id_hosp.R

gistfile1.txt

# Before running, activate env:

# export CONDA_ENVS_PATH=/nfs/projects/n/nsaph_common/conda/envs/
# export CONDA_PKGS_PATH=/nfs/projects/n/nsaph_common/conda/pkgs/
# source activate nsaph


## Code to ID hospitalizations

library(data.table)
library(fst)
library(devtools)
library(NSAPHutils)

library(lubridate)
#set_threads() # why this doesn't work?
library(icd)

# R version on RCE: 3.5.2

outcomes <- list()
outcomes[["all_kidney"]] <- list()
outcomes[["all_kidney"]][["icd9"]] <- expand_range("580","599")
outcomes[["all_kidney"]][["icd10"]] <- expand_range("N00","N39")
outcomes[["ckd"]] <- list()
outcomes[["ckd"]][["icd9"]] <- children("585")
outcomes[["ckd"]][["icd10"]] <- children("N18")
outcomes[["aki"]] <- list()
outcomes[["aki"]][["icd9"]] <- children("584")
outcomes[["aki"]][["icd10"]] <- children("N17")
outcomes[["glomerular"]][["icd9"]] <- expand_range("580", "583")
outcomes[["glomerular"]][["icd10"]] <- expand_range("N00", "N08")

# add diabetes 
outcomes[["diabetes"]] <- list()
outcomes[["diabetes"]][["icd9"]] <- children("250") 
outcomes[["diabetes"]][["icd10"]] <- c(children("E08"),children("E09"), children("E10"), children("E11"), children("E12"), children("E13") )
#expand_range("N08", "N13") 
#outcomes[["diabetes"]][["icd10"]] <- expand_range("E08", "E13") # <- this won't work, need to check ICD package

#add co-morbidities
#Circulatory system disease: ICD-9 390-459 / ICD-10 I00-I99
outcomes[["csd"]] <- list()
outcomes[["csd"]][["icd9"]] <- expand_range("390", "459") 
outcomes[["csd"]][["icd10"]] <- expand_range("I00", "I99") 
#Ischemic heart disease: ICD-9 410-414 / ICD-10 I20-I25
outcomes[["ihd"]] <- list()
outcomes[["ihd"]][["icd9"]] <- expand_range("410", "414") 
outcomes[["ihd"]][["icd10"]] <- expand_range("I20", "I25") 
#Pneumonia: ICD-9 480-486 / ICD-10 J12-J18
outcomes[["pneumonia"]] <- list()
outcomes[["pneumonia"]][["icd9"]] <- expand_range("480", "486") 
outcomes[["pneumonia"]][["icd10"]] <- expand_range("J12", "J18") 
#Heart failure: ICD-9 428/ ICD-10 I50
outcomes[["hf"]] <- list()
outcomes[["hf"]][["icd9"]] <- children("428") 
outcomes[["hf"]][["icd10"]] <- children("I50") 
#Acute myocardial infarction: ICD-9 410/ ICD-10 I21
outcomes[["ami"]] <- list()
outcomes[["ami"]][["icd9"]] <- children("410")  
outcomes[["ami"]][["icd10"]] <- children("I21")   
#Cerebrovascular diseases: ICD-9 430-438/ ICD-10 I60-I69
outcomes[["cerd"]] <- list()
outcomes[["cerd"]][["icd9"]] <- expand_range("430", "438")  
outcomes[["cerd"]][["icd10"]] <- expand_range("I60", "I69")   

#Urinary tract infection: ICD-9 599.0 / ICD-10 N39.0
outcomes[["uti"]] <- list()
outcomes[["uti"]][["icd9"]] <- children("599.0")  
outcomes[["uti"]][["icd10"]] <- children("N39.0")   


## clear out old data in case of re-run

for (outcome in names(outcomes)) {
  for (diag_type in c("primary", "secondary")) {
    file.remove(list.files(file.path("../output_data_test/", outcome, diag_type),
                           pattern = ".fst",
                           full.names = T))
  }
}

# Let's test for a single year (2000)
for (year_ in 2000:2000) {
admissions <- read_data("whanhee/data/hosp_data/", years = year_, columns = c("QID",
                                                           "ADATE",
                                                           "DDATE",
                                                           "DIAG1",
                                                           "DIAG2",
                                                           "DIAG3",
                                                           "DIAG4",
                                                           "DIAG5",
                                                           "DIAG6",
                                                           "DIAG7",
                                                           "DIAG8",
                                                           "DIAG9",
                                                           "DIAG10"))
}
print(admissions)
admissions[, ADATE := dmy(ADATE)]
admissions[, DDATE := dmy(DDATE)]
admissions[, year := year(ADATE)]
admissions <- admissions[year %in% 2000:2016]

