/m3. Multiple comparisons indicated that high level of NO2 was related to only single purpose compared to combination of them (p < 0.05). The concentration of NO2 had a declining linear trend from a single activity to combined activities (p-value for linear trend (p < 0.05).

Multiple food cooking was strongly related to NO2 indoor level compared to any single food preparation (p < 0.05). An increasing linear trend with the number of cooked food was also observed (p-value for linear trend (p < 0.05). Coffee drinking as well as bread and "Kocho" baking were the usual types of food that were frequently cooked.

Increased level of NO2 was significant among households that frequently used firing (p < 0.05). One time of fire use per day was related to GM (GSD) NO2 of 30.8 (3.34) μg/m3, while this was 64.3 (2.79) μg/m3 for any combination of timing of fire in reference to the morning, daytime, evening or nighttime. NO2 had an increasing trend with the frequency of cooking time (p-value for linear trend (p < 0.05).

In a mixed model linear regression, type of ecology, season, type of fuel, frequency of fire events and number of foods cooked per day were able to explain overall variations in NO2 concentrations (Table 4). A household being in a highland, wet season, use of crop residues, any time of having a fire event, frequency of food items, and interaction between ecology and season emerged as predictors of indoor NO2 concentration. The purpose of fire events did not make any effect.
Table 4

Estimates of fixed effects of fire use characteristics in households having fire events during last 24 hours, Butajira, Ethiopia, 2000-2002 (n = 3849)


95% Confidence Interval of slope

Indoor Fire use characteristics


Estimate (Slope)

Std. Error



Lower Bound







p < 0.0001









p < 0.0001











p < 0.001





Biomass fuel type


   Only wood







   Only Crop residues







   Only dung


Purpose of fire events
















   Insect repellent


Frequency of fire events


   One time per day




p < 0.0001



   Two times per day







   Three times per day







   Four times per day


Frequency of food items


   One food item




p < 0.001



   Two food items




p < 0.0001



   Three food items




p < 0.0001



   Four food items


Highland * Dry season




p < 0.0001



Highland * wet season


Dependent Variable: indoor concentration in linear scale transformed to log 10; †: Reference group

Level of NO2 by time of measurement and housing structure

The time for NO2 measurement from our database and variables on housing (calculated volume, window, and kitchen) from Butajira DSS database were extracted for the analysis. The mean (SD) NO2 measurements during our study period was 4.37 (1.90) per household, while the median was 5. Nearly 70% and 56% of the households had at least 4 and 5 measurements of NO2, respectively. There was a significant difference in the repeated measurements of NO2 concentrations over time [Wilk's Lambda = 0.993, (F(4,2047) = 3.46, p < 0.05 with multivariate eta squared = 0.007 and observed power of 89%]. The five NO2 measurements used for comparison in ascending order of time 1 to time 5 (n = 2186) had GM (GSD) as follows: 68.6 (2.58), 70.3 (2.54), 69.1 (2.47), 70.1 (2.49), and 65.2 (2.62) μg/m3 (n = 2051) Post hoc pair-wise comparison showed the overall difference was accounted between time 2 and time 5, and time 4 and time 5. The repeated NO2 measurements also differed by ecology, (F(1,2049) = 260.5, p < 0.05) and by location of households (locally called 'peasant associations'), [F(8,2042) = 48.9, p < 0.05)].

The calculated volume of "tukul" was linearly related to NO2 concentration [β (95% CI): 0.104 (0.055, 0.153]. The addition of window in the 2nd model did not show any association, while kitchen in the 3rd model showed a significant relationship with indoor NO2. The indoor volume showed positive relationship, while kitchen was negatively related to indoor NO2. Indoor volume alone indicated NO2 to vary by about 1.0 μg/m3 for every 10 m3. Both volume and kitchen were able to explain less than 1% variations in NO2 [adjusted R2 = 0.008] (Table 5).
Table 5

A relationship between indoor NO2 and housing physical structures, Butajira, Ethiopia, 2000-2002 (n = 2882)



Regression Coefficients

(Beta), (95%CI)

Std. Error

of Beta



p-value for beta

Model R2






0.104 (0.055,0.153)



p < 0.0001





0.112 ((0.061,0.162)



p < 0.0001





-0.018 (-0.046,0.010)








0.102 (0.51,0.152)



p < 0.0001





-0.022 (-0.05,0.005)








