diff --git a/03_INTERVENTIONAL_M0/INTERVENTIONAL_M0_02_QC_01_data_integrity.R b/03_INTERVENTIONAL_M0/INTERVENTIONAL_M0_02_QC_01_data_integrity.R
new file mode 100644
index 0000000000000000000000000000000000000000..33345d8620b40c1360b91d0104e2c8c16279f376
--- /dev/null
+++ b/03_INTERVENTIONAL_M0/INTERVENTIONAL_M0_02_QC_01_data_integrity.R
@@ -0,0 +1,676 @@
+## INTERVENTIONAL_02_QC_01_curated_data_integrity.R
+## Date : 2025/05/19
+## Author : Thomas Obadia
+##
+## This script processes the raw (so far) dataset that follows a data
+## dump by INTERVENTIONAL_M0_01_dump_REDCap_database.R and generates
+## a set of data quality indicators.
+##
+## The first section of this script will subset the base dataset into
+## multiple, smaller versions that will be used for QC on different
+## aspects:
+## - Screening, eligibility and content
+## - Allocation of unique Participant ID
+## - Demographics
+## - To be continued...
+######################################################################
+
+
+
+
+
+######################################################################
+### SOURCE THE DATA
+######################################################################
+source("./03_INTERVENTIONAL_M0/INTERVENTIONAL_M0_01_dump_REDCap_database.R")
+
+
+
+
+
+######################################################################
+### DATA SUBSET
+######################################################################
+## Data subset for QC-ing the screening & eligibility process
+DAT_QC_SCREENING_ELIGIBILITY_CONSENT <- dat_interventional_m0_raw %>%
+ filter(redcap_event_name %in% c("Baseline Survey (M0) (Arm 1: Intervention)",
+ "1")) %>%
+ select(record_id,
+
+ redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance,
+
+ country, clusterid, hid, subjid,
+
+ eligibility_chk_01, eligibility_chk_02, eligibility_chk_03, consent,
+ eligibility_chk_sum,
+
+ uniqueid, uniqueid2,
+
+ enroldat)
+
+
+## Data subset for process of allocating a unique Participant ID
+DAT_QC_ALLOCATING_PARTICIPANT_ID <- DAT_QC_SCREENING_ELIGIBILITY_CONSENT %>%
+ filter(eligibility_chk_sum == 4)
+
+
+## Data subset for participant demographics
+# All rows are preserved, but further filtering will happen when calculating
+# the QC flags to retain only entries for participants who were properly
+# enrolled in the study.
+DAT_QC_DEMOGRAPHICS <- dat_interventional_m0_raw %>%
+ select(record_id,
+
+ redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance,
+
+ country, clusterid, hid, subjid, uniqueid,
+
+ sex,
+ dob_yn, dob, agey_estimate, agem_estimate, agey)
+
+
+## Data subset for malaria diagnostics and sample collection
+# All rows are preserved, but further filtering will happen when calculating
+# the QC flags to retain only entries for participants who were properly
+# enrolled in the study.
+DAT_QC_MALARIA_DIAG_AND_SAMPLE <- dat_interventional_m0_raw %>%
+ select(record_id,
+
+ redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance,
+
+ country, clusterid, hid, subjid, uniqueid,
+
+ bednet_yn, bednet_uselastn_yn,
+ malaria_last6m_yn, malaria_last6m_diag,
+ malaria_drugs_last6m_yn, malaria_drugs_last6m_datdd, malaria_drugs_last6m_datmm, malaria_drugs_last6m_datyyyy,
+
+ vs_fever_last2d_bls_yn, vs_temp_bls,
+
+ lb_rdt_done_bls_yn, lb_rdt_result_bls_e, lb_rdt_result_bls_m,
+ referred_bls_yn, referred_name_bls,
+ treated_bls_yn, starts_with("treated_name_bls___"), treated_name_other_bls,
+
+ starts_with("lb_bloodsample_collected_bls___"), lb_bloodsample_notcollected_det,
+ lb_samplecode_bls, lb_samplecode_bls_checked
+ )
+
+
+
+
+
+######################################################################
+### PARTICIPANT SCREENING + ELIGIBILITY & CONSENT
+######################################################################
+### Here we derive row-wise flags that will identify data inconsistencies that
+### require clarification from the field team.
