Pilot study of a community-based PM2.5 educational intervention: impacts on behavior, exposure, and pulmonary function Authors: Hyung Doo Kim, Myung-Sook Park, Woo-Jin Kim, Jiho Lee, Myung Ho Lim, Sumi Chae, Dong-Wook Lee, Hwan-Cheol Kim Published: 2026 ABSTRACT AbstractBackgroundParticulate matter (PM) is a recognized Group 1 carcinogen and is also associated with adverse cardiopulmonary and neurological outcomes. However, community-based strategies to reduce exposure among vulnerable populations remain limited in Korea. This pilot study evaluated the feasibility and effectiveness of a structured, nurse-delivered educational intervention to improve protective behaviors, reduce indoor PM2.5, and enhance pulmonary function among older adults.MethodsA cluster non-randomized controlled trial, involving 59 participants (intervention group, n = 30; control group, n = 29; aged ≥ 65 years), was conducted within the Municipal Home Visit Health Management Program in Incheon, South Korea. The intervention group received a standardized education module based on the Health Belief Model, practical training, and text message alerts, whereas the controls received usual care. Outcomes, including awareness, protective behaviors, indoor air quality (continuous PM monitoring), and spirometry parameters, were measured at baseline and at 6 weeks.ResultsThe intervention group showed significant improvements in predicted FEV1% (84.6 ± 26.2 to 95.8 ± 23.6, p = 0.008) and FEV6% (77.2 ± 23.0 to 88.9 ± 21.2, p = 0.012). Indoor PM2.5 concentrations during the final week of monitoring were lower in the intervention group than in controls (11.1 ± 5.7 µg/m3 vs. 19.6 ± 15.8 µg/m3; p = 0.010), with a significantly lower exceedance rate (1.63 vs. 8.91 events, p = 0.042). Behavioral outcomes improved more in the intervention group, although most differences were not statistically significant.ConclusionsIn this 6-week pilot trial, a brief community-based educational intervention integrated into routine home-visit nursing services was feasible, well-received, and associated with short-term improvements in indoor air quality and selected pulmonary function indices among older adults. These preliminary findings warrant confirmation in larger and longer trials before scalable participatory education programs can be recommended for sustained reduction of PM2.5-related health risks.Trial registrationClinical Research Information Service (CRIS) KCT0009917, Registered 8 November 2024.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-27864-6.Keywords: Particulate matter, Community-based intervention, Older adults, Indoor air quality Introduction PM2.5 (particulate matter ≤ 2.5 μm in aerodynamic diameter) is a complex mixture of microscopic solids suspended in the air. Owing to its small size, PM2.5 can penetrate deep into the lungs and even enter the bloodstream, leading to significant adverse health effects [1]. The International Agency for Research on Cancer (IARC) classified particulate matter from outdoor air pollution as carcinogenic to humans (IARC Group 1) [2]. Despite ongoing improvements, the average annual PM2.5 concentration in South Korea, at 18 µg/m³ in 2023, remains considerably higher than the WHO guideline value (5 µg/m³) and exceeds levels observed in other OECD cities such as Paris and London (13 µg/m³) [3, 4]. Thus, substantial efforts to mitigate associated health impacts remain critical. The health implications of PM2.5 exposure, encompassing both short- and long-term effects, are well-documented. Short-term exposure has been consistently linked to increased cardiovascular and respiratory mortality, contributing to 1 million premature deaths annually [5, 6]. Long-term exposure has similarly established relationships with detrimental health effects including cardiovascular diseases and cancers [7, 8]. Furthermore, recent research underscores potential neurological and psychiatric effects of prolonged PM exposure, suggesting heightened vulnerability among individuals with degenerative diseases, including those that cause dementia [9, 10]. Older adults are particularly vulnerable to PM2.5 exposure because of age-related decline in cardiopulmonary reserve. Indoor air quality and exposure-related behaviors represent tractable and actionable targets for health intervention for older adults. Prior intervention studies have shown that combustion sources such as wood-stove heating substantially elevate indoor PM2.5 in homes [11], and that household-level interventions combining behavioral education with environmental modification can yield meaningful reductions in indoor particulate matter (PM) and improvements in respiratory outcomes [12]. Internationally, community-level interventions are promising strategies for bridging the gap between research and practical applications. To date, policy interventions to address PM-related health risks have predominantly focused on emissions control and ambient air quality management. While some initiatives such as mask distribution, air purifier installations, and public advisories during high-PM episodes have been implemented, these efforts have lacked a coordinated approach to directly enhance individual and community-level health management practices. Research interventions using devices (air purifiers), masks, or smartphone-based informational approaches have been explored primarily at the individual or institutional level, resulting in fragmented, short-lived outcomes rather than sustainable community-level changes. In the United States, emergency notification systems linked to air quality alerts in New York City and targeted interventions during severe air pollution episodes exemplify integrated approaches that combine community engagement with sustained environmental health strategies [13]. Therefore, there is an urgent need to translate PM-related health research into sustainable community-based interventions that actively engage local health institutions and community services. Leveraging established health promotion frameworks such as the national Home-Visit Health Management Program could provide an effective platform for these interventions. In particular, initiatives that are “connected and participatory,” directly involving community stakeholders, may significantly enhance their sustainability and effectiveness. This pilot study aimed to assess the feasibility and effectiveness of a structured community-based educational intervention targeting PM2.5 exposure. Specifically, this study evaluated changes in protective health behaviors, indoor air quality (personal PM2.5), and pulmonary function among older adults participating in community health management programs. The outcomes of this research will aid in the design of integrated, sustainable interventions that not only reduce the health risks associated with PM2.5 but also strengthen community resilience and participatory health management practices. Methods Study design and setting This cluster non-randomized controlled trial was conducted in the Dong-gu District, Incheon (Republic of Korea), within the municipal Home-Visit Health Management Program implemented by the Dong-gu Community Health Center. Seven visiting nurses made quarterly home visits to pre-enrolled clients; because each nurse’s weekly roster is fixed, clusters were defined by the calendar week of the scheduled visit. Clients seen in odd-numbered weeks (intervention clusters) received, in addition to usual care, a structured PM2.5 education module integrated into the routine nursing visit, whereas those visited in even-numbered weeks (control clusters) continued to receive standard services only. At enrolment (September 23–November 11, 2024) and again 6 weeks later, participants completed a standardized questionnaire and underwent portable spirometry. In addition, a low-cost optical PM2.5 monitor was installed in the primary living area and mean indoor PM2.5 concentrations recorded between visits were downloaded to quantify exposure change. The community-level nature of the intervention made individual randomization impracticable. Moreover, cluster allocation minimized contamination and mirrored real-world service delivery [14]. