A Deep Dive for Statistical Programmers
Audience: Statistical programmers in clinical data submission roles
Regulators Covered: FDA (United States) and PMDA (Japan)
Standards Referenced: CDISC SDTM IG 3.3, ADaM IG 1.1, J-SDTM, J-ADaM, CDISC CT, JCTB
When FDA or PMDA receives a clinical data package, reviewers do not open the datasets first. They open the documentation that explains the datasets. Without that documentation, every non-standard domain, every custom analysis parameter, and every deviation from the published implementation guide becomes a question — and questions stop the review clock.
The Study Data Standardization Plan (SDSP) for FDA, and PMDA's Form A and Form B, exist to give reviewers context before they open a single XPT file. These documents are not administrative formalities. They are the mechanism by which a sponsor communicates its data architecture to the people responsible for approving the drug.
The distinction between a submission that generates five information requests and one that generates none often comes down to how well these documents were prepared — and how early.
| The core principle: SDSP and Form A tell the agency what to expect before they see the data. Form B tells the agency what they actually received. Together, they eliminate the gap between sponsor intent and reviewer interpretation. |
This article covers all three documents in technical depth: what they contain, when to submit them, how they relate to P21 validation output, what patterns of rejection they prevent, and how they affect review timelines.
The SDSP requirement originates in FDA's December 2014 guidance 'Providing Regulatory Submissions in Electronic Format — Standardized Study Data' and is elaborated in the CDER and CBER Technical Conformance Guide (TCG), last substantively updated in 2022. The requirement applies to IND submissions for Phase 2 and 3 studies, and to all NDA, BLA, sNDA, and sBLA applications.
The SDSP is submitted as part of the IND packet — specifically, as an amendment when a Phase 2 or Phase 3 study initiates. If that window has passed, it is submitted at the Type B pre-NDA meeting as a status document. Once submitted, any material change to the standardization approach requires an SDSP amendment: adding a study, changing an IG version, introducing new non-standard domains, or revising the deviation list.
FDA review divisions use the SDSP to:
The table below maps each required content area to what FDA expects, based on the TCG and patterns observed in FDA information requests.
| Content Area | What to Include | FDA Expectation |
| Standards Catalog | SDTM IG version, ADaM IG version, CDISC CT version and release date | Specific versions — not ranges. E.g., SDTM IG 3.3, ADaM IG 1.1, CT 2023-03-31. FDA cross-checks these against DEFINE.xml |
| Domain List | All planned SDTM domains and ADaM datasets by study, with NSD status flagged | Standard and non-standard domains listed separately. SUPPQUAL domains called out explicitly |
| Non-Standard Domains | NSD short name, label, originating IG rationale reviewed, planned variables and derivations | Justification required — why no standard domain fits. Reference specific IG sections evaluated and ruled out |
| Planned Deviations | Variable name, IG requirement, nature of deviation, study-specific justification | Generic responses like 'system limitation' are insufficient. Reasons must be tied to actual data collection or analysis design |
| Software and Tools | SAS version, Pinnacle 21 (Veeva) conformance version, DEFINE.xml authoring tool and version | Enables traceability. FDA can reproduce validation checks and understand which rule set was applied |
| Data Timing | Planned data cut date per study, submission timeline, study completion date | Must align with clinical protocol milestones and CSR schedule |
| File Format and DEFINE.xml | SAS transport file (.xpt) confirmation, DEFINE.xml version (2.0 or 2.1), reviewer's guide reference | FDA Technical Conformance Guide specifies XPT format. DEFINE.xml version must match the guide version in effect at time of submission |
Most SDSP problems that generate downstream queries fall into one of four categories:
PMDA made CDISC-formatted electronic data mandatory for applications for new drugs submitted on or after April 1, 2020, per PMDA notification dated March 2019. The applicable standards are J-SDTM (Japan SDTM Implementation Guide) and J-ADaM — both built on CDISC foundations but with Japan-specific extensions.
