[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EBX5Y1TM0M6WFDW8MDE2BA":3,"skill-same:01M7EBX5Y1TM0M6WFDW8MDE2BA":62},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":31,"commitSha":32,"skillDirectory":33,"forksCount":14,"repositoryHtmlUrl":34,"repositoryHomepageUrl":34,"ownerHtmlUrl":35,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":36,"install":37,"resources":49},"01M7EBX5Y1TM0M6WFDW8MDE2BA","bio-clinical-biostatistics-cdisc-data",null,"Reads, validates, and prepares CDISC SDTM and ADaM clinical trial data for analysis. Covers SDTM domain joins (DM, AE, EX, VS, LB, DS), ADaM architecture (ADSL, BDS, OCCDS, ADTTE) with traceability, treatment-emergent AE conventions, baseline derivation, SUPPQUAL\u002FNSV handling, Define-XML 2.1, and Pinnacle 21 \u002F CORE validation. Use when working with clinical trial datasets in CDISC SDTM\u002FADaM format, preparing analysis-ready data, or validating for regulatory submission.","Mr-Nobody-Anonymous\u002FAll-skills","Mr-Nobody-Anonymous","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F184253119?v=4","01M7AEXFJHKTYQW5R0ZCQQTWFV","anonymous","01M7C00M3GQR4QKP9GZA42GFAK",0,"MIT","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-cdisc-data\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: bio-clinical-biostatistics-cdisc-data\ndescription: Reads, validates, and prepares CDISC SDTM and ADaM clinical trial data for analysis. Covers SDTM domain joins (DM, AE, EX, VS, LB, DS), ADaM architecture (ADSL, BDS, OCCDS, ADTTE) with traceability, treatment-emergent AE conventions, baseline derivation, SUPPQUAL\u002FNSV handling, Define-XML 2.1, and Pinnacle 21 \u002F CORE validation. Use when working with clinical trial datasets in CDISC SDTM\u002FADaM format, preparing analysis-ready data, or validating for regulatory submission.\ntool_type: python\nprimary_tool: pyreadstat\ngoal_approach_exempt: true\nsource: \"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills\"\nsource_repository: \"GPTomics\u002FbioSkills\"\nsource_path: \"clinical-biostatistics\u002Fcdisc-data-handling\u002FSKILL.md\"\nlicense: \"MIT\"\nimported_at: \"2026-09-20\"\n---\n\n## Version Compatibility\n\nReference examples tested with: pyreadstat 1.2+, pandas 2.1+, numpy 1.26+. CDISC standards referenced: SDTM 2.0 \u002F SDTMIG 3.4 (SDTM 3.0 \u002F SDTMIG 4.0 in public review through April 2026); ADaMIG v1.3 (2021); OCCDS v1.1 (Nov 2021); BDS-for-TTE v1.0; Define-XML 2.1 (FDA-recommended for studies starting on\u002Fafter March 15, 2023); Dataset-JSON v1.1 (Dec 2024; FDA Federal Register notice April 2025); Pinnacle 21 Community 4.0+; CORE (CDISC Open Rules Engine, 2021). Define-XML 2.1 FDA support began March 15, 2021 and is required for studies starting on\u002Fafter March 15, 2023.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- R packages cited (essential for ADaM derivation): admiral (Roche\u002Fopenpharma), metacore, metatools, xportr\n\nIf code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# CDISC SDTM and ADaM Data Handling\n\n**\"Load clinical trial data\"** -> Parse CDISC SDTM domain files; build or consume ADaM analysis-ready datasets; preserve subject-level and event-level structure; respect traceability and validation expectations for regulatory submission.\n- Python: `pyreadstat.read_xport()`, `pd.read_sas()`, `pd.merge()`\n- R: `haven::read_xpt()`, `admiral` for ADaM derivation, `Pinnacle21` or `CORE` for validation\n\n## Aggregation Strategy Taxonomy -- Choose the Right Question\n\n| Strategy | Scientific question answered | Example endpoint | Fails when |\n|----------|------------------------------|------------------|------------|\n| Any event (binary) | Does treatment change probability of experiencing the event at all? | Had any serious AE: Yes\u002FNo | Treatment changes event burden but not anyone-event probability |\n| Event count | Does treatment change burden of events per patient? | Total AE count per subject | Subjects with 1 vs 10 events treated equivalently |\n| Maximum severity | Does treatment shift patients toward more severe manifestations? | Worst AESEV per subject | Confounded with event count (more events -> higher chance of severe) |\n| First event + time | Does treatment delay onset of the event? | Time to first serious AE (TTE) | Multiple events per subject ignored |\n| Rate (events per person-time) | What is the per-time-unit rate? | AEs per subject-year | Requires exposure-time tracking; differential dropout biases rates |\n| Composite (per ICH E9 R1) | Event becomes part of endpoint definition | Death = treatment failure | Direction of components conflict; needs hierarchy |\n\nThese are NOT interchangeable. A drug might not change the proportion with AEs (binary: no effect) but increase events per patient (count: harmful). **The choice must be pre-specified in the SAP based on the scientific question, not analytic convenience.**\n\n## Decision Tree by Scenario\n\n| Scenario | Recommended aggregation | Why |\n|----------|------------------------|-----|\n| Primary safety endpoint, single SAE event | Any event (binary); analyse with logistic | Standard regulatory; cite FDA Safety Reporting Guidance |\n| Total adverse-event burden across study | Event count per subject; analyse with Poisson or negative binomial | Captures all events; sandwich SE recommended |\n| Toxicity grade comparison across arms | Max severity per subject; ordinal logistic with PO check | Preserves grade ordering; cite Brant test for PO |\n| Time-to-first AE (Kaplan-Meier visualisation) | First event + time; censor non-events | See clinical-biostatistics\u002Fsurvival-analysis |\n| Rate of exacerbations per patient-year | Rate via negative binomial with offset for exposure time | Standard in COPD\u002Fasthma trials |\n| Composite endpoint (e.g., MACE) | Component-level definition with hierarchy | Pre-specify per ICH E9(R1) composite strategy |\n| Stratification factor extraction | Use STRATA1, STRATA2 from RANDB or DM SUPP | Must appear in analysis (Kahan-Morris 2012) |\n| Baseline value derivation | VSBLFL='Y' \u002F LBBLFL='Y'; derive latest pre-dose only if flag missing | Trust SDTM flag when present |\n\n## SDTM vs ADaM -- The Regulatory Layer Cake\n\n| Layer | Standard | Purpose | Granularity | Examples |\n|-------|----------|---------|-------------|----------|\n| Source CRF | EDC system | Raw data capture | Form\u002Fpage | Rave, Medidata, Veeva |\n| SDTM | SDTM 2.0 \u002F SDTMIG 3.4 | Tabulation; \"what happened\" | One row per observation | DM, AE, EX, VS, LB, DS |\n| ADaM | ADaMIG v1.3 (2021); v3.0 in development | Analysis-ready; \"one-PROC-away from CSR table\" | Subject (ADSL), parameter-timepoint (BDS), occurrence (OCCDS) | ADSL, ADAE, ADLB, ADTTE, ADRS |\n| TLF | Sponsor SAS \u002F R \u002F Python | Tables, listings, figures for CSR | Output | Statistical methods section, demographic table, primary efficacy |\n\n**The ADaM Fundamental Principles (the \"ROT\" document):**\n\n1. Analysis-ready (one procedure call -> the analysis result)\n2. Traceability (every value links back to SDTM via metadata)\n3. Clear\u002Funambiguous communication via Define-XML\n4. Naming conventions (PARAM, PARAMCD, AVAL, AVALC, BASE, CHG, PCHG, ABLFL, ANL01FL, ...)