for (outcome in names(outcomes)) {
  #primary only
  admissions[DDATE < "2015-10-01", (paste0(outcome, "_primary")) := DIAG1 %in% outcomes[[outcome]][["icd9"]]]
  admissions[DDATE >= "2015-10-01", (paste0(outcome, "_primary")) := DIAG1 %in% outcomes[[outcome]][["icd10"]]]
  #primary or secondary 
  admissions[DDATE < "2015-10-01", (paste0(outcome, "_primarysecondary")) := DIAG1 %in% outcomes[[outcome]][["icd9"]]|
               DIAG2 %in% outcomes[[outcome]][["icd9"]]|
               DIAG3 %in% outcomes[[outcome]][["icd9"]]|
               DIAG4 %in% outcomes[[outcome]][["icd9"]]|
               DIAG5 %in% outcomes[[outcome]][["icd9"]]|
               DIAG6 %in% outcomes[[outcome]][["icd9"]]|
               DIAG7 %in% outcomes[[outcome]][["icd9"]]|
               DIAG8 %in% outcomes[[outcome]][["icd9"]]|
               DIAG9 %in% outcomes[[outcome]][["icd9"]]|
               DIAG10 %in% outcomes[[outcome]][["icd9"]]]
  admissions[DDATE >= "2015-10-01", (paste0(outcome, "_primarysecondary")) := DIAG1 %in% outcomes[[outcome]][["icd10"]]|
               DIAG2 %in% outcomes[[outcome]][["icd10"]]|
               DIAG3 %in% outcomes[[outcome]][["icd10"]]|
               DIAG4 %in% outcomes[[outcome]][["icd10"]]|
               DIAG5 %in% outcomes[[outcome]][["icd10"]]|
               DIAG6 %in% outcomes[[outcome]][["icd10"]]|
               DIAG7 %in% outcomes[[outcome]][["icd10"]]|
               DIAG8 %in% outcomes[[outcome]][["icd10"]]|
               DIAG9 %in% outcomes[[outcome]][["icd10"]]|
               DIAG10 %in% outcomes[[outcome]][["icd10"]]]
  #secondary only
  admissions[DDATE < "2015-10-01", (paste0(outcome, "_secondary")) := DIAG2 %in% outcomes[[outcome]][["icd9"]]|
               DIAG3 %in% outcomes[[outcome]][["icd9"]]|
               DIAG4 %in% outcomes[[outcome]][["icd9"]]|
               DIAG5 %in% outcomes[[outcome]][["icd9"]]|
               DIAG6 %in% outcomes[[outcome]][["icd9"]]|
               DIAG7 %in% outcomes[[outcome]][["icd9"]]|
               DIAG8 %in% outcomes[[outcome]][["icd9"]]|
               DIAG9 %in% outcomes[[outcome]][["icd9"]]|
               DIAG10 %in% outcomes[[outcome]][["icd9"]]]
  admissions[DDATE >= "2015-10-01", (paste0(outcome, "_secondary")) := DIAG2 %in% outcomes[[outcome]][["icd10"]]|
               DIAG3 %in% outcomes[[outcome]][["icd10"]]|
               DIAG4 %in% outcomes[[outcome]][["icd10"]]|
               DIAG5 %in% outcomes[[outcome]][["icd10"]]|
               DIAG6 %in% outcomes[[outcome]][["icd10"]]|
               DIAG7 %in% outcomes[[outcome]][["icd10"]]|
               DIAG8 %in% outcomes[[outcome]][["icd10"]]|
               DIAG9 %in% outcomes[[outcome]][["icd10"]]|
               DIAG10 %in% outcomes[[outcome]][["icd10"]]]
  
}


#adding second ask, among those with secondary AKI which have the following 8 co-morbidities
#if AKI_secondaryONLY==1 (they were admitted for AKI DIAG 2-10), then indicators on what tehir DIAG1 is for same hospitalization
#among those, DIAG=1 cardiovasuclar, then add indicator had cardiovascular as primary during AKI hospitalization.. (etc for 10 conditions)
# 8 co-morbidities

#1. diabetes primary aki secondary
admissions$diabetesaki<-"FALSE"
admissions$diabetesaki[admissions$aki_secondary=="TRUE" & admissions$diabetes_primary=="TRUE"]<-"TRUE"

#2. circulatory system disease primary aki secondary
admissions$csdaki<-"FALSE"
admissions$csdaki[admissions$aki_secondary=="TRUE" & admissions$csd_primary=="TRUE"]<-"TRUE"

#3. ischemic heart disease primary aki secondary
admissions$ihdaki<-"FALSE"
admissions$ihdaki[admissions$aki_secondary=="TRUE" & admissions$ihd_primary=="TRUE"]<-"TRUE"

#4. pnemonia disease primary aki secondary
admissions$pneumoniaaki<-"FALSE"
admissions$pneumoniaaki[admissions$aki_secondary=="TRUE" & admissions$pneumonia_primary=="TRUE"]<-"TRUE"

