WHAT RISK ASSESSMENT CAN TELL US ABOUT THE MORTALITY IMPACTS OF THE KUWAITI OIL FIRES

Abstract

ISEE-367 Introduction: During and after the Gulf War in 1991, the burning of more than 700 oil wells caused a severe air pollution episode in Kuwait and neighboring countries. The Kuwaiti civilian population experienced heavy exposures to fine particles over a period of about 9 months following the fires. To characterize the population exposure and likely mortality risks from the fires, we conducted detailed atmospheric modeling, a population location survey, and a formal elicitation of expert judgment on particulate matter mortality-response. Methods: Our analysis involved three elements: (1) We modeled dispersion of fire plumes and estimated ground-level impacts at census block and neighborhood centroids in Kuwait to assess particulate matter concentrations attributable to oil fires; (2) We combined these modelling concentration estimates with information on population mobility to estimate the distribution of exposures in this population. Population locations were obtained via survey and combined with modeled concentrations in a GIS-based model to account for the war-related migration of Kuwaiti civilians; and (3) We conducted a formal elicitation of expert judgment to estimate and characterize uncertainty in particulate matter mortality-response. Results of these three projects were used to estimate risk and mortality attributable to the oil fires. Results: Our screening estimates, based on the updated American Cancer Society mortality-response coefficient and concentration estimates from an earlier US Department of Defense modeling study, suggested that on the order of 35 deaths were attributable to smoke from the oil fires. Mortality estimates based on the air pollution modeling, population mobility study, and expert elicitation projects described above support this estimate and suggest that the true value could be several times higher. Discussion: The true mortality impacts of the oil fires may never be known. However – (i) analytic epidemiology (to be reported later by Professor Dockery) suggests that mortality among Kuwaitis who remained in the country during this period was elevated in comparison to that observed among those who left the country; (ii) levels of PM10 measured in Kuwait between May and November of 1991 were on the order of 300 μg/m3; (iii) initial air pollution dispersion modeling suggested that approximately 10 μg/m3 of PM2.5 was due to oil fire smoke; our independent modelling indicated that the contibution of the fires may have been 4 to 5 times this large; (iv) standard regulatory approaches for estimating the mortality impact of 10 μg/m3 of smoke would suggest that 35 deaths among the Kuwaiti national population are attributable to this smoke; and (v) the best estimates (i.e., median values) of five of six leading European experts in the epidemiology and toxicology of PM were this large or larger.