\n5. Structural rules (ADSL one row per subject; BDS one row per subject\u002Fparameter\u002Ftimepoint\u002Fanalysis flag)\n\n**Postdoc reading:** the ADaM IG v1.3 PDF (cdisc.org), ADaM ROT, FDA Study Data Technical Conformance Guide (current 2024 version), Pinnacle 21 validation rule catalog, PHUSE Connect 2023-2025 conference proceedings.\n\n## SDTM Domain Overview\n\n| Domain | Level | Description | Key Variables |\n|--------|-------|-------------|---------------|\n| DM | Subject | Demographics (one row per subject) | USUBJID, ARM, ARMCD, ACTARM, ACTARMCD, AGE, SEX, RACE, RFSTDTC, RFXSTDTC, RFENDTC |\n| AE | Event | Adverse events (multiple per subject) | USUBJID, AETERM, AEDECOD, AEBODSYS, AESEV, AESER, AESTDTC, AEENDTC |\n| EX | Event | Drug exposure\u002Fdosing | USUBJID, EXTRT, EXDOSE, EXSTDTC, EXENDTC |\n| VS | Event | Vital signs | USUBJID, VSTESTCD, VSSTRESN, VSBLFL, VISIT |\n| LB | Event | Lab results | USUBJID, LBTESTCD, LBSTRESN, LBSTRESC, LBORRES, LBBLFL, LBSPEC |\n| DS | Event | Disposition | USUBJID, DSDECOD, DSSTDTC |\n| SE | Event | Subject elements (treatment epochs) | USUBJID, ETCD, SESTDTC, SEENDTC |\n| MH | Event | Medical history | USUBJID, MHDECOD, MHCAT |\n| CM | Event | Concomitant medications | USUBJID, CMDECOD, CMSTDTC, CMENDTC |\n\nUSUBJID = STUDYID-SITEID-SUBJID is the universal merge key. Subject-level domains (DM) have one row per USUBJID; event-level domains have multiple.\n\n**ARM vs ACTARM:** ARM is planned treatment from randomisation; ACTARM is actual treatment received. **In crossover designs, ARM differs from ACTARM by definition;** in parallel-arm trials, they diverge when subjects are randomised to one arm but receive another (per-protocol violations). Primary analyses use ARM (ITT); safety uses ACTARM.\n\n**RFSTDTC vs RFXSTDTC:** RFSTDTC is \"first study activity date\" (typically screening start); RFXSTDTC is \"first treatment date.\" For treatment-emergent adverse event (TEAE) calculations, ALWAYS use RFXSTDTC (per ICH E2A) — RFSTDTC includes screening AEs which are not treatment-emergent.\n\n## Reading .xpt Files\n\n```python\nimport pyreadstat\nimport pandas as pd\n\n# pyreadstat (recommended -- handles SAS metadata)\ndm, meta = pyreadstat.read_xport('dm.xpt')\n# meta.column_names, meta.column_labels, meta.variable_value_labels\n\n# pandas built-in (SAS XPORT v5)\ndm = pd.read_sas('dm.xpt', format='xport', encoding='utf-8')\n\n# CSV fallback (common in academic datasets)\ndm = pd.read_csv('DM.csv')\n```\n\nWhen pyreadstat is available, the metadata object provides column labels, value labels, and format information lost with other readers. **Critical for analysis-dataset derivation:** the metadata carries the controlled-terminology codelist, essential for handling values like AESEV ('MILD'\u002F'MODERATE'\u002F'SEVERE') with semantic ordering.\n\n## SAS XPT v5 vs Dataset-JSON -- The 2025-2026 Transition\n\n**SAS XPT v5** is the current FDA submission format but dates to 1995, with constraints:\n\n- 8-character variable names (so `LBSTRESN` is a max-length name)\n- 200-character text values\n- No UTF-8 (ASCII only) -> problematic for multilingual trials\n- Single dataset per file\n\n**Dataset-JSON v1.1 (CDISC, December 2024; FDA Federal Register notice April 2025)** is the modern replacement. PHUSE-CDISC-FDA pilot has demonstrated drop-in feasibility. FDA adoption timeline pending as of mid-2026; EMA and PMDA exploring in parallel.\n\n**Pragmatic position:** for the next ~2 years, SAS XPT v5 will remain the de facto submission format; sponsors should architect for Dataset-JSON migration but maintain XPT compliance.\n\n## Joining Domains -- The Right Way\n\n```python\nimport pandas as pd\n\ndm = pd.read_csv('DM.csv')\nae = pd.read_csv('AE.csv')\n\n# WRONG: merging event-level directly onto subject-level inflates rows\n# RIGHT: aggregate first, then merge\nany_serious = ae.groupby('USUBJID')['AESER'].apply(lambda x: (x == 'Y').any()).reset_index()\nany_serious.columns = ['USUBJID', 'HAD_SERIOUS_AE']\n\nanalysis = dm.merge(any_serious, on='USUBJID', how='left')\nanalysis['HAD_SERIOUS_AE'] = analysis['HAD_SERIOUS_AE'].fillna(False)\n```\n\n**Always use `how='left'` when merging onto DM** to preserve all randomised subjects, even those with no events. Fill missing event indicators with 0 or False.\n\n### Aggregation strategy must follow the scientific question\n\n| Strategy | Scientific question | Example |\n|----------|---------------------|---------|\n| Any event (binary) | Does treatment increase probability of experiencing the event at all? | Had any serious AE: yes\u002Fno |\n| Event count | Does treatment increase event burden per patient? | Total AE count per subject |\n| Maximum severity | Does treatment shift toward more severe manifestations? | Worst AESEV per subject |\n| First event + time | Does treatment delay onset? | Time to first serious AE |\n| Rate (events per person-time) | What is the per-time-unit rate? | AEs per subject-year |\n\nThese are NOT interchangeable. A drug might not change the proportion with AEs (binary: no effect) but increase events per patient (count: harmful). The choice must follow the SAP, not analytic convenience.\n\n```python\n# Count events per subject\nae_counts = ae.groupby('USUBJID').size().reset_index(name='AE_COUNT')\n\n# Maximum severity per subject (map to numeric first -- string max is unreliable)\nseverity_map = {'MILD': 1, 'MODERATE': 2, 'SEVERE': 3}\nae['AESEV_NUM'] = ae['AESEV'].map(severity_map)\nmax_severity = ae.groupby('USUBJID')['AESEV_NUM'].max().reset_index()\n\n# Specific event: COVID-19 adverse event\ncovid_ae = ae[ae['AEDECOD'] == 'COVID-19']\ncovid_ae['AESEV_NUM'] = covid_ae['AESEV'].map(severity_map)\nhad_covid = covid_ae.groupby('USUBJID')['AESEV_NUM'].max().reset_index()\nhad_covid.columns = ['USUBJID', 'COVID_SEVERITY']\n\nanalysis = dm.merge(had_covid, on='USUBJID', how='left')\nanalysis['HAD_COVID'] = analysis['COVID_SEVERITY'].notna().astype(int)\n```\n\n## ADaM Architecture -- The Postdoc Deep Dive\n\n### ADSL (Subject-Level) -- The Spine\n\n**Exactly one row per subject.