+
+## Table with QC queries
+QC_SCREENING_ELIGIBILITY_CONSENT <- DAT_QC_SCREENING_ELIGIBILITY_CONSENT %>%
+ rowwise() %>%
+ mutate(
+ # Missing or incorrect country code
+ QC_COUNTRY_IS_MISSING_OR_INCORRECT = case_when(is.na(country) ~ TRUE,
+ !(country %in% c("E", "Ethiopia",
+ "M", "Madagascar", "Madagasikara")) ~ TRUE,
+ .default = FALSE),
+
+ # Missing Cluster ID or Cluster ID not part of those randomized
+ QC_CLUSTERID_IS_MISSING_OR_INCORRECT = case_when(is.na(clusterid) ~ TRUE,
+ !(
+ (country %in% c("E", "Ethiopia") &
+ clusterid %in% c(NA)) |
+ (country %in% c("M", "Madagascar", "Madagasikara") &
+ clusterid %in% (PVSTATEM_RANDOMIZATION %>% filter(country == "M") %>% pull(cluster)))) ~ TRUE,
+ .default = FALSE),
+
+ # Missing or incorrect House ID
+ QC_HOUSEID_IS_MISSING_OR_INCORRECT = case_when(is.na(hid) ~ TRUE,
+ !grepl(pattern = "^H?\\d{3}$",
+ x = hid) ~ TRUE,
+ .default = FALSE),
+
+ # Consent missing but should have been documented
+ QC_CONSENT_IS_MISSING_BUT_PATIENT_ELIGIBLE = case_when(is.na(consent) &
+ eligibility_chk_01 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_02 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_03 %in% REDCAP_LABELS_RADIO_YES ~ TRUE,
+ .default = FALSE),
+
+ # Consent obtained when it should not have been
+ QC_CONSENT_OBTAINED_BUT_PATIENT_NOT_ELIGIBLE = case_when(consent %in% REDCAP_LABELS_RADIO_YES & eligibility_chk_sum != 4 ~ TRUE,
+ .default = FALSE),
+
+ # Participant consented but Participant ID is missing in at least
+ # one of the two fields
+ QC_PATIENT_CONSENTED_BUT_UNIQUEID_IS_MISSING = case_when(consent %in% REDCAP_LABELS_RADIO_YES & (is.na(uniqueid) | is.na(uniqueid2)) ~ TRUE,
+ consent %in% REDCAP_LABELS_RADIO_YES & !is.na(uniqueid) & !is.na(uniqueid2) ~ FALSE,
+ .default = NA),
+
+ # Participant consented but Participant ID is not identical across the two
+ # uniqueid fields
+ QC_PATIENT_CONSENTED_BUT_UNIQUEID_IS_MISMATCH = case_when(consent %in% REDCAP_LABELS_RADIO_YES & uniqueid != uniqueid2 ~ TRUE,
+ consent %in% REDCAP_LABELS_RADIO_YES & uniqueid == uniqueid2 ~ FALSE,
+ .default = NA),
+
+ # Participant was properly enrolled, allocated an ID, but date of
+ # enrollment is missing or before 2025-05-01
+ QC_ENROLDAT_IS_MISSING_OR_INCORRECT = case_when(uniqueid == uniqueid2 & (is.na(enroldat) | as.Date(enroldat) < "2025-05-01") ~ TRUE,
+ uniqueid == uniqueid2 & !is.na(enroldat) & as.Date(enroldat) >= "2025-05-01" ~ FALSE,
+ .default = NA)
+ )
+
+
+## Number of participants screened, found eligible, and possibly enrolled
+DAT_QC_SCREENING_ELIGIBILITY_CONSENT %>%
+ mutate(screened = TRUE,
+ eligible = case_when(eligibility_chk_01 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_02 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_03 %in% REDCAP_LABELS_RADIO_YES ~ TRUE,
+ (!is.na(eligibility_chk_01) &
+ !is.na(eligibility_chk_02) &
+ !is.na(eligibility_chk_03) &
+ any(!eligibility_chk_01 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_02 %in% REDCAP_LABELS_RADIO_YES &
+ eligibility_chk_03 %in% REDCAP_LABELS_RADIO_YES)) ~ FALSE,
