New thesis on HEPA air cleaners in school classrooms: 🧵 TL;DR: In this school study, HEPA air cleaners showed little reliable impact on absenteeism, and air-quality improvements were limited during actual classroom use.

The study installed portable air purifiers (HEPA-based, some with extra filtration layers) in many classrooms and compared illness-related absenteeism to control classrooms with no purifiers.

They didn't see much difference in particle counts between control and treatment (T) classrooms when occupied, but they only took a couple of 24hr measurements. Other studies with continual measurement have shown reductions in PM with filters present. Unoccupied = PM reduction

But the big question was absenteeism. Do cleaner classrooms = fewer sick days? Short answer: the evidence is mixed and weak.

Some statistical models show small reductions in illness-related absenteeism in HEPA classrooms—mostly in Fall and Spring. But the effect sizes are tiny (fractions of a day per student).

Other models show no statistically significant effect at all once you adjust for school, classroom, season, and time trends. In other words: the “benefit” disappears depending on how you model the data.

And importantly: ⚠️ In Winter, some models show higher absenteeism in HEPA treated classrooms compared to controls—especially at higher equivalent ventilation rates.

Why the uncertainty? The study ran many overlapping models (different seasons, moving averages, interactions). Results change depending on the model → classic risk of model-dependence (and over-interpretation).

This doesn’t mean the researchers did anything wrong—sensitivity analyses are useful. But without a single pre-specified primary model, statistically significant results should be treated cautiously.

The study does control for important confounders (school effects, classroom effects, seasonality). But it can’t control for individual health, household exposure, immunity, or behaviour—so residual confounding remains likely.