Published ahead of print on May 16, 2007, doi:10.1164/rccm.200605-589OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200605-589OC
Indoor Air Quality in Homes of Patients with Chronic Obstructive Pulmonary Disease1 Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, Scotland, United Kingdom; 2 Chest Clinic, Aberdeen Royal Infirmary, NHS Grampian, Aberdeen, Scotland, United Kingdom; 3 Aberdeen City Council, Aberdeen, Scotland, United Kingdom; 4 Institute of Occupational Medicine, Edinburgh, Scotland, United Kingdom; and 5 Department of Environmental and Occupational Medicine, University of Aberdeen, Aberdeen, Scotland, United Kingdom Correspondence and requests for reprints should be addressed to Dr. Liesl M. Osman, Ph.D., Senior Research Fellow, Chest Clinic, Aberdeen Royal Infirmary, Foresterhill, Aberdeen, AB25 2ZN Scotland, UK. E-mail: med078{at}abdn.ac.uk
Rationale: Outdoor air quality is associated with respiratory morbidity and mortality. Less is known of the relationship of indoor air quality to respiratory health of groups vulnerable to outdoor air, such as those with chronic obstructive pulmonary disease (COPD). Objectives: To investigate among patients with COPD the association of health status with indoor air quality in their homes. Methods: Observational study of indoor environmental characteristics of homes of 148 patients with severe COPD in North East Scotland. Measurements and Main Results: Airborne living room levels of particulate matter with a diameter of 2.5 µm or less (PM2.5) (µg/m3) were measured over 8 to 14 hours using DustTrak monitors. Nitrogen dioxide exposure (ppb) in living rooms was measured over 1 week. Endotoxin (EU [endotoxin units]/mg) in living room dust was measured. Health status of participants was assessed by the St. George's Respiratory Health Questionnaire (symptoms, activity limitation, and disease impact). The mean age of participants was 69 years. Approximately 45% were male, 39% were smokers, and 49% lived in smoking households. Average indoor PM2.5 levels were 18 µg/m3, nitrogen dioxide was 7.8 ppb, and endotoxin levels were 95.8 EU/mg of dust. PM2.5 was significantly higher in smoking households (P < 0.001) and was associated with higher levels of endotoxin and NO2. PM2.5 was significantly associated with increased symptom burden (P < 0.01), with greater effect for current smokers. Endotoxin and nitrogen dioxide exposure were not related to health status. Conclusions: Higher levels of PM2.5 are associated with worse health status of these patients with severe COPD. Indoor levels of PM2.5 are significantly higher in homes with smokers.
Key Words: respiratory disease • particulates • morbidity • environmental tobacco smoke
Chronic obstructive pulmonary disease (COPD) is the fifth leading cause of death worldwide (1). The effect of outdoor particulate matter with a diameter of 10 µm or less (PM10) on respiratory mortality and hospital admission is well documented (2, 3), and there is increasing evidence that the smaller size fraction of particles (PM2.5) may be the important fraction. Studies (4, 5) from the United States have shown an increase in COPD mortality for every 10-µg/m3 increase in PM2.5 outdoor levels, over a range of 13 to 23 µg/m3. There is also evidence of a relationship between inflammatory markers in patients with COPD and outdoor PM2.5 levels (6). Although exposure to outdoor pollution is important, most people spend the greater part of their time indoors. Indoor concentrations of 15 µg/m3 PM2.5 in households of nonsmoking patients with COPD in Holland were similar to those in Boston, Massachusetts (7), at 17.2 µg/m3. However, homes without smokers are probably not typical of most COPD patient households where exposure to environmental tobacco smoke (ETS) can be high. There are no guidelines for indoor recommended levels of PM2.5, but, in 2006, the U.S. Environmental Protection Agency (EPA) proposed recommended maximum outdoor air quality levels of PM2.5 of 35 µg/m3 over 24 hours, and a daily average level of 15 µg/m3 annually. Recent World Health Organization air quality guidelines (8) recommend even lower maximum outdoor air quality levels of PM2.5 of 25 µg/m3 over 24 hours and an annual mean of 10 µg/m3. Other indoor factors that influence respiratory health include the oxidant gas nitrogen dioxide (NO2) and endotoxin. Exposure to NO2 indoors is associated with markers of respiratory health in children (9), although evidence of respiratory effects on adults is conflicting (10, 11). Endotoxin is a major proinflammatory agent and is significantly associated with symptoms and reduced lung function among asthmatic populations (12–14). ETS is a contributor to both PM2.5 (15, 16) and airborne endotoxin levels (17, 18), and it has been suggested (9) that NO2 and ETS interact to increase the risk of respiratory symptoms among children. Among patients with COPD, exposure to ETS is associated with respiratory symptoms, decreased lung function, and impaired health status (19, 20), and indoor biomass exposure has been shown to be related to development of COPD independent of exposure to ETS (20). The present study investigated the health status of patients with COPD in relation to indoor air quality in these patients' homes as assessed by measurement of PM2.5, nitrogen dioxide, and settled dust endotoxin levels. The study was carried out in the North East of Scotland and is part of a larger study evaluating the impact of home improvements on respiratory health of these patients. Some of the results described in this article have been presented in abstract form (21, 22).