-0.5.3 (-0.89,-0.018)



p < 0.01






0.093 ((0.043,0.142)



p < 0.0001








p < 0.01




Selected household characteristics affecting the level of indoor air pollution have their own role in changing the level of NO2 in the context of our study area. Nearly all households in the study area used biomass fuel in the form of firewood, crop residues, and animal dung among which the first two predominated. The population in the study area is known for using wood most of the time throughout the year, crop residues such as stocks of maize and barley during harvest times, and animal dung during summer [46]. While biomass fuel sources are relatively cheap and easily available locally, fuels of fossil origin such as kerosene was only used to light local lamps for the interior of housing units at night. Kerosene is less affordable for rural residents to use it for cooking purposes compared to residents of urban areas such as Addis Ababa [47]. The use of biomass fuel as a primary source of household energy is consistent with findings of studies in other developing countries [1, 10, 19, 28, 38, 48].

Biomass fuel was extensively used for cooking traditional foods compared to other purposes of having fire events. Cooking was the most frequent household fuel use activity reported during all times of the day. Heating was the second most frequent activity; whereas only about a tenth of the households used biomass fuel to repeal mosquitoes at night. The use of heating indoor space predominated in colder villages, mainly in the highlands, while repealing mosquitoes prevailed in the lowlands. It is evident from the data that home heating in the highlands might have caused an additional fuel use burden, which possibly contributed to the increased concentration of indoor NO2 compared to the relatively low fuel use burden required for repelling insects in the lowlands. The extreme temperature difference [49] might have contributed to the variation in the use of fuel for heating between the two ecological settings. Low temperature in early mornings and nights is common in the highlands of Ethiopia. Villages in the lowlands inherently possess a risk to malaria caused by mosquito bites and, therefore, there is a cultural practice in the study area for using indoor firing events to repel mosquitoes [42, 50].

Traditional foods that do not require a stock for more than a day were routinely cooked. The cooking time was equally important in all cases of cooking which involved commonly the mornings, mid-days and evenings. Traditional coffee and bread making are also common daily practices in the study area. Coffee in each household was served for a group of neighborhoods nearly on daily basis, which is a cultural heritage of Ethiopia. The relative time and cost of preparing these food items are a bit less than that for "injera" and "wat" which are widely used in other parts of Ethiopia, especially in the temperate and highland areas. The practice of "injera" and "wa t" is expensive and the raw material, locally called "teff" is considered as a cash crop for the rural residents in the study area. The linear relationship between the number of food items cooked and frequency of cooking with the level of IAP is obvious given the increased respective amount of biomass fuels and the corresponding higher emission of other pollutants including nitrogen dioxide. This is in part a reflection of "dose-response" relationship.

Cooking and heating are the main household activities that lead to the excess NO2 concentration due to solid biomass fuel use in general, and in the highland areas in particular. In other studies, given the range of the purpose of biomass fuel use, cooking has been implicated as the main factor for the greater proportion of exposure to IAP in developing countries [4, 51]. Biomass fuel emits about 50 times more pollution during cooking compared to cleaner fuels [4], while the exposure magnitude of breathing in pollutants could be twice more for the same population [7]. Therefore, the magnitude of health risk due to biomass combustion can reach as much as 2-3 times greater than the risk among clean fuel users [4, 9]. Indoor smoke from biomass fuel is attributed to loss of healthy life in poor countries due to known health outcomes such as ARI, acute lower respiratory infections, and chronic obstructive lung diseases [52]. It is possible to speculate based on our findings that higher degree of exposure to indoor air smoke goes to mothers and children who often spend most of their time indoors. This implies that the attainment of Child Health MDGs would be a challenge in developing countries, like Ethiopia.

Assumption of the within-subjects (within a household over time) variation in indoor NO2 concentration by time was certainly important given the possible differences in the exposure to various fuel characteristics within households. This determined the presence of differences in exposure factors. The repeated between subject difference in NO2 demonstrated by ecological factors and seasonality is an important effect that requires closer attention for designing an appropriate intervention, as well. The study revealed only an interaction between ecology and seasonality affecting the indoor NO2 concentration by time. This was supported by our data that indicated variations of NO2 to be dependent on the type of fuel in the bivaraite analysis, although this association remained significant for a crop residue in a linear mixed model.

