USPI, DTS, SmPC, CT.gov, and the Final Push to Drug Approval
Deep Dive Series — CDISC Open Source Ecosystem clinstandards.org | Published: March 2026
Abstract
The period between submission acceptance and drug approval is among the most intense in the life of a clinical program — and statistical programmers are far more central to it than most realize. Beyond the primary NDA/BLA, a constellation of regulatory-driven documents must be produced, reconciled, and maintained: the US Prescribing Information (USPI), Drug Trial Snapshot (DTS), Patient Information (PI/Medication Guide), ClinicalTrials.gov result posting, the EU Summary of Product Characteristics (SmPC), and a series of agency-specific labeling and communication requirements. Each has its own format, statutory deadline, clinical data dependency, and programmer touchpoint. This article explains what each document is, why it exists, what clinical outputs it draws from, what the programmer's specific role is, and what the timing expectations look like. Written entirely from primary regulatory sources.
References
In most clinical programming teams, the months between NDA/BLA submission and PDUFA date are mentally classified as "waiting." In practice, they are nothing of the sort. The regulatory review cycle triggers a cascade of parallel deliverables — some mandatory by statute, some agency-requested, some commercially driven — each requiring verified clinical data drawn from the exact same datasets that supported the submission. The difference is that these outputs are often produced under extreme time pressure, with non-programmer stakeholders (Medical Writing, Regulatory Affairs, Commercial) as primary drivers, and with formatting requirements that look nothing like a conventional TFL.
Programmers who understand this landscape before it arrives are exponentially more effective than those who encounter it reactively. This article is designed to produce the former.
Each document in the closeout package serves a distinct audience and regulatory mandate:
| Document Primary Audience Regulatory Basis Key Clinical Data | |||
| USPI (US Prescribing Information) | Prescribers | 21 CFR Part 201; FDA PLR (Physician Labeling Rule) | Efficacy and safety from pivotal trials; all clinical pharmacology data |
| Drug Trial Snapshot (DTS) | General public | FDA Reauthorization Act 2017 (FDARA); 21 USC 355(d)(7) | Participant demographics, efficacy summary, safety highlights per trial |
| Medication Guide / PPI | Patients | 21 CFR 208; FDA REMS authority | Serious risks, dosing, what to do if adverse event occurs |
| ClinicalTrials.gov Results | General public, researchers | FDAAA 801 (42 USC 282); Final Rule (2016) | Summary results from all registered studies including Phase II/III |
| SmPC (EU) | Prescribers (EU/EEA) | EU Regulation 726/2004; Directive 2001/83/EC | Same pivotal data as USPI but formatted per EMA QRD template |
| RMP Quantitative Annex | EMA | EMA GVP Module V | Safety data, exposure, identified/potential risks |
| JPI (Japan) | Prescribers (Japan) | PMDA MHW regulations | Japanese patient subgroup data; bridging data where applicable |
| Agency Information Requests | FDA reviewers | Ad hoc — issued via Information Request letters | Study-specific; any table, listing, or analysis requested during review |
Table 1.1. The regulatory closeout document web. Each document has a distinct regulatory basis, audience, and clinical data dependency.
Statistical programmers play two distinct roles in closeout:
Data custodian: The submission datasets (ADaM, SDTM) are the single source of truth. Any number in any label document must be traceable to a submission dataset or a protocol-specified analysis. When Regulatory Affairs asks "where does the 68.3% responder rate on page 12 of the label come from?", the programmer must have the answer in under an hour.
Output generator: Most closeout documents require new outputs not produced during the primary submission — subgroup breakdowns, demographic summaries by race/ethnicity, reformatted tables meeting FDA or EMA-specific display requirements, or program-generated text that populates structured label templates. These are programming deliverables as real as any CSR TFL.
NOTE: The source of truth for all label numbers is the submitted ADaM datasets and the final CSR. Any deviation — even rounding — must be explicitly justified and approved by the statistician and regulatory lead. "Close enough" is not acceptable in label negotiation.
The US Prescribing Information (USPI), commonly called the "label" or "package insert," is the primary official document describing a drug's approved indications, clinical pharmacology, efficacy, safety, dosing, and contraindications for US healthcare providers. It is a legally binding regulatory document — every claim must be substantiated by data in the application.
The USPI is governed by the Physician Labeling Rule (PLR), finalized by FDA in 2006 (21 CFR 201.56 and 201.57), which established the standardized PLR format required for all new molecular entities and new biologics. The PLR replaced the older pre-2006 format and introduced the Highlights of Prescribing Information section (the "highlights") as a mandatory front-page summary.