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Inha University Hospital (IRB No. 2024-07-009, approved on August 5, 2024). The protocol was registered with the Clinical Research Information Service (CRIS; KCT0009917, November 8, 2024). Written informed consent was obtained from all participants prior to their enrollment in the study. Eligibility criteria The study participants were community-dwelling older adults aged ≥ 65 years who were receiving home-visit nursing services through the national Community Health Service Program in Dong-gu, Incheon, Republic of Korea. Participant recruitment was conducted from August 2024 to December 2024 in collaboration with public health centers and affiliated home-visiting nurses (Fig. 1). Flow chart of the study participants The eligibility criteria included: (1) age ≥ 65 years, (2) already enrolled in the home-visit nursing service program, comprising both the regular care group (community-dwelling older adults with managed chronic conditions such as hypertension, diabetes, or stable cardiopulmonary disease) and the self-care support group (older adults with elevated risk factors but without a formal chronic disease diagnosis), (3) ability to perform daily activities independently, defined as a full score on the activities of daily living (ADL) scale, and (4) voluntary provision of written informed consent after receiving a full explanation of the study. The inclusion of both program tiers was intentional, ensuring the study population reflected the broader spectrum of older adults served by the program. The exclusion criteria were as follows: (1) receiving long-term care insurance, (2) active treatment for malignancy, (3) current use of psychiatric medications, or (4) judged by the research team or visiting nurses as physically or cognitively unfit for participation. Participants were recruited during routine home visits by the visiting nurses. Trained research staff accompanied the nurses to explain the study and collect contact information from eligible and interested individuals. Follow-up telephone screening was conducted, and participants who met all the criteria were revisited within 7 days for written informed consent and baseline survey administration. Allocation This was a cluster non-randomized controlled trial. Clusters were defined by the nurses’ visiting weeks. Sixty older adults were initially enrolled and allocated to even- or odd-numbered weeks, depending on when community center nurses visited the service recipients. The intervention (n = 30) and control groups (n = 30) were determined based on the community nurses’ visiting schedule. During follow-up, one participant in the control group was lost to follow-up, resulting in a final analytical sample of 59 participants (intervention group: n = 30; control group: n = 29). Baseline demographic and health-related data including age, sex, marital status, education level, body mass index (BMI), smoking history, and alcohol consumption were collected through structured face-to-face interviews conducted by trained research personnel during the initial home visit. The data collection protocol and instruments were standardized across the intervention and control groups. Sample size calculation This pilot study was not intended to provide a definitive estimate of the clinical effectiveness of the educational intervention. Rather, its primary objectives were to determine whether the program could be seamlessly integrated into the municipality’s ongoing Home-Visit Health Management service and to identify any logistical barriers to participant recruitment, follow-up, or data collection ahead of the planned full-scale trial. According to Johanson and Brooks, although statistical power increases with sample size, the gain is not linear; therefore, they recommend approximately 30 participants per arm for feasibility studies [15]. Based on this guideline, we set a target sample size of 60, with 30 individuals each in the intervention and control groups, to obtain adequate information on recruitment and retention rates and protocol adherence without expending unnecessary resources. Intervention Participants in the intervention group received a structured, multifaceted educational intervention designed to improve awareness of PM-related health risks and promote protective health behaviors. The intervention was delivered in person by trained home-visiting nurses during routine home visits, using standardized materials and following the guidance provided by the research team. The educational content was grounded in the health belief model (HBM), a well-established behavioral theory that emphasizes perceived susceptibility, severity, benefits, barriers, cues to action, and self-efficacy [16–18]. Each HBM construct was operationalized as follows: Perceived susceptibility was enhanced through individualized risk narratives linking participants’ age (≥ 65 years old) and comorbidities to elevated PM-related morbidity. Perceived severity was emphasized through lay-friendly explanations of acute cardiopulmonary events and premature mortality attributable to PM exposure. To accentuate perceived benefits, nurses provided concrete “what–when–how” action plans (e.g., limiting outdoor activities, using certified KF94 masks, and activating portable air cleaners), along with data showing potential reductions in symptom days. Perceived barriers (cost of masks or air purifiers, limited digital literacy) were proactively addressed by supplying government-approved masks, demonstrating low-cost ventilation techniques, and offering step-by-step guidance on smartphone AQI applications. SMS reminders and, when needed, telephone follow-ups served as tailored cues to action. Finally, self-efficacy was nurtured through hands-on mask-fitting practice and role-play scenarios that allowed participants to rehearse making decisions about whether to stay indoors during high-PM alerts. The educational topics included (1) the definitions and sources of fine and ultrafine PM (PM2.5 and PM10); (2) health impacts of PM exposure, especially in older adults; (3) importance of regularly checking air quality forecasts; and (4) recommended protective behaviors during days with high PM levels, such as reducing outdoor activities, wearing masks, improving indoor air ventilation, and using air purifiers. The educational session incorporated printed visual aids, verbal explanations, and interactive discussions and typically lasted for 10 min, depending on participant engagement. The study team trained the nurses in both content delivery and motivational communication techniques to ensure consistency and fidelity of the intervention. In addition to the educational session, participants in the intervention group received a set of government-approved facemasks suitable for use during episodes of high PM concentration. Furthermore, during the intervention period, participants received text messages (SMS) whenever fine