dispositions-and-listings
dispositions-and-listings.Rmd
library(sweets)
#> Welcome 🍩 to 🍰 sweets 🍭
library(dmcognigen)
library(dplyr,warn.conflicts = FALSE)During the data build records are flagged with a deletion flag variable called DELFN and removed. Records that have been removed are stored in the delimp directory. To do the disposition counts the removed records are added back to the data build. For this vignette we are starting at this step where deleted delimp records have already been added back to the built analysis ready dataset.
1) Load a dataset
data(broken_pk, package = "sweets")Check the deletion flags.
broken_pk %>%
cnt(DELFN,DELFNC,prop = FALSE)
#> # A tibble: 6 × 4
#> DELFN DELFNC n n_cumulative
#> <dbl> <chr> <int> <int>
#> 1 0 Analysis Record 2320 2320
#> 2 7 Placebo Group Not Included in the Analysis 1204 3524
#> 3 9 Missing Concentration Value 8 3532
#> 4 10 Missing Sample Date and/or Time 8 3540
#> 5 14 Missing Dose Date or Time 14 3554
#> 6 50 Analyst-Identified Outliers 2 3556Check DVID
We want to track and count the removal of concentration records, not dose records (i.e. DVID=0 & EVID=1). DELFN = 0 are records kept for the analysis ready dataset build. Records with Non zero DELFN were removed, stored in delimp and combined back for the disposition and listings.
broken_pk %>%
cnt(DVID,DVIDC,prop = FALSE)
#> # A tibble: 2 × 4
#> DVID DVIDC n n_cumulative
#> <dbl> <chr> <int> <int>
#> 1 0 Dose 254 254
#> 2 1 Xanomeline Concentration (ug/mL) 3302 3556For the deletions that involve dose records, deletion flags are carried from the dose record (DVID=0, EVID=1) onto the affected concentration records (DVID=1) after the dose record of concern. Here we see all the concentration records after the missing dose DTTM have flag 14.
broken_pk %>%
group_by(USUBJID) %>%
filter(any(DELFN==14)) %>%
select(USUBJID,DTTM,DVID,TSFD,DELFN)
#> # A tibble: 14 × 5
#> # Groups: USUBJID [1]
#> USUBJID DTTM DVID TSFD DELFN
#> <chr> <dttm> <dbl> <dbl> <dbl>
#> 1 01-709-1326 NA 0 0 14
#> 2 01-709-1326 2013-04-05 00:05:00 1 0.0833 14
#> 3 01-709-1326 2013-04-05 00:30:00 1 0.5 14
#> 4 01-709-1326 2013-04-05 01:00:00 1 1 14
#> 5 01-709-1326 2013-04-05 01:30:00 1 1.5 14
#> 6 01-709-1326 2013-04-05 02:00:00 1 2 14
#> 7 01-709-1326 2013-04-05 04:00:00 1 4 14
#> 8 01-709-1326 2013-04-05 06:00:00 1 6 14
#> 9 01-709-1326 2013-04-05 08:00:00 1 8 14
#> 10 01-709-1326 2013-04-05 12:00:00 1 12 14
#> 11 01-709-1326 2013-04-05 16:00:00 1 16 14
#> 12 01-709-1326 2013-04-06 00:00:00 1 24 14
#> 13 01-709-1326 2013-04-06 12:00:00 1 36 14
#> 14 01-709-1326 2013-04-07 00:00:00 1 48 142) Subset the data build to concentration i.e. sample records only
Just concentration records are left
3) Make a disposition table
Here we are splitting by STUDYID
