Skip to contents

Add Variables from an Additional Dataset Based on Conditions from Both Datasets

Source: R/derive_joined.R
derive_vars_joined.Rd

The function adds variables from an additional dataset to the input dataset. The selection of the observations from the additional dataset can depend on variables from both datasets. For example, add the lowest value (nadir) before the current observation.

Usage

derive_vars_joined(
  dataset,
  dataset_add,
  by_vars = NULL,
  order = NULL,
  new_vars = NULL,
  tmp_obs_nr_var = NULL,
  join_vars = NULL,
  join_type,
  filter_add = NULL,
  first_cond_lower = NULL,
  first_cond_upper = NULL,
  filter_join = NULL,
  mode = NULL,
  exist_flag = NULL,
  true_value = "Y",
  false_value = NA_character_,
  missing_values = NULL,
  check_type = "warning"
)

Arguments

dataset

Input dataset

The variables specified by the by_vars argument are expected to be in the dataset.

Default value

none

dataset_add

Additional dataset

The variables specified by the by_vars, the new_vars, the join_vars, and the order argument are expected.

Default value

none

by_vars

Grouping variables

The two datasets are joined by the specified variables.

Variables can be renamed by naming the element, i.e. by_vars = exprs(<name in input dataset> = <name in additional dataset>), similar to the dplyr joins.

Default value

NULL

order

Sort order

If the argument is set to a non-null value, for each observation of the input dataset the first or last observation from the joined dataset is selected with respect to the specified order. The specified variables are expected in the additional dataset (dataset_add). If a variable is available in both dataset and dataset_add, the one from dataset_add is used for the sorting.

If an expression is named, e.g., exprs(EXSTDT = convert_dtc_to_dt(EXSTDTC), EXSEQ), a corresponding variable (EXSTDT) is added to the additional dataset and can be used in the filter conditions (filter_add, filter_join) and for join_vars and new_vars. The variable is not included in the output dataset.

For handling of NAs in sorting variables see Sort Order.

Permitted values

list of expressions created by exprs(), e.g., exprs(ADT, desc(AVAL)) or NULL

Default value

NULL

new_vars

Variables to add

The specified variables from the additional dataset are added to the output dataset. Variables can be renamed by naming the element, i.e., new_vars = exprs(<new name> = <old name>).

For example new_vars = exprs(var1, var2) adds variables var1 and var2 from dataset_add to the input dataset.

And new_vars = exprs(var1, new_var2 = old_var2) takes var1 and old_var2 from dataset_add and adds them to the input dataset renaming old_var2 to new_var2.

Values of the added variables can be modified by specifying an expression. For example, new_vars = LASTRSP = exprs(str_to_upper(AVALC)) adds the variable LASTRSP to the dataset and sets it to the upper case value of AVALC.

If the argument is not specified or set to NULL, all variables from the additional dataset (dataset_add) are added.

Permitted values

list of variables or named expressions created by exprs()

Default value

NULL

tmp_obs_nr_var

Temporary observation number

The specified variable is added to the input dataset (dataset) and the additional dataset (dataset_add). It is set to the observation number with respect to order. For each by group (by_vars) the observation number starts with 1. The variable can be used in the conditions (filter_join, first_cond_upper, first_cond_lower). It can also be used to select consecutive observations or the last observation.

The variable is not included in the output dataset. To include it specify it for new_vars.

Default value

NULL

join_vars

Variables to use from additional dataset

Any extra variables required from the additional dataset for filter_join should be specified for this argument. Variables specified for new_vars do not need to be repeated for join_vars. If a specified variable exists in both the input dataset and the additional dataset, the suffix ".join" is added to the variable from the additional dataset.

If an expression is named, e.g., exprs(EXTDT = convert_dtc_to_dt(EXSTDTC)), a corresponding variable is added to the additional dataset and can be used in the filter conditions (filter_add, filter_join) and for new_vars. The variable is not included in the output dataset.

The variables are not included in the output dataset.