The PLR format mandates the following section structure in this order:
| Section Number Content Programmer Data Source | |||
| Highlights of Prescribing Information | (unnumbered) | Key facts: indication, dosage, warnings, adverse reactions, use in specific populations | Pulled from sections below — must be internally consistent |
| Table of Contents | (unnumbered) | Full section listing | Auto-generated by Medical Writing |
| Indications and Usage | 1 | Approved indication(s) with population definition | Efficacy endpoints from pivotal trial(s) |
| Dosage and Administration | 2 | Recommended dose, route, schedule; dose modifications | Clinical pharmacology / dose-finding data |
| Dosage Forms and Strengths | 3 | Available formulations | Manufacturing / CMC |
| Contraindications | 4 | Conditions where drug must not be used | Safety datasets (ADAE, laboratory data) |
| Warnings and Precautions | 5 | Serious risks requiring monitoring or action | ADAE, ADLB, ADEG, ADCM, specific safety analyses |
| Adverse Reactions | 6 | Adverse reaction tables from clinical trials; postmarketing (if applicable) | ADAE; treatment-emergent AE tables |
| Drug Interactions | 7 | PK/PD interactions | Clinical pharmacology data; ADCM |
| Use in Specific Populations | 8 | Pregnancy, lactation, pediatrics, geriatrics, renal/hepatic impairment | Subgroup analyses from ADaM; dedicated PK studies |
| Drug Abuse and Dependence | 9 | Controlled substance scheduling and abuse potential (if applicable) | Dedicated studies |
| Overdosage | 10 | Signs, management | Clinical data; pharmacology |
| Description | 11 | Chemical/structural information | CMC |
| Clinical Pharmacology | 12 | MOA, PK, PD | Clinical pharmacology datasets (ADPC, ADPP) |
| Nonclinical Toxicology | 13 | Carcinogenicity, reproductive toxicology | Preclinical datasets |
| Clinical Studies | 14 | Key clinical trial results — the most data-intensive section for programmers | Pivotal trial ADaM datasets; CSR tables |
| How Supplied / Storage | 16 | Packaging and storage information | Manufacturing |
| Patient Counseling Information | 17 | What to tell patients | Links to Medication Guide if required |
Table 2.1. PLR USPI section structure with primary programmer data sources.
Section 6 of the USPI is where statistical programmers spend the majority of their label-support time. It requires:
6.1 Clinical Trials Experience — An adverse reaction table listing all TEAEs occurring at ≥5% (or a protocol-defined threshold) in the drug arm, with a comparison to placebo or active comparator. The format is standardized but not templated — FDA negotiates the exact presentation during labeling discussions.
Key requirements for Section 6 adverse reaction tables:
Example Section 6 Table format:
Table 6.1 Adverse Reactions Occurring in ≥5% of Patients Receiving DRUGX and
at a Higher Rate Than Placebo in Study PIVOTAL-001
DRUGX 100 mg PlaceboSystem Organ Class (N=312) (N=156)
Preferred Term % %
─────────────────────────────────────────────────────
Gastrointestinal disorders
Nausea 23 8
Diarrhea 18 6
Vomiting 11 4
General disorders
Fatigue 15 9
Pyrexia 8 3
Investigations
ALT increased 7 2
─────────────────────────────────────────────────────
Figure 2.1. Example Section 6 adverse reaction table format. SOCs are listed alphabetically; PTs are alphabetical within SOC. Note: "at a higher rate than placebo" — PTs below the threshold in the active arm OR with lower rates than placebo are excluded per FDA convention.
WARNING: The denominator for Section 6 must exactly match the safety analysis set denominator stated in the label text. If the label states "safety analysis set (N=312)", the table header must show N=312. Any mismatch — including subjects who received drug but were excluded from the analysis set — will be flagged by FDA during label review. Confirm denominator logic with the statistician before any label table is finalized.
Section 14 is the second major programmer deliverable within the USPI. It presents the pivotal trial results in a format that substantiates the indication statement in Section 1. Key content:
KM Figures for the Label: FDA expects Kaplan-Meier curves in Section 14 that include:
These KM figures must be generated from the submission ADaM datasets (ADTTE) using the exact same analysis as the primary submission. Generating a new KM from a different dataset or with different censoring rules — even if visually identical — is a protocol deviation.
The USPI is not written once and submitted — it goes through multiple rounds of negotiation with FDA:
| Stage Timing Programmer Involvement | ||
| Initial proposed label submitted with NDA/BLA | Day 0 (submission) | Section 6 and 14 tables already drafted; KM figures submitted |
| FDA issues labeling comments / proposed revisions | ~Month 8–10 of review | Programmer may need new analyses to support or rebut FDA's proposed language |
| Sponsor responds to FDA labeling comments | Typically 30–60 days after FDA comments | Revised tables, new subgroup analyses, updated figures |
| Labeling negotiation meeting (if needed) | Month 10–11 | Programmer may need to prepare analyses for real-time discussion |
| Final label agreed | Near PDUFA date | All tables locked; programmer produces final clean outputs |
| Approval — label finalized | PDUFA date (±) | Label published on DailyMed; programmer archives all supporting outputs |
Table 2.2. USPI negotiation timeline and programmer involvement.