+ .default = NA),
+ consent = case_when(consent %in% REDCAP_LABELS_RADIO_YES ~ TRUE,
+ consent %in% REDCAP_LABELS_RADIO_NO ~ FALSE,
+ .default = NA),
+ enrolled_ok = case_when(consent & uniqueid == uniqueid2 ~ TRUE,
+ .default = NA)) %>%
+ group_by(country, clusterid) %>%
+ summarize(n_screened = sum(screened, na.rm = TRUE),
+ n_eligible = sum(eligible, na.rm = TRUE),
+ n_eligible_NA = sum(is.na(eligible)),
+ n_consent = sum(consent, na.rm = TRUE),
+ n_consent_NA = sum(is.na(consent)),
+ n_enrolled_ok = sum(enrolled_ok, na.rm = TRUE))
+
+
+
+
+
+######################################################################
+### ALLOCATION OF PARTICIPANT ID
+######################################################################
+### This section will run quality routines on the allocation of unique
+### participant identifiers, therefore covering the set of participants that
+### were deemed eligible and should theorietically have received a
+### Participant ID.
+
+## Table with QC queries
+DAT_QC_ALLOCATING_PARTICIPANT_ID <- DAT_QC_ALLOCATING_PARTICIPANT_ID %>%
+ rowwise() %>%
+ mutate(
+ # Record ID does not match the expected structure
+ QC_RECORD_ID_HAS_INCORRECT_STRUCTURE = case_when(!grepl(pattern = "^P-\\d{5}$",
+ x = record_id) ~ TRUE,
+ grepl(pattern = "^P-\\d{5}$",
+ x = record_id) ~ FALSE,
+ .default = NA),
+
+ # UniqueID is missing in one or more of the two fields
+ QC_UNIQUEID_OR_UNIQUEID2_IS_MISSING = case_when(is.na(uniqueid) | is.na(uniqueid2) ~ TRUE,
+ !is.na(uniqueid) & !is.na(uniqueid2) ~ FALSE,
+ .default = NA),
+
+ # UniqueID has incorrect structure in one or more of the two fields
+ QC_UNIQUEID_OR_UNIQUEID2_HAS_INCORRECT_STRUCTURE = case_when(!grepl(pattern = paste0("^",
+ "(?:",
+ paste0(PVSTATEM_RANDOMIZATION$country,
+ "-",
+ PVSTATEM_RANDOMIZATION$cluster,
+ collapse = "|"),
+ ")",
+ "-",
+ "\\d{3}",
+ "$"),
+ x = uniqueid) &
+ !grepl(pattern = paste0("^",
+ "(?:",
+ paste0(PVSTATEM_RANDOMIZATION$country,
+ "-",
+ PVSTATEM_RANDOMIZATION$cluster,
+ collapse = "|"),
+ ")",
+ "-",
+ "\\d{3}",
+ "$"),
+ x = uniqueid2) ~ TRUE,
+ grepl(pattern = paste0("^",
+ "(?:",
+ paste0(PVSTATEM_RANDOMIZATION$country,
+ "-",
+ PVSTATEM_RANDOMIZATION$cluster,
+ collapse = "|"),
+ ")",
+ "-",
+ "\\d{3}",
+ "$"),
+ x = uniqueid) &
+ grepl(pattern = paste0("^",
+ "(?:",
+ paste0(PVSTATEM_RANDOMIZATION$country,
+ "-",
+ PVSTATEM_RANDOMIZATION$cluster,
+ collapse = "|"),
+ ")",
+ "-",
+ "\\d{3}",
+ "$"),
+ x = uniqueid2) ~ FALSE,
+ .default = NA),
+
+ # UniqueID has incorrect structure in one or more of the two fields
+ QC_UNIQUEID_DOES_NOT_MATCH_UNIQUEID2 = case_when(uniqueid != uniqueid2 ~ TRUE,
+ uniqueid == uniqueid2 ~ FALSE,
+ .default = NA)
+ ) %>%
+ ungroup %>%
+
+ # QC aggregated by a specific factor, to check uniqueness
+ group_by(uniqueid) %>%
+ mutate(
+ # UniqueID is not unique across enrolled participants
+ QC_UNIQUEID_IS_NOT_UNIQUE = n() > 1
+ ) %>%
+ ungroup %>%
+ # QC aggregated by a specific factor, to check uniqueness
+ group_by(uniqueid2) %>%
+ mutate(
