Abstract
Introduction: Interval censored failure time data periodically occur in medical and biological studies. The survival time of interest is observed only as it belongs to an interval rather than being exactly known.
Methods: In the past two decades many methods has been developed to analyze these data. We consider several tests, such as these three: nonparametric generalized log-rank test, a parametric score test and imputation method, compared to interval censored failure data. through a Monte Calro study.
Results: Simulation results indicate that the parametric test and the conventional imputation test outperform …show more content…
The survival time of interest is observed only as it belongs to an interval rather than being exactly known. For example, consider HIV infection. A person may take two examinations to assess their health. Therefore, if the first examination rejects the virus symptoms and signs of infection, further examinations are observed. After that time the virus enters the body and we are not informed about the exact time of entrance, but we are just aware about the interval which is after the first examination and before the second examination, thus producing an interval censored observation. Another example may be considered as breast cancer patients. These patients regularly dates the physician every 4 to 6 months for monitoring, even though they may not be present in some visits. The event here is the time to breast retraction. We only know that the event has occurred after the last visit and before the next one. Finally, these data are considered as interval censored. In fact, when we have some information about individual survival time, but we don’t know the time exactly, censoring …show more content…
However, for interval censored data analysis several techniques have been developed such as generalized log-rank test and several authors have discussed this problem. Peto and Peto() proposed a method for comparing two groups under Lehmann-type alternatives. In their approach interval censored data and exact observations are allowed to be used. In this case, they reduce a score test for interval censored data which is reffered to as log-rank test. Finklstein() used a regression method under the proportional hazard model and developed a parametric score test when the covariates are treatment indicators. Her approach allows K-Sample treatment comparisons. Sun() developed a nonparametric test without assuming the proportional hazard model that were improved by Zhao and Sun(). In this approach data are to be used in both types of censored and exact observation. Sun et al() proposed a new class of generalized log-rank tests that would not allow exact observations to attend. Zhao et al() modified the so-called generalized log-rank test which has been studied by Sun et al() to allow the presence of exactly observed