Permitted values

list of variables or named expressions created by exprs()

Default value

NULL

join_type

Observations to keep after joining

The argument determines which of the joined observations are kept with respect to the original observation. For example, if join_type = "after" is specified all observations after the original observations are kept.

For example for confirmed response or BOR in the oncology setting or confirmed deterioration in questionnaires the confirmatory assessment must be after the assessment. Thus join_type = "after" could be used.

Whereas, sometimes you might allow for confirmatory observations to occur prior to the observation. For example, to identify AEs occurring on or after seven days before a COVID AE. Thus join_type = "all" could be used.

Permitted values

"before", "after", "all"

Default value

none

filter_add

Filter for additional dataset (dataset_add)

Only observations from dataset_add fulfilling the specified condition are joined to the input dataset. If the argument is not specified, all observations are joined.

Variables created by order or new_vars arguments can be used in the condition.

The condition can include summary functions like all() or any(). The additional dataset is grouped by the by variables (by_vars).

Permitted values

a condition

Default value

NULL

first_cond_lower

Condition for selecting range of data (before)

If this argument is specified, the other observations are restricted from the first observation before the current observation where the specified condition is fulfilled up to the current observation. If the condition is not fulfilled for any of the other observations, no observations are considered.

This argument should be specified if filter_join contains summary functions which should not apply to all observations but only from a certain observation before the current observation up to the current observation. For an example see the last example below.

Default value

NULL

first_cond_upper

Condition for selecting range of data (after)

If this argument is specified, the other observations are restricted up to the first observation where the specified condition is fulfilled. If the condition is not fulfilled for any of the other observations, no observations are considered.

This argument should be specified if filter_join contains summary functions which should not apply to all observations but only up to the confirmation assessment. For an example see the last example below.

Default value

NULL

filter_join

Filter for the joined dataset

The specified condition is applied to the joined dataset. Therefore variables from both datasets dataset and dataset_add can be used.

Variables created by order or new_vars arguments can be used in the condition.

The condition can include summary functions like all() or any(). The joined dataset is grouped by the original observations.

Permitted values

a condition

Default value

NULL

mode

Selection mode

Determines if the first or last observation is selected. If the order argument is specified, mode must be non-null.

If the order argument is not specified, the mode argument is ignored.

Permitted values

"first", "last", NULL

Default value

NULL

exist_flag

Exist flag

If the argument is specified (e.g., exist_flag = FLAG), the specified variable (e.g., FLAG) is added to the input dataset. This variable will be the value provided in true_value for all selected records from dataset_add which are merged into the input dataset, and the value provided in false_value otherwise.

Permitted values

Variable name

Default value

NULL

true_value

True value

The value for the specified variable exist_flag, applicable to the first or last observation (depending on the mode) of each by group.

Permitted values

An atomic scalar

Default value

"Y"

false_value

False value

The value for the specified variable exist_flag, NOT applicable to the first or last observation (depending on the mode) of each by group.

Permitted values

An atomic scalar

Default value

NA_character_

missing_values

Values for non-matching observations

For observations of the input dataset (dataset) which do not have a matching observation in the additional dataset (dataset_add) the values of the specified variables are set to the specified value. Only variables specified for new_vars can be specified for missing_values.

Permitted values

named list of expressions, e.g., exprs(BASEC = "MISSING", BASE = -1)

Default value

NULL

check_type

Check uniqueness?

If "warning" or "error" is specified, the specified message is issued if the observations of the (restricted) joined dataset are not unique with respect to the by variables and the order.

This argument is ignored if order is not specified. In this case an error is issued independent of check_type if the restricted joined dataset contains more than one observation for any of the observations of the input dataset.

Permitted values

"none", "warning", "error"

Default value

"warning"

Value

The output dataset contains all observations and variables of the input dataset and additionally the variables specified for new_vars from the additional dataset (dataset_add).