+ # UniqueID2 is not unique across enrolled participants
+ QC_UNIQUEID2_IS_NOT_UNIQUE = n() > 1
+ ) %>%
+ ungroup
+
+
+
+
+
+######################################################################
+### PARTICIPANT DEMOGRAPHICS
+######################################################################
+### This section will run quality routines on demographcis for participant
+### that passed screening and were enrolled into the study.
+###
+### For now, the data used as input is *not* restricted to the subset of
+### participants who were allocated a proper, unique Participant ID.
+### This table is furtehr restricted to participants who have been allocated
+### a Participant ID, *BUT NOT TO THOSE WITH A VALIDATED, UNIQUE PARTICIPANT ID*.
+###
+### This allows for looking at demographics for all participants enrolled,
+### although some of them may have ID-related QC to solved in the first place.
+## Table with QC queries
+DAT_QC_DEMOGRAPHICS <- QC_SCREENING_ELIGIBILITY_CONSENT %>%
+ select(record_id,
+
+ redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance,
+
+ enroldat) %>%
+ right_join(DAT_QC_DEMOGRAPHICS) %>%
+ filter(!is.na(uniqueid)) %>%
+ rowwise() %>%
+ mutate(
+ # Sex is missing
+ QC_SEX_IS_MISSING = case_when(is.na(sex) ~ TRUE,
+ !is.na(sex) ~ FALSE,
+ .default = NA),
+
+ # DOB is unknown and self-reported age is not available
+ QC_DOB_UNKNOWN_BUT_ESTIMATED_AGE_IS_MISSING = case_when(dob_yn %in% REDCAP_LABELS_RADIO_NO & is.na(agey_estimate) ~ TRUE,
+ dob_yn %in% REDCAP_LABELS_RADIO_NO & !is.na(agey_estimate) ~ FALSE,
+ .default = NA),
+
+ # DOB is presumably known but content is empty or invalid
+ QC_DOB_KNOWN_BUT_DOB_IS_MISSING_OR_INCORRECT = case_when(dob_yn %in% REDCAP_LABELS_RADIO_YES &
+ (is.na(dob) | !grepl(pattern = "^(((\\d{2}([13579][26]|[2468][048]|04|08)|(1600|2[048]00))([-\\/])02(\\6)29)|(\\d{4}([-\\/])((0[1-9]|1[012])(\\9)(0[1-9]|1\\d|2[0-8])|((0[13-9]|1[012])(\\9)(29|30))|((0[13578]|1[02])(\\9)31))))$",
+ x = dob)) ~ TRUE,
+ dob_yn %in% REDCAP_LABELS_RADIO_YES &
+ !is.na(dob) & grepl(pattern = "^(((\\d{2}([13579][26]|[2468][048]|04|08)|(1600|2[048]00))([-\\/])02(\\6)29)|(\\d{4}([-\\/])((0[1-9]|1[012])(\\9)(0[1-9]|1\\d|2[0-8])|((0[13-9]|1[012])(\\9)(29|30))|((0[13578]|1[02])(\\9)31))))$",
+ x = dob) ~ FALSE,
+ .default = NA),
+
+ # DOB is self-reported by months are missing in young child
+ QC_AGEM_IS_MISSING = case_when(!is.na(agey_estimate) & agey_estimate < 5 & is.na(agem_estimate) ~ TRUE,
+ !is.na(agey_estimate) & agey_estimate >= 5 ~ FALSE,
+ .default = NA),
+
+ # Age is not available
+ QC_AGEY_IS_MISSING = case_when(is.na(agey) ~ TRUE,
+ !is.na(agey) ~ FALSE,
+ .default = NA),
+
+ # Age is available but too high or too low
+ QC_AGEY_IS_OUT_OF_BOUNDS = case_when(!is.na(agey) & (agey >= 100 | agey < 1) ~ TRUE,
+ !is.na(agey) & agey < 100 & agey >= 1 ~ FALSE,
+ .default = NA),
+
+ # Age is available but does not correspond to known DOB
+ QC_AGEY_IS_INCORRECT = case_when(!is.na(dob) & agey != round(as.numeric(difftime(time1 = enroldat,
+ time2 = dob,
+ units = "days")),
+ digits = 1) ~ TRUE,
