Hi all,
I have a question about multiple imputation within the MICE package. I want to
use passive imputation for my variable called X, because it is calculated out
of multiple variables, namely Y, Z. Let's give an example with BMI. I know,
that if I want to use passive imputation for BMI, I can use the following
command:
meth["BMI"] <- "~I(weight/(height/100)^2)"
pred[c("weight", "length"), "BMI"] <- 0
imp <- mice(Data, method = meth, predictorMatrix = pred, maxit = 10, m = 5)
But what If the formula for a variable is much more complicated, like with
CKD-epi. The formula is:
egfr_crea_ckdepi = function(creatinine, age, is_female, is_male) {
if (length(creatinine) != length(age) ||
length(creatinine) != length(is_female) ||
length(creatinine) != length(is_black)) {
stop("input vector lengths must match!")
}
a = ifelse(!is_black, ifelse(!is_female, 141, 144), ifelse(!is_female(163,
166)))
b = ifelse(!is_female, 0.9, 0.7)
c = ifelse(!is_female, ifelse(creatinine <= 0.9, -0.411, -1.209),
ifelse(creatinine <= 0.7, -0.329, -1.209))
a * ((creatinine / b) ^ c) * (0.993 ^ age)
}
How can I then use this function for the passive imputation of CKD-epi? (My
variable is a different variable, but it has the same principle). I have tried
to do:
meth["CKD-epi"] <- "~I(egfr_crea_ckdepi)"
But that does not work. Using the whole function as written above also does not
work. Does anybody know how I can solve this issue?
Thank you in advance.
Best,
Lisa
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