Details

  1. The variables specified by order are added to the additional dataset (dataset_add).

  2. The variables specified by join_vars are added to the additional dataset (dataset_add).

  3. The records from the additional dataset (dataset_add) are restricted to those matching the filter_add condition.

  4. The input dataset and the (restricted) additional dataset are left joined by the grouping variables (by_vars). If no grouping variables are specified, a full join is performed.

  5. If first_cond_lower is specified, for each observation of the input dataset the joined dataset is restricted to observations from the first observation where first_cond_lower is fulfilled (the observation fulfilling the condition is included) up to the observation of the input dataset. If for an observation of the input dataset the condition is not fulfilled, the observation is removed.

    If first_cond_upper is specified, for each observation of the input dataset the joined dataset is restricted to observations up to the first observation where first_cond_upper is fulfilled (the observation fulfilling the condition is included). If for an observation of the input dataset the condition is not fulfilled, the observation is removed.

    For an example see the last example in the "Examples" section.

  6. The joined dataset is restricted by the filter_join condition.

  7. If order is specified, for each observation of the input dataset the first or last observation (depending on mode) is selected.

  8. The variables specified for new_vars are created (if requested) and merged to the input dataset. I.e., the output dataset contains all observations from the input dataset. For observations without a matching observation in the joined dataset the new variables are set as specified by missing_values (or to NA for variables not in missing_values). Observations in the additional dataset which have no matching observation in the input dataset are ignored.

Note: This function creates temporary datasets which may be much bigger than the input datasets. If this causes memory issues, please try setting the admiral option save_memory to TRUE (see set_admiral_options()). This reduces the memory consumption but increases the run-time.

See also

derive_var_joined_exist_flag(), filter_joined()

General Derivation Functions for all ADaMs that returns variable appended to dataset: derive_var_extreme_flag(), derive_var_joined_exist_flag(), derive_var_merged_ef_msrc(), derive_var_merged_exist_flag(), derive_var_merged_summary(), derive_var_obs_number(), derive_var_relative_flag(), derive_vars_cat(), derive_vars_computed(), derive_vars_merged(), derive_vars_merged_lookup(), derive_vars_transposed()

Examples

library(tibble)
library(lubridate)
library(dplyr, warn.conflicts = FALSE)
library(tidyr)

# Add AVISIT (based on time windows), AWLO, and AWHI
adbds <- tribble(
  ~USUBJID, ~ADY,
  "1",       -33,
  "1",        -2,
  "1",         3,
  "1",        24,
  "2",        NA,
)

windows <- tribble(
  ~AVISIT,    ~AWLO, ~AWHI,
  "BASELINE",   -30,     1,
  "WEEK 1",       2,     7,
  "WEEK 2",       8,    15,
  "WEEK 3",      16,    22,
  "WEEK 4",      23,    30
)

derive_vars_joined(
  adbds,
  dataset_add = windows,
  join_type = "all",
  filter_join = AWLO <= ADY & ADY <= AWHI
)
#> # A tibble: 5 × 5
#>   USUBJID   ADY AVISIT    AWLO  AWHI
#>   <chr>   <dbl> <chr>    <dbl> <dbl>
#> 1 1         -33 NA          NA    NA
#> 2 1          -2 BASELINE   -30     1
#> 3 1           3 WEEK 1       2     7
#> 4 1          24 WEEK 4      23    30
#> 5 2          NA NA          NA    NA

# derive the nadir after baseline and before the current observation
adbds <- tribble(
  ~USUBJID, ~ADY, ~AVAL,
  "1",        -7,    10,
  "1",         1,    12,
  "1",         8,    11,
  "1",        15,     9,
  "1",        20,    14,
  "1",        24,    12,
  "2",        13,     8
)

derive_vars_joined(
  adbds,
  dataset_add = adbds,
  by_vars = exprs(USUBJID),
  order = exprs(AVAL),
  new_vars = exprs(NADIR = AVAL),
  join_vars = exprs(ADY),
  join_type = "all",
  filter_add = ADY > 0,
  filter_join = ADY.join < ADY,
  mode = "first",
  check_type = "none"
)
#> # A tibble: 7 × 4
#>   USUBJID   ADY  AVAL NADIR
#>   <chr>   <dbl> <dbl> <dbl>
#> 1 1          -7    10    NA
#> 2 1           1    12    NA
#> 3 1           8    11    12
#> 4 1          15     9    11
#> 5 1          20    14     9
#> 6 1          24    12     9
#> 7 2          13     8    NA