** Every other ADaM dataset must merge to ADSL on USUBJID. Standard variables:\n\n- **USUBJID** -- universal subject ID\n- **TRT01A \u002F TRT01P \u002F ACTARMCD \u002F ARMCD** -- planned and actual treatment, period 1\n- **TRTSDT \u002F TRTEDT** -- treatment start\u002Fend dates (derived from EX, not SDTM)\n- **AGE, SEX, RACE, ETHNIC** -- demographics from DM\n- **RANDDT** -- randomisation date\n- **DCSREAS \u002F DCSREASP \u002F DCSREASCD** -- discontinuation reason (coded + verbatim)\n- **Population flags:** ITTFL, FASFL, SAFFL, PPROTFL, EFFFL (Y\u002FN flags for analysis populations)\n- **Stratification factors:** STRATA1, STRATA2 (from randomisation)\n- **Baseline covariates** that will be used as model covariates downstream\n\n### BDS (Basic Data Structure) -- Long Format Analysis Data\n\n**One row per subject per parameter per analysis timepoint per analysis flag.** Used for ADVS, ADLB, ADEFF, ADQS, ADTTE.\n\nRequired variables:\n\n- **USUBJID, STUDYID** -- merge keys\n- **PARAM, PARAMCD, PARAMN** -- parameter name, code, number\n- **AVISIT, AVISITN** -- analysis visit name, number\n- **ADT, ADY** -- analysis date, analysis day (relative to TRTSDT)\n- **AVAL, AVALC** -- analysis value (numeric, character)\n- **BASE** -- baseline value (replicated per subject\u002Fparameter)\n- **CHG, PCHG** -- change from baseline, percent change\n- **ABLFL** -- 'Y' for the baseline record\n- **ANL01FL, ANL02FL** -- analysis flags for primary\u002Fsecondary analyses\n- **DTYPE** -- derivation type ('LOCF', 'WOCF', 'AVERAGE', 'BOCF', or null for original)\n- **BASETYPE** -- when multiple baselines per subject\u002Fparameter (crossover)\n- **EPOCH** -- study period (SCREENING, TREATMENT, FOLLOW-UP)\n\n```python\n# Example: derive ADLB BDS structure from LB SDTM\nimport pandas as pd\n\nlb = pd.read_csv('LB.csv')\nadsl = pd.read_csv('ADSL.csv')\n\n# Filter to active tests\nadlb = lb[lb['LBTESTCD'].isin(['ALT', 'AST', 'CREAT', 'HGB'])].copy()\nadlb['AVAL'] = adlb['LBSTRESN']\nadlb['PARAM'] = adlb['LBTEST']\nadlb['PARAMCD'] = adlb['LBTESTCD']\n\n# Merge subject-level treatment from ADSL\nadlb = adlb.merge(adsl[['USUBJID', 'TRT01A', 'TRTSDT']], on='USUBJID')\n\n# Compute analysis day\nadlb['ADT'] = pd.to_datetime(adlb['LBDTC'], errors='coerce')\nadlb['TRTSDT'] = pd.to_datetime(adlb['TRTSDT'], errors='coerce')\nadlb['ADY'] = (adlb['ADT'] - adlb['TRTSDT']).dt.days + 1  # Day 1 = first dose\n\n# Set ABLFL from LBBLFL\nadlb['ABLFL'] = (adlb['LBBLFL'] == 'Y').map({True: 'Y', False: None})\n\n# Derive BASE per subject\u002Fparameter\nbaselines = adlb[adlb['ABLFL'] == 'Y'][['USUBJID', 'PARAMCD', 'AVAL']]\nbaselines.columns = ['USUBJID', 'PARAMCD', 'BASE']\nadlb = adlb.merge(baselines, on=['USUBJID', 'PARAMCD'], how='left')\n\n# Compute CHG and PCHG\nadlb['CHG'] = adlb['AVAL'] - adlb['BASE']\nadlb['PCHG'] = 100 * adlb['CHG'] \u002F adlb['BASE']\n```\n\n### OCCDS (Occurrence Data Structure) -- One Row per Event\n\nOCCDS v1.1 (Nov 2021) handles AE, CM, MH. Variables: AEDECOD, AEBODSYS, AESEV, AESER, ASTDT (analysis start date), AENDT, TRTEMFL.\n\n**OCCDS v1.1 added TRTEM01FL through TRTEM##FL** for multi-period treatment-emergent flags — essential for crossover and multi-phase studies where a single TRTEMFL is ambiguous.\n\n### ADTTE (Time-to-Event) -- The CNSR Convention Trap\n\nThe **ADaM BDS for TTE v1.0** uses BDS structure with extra variables for survival analysis:\n\n- **STARTDT** -- time origin (typically TRTSDT for OS; RANDDT for PFS; response date for DOR)\n- **ADT** -- analysis date (event date if event, censoring date if censored)\n- **AVAL** = ADT - STARTDT (+1 if \"first day = day 1\" convention)\n- **AVALU** = 'DAYS' (or 'MONTHS' for some endpoints)\n- **CNSR** -- censoring indicator. **CONVENTION: CNSR = 0 for events; positive integers for censoring**, integer encodes censoring reason\n- **EVNTDESC** -- text description ('Death due to disease', 'Last alive contact')\n- **CNSDTDSC, SRCDOM, SRCVAR, SRCSEQ** -- traceability back to SDTM source\n\n**The CNSR convention is OPPOSITE to most statistical packages**, which use 1 = event. R `survival::Surv(time, event)` expects event=1; SAS PROC LIFETEST takes CENSORED= statement that's opposite to CNSR convention. **This is a perpetual bug source.** When passing ADTTE to analysis:\n\n```python\n# Convert CDISC ADTTE CNSR to R\u002FPython statistical convention\nadtte['event'] = (adtte['CNSR'] == 0).astype(int)  # 1 = event for survival packages\n```\n\n### Define-XML 2.1\n\nEvery ADaM dataset requires variable-level metadata in Define-XML 2.1 (FDA-required for studies starting on\u002Fafter March 15, 2023; support began March 15, 2021). Fields per variable:\n\n- **Origin** -- CRF, derived, predecessor SDTM variable\n- **Derivation rule** -- free text or controlled algorithm\n- **Codelist** -- linked controlled terminology\n- **Length, datatype, label**\n\nThe FDA reviewer's Analysis Data Reviewer's Guide (ADRG) is now expected in every NDA\u002FBLA — walks reviewer through how each analysis dataset was built.\n\n**Two-level traceability expectation:** SDTM raw -> ADaM analysis-ready, with no orphan derivations. FDA reviewers explicitly trace AE counts in CSR table -> ADAE rows -> AE SDTM rows. Any break is a flag.\n\n## Treatment-Emergent AE -- The Convention Variation\n\n**ICH E2A (1995)** defines an AE generically. **TEAE is sponsor-defined:**\n\n```\nTRTEMFL = 'Y' if AE.ASTDT >= TRTSDT AND AE.ASTDT \u003C= TRTEDT + X days\n```\n\nX = post-treatment follow-up window. Common values:\n\n- Small molecules: 28 or 30 days\n- Biologics with extended half-life: longer (e.g., 60-90 days for mAbs)\n- Cell\u002Fgene therapy: indefinite (lifelong monitoring expected)\n\n**Sponsor variation:**\n\n- Day-of-first-dose AE included as TEAE (FDA preference) vs excluded (some EMA reviewers)\n- Partial-date imputation: impute day 15 if only month\u002Fyear known, vs censor as missing\n- Worsening of pre-existing AE: flagged via SEV change vs requires new PT (preferred term)\n\n**MedDRA SOC\u002FPT hierarchy:** AEs coded to MedDRA Preferred Terms (PT), grouped by System Organ Class (SOC). Clinically related PTs (e.g., 'Diarrhea' \u002F 'Frequent bowel movements' \u002F 'Loose stools') often combined via Standardized MedDRA Queries (SMQs) or sponsor-defined groupings. ADAE typically carries both AEDECOD (PT) and SMQ\u002Fgroup flags.\n\n## Baseline Derivation\n\n**ABLFL = 'Y'** marks the record whose AVAL becomes BASE for all other records of the same subject\u002Fparameter.\n\n**Standard rule:** last non-missing assessment on or before first dose (TRTSDT). If protocol mandates a specific baseline visit ('Day 1 pre-dose'), that visit's record is flagged.\n\n```python\n# Derive ABLFL when SDTM baseline flag is missing\u002Finconsistent\nimport pandas as pd\n\nvs['VSDTC_dt'] = pd.to_datetime(vs['VSDTC'], errors='coerce')\nvs = vs.merge(adsl[['USUBJID', 'TRTSDT']], on='USUBJID')\nvs['is_pre_treatment'] = vs['VSDTC_dt'] \u003C= pd.to_datetime(vs['TRTSDT'])\n\n# Latest pre-treatment value per subject\u002Fparameter\nbaseline_records = (vs[vs['is_pre_treatment'] & vs['VSSTRESN'].notna()]\n                    .sort_values('VSDTC_dt')\n                    .groupby(['USUBJID', 'VSTESTCD'])\n                    .tail(1))\nbaseline_records['derived_ABLFL'] = 'Y'\n```\n\n**Critical detail:** filter on VSBLFL='Y' (or LBBLFL='Y') as the primary source. Only fall back to derivation when the flag is missing. Trust the SDTM flag when present; CRF-level baseline designation embeds clinical judgement the analyst cannot reconstruct.