Existence and uniqueness of relative incidence estimates in case-series analysis

Case-series analysis is used to estimate relative incidences of clinical events in defined time intervals after vaccination compared to a control period. It has advantages, in terms of both a reduction in data collection effort, because it uses only data on cases, and a reduction in the resultant variances of estimates, due to individuals being self-controlled. The existence and uniqueness of relative incidence estimates in case-series analysis are investigated. For the relative incidence of a clinical event, a simple condition for existence and uniqueness of the estimate of the parameter vector in a case-series model is established. An algorithm is developed to examine the established condition, which provides a clue for remedy when the condition for existence and uniqueness is not satisfied.