polypharmacy (version 1.0.0)

data_process: Create the table of the drug treatments

Description

Reads a table of successive drug delivery records (usually extracted from a pharmacy or a health insurance information system) and creates the table required for the calculation of the polypharmacy indicators by applying various user-defined arguments, incorporating hospital stays into the treatment periods and reconstruct continuous treatment periods by merging quasi continuous and/or overlapping drugs deliveries.

Usage

data_process(
  Rx_deliv,
  Rx_id,
  Rx_drug_code,
  Rx_drug_deliv,
  Rx_deliv_dur,
  Cohort = NULL,
  Cohort_id = NULL,
  Hosp_stays = NULL,
  Hosp_id = NULL,
  Hosp_admis = NULL,
  Hosp_discharge = NULL,
  study_start = NULL,
  study_end = NULL,
  grace_fctr = 0.5,
  grace_cst = 0,
  max_reserve = NULL,
  cores = parallel::detectCores(logical = FALSE),
  ...
)

Arguments

Rx_deliv

Name of the table listing all prescription drugs deliveries including the run-in period. See Details.

Rx_id

Column name of Rx_deliv containing individual unique identifier (any format).

Rx_drug_code

Column name of Rx_deliv that contains the drug unique identifier (any format).

Rx_drug_deliv

Column name of Rx_deliv that contains the dates of the drug delivery (Date format, see Details).

Rx_deliv_dur

Column name of Rx_deliv that contains the duration of the delivery (integer number).

Cohort

Name of the table providing the unique identifiers of the study cohort. Only the ids listed in both the Cohort and the Rx_deliv tables will be returned. if Cohort = NULL, all ids of the Rx_deliv table will be returned.

Cohort_id

Column name of Cohort containing individual<U+2019>s unique identifiers (same format as Rx_id). If Cohort is not NULL and Cohort_id is NULL, Cohort_id will take the same value as Rx_id.

Hosp_stays

Name of the table listing all hospital stays. (see Details for possible format).

Hosp_id

Column name of Hosp_stays containing individual<U+2019>s unique identifier (same format as Rx_id). If Hosp_stays is not NULL and Hosp_id is NULL, Hosp_id will take the same value as Rx_id.

Hosp_admis

Column name of Hosp_stays that contains the date of admission in hospital (Date format, see Details).

Hosp_discharge

Column name of Hosp_stays that contains the date of discharge from hospital (Date format, see Details).

study_start, study_end

Defines the first and last day of the study period for which the polypharmacy indicator(s) need to be calculated. All treatment periods prior to study_start and past study_end are not transcribed into the result table (Date format, see Details).

grace_fctr, grace_cst

Numbers \(\ge\) 0. Two types of grace periods can be applied. One is proportional to the treatment duration of the latest delivery (grace_fctr) and the other is a constant number of days (grace_cst).

max_reserve

An integer number \(\ge\) 0 or NULL. Longest treatment duration, in days, that can be stored from successive overlapping deliveries. When max_reserve = NULL no limit is applied. When max_reserve = 0 no accumulation of extra treatment duration is accounted for.

cores

The number of cores to use when executing data_process(). See detectCores.

...

Additional arguments. See Details. Should not be used.

Value

data.table with four (4) variables:

• The individual unique identifier which name is defined by Rx_id.

• The drug unique identifier which name is defined by Rx_drug_code.

• tx_start: The date of initiation of the reconstructed continued treatment (format as date).

• tx_end: The date of the last day of the reconstructed continued treatment (format as date).

Details

Variables:

• Rx_id, Cohort_id and Hosp_id columns must be of the same class (integer, numeric, character, ...).

• Rx_drug_deliv, Hosp_admis and Hosp_discharge can be 1) as.Date('yyyy-mm-dd'), 2) as.character('yyyy-mm-dd') or 3) as.integer() where 0 is January 1, 1970.

Arguments:

• study_start and study_end can be 1) as.Date('yyyy-mm-dd'), 2) as.character('yyyy-mm-dd') or 3) as.integer() where 0 is January 1, 1970.

Hospital stays: Drug availability is assumed to continue during the hospital stay as it is on the day prior admission. The patient is assumed to resume the consumption of the drugs delivered by community pharmacists (as recorded in Rx_deliv) the day after hosp_discharge. Grace period is always zero (0) for hospital stays.

Run-in period: A run-in period is necessary to account for the medications that are available to the individuals on the day of study_start. It is recommended to include a run-in period of about 6 months (e.g. 7 months to account for possible delays) as some drugs are delivered for up to 6 months at once.

Grace period: The grace period is used to determine if two successive deliveries can be considered as a continuous treatment even if there is a gap of several days for which no treatment is apparently available. Two successive deliveries of an identical drug are considered part of a single continuous treatment if the next delivery doesn<U+2019>t occur more than grace_cst + (grace_fctr <U+00D7> Rx_deliv_dur) days after the end of the latest drug delivery. The availability of extra drugs accumulated over the successive deliveries is accounted for prior to evaluating the duration of the gap between deliveries.

Performance For better performance, date columns are converted to integer numbers.

... verif_cols=FALSE : For better performance, you can avoid columns class checking with verif_cols=FALSE. Not recommended.

