PNNL

Abstract

Outcome Metrics Methodology defines a way to evaluate outcome metrics associated with scenario analyses related to biological countermeasures. Previous work developed a schema to allow evaluation of common elements of impacts across a wide range of potential threats and scenarios. Classes of metrics were identified that could be used by decision makers to differentiate the common bases among disparate scenarios. Typical impact metrics used in risk calculations include the anticipated number of deaths, casualties, and the direct economic costs should a given event occur. There are less obvious metrics that are often as important and require more intensive initial work to be incorporated. This study defines a methodology for quantifying, evaluating, and ranking metrics other than direct health and economic impacts. As has been observed with the consequences of Hurricane Katrina, impacts to the culture of specific sectors of society are less obvious on an immediate basis but equally important over the ensuing and long term. Culture is used as the example class of metrics within which • requirements for a methodology are explored • likely methodologies are examined • underlying assumptions for the respective methodologies are discussed • the basis for recommending a specific methodology is demonstrated. Culture, as a class of metrics, is shown to consist of political, sociological, and psychological elements that are highly valued by decision makers. In addition, cultural practices, dimensions, and kinds of knowledge offer complementary sets of information that contribute to the context within which experts can provide input. The quantification and evaluation of sociopolitical, socio-economic, and sociotechnical impacts depend predominantly on subjective, expert judgment. Epidemiological data is limited, resulting in samples with statistical limits. Dose response assessments and curves depend on the quality of data and its relevance to human modes of exposure. With uncertain data and limited common units, the aggregation of results is not inherently obvious. Candidate methodologies discussed include statistical, analytical, and expert-based numerical approaches. Most statistical methods require large amounts of data with a random distribution of values for validity. Analytical methods predominate wherein structured data or patterns are evident and randomness is low. The analytical hierarchy process is shown to satisfy all requirements and provide a detailed method for measurement that depends on expert judgment by decision makers.