This was the initial cross-sectional component of the HEARTH (Home Environment and Respiratory Health) Study, an intervention study that included both intervention and nonintervention arms. The data presented here are for all participants regardless of their future randomization status. The results from the intervention study will be published at a later stage.
Time Period
Sample
Indoor Air Quality Monitoring Indoor PM2.5 (µg/m3) was measured using a DustTrak (TSI, Inc., Shoreview, MN) light scattering monitor. Readings were taken every 5 minutes. The DustTrak was placed in participants' living rooms, usually 1 to 1.5 m high, between 12 P.M. and 5 P.M. and collected the following day between 9 A.M. and 12 P.M. Battery power allowed monitoring for up to 18 hours; collection times were organized to allow a minimum of 12 hours of monitoring. DustTrak readings were corrected by a factor of three. This has been found (23, 24) to be the calibration factor for light scattering monitors when measured against gravimetric samplers. Two DustTrak machines were used. Intermachine reliability was tested in four homes and ranged from 0.92 to 0.99, with a mean of 0.96. Outdoor PM2.5 (µg/m3) readings were collected from a DustTrak monitor at the Department of Environmental and Occupational Medicine, which is based approximately 3 miles from the Aberdeen City center, in a medium density suburban area. Readings were taken every 5 minutes. Averages were calculated from daily averages. These readings were also adjusted by a factor of 3. Endotoxin dust samples were collected using a Morphy Richards 2,000-W vacuum cleaner with HEPA filtration (Morphy Richards, Ltd., Mexborough, Yorkshire, UK), following the methods used in the U.S. Department of Housing National Survey of Lead and Allergens in Housing (23). An area in front of the living room sofa measuring 2 x 1 m was vacuumed for 5 minutes. Dust samples were sealed in plastic bags, refrigerated at temperatures lower than 5°C, and sent within 2 days to the Institute of Occupational Medicine in Edinburgh for batch analysis. The detection limit for this assay is 0.0625 endotoxin units (EU)/ml. Indoor NO2 was measured with passive samplers (Palmes tubes) in the living room and bedroom. Sample tubes were placed away from windows and doors, usually at 1 to 1.5 m height. Samples were left for 1 week. On collection, diffusion tubes were capped and sealed in plastic bags; date and time of collection were recorded, and samples were refrigerated at temperatures lower than 5°C. Concentrations are expressed as parts per billion (ppb). Salivary cotinine samples were collected at the first home visit. Samples were refrigerated and then assayed using Salimetrics High Sensitivity Salivary Cotinine Enzyme Immunoassay kit (Salimetrics Inc., State College, PA).
