Assessing Strong PM Source Levels and Patterns of Exposure Using Log-Probability Analyses of Cookstove Study Data

Abstract

Rodes, C.1., Chartier, R. 1 Background Chronic diseases with long latency have been shown to be associated with elevated PM exposure levels. But data are needed to evaluate how acute exposure patterns influence pneumonias. Strong intermittent sources such as solid fuel smoke from indoor cooking fires create widely varying patterns in concert with high mean levels. How should we measure and analyze these exposure data to devise the most robust exposure-response health studies? Aims Measure integrated and time series PM data and place into log-probability formats to gauge the information content. Determine the influence of co-factors on the distributional parameters including data variability (GSD's). Examine the information content gained or lost by adding personal level monitoring for the cook, to a fixed location indoor monitor. Methods Utilize the RTI MicroPEM exposure monitor in personal and indoor sampling modes to collect filters and measure real-time data in indoor air pollution (IAP) settings. Apply temporal analyses and log-probability graphing, computing and comparing GSD values for each time period, location, and scenario. Results Time series graphs were not very informative. Log-probability formatting clearly divided background and burn period exposures (distinctly different GSD's), and defined the probit level thresholds. The inflection point between the log-normal lines represented the fraction of time the stove operation influenced exposures. Burn period GSD levels were influenced by the presence of multiple stoves and the level of air exchange. Conclusions Log-normal analyses provided new insights for IAP settings and explained how the exposure variability changed between background and cookstove burn periods. Utilizing the inflection point eliminates the need for stove burn monitoring. The log-probability output provides a strong format for gauging the importance of exposure patterns along with level data across participants, stove types, and studies.