During and after the Gulf War in 1991, there was a severe air pollution episode due to more than 700 ignited oil wells. The civilian population experienced heavy exposures with high peaks of fine particles in a large region along the Persian Gulf. There is a large base of scientific literature about the effects of fine particles in ambient air. However, it is not clear how applicable this information is on a special event with a particular pollutant composition, exposure pattern, and population structure. The mortality due to this exposure in Kuwait was estimated in this study using a formal elicitation of expert judgement.
Ten leading air pollution experts in Europe were asked to participate in the study, and six accepted. They were familiarized to the subject and the procedure in a workshop. They were interviewed using a fixed set of questions related to air pollution effects in various settings. The estimates were asked in the form of probability distrubutions. Two questions were directly related to the effects of the oil fires on the civilian population in Kuwait. The experts were asked to explicitly report their reasoning for their estimates. In addition, they were asked air pollution questions for which the right answer was available to the interviewer after the interviews. The experts were given performance scores based on the calibration and precision they showed in these questions. The assessments of the experts were then combined by weighting each expert by his or her score.
The predominant picture was a greater degree consistency than often observed in other studies. This was especially true with questions on long-term exposure-response coefficients in the US or Europe. With the specific question son Kuwait, the estimates differed more. Major reasons for this were different views on (a) the duration of effect in the population, (b) the potency of the fine particles from the oil fires, (c) the effect in the young, and (d) the ture personal exposures given the modeled outdoor concentrations.
As a group, the experts were well calibrated, although there were significant differences between the performance scores. The performance-weighted combination of experts produced the most calibrated estimates. The combined estimate for the number of attributable deaths in Kuwait had very long tails up and down. The interquartal range was from 13 to 82.