\n\n**BASETYPE** required when more than one baseline exists per subject\u002Fparameter (crossover studies, multi-period trials). Distinguishes \"Period 1 Baseline\" vs \"Period 2 Baseline.\"\n\n**DTYPE** values per CDISC controlled terminology: 'LOCF' (last observation carried forward), 'WOCF' (worst), 'AVERAGE', 'BOCF' (baseline observation carried forward), null for original. **Pinnacle 21 flags any DTYPE value not in CT.**\n\n## SUPPQUAL and the NSV Transition\n\n**SUPPQUAL (supplemental qualifiers)** is the legacy mechanism for sponsor-defined variables that don't fit standard SDTM domain columns. Long-format QNAM\u002FQVAL pairs:\n\n```python\nsupp = pd.read_sas('suppae.xpt', format='xport', encoding='utf-8')\nsupp_pivot = supp.pivot_table(\n    index='USUBJID', columns='QNAM', values='QVAL', aggfunc='first'\n).reset_index()\nae_enriched = ae.merge(supp_pivot, on='USUBJID', how='left')\n```\n\nFor record-level SUPPQUAL (where IDVAR and IDVARVAL identify specific rows):\n\n```python\nsupp_record = supp[supp['IDVAR'] == 'AESEQ'].copy()\nsupp_record['AESEQ'] = supp_record['IDVARVAL'].astype(float)\nsupp_pivot_record = supp_record.pivot_table(\n    index=['USUBJID', 'AESEQ'], columns='QNAM', values='QVAL', aggfunc='first'\n).reset_index()\nae_enriched = ae.merge(supp_pivot_record, on=['USUBJID', 'AESEQ'], how='left')\n```\n\n**The 2024-2026 SUPP transition:** Therapeutic Area User Guides (TAUGs) increasingly use NS-- domain extensions or **Non-Standard Variables (NSV) Registry**-listed variables directly in the parent domain, instead of QNAM\u002FQVAL pairs in SUPP--. Not a hard deprecation but the direction is clear. The Non-Standard Variables Registry at cdisc.org is the new canonical place to look up sponsor-extension variables.\n\n## Date Handling -- The Partial-Date Reality\n\n```python\ndm['RFSTDT'] = pd.to_datetime(dm['RFSTDTC'], errors='coerce')\nae['AESTDT'] = pd.to_datetime(ae['AESTDTC'], errors='coerce')\nae['AEENDT'] = pd.to_datetime(ae['AEENDTC'], errors='coerce')\n\n# Days from randomization to AE onset\nae_with_ref = ae.merge(dm[['USUBJID', 'RFSTDT']], on='USUBJID')\nae_with_ref['AE_ONSET_DAY'] = (ae_with_ref['AESTDT'] - ae_with_ref['RFSTDT']).dt.days\n```\n\n**SDTM dates are ISO 8601 strings.** Partial dates (e.g., '2023-03' without day) are common. `errors='coerce'` converts these to NaT rather than raising errors. For analysis requiring complete dates, CDISC conventions impute missing day as the 1st for start dates and the last day of the month for end dates, but imputation rules should match the SAP.\n\n**SDTM records include EPOCH** (SCREENING, TREATMENT, FOLLOW-UP). For TEAEs, filter AEs to onset during or after the treatment epoch. Including pre-treatment AEs confounds the treatment effect estimate.\n\n## Validation -- Pinnacle 21 and CORE\n\n**Pinnacle 21 (Certara, formerly OpenCDISC)** is the de facto FDA submission validation standard. Validates against FDA Validation Rules + CDISC IG conformance + Define-XML schema. Severity tiers:\n\n- **Reject** -- submission will not be accepted\n- **Error** -- must justify\n- **Warning** -- should investigate\n\n**FDA Validation Rules** are published quarterly by FDA Office of Translational Sciences; Pinnacle 21 wraps these into its rule engine.\n\n**CORE (CDISC Open Rules Engine, 2021)** is a newer open-source alternative using YAML-defined rules from the CDISC Rules Catalog. Gaining traction but not yet at Pinnacle-21 parity for confirmatory submissions.\n\n```bash\n# Pinnacle 21 Community (free; appropriate for non-pivotal trials)\np21-community validate --rules sdtmig-3.4 --output-dir validation_output study_data\u002F\n```\n\n## Population Flags\n\n| Flag | Source | Purpose |\n|------|--------|---------|\n| ITTFL | DM all randomised | Primary efficacy population (ICH E9 default) |\n| FASFL | ITT minus eligibility failures + no post-baseline | Practical primary (FAS = Full Analysis Set) |\n| SAFFL | EX (received at least one dose) | Safety analysis (AE reporting) |\n| PPROTFL | SE + DS + protocol-violation list | Per-protocol; sensitivity only |\n| EFFFL | Sponsor-defined | Modified ITT variants |\n\n**FAS vs ITT subtlety:** FAS may exclude post-randomisation subjects (ineligibility, no post-baseline efficacy); ITT cannot. Many SAPs equate them; FDA may insist on stricter ITT at submission. Pre-specify both with explicit FAS exclusion criteria in the protocol.\n\n## Missing Data Considerations -- The Clinical Reasoning Layer\n\nBefore any imputation\u002Fcomplete-case decision, **examine the DS (Disposition) domain to tabulate reasons for discontinuation by treatment arm.** If discontinuation rates or reasons differ between arms, missing data is likely informative (MNAR) and standard MMRM-MAR is questionable.\n\n```python\nds = pd.read_csv('DS.csv')\ndropouts = ds[ds['DSDECOD'] != 'COMPLETED']\ndropouts_by_arm = dropouts.merge(adsl[['USUBJID', 'ARM']], on='USUBJID')\ndiscontinuation_reasons = dropouts_by_arm.groupby(['ARM', 'DSDECOD']).size().unstack(fill_value=0)\n```\n\nThis is the data-quality precursor to choosing the estimand strategy in trial-reporting (see ICH E9(R1)) — missing patterns informed by DS drive the choice between treatment-policy, hypothetical, or composite ICE strategies.\n\n## Common Pitfalls\n\n| Pitfall | Symptom | Solution |\n|---------|---------|----------|\n| Event-level merged onto subject-level without aggregation | Row count inflates after merge | Aggregate first, then merge |\n| First chronological record used as baseline | Misclassified baseline | Filter on VSBLFL='Y' \u002F LBBLFL='Y'; derive only if missing |\n| Character (xxORRES) used for analysis | Inconsistent numeric coercion | Use xxSTRESN (numeric standardised); missing xxSTRESN with present xxORRES means 'NOT DONE' or '\u003CLLOQ' |\n| ARM used in safety analysis | Crossover or actual-treatment differs | Use ACTARM for safety; ARM for ITT efficacy |\n| RFSTDTC used as TEAE reference | Includes screening AEs | Use RFXSTDTC (first treatment); cite ICH E2A |\n| ADTTE CNSR confused with stat-pkg convention | Wrong event\u002Fcensoring assignment | CDISC: CNSR=0 means event; convert: `event = (CNSR == 0).astype(int)` |\n| Partial date parsing error | NaT in date column | `pd.to_datetime(..., errors='coerce')` |\n| SUPPQUAL granularity confusion | Wrong rows merged | Check IDVAR before choosing subject vs record-level merge |\n| Non-standard column names treated as standard SDTM | Missing variables | Inspect actual columns; map to semantic roles |\n| Pinnacle 21 not run before submission | FDA reject | Always validate against current SDTMIG and FDA Validation Rules before submission |\n\n### Common non-standard column mappings\n\n| Standard SDTM | Common alternatives | Role |\n|---------------|--------------------|------|\n| ARM \u002F ARMCD | TRTGRP, TRT01P, treatment, group | Treatment assignment |\n| AEDECOD | AEPT, ae_term, preferred_term | AE preferred term |\n| AESEV (text) | AESEV (numeric 1-4), severity, AETOXGR | Severity \u002F toxicity grade |\n| USUBJID | SUBJID, subject_id, patient_id | Subject identifier |\n| RFXSTDTC | trt_start, first_dose_date | First treatment date |\n| LBSTRESN | lab_value_num, result_numeric | Lab numeric result |\n\n## Quantitative Thresholds and Conventions\n\n| Threshold\u002FConvention | Source | Rationale |\n|----------------------|--------|-----------|\n| TEAE window: 28-30 days post-treatment for small molecules | ICH E2A; sponsor convention | Mode of action, half-life inform window |\n| RFXSTDTC for TEAE reference (not RFSTDTC) | ICH E2A | RFSTDTC includes screening; TEAE is post-treatment |\n| ABLFL='Y' for last non-missing pre-dose | CDISC ADaM IG v1.3 | Standard baseline definition; trust SDTM flag |\n| ARM (planned) for ITT efficacy; ACTARM for safety | ICH E9 | Crossover\u002FPP-violation handling |\n| Pinnacle 21 validation before submission | FDA Study Data Technical Conformance Guide | Standard quality gate; reject errors block acceptance |\n| Define-XML 2.1 for studies starting >=March 15, 2023 | FDA Study Data Standards Catalog | Older 2.0 still accepted for prior studies |\n| Dataset-JSON v1.1 (Dec 2024; FDA notice April 2025) | CDISC + FDA Federal Register | Modern replacement for XPT v5; timeline pending |\n| CNSR=0 for events, positive integers for censoring | ADaM BDS-for-TTE v1.0 | OPPOSITE of R `survival` and most stat packages |\n\n## Anticipated Reviewer Pushback\n\n| Pushback | Response |\n|----------|----------|\n| \"RFXSTDTC or RFSTDTC for TEAE?\" | RFXSTDTC per ICH E2A; RFSTDTC would include screening AEs |\n| \"Baseline from VSBLFL or derived?\" | VSBLFL when present; documented derivation rule when missing |\n| \"Pinnacle 21 errors?\" | All errors resolved or justified; warnings reviewed and documented |\n| \"Define-XML 2.1 ADRG provided?\" | Yes — analysis dataset traceability documented to variable level |\n| \"ITT vs FAS reconciliation?\" | Pre-specified in protocol with explicit FAS exclusion criteria |\n| \"OCCDS v1.1 multi-period flags?\" | TRTEM01FL...TRTEM##FL pre-specified for crossover periods |\n| \"ADTTE CNSR convention conversion documented?\" | Explicit: CDISC CNSR=0 means event; convert to event=1 for downstream R\u002FPython |\n\n## References\n\n- CDISC. 2021. Analysis Data Model Implementation Guide (ADaMIG) v1.3.\n- CDISC. 2021. Occurrence Data Structure (OCCDS) v1.1.\n- CDISC. 2012. ADaM Basic Data Structure for Time-to-Event Analyses v1.0.\n- CDISC. 2024. Dataset-JSON v1.1.\n- FDA. 2024. Study Data Technical Conformance Guide.\n- FDA Federal Register Notice. April 2025. Dataset-JSON Pilot Comment Request.\n- ICH. 1995. E2A: Clinical Safety Data Management -- Definitions and Standards for Expedited Reporting.\n- ICH. 1998. E9: Statistical Principles for Clinical Trials.\n- ICH. 2019. E9(R1) Addendum on Estimands and Sensitivity Analysis.\n- Pinnacle 21 (Certara). 2024. Community Edition Validation Rules.\n- PHUSE\u002FCDISC. 2024. Dataset-JSON Pilot Reports.\n\n## Related Skills\n\n- clinical-biostatistics\u002Flogistic-regression - Model binary outcomes from prepared ADaM\u002FSDTM data\n- clinical-biostatistics\u002Ftrial-reporting - Use prepared analysis datasets for ICH E9(R1) estimands and CONSORT 2025 reporting\n- clinical-biostatistics\u002Fmissing-data-sensitivity - DS-domain reasoning informs estimand choice\n- clinical-biostatistics\u002Fsurvival-analysis - ADTTE CNSR convention; time-to-event preparation\n- expression-matrix\u002Fmetadata-joins - General metadata joining patterns\n","## Version Compatibility\n\nReference examples tested with: pyreadstat 1.2+, pandas 2.1+, numpy 1.26+. CDISC standards referenced: SDTM 2.0 \u002F SDTMIG 3.4 (SDTM 3.0 \u002F SDTMIG 4.0 in public review through April 2026); ADaMIG v1.3 (2021); OCCDS v1.1 (Nov 2021); BDS-for-TTE v1.0; Define-XML 2.1 (FDA-recommended for studies starting on\u002Fafter March 15, 2023); Dataset-JSON v1.1 (Dec 2024; FDA Federal Register notice April 2025); Pinnacle 21 Community 4.0+; CORE (CDISC Open Rules Engine, 2021). Define-XML 2.1 FDA support began March 15, 2021 and is required for studies starting on\u002Fafter March 15, 2023.\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- R packages cited (essential for ADaM derivation): admiral (Roche\u002Fopenpharma), metacore, metatools, xportr\n\nIf code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.\n\n# CDISC SDTM and ADaM Data Handling\n\n**\"Load clinical trial data\"** -> Parse CDISC SDTM domain files; build or consume ADaM analysis-ready datasets; preserve subject-level and event-level structure; respect traceability and validation expectations for regulatory submission.\n- Python: `pyreadstat.read_xport()`, `pd.read_sas()`, `pd.merge()`\n- R: `haven::read_xpt()`, `admiral` for ADaM derivation, `Pinnacle21` or `CORE` for validation\n\n## Aggregation Strategy Taxonomy -- Choose the Right Question\n\n| Strategy | Scientific question answered | Example endpoint | Fails when |\n|----------|------------------------------|------------------|------------|\n| Any event (binary) | Does treatment change probability of experiencing the event at all? | Had any serious AE: Yes\u002FNo | Treatment changes event burden but not anyone-event probability |\n| Event count | Does treatment change burden of events per patient? | Total AE count per subject | Subjects with 1 vs 10 events treated equivalently |\n| Maximum severity | Does treatment shift patients toward more severe manifestations? | Worst AESEV per subject | Confounded with event count (more events -> higher chance of severe) |\n| First event + time | Does treatment delay onset of the event? | Time to first serious AE (TTE) | Multiple events per subject ignored |\n| Rate (events per person-time) | What is the per-time-unit rate? | AEs per subject-year | Requires exposure-time tracking; differential dropout biases rates |\n| Composite (per ICH E9 R1) | Event becomes part of endpoint definition | Death = treatment failure | Direction of components conflict; needs hierarchy |\n\nThese are NOT interchangeable. A drug might not change the proportion with AEs (binary: no effect) but increase events per patient (count: harmful). **The choice must be pre-specified in the SAP based on the scientific question, not analytic convenience.