Abstracts: ISEE 22nd Annual Conference, Seoul, Korea, 28 August–1 September 2010: Air Pollution - Exposure Characterization and Health Effects
The dissemination of high efficiency cooking/heating systems is among the most cost-effective interventions to reduce the burden of disease caused by the exposure to indoor air pollution (IAP). Measuring the population dynamics of time-location and personal-activity behavior that affect IAP exposure is critical to quantify the effectiveness of such interventions. We obtained marginal estimates by age, sex, and stove type (improved/open fire) of time spent in the kitchen and personal exposure to PM2.5. We estimated the variance components of individual and shared time budgets and calculated the probabilities of exceedance and overexposure for the population.
We used 2 novel sensor technologies developed at UC Berkeley to collect minute-by-minute samples of personal exposure: the Time-Activity Monitoring System (TAMS) and the Particulate-and-Temperature System (PATS). We deployed the devices on 61 homes (36-open fire, 25-stove) of the CRECER Guatemala study and collected quarterly measurements over 2-years on each adult women, baby, and younger child. We obtained marginal estimates using GEE model, and estimated the variance components using a random effects model. The probabilities of exceeding air quality guidelines were calculated from the population distributions.
For adult women, median 24-hour kitchen concentrations of PM2.5 for the fire and stove groups were different at the 5% level, but the differences in kitchen time-activity were not significant (4.1 and 3.6 hours for fire and stove, respectively). Intraclass correlation coefficients for time-activity were: 18% for the fire group and almost 0% for stoves. The probability that a randomly selected measurement in the stove group exceeds the PM2.5 24-hour EPA-NAAQS (0.035 mg/m3) is 48.8%, while the probability that a typical person in the stove group would be overexposed is 93.0%.
Our results highlight the importance of behavioral differences between groups and within subjects to assess the impact of IAP interventions and to determine sample size requirements for effective monitoring.