particle advisories were issued by the local government. These messages served as real-time cues to action, reminding participants of the appropriate protective behaviors to follow. All educational materials were developed by the study team based on national PM guidelines, community health standards, and prior needs assessments of the target population. The control group received usual care under the existing home-visit nursing service, without any PM-specific education or materials during the study period. Data collection Data were collected at two time points: baseline (before the intervention) and follow-up (after the intervention). All data collection procedures were conducted during home visits by trained research staff in collaboration with home visiting nurses. To minimize measurement bias and taking into account the advanced age of the participants, questionnaires were administered face-to-face with verbal explanations provided for each item. Structured questionnaires were used to collect the following information: sociodemographic and health-related characteristics, including age, sex, education level, marital status, BMI, smoking, alcohol consumption, and presence of chronic conditions. The questionnaire used in this study is available in supplementary materials (Supplementary Table S1–2). Awareness and attitudes toward fine PM were assessed by the participants’ understanding of the term PM, awareness of air quality forecasting systems, perception of the health risks associated with PM exposure, and perceived self-efficacy in protecting oneself from PM-related harm. These were measured using binary response options (yes/no) and 4-point Likert scales (1 = ‘not at all’ to 4 = ‘very much’). Protective behaviors related to PM exposure were assessed by the frequency and perceived effectiveness of (a) avoiding outdoor activities on high-PM days, (b) checking PM forecasts, (c) ventilating indoor spaces appropriately, and (d) wearing a mask. Each behavior was assessed using items on perceived necessity, actual practice frequency, and perceived effectiveness, all rated on 4-point Likert scales. In addition to the questionnaire, an environmental exposure assessment was conducted to objectively measure the participants’ indoor air quality. At the time of baseline data collection, a real-time air quality monitor (IAQ-CW LTE; K-weather, Korea) was installed in each participant’s house. The device continuously measured and recorded PM (PM2.5 and PM10), temperature, humidity, carbon dioxide (CO2), volatile organic compounds (VOCs), and noise levels at 1-minute intervals using the light-scattering method. Measurements were transmitted to the research team for offline analysis. Participants did not have access to numerical pollutant concentrations; the device displayed only a four-level color-coded LED indicator reflecting an integrated indoor comfort index that combines PM, CO2, VOCs, and thermal parameters, rather than a pollutant-specific real-time reading. The monitor was operated for approximately 6 weeks and data were collected during the follow-up visit. Behavioral and lifestyle factors, including smoking status, were assessed only at baseline and at the 6-week follow-up; intermediate-time-point interviews were not conducted. Continuous indoor PM monitoring at 1-minute intervals over the entire inter-visit period was used as an indirect indicator of temporal change in household exposure-related behavior. All instruments and protocols were standardized across the intervention and control groups. Prior to the study, the research staff received training in data collection and instrument operation. Daily mean outdoor PM2.5 and PM10 concentrations corresponding to the indoor monitoring period were obtained from the Songnim urban ambient air-quality monitoring station (Dong-gu, Incheon), a national monitoring site operated under the Incheon Institute of Health and Environment. Hourly observations were aggregated to 24-hour arithmetic means to align the temporal resolution with the indoor measurements. Because all participants resided within Dong-gu, Incheon, the Songnim station served as the common urban reference for every household, and each participant’s daily indoor mean was paired with the same-day outdoor value from this station for all outdoor–indoor analyses. Statistical analysis Descriptive statistics were used to summarize participants’ baseline characteristics. Categorical variables were presented as frequencies and percentages, and continuous variables as means and standard deviations. Baseline differences between the intervention and control groups were assessed using chi-square or Fisher’s exact tests for categorical variables and independent t-tests for continuous variables. Paired t-tests were used to evaluate within-group changes between baseline and follow-up. Between-group differences in change scores (follow-up minus baseline) were analyzed using independent t-tests. For multiple-item behavioral and attitudinal outcomes measured on Likert scales, mean differences and standard errors were calculated. To evaluate environmental exposure, we calculated the frequency of exceedance events, defined as the proportion of hourly PM2.5 (≥ 50 µg/m3) or PM10 (≥ 100 µg/m3) measurements that surpassed threshold values over the entire monitoring period. The exceedance event rates (expressed as percentages) were compared between groups using independent t-tests. Variables showing statistically significant baseline imbalance (P < 0.05) were prespecified as covariates for sensitivity analysis. To examine whether between-group differences in indoor air quality parameters were independent of baseline characteristics that differed between groups, analysis of covariance (ANCOVA) was performed for all measured indoor environmental parameters with marital status and BMI as covariates. Adjusted means with standard errors were reported, with marginal means evaluated at the mean values of the covariates. Linear mixed-effects models (LMMs) were employed to account for repeated measurements over time and assess the effects of intervention while controlling for potential confounders. Group (intervention vs. control), time (baseline vs. follow-up), and interaction were included as fixed effects, with random intercepts specified for each participant. This approach allowed us to model both intra- and inter-group differences. The group × time interaction term was the primary parameter of interest, representing the between-group difference in change over time. Fixed effects were prespecified a priori based on the study design. No data-driven model selection was performed. To examine how ambient concentrations explain individual indoor PM exposure, linear mixed-effects models fit the pooled person-day dataset, with log-transformed indoor concentration regressed on log-transformed same-day outdoor concentration, group, and their interaction; a random intercept for participant; and a participant-specific random slope for outdoor concentration. Proportionality was evaluated by testing whether the log-log slope differed from unity. Variance explained was partitioned into marginal R² (fixed effects only) and conditional R² (fixed plus random effects) following Nakagawa and Schielzeth [19]. Likelihood-ratio tests assessed the contributions of random-effect components. Supporting analyses included per-participant log-log regressions, group comparisons of indoor/outdoor geometric-mean ratios using Mann-Whitney U tests, and identification of person-days with standardized residuals exceeding ± 