Patient Data
SGRQ Smoking status was identified through responses to the questions, which included the following: "Have you smoked more than 100 cigarettes in total in your life?" ("Yes" or "No") and "Are you a current smoker?" Self-reported smoking status was verified through cotinine assay. Respondents who reported that they were not current smokers, but had a salivary cotinine level above 20 µg/L, were reclassified as current smokers (28). All patients had been reviewed by a hospital physician within the past 2 years, and had a recorded diagnosis of COPD. Lung function measurements (FEV1 and FVC) of subjects at time of review were collected from clinical records as were details of hospital admissions for COPD exacerbations in the 12 months before a patient entered the study. Social deprivation was assessed by the Carstairs deprivation index (29). This is derived from postal codes and is calculated from four census variables: car ownership, household overcrowding, head of household in social class IV or V (partly-skilled or unskilled occupations, respectively), and male unemployment. The index is a standardized score, with 0 as the national mean score and a standard deviation of 3.5. A positive score indicates greater disadvantage than average. The Carstairs deprivation index is significantly related to differences in prevalence of chronic health conditions (30) and is an index standardly used to examine social deprivation in British health evaluation studies.
Ethical Approval
Analysis Initial analysis showed that distributions of SGRQ scores were acceptable for parametric analysis of the subscales for impact and symptoms, but activities scores were more skewed. Analysis of residual versus predicted values for the models for symptoms and activities showed a tendency to a funnel distribution (i.e., predictions had greater error for low values of the independent variables). This was not surprising, because symptoms and activities scores were positively skewed and there were few observations with high exposures to the environmental measures. Two alternative approaches were examined to address this problem in the regression models: squaring the skewed SGRQ scores and taking logarithms of the exposure variables. The latter approach was more effective in ensuring that the assumptions of linear regression were satisfied, and so results from this approach are presented.
Participants Five hundred and thirty-four patients meeting the study criteria were identified from hospital records and were invited to take part in an initial housing survey. Two hundred and fifty-four agreed. Of these, 46 were not surveyed for a variety of reasons, including difficulty in contacting to arrange survey, withdrawing from the study, or death. Two patients were found not to have COPD on review of case notes, leaving 206 homes. Mean Carstairs deprivation score of those agreeing to be surveyed was close to the national average and did not differ from those not wishing to participate (P = 0.16). Participants who met the criteria for the main intervention study were invited to take part in a 1-week monitoring of air quality; 148 (72%) participants who did not differ in age, sex, or deprivation score from the original 206 participants agreed to take part. One hundred eight participants reported that they were not current smokers. Eighteen of these had salivary cotinine levels above 20 µg/L (27–420 µg/L) and were reclassified as smokers. Mean percent-predicted FEV1 of participants was 43%, with an SD of 18.6%. Using GOLD (Global Initiative for Chronic Obstructive Lung Disease) criteria, 91 (61%) were classified as moderate, 39 (26%) as severe, and 18 (12%) as having mild COPD. Demographic and clinical characteristics of participants are shown in Table 1. Table 2 shows environmental measures and their interrelationships. Table 3 shows environmental measures in relation to household smoking status.
PM2.5 in Homes of Patients with COPD High levels of PM2.5 were recorded in the homes of patients with COPD. Table 2 shows that maximum levels were more than four times the EPA recommendation for maximum levels over 24 hours. Figure 1 and Table 3 show that these high levels were associated with the presence of smokers in the home.
A significant cause of PM2.5 in nonsmoking households appeared to be smoking by visitors. Figure 2 shows a typical pattern of PM2.5 in a nonsmoking household. This home had an average PM2.5 of 16 µg/m3 over the 12 hours measured, but the average disguises the peak value of 446 µg/m3 between 8 and 9.30 P.M. in the evening, which we believe to be due to a smoking visitor.
Outdoor PM2.5 Over the study period, monthly median 24-hour outdoor values of PM2.5 ranged from 4 to 17 µg/m3.
Models of Health Status and Environmental Exposures
Figure 3 models the change (worsening) in SGRQ associated with increase in exposure levels of PM2.5, for the models derived from average and maximum PM2.5 levels. It shows that the difference in median exposure levels between smokers and nonsmokers is associated with a 6-point difference in SGRQ symptom score.
When the analyses were stratified by smoking status, the results were in the same direction as in the total model, but the size of the coefficients indicated that the effect of PM2.5 on symptoms was increased for smokers, and reduced for nonsmokers. Therefore, the changes in average or maximum PM2.5 necessary for a predicted 4-point decrease in SGRQ symptom scores were smaller in smokers than in nonsmokers.
