Wood stove use and other determinants of personal and indoor exposures to particulate air pollution and ozone among elderly persons in a Northern Suburb

Abstract A six‐month winter‐spring study was conducted in a suburb of the northern European city of Kuopio, Finland, to identify and quantify factors determining daily personal exposure and home indoor levels of fine particulate matter (PM2.5, diameter <2.5 µm) and its light absorption coefficient (PM2.5abs), a proxy for combustion‐derived black carbon. Moreover, determinants of home indoor ozone (O3) concentration were examined. Local central site outdoor, home indoor, and personal daily levels of pollutants were monitored in this suburb among 37 elderly residents. Outdoor concentrations of the pollutants were significant determinants of their levels in home indoor air and personal exposures. Natural ventilation in the detached and row houses increased personal exposure to PM2.5, but not to PM2.5abs, when compared with mechanical ventilation. Only cooking out of the recorded household activities increased indoor PM2.5. The use of a wood stove room heater or wood‐fired sauna stove was associated with elevated concentrations of personal PM2.5 and PM2.5abs, and indoor PM2.5abs. Candle burning increased daily indoor and personal PM2.5abs, and it was also a determinant of indoor ozone level. In conclusion, relatively short‐lasting wood and candle burning of a few hours increased residents’ daily exposure to potentially hazardous, combustion‐derived carbonaceous particulate matter.

size distribution, and chemical composition, as well as building characteristics. 1,6,7 The systematic review of Janssen et al 8  and sunlight are the key actors. 10 In this study area, traffic exhaust emissions and residential wood combustion were the main sources of precursors of tropospheric ozone. As ozone formation requires sunlight for completion, outdoor ozone concentration is low in winter, especially in high latitudes, and increases considerably in spring. 11 The concentrations are usually somewhat lower in city centers compared to suburban and rural areas, because ozone reacts with NO originating as the primary nitrous emission product from tailpipes. 11 In short-term health studies, ozone has been associated with both respiratory and cardiovascular health outcomes including mortality and hospital admissions, 6,12,13 and also with out-of-hospital cardiac arrest, according to the systematic review and meta-analysis conducted by Zhao et al. 14 However, in a recent quantitative review of Atkinson et al, 15 annual outdoor ozone levels were not systematically associated with morbidity.
Generally, there is a lack of information on ozone levels and its determinants in home indoor environments in the northern subarctic climate.
In most short-term studies, exposure assessment is based on air pollution measurements at one or few central sites. However, central site measurements may not effectively illustrate home outdoor concentrations, especially in areas with intense low-height emissions from local sources such as residential wood combustion or traffic emissions in major roads. [16][17][18][19] Corresponding weak relationships have also been found between outdoor and personal exposure concentrations of PM 2.5 . 20,21 Despite significant positive correlations observed between indoor and outdoor ozone concentrations, 22,23 the indoor ozone levels have been low compared to the outdoor levels and are strongly influenced by the building ventilation and airtightness of its shield. Since people spend most of their daily time indoors in developed countries, their total personal exposure depends mostly on indoor concentrations of air pollutants, in other words on infiltrated outdoor air pollutants and indoor-generated pollutants.
In the review article of Morawska et al, 24 it has been assessed that 10%-30% of the total burden of disease from particulate matter exposure in developed countries resulted from indoor-generated particles. However, there is far less information available on exposures to indoor-generated PM 2.5 than there is on outdoor PM 2.5 .
Smoking and cooking have been suggested to be the main sources of indoor air particles at home. [25][26][27][28][29] Other possible indoor sources of air pollution are the use of personal care products, 30,31 candle burning, 26,29 wood burning, 21 home cleaning, 28 pets, 30 and gypsum/wallboard. 31 Regardless, in developed countries, there are still a limited number of studies where personal and indoor determinants of wood burning are studied. And those already reported are in principally in areas with high concentrations. 21 The objective of this study was to identify and quantify factors determining personal exposure and indoor levels of PM 2.5 and its light absorption coefficient (PM 2.5abs ) as well as indoor ozone concentration in houses located in a suburb of a Northern European city that has low annual air pollution levels. Our special interest was to evaluate the relevance of wood combustion as an exposure determinant in a suburb where wood is mainly burned in secondary heating devices and in sauna stoves.