2 standard deviations (SDs) as candidate indoor-source events. All statistical analyses were performed using R software (version 4.4.2). Statistical significance was set at P < 0.05. Results General characteristics of study participants A total of 59 participants were enrolled in the study, of whom 29 were assigned to the control group and 30 to the intervention group (Fig. 1). One participant from the control group was lost to follow-up. Table 1 summarizes the general characteristics of the study participants. There were no significant differences between the two groups in terms of sex, age distribution, education level, smoking status, or alcohol consumption. Most participants in the control and intervention groups were women (79.3% and 73.3%, respectively), and most were aged between 75 and 84 years (75.9% and 63.3%, respectively). Education levels were generally low, with over 70% of the participants in both groups having completed elementary school or less. Smoking and alcohol consumption were also similar across the groups, with over 75% of participants in both groups having never smoked. However, significant differences were observed in marital status and BMI categories. A greater proportion of participants in the intervention group were currently married compared with the control group (60.0% vs. 20.7%, p = 0.005). Regarding BMI, the control group had a higher proportion of participants classified as obese (≥ 25.0 kg/m2) than the intervention group (48.3% vs. 23.3%, P = 0.030). Table 1General characteristics of the study participants Variables Control (n = 29)Intervention (n = 30) P N % N % Sex 0.815 Men620.7826.7 Women2379.32273.3 Age (years) 0.639 65–74310.3620.0 75–842275.91963.3 85+26.9516.7 Education 0.683 Elementary School or Below2275.92273.3 Middle School517.2413.3 High School or Higher26.9413.3 Marriage 0.005 Married620.71860.0 Divorced or widowed2275.91240.0 Others / No response13.50- Body Mass Index (kg/m 2 ) 0.030 Low (≤ 18.5)310.30- Normal (18.6–22.9)620.71136.7 Overweight (23.0–24.9)620.71240.0 Obesity (≥ 25)1448.3723.3 Smoking 0.999 Current26.926.7 Ex-smoker517.2516.7 Never2275.92376.7 Alcohol Consumption 0.775 Yes1034.51343.3 No1965.51756.7P‑values were calculated using the chi‑squared test General characteristics of the study participants P‑values were calculated using the chi‑squared test Awareness and protective behaviors related to fine particles Participants’ awareness, attitudes, and protective behaviors related to PM were assessed at baseline and follow-up using structured questionnaires (Table 2). In terms of awareness, the proportion of participants who correctly understood the definition of PM increased in the intervention group from 20.0% to 33.3%, whereas it decreased in the control group from 24.1% to 13.8%. Awareness of the recommended protective actions also improved slightly in both groups, with the intervention group showing an increase from 70.0% to 83.3%; however, the between-group differences were not significant. The perceived health risks associated with PM exposure remained high in both groups at follow-up, with mean scores approaching the maximum (4.0). In the intervention group, the perceived self-efficacy to reduce PM-related health risks increased from 3.2 ± 1.1 to 3.6 ± 0.9, compared with an increase from 3.0 ± 1.1 to 3.3 ± 1.1 in the control group. Behavioral outcomes such as avoiding outdoor activity on high-PM days, checking PM levels, ventilating appropriately, and wearing masks showed improvements in the intervention group across multiple dimensions (perceived necessity, frequency of practice, and perceived effectiveness). For example, the frequency of checking PM levels increased from 1.4 ± 1.0 to 2.2 ± 1.4 in the intervention group versus 1.5 ± 1.0 to 1.6 ± 1.1 in the control group. Table 2Participants’ PM knowledge and attitudes toward protective behaviors: perceived necessity, practice frequency, and protective effectiveness Items Control (N = 29)Intervention (N = 30) P BaselineFollow-upBaselineFollow-upN (%)orMean ± SDN (%)orMean ± SDN (%)orMean ± SDN (%)orMean ± SD Knowledge of PM Knowledge of what PM is7 (24.1)4 (13.8)6 (20.0)10 (33.3)0.054 Awareness of PM forecasting12 (41.4)12 (41.4)10 (33.3)10 (33.3)0.999 Awareness of actions to avoid PM exposure20 (69.0)25 (86.2)21 (70.0)25 (83.3)0.965 Perceived health harm of PM (Score: 1–4)3.9 ± 0.34.0 ± 03.8 ± 0.54.0 ± 00.354 Self-efficacy to reduce PM risks (Score: 1–4)3.0 ± 1.03.3 ± 1.13.2 ± 1.13.6 ± 0.90.716 Avoidance of outdoor activities on high-PM days Perceived necessity3.6 ± 0.83.7 ± 0.73.5 ± 1.04.0 ± 00.095 Frequency of practice3.1 ± 1.33.3 ± 1.22.8 ± 1.23.6 ± 1.00.071 Perceived protective effectiveness3.8 ± 0.63.7 ± 0.63.2 ± 1.14.0 ± 0< 0.001 Checking ambient PM levels Perceived necessity2.7 ± 1.23.2 ± 1.12.7 ± 1.33.7 ± 0.80.130 Frequency of practice1.5 ± 1.01.6 ± 1.11.4 ± 1.02.2 ± 1.40.009 Perceived protective effectiveness3.2 ± 1.23.2 ± 1.03.0 ± 1.23.8 ± 0.60.013 Indoor ventilation at high-PM days Perceived necessity2.2 ± 1.43.0 ± 1.22.4 ± 1.43.3 ± 1.20.785 Frequency of practice2.1 ± 1.42.6 ± 1.41.7 ± 1.12.8 ± 1.30.073 Perceived protective effectiveness2.2 ± 1.43.0 ± 1.22.4 ± 1.33.3 ± 1.30.777 Mask wearing on high-PM days Perceived necessity3.8 ± 0.83.9 ± 0.44.0 ± 04.0 ± 00.444 Frequency of practice3.6 ± 1.03.9 ± 0.43.7 ± 0.84.0 ± 0.21.000 Perceived protective effectiveness3.8 ± 0.84.0 ± 0.23.8 ± 0.44.0 ± 01.000P‑values reflect the Group × Time interaction term from linear mixed-effects modelScale for scoring items: 1 = Not at all; 4 = Very muchAbbreviations: PM Particulate matter Participants’ PM knowledge and attitudes toward protective behaviors: perceived necessity, practice frequency, and protective effectiveness P‑values reflect the Group × Time interaction term from linear mixed-effects model Scale for scoring items: 1 = Not at all; 4 = Very much Abbreviations: PM Particulate matter Changes in pulmonary function parameters The changes in pulmonary function test results and FeNO levels are summarized in Table 3. Significant group-by-time interactions were observed for percent-predicted FEV1 and FEV6. In the intervention group, FEV1 (% predicted) increased from 84.6 ± 26.2 at baseline to 95.8 ± 23.6 at follow-up, whereas it slightly decreased in the control group from 83.6 ± 22.0 to 81.4 ± 22.8 (P for interaction = 0.008). Similarly, FEV6 (% predicted) increased in the intervention group from 77.2 ± 23.0 to 88.9 ± 21.2, while remaining largely unchanged in the control group (75.4 ± 23.2 to 74.4 ± 25.4; P for interaction = 0.012). Other parameters, including absolute values of FEV1 and FEV6, FEV1/FEV6 ratio, and FeNO levels, did not show significant differences between the groups. FeNO levels decreased slightly in the control group but increased numerically in the intervention group; however, this difference did not reach statistical significance (P for interaction = 0.104). Table 3Changes in pulmonary function test parametersParametersControl (N = 29)Intervention (N = 30) P * BaselineFollow-upBaselineFollow-upMean ± SDMean ± SDMean ± SDMean ± SDFeNO27.9 ± 23.324.2 ± 19.932.3 ± 20.843.4 ± 36.90.104FEV11.5 ± 0.41.4 ± 0.52.4 ± 4.61.7 ± 0.50.489FEV1 (%Pred)83.6 ± 22.081.4 ± 22.884.6 ± 26.295.8 ± 23.60.008FEV61.7 ± 0.61.7 ± 0.72.8 ± 5.42.1 ± 0.60.502FEV6 (%Pred)75.4 ± 23.274.4 ± 25.477.2 ± 23.088.9 ± 21.20.012FEV1/FEV60.9 ± 0.10.9 ± 0.10.9 ± 0.10.9 ± 0.10.782FEV1/FEV6 (%Pred)112.9 ± 10.6112.7 ± 11.2110.8 ± 11.8110.0 ± 13.40.829Abbreviations: FeNO Fractional exhaled nitric oxide, FEV₁ Forced expiratory volume in 1 s, FEV₁% Percentage of predicted