NOTE: FDA labeling comments frequently request analyses not pre-specified in the SAP — for example, "provide adverse reaction rates by age group (< 65, ≥ 65)" or "provide the subgroup analysis for patients with prior treatment." These are ad hoc programming requests that must be fulfilled rapidly (often within 5–10 business days). Maintaining clean, well-documented ADaM programs is the only way to execute these efficiently.
The final USPI is published in Structured Product Labeling (SPL) format — an HL7-based XML format required by FDA for all approved drug labels. SPL is the format used in DailyMed (the FDA's official label repository) and is the machine-readable version of the label.
Statistical programmers do not typically author SPL directly — this is done by Regulatory Affairs using specialized SPL authoring tools. However, programmers must ensure that every number in the SPL-rendered label exactly matches the source analysis. Post-approval discrepancies between the SPL label and the submitted analyses are a regulatory compliance issue.
The Drug Trial Snapshot (DTS) is an FDA-mandated plain-language summary of the clinical trials that supported a drug's approval, with specific emphasis on the representation of diverse participant populations — particularly by sex, race, and age. It is published on the FDA website at the time of approval and updated as new labeling supplements are approved.
The legal basis is the FDA Reauthorization Act of 2017 (FDARA), specifically Section 505(d)(7) of the Federal Food, Drug, and Cosmetic Act (21 USC 355(d)(7)), which requires FDA to publish a Drug Trial Snapshot for each new molecular entity (NME) or new biologic approved.
The DTS is a structured web-formatted document, not a PDF. It contains three mandatory components:
Component 1: The Basics
Component 2: The Results
Component 3: Using this Snapshot
The most data-intensive component for programmers is the participant demographics table, which must report:
| Demographic Category Required Breakdowns Source Dataset | ||
| Sex | Male / Female (% of each trial arm) | ADSL (SEX) |
| Age | Median age; % ≥65; % ≥75 | ADSL (AGE, AGEGR1) |
| Race | White / Black or African American / Asian / American Indian or Alaska Native / Native Hawaiian or Other Pacific Islander / Multiracial / Unknown | ADSL (RACE) |
| Ethnicity | Hispanic or Latino / Not Hispanic or Latino / Unknown | ADSL (ETHNIC) |
Table 3.1. DTS demographics requirements. All categories are mandatory; "Unknown/Not Reported" must be shown when applicable — it cannot be rolled into another category or suppressed.
NOTE: FDA is intensely focused on the race/ethnicity reporting in the DTS. The five OMB race categories are mandatory, and "Unknown" must be reported separately. A common programming error is categorizing "Other" races into "Unknown" or aggregating multiracial subjects. The ADSL RACE and ETHNIC variables must align with the FDA's OMB-based categorization scheme. If CDASH collection used different terminology, the mapping must be pre-specified and documented.
The DTS efficacy section reports the primary endpoint result in plain language. For example:
"In the main study, 68% of patients who received DRUGX responded to treatment, compared with 32% of patients who received placebo. This difference was statistically significant."
The 68% and 32% values must exactly match Section 14 of the USPI and the submitted CSR. The DTS is cross-checked against the approved label by FDA's Drug Information team — discrepancies are not tolerated and require a label update (a regulatory action in itself).
| Milestone Timing Responsible Party Programmer Role | |||
| Sponsor drafts DTS | During NDA review cycle (typically Month 6–9) | Medical Writing / Regulatory | Provide demographics table data; verify efficacy numbers match label |
| FDA reviews DTS | During label negotiation period | FDA Drug Information | Respond to queries about numbers |
| DTS finalized | Near PDUFA date | FDA publishes | Final data verification |
| DTS posted | Day of approval | FDA website | Archive supporting programs |
Table 3.2. DTS process timeline. The DTS is typically finalized in parallel with the USPI negotiations.
FDA recognizes two distinct types of patient-directed labeling, governed by different regulations:
| Document Regulatory Basis When Required Audience | |||
| Medication Guide (MedGuide) | 21 CFR Part 208 | When FDA determines the drug poses serious and significant public health concern requiring patient information for safe use | Patients; must be dispensed by pharmacist with every prescription fill |
| Patient Package Insert (PPI) | 21 CFR 201.57(c)(18) | At FDA's discretion or sponsor's proposal | Patients; distributed with drug |
| Instructions for Use (IFU) | Device combination products; biologics | When drug requires a specific administration technique (e.g., auto-injector, implant) | Patients/caregivers |
Table 4.1. Patient-directed labeling types. Medication Guides are the most common and most rigorous.
A Medication Guide must contain the following sections in plain language (8th-grade reading level is the standard target):
The serious side effects section is where statistical programmers contribute: the MedGuide must include a list of adverse reactions that patients should watch for and report immediately. This list is derived from the serious adverse events in the safety datasets (ADAE, filtered on AESER = 'Y' or AE severity = Grade 3/4 under CTCAE grading) and aligned with the Warnings and Precautions section of the USPI.