disposition_table <- broken_pk_samples %>%
sweet_disposition(subjid = USUBJID,group_vars = STUDYID)
disposition_table %>%
print(n = 15,width = Inf)
#> # A tibble: 12 × 8
#> # Groups: STUDYID [2]
#> `Flag #` `Reason for Deletion` STUDYID `Samples Excluded` `Subjects Affected` `Subjects Excluded` `Remaining Samples` `Remaining Subjects`
#> <dbl> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 0 Concentration Records Received CDISCPILOT01 0 0 0 2028 156
#> 2 0 Concentration Records Received CDISCPILOT02 0 0 0 1274 98
#> 3 7 Placebo Group Not Included in the Analysis CDISCPILOT01 689 53 53 1339 103
#> 4 7 Placebo Group Not Included in the Analysis CDISCPILOT02 429 33 33 845 65
#> 5 9 Missing Concentration Value CDISCPILOT01 8 2 0 1331 103
#> 6 9 Missing Concentration Value CDISCPILOT02 0 0 0 845 65
#> 7 10 Missing Sample Date and/or Time CDISCPILOT01 8 2 0 1323 103
#> 8 10 Missing Sample Date and/or Time CDISCPILOT02 0 0 0 845 65
#> 9 14 Missing Dose Date or Time CDISCPILOT01 0 0 0 1323 103
#> 10 14 Missing Dose Date or Time CDISCPILOT02 13 1 1 832 64
#> 11 50 Analyst-Identified Outliers CDISCPILOT01 1 1 0 1322 103
#> 12 50 Analyst-Identified Outliers CDISCPILOT02 1 1 0 831 64Someone changed their mind and now we want to keep DELFN 50 (Analyst-Identified Outliers) records in the dataset. Let’s set DELFN 50 as a ‘count and keep’ flag. Notice how the 2 samples that were previously removed are now only ‘Affected’ and not ‘Excluded’
}
broken_pk_samples %>%
sweet_disposition(subjid = USUBJID,
group_vars = STUDYID,
cnt_n_keeps = 50) %>%
print(n = 15,width = Inf)
#> # A tibble: 12 × 8
#> # Groups: STUDYID [2]
#> `Flag #` `Reason for Deletion` STUDYID `Samples Excluded` `Subjects Affected` `Subjects Excluded` `Remaining Samples` `Remaining Subjects`
#> <dbl> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 0 Concentration Records Received CDISCPILOT01 0 0 0 2028 156
#> 2 0 Concentration Records Received CDISCPILOT02 0 0 0 1274 98
#> 3 7 Placebo Group Not Included in the Analysis CDISCPILOT01 689 53 53 1339 103
#> 4 7 Placebo Group Not Included in the Analysis CDISCPILOT02 429 33 33 845 65
#> 5 9 Missing Concentration Value CDISCPILOT01 8 2 0 1331 103
#> 6 9 Missing Concentration Value CDISCPILOT02 0 0 0 845 65
#> 7 10 Missing Sample Date and/or Time CDISCPILOT01 8 2 0 1323 103
#> 8 10 Missing Sample Date and/or Time CDISCPILOT02 0 0 0 845 65
#> 9 14 Missing Dose Date or Time CDISCPILOT01 0 0 0 1323 103
#> 10 14 Missing Dose Date or Time CDISCPILOT02 13 1 1 832 64
#> 11 50 Analyst-Identified Outliers CDISCPILOT01 0 1 0 1323 103
#> 12 50 Analyst-Identified Outliers CDISCPILOT02 0 1 0 832 64Pool disposition
pooling a disposition with pool_disposition() will ignore any group stratification set in your disposition table and add a Grand Total row.
