Short-term dynamics of indoor and outdoor endotoxin exposure: Case of Santiago, Chile, 2012

Abstract

Indoor and outdoor endotoxin in PM2.5 was measured for the very first time in Santiago, Chile, in spring 2012.Average endotoxin concentrations were 0.099 and 0.094 [EU/m3] for indoor (N = 44) and outdoor (N = 41)samples, respectively; the indoor–outdoor correlation (log-transformed concentrations) was low: R = −0.06,95% CI: (−0.35 to 0.24), likely owing to outdoor spatial variability.A linear regression model explained 68% of variability in outdoor endotoxins, using as predictors elementalcarbon (a proxy of traffic emissions), chlorine (a tracer of marine air masses reaching the city) and relativehumidity (a modulator of surface emissions of dust, vegetation and garbage debris). In this study, for the firsttime a potential source contribution function (PSCF) was applied to outdoor endotoxin measurements. Windtrajectory analysis identified upwind agricultural sources as contributors to the short-term, outdoor endotoxinvariability. Our results confirm an association between combustion particles from traffic and outdoor endotoxinconcentrations.For indoor endotoxins, a predictive model was developed but it only explained 44% of endotoxin variability; thesignificant predictors were tracers of indoor PM2.5 dust (Si, Ca), number of external windows and number ofhours with internal doors open. Results suggest that short-term indoor endotoxin variability may be driven byhousehold dust/garbage production and handling. This would explain the modest predictive performance ofpublished models that use answers to household surveys as predictors. One feasible alternative is to increasethe sampling period so that household features would arise as significant predictors of long-termairborne endotoxinlevels.