Overview of Time-to-Event Endpoint Methodology

Document Type

Contribution to Book

Publication Date

4-23-2009

Publication Title

Design and Analysis of Clinical Trials with Time-to-Event Endpoints

ISBN

9781420066395

Abstract

In many clinical trials and other experimental settings, the primary endpoint is some critical event. In analgesic studies such as postdental extraction pain [1] or postpartum pain, the event is relief of pain. In an anti-infective study such as chronic urinary tract infection [2], the event is microbiological cure or abatement of clinical signs and symptoms of the infection. In oncology studies such as carcinoma of the breast [3], head and neck cancer [4], liver cancer [5], colorectal cancer [6], etc., the primary event is death. However, if the primary treatment is surgery which leaves the patient with no evidence of disease, another important event in patient follow-up is evidence of disease or recurrence [7]. In the study of bladder tumors, for example, important events in patient follow-up are remission or progression [8]. In transplantation studies involving renal allografts [9], although death is ‘‘primary,’’ rejection may be the primary event reflecting the experimental objective. Similarly, in long-term chronic toxicity studies of rodents, tumor development may directly reflect the experimental objective more so than death. In studies such as these, participants are usually followed for a specified length of time. Depending on the disease being studied, the frequency of follow-up may vary; e.g., daily, weekly, monthly, quarterly, biannually, or over unequally spaced intervals. Seldom if ever would follow-up be less frequent than quarterly. Follow-up on each participant permits the recording of the observed occurrence of the event or nonoccurrence of the event during the opportunity to observe. If the event is not observed to occur, it could be that the observation time was censored due to subject withdrawal prior to the end of the study (moved away from the area, died from causes unrelated to disease or treatment under study, adverse event unrelated to study conditions that prevented subject’s continued study participation, etc.) or due to the study ending. Figure 1.1 illustrates nine patients who entered a clinical trial where the event was observed on five patients and the event was censored on four patients. The left portion of the line for each patient represents the beginning of the observation time (time of randomization or time of beginning the treatment intervention) and the right endpoint represents the last time the status (event or censored) of the patient was known. Time 0 represents the calendar time the study started (with entry of the first patient). The fact that the left endpoints of the lines are staggered reflects that patients usually enter a clinical trial at different calendar times. Time-to-event data from Figure 1.1 available for analysis are the lengths of the patient lines and the event status. This effectively scales event times for all patients back to the origin.


Link to Full Text

Share

COinS