WARNING: The Medication Guide is a legal document. Its serious adverse reaction list must be consistent with Section 5 (Warnings and Precautions) of the USPI. Any AE category listed in Section 5 must appear in the MedGuide, and vice versa. Programmers who produce the Section 5 safety analyses and the MedGuide supporting data must cross-check these two lists explicitly — inconsistency between the MedGuide and the USPI is a labeling error.
Some drugs require a Risk Evaluation and Mitigation Strategy (REMS), which is a drug safety program required by FDA when a drug's benefits can only outweigh its risks if specific measures are taken to mitigate those risks. REMS programs can include:
Programmer role in REMS: If a REMS program includes mandatory safety monitoring (e.g., regular lab tests, pregnancy testing), the clinical safety data justifying the REMS requirements must be quantified from the submission datasets. The frequency, severity, and reversibility of the safety signal driving the REMS requirement must be documented with specific numbers from ADAE, ADLB, or relevant safety ADaM datasets.
ClinicalTrials.gov results posting is governed by the FDA Amendments Act of 2007 (FDAAA 801), codified at 42 USC 282(j), and operationalized by the NIH Final Rule (42 CFR Part 11), effective January 2017. Non-compliance carries civil monetary penalties of up to $12,608 per day (adjusted for inflation) per violation. This is not a soft requirement — it is federal law with financial penalties that have been enforced.
Key provisions:
The ClinicalTrials.gov results database requires the following structured data elements:
| Results Section Required Data Elements Programmer Source | ||
| Participant Flow | Number enrolled, number completing each period, number analyzed, reasons for discontinuation by arm | ADSL (disposition variables); ADPP |
| Baseline Characteristics | Demographics by arm: age (mean/SD or median/IQR), sex, race, ethnicity, and any baseline disease characteristics specified as primary analysis variables | ADSL |
| Outcome Measures | For each pre-registered outcome: title, description, time frame, unit of measure, result value by arm, statistical analysis (p-value, CI, parameter estimate) | ADaM datasets (ADRS, ADTTE, ADLB, etc.) |
| Adverse Events | All-cause mortality; serious adverse events by preferred term (with n and %) in each arm; other adverse events at ≥5% threshold | ADAE |
| Agreements | Any agreements restricting publication | Legal / Regulatory |
Table 5.1. ClinicalTrials.gov required results data elements and programmer sources.
The Outcome Measures section is where programmers must be most precise. Each pre-registered primary and secondary endpoint must have results posted. The requirements:
For continuous endpoints (e.g., change from baseline in a scale score):
For binary endpoints (e.g., ORR, response rate):
For time-to-event endpoints (e.g., PFS, OS):
Statistical analysis fields:
WARNING: ClinicalTrials.gov results must match the CSR and the USPI. FDA cross-references the posted results against the approved label. A discrepancy — even in rounding or the number of decimal places — is a compliance finding. The FDA's Drug Trial Snapshot also cross-references ClinicalTrials.gov. All three documents (ClinicalTrials.gov, DTS, USPI) must be internally consistent.
The adverse event section on ClinicalTrials.gov has specific requirements that differ from the USPI:
| AE Category ClinicalTrials.gov Requirement USPI Requirement | ||
| Serious Adverse Events | All SAEs by PT, regardless of frequency — number and % in each arm | Not all SAEs listed — only those meeting the labeling threshold or considered drug-related |
| Non-Serious Adverse Events | All TEAEs at ≥5% in any arm — number and % | TEAEs at ≥5% in drug arm and at higher rate than comparator |
| Deaths | All-cause mortality separately reported | Deaths may be in AE table or narrative, depending on labeling |
| Timeframe | Must specify the collection timeframe (e.g., "treatment-emergent AEs from first dose through 30 days after last dose") | Stated in Section 6 header |
Table 5.2. Adverse event reporting comparison: ClinicalTrials.gov vs. USPI. The denominators and inclusion criteria differ.
The entry process for results is typically a shared responsibility:
Programmers should expect to receive a results data shell — a structured spreadsheet or template with field names corresponding to ClinicalTrials.gov data elements — and populate it from the final ADaM datasets. This shell is then entered into the ClinicalTrials.gov Protocol Registration and Results System (PRS) by Regulatory Affairs.
Key field-level precision requirements that trip up programmers:
If a sponsor needs to delay results posting (e.g., to protect proprietary information before approval, or while preparing a new indication application), they must file a certification of delay with ClinicalTrials.gov stating the reason. The maximum allowable delay for an approved drug is until the date of approval or 30 days after, whichever is earlier. After approval, the results must be posted promptly — there is no grace period post-approval for delay.
The Summary of Product Characteristics (SmPC) is the EU equivalent of the USPI. It is the official scientific document summarizing information for healthcare professionals on how to use a medicinal product safely and effectively. The SmPC is mandated by Directive 2001/83/EC (as amended) and EU Regulation 726/2004 for centrally authorized products.
