The Health Impact of Indoor Air Filtration in Healthy Adults and Asthmatic Children

Thumbnail Image




Cui, Xiaoxing

Journal Title

Journal ISSN

Volume Title

Repository Usage Stats



The World Health Organization (WHO) estimated that 92% of the world’s population lived in areas with outdoor air pollution levels exceeding the WHO guidelines in 2014. Although the ultimate solution is to control emission sources, exposure reduction strategies at the individual level can address more immediate needs. As people typically spend approximately 80% of their time indoors, improvement of indoor air quality, such as using air filtration technologies, may lead to reduced total exposure. However, evidence is limited to support the effectiveness of air filtration in bringing beneficial health effects to the users. The goal of this dissertation research is to evaluate the health impact of indoor air filtration in healthy adults and asthmatic children using two randomized, double-blind, crossover trials (Aim 1 and Aim 2). Utilizing data from these two trials and other data, this dissertation also addresses two methodological questions. Aim 3 examines the relationship between free and total malondialdehyde, a biomarker of oxidative stress commonly used in air pollution research, in various types of human specimens. Aim 4 evaluates the relationship between the subjective evaluation of asthma control by childhood asthma control test (C-ACT) and objective indicators of lung pathophysiology using longitudinal measurements.

In Aim 1 of this dissertation, a double-blinded, randomized, crossover study was conducted to examine the cardiorespiratory health effect of a sporadic, overnight use of indoor air filtration. Seventy healthy non-smoking adults were recruited from the medical and nursing students who were living in the same dormitory building in a suburb of Shanghai, China. The participants were aged 19 to 26 years old and included 41 (59%) females. Each participant received a true and a sham indoor air filtration session in a randomized sequence. Participants and research staff that performed health assessments were blinded to this sequence until the end of the study. Each filtration session was approximately 13 hours long. True and sham filtration sessions were separated by a two-week “washout” interval. During the study period, outdoor PM2.5 concentrations ranged from 18.6 to 106.9 µg/m3, which overlapped with levels measured in Western Europe and North America. Compared to sham filtration, true filtration on average decreased indoor PM2.5 concentration by 72.4% to 10.0 µg/m3 and particle number concentration by 59.2% to 2316/cm3. For lung function measured immediately after the end of filtration, true filtration significantly lowered airway impedance at 5 Hz (Z5) by 7.1% [95% CI: 2.4%, 11.9%], airway resistance at 5 Hz (R5) by 7.4% [95% CI: 2.4%, 12.5%], and small airway resistance (R5-R20) by 20.3% [95% CI: 0.1%, 40.5%], reflecting improved airway mechanics especially for the small airways. However, no significant improvements for spirometric indicators of lung function (FEV1, FVC) were observed. True filtration also significantly lowered von Willebrand factor (VWF) by 26.9% [95% CI: 7.3%, 46.4%] 24 hours after the end of filtration, indicating reduced risk for thrombosis. Stratified analysis in male and female participants showed that true filtration significantly decreased pulse pressure by 3.3% [95% CI: 0.8%, 7.4%] in females, and significantly reduced VWF by 42.4% [95% CI: 17.4%, 67.4%] and interleukin-6 by 22.6% [95% CI: 0.4%, 44.9%] in males. Effect modification analyses indicated that filtration effects in male and female participants were not significantly different. These findings suggest that a single overnight residential air filtration, capable of reducing indoor particle concentrations substantially, can lead to improved airway mechanics and reduced thrombosis risk.

Air pollution exposure is a well-established risk factor for asthma exacerbation. In individuals with asthma, indoor air filtration has only been evaluated for allergen removal in areas with low outdoor levels of PM2.5. As asthmatic individuals may be responsive to short-term changes in air pollution levels, it is not clear whether the exposures occurring outside the home environment can override the potential health benefits of residential indoor air filtration. In Aim 2 of this dissertation, I investigated the respiratory impact of residential indoor air filtration in asthmatic children living with moderate levels of outdoor PM2.5. This double-blind, randomized crossover trial recruited 43 participants (40% females) aged 5 to 13 with mild or moderate asthma. From February to April 2017, each participant used a true filtration device and a sham filtration device, respectively in their bedrooms for two weeks. A two-week washout period separated the two sessions. During the study, the average PM2.5 concentrations for outdoor, indoor with sham filtration, and indoor with true filtration were 56.5 µg/m3, 34.1µg/m3, and 14.8 µg/m3, respectively. Compared to sham filtration, on average, true filtration reduced indoor PM2.5 concentrations by 53.3%. Concomitantly, true filtration significantly decreased fractional exhaled oxide (FeNO) by 5.1 [95% CI: 0.5, 9.6] ppb, airway impedance at 5Hz (Z5) by 14.7% [3.7%, 25.8%], respiratory resistance at 5Hz (R5) by 22.4% [9.6%, 35.2%], and small airway resistance (R5-R20) by 40.6% [10.2,% 70.9%]. Peak expiratory flow (PEF) measurements were made twice daily throughout each two-week intervention session. PEF values during true filtration were 1.5% [0.7%, 2.4%] higher than the measurements during sham filtration. True filtration also introduced a non-significant increase on FEF25-75 by 5.80% [-4.09%, 15.68%]. Stratified analyses show that the participants with blood eosinophil count ≤ 500/mm3, compared to those with blood eosinophil count > 500/mm3, had a 24.3% greater increase of FEF25-75 by true filtration; and that the participants allergic to dust mite had smaller improvements in small airway resistance (R5 and R5-R20). These findings suggest that a two-week long intervention of indoor air filtration, capable of significantly reducing indoor PM2.5 concentrations, can lead to reduced respiratory inflammation, improved lung function (PEF) and airway mechanics in asthmatic children.