pooled_disposition <- disposition_table %>%
pool_disposition()
#> ℹ Grand Total of `Subjects Affected` is unique subjects affected
pooled_disposition %>%
print(n = 15, width = Inf)
#> # A tibble: 7 × 6
#> `Reason for Deletion` `Samples Excluded` `Subjects Affected` `Subjects Excluded` `Remaining Samples` `Remaining Subjects`
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 Concentration Records Received 0 0 0 3302 254
#> 2 Placebo Group Not Included in the Analysis 1118 86 86 2184 168
#> 3 Missing Concentration Value 8 2 0 2176 168
#> 4 Missing Sample Date and/or Time 8 2 0 2168 168
#> 5 Missing Dose Date or Time 13 1 1 2155 167
#> 6 Analyst-Identified Outliers 2 2 0 2153 167
#> 7 Grand Total 1149 93 87 2153 167Filtered disposition
For most projects it makes sense to include a “report” version of the disposition table with only DELFN greater than 7
report_disposition <- disposition_table %>%
filter(`Flag #`> 7)
report_disposition %>%
print(n = 15, width = Inf)
#> # A tibble: 8 × 8
#> # Groups: STUDYID [2]
#> `Flag #` `Reason for Deletion` STUDYID `Samples Excluded` `Subjects Affected` `Subjects Excluded` `Remaining Samples` `Remaining Subjects`
#> <dbl> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 9 Missing Concentration Value CDISCPILOT01 8 2 0 1331 103
#> 2 9 Missing Concentration Value CDISCPILOT02 0 0 0 845 65
#> 3 10 Missing Sample Date and/or Time CDISCPILOT01 8 2 0 1323 103
#> 4 10 Missing Sample Date and/or Time CDISCPILOT02 0 0 0 845 65
#> 5 14 Missing Dose Date or Time CDISCPILOT01 0 0 0 1323 103
#> 6 14 Missing Dose Date or Time CDISCPILOT02 13 1 1 832 64
#> 7 50 Analyst-Identified Outliers CDISCPILOT01 1 1 0 1322 103
#> 8 50 Analyst-Identified Outliers CDISCPILOT02 1 1 0 831 64Write disposition
disposition_tabs <- list("pooled"= pooled_disposition,
"report"= report_disposition,
"programmer"= disposition_table,
)
writexl::write_xlsx(disposition_tabs,"disposition.xlsx")🍭 It is best practice to refer back to the counts in your code and check the results against your newly generated disposition table
4) Deletion Listings
Setup the report
Initialize the deletion listings report by calling
build_report()
del_report <- build_report(file_name = "pk-deletion-listings")🍩 Listings are project specific. Always have a discussion about what should be included.
Check the flags and reasons in the disposition with samples excluded > 0. We don’t need empty tables in our listings (e.g. CDISCPILOT02 Missing Concentration Value). These remaining 7 rows indicate there should be 7 tables in our listings.
disposition_table %>%
filter(`Samples Excluded`>0) %>%
select(STUDYID,`Flag #`,`Reason for Deletion`,`Samples Excluded`)
#> # A tibble: 7 × 4
#> # Groups: STUDYID [2]
#> STUDYID `Flag #` `Reason for Deletion` `Samples Excluded`
#> <chr> <dbl> <chr> <dbl>
#> 1 CDISCPILOT01 7 Placebo Group Not Included in the Analysis 689
#> 2 CDISCPILOT02 7 Placebo Group Not Included in the Analysis 429
#> 3 CDISCPILOT01 9 Missing Concentration Value 8
#> 4 CDISCPILOT01 10 Missing Sample Date and/or Time 8
#> 5 CDISCPILOT02 14 Missing Dose Date or Time 13
#> 6 CDISCPILOT01 50 Analyst-Identified Outliers 1
#> 7 CDISCPILOT02 50 Analyst-Identified Outliers 1Add listing tables
Again we take our broken_pk_samplesdataset consisting of
the analysis ready dataset plus the deleted records that have been
removed during the build and now added back. Listings are simply a
consolidation of one or more tables. These tables need to be customized
to show variables that make sense for each deletion reason. Perhaps we
would want to show TRT or ACTARM for
Placebo Group Not Included in the Analysis or
DTTM for Missing Sample Date and/or Time.
🍰 Try to use source variables over build variables in your listings.