FEV₁, FEV₆ Forced expiratory volume in 6 s, FEV₆% Percentage of predicted FEV6, FEV1/FEV6, ratio of FEV1 to FEV6; FEV1/FEV6%, percentage of predicted FEV₁/FEV₆* P-value for the group × time interaction term from linear mixed-effects models Changes in pulmonary function test parameters Abbreviations: FeNO Fractional exhaled nitric oxide, FEV₁ Forced expiratory volume in 1 s, FEV₁% Percentage of predicted FEV₁, FEV₆ Forced expiratory volume in 6 s, FEV₆% Percentage of predicted FEV6, FEV1/FEV6, ratio of FEV1 to FEV6; FEV1/FEV6%, percentage of predicted FEV₁/FEV₆ * P-value for the group × time interaction term from linear mixed-effects models Changes in indoor air quality Indoor air quality was assessed before and after the intervention by measuring the concentrations of various environmental parameters within the participants’ homes, including PM2.5, PM10, CO2, VOCs, temperature, humidity, and noise levels. Across the entire monitoring period, the intervention group showed significantly lower mean concentrations of both PM2.5 (12.3 ± 7.9 vs. 19.6 ± 15.8 µg/m3; P = 0.016) and PM10 (17.0 ± 8.6 vs. 25.6 ± 16.8 µg/m3; P = 0.030) than the control group, while indoor temperature was higher in the intervention group (24.5 ± 3.2 °C vs. 22.7 ± 3.0 °C; P = 0.025) (Table 4). No significant between-group differences were observed for CO2, VOCs, humidity, or noise level. The mean monitoring duration was 47.5 ± 3.4 days for the control group and 45.0 ± 4.3 days for the intervention group (P = 0.016). Table 4The results of indoor air quality measurement of the study participantsParametersControl Group(N = 29)Intervention Group(N = 30) P Mean ± SDMean ± SDMeasured Days47.5 ± 3.445.0 ± 4.30.016PM2.5 (µg/m3)19.6 ± 15.812.3 ± 7.90.016PM10 (µg/m3)25.6 ± 16.817.0 ± 8.60.030CO2 (ppm)888 ± 379845 ± 2460.601VOC (ppb)553 ± 616544 ± 5650.951Temperature (°C)22.7 ± 3.024.5 ± 3.20.025Humidity (%)43.7 ± 7.945.1 ± 11.20.582Noise Level (dB)55.1 ± 4.556.2 ± 4.60.366High PM2.5 Event8.91 ± 13.681.63 ± 2.850.042High PM10 Event1.47 ± 5.320.66 ± 1.920.260Abbreviations: PM2.5 Particulate matter with aerodynamic diameter ≤ 2.5 μm, PM10 Particulate matter with aerodynamic diameter ≤ 10 μm, CO2 Carbon dioxide, VOC Volatile organic compounds The results of indoor air quality measurement of the study participants Abbreviations: PM2.5 Particulate matter with aerodynamic diameter ≤ 2.5 μm, PM10 Particulate matter with aerodynamic diameter ≤ 10 μm, CO2 Carbon dioxide, VOC Volatile organic compounds To assess whether the observed between-group differences in indoor PM concentrations were attributable to baseline imbalances in marital status and BMI, analysis of covariance was conducted with these variables as covariates (Supplementary Table S3). After adjustment, the between-group differences in PM2.5 (adjusted means: control 19.7 ± 2.4 µg/m3 vs. intervention 13.3 ± 2.3 µg/m3; P = 0.064) and PM10 (25.2 ± 2.6 µg/m3 vs. 18.8 ± 2.5 µg/m3; P = 0.083) did not reach statistical significance, although the direction and magnitude of the differences were preserved. Indoor temperature remained significantly higher in the intervention group after adjustment (24.2 ± 0.6 °C vs. 22.4 ± 0.6 °C; P = 0.036). The lack of significant between-group differences for CO2, VOCs, humidity, and noise level persisted in the adjusted analyses. To clarify whether between-group differences in indoor PM emerged immediately at baseline or developed over the course of the intervention, indoor environmental parameters during the first and last weeks of monitoring were compared (Supplementary Table S4). During the first week, between-group differences in PM2.5 (18.7 ± 17.1 vs. 11.8 ± 10.3 µg/m3, P = 0.071) and PM10 (23.2 ± 17.2 vs. 16.1 ± 11.4 µg/m3, p = 0.070) did not reach statistical significance, although the intervention group already showed lower mean values; only indoor temperature differed significantly between groups at this initial time point. By the last week, the intervention group exhibited significantly lower concentrations of both PM2.5 (11.1 ± 5.7 vs. 19.6 ± 15.8 µg/m³, P = 0.010) and PM10 (15.3 ± 7.0 vs. 26.0 ± 21.4 µg/m³, P = 0.015), a temporal pattern consistent with progressive behavioral uptake during the intervention rather than baseline imbalance. In addition to the average concentration levels, the frequency of exceedance events (Supplementary Figure S1), defined as instances in which indoor PM levels surpassed the recommended thresholds, was also compared between the groups (Table 4). The intervention group demonstrated a significantly lower PM2.5 exceedance rate compared with the control group (1.63 ± 2.85 events vs. 8.91 ± 13.68 events, P = 0.042). Although the PM10 exceedance rate was also lower in the intervention group (0.66 ± 1.92 events) than in the control group (1.47 ± 5.32 events), the difference did not reach statistical significance (P = 0.260). Outdoor and indoor relationship of PM2.5 and PM10 concentrations In the linear mixed-effects models, the log-log slope of indoor on same-day outdoor concentration was 1.01 (95% CI, 0.97–1.04) for PM2.5 and 1.16 (1.11–1.20) for PM10. The PM2.5 slope was statistically indistinguishable from unity, whereas the PM10 slope significantly exceeded unity, indicating a near-proportional response for PM2.5 and a slightly amplified indoor response for PM10. The outdoor × group interaction was not statistically significant for either pollutant (P = 0.70 for PM2.5; P = 0.59 for PM10), indicating that the intervention did not alter the outdoor-to-indoor coupling slope itself. Variance partitioning revealed that ambient concentration alone was an incomplete predictor of individual indoor exposure. Outdoor PM explained 47.7% (PM2.5) and 50.0% (PM10) of total indoor variance, rising to 88.3% and 80.8% once between-participant differences were accommodated. Between-participant heterogeneity was substantial: the intraclass correlation was 0.78 for PM2.5 and 0.62 for PM10. Per-participant log-log regressions showed wide variation in fit, with median R2 of 0.68 (IQR, 0.46–0.82) for PM2.5 and 0.61 (IQR, 0.42–0.71) for PM10; outdoor concentration explained less than half of indoor variance in 32.2% (PM2.5) and 42.4% (PM10) of homes. Indoor/outdoor geometric-mean ratios were consistently below one and tended to be lower in the intervention group than in controls, although these between-group differences were not statistically significant. Approximately 5.0% (PM2.5) and 4.7% (PM10) of person-days exhibited standardized residuals exceeding ± 2 SD, with 67 (PM2.5) and 57 (PM10) of these representing upward excursions where indoor concentrations substantially exceeded outdoor-based predictions, a pattern consistent with episodic indoor activities such as cooking, cleaning, or insufficient ventilation. Taken together, these findings indicate that ambient PM is a dominant but not deterministic driver of residential indoor exposure, and that a meaningful fraction of indoor variability reflects home-specific and behavior-specific contributions (Table 5). Table 5Outdoor-indoor relationship of PM2.5 and PM10 concentrationsDomainIndicatorPM2.5PM10ProportionalityOutdoor→Indoor slope(95% CI)1.01(0.97–1.04)1.16(1.11–1.20)Outdoor × Group interaction0.700.59Variance explainedMarginal R2(Outdoor concentration)0.480.50Conditional R2(Outdoor + person)0.880.81Between-person heterogeneityICC / random-slope SD0.78 / 0.380.62 / 0.31Individual fitPer-home R2, median(IQR)0.68(0.46–0.82)0.61(0.42–0.71)Participants with R2 < 0.532%42%Indoor-origin signalsI/O geometric-mean ratio(control vs. intervention)0.60 vs. 0.570.45 vs. 