A Model for Repeated Clustered Data with Informative Cluster Size Using Poisson Distribution

• Activity Number: 656
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Ana-Maria Iosif*+ and Allan Sampson
• Companies: University of California at Davis and University of Pittsburgh
• Keywords: clustered data ; informative cluster size ; repeated ; Poisson ; recurrent

Abstract

In many chronic conditions with recurrent events such as epilepsy or migraine headaches, both the number of events and the severity of each event tend to depend on the underlying condition severity. Data are usually collected over follow-up intervals, and at each assessment the outcome for an individual is in fact a cluster of severities. The size of the cluster is a random variable determined by the number of events experienced during that follow-up interval and might be informative about the condition severity. A change in the underlying condition severity is reflected in simultaneous changes in both the number of events experienced and their severities. In this paper, we introduce models to analyze clustered data gathered repeatedly on both the frequency of an event and its severity. We propose a general maximum likelihood framework using a shared parameter approach, with the Poisson distribution modeling the number of events, and derive estimators with good asymptotic properties. Our simulations under scenarios with sample sizes in a small to medium range show that the estimators also have good finite sample properties. We implement this method to a real life data set.