Ecology, season, type of biomass fuel, the purpose and time of having fire events, and the frequency of cooked food items were all found to affect the level of indoor air in a bivariate analysis, while this association was consistent in a linear mixed model except for the type of biomass fuel for wood. In western countries, the type of fuel (gas or electricity), occupancy density, the number of cooked meals, frequency of cooking, season and income were found to significantly impact indoor NO2 concentrations [20, 2426, 53]. Those studies have indicated the presence of strong link among factors responsible for the increased level of indoor NO2 both in the developed and developing countries. The new finding in our study is the presence of ecology as a factor that predominately affected the level of NO2. The high level of NO2 in the highland areas can also be explained by high proportion of wood fuel use. Wood is at least better than crop residues and animal dung in the energy ladder [54] and provides relatively better energy efficiency. When wood is used, it oxidizes relatively more indoor air nitrogen because of the relatively high combustion temperature than others.

The effect of housing volume was found to show little importance in affecting IAP as measured by NO2, which was against our hypothesis. Together with the presence of window and kitchen in the multiple linear models, there was only very small proportion of explainable variance (less than 1%) in NO2 concentration despite the statistical significance. The computed model was not able to indicate a practical relevance in explaining the direction and strength of the association between the magnitudes of indoor air pollution and the physical housing characteristics in our study area. The significant difference, however, could only be explained due to large sample size that could have picked up small differences for calculating p-value. Rural housing units, due to their nature of construction, allow the easily passage of indoor smoke through their thatched roof, open eves and unplastered or partially plastered wall, restricting the continued built up of indoor air pollutants. It is quite common to observe visually the penetration of intense smoke through such structures during active cooking times in early mornings when there is good visual contrast (personal observation). Windows in the majority of housing units in the area are represented by just small circular holes (usually < 5% of the floor area, which is often closed due to the fear of wind drafts). Furthermore, opening of windows is culturally believed to affect resident's heath in our study settings.

Lack of assessment for additional air pollutants such as PM and the absence of real time measurement for the indoor air pollution were major limitations of this study. Nevertheless, the present study has shown that ecology, season, purpose of fire events, the frequency of cooked food items and the frequency of fire events as being predictors of indoor NO2 in a rural setting.


Agro-ecology, season of the year, fire use characteristics and the physical structure of the housing were found to be important determinants of indoor air pollution in this study.

Further study on personal exposure assessment using NO2 and PM is highly recommended. In addition, relating indoor NO2 levels with commonly seen childhood diseases, such as respiratory symptoms, is another area of significant relevance for a research given the high level of IAP in our study settings. Finally, the provision of separate kitchen, improved stoves with hood, and presence of window in kitchens are highly advised to manage low level of IAP.

List of abbreviations


ambient air quality


Demographic Surveillance System


Environmental Protection Agency


geometric mean


geometric standard deviation


gram per cubic meter


indoor air pollution


Millennium Development Goal


micro grams per cubic meter


millilitre per minute


N-(1-naphtyl) ethylenediammonium


nitrogen dioxide


particulate matter


aerodynamic diameter of 10 microns PM


parts per billion


parts per million


respirable suspended matter


World Health Organization.



The Authors thank Sida/Sarec and the Umea International School of Public Health for funding this study. We are grateful for the Butajira Rural Health Program, including all enumerators, field supervisors and laboratory technicians for their contribution in data collection and data quality assurance. The encouragement extended by the School of Public Health, the Medical Faculty of Addis Ababa University is highly respected. We are also grateful to all colleagues at the School of Public Health, Addis Ababa University for their continued encouragement to complete the study. Ato Bizu Gelaye, Dr Mesganaw Fantahun and Dr Damen H/Mariam deserve respect for their encouragement in the analysis and write up of the manuscript.

