In this study, the investigators tested whether portable air cleaners can improve the air quality in primary school classrooms. Poor air quality can affect children's respiratory health and learning. Classroom air quality is a key concern, as children spend a large part of their day there, often with many students in one room. The investigators studied air quality in 180 classrooms in 29 Dutch primary schools. Within each school, classrooms were grouped in threes based on similar characteristics, such as size, number of students, and ventilation system. Classrooms in each group were randomly assigned to receive either: an air cleaner with a filter that traps particles from the air, an air cleaner that uses an electric charge to remove particles from the air, or no air cleaner (control). All devices were tested for safety in a laboratory beforehand. Measurements were taken in all classrooms before the devices were switched on, to provide a refence point. The air cleaners were then operated for nine weeks. After that, air cleaners were switched off for two weeks while additional measurements continued. To measure the effect of air cleaners, special cloths were placed in classrooms to collect dust and tiny droplets from the air, which were later analyzed in a laboratory for bacteria and viruses. This included viruses that commonly cause respiratory infections, such as flu and respiratory syncytial virus (RSV), and bacteria originating from the human skin, nose, or mouth that can spread through the air. Small sensors also tracked tiny particles in the air (particulate matter). The investigators collected additional information on classroom attendance and parent-reported respiratory symptoms to explore possible links. This study provides real-world evidence on how portable air cleaners can affect classroom air quality and may help guide decisions to improve indoor air environments in schools.

The cluster-randomized trial evaluated the effectiveness of portable air cleaners in 180 classrooms across 29 Dutch primary schools. Classrooms were grouped within schools based on similar characteristics and randomly assigned to receive either HEPA-filter air cleaners, ionization/plasma air cleaners, or no air cleaners. The study contained a three-week baseline, followed by three repeated three-week intervention periods and a two-week post-intervention period, resulting in 14 weeks of measurements per school. Air cleaners were pre-tested for safety and operated at a clean air delivery rate (CADR) of approximately 400-500 m³/h during intervention. Airborne dust was collected using electrostatic dust fall collectors (EDCs), a passive air sampling method, and analyzed for respiratory viruses, human-associated bacteria and a general bacterial marker using quantitative polymerase chain reaction (qPCR). In a subset of eleven classrooms, active air sampling was performed alongside EDCs, to validate and quantify passive measurements. Low-cost indoor air sensors continuously monitored particulate matter, CO₂, temperature, relative humidity, and volatile organic compounds (VOCs). Retrospective parent-reported respiratory symptoms and classroom absenteeism data were gathered by surveys. Outcomes will be analyzed using hierarchical mixed-effects models.

Inclusion Criteria:

1. Primary schools with at least three comparable classrooms between grade 3-8
2. Regular primary education classrooms
3. Primary schools located within reasonable travel distance of Utrecht, the Netherlands

Exclusion Criteria:

1. Schools for special education
2. Secondary schools
3. Kindergarten classes (grade 1&2; approximately ages 4-6 years)
4. Schools unable or unwilling to provide the required study information

Laboratory personnel performing extraction and analysis of microbial air samples were blinded to the classroom intervention allocation. Participants (students and teachers) and investigators were not blinded because the presence of air cleaning devices in classrooms was visible. Analysis of particulate matter (PM) measurements was conducted using sensor data and was not blinded to intervention status.