ABSTRACT
This chapter reviews some specialized topics involving interval-censored observations. Interval-censoring complicates matters especially for non- and semiparametric multistate models. A discrete survival model was fitted to these data by considering the interval-censored outcomes as grouped data. Rounding is a trivial example of interval censoring. Biomarker data are often subject to interval censoring because of limits of detection or quantification and regularly used as a regressor in all kinds of models. A parametric and a nonparametric approach for estimating the distribution of the interval-censored covariate were considered. Interval-censoring was dealt with in a classical parametric manner, while the marginal likelihood is computed with Gaussian quadrature implemented in the SAS procedure NLMIXED. The longitudinal analysis of caries experience, which is caries in the past or the present, can be done with a survival model whereby the response is interval-censored. Despite the availability of statistical software, interval-censored data are often not analyzed in an appropriate manner.
