Hi Seyit, welcome!
You have to use fct_drop() in combination with mutate()
library(dplyr)
library(forcats)
gss_cat <- data.frame(relig = c("Protestant", "Catholic", "None", "Hindu"))
gss_cat %>%
filter(relig %in% c("Protestant","Catholic","None")) %>%
mutate(relig = fct_drop(relig)) %>%
sjmisc::frq(relig, show.na = FALSE)
#>
#> # relig <categorical>
#> # total N=3 valid N=2 mean=2.00 sd=1.00
#>
#> val frq raw.prc valid.prc cum.prc
#> Catholic 1 33.33 33.33 33.33
#> None 1 33.33 33.33 66.67
#> Protestant 1 33.33 33.33 100.00
Created on 2019-03-27 by the reprex package (v0.2.1)