|
Clinical data on the prognosis of diseases with a clinical history marked by exacerbations and remissions are best derived from a longitudinal study of a group of patients from the onset of disease. For instance, in Systemic Lupus Erythmatouses (SLE), an autoimmune disease, the clinical pivotal issues are correlated with the transitions between exacerbations (aggravation) and remissions (free of the disease condition). In other words, patients suffering from SLE may have a sudden change in the disease activity when there is a worsening of the disease condition or the patient may be free of the disease condition for some period of time which is usually not known. The usual survival analyses or incidence rates by subgroups will not address the impact of transitions [11]. They are restricted by assumptions of non-informative censoring and limit the description of disease to permanent transition from one state (alive) to another (dead). Many clinical studies involve complex changes other than death, for example, relapse, recurrence, recovery, and remission. Investigators will force this complexity to fit with Kaplan–Meier methods by evaluating time to recurrence or relapse. This fragments the investigation and if explored out of context with other outcomes, results in incomplete evaluation of the data. When a person is in relapse, this time often offers some information on the time of death as relapse may lead to overwhelming infection or kidney failure when the patient may eventually die. Relapse also represents a competing risk. The objective for most competing risks is to estimate the time of failure from a particular cause when other causes of failure are not in effect. Thus, two limitations of conventional survival analysis methods currently used for investigation of clinical studies become apparent; (1) they are inadequate to describe the complexity of disease beyond two simple states of alive or dead (or some isolated intermediate endpoint), and (2) they are inadequate to handle the complicated issue of observations.
|