A generalized log-rank test for interval-censored failure time data via multiple imputation.

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Trials

Background:

• Interval-censored (IC) data arise when failure times are known only within intervals, common in longitudinal studies.
• Comparing survival distributions with IC data presents unique statistical challenges.

Purpose of the Study:

• To propose a modified log-rank test for comparing two or more interval-censored samples.
• To evaluate the performance of the proposed test against existing methods.

Main Methods:

• A modified log-rank statistic and covariance matrix are computed using a multiple imputation technique.
• Simulation studies are conducted to assess test performance.
• The method is illustrated with a breast cancer patient dataset.

Main Results:

• The proposed log-rank test demonstrates performance comparable to Finkelstein's test.
• The new method outperforms existing log-rank type tests by Sun and Zhao and Sun.
• Performance differences are attributed to novel multiple imputation and covariance matrix estimation methods.

Conclusions:

• The proposed modified log-rank test offers a robust and effective approach for analyzing interval-censored failure time data.
• This method provides a valuable alternative for researchers dealing with periodic follow-up data.
• The technique is applicable to various clinical and epidemiological studies involving time-to-event data.