MATERIAL AND METHODS

Study subjects

The BAMSE cohort study design is described in detail elsewhere.¹⁴ Briefly, the cohort comprises 4089 children born between 24 February 1994 and 20 November 1996 in certain parts of central and northwestern Stockholm. Both urban and suburban districts were represented, including areas with varying types of traffic exposure, different types of buildings, dwellings with and without gas stoves for cooking, as well as differences in socioeconomy. Access to a community population register ensured that the parents of all children born in the area during the period under study could be approached.

When the children were 2 months old (median age), their parents answered a questionnaire on housing characteristics, various indoor environmental exposures, and atopic diseases affecting the child’s parents. The questions have been used in earlier studies.^15,16 When the children were 1 and 2 years old the parents answered symptom questionnaires. At this time point children with recurrent wheezing according to our definition and two healthy controls, matched by day of birth, were identified from the cohort. The response rate was high: 93% answered all three questionnaires. Our definition of recurrent wheezing stipulated three or more reported episodes of wheezing not associated with common colds after 3 months of age, combined with either symptoms of bronchial hyperreactivity (cough during sleep, play and laughter, etc) or inhaled steroid treatment. Furthermore, to be included in the case-control study, both cases and controls had to reside in the same dwelling as when they were born.

In total, 321 children with recurrent wheezing according to the definition were identified, but only 181 of these, 65 girls and 116 boys, still lived in their first home at the time of recruitment to the case-control study. Controls were matched to cases on day of birth and the control group included 359 children (189 girls and 170 boys). Two hundred and ninety four of the 540 children were recruited at the age of 1 year, including 98 cases and 196 controls. Three controls, selected at 1 year of age fulfilled the criteria of recurrent wheezing at the age of 2, and were also included as cases at that age.

Residential areas were classified into three different groups, mainly based on geographical location and building type. Areas classified as “urban” had narrow streets and mainly old block buildings with enclosed courtyards, “semi-urban” had blocks of flats or terraced houses, and “suburban areas” had predominantly single family houses (bungalows, detached, semi-detached, etc). According to this classification, 129 children (23.9%) lived in “urban areas”, 274 (50.7%) in “semi-urban areas”, and 137 (25.4%) in “suburban areas”. The distance between the city area and the most distant suburban area is about 25 km.

Environmental measurements

In the first winter season following recruitment in the case-control study, the dwelling of each child was inspected by an environmental health officer and data on building construction, ventilation system, water damage, etc were collected using a standardised form. Furthermore, parents answered questions about renovations of the dwelling and about lifestyle factors, such as smoking, household pets, etc.

During a four week period (mean 28 days, range 26–31), NO₂ was measured indoors, in the main living room at about 1.7–1.8 m above the floor, and outdoors, outside the window of this room, using a passive method (Palmes tube).^17,18 To assess the precision, double measurements were performed at random in some of the 540 dwellings (132 (24.4%) of dwellings for indoor NO₂ and 121 (22.4%) for outdoor NO₂). The intra-class correlation coefficient for both indoor and outdoor NO₂ was 0.99.¹⁹ Furthermore, a comparison was made with NO₂ levels recorded by central urban monitors using the chemiluminescent nitrogen oxide analyser, model AC 31 M (Environnement SA, France) and DOAS (differential optical absorption spectroscopy), model AR500 (OPSIS AB, Sweden).^20,21

Indoor air exchange rate per hour (ACH) was measured simultaneously in all dwellings during the same period as the NO₂ measurements, using a passive tracer gas technique.²² The method is described in detail elsewhere.^23,24

