ABSTRACT
The combination of disease state, patient population, and types of endpoints considered in a clinical outcomes trial determine the commonly used classification of such studies as either treatment or prevention trials. This chapter will focus
specifically on the prevention paradigm and examine design and analytical issues particularly for cardiovascular outcomes trials. This chapter will explore in some detail issues relating to definition of study
endpoints and event rate estimation; further discussion will focus on exposure considerations as well as the analysis of composite endpoints and accommodating the possibility of multiple events per patient. Each of these broad topics present a number of practical challenges that warrant careful consideration. In the latter part of the chapter, topics of interest that all relate to prevention trials more generally are discussed as well. In Section 13.3, the Raloxifene Use for The Heart (RUTH) study will be used to illustrate concepts. Prevention trials typically focus on individuals with a chronic condition in which
the goal of intervention is to prevent an undesirable event for which this population is at risk, but which may take years to develop. The chosen terminology emphasizes the most desirable outcome of risk elimination, though in practical terms the context in which the term ‘‘prevention’’ is used is often synonymous with ‘‘risk reduction,’’ and in some cases this is essentially tantamount to a significant delay in event onset. This contrasts with the treatment trial paradigm, which focuses on interventions for the amelioration of acute conditions, or the management of sequelae of acute injury. The above notwithstanding, a patient in the prevention setting may in some cases be subclassified as belonging to a secondary versus a primary prevention population, depending on whether or not he=she has already experienced a key event of interest at some point previously and is at risk for subsequent events, as opposed to being at risk for the first occurrence. This distinction illustrates that the division between prevention and treatment trials is not necessarily a sharp one. Furthermore, both types of trials may be designed to provide information on the same kind of measure, such as a risk reduction or relative hazard. Nonetheless, the following key differences are noteworthy. Patient populations in prevention trials tend to be more broadly defined, healthier, and experience lower event rates than those typical of pure treatment trials. As a consequence, these trials tend to be larger, and=or involve longer exposure durations. An added challenge may be faced in the definition and complexity of a suitable primary endpoint, which will invariably constitute a composite of numerous relevant components as opposed to a single clinical event type.