+ !is.na(dob) & agey == round(as.numeric(difftime(time1 = enroldat,
+ time2 = dob,
+ units = "days")),
+ digits = 1) ~ FALSE,
+ .default = NA)
+ )
+
+
+
+
+
+######################################################################
+### MALARIA DIAGNOSTIC & SAMPLE COLLECTION
+######################################################################
+DAT_QC_MALARIA_DIAG_AND_SAMPLE <- QC_SCREENING_ELIGIBILITY_CONSENT %>%
+ select(record_id,
+
+ redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance) %>%
+ right_join(DAT_QC_MALARIA_DIAG_AND_SAMPLE) %>%
+ filter(!is.na(uniqueid)) %>%
+ rowwise() %>%
+ mutate(
+ # Availability of bednet is missing
+ QC_BEDNET_AVAILABILITY_IS_MISSING = case_when(is.na(bednet_yn) ~ TRUE,
+ !is.na(bednet_yn) ~ FALSE,
+ .default = NA),
+
+ # Bednet is available but participant dit not report on usage last night
+ QC_BEDNET_AVAILABLE_BUT_LAST_USAGE_IS_MISSING = case_when(bednet_yn %in% REDCAP_LABELS_RADIO_YES & is.na(bednet_uselastn_yn) ~ TRUE,
+ bednet_yn %in% REDCAP_LABELS_RADIO_YES & !is.na(bednet_uselastn_yn) ~ FALSE,
+ .default = NA),
+
+ # Report of a malaria episode over the past 6 months is missing
+ QC_MALARIA_REPORT_OVER_PAST_6_MONTHS_IS_MISSING = case_when(is.na(malaria_last6m_yn) ~ TRUE,
+ !is.na(malaria_last6m_yn) ~ FALSE,
+ .default = NA),
+
+ # Patient had malaria over the past 6 months but information about
+ # diagnostic is missing
+ QC_MALARIA_REPORTED_OVER_PAST_6_MONTHS_BUT_DIAG_IS_MISSING = case_when(malaria_last6m_yn %in% REDCAP_LABELS_RADIO_YES & is.na(malaria_last6m_diag) ~ TRUE,
+ malaria_last6m_yn %in% REDCAP_LABELS_RADIO_YES & !is.na(malaria_last6m_diag) ~ FALSE,
+ .default = NA),
+
+ # Patient had malaria over the past 6 months but did not report on any date
+ QC_MALARIA_REPORTED_OVER_PAST_6_MONTHS_BUT_DATE_IS_MISSING = case_when(malaria_last6m_yn %in% REDCAP_LABELS_RADIO_YES &
+ is.na(malaria_drugs_last6m_datdd) &
+ is.na(malaria_drugs_last6m_datmm) &
+ is.na(malaria_drugs_last6m_datyyyy) ~ TRUE,
+ malaria_last6m_yn %in% REDCAP_LABELS_RADIO_YES &
+ (!is.na(malaria_drugs_last6m_datdd) |
+ !is.na(malaria_drugs_last6m_datmm) |
+ !is.na(malaria_drugs_last6m_datyyyy)) ~ FALSE,
+ .default = NA),
+
+ # Patient did not report on possible fever in the preceding 2 days
+ QC_SELF_REPORTED_FEVER_IS_MISSING = case_when(is.na(vs_fever_last2d_bls_yn) ~ TRUE,
+ !is.na(vs_fever_last2d_bls_yn) ~ FALSE,
+ .default = NA),
+
+ # Patient temperature is missing
+ QC_TEMPERATURE_IS_MISSING = case_when(is.na(vs_temp_bls) | is.na(as.numeric(vs_temp_bls)) ~ TRUE,
+ !is.na(vs_temp_bls) & !is.na(as.numeric(vs_temp_bls)) ~ FALSE,
+ .default = NA),
+
+ # Patient temperature is missing
+ QC_TEMPERATURE_IS_OUT_OF_BOUNDS = case_when(vs_temp_bls < 33 | vs_temp_bls > 42 ~ TRUE,
+ vs_temp_bls >= 33 & vs_temp_bls <= 42 ~ FALSE,
+ .default = NA),
+
+ # Malaria RDT was not done or is missing in patient with a fever and should
+ # have been administered
+ QC_FEVER_REPORTED_BUT_RDT_IS_MISSING_OR_NOT_DONE = case_when((vs_fever_last2d_bls_yn %in% REDCAP_LABELS_RADIO_YES | vs_temp_bls >= 37.5) &