If your dataset has labels, the label will appear as the variable
name on the table. To turn this off populate ... of
add_listing_to_report() using
use_attributes = 'none'. See reporter::create_table()
for additional table specifications. If you need a quick way to add,
format, or modify an existing label add reporter::define()
customization to listing_table()
# call the initialized deletion report
# build listing tables and add them inside add_listing_to_report()
# Add tables to the report
del_report <- del_report %>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT01', DELFN == 7) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(STUDYID,USUBJID,TRT),
title = "Study 01 Placebo Group Not Included in the Analysis")
) %>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT02', DELFN == 7) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(STUDYID,USUBJID,TRT,ACTARM),
title = "Study 02 Placebo Group Not Included in the Analysis")
)%>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT01', DELFN == 9) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(USUBJID,DTTM,VISIT,PCTPT,PCTEST,PCSTRESN),
title = "Study 01 Missing Concentration Value")
) %>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT01', DELFN == 10) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(USUBJID,DTTM,VISIT,PCTPT,PCTEST,PCSTRESN),
title = "Study 01 Missing Sample Date and/or Time")
) %>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT02', DELFN == 14) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(USUBJID,DTTM,VISIT,PCTPT,PCTEST,DVIDC,DV),
title = "Study 02 Missing Sample Date and/or Time")
)%>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT01', DELFN == 50) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(USUBJID,DTTM,VISIT,PCTPT,PCTEST,DVIDC,DV),
title = "Study 01 Analyst-Identified Outliers")
) %>%
add_listing_to_report(
listing_table(
broken_pk_samples %>%
filter(STUDYID=='CDISCPILOT02', DELFN == 50) %>%
arrange(STUDYID, USUBJID, DTTM) %>%
select(USUBJID,DTTM,VISIT,PCTPT,PCTEST,DVIDC,DV),
title = "Study 02 Analyst-Identified Outliers")
)
#> ✔ Listing added to report.
#> ℹ There are now 1 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 689 rows 3 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 01 Placebo Group Not Included in the Analysis'
#> ✔ Listing added to report.
#> ℹ There are now 2 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 429 rows 4 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 02 Placebo Group Not Included in the Analysis'
#> ✔ Listing added to report.
#> ℹ There are now 3 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 8 rows 6 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 01 Missing Concentration Value'
#> ✔ Listing added to report.
#> ℹ There are now 4 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 8 rows 6 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 01 Missing Sample Date and/or Time'
#> ✔ Listing added to report.
#> ℹ There are now 5 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 13 rows 7 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 02 Missing Sample Date and/or Time'
#> ✔ Listing added to report.
#> ℹ There are now 6 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 1 rows 7 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 01 Analyst-Identified Outliers'
#> ✔ Listing added to report.
#> ℹ There are now 7 listings total.
#> ℹ New listing attributes:
#> # A table specification:
#> - data: tibble 'listing' 1 rows 7 cols
#> - show_cols: all
#> - use_attributes: all
#> - title 1: 'Study 02 Analyst-Identified Outliers'get_listing_info() allows you to see the listings added
to the deletion report. This can be useful when troubleshooting or
comparing disposition counts to listings.id_var argument
adds subject counts. Default is USUBJID. Change it to
ID or SUBJID etc. based on your dataset or
ignore it by setting to none
get_listing_info(del_report)
#> # A tibble: 7 × 5
#> Title listing n_USUBJID n n_cumulative
#> <chr> <named list> <int> <int> <int>
#> 1 Study 01 Placebo Group Not Included in the Analysis <tibble [689 × 3]> 53 689 706
#> 2 Study 02 Placebo Group Not Included in the Analysis <tibble [429 × 4]> 33 429 1149
#> 3 Study 01 Missing Concentration Value <tibble [8 × 6]> 2 8 9
#> 4 Study 01 Missing Sample Date and/or Time <tibble [8 × 6]> 2 8 17
#> 5 Study 02 Missing Sample Date and/or Time <tibble [13 × 7]> 1 13 720
#> 6 Study 01 Analyst-Identified Outliers <tibble [1 × 7]> 1 1 1
#> 7 Study 02 Analyst-Identified Outliers <tibble [1 × 7]> 1 1 707
get_listing_info(del_report,id_var = "none") %>%
dplyr::slice(7) %>%
tidyr::unnest(listing)
#> "none" is not a variable in deletion report and will be ignored
#> # A tibble: 1 × 8
#> Title USUBJID DTTM VISIT PCTPT PCTEST DVIDC DV
#> <chr> <chr> <dttm> <chr> <chr> <chr> <chr> <dbl>
#> 1 Study 02 Analyst-Identified Outliers 01-718-1254 2013-07-10 00:30:00 BASELINE 30 Min Post-dose XANOMELINE Xanomeline Concentration (ug/mL) 0.507Write Listings
file path by default will be the path specified during report
initialization with build_report. The ... is
for extra arguments to be passed to reporter::write_report()
# Write Listings
write_listings_report(del_report)