Air filters in early childhood

We will install high-efficiency particulate air (HEPA) filters in preschool classrooms to reduce children's exposure to indoor air pollution. The intervention targets classrooms in public schools in Bogotá, Colombia, with children aged 4–6 enrolled in the “transición” grade.

Out of 266 selected schools, approximately 105 will receive air purifiers. Teachers in those classrooms will be trained on the proper use and maintenance of filters.

2025-08-01

2025-12-31

Cognitive assessments will be administered to all students to measure learning outcomes, and teacher surveys will collect information on classroom environmental conditions and children's respiratory health.

Cognitive Test Scores
Measured using a structured assessment designed for children aged 4–6, capturing early learning competencies such as number and letter recognition, memory, shape and color identification, and basic reasoning skills

Respiratory Health Indicators
Based on teacher-reported data on the frequency of symptoms such as persistent cough, wheezing, asthma, or respiratory infections among children in the classroom.

School Attendance
Based on administrative records, measuring days missed by individual students over the academic year

The cognitive test score will be constructed using item response theory from a custom-designed assessment instrument for children aged 4–6. The test includes modules on:

Personal awareness (e.g., name, age, location)

Visual recognition and memory (object and order recall)

Shape, size, and color identification

Number and letter recognition

Quantitative reasoning (counting, comparing quantities)

Vocabulary and listening comprehension

Puzzle completion (fine motor and problem-solving skills)

This study employs a randomized controlled trial (RCT) to assess the effect of indoor air purification on early childhood learning and health outcomes in Bogotá, Colombia.

We selected 266 public preschools with eligible “transición” classrooms (ages 4–6). Following baseline cognitive assessments and consent collection, approximately 105 schools will be randomly assigned to the treatment group, where HEPA air purifiers will be installed in one classroom per school. The remaining schools will form the control group and will not receive any intervention during the study period. The final sample size will depend on the number of schools that consent to be part of the study.

All students will complete a brief cognitive test tailored to their developmental stage. Teachers will complete a brief survey on classroom air quality and the respiratory health of children.

Randomization is stratified by locality to ensure geographic balance. The primary objective is to determine whether improving air quality through the use of HEPA filters enhances cognitive performance and health indicators in young children.

Randomization will be conducted in office by a computer using a reproducible, stratified randomization procedure. Schools will be stratified by locality before random assignment to treatment (HEPA filter installation) or control group.

School

Yes

266 schools (public preschools in Bogotá, Colombia).

Approximately 5,320 students in grade transición (ages 4–6)

Control group: ~161 schools
Treatment group (HEPA filters): ~105 schools

(Exact numbers may vary depending on consent and school participation rates)

We are planning to randomly assign 105 treatment early childhood centers. While there are 363 centers in total, and therefore 258 controls, given the lack of administrative data, we will collect data from 150 randomly selected control centers for most outcomes (and is the number we used for our power calculations). As the experiment is randomized at the early childhood center level, but the main results are observed at the student level (i.e., scores from standardized tests), it must be taken into account that these results have some correlation with each other. This is done by clustering the standard errors at the school level. We assume for the calculations that the power (β) is equal to 0.8 and the level (α) is 0.05. Using historical data, we know that the intracluster correlation coefficient for student test scores (ρ) is 0.035, and we have an average of 30 students per school. The control variables (e.g., students’ age and gender, socio-economic level, strata fixed effects, baseline scores, and historical pollution levels) explain 30% of the variance in the final result and reduce the intra-cluster correlation to 0.013. Under these assumptions, using the formula described above, the MDE is 0.0924 without controlling for covariates. When including these control variables, the MDE is 0.0636.