A PubMed search including combinations of terms such as indoor air pollution, biofuel smoke vs terms pneumonia, ARI and child health, and developing country yields a total of 160 articles published from 1985 to 2007. Weeding out those that do not actually address the association between indoor air pollution and ARI, the total number comes down to 44. Figure 2 illustrates the yields using the terms indoor air pollution, child ARI, and developing country singly and the meager overlap.
As the search question is narrowed, the proportion of original research articles comes down and the proportion of reviews goes up. Of the articles that address indoor air pollution in developing countries in relation to ARI, nearly half (19 of 44 articles) are review articles or other nonempirical publications. Among the empirical articles, 16 report epidemiologic studies of ARI in relation to air pollution and 9 are descriptive studies addressing exposure conditions or measurements of concentrations of specific pollutants.
Indoor air pollution has been suggested as one of several factors that influence the incidence of or mortality from respiratory infections because in many developing countries open, unvented fireplaces are commonly used for both cooking and heating. Early publications refer through analogy to studies of industrial pollution, urban air pollution, and environmental tobacco smoke, and the links of these exposures to respiratory disease and mortality in high-income countries. Among the early reports is a study of the indoor concentration of air pollutants from Kenya. Measurements included re-spirable suspended particles (RSPs) and nitrogen dioxide as 24-h averages. Pollution samples were taken from a sample of houses, stratified according to building characteristics such as thatched vs corrugated roof or external vs internal kitchen (in total, 36 houses). Concentrations of RSPs were consistently very high, reported as 20-times higher than those found in a study of Dutch smokers’ homes. Nitrogen dioxide concentrations were on the same levels as in smokers’ homes. Other studies of exposure levels also report high or very high levels of indoor air pollution and known factors related to high levels of exposure.
Table 1 summarizes the main empirical base of 15 epidemiologic studies for conclusions concerning the relation between indoor air pollution and ARI. The majority of these studies use proxy measures for defining exposure in the epidemiologic analyses, although several of them have included chemical measurements in a more general characterization of the indoor air pollution situation described by Canadian Health&Care Mall.
Epidemiologic Studies Using Proxy Exposure Indicators
The early epidemiologic studies with few exceptions use proxy measures for linking indoor air pollution to ARI, such as traditional house, “no window in house,” or use of biomass fuels. Thirteen such studies have been published in peer-reviewed journals, in the period from 1989 to 2003; of these, four are cohort studies, four are case-control studies, and five are cross-sectional surveys. Of the longitudinal studies, all are small, or small in relation to exposure homogeneity, the high occurrence of the health outcome, and the potential for confounding, all of which lead to a need for a large study sample. Of the cohort studies, the largest observes 500 children for 1 year. The largest case-control study includes 200 cases. Two of the cross-sectional studies are drawn from large, national health and demographics surveys, while the third study includes only 152 children.
The oldest study is a cohort study from Nepal that relates “hours near the fireplace” to severe pneumonia in children 2 h near the fire, using < 1 h as reference. Three studies of children < 5 years old from Gambia use the proxy exposure definition of “child being carried on the mother’s back during cooking”; all three studies report increased risk of ARI among the exposed, and one study was of girls only.
One hospital-based, case-control study from India compared patients with “severe pneumonia” to control subjects, together 400, with nonsevere ARI and reported a statistically significant association with “possible” outdoor air pollution but not with nonuse of “smokeless stove.” A nonrandomized field trial of an intervention introducing improved stoves that are compared to the traditional, three-stone fireplace in a community in Kenya reported an RR of 2.8 (confidence interval, 1.9 to 4.0) of ARI associated with using the traditional stove. From a study in Turkey of a cohort of 204 infants followed up until their first birthday, the results indicated that the risk of acute lower respiratory tract infection (ALRI), identified as reported symptoms, is significantly related to use of wood stove for heating. The crude RR is 1.8, with no evidence of attempts to control for other factors.