# add highest hemoglobin value within two weeks before AE,
# take earliest if more than one
adae <- tribble(
  ~USUBJID, ~ASTDY,
  "1",           3,
  "1",          22,
  "2",           2
)

adlb <- tribble(
  ~USUBJID, ~PARAMCD, ~ADY, ~AVAL,
  "1",      "HGB",       1,   8.5,
  "1",      "HGB",       3,   7.9,
  "1",      "HGB",       5,   8.9,
  "1",      "HGB",       8,   8.0,
  "1",      "HGB",       9,   8.0,
  "1",      "HGB",      16,   7.4,
  "1",      "HGB",      24,   8.1,
  "1",      "ALB",       1,    42,
)

derive_vars_joined(
  adae,
  dataset_add = adlb,
  by_vars = exprs(USUBJID),
  order = exprs(AVAL, desc(ADY)),
  new_vars = exprs(HGB_MAX = AVAL, HGB_DY = ADY),
  join_type = "all",
  filter_add = PARAMCD == "HGB",
  filter_join = ASTDY - 14 <= ADY & ADY <= ASTDY,
  mode = "last"
)
#> # A tibble: 3 × 4
#>   USUBJID ASTDY HGB_MAX HGB_DY
#>   <chr>   <dbl>   <dbl>  <dbl>
#> 1 1           3     8.5      1
#> 2 1          22     8        8
#> 3 2           2    NA       NA

# Add APERIOD, APERIODC based on ADSL
adsl <- tribble(
  ~USUBJID, ~AP01SDT,     ~AP01EDT,     ~AP02SDT,     ~AP02EDT,
  "1",      "2021-01-04", "2021-02-06", "2021-02-07", "2021-03-07",
  "2",      "2021-02-02", "2021-03-02", "2021-03-03", "2021-04-01"
) %>%
  mutate(across(ends_with("DT"), ymd)) %>%
  mutate(STUDYID = "xyz")

period_ref <- create_period_dataset(
  adsl,
  new_vars = exprs(APERSDT = APxxSDT, APEREDT = APxxEDT)
)

period_ref
#> # A tibble: 4 × 5
#>   STUDYID USUBJID APERIOD APERSDT    APEREDT   
#>   <chr>   <chr>     <int> <date>     <date>    
#> 1 xyz     1             1 2021-01-04 2021-02-06
#> 2 xyz     1             2 2021-02-07 2021-03-07
#> 3 xyz     2             1 2021-02-02 2021-03-02
#> 4 xyz     2             2 2021-03-03 2021-04-01

adae <- tribble(
  ~USUBJID, ~ASTDT,
  "1",      "2021-01-01",
  "1",      "2021-01-05",
  "1",      "2021-02-05",
  "1",      "2021-03-05",
  "1",      "2021-04-05",
  "2",      "2021-02-15",
) %>%
  mutate(
    ASTDT = ymd(ASTDT),
    STUDYID = "xyz"
  )

derive_vars_joined(
  adae,
  dataset_add = period_ref,
  by_vars = exprs(STUDYID, USUBJID),
  join_vars = exprs(APERSDT, APEREDT),
  join_type = "all",
  filter_join = APERSDT <= ASTDT & ASTDT <= APEREDT
)
#> # A tibble: 6 × 6
#>   USUBJID ASTDT      STUDYID APERIOD APERSDT    APEREDT   
#>   <chr>   <date>     <chr>     <int> <date>     <date>    
#> 1 1       2021-01-01 xyz          NA NA         NA        
#> 2 1       2021-01-05 xyz           1 2021-01-04 2021-02-06
#> 3 1       2021-02-05 xyz           1 2021-01-04 2021-02-06
#> 4 1       2021-03-05 xyz           2 2021-02-07 2021-03-07
#> 5 1       2021-04-05 xyz          NA NA         NA        
#> 6 2       2021-02-15 xyz           1 2021-02-02 2021-03-02

