how to show an event happened between two dates in R

问题

My question seems simple and I hope it is.

I have a dataframe that has the date of diagnosis of a disease, a binary variable indicating which drug the patient was on (or exposed and unexposed group), a start and stop date for the drug, and an overall stop date.

ID    Diag_date    Treatment    End.date    Drug.start    drug.end
1        NA           0         15/03/2002   01/01/2002    01/02/2002
1        NA           1         15/03/2002   01/02/2002    01/03/2002
1        NA           0         15/03/2002   01/03/2002    NA
2    01/04/2002       1         01/05/2002   01/01/2015    01/02/2002
2    01/04/2002       0         01/05/2002   01/02/2002    01/03/2002
2    01/04/2002       0         01/05/2002   01/03/2002    NA

As you can see the date of diagnosis is non time-varying, but the drug start and stop dates are.

Preferably I want an answer to two questions:

1.) How do i transfer the overall End.date to the final drug.end for each ID?

2.) How do I create a binary column that shows if the diagnosis date occurs in the interval between Drug.start and Drug.end?

I wish my final data to look like the following:

ID    Diag_date    Treatment    End.Date     Drug.start    Drug.end    Event
1        NA           0         15/03/2002   01/01/2002    01/02/2002   0
1        NA           1         15/03/2002   01/02/2002    01/03/2002   0
1        NA           0         15/03/2002   01/03/2002    15/03/2002   0 
2    01/04/2002       1         01/05/2002   01/01/2015    01/02/2002   0
2    01/04/2002       0         01/05/2002   01/02/2002    01/03/2002   0
2    01/04/2002       0         01/05/2002   01/03/2002    01/05/2002   1

Not everyone has a diagnosis date because not everyone in the sample had the disease. The code I wrote is the following:

 for (i in 1:nrow(df)) {
   if ((df$Diag_date[i] >= df$Drug.start[i]) && ( df$Diag_date[i] <= df$Drug.stop[i])) {
df$Event[i] <- 1
} else {
df$Event[i] <- 0
  }
}

the error i get when I run this code is:

missing value where TRUE/FALSE needed

Any help would be much appreciated.

回答1:

You can try

library(dplyr)
df1 %>%
    mutate_each(funs(as.Date(., '%d/%m/%Y')), matches('start|end|date')) %>%
     mutate(drug.end= as.Date(ifelse(is.na(drug.end), End.date, 
                         drug.end),origin='1970-01-01'),
    Event= as.integer((Diag_date >= Drug.start & Diag_date<=drug.end) & 
                        !is.na(Diag_date))) #%>%
    #mutate_each(funs(format(., '%d/%m/%Y')), matches('start|end|date'))
 #    ID  Diag_date Treatment   End.date Drug.start   drug.end Event
 #1  1       <NA>         0 2002-03-15 2002-01-01 2002-02-01     0
 #2  1       <NA>         1 2002-03-15 2002-02-01 2002-03-01     0
 #3  1       <NA>         0 2002-03-15 2002-03-01 2002-03-15     0
 #4  2 2002-04-01         1 2002-05-01 2015-01-01 2002-02-01     0
 #5  2 2002-04-01         0 2002-05-01 2002-02-01 2002-03-01     0
 #6  2 2002-04-01         0 2002-05-01 2002-03-01 2002-05-01     1

As @David Arenburg mentioned, it is better to keep the 'date' columns as the 'Date' class. If you need it in the 'character' format, just uncomment the last line and run it.

NOTE: Removed the group_by as it was not needed

回答2:

Her's a possible data.table equivalent

library(data.table)

# Converting to dates
Dates <- names(df)[c(2, 4:6)]
setDT(df)[, (Dates) := lapply(.SD, as.IDate, format = "%d/%m/%Y"), .SDcols = Dates] 

# First question
df[is.na(drug.end), drug.end := End.date] 

# Second question
df[Diag_date >= Drug.start & Diag_date <= drug.end, Event := 1L]

回答3:

Akrun's answer suffices for the problem in hand. Suggesting a more straight forward code.

A <- read.table("clipboard", header = T)
Dates <- c("Diag_date", "End.date", "Drug.start", "drug.end")


A[,Dates] <- lapply(A[,Dates],function(x) as.Date(x, format = "%d/%m/%Y"))
A$drug.end[is.na(A$drug.end)] <- as.character(A$End.date[is.na(A$drug.end)])
A$Event <- as.numeric((A$Diag_date >= A$Drug.start & A$Diag_date<=A$drug.end) & !is.na(A$Diag_date))

来源：https://stackoverflow.com/questions/31262707/how-to-show-an-event-happened-between-two-dates-in-r

标签

date

intervals