Known Issues in PMDA Validation Engine 1810.3
Pharmaceuticals and Medical Devices Agency (PMDA) study data validation rules use implementation from the Pinnacle 21 (P21) tool. For regulatory submissions filed after March 31st, 2020, sponsors need to use the P21 validation engine version 1810.3 (engine 1511.6 is still optional until March 31st, 2021). Reported validation issues are grouped by PMDA Severity. Rejection issues must be fixed. All Errors must be explained in Reviewers Guide. During the submission process, the PMDA performs an independent validation of the study data and reconciles their results with the results reported by a sponsor. All inconsistencies are investigated, discussed, and resolved during a series of consultation meetings right before submission. Major problems may delay submission.
With a focus on consistency of validation results, the PMDA requests to report all validation results including false-positive validation messages due to application issues in P21 validation engine. PMDA has used the same validation engine version for several years, which helps the industry adopt a specific version of the validation rules. However, application issues fixes are not available to the industry until the release of the next version of the validation engine.
The purpose of this publication is to list known issues which can produce false-positive validation results in the P21 engine version 1810.3. Sponsors and data vendors may consider this publication as a reference of official acknowledgement of issues by Pinnacle 21 and use this for explanation of some false-positive validation results in the Reviewer’s Guide.
Note that most issues in this list are already fixed in more recent versions of the P21 engine. Please see our additional resources and contact Pinnacle 21 if you need additional support.
#1. FA domain
Issue: Validator incorrectly recognizes FA domain as a custom Findings domain instead of Findings About domain and applies some non-applicable rules. In some cases, algorithms of validation rule may be modified to FA domain to custom Findings domain
- SD0058: FAOBJ Variable appears in dataset, but is not in SDTM model
- FAOBJ variable is present only in FA domain, but is missing in standard specifications for custom Findings domain
- SD1044: No FABLFL variable in custom Findings domain (Warning)
- This rule may be applicable only for custom Findings domain
- SD1117: Duplicate records (Warning)
- FAOBJ variable is removed from a list of potential Keys
- SD1272: FATESTCD equals 'OTHER'
- SD1275: FATESTCD equals 'MULTIPLE'
#2. AP domains
Issue: Validation engine does not support Associated Persons (AP) domains. It incorrectly recognizes them as split AP domain.
- Validator engine merges APDM and APAE datasets into one AP (split) domain which will be considered as custom Events domain for validation. Many false-positive validation messages will be generated. For example, “ARM is non-SDTM variable” (in Events domain), “Missing value for Required AETERM variable” (for APDM records), “Inconsistent value for Domain” (expecting “AP” instead of “DM” and “AP”).
- SUPPQUAL datasets SQAPDM, SQAPMH are incorrectly recognized as split “SQ” domain
#3. "…date is after the latest Disposition date"
Issue: SD0080 (AE start date is after the latest Disposition date) and SD0082 (EX start date is after the latest Disposition date) rules reports 3 different types of false-positive messages:
- When Disposition date (DSSTDTC) is partially missing
- For example, AESTDTC=2020-01-01 and DSSTDTC=2020-01
- When AE or EX Start Date includes Time
- For example, AESTDTC=2020-01-01T15:11 and DSSTDTC=2020-01-01
- Subject has AE (EX) records but does not have any DS records
#4. SD1117: Duplicate records
Issue: Validation rule utilizes generic set of key variables for all Findings domains
Sometimes it is not enough
- For example, --NOMDY, --SPID, non-standard variables in SUPPQUALs
#5. DD0031, DD0032 for Null Flavor
Issue: Validation rules fire false positive when user does not create the user defined codelist for Null Flavor
- CDISC SDTM CT does not officially contain a codelist for “Null Flavor”. P21 linked the “Null Flavor” and its “Codelist” to SDTM Terminology. But this linkage will not be available in the later engines.