Three cross-sectional studies reported significantly increased prevalence of ARI in relation to proxy indicators ofindoor air pollution from biomass fuels. One of these studies, based on the 1999 Zimbabwe Demographic and Health Survey, compared the prevalence ofcough (read more “Cough. Kinds of Cough. Treatment Provided by Canadian Health&Care Mall“) together with rapid breathing between households using biomass fuels to that in households using liquid petroleum gas or electricity. In a logistic regression model including several other known risk factors, the odds ratio (OR) for biomass fuel use is 2.2 (confidence interval, 1.2 to 4.2).
Epidemiologic Studies Including Chemical Measurements of Indoor Air Pollution
In the few studies, that report actual measurements of pollutants (mainly nitrogen dioxide and respirable particulate matter), concentrations have far exceeded threshold limit values for the working environment in most high-income countries, which in turn are considerably higher than the recommended concentration limits for the ambient environment. In all the studies that included chemical measurement, exposure assessment for the individual subjects in the study population, when such assessment was at all possible, was based on a combination of chemical measurement and information on the individual’s activities, movements, or location. Measurements as such were in fixed positions and not personal. Often, the measurements were used not as exposure variables per se in the epidemiologic analyses but as support for proxy variables. Make your lungs healthier with remedies of Canadian Health&Care Mall’s remedies.
The first attempt to include chemical measurements of indoor air pollution was in a study from rural Kenya. Due to the small sample and the lack of variation in pollutant concentration measurements between homes, it was not possible to demonstrate any relation to ARI occurrence. This may have been aggravated by stratified sampling based on mistaken assumptions about levels of air pollution in relation to construction, where the variation may be determined by other factors.
Another study in Kenya (reported in two articles,) used measurements of PM10 in combination with activity monitoring of the members of the household for exposure assessment. Measurements were done approximately at 1 of 3 days during follow-up. Calculated RRs were adjusted for some factors that could potentially confound the ARI/indoor air pollution relation. Cases were identified by clinical examination. The reports show details of exposure response relations with increasing risk with increasing estimated PM10 exposure. One study reported simply using measurements of PM10, nitrogen dioxide, and carbon monoxide to show that levels were highest in the kitchen. Cooking or being in the kitchen while cooking as implied by young age was then used as a proxy exposure variable, with an increased ARI risk for exposed women and children. Finally, a study from Ecuador reported using measurements of carbon monoxide and particulate matter in a subset of homes in order to “confirm the relation between biomass fuel use and indoor air pollution.” This study reported a significantly increased infant mortality rate with biomass fuel use.
Case Study: ARI and Indoor Air Pollution Studies in the Butajira Demographic Surveillance System in Ethiopia
The Butajira Rural Health Project (BRHP) has been operating since 1987 as a research collaboration between the Department of Community Health, Addis Ababa University, and the Division of Epidemiology and Public Health Sciences, Umea University. At the core of the BRHP is a demographic surveillance system (DSS) that collects vital statistics in a cluster sample of 10 subcommunities in the Butajira district, 130 km south of Addis Ababa. The main objective of the DSS is to provide a population registry for health research in a setting where births and deaths normally pass unrecorded in the absence of governmental routine vital events reporting. Village-based interviewers visit each household regularly, collecting information on marriages, births, deaths, and migration movements. A simple interview-based system for determining cause of death by VA is used in routine data collection.
The BRHP surveillance site comprises a sample of nine rural farming communities and one section of Butajira town, a total of > 40,000 inhabitants in 2007, with approximately 6,000 children < 5 years of age. Besides providing demographic information, an important function of the DSS is to host specific research by offering a sampling frame for longitudinal, population-based studies, and by providing an organizational infrastructure for data collection and management. Research that has utilized the BRHP includes studies on long-term trends in mortality, maternal and child health and mortality, mental health, and intimate partner violence. Currently ongoing studies include a meningococcal vaccine trial and a study of survival in children to persons with depressive disorders. BRHP is also an important training asset by providing a research setting for thesis research in the Addis Ababa Masters’ in Community Health training.