+ (is.na(lb_rdt_done_bls_yn) | lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_NO) ~ TRUE,
+ (vs_fever_last2d_bls_yn %in% REDCAP_LABELS_RADIO_NO & vs_temp_bls < 37.5) &
+ lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_NO ~ FALSE,
+ .default = NA),
+
+ # Malaria RDT was done but Pf/pan result is missing (Madagascar only)
+ QC_RDT_PF_PAN_DONE_BUT_RESULT_IS_MISSING = case_when(lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ is.na(lb_rdt_result_bls_m) ~ TRUE,
+ lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ !is.na(lb_rdt_result_bls_m) ~ FALSE,
+ .default = NA),
+
+ # Malaria Pf/pan RDT was positive but Pf/Pv result is missing (Madagascar
+ # only)
+ QC_RDT_PF_PAN_POSITIVE_BUT_RDT_PF_PV_RESULT_IS_MISSING = case_when(lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) &
+ is.na(lb_rdt_result_bls_e) ~ TRUE,
+ lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) &
+ !is.na(lb_rdt_result_bls_e) ~ FALSE,
+ .default = NA),
+
+ # Malaria RDT was done but Pf/Pv result is missing (Madagascar only)
+ QC_RDT_PF_PV_DONE_BUT_RESULT_IS_MISSING = case_when(lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ ((country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) &
+ is.na(lb_rdt_result_bls_e)) |
+ (country %in% REDCAP_LABELS_RADIO_COUNTRY_E &
+ is.na(lb_rdt_result_bls_e))) ~ TRUE,
+ lb_rdt_done_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ ((country %in% REDCAP_LABELS_RADIO_COUNTRY_M &
+ lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PV_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_111) &
+ !is.na(lb_rdt_result_bls_e)) |
+ (country %in% REDCAP_LABELS_RADIO_COUNTRY_E &
+ !is.na(lb_rdt_result_bls_e))) ~ FALSE,
+ .default = NA),
+
+ # Patient had a positive malaria RDT but detail about referral are missing
+ QC_RDT_POSITIVE_BUT_REFERRAL_STATUS_IS_MISSING = case_when((lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) |
+ lb_rdt_result_bls_e %in% c(REDCAP_LABELS_RADIO_RDT_PF_PV_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_111)) &
+ is.na(referred_bls_yn) ~ TRUE,
+ (lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) |
+ lb_rdt_result_bls_e %in% c(REDCAP_LABELS_RADIO_RDT_PF_PV_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_111)) &
+ !is.na(referred_bls_yn) ~ FALSE,
+ .default = NA),
+
+ # Patient was referred after a positive malaria RDT but referral location
+ # is missing
+ QC_PATIENT_REFERRED_BUT_REFERRAL_LOCATION_IS_MISSING = case_when(referred_bls_yn %in% REDCAP_LABELS_RADIO_YES & is.na(referred_name_bls) ~ TRUE,
+ referred_bls_yn %in% REDCAP_LABELS_RADIO_YES & !is.na(referred_name_bls) ~ FALSE,
+ .default = NA),
+
+ # Patient had a positive malaria RDT but detail about treatment are missing
+ QC_RDT_POSITIVE_BUT_TREATMENT_STATUS_IS_MISSING = case_when((lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) |
+ lb_rdt_result_bls_e %in% c(REDCAP_LABELS_RADIO_RDT_PF_PV_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_111)) &
+ is.na(treated_bls_yn) ~ TRUE,
+ (lb_rdt_result_bls_m %in% c(REDCAP_LABELS_RADIO_RDT_PF_PAN_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PAN_111) |
