Is there a way to select cases based on matching criteria and a new event date >365 days from last?

Question

I have the following variables over 5 years. Name, DOB, DateDX, LabTest, LabResult

I need to deduplicate to count matching lab tests and results only once per patient within a 365 day period.

The last line of code deduplicates all of the cases counting each unique patient/LabTest/LabResult only once. But I need to add the >365 day time period based on DateDX.

I've looked into creating a loop or a function, but I find these difficult to think through. example I found looks something like this, but I'm lost in trying to adapt it.

# function for identifying dates. An 

cases<-function(x,lag=12){
  return(diff(log(x),lag))
}

j = 1

for(i in MyData){
  n = which(Last(pp1) == i)
  returnsmatrix[,j] =  rets(Last[,n],1)
  j=j+1
}

here is my code so far

library(readr)
library(data.table)
library(plyr)
library(dplyr)
library(dtplyr)
library(tidyr)
library(htmtools)

data <- group_by(data, TestResult)
datasummary <- summarise(data,
                    Test1=length(which(TestType=="1")),
                    Test2=length(which(TestType=="2")),
                    Test3=length(which(TestType=="3")),
                    Test4=length(which(TestType=="4"))
)

dedupdata <- data[!duplicated(data[,c("Name", "DOB", "LabTest", "LabResult")]),]

Can i just add a criteria that will look at the data of the row and select the nex match where DateDX => 365?

I expect to be able to get multiple TestType and TestResults back for a patient that will match but they need to be at least one year a part based on DateDX.

Do they really need to be at least one year apart, or can they be about one year apart? I ask because if a patient only comes in once per year, and always in the first week of February, odds are that some of those visits will be < 365 days since the last visit. You may end up removing records that you wanted to keep. — Joe, Aug 21 '19 at 03:04
For this particular case counting problem I need them to be at least 365 days apart. I struggle with this to, but it is in the case definition. Thanks! — kkaub, Aug 21 '19 at 16:04
do you have any example data (with expected outcome)? would something like this be helpful? https://stackoverflow.com/questions/39317354/how-to-filter-rows-based-on-difference-in-dates-between-rows-in-r — Ben, Aug 21 '19 at 19:59

Is there a way to select cases based on matching criteria and a new event date >365 days from last?

0 Answers0