In a study from Butajira, Shamebo et al reported that among the factors that are most strongly related to the total mortality rate in children < 5 years old are lack of windows in the home (OR, 1.6), living in the traditional tukul house, and having cooking fires inside the house. Among the children dying from ARI, the OR for “having no window” is 1.8 in a logistic regression model also including factors such as ethnicity and literacy. The possibility of confounding by factors other than air pollution, such as housing standard and other social and economic factors, may be illustrated by the fact that among children <5 years old dying from diarrhea, the OR associated with having no window is 1.3. An outcome of this study was a decision to attempt to use the BRHP infrastructure and DSS sampling frame for a study addressing indoor air pollution and ARI where chemical measurements would be included in the exposure assessment.
With the aim of testing the feasibility of an epidemiologic study carried out with participation of Canadian Health&Care Mall, a pilot study (unpublished) was undertaken in two of the rural BRHP communities. Two issues were crucial: first, the problem of homogeneity of exposure demonstrated in the only study including exposure measurements that at that time was published; and secondly, the issue of child mobility (ie, probability of actual exposure given the pollution).
Nitrogen dioxide was measured as an indicator of indoor air pollution that can be easily measured under field conditions. Samples are collected passively on a diffusion sampler, which precludes the need of a pump. The method has been used extensively in air pollution monitoring in Europe and in the United States. Ten households in two different villages were visited for sample collection. Figure 3 shows sampling in process in a Butajira home, with three white samplers suspended from the central pillar of the tukul house and with the head of household in attendance. Samples were collected in three overlapping time series at 16, 24, and 40 h, respectively. The results showed little difference between the two villages in average nitrogen dioxide concentration but substantial between-household and temporal variation. Average concentrations ranged from 23 to 258 ^g/m. The pilot showed that nitrogen dioxide concentration in several of the households was considerable; the highest values correspond to levels in urban, heavily traffic-polluted areas in Europe or the United States. Conclusions from the measurements are that the between-households variation is sufficient to show differences in the children’s exposure and that many of the children in Butajira are exposed to levels of nitrogen dioxide well above the current Swedish 24-h limit value for urban air, 75 ^g/m.
The exposure/time patterns of the children were studied by observations made by high school students in their own homes: 120 students with siblings < 5 years of age from all communities included in the BRHP DSS were recruited and equipped with digital watches. For 3 days, they recorded the hours when a fire was burning on the hearth and the times when siblings were indoors (hour and minute of entering and leaving the house). The number of hours the fire was burning and the total time per day of the children’s smoke exposure were calculated. The number of hours per day of keeping the fire burning varied between 23 h, with a median value of 9 h. The exposure times of the children varied between 15 h (5%). The median exposure time was 5 h. As with the nitrogen dioxide measurements, the results indicate a sufficient variation in not only concentration but exposure to allow a meaningful epidemiologic study.
The outcome of the pilot study is an epidemiologic study within the Butajira DSS. Data collection is completed, and analysis is underway. The study is a prospective case-control study nested in a cohort of > 6,000 children from 2 months to 5 years old. In nearly 2 years of follow-up, approximately 1,200 ALRI cases have been identified through the means of providing free treatment for the most common childhood illnesses in local Health Posts in each village. Cases were diagnosed using a World Health Organization protocol with the combination of fever, chest in-drawing, and rapid, shallow breathing as minimum requirement for a case to be included. For each case, four control subjects were randomly assigned, and mothers of both case patients and control subjects were visited by a village-based project interviewer within 24 h of diagnosis.