SGRQ activity limitation scores
SGRQ disease impact scores
This study found high levels of airborne PM2.5 in the homes of patients with COPD. The highest levels of PM2.5 were, on average, four times the maximum recommended by the EPA for 24-hour periods. Levels of PM2.5 were strongly associated with the presence of smokers in the household. Higher PM2.5 levels were related to poorer health status of smoking and nonsmoking participants. NO2 levels were not associated significantly with health status in this study. Levels in bedrooms and living rooms were lower than those found by Jarvis and colleagues (11), which were similarly measured over 1 week (6–7 ppb in our study compared with 13 ppb in Jarvis and colleagues' study). Jarvis and colleagues also found no relationship between NO2 and respiratory health status, confirming that NO2 exposures at this low level are unlikely to cause health effects. Although Morrow and coworkers (31) found that exposure of 300 ppb of NO2 over 4 hours with intermittent exercise was associated with decreased FEV1 at P < 0.10, Gong and colleagues (32) found no significant response to 400 ppb of NO2 over 2 hours with intermittent exercise in elderly volunteers with COPD. Settled dust endotoxin levels were high and were associated with airborne levels of PM2.5. The median level of 96 EU/mg was similar to Blanc and colleagues' (12) finding of 42 EU/mg and Topp and coworkers' (33) finding of levels of 20 to 30 EU/mg in repeated measures over 3 years, and also is similar to levels found in the National Allergy study (14) of 63 EU/mg for living room endotoxin. In unadjusted analysis, endotoxin had a borderline association with health status (SGRQ impact), but this association disappeared when a more robust logarithmically transformed model was used. The present study suggests that the main source of PM2.5 in the homes studied was ETS. Other studies in respiratory patients (19, 36) reported that exposure to cigarette smoke was associated with poorer health status and asthma-specific quality of life and with increased risk of hospital admissions and emergency visits. In an elderly Italian population, Simoni and coworkers (19) also found that exposure to ETS was a significant risk factor for respiratory symptoms. A difference of 4 points in SGRQ scores is regarded as clinically significant (27), and Figure 3 shows that a change from the median PM2.5 exposure in smoking homes (71 µg/m3) to median exposure in nonsmoking homes (8 µg/m3) is associated with a 6-point improvement (reduction) in SGRQ symptom scores. The stratified analysis showed that exposure effects were consistent between smokers and nonsmokers, but that the effect on symptom score of an increase in PM2.5 was greater for smokers than for nonsmokers. Previous studies (7, 11, 33–35) of indoor air quality have tended to omit smokers, but the present study shows that excluding households with smokers is likely to underestimate the impact of indoor air quality on health in patients with COPD. Almost half the participants in the present study lived in "smoking environments." Approximately 31% of smokers and 17% of nonsmokers lived in households in which another member smoked. In addition, individual monitoring results showed that, even in nonsmoking households, short-lived high values of PM2.5 were recorded, probably associated with visits to the home by smokers. Health status did not differ between smoking and nonsmoking participants, but the PM2.5 effect was greater among continuing smokers. Smokers were younger than nonsmokers, and hence should have had some advantage in health status, but smokers were likely to live with other smokers, leading to very high levels of PM2.5 in their homes. Therefore, the apparently greater effect among smokers may be an artifact of the higher levels of PM2.5 in their homes. Figure 3 shows that the effect of PM2.5 on symptoms is not linear. It increases sharply between levels typical of smoking and nonsmoking households, and then flattens out at the high levels typical of smoking households. Finally, ETS has differences compared with mainstream smoke. ETS is composed of approximately 85% sidestream smoke and 15% mainstream smoke. Sidestream smoke is more toxic than mainstream smoke (39). Mass for mass, ETS is four times more toxic than mainstream smoke. We found a weak relationship between settled dust endotoxin in the living room and symptoms in our population, which did not remain when a logarithmically transformed model was used. There was a moderate association between PM2.5 and endotoxin levels. Among adults with asthma, levels of settled dust endotoxin are related to FEV1 (12) and to asthma symptoms and wheezing (14). Thorne and colleagues, in the National Allergy study (14), found no relationship of allergy status to endotoxin load and concluded that the significant effect of endotoxin exposure in asthma is on airway inflammation, independent of atopy. Airborne endotoxin exposure is known to be a risk factor for causing COPD (36, 37), but Park and colleagues (38) found