Statistical analysis

Statistical analysis was carried out with STATA software.²⁵ A conditional logistic multivariate regression model was used to control for potential confounding factors. Crude and adjusted odds ratios (OR) and 95% confidence intervals (CI) were estimated. Cases and controls were matched for day of birth. Analyses were adjusted for gender and heredity (no/yes), maternal age (median, <26/⩾26 years), maternal smoking (⩾1 cigarette/day during pregnancy and/or when answering the questionnaire, at median age 2 months of their child), for any breast feeding (<6 months/⩾6 months), and building age (1939 or earlier, 1940–75, 1976 or later). Heredity was defined as doctor diagnosed asthma and/or rhinitis together with allergy to furred pets or to pollen, in one or both parents. When stratifying for gender the matched data sets were dissolved and unconditional logistic regression was used adjusted as above. In a separate regression analysis an interaction term between NO₂ and environmental tobacco smoke (ETS) was calculated. ETS was defined from the dichotomous question “do you have any smoker in the family”, asked by the inspector when investigating the home. A p value of <0.05 for the interaction coefficient was considered statistically significant, indicating different effects of NO₂ exposure in children exposed and unexposed to ETS. The Spearman rank correlation coefficient was calculated for comparison of measured outdoor and indoor levels of NO₂, and for the correlation between indoor NO₂ levels and ventilation rate, presented as air change rate per hour (ACH).

Measurements were originally performed for all children in the case-control study. Due to technical failure, some results of NO₂ measurements were lost (7.2% (40/540) of the indoor NO₂ measurements and 7.7% (42/540) of the outdoor NO₂ measurements). In order to reduce the loss of efficiency we used a multiple imputation method, in which we first generated five copies of the original data set; each of those with missing values were replaced by values randomly generated by the imputation model.²⁶ After performing identical conditional logistic regressions on each of five data sets, the results were combined to produce overall estimates and standard errors. Multiple imputation was performed using NORM.²⁷

To compensate for the fact that NO₂ was measured at each site only once (albeit over a four week period), and thus was influenced by weather conditions during that time period, an adjustment factor was calculated to estimate the average outdoor wintertime exposure. Using daily measurements from three central urban background monitors, the adjustment factor was calculated as the difference between the average NO₂ level of each measurement period, and the average NO₂ level for the entire period. The adjustment factor was then subtracted from each measured value, resulting in a value adjusted for time period.

RESULTS

Table 1 presents the variables used in the regression model, including gender, allergic heredity, maternal age and smoking, duration of breast feeding, and building age. The association between building related exposure such as building age and construction will be further explored in a separate paper.

The mean NO₂ level for all outdoor measurements was 21.8 µg/m³ (SD 8.5). Levels were highest in urban areas and decreased in areas less exposed to traffic; this was true both for outdoor and indoor levels (table 2). Mean urban background NO₂ during the four winter seasons, measured with three monitors at roof level, was 26.0 µg/m³ (SD 9.9), whereas corresponding mean outdoor NO₂ levels for home measurements within the same area, measured at different levels from the ground, was 31.5 µg/m³ (SD 6.0). Gas stoves were mainly present in buildings erected before 1940, which predominated in the urban area. Within this area, mean levels of NO₂ in homes where gas stoves were used was 22.6 µg/m³, compared with 16.4 µg/m³ in homes without gas. In homes where gas was not used and where no family member smoked, a correlation was seen between measured indoor and outdoor levels of NO₂ (r = 0.69, p < 0.001). For dwellings equipped with a gas stove there was no such correlation (r = 0.13, p = 0.43, smokers included; and r = 0.06, p = 0.75, when smokers were excluded). A weak positive correlation was also found for air change rate (ACH) and indoor NO₂ levels in dwellings without a gas stove or smoking family members (r = 0.34, p < 0.001), whereas no correlation could be shown for dwellings with gas stoves (r = −0.04, p = 0.77).

For the whole group of children, no clear association was seen between exposure to NO₂ and recurrent wheezing during the first two years. However, there was a tendency towards an increased risk for the highest quartile of exposure, both for indoor and outdoor measurements (OR 1.51, 95% CI 0.81 to 2.82; and OR 1.60, 95% CI 0.78 to 3.26, respectively; table 3). Analyses performed with the imputed data set did not change odds ratios essentially, compared with results based on the measured data sets, but the confidence intervals became narrower (data not presented). Nor did analyses using data based on calculated annual average mean exposure change the OR markedly.