Before submitting a JNDA (Japanese New Drug Application), sponsors must complete an RS Consultation — the formal pre-submission meeting on electronic research data standardization. Form A is the document submitted at that meeting. Without an RS Consultation on record, the JNDA application will not pass PMDA's administrative screening, regardless of data quality.
| Timing requirement: PMDA recommends requesting the RS Consultation at least 12 months before the planned JNDA submission date. For multi-regional trials where Japan data is a subset, this means starting Form A preparation when Phase 3 SDTM programming begins. |
J-SDTM is not a replacement for SDTM — it is SDTM with PMDA-specific amendments. The differences that affect programming work most directly:
| Content Area | Form A Requirement | Difference from FDA SDSP |
| Standards Catalog | J-SDTM IG version, J-ADaM IG version, CDISC CT (including JCTB) version and date | J-prefix versions reflect PMDA-specific extensions. JCTB (Japan Controlled Terminology Backbone) entries required for Japan-specific coding categories |
| Trial Information | Study code, phase, indication, target population, enrollment countries | Japan-specific: includes ethnicity factor analysis requirement for bridging studies under ICH E5 framework |
| Domain List with Row Estimates | Planned SDTM domains and ADaM datasets per study with estimated row counts | Row count estimates are required in Form A. PMDA uses these to verify data completeness at JNDA screening |
| Non-Standard Domains | NSD name, label, purpose, planned variables — provided in both Japanese and English | Bilingual requirement. Japanese variable label translation is mandatory and must use PMDA-accepted medical terminology |
| Deviation Plan | Planned deviations from J-SDTM or J-ADaM guidance, with justification referencing PMDA guideline section numbers | Must reference specific PMDA electronic data submission guideline sections, not only CDISC IG sections |
| Anonymization Plan | Methodology for handling personally identifiable information under Japan's Act on Protection of Personal Information (APPI) | No FDA equivalent. Explicit description of subject ID masking, date shifting approach, and data handling procedure required |
| RS Consultation Reference | Reference number and date of the RS Consultation meeting at which Form A is submitted | Links Form A to the formal PMDA consultation record. Without this reference, JNDA screening will flag missing consultation history |
Form B is submitted with the JNDA application. It converts the prospective plan in Form A into the actual record of what was submitted. PMDA reviewers use Form B as their first point of reference when beginning the data review — before they open a dataset.
The relationship between Form A and Form B matters practically. If Form B shows datasets, domains, or deviations that were not in Form A, PMDA must determine whether the gap was an oversight or a sign that something changed during the study. Either way, it generates queries. A good change log in Form B answers those questions before they become formal queries.
FDA does not use a 'Form B' by that name. The closest FDA equivalent is the Study Data Reviewer's Guide (SDRG), submitted with the NDA. The SDRG provides dataset-level guidance to FDA reviewers — traceability from ADaM endpoints to CSR tables, primary analysis dataset identification, key variable definitions. Form B's reviewer notes section serves the same function for PMDA.
The coverage mapping:
| Content Area | Form B Requirement | Relationship to Form A |
| Standards Actually Used | Actual J-SDTM IG, J-ADaM IG, and CT versions present in submitted datasets and DEFINE.xml | Must match Form A unless a version change is explicitly documented with justification in the change log |
| Domains Actually Submitted | Final list of SDTM and ADaM datasets with actual row counts from submitted XPT files | PMDA compares against Form A row estimates. Differences exceeding 10% without explanation trigger completeness queries |
| Actual Deviations | All deviations implemented at submission, including any not documented in Form A, each with justification | New deviations not in Form A require explanation of when the decision was made and why it was not communicated at RS Consultation |
| Known Issues Section | Remaining P21 or PMDA validation errors and warnings, listed by rule ID, with explanations | This section has no counterpart in Form A. It is the primary mechanism for pre-empting reviewer queries about P21 findings |
| Data Cut Information | Actual data lock date, data cutoff date, and interim vs. final dataset distinction where applicable | Final dates vs. planned dates from Form A. Any difference must be noted and explained |
| Reviewer Notes | Dataset-specific guidance: traceability to CSR, primary endpoint dataset identification, key variable explanations | Functions as FDA's Study Data Reviewer's Guide (SDRG). No counterpart in Form A |
| Change Log | Summary of all material changes between Form A and Form B, with date and reason for each change | Required section unique to Form B. Absence of a change log when differences exist is itself a deficiency finding |
The table below puts SDSP, Form A, and Form B side-by-side on the dimensions that affect submission planning and programming decisions.