0.37Person-days with |residual| > 2 SD5.0% (67)4.7% (57)n = 59 participants over 105 measurement days; same-day matched person-day observations. Models: linear mixed-effects with log-transformed indoor concentration as outcome. All random-effect likelihood-ratio tests (random intercept; random slope) were P < 0.001 for both pollutants. ICC = intraclass correlation; SD = standard deviation; R² = coefficient of determination; IQR = interquartile range. Marginal and conditional R² follow Nakagawa and Schielzeth (2013). P values for I/O geometric-mean ratio comparison are from Mann–Whitney U testsAbbreviations: β₁ Log-log regression slope of indoor on outdoor concentration, CI Confidence interval, ICC Intraclass correlation coefficient, IQR Interquartile range, I/O Indoor-to-outdoor, LR Likelihood-ratio, PM2.5 Particulate matter ≤ 2.5 μm in aerodynamic diameter, PM10 Particulate matter ≤ 10 μm in aerodynamic diameter, R2 Coefficient of determination, SD Standard deviation Outdoor-indoor relationship of PM2.5 and PM10 concentrations Outdoor→Indoor slope (95% CI) 1.01 (0.97–1.04) 1.16 (1.11–1.20) Marginal R2 (Outdoor concentration) Conditional R2 (Outdoor + person) Per-home R2, median (IQR) 0.68 (0.46–0.82) 0.61 (0.42–0.71) I/O geometric-mean ratio (control vs. intervention) n = 59 participants over 105 measurement days; same-day matched person-day observations. Models: linear mixed-effects with log-transformed indoor concentration as outcome. All random-effect likelihood-ratio tests (random intercept; random slope) were P < 0.001 for both pollutants. ICC = intraclass correlation; SD = standard deviation; R² = coefficient of determination; IQR = interquartile range. Marginal and conditional R² follow Nakagawa and Schielzeth (2013). P values for I/O geometric-mean ratio comparison are from Mann–Whitney U tests Abbreviations: β₁ Log-log regression slope of indoor on outdoor concentration, CI Confidence interval, ICC Intraclass correlation coefficient, IQR Interquartile range, I/O Indoor-to-outdoor, LR Likelihood-ratio, PM2.5 Particulate matter ≤ 2.5 μm in aerodynamic diameter, PM10 Particulate matter ≤ 10 μm in aerodynamic diameter, R2 Coefficient of determination, SD Standard deviation Discussion Main findings This pilot study suggests that a brief community-based educational intervention integrated into routine home-visit nursing services is feasible and may be associated with favorable short-term changes in indoor PM exposure and selected pulmonary function indices among older adults. Regarding pulmonary function, FEV1% and FEV6% significantly improved in the intervention group. Simultaneously, indoor air quality markedly improved in the intervention group. Average indoor PM2.5 and PM10 concentrations were significantly lower in the intervention group than in controls in unadjusted analyses. However, after adjustment for marital status and BMI through ANCOVA, these between-group differences did not retain statistical significance (PM2.5: P = 0.064; PM10: P = 0.083), although the magnitude and direction of the effects were preserved (Supplementary Table S3). The attenuation of significance may partly reflect the modest sample size and the residual confounding influence of these covariates, which differed between groups at baseline. Notably, indoor temperature remained significantly higher in the intervention group even after covariate adjustment (P = 0.036), suggesting that ventilation- or thermal-management-related behavioral changes were more robust to confounding than absolute concentration differences. These findings underscore the importance of interpreting unadjusted group differences with caution and highlight the need for larger trials with covariate-balanced allocation. The role of marital status as a potential effect modifier, which may operate through spousal support for behavioral change, warrants explicit examination in adequately powered future studies. Although some improvements were observed in awareness, perceived self-efficacy, and health-protective behaviors, such as increased frequency of checking PM levels or wearing masks, most of these changes were not significant between the groups. Overall, these findings suggest that targeted education can contribute to measurable improvements in environmental exposure and respiratory health. The improvement in predicted pulmonary function measures supports the potential of behaviorally driven interventions to improve health outcomes in the community. Mechanism of indoor PM reduction The reduction in indoor PM did not appear to operate through altered outdoor-to-indoor coupling: the log-log slope was essentially unity for PM2.5 (1.01; 95% CI, 0.97–1.04) and significantly higher than unity for PM10, and the outdoor × group interaction was not significant for either pollutant. Rather, the effect appears to act through the indoor-source exposure. Ambient concentration alone explained roughly half of the total indoor variance (marginal R2 = 0.48 for PM2.5, 0.50 for PM10), rising to approximately 80–88% only after between-home differences were accommodated. Approximately 5% of person-days exhibited indoor concentrations exceeding outdoor-based predictions by more than two standard deviations, a pattern consistent with episodic cooking, cleaning, or insufficient ventilation. This mechanistic profile is consistent with the educational program’s emphasis on ventilation timing during high-PM episodes, cooking-related exposure reduction, and household cleaning practices, rather than measures that would alter ambient infiltration. This profile also helps explain why no parallel between-group differences emerged for CO2, VOCs, humidity, or noise level. The educational content was explicitly PM-focused rather than promoting generic indoor environmental management. CO2 concentrations in both groups fell within ranges typical of normally occupied residential settings and were unlikely to be substantially altered by the brief, episode-specific ventilation guidance delivered. Indoor VOCs concentrations showed extreme between-participant variability, reflecting the diversity of household sources, few of which were targeted by the intervention; such variability further reduces the statistical detectability of any modest intervention effect. Temperature, humidity, and noise level are not direct consequences of PM-protective behavior. Together, these considerations indicate that the absence of between-group differences in non-PM environmental parameters is consistent with, rather than contradictory to, the PM-specific behavioral pathway through which the intervention plausibly operated. Educational interventions for PM2.5 exposure reduction: A review of previous studies To interpret our findings, we reviewed previous studies on educational interventions aimed at reducing PM2.5 exposure. Several studies among older adults have evaluated whether education can reduce personal or indoor PM2.5 exposure. In the study of rural Navajo and Nez Perce elders, participants who received only behavioral education about air quality hazards showed an approximately 30% reduction in personal PM2.5 exposure, an effect that did not reach statistical significance, whereas those provided with HEPA filtration units achieved a significant 50% decrease in indoor PM2.5 levels [11]. In community samples, home-based interventions combining behavior change education with mechanical controls have yielded a larger PM2.5 decline and modest improvements in pulmonary function. A systematic review reported that multifaceted programs, including cleaners, burning