Assessments of systemic health issues can conveniently be structured on the basis of a causal chain, representing the key links between primary sources and ultimate impacts (see link to The causal chain for more information). 

The position of scenarios in relation to the causal chain may vary.  Some are exogenous (upper figure), and describe only the external context or driving forces that influence this chain (e.g. policy developments, technological changes). Most, however, are endogenous (lower figure) in that they describe links within the chain – for example, levels of (or changes in) source activities, releases, ambient concentrations and/or exposure.

As these figures show, the position of the scenario in relation to the causal chain fundamentally affects the character and scope of the assessment.  This is because the scenarios give the fixed assumptions for the assessment.  The assessment itself takes the conditions thus specified as given, and considers only the ‘downstream’ effects, as they work through the causal chain.

• Exogenous scenarios define the external conditions (i.e. context) but tell nothing direct about the changes that might occur within the causal chain.  These all have to be deduced as part of the assessment – for example, by modelling the way the external influencers work their way from source to impact. 
• Endogenous scenarios, in contrast, define some aspects of the causal chain – those which are specified as part of the scenario.  The assessment is only concerned with estimating the subsequent implications .  In the case of an emissions scenario, therefore, expected changes in emission are predefined, and the assessment focuses on estimating how these work through to affect environmental concentrations and exposures, and thence health effects.  In the case of an exposure scenario (i.e. where changes in exposure are specified), the assessment only has to consider the link from exposure to health impact.

More detailed and encompassing scenarios might seem useful, since they define more clearly the conditions under which the assessment should be done.  They do so, however, at the cost of predefining more of the results, and limiting the independence of the assessment.  A balance needs to be sought, such that the scenario defines only what can legitimately be taken as given, while the assessment should explore and discover what is not known.