**\n\n## Decision Tree by Scenario\n\n| Scenario | Recommended aggregation | Why |\n|----------|------------------------|-----|\n| Primary safety endpoint, single SAE event | Any event (binary); analyse with logistic | Standard regulatory; cite FDA Safety Reporting Guidance |\n| Total adverse-event burden across study | Event count per subject; analyse with Poisson or negative binomial | Captures all events; sandwich SE recommended |\n| Toxicity grade comparison across arms | Max severity per subject; ordinal logistic with PO check | Preserves grade ordering; cite Brant test for PO |\n| Time-to-first AE (Kaplan-Meier visualisation) | First event + time; censor non-events | See clinical-biostatistics\u002Fsurvival-analysis |\n| Rate of exacerbations per patient-year | Rate via negative binomial with offset for exposure time | Standard in COPD\u002Fasthma trials |\n| Composite endpoint (e.g., MACE) | Component-level definition with hierarchy | Pre-specify per ICH E9(R1) composite strategy |\n| Stratification factor extraction | Use STRATA1, STRATA2 from RANDB or DM SUPP | Must appear in analysis (Kahan-Morris 2012) |\n| Baseline value derivation | VSBLFL='Y' \u002F LBBLFL='Y'; derive latest pre-dose only if flag missing | Trust SDTM flag when present |\n\n## SDTM vs ADaM -- The Regulatory Layer Cake\n\n| Layer | Standard | Purpose | Granularity | Examples |\n|-------|----------|---------|-------------|----------|\n| Source CRF | EDC system | Raw data capture | Form\u002Fpage | Rave, Medidata, Veeva |\n| SDTM | SDTM 2.0 \u002F SDTMIG 3.4 | Tabulation; \"what happened\" | One row per observation | DM, AE, EX, VS, LB, DS |\n| ADaM | ADaMIG v1.3 (2021); v3.0 in development | Analysis-ready; \"one-PROC-away from CSR table\" | Subject (ADSL), parameter-timepoint (BDS), occurrence (OCCDS) | ADSL, ADAE, ADLB, ADTTE, ADRS |\n| TLF | Sponsor SAS \u002F R \u002F Python | Tables, listings, figures for CSR | Output | Statistical methods section, demographic table, primary efficacy |\n\n**The ADaM Fundamental Principles (the \"ROT\" document):**\n\n1. Analysis-ready (one procedure call -> the analysis result)\n2. Traceability (every value links back to SDTM via metadata)\n3. Clear\u002Funambiguous communication via Define-XML\n4. Naming conventions (PARAM, PARAMCD, AVAL, AVALC, BASE, CHG, PCHG, ABLFL, ANL01FL, ...)\n5. Structural rules (ADSL one row per subject; BDS one row per subject\u002Fparameter\u002Ftimepoint\u002Fanalysis flag)\n\n**Postdoc reading:** the ADaM IG v1.3 PDF (cdisc.org), ADaM ROT, FDA Study Data Technical Conformance Guide (current 2024 version), Pinnacle 21 validation rule catalog, PHUSE Connect 2023-2025 conference proceedings.\n\n## SDTM Domain Overview\n\n| Domain | Level | Description | Key Variables |\n|--------|-------|-------------|---------------|\n| DM | Subject | Demographics (one row per subject) | USUBJID, ARM, ARMCD, ACTARM, ACTARMCD, AGE, SEX, RACE, RFSTDTC, RFXSTDTC, RFENDTC |\n| AE | Event | Adverse events (multiple per subject) | USUBJID, AETERM, AEDECOD, AEBODSYS, AESEV, AESER, AESTDTC, AEENDTC |\n| EX | Event | Drug exposure\u002Fdosing | USUBJID, EXTRT, EXDOSE, EXSTDTC, EXENDTC |\n| VS | Event | Vital signs | USUBJID, VSTESTCD, VSSTRESN, VSBLFL, VISIT |\n| LB | Event | Lab results | USUBJID, LBTESTCD, LBSTRESN, LBSTRESC, LBORRES, LBBLFL, LBSPEC |\n| DS | Event | Disposition | USUBJID, DSDECOD, DSSTDTC |\n| SE | Event | Subject elements (treatment epochs) | USUBJID, ETCD, SESTDTC, SEENDTC |\n| MH | Event | Medical history | USUBJID, MHDECOD, MHCAT |\n| CM | Event | Concomitant medications | USUBJID, CMDECOD, CMSTDTC, CMENDTC |\n\nUSUBJID = STUDYID-SITEID-SUBJID is the universal merge key. Subject-level domains (DM) have one row per USUBJID; event-level domains have multiple.\n\n**ARM vs ACTARM:** ARM is planned treatment from randomisation; ACTARM is actual treatment received. **In crossover designs, ARM differs from ACTARM by definition;** in parallel-arm trials, they diverge when subjects are randomised to one arm but receive another (per-protocol violations). Primary analyses use ARM (ITT); safety uses ACTARM.\n\n**RFSTDTC vs RFXSTDTC:** RFSTDTC is \"first study activity date\" (typically screening start); RFXSTDTC is \"first treatment date.\" For treatment-emergent adverse event (TEAE) calculations, ALWAYS use RFXSTDTC (per ICH E2A) — RFSTDTC includes screening AEs which are not treatment-emergent.\n\n## Reading .xpt Files\n\n```python\nimport pyreadstat\nimport pandas as pd\n\n# pyreadstat (recommended -- handles SAS metadata)\ndm, meta = pyreadstat.read_xport('dm.xpt')\n# meta.column_names, meta.column_labels, meta.variable_value_labels\n\n# pandas built-in (SAS XPORT v5)\ndm = pd.read_sas('dm.xpt', format='xport', encoding='utf-8')\n\n# CSV fallback (common in academic datasets)\ndm = pd.read_csv('DM.csv')\n```\n\nWhen pyreadstat is available, the metadata object provides column labels, value labels, and format information lost with other readers. **Critical for analysis-dataset derivation:** the metadata carries the controlled-terminology codelist, essential for handling values like AESEV ('MILD'\u002F'MODERATE'\u002F'SEVERE') with semantic ordering.\n\n## SAS XPT v5 vs Dataset-JSON -- The 2025-2026 Transition\n\n**SAS XPT v5** is the current FDA submission format but dates to 1995, with constraints:\n\n- 8-character variable names (so `LBSTRESN` is a max-length name)\n- 200-character text values\n- No UTF-8 (ASCII only) -> problematic for multilingual trials\n- Single dataset per file\n\n**Dataset-JSON v1.1 (CDISC, December 2024; FDA Federal Register notice April 2025)** is the modern replacement. PHUSE-CDISC-FDA pilot has demonstrated drop-in feasibility. FDA adoption timeline pending as of mid-2026; EMA and PMDA exploring in parallel.\n\n**Pragmatic position:** for the next ~2 years, SAS XPT v5 will remain the de facto submission format; sponsors should architect for Dataset-JSON migration but maintain XPT compliance.\n\n## Joining Domains -- The Right Way\n\n```python\nimport pandas as pd\n\ndm = pd.read_csv('DM.csv')\nae = pd.read_csv('AE.csv')\n\n# WRONG: merging event-level directly onto subject-level inflates rows\n# RIGHT: aggregate first, then merge\nany_serious = ae.groupby('USUBJID')['AESER'].apply(lambda x: (x == 'Y').any()).reset_index()\nany_serious.columns = ['USUBJID', 'HAD_SERIOUS_AE']\n\nanalysis = dm.merge(any_serious, on='USUBJID', how='left')\nanalysis['HAD_SERIOUS_AE'] = analysis['HAD_SERIOUS_AE'].fillna(False)\n```\n\n**Always use `how='left'` when merging onto DM** to preserve all randomised subjects, even those with no events. Fill missing event indicators with 0 or False.