All approaches have in common that predictions are based on knowledge of, and extrapolation from, previous events. Information on the web of causation and the size and direction of different impacts can be derived from studying the consequences of decisions, which have already been implemented, or events, which have already occurred. This gives better understanding of how systems react, which can be used to make better predictions for future policy. The study of consequences of policies that have been implemented is usually referred to as retrospective Health Impact Assessment, although it could equally be described as an evaluation, despite the claims of some (Milner and Marples, 1997) that these are distinct activities.