| Feature | SDSP (FDA) | Form A (PMDA) | Form B (PMDA) |
| Submitted When | At IND, or before Phase 2/3 study start; updated at Type B pre-NDA meeting | At RS Consultation meeting (12+ months before planned JNDA) | With the NDA / JNDA application package |
| Purpose | Plan (prospective) | Plan (prospective) | As-submitted record (retrospective) |
| Regulatory Agency | FDA (United States) | PMDA (Japan) | PMDA (Japan) |
| Language | English | Japanese + English (bilingual) | Japanese + English (bilingual) |
| Document Format | FDA template available; structured free-form accepted | Structured PMDA form (published template) | Structured PMDA form (published template) |
| NSD Documentation | Yes — planned NSD domains with IG justification | Yes — planned NSD domains, bilingual labels required | Yes — actual NSD domains submitted with row counts |
| Deviation Documentation | Yes — planned deviations | Yes — planned deviations (reference PMDA guidelines) | Yes — actual deviations, including any new post-Form A |
| Anonymization Section | Not required | Required (APPI compliance) | Required (implementation confirmation) |
| Known Issues Log | Not standard in SDSP (covered in SDRG) | Not in Form A | Required section |
| CT Reference | CDISC CT version + release date | CDISC CT + JCTB version + release date | As-submitted CT version (must match DEFINE.xml) |
| Predecessor Document | None — first formal data standards contact | Pre-consultation notes (informal) | Form A — formal reconciliation required |
For a Phase 2/3 program, the first SDSP goes in with the Phase 2 IND amendment that initiates the pivotal study. At that point it is a plan. It gives FDA the opportunity to flag concerns before data is locked. An FDA comment on SDSP content at IND stage is a far cheaper problem than an information request at NDA review.
For Type B pre-NDA meetings, the SDSP is included in the briefing package as a status document: this is what we planned, this is what we implemented, and these are the known deviations. FDA reviewers at that meeting use the SDSP to assess whether the incoming NDA data package will be reviewable without significant queries.
Based on FDA's published technical rejection patterns and recurring information request content:
The SDSP should be treated as a living document during the IND phase. Update it when:
FDA review teams typically pull the most recent SDSP from the IND file before NDA review begins. An SDSP that is two years out of date relative to the submitted data is visible immediately and will generate reconciliation questions.
The RS Consultation is a formal process with a defined sequence. For a standard JNDA program:
Statistical programmers supporting a Japan submission need to plan translation timelines carefully:
PMDA queries arising from Form A or Form B gaps follow predictable patterns:
The following ten scenarios represent data situations statistical programmers encounter across Phase 2/3 programs. For each, the table identifies the documentation challenge, how it appears in P21 output, and what the agency response looks like with and without proper SDSP or Form A/B coverage.