Examples

# NOT RUN {
### Standard evaluation
data_process(
  Rx_deliv = sample_Rx_unprocessed, Rx_id = "id", Rx_drug_code = "code",
  Rx_drug_deliv = "start", Rx_deliv_dur = "duration",
  cores = 1L
)

### Hospitalisation stays
rx1 <- data.frame(
  id = c(1L, 3:8),
  code = LETTERS[c(1, 3:8)],
  date = as.Date(c("2001-01-15", "2003-03-15", "2004-04-15", "2005-05-15",
                   "2006-06-15", "2007-07-15", "2008-08-15")),
  duration = 10L
)
hosp1 <- data.frame(
  ID = 3:8,
  ADM = as.Date(c("2003-03-10", "2004-04-25", "2005-05-12",
                  "2006-06-20", "2007-07-26", "2008-08-01")),
  DEP = as.Date(c("2003-03-14", "2004-04-30", "2005-05-17",
                  "2006-06-30", "2007-07-30", "2008-08-13"))
)
data_process(
  Rx_deliv = rx1, Rx_id = "id", Rx_drug_code = "code",
  Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
  Hosp_stays = hosp1, Hosp_id = "ID", Hosp_admis = "ADM", Hosp_discharge = "DEP",
  study_start = "2001-01-01", study_end = "2008-12-31",
  cores = 1L
)
# Many drug codes
rx2 <- data.frame(
  id = 1L,
  code = c(111L, 222L, 222L, 333L, 444L),
  date = as.Date(c("2001-01-15", "2002-02-15", "2002-03-01", "2004-04-07", "2004-05-05")),
  duration = as.integer(c(10, 10, 10, 30, 10))
)
hosp2 <- data.frame(
  id = 1L,
  adm = as.Date(c("2000-01-01", "2000-01-15", "2001-01-01", "2002-02-23", "2004-04-15")),
  dep = as.Date(c("2000-01-31", "2000-01-31", "2001-01-10", "2002-02-28", "2004-05-15"))
)
data_process(
  Rx_deliv = rx2, Rx_id = "id", Rx_drug_code = "code",
  Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
  Hosp_stays = hosp2, Hosp_id = "id", Hosp_admis = "adm", Hosp_discharge = "dep",
  study_start = "2001-01-01", study_end = "2008-12-31",
  cores = 1L
)

### Study dates - start and end
rx3 <- data.frame(id = 1:3,
                  code = "A",
                  date = as.Date(c("2020-01-01", "2020-06-06", "2020-12-22")),
                  duration = 10L)
# NULLs
data_process(Rx_deliv = rx3, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = NULL, study_end = NULL,
             cores = 1)
# Not NULLs
data_process(Rx_deliv = rx3, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = "2020-06-10", study_end = NULL,
             cores = 1)
data_process(Rx_deliv = rx3, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = NULL, study_end = "2020-06-10",
             cores = 1)
data_process(Rx_deliv = rx3, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = "2020-01-05", study_end = "2020-12-25",
             cores = 1)

### Grace factor
rx4 <- data.frame(id = c(rep(1, 3), rep(2, 3)),
                  code = "A",
                  date = as.Date(c("2000-01-01", "2000-01-17", "2000-01-31",
                                   "2000-06-01", "2000-06-23", "2000-07-16")),
                  duration = as.integer(c(10, 10, 10, 15, 15, 15)))
# 50% of duration
data_process(Rx_deliv = rx4, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             grace_fctr = 0.5,
             cores = 1)
# 0% of duration
data_process(Rx_deliv = rx4, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             grace_fctr = 0,
             cores = 1)

### Grace constant
rx5 <- data.frame(id = 1,
                  code = "A",
                  date = as.Date(c("2000-01-01", "2000-01-14", "2000-01-25")),
                  duration = as.integer(c(10, 10, 6)))
# 2 days
data_process(Rx_deliv = rx5, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             grace_fctr = 0, grace_cst = 2,
             cores = 1)
# 3 days
data_process(Rx_deliv = rx5, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             grace_fctr = 0, grace_cst = 3,
             cores = 1)

### Max reserve
rx6 <- data.frame(id = as.integer(c(1, 1, 3, 3, 3, 5, 5)),
                  code = "A",
                  date = as.Date(c("2000-01-01", "2000-01-31",
                                   "2000-03-03", "2000-03-15", "2000-03-30",
                                   "2000-05-05", "2000-05-05")),
                  duration = as.integer(c(30, 30,
                                          30, 30, 30,
                                          90, 90)))
# 0 days
data_process(Rx_deliv = rx6, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = NULL, study_end = "2000-12-31",
             grace_fctr = 0, grace_cst = 0,
             max_reserve = 0,
             cores = 1)
# 60 days
data_process(Rx_deliv = rx6, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = NULL, study_end = "2000-12-31",
             grace_fctr = 0, grace_cst = 0,
             max_reserve = 60,
             cores = 1)
# Inf days
data_process(Rx_deliv = rx6, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             study_start = NULL, study_end = "2000-12-31",
             grace_fctr = 0, grace_cst = 0,
             max_reserve = NULL,
             cores = 1)

### Combine Hospital stays and Grace factor
rx7 <- data.frame(id = c(1L, 1L, 1L, 2L),
                  code = "A",
                  date = c("2000-01-01", "2000-02-20", "2000-04-11", "2002-02-02"),
                  duration = as.integer(c(30, 30, 30, 15)))
hosp7 <- data.frame(id = 1L,
                    adm = c("2000-01-11", "2000-02-21"),
                    dep = c("2000-01-15", "2000-02-25"))
data_process(Rx_deliv = rx7, Rx_id = "id", Rx_drug_code = "code",
             Rx_drug_deliv = "date", Rx_deliv_dur = "duration",
             Hosp_stays = hosp7, Hosp_id = "id",
             Hosp_admis = "adm", Hosp_discharge = "dep",
             study_start = "2000-01-01", study_end = "2002-12-31",
             grace_fctr = 0.5, grace_cst = 0, max_reserve = NULL,
             cores = 1)
# }