that there was only a weak correlation (r = 0.3, P < 0.05) between floor (dust) endotoxin and airborne endotoxin. To our knowledge, there are no studies examining health status in patients with COPD in relation to domestic airborne endotoxin. Our results suggest that it would be of interest to explore the relationship between health status of patients with COPD in smoking households and airborne endotoxin in these homes. Our study has advantages and limitations. It was a representative sample of patients with severe COPD living in their own homes. Exposures to environmental agents involved direct measurement of PM2.5, NO2, and endotoxin, which allowed comparison of objective measurements and health status. The salivary cotinine assay validated the self-report of household smoking. A further advantage is that the specific a priori aim of the study was to explore the relationship of PM2.5, endotoxin, and NO2, and hence the multivariate analyses were not derived by post hoc significance testing. A limitation of the study was that health status was related to indoor environment measurement taken at one point in time. Air quality will vary in these patients' homes to a greater extent than is identified by 1-day PM2.5 monitoring, or one sample of endotoxin. For instance Topp and colleagues (33) found correlations of settled dust endotoxin levels at different time points over a 6-year period to range from 0.35 to 0.51, and concluded that single-point measurement does not represent long-term exposure. PM2.5 levels will vary depending on who is living in or visiting the patient's home. However, this variability will tend to weaken associations with health status measured at one point in time, and, in the present study, the association between health status outcomes and PM2.5 was strong. Models derived from average and maximum PM2.5 levels returned similar coefficients, suggesting a consistent effect size of PM2.5. We did not assess ventilation behavior, but outdoor pollution would have contributed to indoor levels of PM2.5. Average monthly outdoor PM2.5 levels over the study period varied between 4 and 17 µg/m3, compatible with the average levels found in the nonsmoking households in the study, but well below averages found in smoking households. This suggests that the main contributor to high indoor levels of PM2.5 was personal behavior within the home, particularly cigarette smoking. Other potential sources include disturbed floor dust and fugitive fumes from an adjacent kitchen. In the calibration studies for PM2.5 estimates from light scattering monitors, measurement error is small, and the magnitude of the relative error remains the same over the whole range of PM2.5 values recorded (39, 40). This supports the validity of the calibrated measurement. However, we must be cautious before making quantitative estimates of effects of PM2.5 on health status, or deriving any recommendations for minimum acceptable levels of PM2.5 indoors or outdoors. The present study is cross-sectional and cannot show whether the poorer health status associated with poor indoor air quality will translate to increased health service use. As part of a larger study, outcomes in the group described in this article will be followed over 2 years. Nonetheless, the present results show that a health status burden is associated with indoor particulate levels, and strongly suggest that ETS is a major contributor to high indoor levels of PM2.5. The results of the present study suggest an important effect of indoor PM2.5 on health of patients with COPD by showing an association between health status and particulate exposure. Further investigation is needed to test whether the relationships found in these baseline results are reliable over time and whether exposures are associated with increased frequency of exacerbations.
The authors thank the patients who took part in this study. They also thank Gordon Kyle, Chief Executive of Castlehill Housing Association, Karen Milne, Project Manager, and Mrs. Val Bennett, Energy Surveyor, Aberdeen Care & Repair, Castlehill Housing Association. They thank Dr. Gordon Prescott, Senior Lecturer in Medical Statistics of the Department of Public Health, University of Aberdeen, for statistical advice and the development of logarithmically transformed models of exposure.
Supported by the Eaga Partnership Charitable Trust. This article has an online supplement, which is accessible from this issue's table of contents at www.thoracic.org Originally Published in Press as DOI: 10.1164/rccm.200605-589OC on May 16, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form May 2, 2006; accepted in final form May 15, 2007
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
5 µg/m3), where these models predicted higher (worse) symptom scores. The coefficients of 5.3 and 6.0 indicate that a 1-unit reduction in log PM2.5 or, equivalently, a 10-fold decrease in PM2.5 would reflect a significant decrease of 5.3 or 6.0 SGRQ points, respectively (P = 0.001) (Tables 
PM2.5 average in nonsmoking household was 8 µg/m3;