| # | Data Situation | SDSP / Form A Implication | P21 Flag | Without Documentation | With Documentation |
| 1 | Phase 3 insomnia trial collects continuous actigraphy data from wearables. No standard CDISC domain covers this data type. | Create NSD 'AC' domain. Document variable list, label, OIDM relationship, and derivation in SDSP or Form A. | SD0083: Non-standard domain AC not defined in implementation guide | FDA issues information request; review clock pauses pending clarification of AC domain content | SD0083 expected and explained. Reviewer proceeds to data review without query |
| 2 | WOCF (Worst Observation Carried Forward) imputation applied to primary endpoint. This is not a standard ADaM DTYPE value. | Document custom DTYPE 'WOCF' in SDSP deviation table with SAP section cross-reference and derivation algorithm description. | AD0062: DTYPE value WOCF is not a recognized ADaM value | Query: 'Clarify meaning and derivation of WOCF records in ADEFF.' Review pauses for up to 30 days per cycle. | AD0062 documented. SAP reference included. Reviewer understands imputation method without query. |
| 3 | Two Phase 3 trials pooled into an integrated efficacy summary. Study A used SDTM IG 3.2; Study B used SDTM IG 3.3. | Document both IG versions per study in SDSP standards catalog. Explain pooled analysis handling in ADaM and DEFINE.xml mapping strategy. | CT0009: Controlled terminology version inconsistency detected across studies in integrated submission | FDA raises data integrity concern on pooled dataset traceability. Major amendment with resubmission required. | SDSP version mapping table clarifies traceability per study. No amendment needed; reviewers proceed. |
| 4 | Japan bridging study requires ADEX (Exposure-Response Analysis Dataset) per J-ADaM guidance. This dataset is not in the global ADaM package. | Document ADEX in Form A as a Japan-only ADaM dataset with J-ADaM guideline section reference. | PMDA-JP01: Required ADEX dataset for bridging study not found in submission package | PMDA flags application as incomplete at screening. JNDA not accepted for formal review. | Form A explicitly lists ADEX as Japan-specific. PMDA reviewer locates dataset immediately upon receipt. |
| 5 | Phase 1 multiple-ascending dose study with 6 dose cohorts. ADSL ARM variable cannot meaningfully represent all levels within standard structure. | Document custom ARMCD coding scheme in SDSP. Provide protocol dose table cross-reference. Explain pooling logic. | SD0008: ARMCD value not found in sponsor-defined codelist; codelist extension not declared | FDA requests clarification of dose assignment logic. Data not interpretable until amendment filed. | SD0008 explained by codelist extension declared in SDSP. ARMCD values align with protocol appendix. |
| 6 | Proprietary 23-item patient-reported outcome (PRO) scale does not appear in CDISC controlled terminology. Custom codelist required. | Declare sponsor-defined codelist in DEFINE.xml and reference it in SDSP with instrument validation publication citation. | CT0019: Codelist referenced in DEFINE.xml not found in CDISC CT library | FDA marks PRO data as non-interpretable. Full data re-annotation and DEFINE.xml revision required. | CT0019 documented. DEFINE.xml includes full sponsor codelist. PRO dataset accepted without further action. |
| 7 | Legacy compound acquisition: data originally in SAS v6 formats, converted to SDTM. Legacy studies use SDTM IG 3.1.3; new studies use SDTM IG 3.3. | Document conversion methodology in SDSP. List legacy studies under SDTM IG 3.1.3 with version-split rationale. Note any variable mapping changes. | SD0010: Dataset metadata version inconsistent with expected SDTM IG specification in DEFINE.xml header | FDA cannot validate legacy datasets. Data integrity questions across IG versions require amendment before review continues. | SDSP version mapping addresses legacy data explicitly. FDA accepts multi-version submission with documented rationale. |
| 8 | SUPPQUAL used for a protocol deviation sub-category variable named DVCAT, which conflicts with the standard DV domain variable DVCAT. | Document the SUPPQUAL variable name conflict in SDSP deviation table. Explain that DVCAT in SUPPQUAL carries additional categorization beyond what DV.DVCAT captures. | SD0089: SUPPQUAL contains a variable (DVCAT) that is a standard SDTM variable name | FDA flags potential data duplication or mismatch between DV.DVCAT and SUPPQUAL.DVCAT. Structural query issued. | SD0089 pre-explained in SDSP. Reviewer understands dual-purpose variable design and moves on. |
| 9 | Response-adaptive randomization trial: realized treatment arms diverge from protocol-defined arms because allocation probabilities shift mid-study. | Document adaptive ARM coding methodology in SDSP. Explain how ACTARMCD will reflect realized versus planned arms, and link to adaptive design SAP. | SD0006: ACTARMCD contains values not present in expected value set from protocol arm definitions | FDA queries discrepancy between randomization model documentation and ADSL ARM values. Amendment filed, 3-month delay. | SDSP explains adaptive ARM coding scheme. Reviewer accepts the ARM structure without query. |
| 10 | Multi-regional trial (US + Japan). P21 run with PMDA conformance profile identifies 47 warnings not in the FDA profile, including anonymization and ADEX flags. | SDSP covers FDA-profile deviations. Form A covers PMDA-specific deviations. Both documents cross-reference the same underlying global trial datasets. | PMDA-JP01, PMDA-JP02, CT0019: flagged in PMDA conformance profile only; absent from FDA profile | PMDA rejects JNDA at screening. No RS Consultation on file, Form A not submitted. Application must restart. | Form A pre-documents all PMDA-specific deviations. PMDA screening passes; known issues addressed in Form B. |
Pinnacle 21 (Veeva conformance validation) runs against FDA's published CDISC validation rules and, in the PMDA conformance profile, against PMDA-specific rules. A P21 finding that is documented in the SDSP or Form A/B is an anticipated finding — reviewers see the explanation alongside the flag and move on. A P21 finding with no documentation is an open question that becomes a formal query.