Stratification for age when children were included in the case-control study indicated a dose-response trend, particularly for children who fulfilled the criterion of recurrent wheezing at 2 years of age (table 4). Again, the findings were consistent after adjustment for annual average NO₂ levels. No consistent indications of gender related effects were found (not in table).

Conditional logistic regression analysis using continuous outdoor NO₂ levels resulted in an odds ratio of 1.20 (95% CI 0.90 to 1.60) for an increase of 10 µg/m³ of NO₂. Corresponding analysis of the association between indoor NO₂ levels and recurrent wheezing gave an odds ratio of 1.06 (95% CI 0.74 to 1.52) for an increase of 10 µg/m³ of NO₂. For children taken ill after the age of 1, outdoor OR was 1.73 (95% CI 1.05 to 2.82) and indoor OR 0.92 (95% CI 0.54 to 1.60), respectively, for an increase of 10 µg/m³ of NO₂.

An interaction between indoor NO₂ exposure and ETS, defined as “any smoking family member” was found in relation to recurrent wheezing (p = 0.04). When dividing NO₂ exposure on the 4th quartile, OR for exposure to both NO₂ and ETS was 3.10 (95% CI 1.32 to 7.30), compared with children unexposed to ETS and in the lower quartile of NO₂. OR for children exposed to higher NO₂ levels only was 1.09 (95% CI 0.62 to 1.91) and OR for exposure to ETS only was 0.96 (95% CI 0.51 to 1.83) (tables 5 and 6). The corresponding interaction for outdoor NO₂ levels did not reach statistical significance (p = 0.15), but the OR for exposure to both outdoor NO₂ and ETS was 3.60 (95% CI 1.37 to 9.45), compared with children in the lowest exposure category for both agents. No marked differences in risk estimates were obtained when no adjustment was made for mothers smoking during pregnancy.

Gas was used for cooking only. When the children were newborn 49 dwellings were equipped with a gas stove, but at the time of investigation gas was used in 46 of 540 homes (8.52%). Only four of the homes with a gas stove were equipped with separate stove ventilation connected to a mechanical exhaust ventilation system, and another 22 kitchens had carbon filter ventilators. The adjusted OR for recurrent wheezing associated with the use of a gas stove was 1.72 (95% CI 0.77 to 3.87), without any differences in gender. No difference appeared when analysing presence of a gas stove in the home when the children were newborn, compared with use of a gas stove at the time of the investigation of the home.

DISCUSSION

This study focuses on the association between exposure to air pollution, including NO₂, and recurrent wheezing in children up to 2 years of age. Effects of outdoor and indoor exposure were evaluated and a relation between exposure to NO₂ and recurrent wheezing is suggested. When the material was stratified for age, the association between wheezing and NO₂ exposure appeared stronger among children who did not fulfil the criteria until the age of 2. This might be explained by reduced misclassification of disease in the older children, or it may be that prolonged exposure time is required for ambient air pollution to induce recurrent wheezing at those exposure levels. It should be noted that NO₂ is used as an indicator of air pollution exposure and other components may be of importance, such as airborne particulate matter and ozone. However, it is also worth noting that the NO₂ levels measured in our study lie well below the current European guidelines for annual exposure (40 µg/m³).²⁸

In contrast, a Norwegian study on bronchial obstruction in children, as well as a study of childhood asthma in four regions of Scandinavia (PEACE project), by Forsberg et al did not show any consistent pattern in the relation between air pollution, including NO₂, and asthma.^4,29 However, a newly presented international collaborative study on the impact of traffic related air pollution and asthma (TRAPCA), in turn, indicates an association between traffic related air pollution and cough. In contrast to our results, the associations between symptoms and NO₂ exposure were found to be weaker for the second year of life.³⁰ The reasons for the apparent inconsistency between different study results remain unclear.