Authors’ Affiliations

School of Public Health, College of Health Sciences, Addis Ababa University
Umeå International School of Public Health, Umeå University
Addis Continental Institute of Public Health
Department of Pharmacology, Medical Faculty, Addis Ababa University


  1. Rehfuess E, Mehta S, Prüss-Üstün A: Assessing Household Solid Fuel Use: Multiple Implications for the Millennium Development Goals. Environ Health Perspect. 2006, 114: 373-378.View ArticleGoogle Scholar
  2. Murray C, Lopez A: Global mortality, disability, and the contribution of risk factors: Global Burden of Disease Study. Lancet. 1997, 349: 1436-1442. 10.1016/S0140-6736(96)07495-8.View ArticleGoogle Scholar
  3. Murray CJL, Lopez AD: Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet. 1997, 349: 1269-1276. 10.1016/S0140-6736(96)07493-4.View ArticleGoogle Scholar
  4. Smith KR, Samet JM, Romieu I, Bruce N: Indoor air pollution in developing countries and acute lower respiratory infections in children. Thorax. 2000, 55: 518-532. 10.1136/thorax.55.6.518.View ArticleGoogle Scholar
  5. World Health Organization: World Health Report: Reducing Risks, Promoting Healthy Life. 2002, Geneva. World Health OrganizationGoogle Scholar
  6. World Health Organization: Indoor air pollution: the killer in the kitchen. Joint statement WHO/UNDP/5. 2004, Geneva. World Health OrganizationGoogle Scholar
  7. Smith KR: National burden of disease in India from indoor air pollution. PNAS. 2000, 97: 13286-13293. 10.1073/pnas.97.24.13286.View ArticleGoogle Scholar
  8. Pandey MR, Smith KR, Boleij JSM, Wafula EM: Indoor air pollution in developing countries and acute respiratory infections in children. Lancet. 1989, i: 427-429. 10.1016/S0140-6736(89)90015-9.View ArticleGoogle Scholar
  9. Zhang J, Smith KR: Indoor air pollution from household fuel combustion in China: a review presentation. The 10th International Conference on Indoor Air Quality and Climate, September 4-9, Beijing, China. 2005Google Scholar
  10. Dasgupta S, Huq M, Khaliquzzaman M, Pandey K, Wheeler D: Indoor air quality for poor families: new evidence from Bangladesh. Indoor Air. 2006, 16: 426-444. 10.1111/j.1600-0668.2006.00436.x.View ArticleGoogle Scholar
  11. Zhang J, Smith KR: Indoor air pollution: a global concern. British Medical Bulletin. 2003, 68: 209-225. 10.1093/bmb/ldg029.View ArticleGoogle Scholar
  12. Bruce N, Perez-Padilla R, Albalak R: Indoor air pollution in developing countries: a major environmental and public health challenge. Bull World Health Organ. 2000, 78: 1078-1092.Google Scholar
  13. Smith KR: Indoor air pollution in developing countries: recommendations for research. Indoor Air. 2002, 12: 198-207. 10.1034/j.1600-0668.2002.01137.x.View ArticleGoogle Scholar
  14. Pandey MR: Domestic smoke pollution and chronic bronchitis in a rural community of the hill Region of Nepal. Thorax. 1984, 39: 337-339. 10.1136/thx.39.5.337.View ArticleGoogle Scholar
  15. Mishra V: Indoor air pollution from biomass combustion and acute respiratory illness in preschool age children in Zimbabwe. Int J Epidemiol. 2003, 32: 847-852. 10.1093/ije/dyg240.View ArticleGoogle Scholar
  16. Albalak R, Frisancho AR, Keeler GJ: Domestic biomass fuel combustion and chronic bronchitis in two rural Bolivian villages. Thorax. 1999, 54: 1004-1008.View ArticleGoogle Scholar
  17. Schel MA, Hessen JO, Smith KR, Briuce N, Mccracken , Lopez V: Childhood asthma and indoor wood smoke from cooking in Guatemala. Journal of exposure analysis and environmental epidemiology. 2004, 14: S110-S117. 10.1038/sj.jea.7500365.View ArticleGoogle Scholar
  18. Shrestha I, Shrestha S: Indoor Air pollution from biomass fuels and respiratory Health of the exposed population in Nepalese households. Int J Occup Environmental Health. 2005, 11: 150-160.View ArticleGoogle Scholar