The table below covers common P21 rule categories and their connection to SDSP and PMDA form documentation. Rule IDs follow Pinnacle 21 naming conventions and are representative; exact IDs vary by P21 version.
| Rule ID | Description | Severity | SDSP / Form A Addresses? | Form B Covers? | How to Address in Documentation |
| SD0083 | Non-standard domain not defined in implementation guide | Warning | Yes — NSD section | Yes — actual NSD list | List domain in NSD table with OIDM relationship, variable list, label, and study-specific justification. Form B confirms actual variables match the plan. |
| SD0005 | Required SDTM variable is missing from dataset | Error | Yes — deviations section | Yes — known issues | Document the missing variable, state why it does not apply to this study (e.g., AVAL absent because there is no numeric result in this findings domain). Add DEFINE.xml notes at value level. |
| SD0008 | Variable value not in sponsor-defined or CDISC codelist | Warning | Yes — codelist deviation | Yes — known issues | Extend the codelist in DEFINE.xml with a sponsor-defined codelist. Reference the extension in SDSP. If new values appear after Form A submission, add them to Form B change log with reason. |
| SD0089 | SUPPQUAL contains a variable that is also a standard SDTM variable name | Warning | Yes — deviations section | Yes — known issues | Explain in SDSP why the variable appears in SUPPQUAL rather than the parent domain — typically additional context or sub-classification beyond what the standard variable captures. Name the variable and parent domain explicitly. |
| AD0062 | DTYPE value is not a recognized ADaM DTYPE | Warning | Yes — custom DTYPE deviation | Yes — known issues | Define custom DTYPE in SDSP with SAP section reference and derivation algorithm. In Form B, confirm actual record count and derivation matches the SAP specification. |
| AD0019 | PARAMCD not found in expected datasets or does not follow expected pattern | Error | Yes — ADaM dataset plan | Yes — known issues | Cross-reference PARAMCD list in SDSP to the efficacy endpoints in the SAP and protocol. PMDA Form A specifically requires PARAMCD consistency across all efficacy datasets. |
| CT0001 | Controlled terminology version not found or inconsistent with submission package | Error | Yes — standards catalog | Yes — standards used section | Specify exact CT version (e.g., 2023-03-31) in SDSP. If CT version changes between Form A submission and JNDA, document the change in Form B with the date of change decision. |
| CT0019 | Codelist in DEFINE.xml not found in CDISC CT library | Warning | Yes — sponsor codelist section | Yes — known issues | Distinguish between CDISC codelist extensions and entirely new sponsor codelists. Both require SDSP declaration. PMDA Form A requires bilingual codelist descriptions for any new sponsor codelist. |
| DEFINE0001 | Dataset description or value-level metadata missing from DEFINE.xml | Error | Partial — tool reference | Yes — reviewer notes | SDSP references the DEFINE.xml authoring tool. Form B reviewer notes section must confirm DEFINE.xml version and that value-level metadata is complete for all custom variables and sponsor codelists. |
| SD0010 | DOMAIN variable value is inconsistent with the dataset file name | Error | Yes — domain list (preventive) | Yes — actual domains submitted | This is a structural error caught during QC and fixed before submission. SDSP's domain list is the programming reference that prevents the mismatch. Any remaining instance must be in Form B with a structural explanation. |
| PMDA-JP01 | Required ADEX dataset for J-ADaM bridging study submission not found | Error | Yes — Form A domain list | Yes — Form B domain list | PMDA-specific rule. Form A must list ADEX as a required dataset for any bridging submission under ICH E5. Form B confirms ADEX was generated with actual row count and data lock date. |
| PMDA-JP02 | Datasets contain subject identifiers not masked per APPI requirements | Error | Yes — Form A anonymization plan | Yes — Form B anonymization confirmation | Form A anonymization plan must describe the ID masking method. Form B includes a confirmation statement that the masking was implemented as planned. This error has no FDA equivalent — it is Japan-specific. |
The target is not to eliminate all P21 findings. Some are inherent to the data design and cannot be resolved without compromising data integrity or analysis methodology. The target is to explain every remaining finding before the agency encounters it.
