Time difference check in R-CodePudding

I have a DF_1 that shows the hospital admission date (date and time), the hospital discharge date (date and time) and whether the patient is a hospital readmission less than 30 days. Look:

ID <- c(111,222,222,333,444,444,555,666,1010,1010,1010)
PATIENT_ADMISSION <- c('18/03/2022 15:30','24/03/2022 12:28','27/03/2022 01:38','31/03/2022 08:53','16/04/2022 22:45','22/04/2022 13:15','05/04/2022 05:44','30/03/2022 06:16','10/01/2022 17:30','16/03/2022 22:00','08/04/2022 14:49')
PATIENT_DISCHARGE <- c('01/04/2022 11:20','26/03/2022 12:56','27/03/2022 17:52','01/04/2022 16:15','17/04/2022 12:26','25/04/2022 14:54','05/04/2022 11:44','07/04/2022 05:23','12/01/2022 06:35','06/04/2022 11:35','12/04/2022 12:36')
PATIENT_READMISSION_30D <- c('N','N','Y','N','N','Y','N','N','N','N','Y')
DF_1 <- data.frame(ID,PATIENT_ADMISSION,PATIENT_DISCHARGE,PATIENT_READMISSION_30D)

I want to include one more information in DF_1: I want to know if this readmission (PATIENT_READMISSION_30D = Y) happened within 72 hours. Thus, my DF_1 would have one more variable and would be presented as follows:

ID <- c(111,222,222,333,444,444,555,666,1010,1010,1010)
PATIENT_ADMISSION <- c('18/03/2022 15:30','24/03/2022 12:28','27/03/2022 01:38','31/03/2022 08:53','16/04/2022 22:45','22/04/2022 13:15','05/04/2022 05:44','30/03/2022 06:16','10/01/2022 17:30','16/03/2022 22:00','08/04/2022 14:49')
PATIENT_DISCHARGE <- c('01/04/2022 11:20','26/03/2022 12:56','27/03/2022 17:52','01/04/2022 16:15','17/04/2022 12:26','25/04/2022 14:54','05/04/2022 11:44','07/04/2022 05:23','12/01/2022 06:35','06/04/2022 11:35','12/04/2022 12:36')
PATIENT_READMISSION_30D <- c('N','N','Y','N','N','Y','N','N','N','N','Y')
PATIENT_READMISSION_72H <- c('','','Y','','','N','','','','','Y')
DF_1 <- data.frame(ID,PATIENT_ADMISSION,PATIENT_DISCHARGE,PATIENT_READMISSION_30D,PATIENT_READMISSION_72H)

Therefore, I would like to know how it is possible to check and include this new variable.

CodePudding user response：

You can use difftime to calculate the time difference in hours and dplyr::case_when to satisfy the conditions:

 # ensure proper format for dates
DF_1[2:3] <- lapply(DF_1[2:3], lubridate::dmy_hm)

#DF_1$test <- difftime(DF_1$PATIENT_DISCHARGE, DF_1$PATIENT_ADMISSION, units = "hours")

DF_1 %>% mutate(PATIENT_READMISSION_72H = case_when(
  PATIENT_READMISSION_30D == "N" ~ "",
  difftime(PATIENT_DISCHARGE, PATIENT_ADMISSION, units = "hours") <= 72 ~ "Y",
  difftime(PATIENT_DISCHARGE, PATIENT_ADMISSION, units = "hours") > 72 ~ "N"
))

Output:

#      ID   PATIENT_ADMISSION   PATIENT_DISCHARGE PATIENT_READMISSION_30D PATIENT_READMISSION_72H
# 1   111 2022-03-18 15:30:00 2022-04-01 11:20:00                       N                            
# 2   222 2022-03-24 12:28:00 2022-03-26 12:56:00                       N                            
# 3   222 2022-03-27 01:38:00 2022-03-27 17:52:00                       Y                           Y
# 4   333 2022-03-31 08:53:00 2022-04-01 16:15:00                       N                            
# 5   444 2022-04-16 22:45:00 2022-04-17 12:26:00                       N                            
# 6   444 2022-04-22 13:15:00 2022-04-25 14:54:00                       Y                           N
# 7   555 2022-04-05 05:44:00 2022-04-05 11:44:00                       N                            
# 8   666 2022-03-30 06:16:00 2022-04-07 05:23:00                       N                            
# 9  1010 2022-01-10 17:30:00 2022-01-12 06:35:00                       N                            
# 10 1010 2022-03-16 22:00:00 2022-04-06 11:35:00                       N                            
# 11 1010 2022-04-08 14:49:00 2022-04-12 12:36:00                       Y                           N

Note that line 11 differs from your desired output (a "N" instead of "Y") but that is because in the provided data, the values of PATIENT_DISCHARGE and PATIENT_ADMISSION are > 72 hours apart (93.78333 hours). If there is another reason why this should be "Y" please let me know and I will modify my answer. You can see the times if you un-comment the DF_1$test line in the provided code.

CodePudding user response：

This is a proposal, that uses {lubridate}:


DF_1 %>% 
  mutate(across(c(PATIENT_ADMISSION, PATIENT_DISCHARGE), 
                . %>% as_datetime(format = "%d/%m/%Y %H:%M"))) %>%
  group_by(ID) %>% 
  mutate(
    PATIENT_READMISSION_72H = 
      if_else(difftime(PATIENT_DISCHARGE, PATIENT_ADMISSION, units = "hours") <= 72, "Y", "N")
  ) %>% 
  ungroup() %>% 
  print(width=Inf)