  19. Derege K: Socio-demographic characteristics and indoor air pollution as risk factors for acute lower respiratory infections in under-five children in Addis Ababa. Ethiop J Health Dev. 1997, 11: 315-325.Google Scholar
  20. Goldstein BD, Melia RJ, Chinn S, Florey CV, Clark D, John HH: The relation between respiratory illness in primary schoolchildren and the use of gas for cooking--II. Factors affecting nitrogen dioxide levels in the home. J Epidemiol. 1979, 8: 339-345. 10.1093/ije/8.4.339.Google Scholar
  21. Belanger K, Gent JF, Triche EW, Bracken MB, Leaderer BP: Association between nitrogen dioxide exposure with respiratory symptoms in children with asthma. Am J Respir Crit care Med. 2006, 173: 297-203. 10.1164/rccm.200408-1123OC.View ArticleGoogle Scholar
  22. García A, Pichini S, Basagaña X, Puig C, Vall O, Torrent M, Harris J, Sunyer J, Cullinan P: Concentrations and determinants of NO2 in homes of Ashford, UK and Barcelona and Menorca, Spain. Indoor Air. 2004, 14: 298-304. 10.1111/j.1600-0668.2004.00256.x.View ArticleGoogle Scholar
  23. Hagenbjork GA, Forsberg B, Hestvik G, Karlsson D, Wahlberg S, Sandstrom T: Measurements of indoor and outdoor nitrogen dioxide concentrations using a diffusive sampler. Analyst. 1996, 121: 1261-1264. 10.1039/an9962101261.View ArticleGoogle Scholar
  24. Zota A, Adamkiewicz G, Levy JI, Spengler JD: Ventilation in public housing: implications for indoor nitrogen dioxide concentrations. Indoor Air. 2005, 15: 393-401. 10.1111/j.1600-0668.2005.00375.x.View ArticleGoogle Scholar
  25. Hansel NN, Breysse PN, McCormack MC, Matsui EC, Curtin-Brosnan J, Williams DL, Moore JL, Cuhran JL, Diette GB: A Longitudinal Study of Indoor Nitrogen Dioxide Levels and Respiratory Symptoms in Inner-City Children with Asthma. Environ Health Perspect. 2008, 116: 1428-1432.View ArticleGoogle Scholar
  26. Levy JI, Lee K, Yanagisawa K, Hutchinson DP, Spengler J: Determinants of nitrogen dioxide Concentrations in indoor ice skating rinks. Am J Public Health. 1998, 88: 1781-1786. 10.2105/AJPH.88.12.1781.View ArticleGoogle Scholar
  27. World Health Organization: WHO Air Quality Guidelines Global Update 2005 for articulate matter, ozone, nitrogen dioxide and sulfur dioxide. Geneva. 2006, []Google Scholar
  28. Balakrishnan K, Parikh J, Sankar S, Padmavathi R, Srividya K, Venugopal V, Prasad S, Pandey VL: Daily average exposures to respirable particulate matter from combustion of biomass fuels in rural households of Southern India. Environ Health Perspect. 2002, 110: 1069-1075.View ArticleGoogle Scholar
  29. Ellegard A: Cooking fuel smoke and respiratory symptoms among women in low-income areas in Maputo. Environ Health Perspec. 1996, 104: 980-985. 10.2307/3433146.View ArticleGoogle Scholar
  30. Wafula EM, Onyango FE, Thairu H, Boleij JS, Hoek F, Ruigewaard P, Kagwanja S, De Koning H, Pio A, Kimani E, et al.: Indoor air pollution in a Kenyan village. East Afr Med J. 1990, 67: 24-32.Google Scholar
  31. Center for Entrepreneurship in International Health and Development, School of Public Health (CEIHD), University of California, Berkley: Indoor air pollution monitoring in Ghana. []
  32. McCracken J, Charron D: Evaluation of the Efficacy and Effectiveness of the EcoStove for reducing indoor air pollution exposures among Nicaraguan women. 2003, []Google Scholar
  33. U.S Environmental Protection Agency: National Ambient Air Quality Standards (NAAQS). []
  34. Central Statistical Agency: The Federal Democratic Republic of Ethiopia. Welfare Monitoring Survey 2004. Analytical Report. 2004, Addis Ababa. Central Statistical AgencyGoogle Scholar
  35. Central Statistical Agency and ORC Macro: Ethiopia demographic and health survey 2005. 2006, Addis Ababa, Ethiopia and Calverton, Maryland, USA. Central Statistical Agency and ORC MacroGoogle Scholar