#5. heart failure primary aki secondary
admissions$hfaki<-"FALSE"
admissions$hfaki[admissions$aki_secondary=="TRUE" & admissions$hf_primary=="TRUE"]<-"TRUE"

#6. acute myocardial infraction primary aki secondary
admissions$amiaki<-"FALSE"
admissions$amiaki[admissions$aki_secondary=="TRUE" & admissions$ami_primary=="TRUE"]<-"TRUE"

#7. cerberovascular disease  primary aki secondary
admissions$cerdaki<-"FALSE"
admissions$cerdaki[admissions$aki_secondary=="TRUE" & admissions$cerd_primary=="TRUE"]<-"TRUE"

#8. uti  primary aki secondary
admissions$utiaki<-"FALSE"
admissions$utiaki[admissions$aki_secondary=="TRUE" & admissions$uti_primary=="TRUE"]<-"TRUE"

# first ask: 
# this paper is on AKI, among those with AKI secondary (DIAG1-10) an 
# indicator if diabetes (DIAG1-10) or CKD (DIAG1-10) prior to AKI hospitalization
# indicator that if they had AKI_secondary (either before oct 2015 or after), 
# did they have DIABETES_secondary before AKI_secondary? 

# if AKI_secondary==1 (they were admitted for AKI primary or secondary), take date for first admission for AKI
# admissions$DDATE (for first admission AKI)> admissions$DDATE (for diaebtes primary or secondary)
# --> diabetes_prior_aki (indicator)
# want to add indicator that they had diabetes prior to AKI

#steps
#check if someone with same QID has one row with admissions$aki_primarysecondary=="TRUE" 
#and a different row with admissions$diabetes_primarysecondary=="TRUE", and admi_date of first row > admission date of second row
#add indicator admissions$diabeteshosp_prior_aki only for hospitalization corresponding to aki


#for testing add duplicated QID
admissions<-rbind(admissions, admissions[100,])
admissions[101]$DIAG1<-"N08"
admissions[101]$ADATE<-admissions$ADATE[100]-10
admissions[101]$diabetes_primarysecondary<-"TRUE"
admissions[101]$aki_primarysecondary<-"FALSE"
#identify indviduals with multiple admissions
mult_admissions_qid<-admissions$QID[which(table(admissions$QID)>1)]


#qid of those with diabetes prior to aki
akidiabetesqid<-c()
#for loop to see which of the ones with multiple admission have AKI and prior diabetes
for (i in unique(mult_admissions_qid)){
  hos_sub<-admissions[admissions$QID==i,] # program stuck here
  #check if they have AKI hospitalization 
  if (any(hos_sub$aki_primarysecondary)) {
    #check if also have diabetes hospitalization 
    if (any(hos_sub$diabetes_primarysecondary)) {
      #check if diabetes happened prior to aki
      #take first aki hospitalization
      dateaki<-min(hos_sub$ADATE[which(hos_sub$aki_primarysecondary=="TRUE")])
      #take first diabetes hospitalization 
      datediabetes<-min(hos_sub$ADATE[which(hos_sub$diabetes_primarysecondary=="TRUE")])
      
      if (datediabetes<dateaki) {
        akidiabetesqid<-c(akidiabetesqid, i)
      }
      
    }
  }
}
admissions$diabeteshosp_prior_aki<-"FALSE"
#add indicator only to the hospitalization with AKI to indicate prior diabetes
admissions$diabeteshosp_prior_aki[admissions$aki_primarysecondary=="TRUE" & admissions$QID %in% akidiabetesqid]<-"TRUE"

#ana please add same to define prior CKD

#print(admissions)

# declare diabetes_prior_aki TRUE if 
# aki_DDATE > diabetes_DDATE
#admissions <- transform(
# admissions, diabetes_prior_aki= ifelse(
#  admissions$aki_secondary == TRUE, 
# TRUE, 
#FALSE)
#)
#print(admissions)

#same for CKD, want to add inidcator that they had CKD prior to AKI

#total of 10 new columns

for (i in 2000:2000) {
    for (outcome in names(outcomes)) {
      for (type in c("primary", "secondary")) {
        varname <- paste0(outcome, "_", type)
        if (file.exists(paste0(file.path("../data/output_data_test2/", outcome, type),
                               "/", varname, "_", i, ".fst"))) {
          year_admissions <- read_fst(paste0(file.path("../data/output_data_test2/", outcome, type),
                                             "/", varname, "_", i, ".fst"))
        } else {
          year_admissions <- NULL
        }
        year_admissions <- rbind(year_admissions, admissions[year == i & get(varname) == T])
        
        if (nrow(year_admissions) != 0) {
          write_fst(year_admissions,paste0(file.path("../data/output_data_test2/", outcome, type),
                                           "/", varname, "_", i, ".fst"))
        }
      }
    }
}