# Add day since last dose (LDRELD)
adae <- tribble(
  ~USUBJID, ~ASTDT,       ~AESEQ,
  "1",      "2020-02-02",      1,
  "1",      "2020-02-04",      2
) %>%
  mutate(ASTDT = ymd(ASTDT))

ex <- tribble(
  ~USUBJID, ~EXSDTC,
  "1",      "2020-01-10",
  "1",      "2020-01",
  "1",      "2020-01-20",
  "1",      "2020-02-03"
)

## Please note that EXSDT is created via the order argument and then used
## for new_vars, filter_add, and filter_join
derive_vars_joined(
  adae,
  dataset_add = ex,
  by_vars = exprs(USUBJID),
  order = exprs(EXSDT = convert_dtc_to_dt(EXSDTC)),
  join_type = "all",
  new_vars = exprs(LDRELD = compute_duration(
    start_date = EXSDT, end_date = ASTDT
  )),
  filter_add = !is.na(EXSDT),
  filter_join = EXSDT <= ASTDT,
  mode = "last"
)
#> # A tibble: 2 × 4
#>   USUBJID ASTDT      AESEQ LDRELD
#>   <chr>   <date>     <dbl>  <dbl>
#> 1 1       2020-02-02     1     14
#> 2 1       2020-02-04     2      2

# first_cond_lower and first_cond_upper argument
myd <- tribble(
  ~subj, ~day, ~val,
  "1",      1, "++",
  "1",      2, "-",
  "1",      3, "0",
  "1",      4, "+",
  "1",      5, "++",
  "1",      6, "-",
  "2",      1, "-",
  "2",      2, "++",
  "2",      3, "+",
  "2",      4, "0",
  "2",      5, "-",
  "2",      6, "++"
)

# derive last "++" day before "0" where all results in between are "+" or "++"
derive_vars_joined(
  myd,
  dataset_add = myd,
  by_vars = exprs(subj),
  order = exprs(day),
  mode = "first",
  new_vars = exprs(prev_plus_day = day),
  join_vars = exprs(val),
  join_type = "before",
  first_cond_lower = val.join == "++",
  filter_join = val == "0" & all(val.join %in% c("+", "++"))
)
#> # A tibble: 12 × 4
#>    subj    day val   prev_plus_day
#>    <chr> <dbl> <chr>         <dbl>
#>  1 1         1 ++               NA
#>  2 1         2 -                NA
#>  3 1         3 0                NA
#>  4 1         4 +                NA
#>  5 1         5 ++               NA
#>  6 1         6 -                NA
#>  7 2         1 -                NA
#>  8 2         2 ++               NA
#>  9 2         3 +                NA
#> 10 2         4 0                 2
#> 11 2         5 -                NA
#> 12 2         6 ++               NA

# derive first "++" day after "0" where all results in between are "+" or "++"
derive_vars_joined(
  myd,
  dataset_add = myd,
  by_vars = exprs(subj),
  order = exprs(day),
  mode = "last",
  new_vars = exprs(next_plus_day = day),
  join_vars = exprs(val),
  join_type = "after",
  first_cond_upper = val.join == "++",
  filter_join = val == "0" & all(val.join %in% c("+", "++"))
)
#> # A tibble: 12 × 4
#>    subj    day val   next_plus_day
#>    <chr> <dbl> <chr>         <dbl>
#>  1 1         1 ++               NA
#>  2 1         2 -                NA
#>  3 1         3 0                 5
#>  4 1         4 +                NA
#>  5 1         5 ++               NA
#>  6 1         6 -                NA
#>  7 2         1 -                NA
#>  8 2         2 ++               NA
#>  9 2         3 +                NA
#> 10 2         4 0                NA
#> 11 2         5 -                NA
#> 12 2         6 ++               NA