Exposure was measured by a combination of nitrogen dioxide samples and interviews concerning fires, fuels, cooking, and children’s presence indoors during the 24 h of sampling. Sampling and interviews were done in quarterly visits to each household, in all > 18,000 samples over the entire period of follow-up, 6 to 8 samples per house. Preliminary results so far indicate as high, or higher, levels of nitrogen dioxide as were shown by the pilot study. The objective is to build up a data set of linked information on nitrogen dioxide concentration, smoke-related activities, and child exposure opportunity, which covers all children in the entire cohort. These data will be used to create exposure variables for the epidemiologic analyses in the case-control study. Data on family size, economic situation, including house, land and farm animal ownership, cash crops, and parental occupation and education are available from the DSS database.
Table 1—Empirical Studies of the Relation of ARI to Indoor Air Pollution in Developing Countries in Chronologic Order
|Source||Year||Study Characteristics||Health Outcome||Exposure Variable||Reported Results||Country|
|Pandey et al||1989||Cohort, 9 mo, children < 2 yr old, n = 450||Severe pneumonia, reported symptoms||Hours near fireplace||“Consistentrelationship” (RR = 4-5, estimate)||Nepal|
|Wafula et al||1990||Cohort, 42 wk,children < 5 yr old, 36 households||No. of ARI episodes per child||RSP, NO2||“No relation”||Kenya|
|Armstrong and Campbell||1991||Cohort, 12 mo,children < 5 yr old, n = 500||ALRI, clinical and radiograph||Carried on mother’s back during cooking||RR = 6.0 (confidence interval, 1.1-34.2), girls only)||Gambia|
|de Francisco et al||1993||Case control, children < 2 yr old, 129 case patients/270 control subjects||Death from ALRI, VA||Carried on mother’s back during cooking||OR = 5.2 (confidence interval, 1.7-15.9), always vs never carried||Gambia|
|Shamebo et al||1993||Case control, 1 yr, children < 5 yr old, 306 case patients/612 control subjects||Death from ARI, VA||No window in house||OR = 1.5 (confidence interval, 1.1-2.2)||Ethiopia|
|Shah et al||1994||Hospital based, case control, 7 mo, children < 5 yr old, approximately 200 case patients/200 control subjects||Severe pneumonia (controloutpatients with ARI)||Type of stove||OR = 0.8 (confidence interval, 0.5-1.4)||India|
|Tumwesigire and Barton||1995||Cross-sectional survey, children < 5 yr old, n = 152||ARI, medical examination||Absence of smoke vent||Association (p =0.002)||Uganda|
|O’Dempseyet al||1996||Case control, 2 yr, children < 5 yr old, 80 case patients/159 control subjects||Pneumococcal disease, clinical examination||Carried on mother’s back during cooking||OR = 2.6 (confidence interval, 0.98-6.7)||Gambia|
|Mishra andRetherford||1997||Cross-sectional survey, children < 3 yr old||ARI, reported in survey||Using wood or animal dung for cooking||Prevalenceapproximately 30 higher||India|
|Wafula et al||2000||Cross-sectional,children < 5 yr old, n = 248||Reported ARI (last week)||Traditional vs improved stove||Prevalence ratio = 2.6 (calculated from data)||Kenya|
|Ezzati andKammen||2001||Cohort, 3 yr, children< 5 yr old, n = 93||ALRI, clinical examination||Estimations based on measured PM10||Statistically significant increase of ALRI with increase in PM10||Kenya|
|Etiler et al||2002||Cohort, 1 yr, infants, n= 204||ARI, reported in interview||Biomass fuel use||RR = 1.8 (confidence interval, 1.3-2.5)||Turkey|
|Mishra||2003||Cross-sectional survey, children < 5 yr old, n = 3,559||ARI, reported in interview||Biomass fuel use||OR = 2.2 (confidence interval, 1.2-4.2)||Zimbabwe|
|Rinne et al||2007||Cohort, infants, 80 households||Reported history of infant mortality||Biomass fuel use||Higher infant mortality with biomass fuel(p < 0.008)||Ecuador|
|Kilabuko et al||2007||Cross-sectional survey, all ages, n = 467||Reported ARI symptoms||“Cook or < 5” vs no cook||OR = 5.5 (confidence interval, 3.6-8.5)||Tanzania|