`rm.terms` : hide specific levels of factor terms involving interaction effects?

[XueWenYan]

I know that when with rm.terms in the plot_models function, we can specifically remove terms, including specific levels of a factor, by using "rm.terms = varname [level name]". I wonder if there's a similar syntax when interaction effects between factors are involved.

For example,The terms in the graph below are automatically generated from sjplot, and for the last 4 interaction terms, I specified "race_ethnicity:sexuality_withNA" in the regression model, and the shown levels are automatically generated from computing. How can I tell sjplot to omit reporting the two circled rows (using rm.terms I imagine?) involving the NA levels in these interaction effects? I wasn't able to find instructions online.

For the NA levels for other main-effect variables in the plot, I'm able to get rid of them by specifying varname[level name].