For centrally authorized products (those authorized via the EMA), the SmPC is prepared by the applicant and reviewed by the Committee for Medicinal Products for Human Use (CHMP). The template format is defined in the EMA QRD (Quality Review of Documents) template, currently in version 10.1.
The SmPC and USPI cover similar scientific ground but differ substantially in structure, required content, and terminology:
| Section USPI (PLR) SmPC (QRD) Key Difference | |||
| Indications | Section 1 | Section 4.1 | SmPC requires precise population definition; USPI may be broader |
| Dosage | Section 2 | Section 4.2 | SmPC requires specific dose adjustment tables for renal/hepatic impairment; USPI may reference prescribing information |
| Contraindications | Section 4 | Section 4.3 | SmPC contraindications often more conservative than USPI |
| Warnings | Section 5 | Section 4.4 (Special warnings) | SmPC uses "Special warnings and precautions" terminology |
| Interactions | Section 7 | Section 4.5 | SmPC requires mechanistic basis for interactions |
| Pregnancy/Lactation | Section 8.1–8.2 | Section 4.6 | SmPC requires full preclinical reproductive toxicology context |
| Adverse Reactions | Section 6 | Section 4.8 | SmPC uses MedDRA frequency bands (see Table 6.1) |
| Overdose | Section 10 | Section 4.9 | Similar content |
| PK | Section 12 | Section 5.2 | SmPC requires more complete PK data tables |
| Clinical Studies | Section 14 | Section 5.1 (Pharmacodynamic properties) | SmPC places clinical efficacy data in Section 5.1 |
| Preclinical | Section 13 | Section 5.3 | SmPC requires more complete nonclinical safety summary |
Table 6.1. USPI vs. SmPC structural comparison.
The SmPC Section 4.8 adverse reactions section uses MedDRA frequency bands — a standardized system for categorizing adverse reaction frequency. This is distinct from the USPI, which presents raw percentages:
| Frequency Band Definition Percentage Equivalent | ||
| Very common | ≥1/10 | ≥10% |
| Common | ≥1/100 to <1/10 | ≥1% to <10% |
| Uncommon | ≥1/1,000 to <1/100 | ≥0.1% to <1% |
| Rare | ≥1/10,000 to <1/1,000 | ≥0.01% to <0.1% |
| Very rare | <1/10,000 | <0.01% |
| Not known | Cannot be estimated from available data | — |
Table 6.2. MedDRA frequency bands for SmPC Section 4.8. Programmer task: categorize each adverse reaction's observed rate into the appropriate frequency band.
Programmer task for SmPC Section 4.8:
The SmPC AE table presents PTs organized by SOC, with the frequency band replacing the percentage. Programmers must:
A common error: using the total exposure-adjusted rate (events per patient-year) rather than the subject-level rate (% of subjects with ≥1 occurrence) for frequency band assignment. The frequency bands are based on subject-level rates.
The CHMP review process for a centrally authorized application (CAP) operates on a 210-day clock with stops. Key programming touchpoints:
| Review Milestone Day Programmer Involvement | ||
| Start of procedure | Day 0 | Application submitted; datasets locked |
| Day 120 List of Questions (LoQ) | Day 120 | Responses to CHMP questions — may require new analyses |
| Day 120 responses submitted | Day 150 (clock stop) | Major programming effort for additional analyses |
| Day 150 (clock restart) | Day 150 | CHMP reviews responses |
| Day 180 List of Outstanding Issues (LoOI) | Day 180 | Final opportunity to address CHMP concerns |
| Day 180 responses / Oral Explanation | Day 210 | Brief additional analyses if needed |
| CHMP Opinion | Day 210 | Label finalized; SmPC agreed |
| EC Decision | Day 210 + 67 days | Marketing Authorization granted |
Table 6.3. CHMP 210-day review clock and programmer involvement milestones.
NOTE: The Day 120 and Day 180 responses are the EU equivalent of FDA's Information Requests — they require new or revised analyses, often within tight timelines (typically 30 days for Day 120 responses, shorter for Day 180). Programs must be well-documented and immediately re-executable against the locked datasets.
EMA may request specific output formats not required by FDA:
The Risk Management Plan (RMP) is a comprehensive document required for all EU marketing authorization applications. It documents what is known and unknown about the safety of a medicine and how any risks will be monitored and minimized. The regulatory basis is EMA GVP (Good Pharmacovigilance Practices) Module V.