+ lb_rdt_result_bls_e %in% c(REDCAP_LABELS_RADIO_RDT_PF_PV_110,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_101,
+ REDCAP_LABELS_RADIO_RDT_PF_PV_111)) &
+ !is.na(treated_bls_yn) ~ FALSE,
+ .default = NA),
+
+ # Patient was treated after a positive malaria RDT but detail about
+ # treatment administered are missing
+ QC_PATIENT_TREATED_BUT_TREATMENT_DESC_IS_MISSING = case_when(treated_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ (treated_name_bls___bs_01 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___bs_01 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___bs_02 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___bs_03 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___ls_01 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___98 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___99 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ treated_name_bls___unk %in% REDCAP_LABELS_CHECKBOX_UNCHECKED) ~ TRUE,
+ treated_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ (treated_name_bls___bs_01 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___bs_01 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___bs_02 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___bs_03 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___ls_01 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___98 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___99 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ treated_name_bls___unk %in% REDCAP_LABELS_CHECKBOX_CHECKED) ~ FALSE,
+ .default = NA),
+
+ # Patient was treated after a positive malaria RDT and detail indicate
+ # a treatment to be further documented, but such description is missing
+ QC_PATIENT_TREATED_BUT_OTHER_DESCRIPTION_IS_MISSING = case_when(treated_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ treated_name_bls___99 %in% REDCAP_LABELS_CHECKBOX_CHECKED &
+ is.na(treated_name_other_bls) ~ TRUE,
+ treated_bls_yn %in% REDCAP_LABELS_RADIO_YES &
+ treated_name_bls___99 %in% REDCAP_LABELS_CHECKBOX_CHECKED &
+ !is.na(treated_name_other_bls) ~ FALSE,
+ .default = NA),
+
+ # Information about blood sample collection is missing
+ QC_BLOOD_SAMPLE_COLLECTION_IS_MISSING = case_when(lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED ~ TRUE,
+ lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_CHECKED ~ FALSE,
+ .default = NA),
+
+ # Blood sample was not collected from patient and no reason was recorded
+ QC_BLOOD_SAMPLE_COLLECTION_WAS_NOT_DONE_WITHOUT_DOCUMENTATION = case_when(lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_CHECKED &
+ is.na(lb_bloodsample_notcollected_det) ~ TRUE,
+ (lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_CHECKED) &
+ (lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED |
+ (lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_CHECKED &
+ !is.na(lb_bloodsample_notcollected_det))) ~ FALSE,
+ .default = NA),
+
+ # Blood sample collection is inconsistent, reporting both collected and