stoves, and air purifiers along with educational components, achieved a 31–53% reduction in indoor PM2.5 and approximately 4.4% average increase in FEV1 [12]. In contrast, stand-alone educational efforts based on behavioral models, such as the transtheoretical model or community health campaigns, have produced inconsistent and generally small decreases in indoor PM2.5, often failing to reach statistical significance. Studies on other vulnerable populations have shown similar patterns. In households with children, adding portable air filters to an educational intervention decreased indoor PM2.5 levels and lowered children’s respiratory infections [11]. In contrast, education alone had little effect on air quality. In cardiac rehabilitation patients, tailored education boosted pollution knowledge but did not change home PM2.5 levels [20]. Impact of educational interventions on respiratory outcomes A key finding of this study was a significant improvement in pulmonary function (FEV1% and FEV6%) in the intervention group. However, this result must be interpreted with caution. As this was a pilot study with a modest sample size, this finding may be attributable to chance rather than a true intervention effect. Furthermore, achieving measurable changes in pulmonary function from a single, brief educational session is generally unexpected, highlighting the need for careful consideration. Despite these caveats, prior research suggests that more intensive and targeted educational programs can produce modest but meaningful gains. An educational intervention study involving 157 Moroccan adults with asthma reported significant increases in FEV1, FVC, and PEFR [21]. Similarly, a recent systematic review of household-level interventions found an average FEV1 increase of approximately 4.4% [12]. These findings indicate that comprehensive education delivered in multiple sessions can lead to behavioral changes that enhance pulmonary function. Our pilot study employed a single educational session tailored to older adults without an active or severe respiratory diagnosis, which may have lacked sufficient intensity or duration to change the outcomes. Nevertheless, it is important to note that pulmonary function can also vary in the short-term depending on fine-particle exposure. Experimental and observational studies have demonstrated that acute exposure to PM can reduce lung function within hours to days [22, 23], suggesting that meaningful changes are biologically plausible, even over relatively short timeframes. In contrast to the FEV-based measures, FeNO showed a numerical increase in the intervention group (32.3 → 43.4 ppb) that did not reach statistical significance (p = 0.104), in a direction opposite to what would be expected from reduced PM exposure alone. According to the American Thoracic Society clinical practice guideline, FeNO is a marker of eosinophilic airway inflammation rather than of total airway inflammation, and its level is influenced by atopy, recent respiratory infections, inhaled corticosteroid use, seasonal aeroallergen exposure, and smoking, all of which can vary independently of ambient or indoor PM concentration [24]. The present study was conducted between late September and mid-November, a season in which aeroallergen exposure changes substantially in Korea, and we did not systematically capture changes in respiratory medication or recent respiratory illness during the intervention period. The directional pattern of FeNO is therefore most plausibly attributable to a combination of inter-individual variability in baseline airway inflammation, seasonal allergen variation, and unmeasured medication exposure rather than to a true intervention effect; the wide standard deviation in the intervention group at follow-up (± 36.9 ppb) is consistent with such heterogeneous influences. This interpretation should be regarded as exploratory, and future trials with larger samples should prospectively capture inhaled corticosteroid prescriptions, respiratory infection history, and atopic status to permit covariate-adjusted analysis of FeNO as a secondary outcome. Sustainability of educational interventions Community-based education programs aimed at reducing indoor PM typically achieve high initial uptake and improvements in knowledge but rarely demonstrate long-term integration or sustained air quality benefits. For instance, a two-session, 45-minute workshop on home ventilation and pollution sources markedly improved participants’ understanding; however, the curriculum was not integrated into local health services for ongoing reinforcement [25]. Similarly, a one-hour home visit based on the CDC–HUD Healthy Housing Manual increased awareness of indoor air hazards but did not improve objective PM2.5 measurements and lacked structured follow-up [26]. In our pilot study, a single 10-minute educational session proved both feasible and well-received among older adults. Importantly, air quality monitoring showed that the average indoor PM level in the intervention group was significantly lower than that in the control group, demonstrating a clear between-group reduction effect. Interestingly, perceived self-efficacy showed only modest, non-significant gains in the intervention group (3.2 → 3.6) relative to controls (3.0 → 3.3; P = 0.716 for the group × time interaction), and several other self-reported attitudes and awareness items shifted minimally. In contrast, the objective indoor PM measurements indicated a different outcome. During the study period, average indoor PM2.5 and PM10 concentrations were significantly different between the intervention and control groups, indicating that practical guidance and timely pollution alerts can prompt real behavior changes, even when participants do not consciously report feeling more informed. One interpretive caveat directly relevant to sustainability concerns the role of the indoor air quality monitor itself, which was installed in both the intervention and control households for measurement purposes. Although the monitor did not display numerical pollutant concentrations to participants, it did provide a four-level color-coded LED indicator reflecting an integrated indoor comfort index. This integrated signal is not pollutant-specific. A deteriorating LED reading could equally reflect CO2, VOCs, or thermal departure rather than PM elevation, so the device is unlikely to have driven PM-specific behavioral responses on its own. Nevertheless, awareness that air quality was being monitored in their homes, together with this generalized environmental feedback, could in principle have generated some behavioral change independently of the educational and SMS-alert components. Some indirect evidence argues against a strong monitor-only effect: indoor PM2.5 concentrations in the control group did not decline between the first and last weeks of monitoring (18.7 → 19.6 µg/m3), whereas a substantial monitor-only behavioral effect would be expected to produce some downward trend in both groups. However, full separation of the educational and monitor components would require a three-arm design (education + monitor vs. monitor-only vs. usual care) that was beyond the scope of this pilot trial. Limitations Several limitations of the present study, stemming from both its pilot scale and its embedded, real-world implementation within an existing community service, warrant explicit consideration when interpreting the findings. First, this was a pilot trial with a modest sample of older adults, which limits statistical power and may have prevented the