The RMP contains a quantitative safety annex — Part II: Safety Specification, Module SII — which is the programmer-relevant component. It requires structured quantitative summaries of safety data.
| RMP Section Required Data Source Dataset Programmer Notes | |||
| Exposure summary | Total patient-years of exposure by dose, duration, age, sex | ADSL, ADEX (exposure dataset) | Patient-years = sum of actual treatment duration in years across all subjects |
| Identified risks | Incidence rates for each identified risk (serious AEs driving the risk profile) | ADAE | Compute incidence per 100 patient-years, not just % of subjects |
| Potential risks | Background rate context; observed vs. expected comparisons | ADAE + external databases | May require SIR (Standardized Incidence Ratio) calculations |
| Missing information | Data gaps by population (pediatric, pregnancy, renal impairment) | ADSL subgroups | Quantify the n exposed in each subgroup to document gaps |
| Efficacy in special populations | Subgroup efficacy data where available | ADaM (ADRS, ADTTE) subgroups | Same analyses as USPI Section 8 but formatted per RMP requirements |
Table 7.1. RMP quantitative annex programmer data requirements.
The RMP typically requires adverse event rates expressed as exposure-adjusted incidence rates (EAIRs) rather than the subject-level percentages used in the USPI. The formula:
EAIR (per 100 patient-years) = (Number of subjects with ≥1 event / Total patient-years of exposure) × 100
where: Total patient-years = sum(treatment duration in days) / 365.25 across all subjects
This requires the ADEX (exposure) dataset, which tracks actual drug exposure duration per subject. If ADEX does not exist or exposure duration is not directly available, it can be derived from ADCM (concomitant medications) or from dose records in the protocol-specific datasets.
The Japanese Package Insert (JPI), known as 添付文書 (Tenpusho), is the official prescribing information document for drugs approved in Japan by the Pharmaceuticals and Medical Devices Agency (PMDA). It follows a structure defined by Ministry of Health, Labour and Welfare (MHLW) guidelines, which was substantially revised in 2019 to align more closely with ICH E3 and international norms.
Key structural elements of the JPI relevant to programmers:
| JPI Section Content Programmer Relevance | ||
| Clinical Pharmacology | PK/PD data; includes Japanese vs. non-Japanese PK comparison if bridging study conducted | PK datasets; bridging analysis |
| Indications | Approved indication in Japan (may differ from US/EU) | Japanese-specific subgroup or bridging trial results |
| Dosage and Administration | Japanese-approved dose (may differ) | Dose-finding data from Japanese-specific studies |
| Adverse Reactions | Japanese clinical trial AE data separately reported | Japanese subpopulation ADAE analysis |
| Clinical Studies | Japanese bridging study or Japanese subgroup analysis | Japan-specific ADaM subsets or dedicated Japanese study datasets |
Table 8.1. JPI sections with programming relevance.
Japan has historically required a bridging strategy to establish that international clinical data can be extrapolated to the Japanese population. The E5 ICH guideline governs this. In the modern era, PMDA increasingly accepts global studies with Japanese subgroup data as sufficient for approval (the multi-regional clinical trial, or MRCT, approach per ICH E17).
Programmer implications of Japanese bridging:
NOTE: PMDA expects Analysis Results Metadata (ARM) — the structured metadata describing how analyses were performed — in the submission. This is distinct from FDA, where ARM is optional (though expected by PMDA as noted in CDISC ARM specifications). If ARM is being produced for PMDA, programmers must ensure the ARM metadata accurately reflects the actual analysis programs, including Japanese subgroup analyses.
During the review period, FDA may issue several types of formal communication that require rapid programming responses:
| Communication Type Timing Content Response Deadline Programmer Implication | ||||
| Information Request (IR) | Any point during review | Specific questions requiring additional data, analyses, or clarifications | Typically 14–30 business days | May require new TFLs, subgroup analyses, or dataset queries |
| Complete Response Letter (CRL) | After PDUFA date if not approved | FDA's reasons for non-approval; may require new studies or major additional analyses | Sponsor determines timeline for resubmission | Can require substantial new analysis; effectively restarts the clock |
| Labeling Comments | Months 8–10 of review | FDA's proposed changes to the label; may include requests for label-specific tables or data | 30–60 days typically | Direct programming deliverable |
| Discipline Review Letters | During review | Division-specific questions (Clinical, Statistics, Clinical Pharmacology divisions each may issue separately) | Varies | Statistics division IRs are most directly programming-relevant |
| Advisory Committee Questions | Before AdComm meeting | Data packages prepared for committee review | Days to weeks | High-priority; FDA may need specific analyses formatted for committee |
Table 9.1. FDA communication types during review and programmer response requirements.
When an IR arrives that requires new analysis, the following protocol should be applied:
When FDA convenes an Advisory Committee to review a NDA/BLA, the agency prepares briefing documents from the submitted datasets. Sponsors also prepare their own briefing documents. These packages often require:
NOTE: AdComm presentations require FDA prior approval of the materials. Sponsor briefing documents are submitted to FDA before the committee meeting and reviewed for accuracy. Every number in an AdComm presentation package must be traceable to the submission datasets with the same rigor as a label number.
