RISE_FJ/Deidentification/consent_FJ-deidentify_data.R

library(digest)

rm(list = ls())

setwd("Z:/Data Files/Practice Data/FJ_consent")
#setwd("S:/R-MNHS-SPHPM-EPM-IDEpi/RISE/4. Surveys/3. Objectives/1. FJ/3/20190930_child sampling/3. Data/1. raw data")

consent <- read_csv (file="RISE_consent_FJ_2019_final.csv")
consent_form1 <- read_csv (file="RISE_consent_FJ_2019_final-consent_form1.csv")
consent_form3 <- read_csv (file="RISE_consent_FJ_2019_final-consent_form3.csv")
consent_childname <- read_csv (file="RISE_consent_FJ_2019_final-consent3_childname.csv")

temp_consent <- consent
temp_consent_form1 <- consent_form1
temp_consent_form3 <- consent_form3
temp_consent_childname <- consent_childname


hashed_id <- function(x, salt) {
  y <- paste(x, salt)
  y <- sapply(y, function(X) digest(X, algo="md5",serialize = FALSE))
  as.character(y)
}

###############################################################################
#
# # CONSENT SURVEY DE-IDENTIFICATION
#
###############################################################################

#de-identify names and phone numbers in consent survey - Jeff F
consent_names <- consent %>%
  select (devicephonenum, return_phone, name_surveyor_other, name_consent_barcode,
          return_name)

for(i in names(consent_names)) {
  temp_consent[[i]] <- ifelse(!is.na(consent_names[[i]]), hashed_id(consent_names[[i]], "RISE"), NA)
}

#de-identify any text fields in consent survey - Jeff F
consent_text <- consent %>%
  select (barcode_location_other)

for(i in names(consent_text)) {
  temp_consent[[i]] <- ifelse(!is.na(consent_text[[i]]), hashed_id(consent_text[[i]], "RISE"), NA)
}

#de-identify any house numbers/names in consent survey - Jeff F
consent_house <- consent %>%
  select (community:community_name, barcode_scan, barcode_scan_text,
          barcode_id, barcode_id_text, house_no_old2, house_no, instanceName)

for(i in names(consent_house)) {
  temp_consent[[i]] <- ifelse(!is.na(consent_house[[i]]), hashed_id(consent_house[[i]], "RISE"), NA)
}

write_csv(temp_consent, path = "Z:/Data Files/Practice Data/Deidentification/FJ_consent/deidentified_RISE_consent_FJ_2019_final.csv")

###############################################################################
#
# # CONSENT FORM 1 SURVEY DE-IDENTIFICATION
#
###############################################################################

#de-identify names and phone numbers in consent form 3 survey - Jeff F
consent_form1_names <- consent_form1 %>%
  select (hhd_head_name, phone_y_no)

for(i in names(consent_form1_names)) {
  temp_consent_form1[[i]] <- ifelse(!is.na(consent_form1_names[[i]]), hashed_id(consent_form1_names[[i]], "RISE"), NA)
}

write_csv(temp_consent_form1, path = "Z:/Data Files/Practice Data/Deidentification/FJ_consent/deidentified_RISE_consent_FJ_2019_final-consent_form1.csv")

###############################################################################
#
# # CONSENT FORM 3 SURVEY DE-IDENTIFICATION
#
###############################################################################

#de-identify names and phone numbers in consent form 3 survey - Jeff F
consent_form3_names <- consent_form3 %>%
  select (guardian_form3)

for(i in names(consent_form3_names)) {
  temp_consent_form3[[i]] <- ifelse(!is.na(consent_form3_names[[i]]), hashed_id(consent_form3_names[[i]], "RISE"), NA)
}

write_csv(temp_consent_form3, path = "Z:/Data Files/Practice Data/Deidentification/FJ_consent/deidentified_RISE_consent_FJ_2019_final-consent_form3.csv")

###############################################################################
#
# # CONSENT CHILD NAME SURVEY DE-IDENTIFICATION
#
###############################################################################

#de-identify names and phone numbers in consent child name survey - Jeff F
consent_childname_names <- consent_childname %>%
  select (child_name)

for(i in names(consent_childname_names)) {
  temp_consent_childname[[i]] <- ifelse(!is.na(consent_childname_names[[i]]), hashed_id(consent_childname_names[[i]], "RISE"), NA)
}

#de-identify dobs in consent child name survey - Jeff F
consent_childname_dobs <- consent_childname %>%
  select (dob)

for(i in names(consent_childname_dobs)) {
  temp_consent_childname[[i]] <- ifelse(!is.na(consent_childname_dobs[[i]]), hashed_id(consent_childname_dobs[[i]], "RISE"), NA)
}
write_csv(temp_consent_childname, path = "Z:/Data Files/Practice Data/Deidentification/FJ_consent/deidentified_RISE_consent_FJ_2019_final-consent3_childname.csv")
Monash-RISE/riseR documentation built on Dec. 11, 2019, 9:49 a.m.