detection of more subtle but meaningful changes in PM2.5 exposure and protective behaviors. The cluster allocation by visiting weeks, while pragmatic within the existing home-visit schedule, did not involve individual-level randomization and may have introduced selection bias; the statistically significant baseline imbalances in marital status and BMI are consistent with this concern. Although the ANCOVA-based sensitivity analysis preserved the direction of intervention effects after adjustment for these covariates, residual confounding from unmeasured selection-related characteristics cannot be excluded. Future trials should use stratified or covariate-adaptive randomization at the individual level to reduce selection bias. Second, the generalizability of these findings warrants careful consideration. Eligibility was restricted to community-dwelling older adults with full functional independence on the Activities of Daily Living scale who were not receiving long-term care insurance benefits, undergoing active treatment for malignancy, or taking psychiatric medications. Although the population spanned both regular care and self-care support tiers of the program, the findings may not generalize to older adults with severe respiratory disease, substantial functional or cognitive impairment, or those receiving facility-based care. Furthermore, individuals already engaged in a community health service may be more health-conscious and more receptive to behavioral guidance than non-enrollees, potentially inflating the apparent effectiveness of the intervention relative to a general older-adult population. Third, several determinants of indoor exposure were not measured directly. Self-reported behavioral and lifestyle variables, including smoking, were captured only at baseline and at 6-week follow-up; intermediate-time-point interviews were not conducted, so transient or progressive behavioral changes during the intervention, including any contribution of altered smoking to indoor PM reduction, could not be directly quantified. Time spent indoors, number of household occupants, ventilation systems, heating type, and outdoor activity patterns were also unmeasured; marital status partially captured cohabitation but is an incomplete proxy and, as discussed in Sect. 4.1, may further operate as an effect modifier through spousal support for behavioral change. Indirect evidence of unmeasured heterogeneity is nonetheless visible in our outdoor-indoor analysis, where between-participant random-slope variability was substantial and approximately one third of participants showed weak outdoor-only fit. The lack of direct measurement was partially offset by continuous PM monitoring, which captured temporal exposure trajectories, and the divergence between groups from the first to the last week of monitoring is consistent with progressive behavioral uptake. Future trials should incorporate brief time-activity diaries, structured questions on household composition and ventilation, and outdoor-activity logs alongside repeated digital self-report. Finally, the 6-week monitoring period precludes inference about whether the observed improvements in indoor PM and pulmonary function would be sustained over longer time horizons; behavioral effects of brief interventions are well known to attenuate without continued reinforcement. The intervention consisted of a single educational session supplemented with automated SMS alerts only during episodes of high PM concentration. Although this minimal format was chosen to maximize feasibility and sustainability within community services, it may not have provided sufficient intensity or reinforcement to foster lasting behavioral changes, and participants may not have developed consistent daily habits for ventilation, mask use, or air-quality monitoring. Information on participants’ baseline pharmacological treatment and on any treatment changes during the intervention period was also not collected, which limits causal interpretation of the pulmonary function and FeNO findings. Future trials should consider multi-session curricula, more immersive or hands-on components, continuous feedback mechanisms, and prospective collection of medication history to strengthen and sustain protective behaviors across larger and more diverse cohorts of older adults. Conclusions In this 6-week pilot trial, a brief community-based educational intervention integrated into a home-visit health management program was feasible, well-received, and associated with short-term improvements in indoor PM2.5 and PM10 concentrations and in selected pulmonary function indices (FEV1% and FEV6%) among community-dwelling older adults. The intervention appeared to act primarily through PM-specific behavioral changes rather than through alteration of ambient-to-indoor coupling, providing a coherent mechanistic basis for the observed indoor PM reduction without parallel changes in other indoor environmental parameters. These findings should, however, be interpreted as hypothesis-generating rather than definitive. Several constraints temper the strength of the conclusions: the modest sample size of a feasibility trial, statistically significant baseline imbalances in marital status and BMI consistent with non-randomized cluster allocation, attenuation of between-group differences after covariate adjustment, a 6-week monitoring period that precludes inference about long-term sustainability, and the present design’s inability to fully disentangle the educational and SMS-alert components from any monitor-induced awareness effect. Several behavioral and lifestyle determinants of indoor exposure were not directly captured. Notwithstanding these caveats, the present results highlight the potential of leveraging existing community health infrastructures to deliver pragmatic, behavior-focused environmental health interventions for vulnerable older populations. Confirmation through larger, individually randomized, multi-arm trials with longer follow-up and prospective measurement of behavioral, pharmacological, and household-level covariates is warranted before broader implementation can be recommended. Supplementary Information Supplementary Material 1. Acknowledgements Not applicable. Declaration of generative AI in scientific writing During the preparation of this work authors used ChatGPT (OpenAI, CA, USA) for language editing and grammar checking. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article. Authors’ contributions Hyung Doo Kim: Writing – Original draft preparation, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Myung-Sook Park, Woo-Jin Kim, Sumi Chae: Formal analysis, Visualization. Jiho Lee, Myung Ho Lim: Writing – Review and Editing, Investigation. Dong-Wook Lee: Writing – Review and Editing, Supervision, Resources. Hwan-Cheol Kim: Writing – Review and Editing, Supervision, Funding acquisition, Resources. Funding This study was supported by a grant from the Korea National Institute of Health [grant no. 2024-ER0911-00]. Data availability The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Inha University Hospital (IRB No. 2024-07-009, approved on August 5, 2024). The protocol was registered with the Clinical Research Information Service (CRIS; KCT0009917, November 8, 2024). Written informed consent was obtained from all participants prior to their enrollment in the study. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Dong-Wook Lee, Email: is2uz@snu.ac.kr. Hwan-Cheol Kim, Email: carpediem@inha.ac.kr. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request.