Missing values in covariates

[yinglittxs]

Dear Professor Chris Jackson: I would like to ask whether, if there are missing values in the covariates, the model will simply remove them like in the Cox model or if it will adopt some other method. Will this affect the HR results? Is it true that not removing the missing values of the covariates should generate similar HR-estimated results as removing them?

[chjackson]

How msm deals with missing data is explained in help(msm), section "na.action". By default it drops records with missing data on variables necessary for the model you have requested - otherwise it would not be able to fit your model. I'm not able to advise whether dropping records with missing data is appropriate in any specific application though.

[yinglittxs]

If we do not write "na.action=na.omit" in the msm model, Is it by default "na.action=na.omit" (i.e. drops records with missing data on variables necessary for the model I have requested)

[chjackson]

Yes. In R function help pages, the default value for a function argument is shown in the "Usage" section.