For FDA submissions: the P21 report is typically submitted as part of the NDA package. Cross-reference remaining errors in the SDRG or in DEFINE.xml notes at the value level. The SDSP provides the framework; the SDRG provides the dataset-specific explanation.
For PMDA submissions: Form B's known issues section is the formal location for this explanation. For each remaining error, list the rule ID, affected dataset and variable, and a statement explaining why the finding is acceptable given the data design. PMDA reviewers check this section at the start of data review.
| Practice note for multi-regional trials: Run P21 with both FDA and PMDA conformance profiles during QC. Some warnings appear in only one profile. Form A and Form B must address PMDA-specific warnings that the SDSP will not cover, since the SDSP only references FDA conformance. |
The table below documents rejection and refusal patterns drawn from publicly documented FDA RTF actions, PMDA screening rejection notices, and recurring information request categories. Timeline estimates reflect typical review cycle times as of 2024–2025.
| Rejection Type | Primary Trigger | Agency | Documentation Fix | Timeline Impact If Not Fixed |
| Refuse to File (RTF) | No SDSP in the IND file and no reference to one in the NDA cover letter. Reviewer cannot establish any conformance baseline for the data package. | FDA | Submit SDSP as an IND amendment. For programs past that point, include SDSP status at the Type B pre-NDA meeting. | 60-day RTF clock plus amendment cycle plus re-submission. Estimated 4–6 month delay minimum. |
| Refuse to File (RTF) | SDSP states SDTM IG 3.1.3 but submitted datasets and DEFINE.xml header reference SDTM IG 3.3. Version mismatch triggers integrity concern on entire data package. | FDA | Update SDSP version to match DEFINE.xml before submission. Lock CT version and IG version together during data lock QC. | Amendment required before NDA is accepted. Typical NDA review clock reset to start date of complete response. |
| JNDA Screening Rejection | Form A was never submitted and no RS Consultation record exists in PMDA's system. PMDA has no pre-submission reference point. | PMDA | Request RS Consultation at least 12 months before planned JNDA submission. Submit Form A as part of that request. | JNDA rejected at administrative screening. Full RS Consultation process must restart. Adds 12–18 months to JNDA timeline. |
| Major Deficiency (FDA) | Non-standard domains present in submission with no NSD documentation in SDSP. Reviewer cannot interpret custom dataset content or variables. | FDA | Document all NSD in SDSP NSD table with OIDM, variable-level DEFINE.xml metadata, and derivation rationale. This must be in place before the NDA is filed. | Complete response letter issued. Typical response cycle adds 6 months per round. Multiple rounds possible. |
| Data Integrity Query | P21 errors present in submission with no acknowledgment in any submission document. Reviewer cannot determine whether errors were known or undetected by the sponsor. | FDA / PMDA | Form B known issues section lists all remaining P21 errors by rule ID with explanations. SDSP deviation table covers the ones that are structural. SDRG covers dataset-specific notes. | Each uncovered error becomes a potential information request. Typical cycle: 30 days to respond, then reviewer resumes. Multiple queries can stack. |
| Form B / Form A Reconciliation Issue | Form B domain list shows 4 datasets that were not in Form A, with no change log explaining the additions. PMDA cannot determine whether these represent scope expansion or errors. | PMDA | Form B change log must account for every addition, removal, or modification compared to Form A. Each entry needs study-specific justification and date of decision. | PMDA issues completeness query; review paused pending sponsor response. Adds 2–4 months per query cycle. |
| CT Version Mismatch | DEFINE.xml references CDISC CT 2022-06-24. SDSP states CT 2023-03-31. Mismatch in one formal document versus another triggers automated data validation failure. | FDA | CT version in SDSP must match the CT version in DEFINE.xml origin field. Lock CT version as part of data lock procedures, not separately. | Information request with 30-day response window minimum. If full resubmission needed, extends review by one full NDA review cycle (10 months). |
