Performing Clinical Analyses | Data Integrity | Risk-Based Monitoring - External Standard - Version 2

Risk-Based Monitoring - External Standard - Version 2
The supplemental data set is used to enter any data (external to the database or otherwise) used by the RBM system. It is modeled after a CDISC event domain, and uses the two-letter prefix RB.
Variables in Supplemental Data Sets
Required Variables: The supplemental RB data set must contain the variables listed in the following table:
Special codes: QUERY, CRFPAGE.
Within the same analysis, subject and site-level variables cannot share the same values for VARIABLE and RBDECOD. In other words, variable names and labels must be distinct.
Subjects without an instance of a particular VARIABLE and RBDECOD are assumed of have 0 instances
One of Enrollment, Disposition, Safety or
Safety and Supplemental Variables: Manually Entered group variables are now called Supplemental Variables. These can include any one of Enrollment, Disposition, Safety, or Supplemental that contribute to the appropriate OVERALL flags. Adverse Events are now called Safety variables and include AEs, SAEs, deaths, and hospitalizations.
Existing supplemental data sets or risk threshold data sets having the old categories are now recoded to change Adverse Events to Safety, and Manually Entered to Supplemental. This recoding occurs when new thresholds are created from old versions, and are done on the fly for thresholds and supplemental data in the RBM report.
Site Level Variables: Risk-Based Monitoring (RBM) supports site-level variables in the supplemental data set. These are identified by having USUBJID missing. Study Site Identifier (SITEID) is now a required variable for this format. Because dates can be provided, site-level variables should be summarized in time trend analyses. Within the same analysis, subject and site-level variables cannot share the same values for VARIABLE and RBDECOD.
Example 1: Protocol deviations computed by SAS programs entered into system
 
Subject 10104 had 2 protocol deviations on December 4th. In site level table, PROTDEV is the sum of all protocol deviations for a site. PROTDEV is the country-level sum of deviations for each country. If subjects, sites or countries are not represented, they have a value of 0. RBMTERM can be used to distinguish between specific events. To count different types of deviations, this would require multiple sets of codes. For example, if you want a count of all and clinically-relevant deviations, the clinically-relevant deviations would need additional rows in the table with a new code and label (for example, CPROTDEV). These frequency variables have additional variables created for total averaged by # randomized subjects (AVPROTDEV) and by # patientweeks (PWPROTDEV).
Computed Eligibility Violations would work in the same way (if a SAS program calculated these, for example). Alternatively, specific hospitalization of interest (that would ultimately end up in the HO domain) could be collected in this manner.
Example 2:QUERY
 
Subject 10104 had 2 queries on December 4th that were resolved the same day. In site level table, QUERY is the sum of all queries for a site. QUERY is country-level sum of queries for each country. If subjects, sites or countries are not represented, they have a value of 0. Frequency variables have additional variables created for total averaged by # randomized subjects (AVQUERY) and by # patientweeks (PWQUERY). The special term QUERY also calculates OQUERY as the count of queries with missing RBENDTC. The word "Overdue" is added to QUERY label listed in RBDECOD (here, resulting in RBEDCOD value of Overdue Query). Similar averages are computed for # randomized subjects and patient weeks. The special term QUERY also calculates RQUERY, which is the average query response time (# days between RBSTDTC and RBENDTC) for all queries for a subject, site, or country, taking into consideration RBFREQ.
Example 3: CRFPAGE
It would be possible to include records for each CRF page a subject is expected to enter. Alternatively, the total completed pages or incomplete pages could be tracked. Here, RBSTDTC represents the date on which the visit occurred, while RBENDTC represents the date on which the page (paper or electronic) was completed.
 
In site level table, CRFPAGE is the sum of all pages for a site. CRFPAGE is the country-level sum of pages for each country. If subjects, sites or countries are not represented, they have a value of 0. Frequency variables have additional variables created for total averaged by # randomized subjects (AVCRFPAGE) and by # patientweeks (PW CRFPAGE). The special term CRFPAGE also calculates OCRFPAGE as the count of incomplete CRF Pages with missing RBENDTC. The word "Overdue" added to CRFPAGE label listed in RBDECOD (here, resulting in RBEDCOD value of Overdue CRF Pages). Similar averages are computed for # randomized subjects and patient weeks. The special term CRFPAGE also calculates RCRFPAGE, which is the average CRF entry time (# days between RBSTDTC and RBENDTC) for all CRF pages for a subject, site, or country, taking into consideration RBFREQ. Above is one row per page. Alternatively,
 
Here, subject 10101 had 20 pages completed for Visit 1 (Visit 1 occurred December 1, but these pages were entered on December 3), though 5 are missing. Duration would be computed as 3 days for the 20 pages. If only below is provided, the CRFPAGE and OCRFPAGE results should be similar, though RCRFPAGE is not be computed.
 
Example 4: SITEDEV
A site-level deviation variable could be added to distinguish between protocol deviations perpetrated by the site and by the subject (PROTDEV above). Here, USUBJID is blank to tell JMP Clinical that this is a site-level variable.