Censoring data – left censoring

Left Censoring. Left censoring occurs where the event of interest occurs before the subject has entered the study (Antonisamy, Christopher & Samuel 2010, p. 215). Such could for instance occur where a follow-up evaluation, for instance to identify recurrence of a disease, reveals such event to have already occurred. In this respect, the actual failure will be a period between a procedure was carried out and the time of screening for recurrence, but the identification of the event (recurrence), is only made at the time of screening (Antonisamy, Christopher & Samuel 2010, p. 215; Klein & Moeschberger, 2003, p. 70).

Interval censoring. Interval censoring occurs where the event of interest is known to have occurred within a specified interval (e.g. the period from a certain year to a different year) but the exact time of occurrence is unknown (Klein & Moeschberger, 2003, p. 69). Such censoring is common where the study has periodic follow-ups, and hence survival time appears between two follow-ups but the exact time cannot be identified (Klein & Moeschberger, 2003, p. 71).

Informative censoring. Informative censoring occurs where the subjects sacrificed due to dropouts, are more likely to achieve the event of interest compared to the remaining subjects (Allison, 2010, pp. 13 – 14). Such censoring leads to bias. For instance, where the subjects censored are likely to have had long times to achieve the event of interest, informative censoring would result into the median survival rate for the group being underestimated (Allison, 2010). The misleading nature of informative censoring is especially heightened for random censoring, where an important requirement is that random censoring be non-informative (Allison, 2010, pp. 13-14).

Challenges presented by censoring during data analysis. Censoring presents various challenges for data analysis. One of these is that censoring results into bias, hence requiring additional assumptions to fill in data that may be missing due to such censoring (Shih 2002). Such assumptions are integrated into various imputation methods used to fill in data for subjects censored. The accuracy of the data analysis would thus be subject to the accuracy of the additional assumptions made for including the data from the censored observations.



Allison, P. D. (2010). Survival analysis using SAS: A practical guide (2nd ed., illustr.). Cary, NC: SAS Institute.

Antonisamy, B., Christopher, S., & Samuel, P. P. (2010). Biostatistics: Principles and practice. New Delhi: Tata McGraw-Hill Education.

Klein, J. P., & Moeschberger, M. L. (2003). Survival analysis: Techniques used for censored and truncated data (2nd ed., illustr. Repr.). New York: Springer.

Lee, E. T. & Wang, J. W. (2003). Statistical methods for survival data analysis (3rd ed., illustr.). Hoboken, NJ: John Wiley & Sons.

Shih, W. J. (2002). Problems in dealing with missing data and informative censoring in clinical trials. Current Controlled Trials in Cardiovascular Medicine, 3(1), 4 –10. doi:10.1186/1468-6708-3-4

find the cost of your paper