| PMDA Anonymization Failure | Submitted datasets contain SUBJID values that match Japanese hospital registration IDs. Violates Japan's Act on Protection of Personal Information. Reported at PMDA data review. | PMDA | Form A anonymization plan describes ID masking methodology. Form B confirms implementation with a signed verification statement. Run APPI compliance check as part of pre-submission QC. | Application placed on clinical hold pending full re-anonymization and resubmission. Can add 6–12 months depending on data volume. |
FDA reviewers work against PDUFA dates; PMDA reviewers against defined review periods. Every information request issued pauses the formal review clock or consumes reviewer time equivalent to stopping it. Reviewers are motivated to find answers in the submission documents rather than issue requests — but only if the answers are there.
An SDSP or Form A/B that answers the question before the reviewer asks it is directly faster than issuing a request, waiting 30 days, reading the response, and resuming. The challenge is that the submission team cannot know in advance exactly which questions the reviewer would have asked. The solution is comprehensive coverage of everything non-standard in the data.
Documented experience across multiple NDA and BLA submissions shows that thorough SDSP preparation reduces formal information requests related to data standards by 60–80%. The same pattern holds for PMDA with Form A and Form B. This is not a guarantee, but the consistency of the effect across programs makes it a standard practice at experienced sponsors.
| Fix Timing | Typical Action Required | Approximate Timeline Impact |
| Before IND (SDSP / Form A draft stage) | SDSP revision, team review, IG version confirmation | Hours to days. No regulatory timeline impact |
| After IND, before NDA or JNDA | SDSP amendment, sometimes protocol-level alignment on analysis methodology | 1–4 weeks. Minimal impact if resolved before data lock |
| After NDA / JNDA submission (information request) | Complete response amendment, additional P21 run, revised datasets or DEFINE.xml | 1–3 months per cycle. Review clock may slip |
| After RTF or JNDA screening rejection | Full resubmission addressing all identified deficiencies | 6–12 months minimum. PDUFA or PMDA review date resets |
These actions have the highest direct impact on review quality and speed:
SDSP, PMDA Form A, and PMDA Form B serve the same fundamental purpose: establishing a shared understanding of data architecture between the sponsor and the reviewing agency before the review begins. Each document does this at a different point in the regulatory process and with a different level of detail.
For FDA, the SDSP is the IND-stage commitment. It defines what standards will be used, what non-standard elements are planned, and what deviations are justified. A well-maintained SDSP transforms P21 warnings from unexplained flags into acknowledged design choices.
For PMDA, Form A is the mandatory RS Consultation deliverable. Without it, the JNDA will not pass screening. Form B is the implementation record submitted with the application, reconciling plan against reality and pre-empting reviewer queries through the known issues section. Both documents have Japan-specific requirements — J-SDTM extensions, APPI anonymization requirements, JCTB controlled terminology, and bilingual labeling — that have no direct FDA equivalent.
Statistical programmers who treat these documents as active programming artifacts — built alongside the datasets and updated when the datasets change — will encounter fewer agency queries, faster review outcomes, and significantly less amendment work at the NDA or JNDA stage.
Source: clinstandards.org | Based on FDA Technical Conformance Guide (2022), PMDA Electronic Data Submission Guidelines, CDISC SDTM IG 3.3, ADaM IG 1.1, J-SDTM and J-ADaM PMDA guidance documents, and CDISC CT/JCTB current releases.
No comments yet. Be the first!
