Definition: Censoring, in public health and biostatistics, refers to the situation where the exact time of an event of interest (e.g., disease onset, recovery, death) is not known for all study participants, but it is known that the event occurred either before or after a certain observation time or within a specific interval. It is a method for handling incomplete data, particularly in survival analysis.
Censoring is a fundamental concept in survival analysis, a branch of statistics used to analyze the duration of time until one or more events occur. It arises when researchers cannot observe the event of interest for every individual in a study. The most common type is right-censoring, where an individual’s event has not occurred by the end of the study period or they drop out (loss to follow-up), so their event time is known to be *at least* a certain duration. Less common are left-censoring, when an event happened before the observation period began, and interval-censoring, meaning the event occurred within a known time interval but the precise time is unknown. These situations are highly prevalent in longitudinal studies, clinical trials, and epidemiological cohort studies.
Ignoring censored data or treating it as if the event did not occur would lead to biased estimates of survival probabilities, incidence rates, and treatment effects, typically underestimating the true event times or overestimating survival. Therefore, specialized statistical methods, such as Kaplan-Meier survival curves and Cox proportional hazards models, are employed to appropriately account for censored observations. These methods allow researchers to utilize all available information, even from participants whose full follow-up is incomplete, thereby improving the accuracy, validity, and generalizability of public health research findings regarding disease prognosis, intervention efficacy, and risk factor assessment.
Key Context:
- Survival Analysis
- Loss to Follow-up
- Kaplan-Meier Estimator and Cox Proportional Hazards Model