The suggested increase in risk of recurrent wheezing was mainly confined to the children most strongly exposed to NO₂ and who in addition had at least one smoking family member. A significant interaction could be shown for children exposed to the upper quartile of indoor NO₂ levels in combination with any smoking family member. A similar interaction was also seen between outdoor levels of NO₂. We are not aware of any other studies that provide information on the interaction between NO₂ and ETS on recurrent wheeze or asthma. Our results also suggest that a rather low degree of exposure to air pollution, including NO₂, may be harmful in combination with other environmental factors.

Nitrogen dioxide was measured during a four week period, approximately the same period for the sets of matched cases and their respective controls. The limited measuring period could result in substantial misclassification of long term exposure. However, analysis of the estimated annual average exposure levels confirmed the results obtained with the measured data. We measured NO₂ at fixed sites, in and outside the dwellings, respectively. Because of this, imprecise assessment of child exposure could have occurred due to the position of the child’s room in relation to the street and the living room, where NO₂ had been measured. Further, some studies indicate that exposure estimated by personal sampling is better correlated to indoor NO₂ levels than to outdoor levels.^31,32 It may thus be expected that exposure estimates based on outdoor measurements will contribute to a dilution of any relation between NO₂ exposure and recurrent wheezing, especially for children living in homes equipped with a gas stove.^33,34

Children exposed to combustion products from gas appliances have in some surveys shown an increased risk of respiratory symptoms.^7–9 However, we found no clear association between the use of gas stoves and recurrent wheezing. Our risk estimate for use of gas stoves was of the same order of magnitude as for those exposed to increased levels of outdoor and indoor NO₂, but the number of gas stoves was low. Therefore, the present results on gas stoves and recurrent wheezing should be interpreted carefully. Nor did we find any clear and consistent differences related to gender, although others have shown such findings.^7,10,35

One notable finding in our study was the confounding effect of the age of the children’s homes on the risk estimates of NO₂ exposure. When we included the age of the building in the regression model, an increased risk of NO₂ exposure was suggested, which was not obvious without this adjustment. Most children exposed to increased levels of NO₂ lived in old brick built apartment blocks, erected before 1940. Children less exposed to NO₂ lived more often in buildings erected after 1940 with other design and construction, including single family houses. The impact of building age has also been discussed by Krämer et al, who found an increased OR for the association between NO₂ and some atopic diseases, when building age was included in the analysis.³⁶ The findings indicate that health effects associated with building age, or more likely building related factors, should be taken into consideration when evaluating the association between NO₂ exposure and health.³⁷ This may be important, especially in areas with rather low NO₂ levels, and in regions with a temperate climate and high demand on energy saving measures with a consequent effect on the indoor environment.

Studies reporting data on air change rate when measuring indoor NO₂ exposure seem to be lacking. In our study, data on ventilation were available. The lack of association between ACH and indoor NO₂ levels in dwellings with a gas stove may have several explanations. We did not register the use of the gas stove in terms of “time in use”, or evaluate the effects of the condition of the stove. For homes without a gas stove, the positive correlation between ACH and measured indoor NO₂ levels is probably explained by outdoor levels of NO₂; the better the ventilation, the more outdoor combustion products come into the dwelling. This will have an impact, especially in more polluted areas. Similar results have been reported by Spengler et al, who measured increased levels of NO₂ inside homes without gas appliances but with a kitchen fan, compared with similar homes without a kitchen fan.³³

Conclusion

In the present study, outdoor sources constitute major determinants of indoor NO₂ levels in dwellings without a gas stove. Although chance cannot be discounted, an increase in recurrent wheezing is suggested for those children in the study most highly exposed to NO₂, even though the levels were lower than the current European recommended annual level of 40 µg/m³. The combination of exposure to ambient air pollution and ETS appears important for development of recurrent wheezing in children.