The following timeline maps the major closeout deliverables against the FDA standard 12-month review clock. Accelerated Approval and Priority Review (6-month) timelines compress this significantly.
| Month FDA Review Activity Programmer Deliverable | ||
| 0 | NDA/BLA submitted; filing review begins | Submission datasets locked; archive created |
| 2 | Filing decision (Refuse to File or accepted) | If RTF: immediate remediation. If accepted: preparation mode. |
| 3–4 | Clinical, statistical, pharmacology reviews begin | Be available for IR; update DTS demographics from ADSL |
| 5–6 | Midcycle review meeting (FDA internal) | Potential early IRs from Statistics division |
| 6–7 | AdComm meeting (if required) | AdComm data package; benefit-risk tables; forest plots |
| 8–9 | Labeling negotiations begin; FDA sends proposed label | USPI Section 6 and 14 tables revised; KM figures updated |
| 8–9 | DTS draft finalized | Demographics tables finalized; efficacy numbers verified |
| 9–10 | ClinicalTrials.gov results posting (if 12 months post-PCD reached) | CT.gov results data shell populated and submitted to Reg Affairs |
| 10–11 | Labeling negotiations (continued); IR responses | Additional subgroup analyses; label tables finalized |
| 11 | Pre-approval inspection (if manufacturing) | No programming role unless data queries arise |
| 11.5 | Final label agreed | All label tables and figures locked; SPL production begins |
| 12 (PDUFA date) | Approval decision | DTS posted; Medication Guide printed; JPI/SmPC synchronized |
| 12+ | Post-approval | Archive programs; prepare for label supplements; RMP updates |
Table 10.1. 12-month NDA/BLA review timeline with programmer deliverable mapping. Priority Review compresses Months 1–6; timelines are indicative.
| Day CHMP Activity Programmer Deliverable | ||
| Day 0 | Application submitted | Datasets locked |
| Day 70 | Preliminary assessment report circulated | — |
| Day 120 | List of Questions (LoQ) issued | MAJOR: new analyses for LoQ responses within 30 days |
| Day 121–150 | Clock stop — sponsor prepares responses | Intensive programming period |
| Day 150 | Responses submitted; clock restarts | — |
| Day 180 | List of Outstanding Issues (LoOI) issued | Minor additional analyses; oral explanation preparation |
| Day 210 | CHMP Opinion | SmPC finalized; EPAR data confirmed |
| Day 210 + 67 | EC Decision — Marketing Authorization granted | SmPC published; CT.gov synchronized |
Table 10.2. CHMP 210-day review clock with programmer touchpoints.
ClinicalTrials.gov deadlines require precise date calculation:
Primary Completion Date (PCD) = date last subject examined/receives intervention for primary endpoint
(pre-registered in the protocol)
Results Posting Deadline = PCD + 12 months (standard)
= PCD + 30 days after approval (if delay certification filed)
For a drug approved on 01-Jan-2026:
If PCD was 01-Jul-2024:
Standard deadline: 01-Jul-2025 (already passed → must post immediately)
Post-approval: within 30 days of approval = 31-Jan-2026
Figure 10.1. ClinicalTrials.gov deadline calculation. The 12-month clock from PCD runs independently of the FDA review clock. Studies where the PCD was more than 12 months before the PDUFA date require posting regardless of approval status.
The following checklist should be completed before the PDUFA date arrives:
| Task Responsible Status Tracking | ||
| Dataset Archive | ||
| Confirm all submission ADaM and SDTM datasets are archived and version-controlled | Lead Programmer | Completed pre-submission |
| Document all define.xml variable definitions to facilitate ad hoc queries | Lead Programmer | Pre-submission |
| Confirm all programs are archived with run logs and reproducibility confirmed | Lead Programmer | Pre-submission |
| USPI Support | ||
| Finalize Section 6 AE tables with current lock denominator | Programmer | Ongoing |
| Generate Section 14 efficacy summary tables aligned with CSR | Programmer | Ongoing |
| Produce KM figures with number-at-risk table per FDA formatting requirements | Programmer | Ongoing |
| Cross-check all label numbers against CSR source tables | Statistician + Programmer | Pre-PDUFA |
| Produce subgroup analyses for Use in Specific Populations (Section 8) | Programmer | Per SAP |
| DTS | ||
| Generate demographics table by arm (race/ethnicity per OMB categories) | Programmer | Month 6–9 |
| Verify DTS efficacy numbers match Section 14 exactly | Programmer | Pre-PDUFA |
| ClinicalTrials.gov | ||
| Calculate PCD and results posting deadline | Regulatory Affairs + Programmer | Pre-submission |
| Complete CT.gov results data shell (all outcome measures, AE tables, participant flow) | Programmer | Per deadline |
| Verify CT.gov results match CSR and label | Statistician + Programmer | Pre-posting |
| SmPC / RMP (EU) | ||
| Generate Section 4.8 frequency band table | Programmer | During CHMP review |
| Compute exposure-adjusted incidence rates for RMP Annex | Programmer | During CHMP review |
| Prepare Japanese subgroup analyses | Programmer (if global study) | Per PMDA timeline |
| IR Readiness | ||
| Maintain executable programs (no manual steps, no hardcoded paths) | All Programmers | Ongoing |
| Document runtime dependencies (SAS version, macro libraries, format catalogs) | Lead Programmer | Pre-PDUFA |
| Establish IR response SOP (triage → development → QC → sign-off) | Lead Programmer + Mgmt | Pre-PDUFA |
Table 11.1. Statistical programmer closeout readiness checklist.