\n\n### Aggregation strategy must follow the scientific question\n\n| Strategy | Scientific question | Example |\n|----------|---------------------|---------|\n| Any event (binary) | Does treatment increase probability of experiencing the event at all? | Had any serious AE: yes\u002Fno |\n| Event count | Does treatment increase event burden per patient? | Total AE count per subject |\n| Maximum severity | Does treatment shift toward more severe manifestations? | Worst AESEV per subject |\n| First event + time | Does treatment delay onset? | Time to first serious AE |\n| Rate (events per person-time) | What is the per-time-unit rate? | AEs per subject-year |\n\nThese are NOT interchangeable. A drug might not change the proportion with AEs (binary: no effect) but increase events per patient (count: harmful). The choice must follow the SAP, not analytic convenience.\n\n```python\n# Count events per subject\nae_counts = ae.groupby('USUBJID').size().reset_index(name='AE_COUNT')\n\n# Maximum severity per subject (map to numeric first -- string max is unreliable)\nseverity_map = {'MILD': 1, 'MODERATE': 2, 'SEVERE': 3}\nae['AESEV_NUM'] = ae['AESEV'].map(severity_map)\nmax_severity = ae.groupby('USUBJID')['AESEV_NUM'].max().reset_index()\n\n# Specific event: COVID-19 adverse event\ncovid_ae = ae[ae['AEDECOD'] == 'COVID-19']\ncovid_ae['AESEV_NUM'] = covid_ae['AESEV'].map(severity_map)\nhad_covid = covid_ae.groupby('USUBJID')['AESEV_NUM'].max().reset_index()\nhad_covid.columns = ['USUBJID', 'COVID_SEVERITY']\n\nanalysis = dm.merge(had_covid, on='USUBJID', how='left')\nanalysis['HAD_COVID'] = analysis['COVID_SEVERITY'].notna().astype(int)\n```\n\n## ADaM Architecture -- The Postdoc Deep Dive\n\n### ADSL (Subject-Level) -- The Spine\n\n**Exactly one row per subject.** Every other ADaM dataset must merge to ADSL on USUBJID. Standard variables:\n\n- **USUBJID** -- universal subject ID\n- **TRT01A \u002F TRT01P \u002F ACTARMCD \u002F ARMCD** -- planned and actual treatment, period 1\n- **TRTSDT \u002F TRTEDT** -- treatment start\u002Fend dates (derived from EX, not SDTM)\n- **AGE, SEX, RACE, ETHNIC** -- demographics from DM\n- **RANDDT** -- randomisation date\n- **DCSREAS \u002F DCSREASP \u002F DCSREASCD** -- discontinuation reason (coded + verbatim)\n- **Population flags:** ITTFL, FASFL, SAFFL, PPROTFL, EFFFL (Y\u002FN flags for analysis populations)\n- **Stratification factors:** STRATA1, STRATA2 (from randomisation)\n- **Baseline covariates** that will be used as model covariates downstream\n\n### BDS (Basic Data Structure) -- Long Format Analysis Data\n\n**One row per subject per parameter per analysis timepoint per analysis flag.** Used for ADVS, ADLB, ADEFF, ADQS, ADTTE.\n\nRequired variables:\n\n- **USUBJID, STUDYID** -- merge keys\n- **PARAM, PARAMCD, PARAMN** -- parameter name, code, number\n- **AVISIT, AVISITN** -- analysis visit name, number\n- **ADT, ADY** -- analysis date, analysis day (relative to TRTSDT)\n- **AVAL, AVALC** -- analysis value (numeric, character)\n- **BASE** -- baseline value (replicated per subject\u002Fparameter)\n- **CHG, PCHG** -- change from baseline, percent change\n- **ABLFL** -- 'Y' for the baseline record\n- **ANL01FL, ANL02FL** -- analysis flags for primary\u002Fsecondary analyses\n- **DTYPE** -- derivation type ('LOCF', 'WOCF', 'AVERAGE', 'BOCF', or null for original)\n- **BASETYPE** -- when multiple baselines per subject\u002Fparameter (crossover)\n- **EPOCH** -- study period (SCREENING, TREATMENT, FOLLOW-UP)\n\n```python\n# Example: derive ADLB BDS structure from LB SDTM\nimport pandas as pd\n\nlb = pd.read_csv('LB.csv')\nadsl = pd.read_csv('ADSL.csv')\n\n# Filter to active tests\nadlb = lb[lb['LBTESTCD'].isin(['ALT', 'AST', 'CREAT', 'HGB'])].copy()\nadlb['AVAL'] = adlb['LBSTRESN']\nadlb['PARAM'] = adlb['LBTEST']\nadlb['PARAMCD'] = adlb['LBTESTCD']\n\n# Merge subject-level treatment from ADSL\nadlb = adlb.merge(adsl[['USUBJID', 'TRT01A', 'TRTSDT']], on='USUBJID')\n\n# Compute analysis day\nadlb['ADT'] = pd.to_datetime(adlb['LBDTC'], errors='coerce')\nadlb['TRTSDT'] = pd.to_datetime(adlb['TRTSDT'], errors='coerce')\nadlb['ADY'] = (adlb['ADT'] - adlb['TRTSDT']).dt.days + 1  # Day 1 = first dose\n\n# Set ABLFL from LBBLFL\nadlb['ABLFL'] = (adlb['LBBLFL'] == 'Y').map({True: 'Y', False: None})\n\n# Derive BASE per subject\u002Fparameter\nbaselines = adlb[adlb['ABLFL'] == 'Y'][['USUBJID', 'PARAMCD', 'AVAL']]\nbaselines.columns = ['USUBJID', 'PARAMCD', 'BASE']\nadlb = adlb.merge(baselines, on=['USUBJID', 'PARAMCD'], how='left')\n\n# Compute CHG and PCHG\nadlb['CHG'] = adlb['AVAL'] - adlb['BASE']\nadlb['PCHG'] = 100 * adlb['CHG'] \u002F adlb['BASE']\n```\n\n### OCCDS (Occurrence Data Structure) -- One Row per Event\n\nOCCDS v1.1 (Nov 2021) handles AE, CM, MH. Variables: AEDECOD, AEBODSYS, AESEV, AESER, ASTDT (analysis start date), AENDT, TRTEMFL.\n\n**OCCDS v1.1 added TRTEM01FL through TRTEM##FL** for multi-period treatment-emergent flags — essential for crossover and multi-phase studies where a single TRTEMFL is ambiguous.\n\n### ADTTE (Time-to-Event) -- The CNSR Convention Trap\n\nThe **ADaM BDS for TTE v1.0** uses BDS structure with extra variables for survival analysis:\n\n- **STARTDT** -- time origin (typically TRTSDT for OS; RANDDT for PFS; response date for DOR)\n- **ADT** -- analysis date (event date if event, censoring date if censored)\n- **AVAL** = ADT - STARTDT (+1 if \"first day = day 1\" convention)\n- **AVALU** = 'DAYS' (or 'MONTHS' for some endpoints)\n- **CNSR** -- censoring indicator. **CONVENTION: CNSR = 0 for events; positive integers for censoring**, integer encodes censoring reason\n- **EVNTDESC** -- text description ('Death due to disease', 'Last alive contact')\n- **CNSDTDSC, SRCDOM, SRCVAR, SRCSEQ** -- traceability back to SDTM source\n\n**The CNSR convention is OPPOSITE to most statistical packages**, which use 1 = event. R `survival::Surv(time, event)` expects event=1; SAS PROC LIFETEST takes CENSORED= statement that's opposite to CNSR convention. **This is a perpetual bug source.** When passing ADTTE to analysis:\n\n```python\n# Convert CDISC ADTTE CNSR to R\u002FPython statistical convention\nadtte['event'] = (adtte['CNSR'] == 0).astype(int)  # 1 = event for survival packages\n```\n\n### Define-XML 2.1\n\nEvery ADaM dataset requires variable-level metadata in Define-XML 2.1 (FDA-required for studies starting on\u002Fafter March 15, 2023; support began March 15, 2021). Fields per variable:\n\n- **Origin** -- CRF, derived, predecessor SDTM variable\n- **Derivation rule** -- free text or controlled algorithm\n- **Codelist** -- linked controlled terminology\n- **Length, datatype, label**\n\nThe FDA reviewer's Analysis Data Reviewer's Guide (ADRG) is now expected in every NDA\u002FBLA — walks reviewer through how each analysis dataset was built.\n\n**Two-level traceability expectation:** SDTM raw -> ADaM analysis-ready, with no orphan derivations. FDA reviewers explicitly trace AE counts in CSR table -> ADAE rows -> AE SDTM rows. Any break is a flag.