Oxidative stress is a core mechanism involved in the cardiorespiratory effects of air pollution. Malondialdehyde (MDA) is a well-established marker of oxidative stress. However, most of its use in existing studies were conducted by researchers in clinical medicine and epidemiology, who were unaware that MDA is present in both unconjugated and conjugated forms. Aim 3 of this dissertation examined the relationship between free MDA (unconjugated MDA) and total MDA (the sum of both unconjugated and conjugated MDA) in various types of human biospecimens. Using bio-banked samples from multiple studies, free MDA and total MDA were measured simultaneously in nasal fluid (N=158), saliva (N=158), exhaled breath condensate (N=40), serum (N=232), and urine samples (N=429). MDA quantification was performed using an HPLC-fluorescence method with high sensitivity and specificity. Due to the right-skewed distribution of free MDA and total MDA, natural-log transformation was performed before subsequent statistical analyses. The relationship between the natural log of free and total MDA was evaluated by R2 of simple linear regression. T-test was used for comparisons of means between two groups. One-way analysis of variance was used in combination with Tukey’s test to compare the natural log of the ratio of free MDA to total MDA across various types of biospecimens. For exhaled breath condensate, serum, urine, nasal fluid and saliva samples, the R2 between free and total MDA were 0.61, 0.22, 0.59, 0.47 and 0.06, respectively; the medians of the free MDA to total MDA ratio were 48.1%, 17.4%, 9.8%, 5.1% and 3.0%, respectively. The free MDA to total MDA ratio in EBC > serum > urine > nasal fluid > saliva (P<0.001 for pairwise comparisons). These findings indicate that, for exhaled breath condensate and urine samples, using either free or total MDA can provide information regarding the level of oxidative stress; however, that is not the case for serum, nasal fluid, and saliva, given the low correlations between free and total MDA. For these types of biospecimens, future research is needed to examine which form of MDA better reflects oxidative stress, in a mechanistic fashion. Given these findings, free MDA in urine samples was measured in the Aim 1 study.

The Childhood Asthma Control Test (C-ACT) is a validated questionnaire that provides a subjective evaluation of asthma control using collective inputs from children and their caregivers. Cross-sectional studies have identified cut-points for discerning different degrees of asthma control. However, few studies have examined whether changes in the C-ACT score are reflective of changes in airway pathophysiology in longitudinal measurements from the same individuals. In Aim 4 of this dissertation, I utilized data from four clinical visits originally conducted for Aim 2. The clinical visits were scheduled bi-weekly over a period of six weeks. Thirty-seven children aged 5 to 10 with mild or moderate asthma (43% female) took the C-ACT. The scores were 24.4 ± 2.4 (mean ± standard deviation) and ranged from 16 to 27. Linear mixed-effects models were used to examine the association between C-ACT and clinical measurements from the same participants. As a change of two points in the C-ACT score is considered as the minimal clinically important difference, I reported the change in biomarkers associated with a two-point decrease in C-ACT score. Results showed that a two-point decrease was significantly associated with a 1.7% [95% CI: 0.1%, 3.3%] decrease in FEV1, a 1.6% [0.5%, 2.8%] decrease in FVC, and a 3.8% [0.0%, 7.6%] increase in airway resistance at 5 Hz (R5). For PEF measurements conducted within two weeks before the C-ACT, a two-point decrease in C-ACT score was significantly associated with a 17.3% [95%CI: 6.8%, 27.8%] increase in the coefficient of variation, while no significant association was observed with the average PEF value (P=0.12). In contrast, within-participant measurements showed no significant associations of C-ACT with respiratory inflammation (FeNO) or any of the small airway function and mechanics parameters (FEF25-75, FEF25-75, R5-R20, X5, and Fres). When examining individual questions of the C-ACT, none of the questions answered by the caregiver were significantly associated with indicators of lung function and respiratory inflammation. In contrast, three out of the four questions answered by the child demonstrated significant or marginally significant associations with FEV1, FVC, FEF25-75, R5, and R20. This suggests that in spite of the relatively young age of the participants, their perception of disease control provides important information for asthma control assessment. Overall, the findings in Aim 4 indicate that subjective evaluation of asthma control using the C-ACT was associated with objective measurements of airway obstruction, airway restriction, respiratory resistance and the variation of airflow limitations in longitudinal measurements. However, the C-ACT is not reflective of changes in respiratory inflammation and small airway mechanics, indicating that its use should be complementary to but not a substitute for these measurements.






Cui, Xiaoxing (2018). The Health Impact of Indoor Air Filtration in Healthy Adults and Asthmatic Children. Dissertation, Duke University. Retrieved from


Dukes student scholarship is made available to the public using a Creative Commons Attribution / Non-commercial / No derivative (CC-BY-NC-ND) license.