On 2013 Jun 17, Mike Fay commented:

This paper shows an improved way to calculate the variance when using multiple imputation with interval censored data. Huang, Lee and Yu show this is better than calculating the variance by using a correction for multiple imputation developed by Rubin for a different situation, that was used in earlier papers.

Another advantage of this test (not studied in the original paper) is that it retains the type I error fairly well even when the assessment times depend on treatment (see my paper with Joanna Shih Fay MP, 2012).

Software to calculate this test is available in the interval R package, http://cran.r-project.org/web/packages/interval/index.html.

On 2013 Jun 17, Mike Fay commented:

This paper shows an improved way to calculate the variance when using multiple imputation with interval censored data. Huang, Lee and Yu show this is better than calculating the variance by using a correction for multiple imputation developed by Rubin for a different situation, that was used in earlier papers.

Another advantage of this test (not studied in the original paper) is that it retains the type I error fairly well even when the assessment times depend on treatment (see my paper with Joanna Shih Fay MP, 2012).

Software to calculate this test is available in the interval R package, http://cran.r-project.org/web/packages/interval/index.html.