| Study design
The study was conducted in a suburb called Jynkkä in the City of Kuopio, Central Eastern Finland, during the heating season be- about 98 000 inhabitants. Around 3400 inhabitants lived in this low population density suburb, which is located 6-8 km south of the city center of Kuopio. Nearly 100% of the houses in this suburb had district heating as their primary heating source.
A central air pollution measurement site was set up to determine air pollutant concentrations prevailing in the study area. The sample intakes were about 5 m above ground level. Sampling inlets were placed 1.5 m above the roof of the container (container roof was about 3.5 m above the ground) in order to allow free air movements around the inlets and prevent vandalism. The nearest relatively busy road (traffic intensity >5000 vehicles per 24 hours) was at a distance of over 1 km. Estimate of traffic intensity was based on a traffic census made by the City of Kuopio, and the distance was Detailed results from the central site outdoor air quality measurements have been previously described by Yli-Tuomi et al. 19 Fine particle sources that were identified in the study area are long/regional-range Information on housing characteristics as well as on subjects' activities, potentially affecting personal exposure or indoor pollutant concentrations, was collected with questionnaires and timeactivity diaries. Baseline questionnaires filled in by the researchers were used to collect information on constant variables during the study period, such as gender, education, home type, ventilation system, and distance between home and collector road. These are called "time-invariant determinants" in this study. Type of air exchange where building air supply and extraction are based on pressure differences inside and outside the building is referred as natural ventilation system. Ventilation is referred to as mechanical when air extraction or both air supply and extraction are carried out using a fan or other mechanical system. These types of ventilation systems are typical for the homes in this study, as they were built mainly in the seventies and eighties (building years: [1976][1977][1978][1979][1980][1981][1982][1983][1984][1985][1986][1987]2000). A self-administrated questionnaire, filled in during each 22-hour measurement period, was used to collect information on potential "time-varying determinants", such as cooking, use of a wood stove (room heater) or sauna stove, smelling of wood smoke odor, and time spent in different microenvironments.
Time-activity diaries were filled in with 15-minute resolution.

| Sampling and sample analysis
Sampling methods have been described in more detail in the paper of Yli-Tuomi et al. 19 Briefly, PM 2.5 samples were collected by using  concentrations were calculated from 9 am to 9 am the next day.

| Statistical analyses
The strength of the relationships between daily personal, home in- were also tested in the final models. In general, no substantial changes in the results were found in various sensitivity analyses of the multi-determinant models. After the exclusion of high residuals (using either 2 or 3 as the cutoff value), the associations became slightly weaker, but the general pattern remained the same. After replacement of the original cooking variables (cooking with frying pan or in oven) with a more general cooking variable, including all cooking activities except for a using microwave or coffee machine, the association with indoor PM 2.5 was no longer significant. Previously modeled outdoor PM 2.5 concentration from wood burning in the study area 19 was not a determinant of personal or indoor PM 2.5 or PM 2.5abs concentration.   According to Fadeyi et al, 38 ozone concentrations in indoor environments are generally low, but they may be strongly influenced by the use of certain types of room air-cleaning devices, some of which generate considerable amounts of ozone. Photocopiers and laser printers may also produce ozone. 11 Only five subjects in our study used a room air cleaner, and only on eight days in total. Thus, the influence of these particular devices on indoor ozone concentration was minor.

| D ISCUSS I ON
Candle burning has been recognized as a major indoor particle source. 26,29 In this study, candle burning affected both personal and indoor PM 2.5abs levels. Candle burning increased personal PM 2.5abs levels 8% and indoor PM 2.5abs 10%. Interestingly, candle burning was also associated with a slightly higher (3%) home indoor ozone concentration in this study. However, we have no obvious explanation for this finding, which also may have appeared merely by chance. In one previous study, a negative association between candle burning and indoor ozone has also been reported. 39 Wood burning was associated with higher levels of personal and indoor PM 2.5abs , and it also increased personal PM Wood combustion is a significant source of ambient air pollution in northern countries during the winter season. In Finland, district heating is common in the cities, but small-scale wood combustion as a secondary heating source and wood-fired sauna stoves are widely used in residential areas. In the study area, the mean outdoor concentration of PM 2.5 from wood burning was 0.9 µg/m 3 on the basis of a previous source apportionment. 19 However, the central site outdoor were elderly persons and almost all of them were retired. Therefore, results may not be generalizable for other age groups, such as school children or the working age population, who spend a notable part of the day outside the home.
Smoking indoors is a dominant source of the PM 2.5 and PM 2.5abs in indoor air, if anyone is smoking indoors. 25,26,41 A strength of this study was the lack of smokers, allowing the estimation of other, more sparsely studied indoor emission sources. Moreover, measurements of both personal and indoor levels of the selected pollutants could be done. Finally, the low level of regional outdoor air pollution enhanced our changes to identify and quantify individual indoor sources, as the absolute amount of particles in outdoor air did not strongly dominate personal exposure and home indoor concentrations.

| CON CLUS IONS
Local outdoor air concentrations of PM 2.5 , PM 2.5abs , and ozone were important determinants of their daily home indoor and personal exposure levels in a suburb of the northern European city of Kuopio, Finland, which has a generally low level of air pollution. Natural ventilation of the detached and row houses increased total personal exposure to PM 2.5 , but not to PM 2.5abs , when compared to mechanical ventilation. Only cooking out of the recorded household activities increased indoor PM 2.5 . The use of a wood stove room heater or wood-fired sauna stove was associated with elevated concentrations of personal PM 2.5 and PM 2.5abs , and indoor PM 2.5abs . Wood and candle burning were significant predictors of residents' higher indoor and personal exposures to potentially hazardous, combustionderived carbonaceous particulate matter.

ACK N OWLED G EM ENTS
This study was supported by the Academy of Finland (contract no. 10155) and intramural funding from THL. Additional funding for Timo Lanki came from the Academy of Finland (122783). Erkki Pärjälä from the City of Kuopio and Pauliina Taimisto are greatly acknowledged for their help with the measurements.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.