Regression analysis of mixed recurrent-event and panel-count data with additive rate models

Biometrics. 2015 Mar;71(1):71-79. doi: 10.1111/biom.12247. Epub 2014 Oct 23.


Event-history studies of recurrent events are often conducted in fields such as demography, epidemiology, medicine, and social sciences (Cook and Lawless, 2007, The Statistical Analysis of Recurrent Events. New York: Springer-Verlag; Zhao et al., 2011, Test 20, 1-42). For such analysis, two types of data have been extensively investigated: recurrent-event data and panel-count data. However, in practice, one may face a third type of data, mixed recurrent-event and panel-count data or mixed event-history data. Such data occur if some study subjects are monitored or observed continuously and thus provide recurrent-event data, while the others are observed only at discrete times and hence give only panel-count data. A more general situation is that each subject is observed continuously over certain time periods but only at discrete times over other time periods. There exists little literature on the analysis of such mixed data except that published by Zhu et al. (2013, Statistics in Medicine 32, 1954-1963). In this article, we consider the regression analysis of mixed data using the additive rate model and develop some estimating equation-based approaches to estimate the regression parameters of interest. Both finite sample and asymptotic properties of the resulting estimators are established, and the numerical studies suggest that the proposed methodology works well for practical situations. The approach is applied to a Childhood Cancer Survivor Study that motivated this study.

Keywords: Additive rate model; Event-history studies; Mixed data; Panel-count data; Recurrent-event data.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Child
  • Data Interpretation, Statistical*
  • Female
  • Humans
  • Incidence
  • Male
  • Neoplasm Recurrence, Local / epidemiology*
  • Neoplasms / epidemiology*
  • Regression Analysis*
  • Survivors / statistics & numerical data*
  • United States / epidemiology