  36. Kumie A, Berhane Y: Crowding in a traditional rural housing ("Tukul") in Ethiopia. Ethiop J Health Dev. 2002, 16: 303-308.Google Scholar
  37. Fantahun M, Kumbi S, Degu G, Kebde Y, Admassu M, Haile W, Hailu S: Dabat Rural Health Project, North west Ethiopia: report of the baseline survey. Ethiop J Health Dev. 2001, 15: 1-27.Google Scholar
  38. Kebede F: Survey of indoor air pollution problems in the rural communities of Jimma, Southwest Jimma. Ethiop J Health Sci. 2002, 12: 1-13.Google Scholar
  39. Jurgen U: Indoor Air Pollution in Rural Tigray, Ethiopia. 1994, []Google Scholar
  40. Center for Entrepreneurship in International Health and Development (CEIHD) School of Public Health, University of California, Berkeley, California in collaboration with Gaia Association: Indoor air pollution monitoring summary for the Gaia Association Clean Cook Stove tests for Somali Region. 2007, Bonga Refugee Camp,, []Google Scholar
  41. Kumie A, Emmelin A, Wahlberg S, Berhane Y, Ali A, Mekonnen E, Brandstrom D: Magnitude of indoor NO2 from biomass fuels in rural settings of Ethiopia. Indoor Air. 2009, 19: 14-21. 10.1111/j.1600-0668.2008.00555.x.View ArticleGoogle Scholar
  42. Ministry of Health, Ethiopia: Health and Health related indicators 2005/2006. 2005, Addis Ababa. Ministry of HealthGoogle Scholar
  43. Willems JJH, Adema EH: Passive monitoring of NH3, SO2, NO2, and Ozone. Allergrini I. Development of analytical techniques for atmospheric pollutants. 1992, Workshop organized by EC and CNR. RomeGoogle Scholar
  44. Hagenbjork GA, Forsberg B, Hestvik G, Karlsson D, Wahlberg S, Sandstrom T: Measurements of indoor and outdoor nitrogen dioxide concentrations using a diffusive sampler. Analyst. 1996, 121: 1261-1264. 10.1039/an9962101261.View ArticleGoogle Scholar
  45. Hagenbjork GA, Roger L, J OL, Doris K: Validation of diffusive sampler for NO2. J Environ Monit. 1999, 1: 349-352. 10.1039/a902937k.View ArticleGoogle Scholar
  46. Berhanu Eshete: Exposure assessment to indoor smoke pollution in a rural Ethiopia. MPH thesis. 1999, Addis Ababa UniversityGoogle Scholar
  47. Central Statistical Authority, Ethiopia: Statistical Abstract. 2007, Addis Ababa. Central Statistical AgencyGoogle Scholar
  48. Balakrishnan K, Sambandam S, Ramaswamy P, Mehta S, Smith KR: Exposure assessment for respirable particulates associated with household fuel use in rural districts of Andhra Pradesh, India. Journal of Exposure Analysis and Environmental Epidemiology. 2004, 14: S14-S25. 10.1038/sj.jea.7500354.View ArticleGoogle Scholar
  49. Kloos H, Woldu Z: The physical and biotic environment: its features and health implications. Epidemiology and Ecology of Health and Disease in Ethiopia. Edited by: BerhaneY, Haile-Mariam D, Kloos H. 2006, Addis Ababa: Shama Books, 10-30.Google Scholar
  50. Adhanom Tedros, Deressa Wakgari, Hanna Karen, Getachew Asfaw, Seboxa Teshale: Malaria. Epidemiology and ecology of health and disease in Ethiopia. Edited by: Berhane Y, Haile-Mariam D, Kloos H. 2006, Addis Ababa: Shama Books, 556-576.Google Scholar
  51. Ezzati M: Indoor air pollution and health in developing countries. Lancet. 2005, 366: 104-106. 10.1016/S0140-6736(05)66845-6.View ArticleGoogle Scholar
  52. Ezzati M: Selected major risk factors and global and regional burden of disease. Lancet. 2002Google Scholar
  53. Baxter LK, Clougherty JE, Laden F, Levy JI: Predictors of concentrations of nitrogen dioxide, fine particulate matter, and particle constituents inside of lower socioeconomic status urban homes. J Expo Sci Environ Epidemiol. 2007, 17: 433-444. 10.1038/sj.jes.7500532.View ArticleGoogle Scholar
  54. United Nations Environment program: GEO: Global Environment outlook. GEO Year Book. []


© Kumie et al; licensee BioMed Central Ltd. 2009

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.