The single most important practice during the closeout phase is cross-document data reconciliation — ensuring that every number referencing clinical data is consistent across all documents:
| Number Must Be Consistent In | |
| Subject counts (N per arm) | USPI Section 6, USPI Section 14, DTS, CT.gov Participant Flow, SmPC Section 4.8 header, RMP exposure |
| Primary endpoint result | USPI Section 14, DTS efficacy, CT.gov Outcome Measures, any AdComm package |
| Adverse reaction rates | USPI Section 6, SmPC Section 4.8 (frequency band must match raw %), CT.gov AE table, MedGuide serious AE list |
| Demographic breakdown | DTS, CT.gov Baseline Characteristics, USPI Section 8 specific population data |
| Deaths | USPI Section 6 (if listed), CT.gov all-cause mortality, RMP identified risks |
Table 11.2. Cross-document data reconciliation requirements. Every cell in this table represents a potential discrepancy finding during regulatory review.
A practical approach: maintain a master numbers table — a single spreadsheet maintained by the lead programmer or biostatistician that lists every key efficacy and safety number, its source (dataset + program + output table reference), and its current value. Every document that references that number is checked against the master table before submission.
| Error Document Affected Root Cause Prevention | |||
| Label N does not match submission safety analysis set N | USPI Section 6, SmPC 4.8 | Late protocol deviations excluded post-lock; wrong population flag | Confirm denominator with statistician; trace to ADSL flag (SAFFL='Y') |
| Race/ethnicity "Other" category rolled up incorrectly | DTS, CT.gov | ADSL RACE variable contains non-OMB categories | Map RACE to OMB categories explicitly in a spec; validate against CRF data |
| ClinicalTrials.gov p-value entered as "<0.001" instead of exact value | CT.gov | Manual entry error; field accepts text | Extract p-value programmatically; provide to Reg Affairs as numeric to 4 decimal places |
| SmPC frequency band wrong because denominator included subjects from non-EU countries | SmPC Section 4.8 | Global study; EU subgroup rate differs from overall | Confirm whether SmPC uses global or EU-only rates; if EU-only, generate EU-subset AE tables |
| DOR median in label uses different censoring rules than CT.gov | USPI Section 14, CT.gov | Different analysis programs used for different outputs | Both must use the primary sensitivity analysis (ANL01FL); trace both to same ADTTE record |
| KM figure in USPI uses different number-at-risk intervals than FDA requested | USPI Section 14 | Communication gap between Medical Writing and Programming | Clarify FDA's preferred intervals (e.g., every 3 months vs. every visit); regenerate figure |
| MedGuide serious AEs list does not include an AE in USPI Section 5 | MedGuide, USPI | Medical Writing produced MedGuide independently | Systematic cross-check of Section 5 → MedGuide before finalization |
| CT.gov results posted before lock datasets are confirmed | CT.gov | Timing pressure; results entered from draft analysis | Confirm CT.gov data shell is sourced from final, locked ADaM only |
| PMDA receives Japanese subgroup analysis with different analysis set than pre-specified | JPI, PMDA dossier | Japan flag added to ADSL after primary SAP locked | Include Japan subgroup flag (COUNTRY='JPN') in primary SAP; confirm pre-lock |
| RMP exposure in patient-years computed from planned (not actual) exposure duration | RMP Annex | ADEX not available; TRTSDT/TRTEDT used without validating against actual dose records | Use actual treatment dates from ADEX or drug administration CRF; validate against ADSL TRTSDT/TRTEDT |
Table 12.1. Common closeout phase errors with prevention strategies.
The most systemic error pattern in the closeout phase is not any individual calculation mistake — it is the fragmentation of outputs across multiple teams, each maintaining their own version of a number. Medical Writing has a draft label table from Month 6 that has not been updated since the database lock. Regulatory Affairs is pulling numbers from a PowerPoint from the Phase III advisory board. The programmer has the correct final table from the locked datasets, but nobody has reconciled all three.
The prevention is organizational as much as technical: establish a single-source principle from the outset of the closeout phase. The lead programmer and lead biostatistician own the master numbers table. Every document that references a clinical data number is reconciled against that table before any regulatory submission.
The regulatory closeout phase is not a denouement — it is a distinct and demanding phase of clinical programming with its own deliverables, standards, and failure modes. These are the essential takeaways:
Published by clinstandards.org — Technical Reference for the CDISC Statistical Programming Community. All content is based on primary regulatory and CDISC sources.
No comments yet. Be the first!