+ # not collected status
+ QC_BLOOD_SAMPLE_COLLECTION_IS_INCORRECT = case_when((lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_CHECKED) &
+ lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_CHECKED ~ TRUE,
+ lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___unk %in% REDCAP_LABELS_CHECKBOX_UNCHECKED &
+ lb_bloodsample_collected_bls___0 %in% REDCAP_LABELS_CHECKBOX_CHECKED ~ FALSE,
+ .default = NA),
+
+ # Blood sample collection was properly conducted but sample code was not
+ # entered or scanned and is missing
+ QC_BLOOD_SAMPLE_COLLECTION_DONE_BUT_SAMPLE_CODE_IS_MISSING = case_when((lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_CHECKED) &
+ is.na(lb_samplecode_bls) ~ TRUE,
+ (lb_bloodsample_collected_bls___samp_0100_dbs %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0500_microt %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_0009_bsmear %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2000_venb %in% REDCAP_LABELS_CHECKBOX_CHECKED |
+ lb_bloodsample_collected_bls___samp_2500_venbhep_2500 %in% REDCAP_LABELS_CHECKBOX_CHECKED) &
+ !is.na(lb_samplecode_bls) ~ FALSE),
+
+ # Blood sample collection was properly conducted and sample code was
+ # recorded, but the sample code entered does not match the Participant ID
+ # and may be erroneous
+ QC_BLOOD_SAMPLE_COLLECTION_DONE_BUT_SAMPLE_CODE_IS_MISMATCH = case_when(!is.na(lb_samplecode_bls) & gsub(pattern = "^((?:E|M)-\\d{2}-\\d{3,4})-(?:A|B|C|D|M(?:0|6|12|18)|PCD)$",
+ replacement = "\\1",
+ lb_samplecode_bls) != uniqueid ~ TRUE,
+ !is.na(lb_samplecode_bls) & gsub(pattern = "^((?:E|M)-\\d{2}-\\d{3,4})-(?:A|B|C|D|M(?:0|6|12|18)|PCD)$",
+ replacement = "\\1",
+ lb_samplecode_bls) == uniqueid ~ FALSE,
+ .default = NA),
+
+ # Blood sample collection was properly conducted and sample code was
+ # recorded, but information about double-checking the code is missing or
+ # says it was not double-checked
+ QC_SAMPLE_CODE_CHECK_IS_MISSING_OR_INCORRECT = case_when(!is.na(lb_samplecode_bls) & (is.na(lb_samplecode_bls_checked) |
+ lb_samplecode_bls_checked %in% REDCAP_LABELS_RADIO_NO) ~ TRUE,
+ !is.na(lb_samplecode_bls) & lb_samplecode_bls_checked %in% REDCAP_LABELS_RADIO_YES ~ FALSE,
+ .default = NA)
+ )
+
+
+
+
+
+
+######################################################################
+### WRITE QC DATA TO OUTPUT DIRECTORY
+######################################################################
+## Name of output files
+
+
+## Write to output files
+
+
+
+
+
+
+######################################################################
+### CLEANUP
+######################################################################
+rm()
+
+
+
+
+
+######################################################################
+### UPDATE DATA_EXTRACT_IS_RECENT_OBS
+######################################################################
+DATA_EXTRACT_IS_RECENT_OBS <- as.logical(difftime(time1 = Sys.Date(),
+ time2 = as.Date(ifelse(exists("DATA_EXTRACT_TS_INT_M0"), DATA_EXTRACT_TS_INT_M0, DATA_EXTRACT_TS_DEFAULT)),
+ units = "days") <= DATA_EXTRACT_EXPIRY_TIME_D)