Period life tables from longitudinal data (cont.)

Left Censoring

Left censoring is necessary when a subject is known to be at risk before the period under study. At the time when that subject enters the study period the lifeline is left censored so that the period before the study is discarded while the time period (person-time) within the study is included.

An example might be a woman diagnosed with breast cancer at age 40 in 2005 but included in a survival study recruiting women from 2010 onwards. The woman's origin start date (diagnosis) would be 2005 but only her period of survival from 2010 would be included in the study. Her survival period has therefore been left censored at 2010 and her person-time of survival counted only from 2010. In an aggregate life table she would enter the table at year 5 after diagnosis.

Using a statistical package it is relatively easy to declare the time of entry to the study as the time when recording of her survival starts. This procedure is known as left censoring of the record.

Look at the diagram on the next page to fix this concept, and that of record splitting and period life tables, in your mind.