\n\n## Treatment-Emergent AE -- The Convention Variation\n\n**ICH E2A (1995)** defines an AE generically. **TEAE is sponsor-defined:**\n\n```\nTRTEMFL = 'Y' if AE.ASTDT >= TRTSDT AND AE.ASTDT \u003C= TRTEDT + X days\n```\n\nX = post-treatment follow-up window. Common values:\n\n- Small molecules: 28 or 30 days\n- Biologics with extended half-life: longer (e.g., 60-90 days for mAbs)\n- Cell\u002Fgene therapy: indefinite (lifelong monitoring expected)\n\n**Sponsor variation:**\n\n- Day-of-first-dose AE included as TEAE (FDA preference) vs excluded (some EMA reviewers)\n- Partial-date imputation: impute day 15 if only month\u002Fyear known, vs censor as missing\n- Worsening of pre-existing AE: flagged via SEV change vs requires new PT (preferred term)\n\n**MedDRA SOC\u002FPT hierarchy:** AEs coded to MedDRA Preferred Terms (PT), grouped by System Organ Class (SOC). Clinically related PTs (e.g., 'Diarrhea' \u002F 'Frequent bowel movements' \u002F 'Loose stools') often combined via Standardized MedDRA Queries (SMQs) or sponsor-defined groupings. ADAE typically carries both AEDECOD (PT) and SMQ\u002Fgroup flags.\n\n## Baseline Derivation\n\n**ABLFL = 'Y'** marks the record whose AVAL becomes BASE for all other records of the same subject\u002Fparameter.\n\n**Standard rule:** last non-missing assessment on or before first dose (TRTSDT). If protocol mandates a specific baseline visit ('Day 1 pre-dose'), that visit's record is flagged.\n\n```python\n# Derive ABLFL when SDTM baseline flag is missing\u002Finconsistent\nimport pandas as pd\n\nvs['VSDTC_dt'] = pd.to_datetime(vs['VSDTC'], errors='coerce')\nvs = vs.merge(adsl[['USUBJID', 'TRTSDT']], on='USUBJID')\nvs['is_pre_treatment'] = vs['VSDTC_dt'] \u003C= pd.to_datetime(vs['TRTSDT'])\n\n# Latest pre-treatment value per subject\u002Fparameter\nbaseline_records = (vs[vs['is_pre_treatment'] & vs['VSSTRESN'].notna()]\n                    .sort_values('VSDTC_dt')\n                    .groupby(['USUBJID', 'VSTESTCD'])\n                    .tail(1))\nbaseline_records['derived_ABLFL'] = 'Y'\n```\n\n**Critical detail:** filter on VSBLFL='Y' (or LBBLFL='Y') as the primary source. Only fall back to derivation when the flag is missing. Trust the SDTM flag when present; CRF-level baseline designation embeds clinical judgement the analyst cannot reconstruct.\n\n**BASETYPE** required when more than one baseline exists per subject\u002Fparameter (crossover studies, multi-period trials). Distinguishes \"Period 1 Baseline\" vs \"Period 2 Baseline.\"\n\n**DTYPE** values per CDISC controlled terminology: 'LOCF' (last observation carried forward), 'WOCF' (worst), 'AVERAGE', 'BOCF' (baseline observation carried forward), null for original. **Pinnacle 21 flags any DTYPE value not in CT.**\n\n## SUPPQUAL and the NSV Transition\n\n**SUPPQUAL (supplemental qualifiers)** is the legacy mechanism for sponsor-defined variables that don't fit standard SDTM domain columns. Long-format QNAM\u002FQVAL pairs:\n\n```python\nsupp = pd.read_sas('suppae.xpt', format='xport', encoding='utf-8')\nsupp_pivot = supp.pivot_table(\n    index='USUBJID', columns='QNAM', values='QVAL', aggfunc='first'\n).reset_index()\nae_enriched = ae.merge(supp_pivot, on='USUBJID', how='left')\n```\n\nFor record-level SUPPQUAL (where IDVAR and IDVARVAL identify specific rows):\n\n```python\nsupp_record = supp[supp['IDVAR'] == 'AESEQ'].copy()\nsupp_record['AESEQ'] = supp_record['IDVARVAL'].astype(float)\nsupp_pivot_record = supp_record.pivot_table(\n    index=['USUBJID', 'AESEQ'], columns='QNAM', values='QVAL', aggfunc='first'\n).reset_index()\nae_enriched = ae.merge(supp_pivot_record, on=['USUBJID', 'AESEQ'], how='left')\n```\n\n**The 2024-2026 SUPP transition:** Therapeutic Area User Guides (TAUGs) increasingly use NS-- domain extensions or **Non-Standard Variables (NSV) Registry**-listed variables directly in the parent domain, instead of QNAM\u002FQVAL pairs in SUPP--. Not a hard deprecation but the direction is clear. The Non-Standard Variables Registry at cdisc.org is the new canonical place to look up sponsor-extension variables.\n\n## Date Handling -- The Partial-Date Reality\n\n```python\ndm['RFSTDT'] = pd.to_datetime(dm['RFSTDTC'], errors='coerce')\nae['AESTDT'] = pd.to_datetime(ae['AESTDTC'], errors='coerce')\nae['AEENDT'] = pd.to_datetime(ae['AEENDTC'], errors='coerce')\n\n# Days from randomization to AE onset\nae_with_ref = ae.merge(dm[['USUBJID', 'RFSTDT']], on='USUBJID')\nae_with_ref['AE_ONSET_DAY'] = (ae_with_ref['AESTDT'] - ae_with_ref['RFSTDT']).dt.days\n```\n\n**SDTM dates are ISO 8601 strings.** Partial dates (e.g., '2023-03' without day) are common. `errors='coerce'` converts these to NaT rather than raising errors. For analysis requiring complete dates, CDISC conventions impute missing day as the 1st for start dates and the last day of the month for end dates, but imputation rules should match the SAP.\n\n**SDTM records include EPOCH** (SCREENING, TREATMENT, FOLLOW-UP). For TEAEs, filter AEs to onset during or after the treatment epoch. Including pre-treatment AEs confounds the treatment effect estimate.\n\n## Validation -- Pinnacle 21 and CORE\n\n**Pinnacle 21 (Certara, formerly OpenCDISC)** is the de facto FDA submission validation standard. Validates against FDA Validation Rules + CDISC IG conformance + Define-XML schema. Severity tiers:\n\n- **Reject** -- submission will not be accepted\n- **Error** -- must justify\n- **Warning** -- should investigate\n\n**FDA Validation Rules** are published quarterly by FDA Office of Translational Sciences; Pinnacle 21 wraps these into its rule engine.\n\n**CORE (CDISC Open Rules Engine, 2021)** is a newer open-source alternative using YAML-defined rules from the CDISC Rules Catalog. Gaining traction but not yet at Pinnacle-21 parity for confirmatory submissions.\n\n```bash\n# Pinnacle 21 Community (free; appropriate for non-pivotal trials)\np21-community validate --rules sdtmig-3.4 --output-dir validation_output study_data\u002F\n```\n\n## Population Flags\n\n| Flag | Source | Purpose |\n|------|--------|---------|\n| ITTFL | DM all randomised | Primary efficacy population (ICH E9 default) |\n| FASFL | ITT minus eligibility failures + no post-baseline | Practical primary (FAS = Full Analysis Set) |\n| SAFFL | EX (received at least one dose) | Safety analysis (AE reporting) |\n| PPROTFL | SE + DS + protocol-violation list | Per-protocol; sensitivity only |\n| EFFFL | Sponsor-defined | Modified ITT variants |\n\n**FAS vs ITT subtlety:** FAS may exclude post-randomisation subjects (ineligibility, no post-baseline efficacy); ITT cannot. Many SAPs equate them; FDA may insist on stricter ITT at submission. Pre-specify both with explicit FAS exclusion criteria in the protocol.\n\n## Missing Data Considerations -- The Clinical Reasoning Layer\n\nBefore any imputation\u002Fcomplete-case decision, **examine the DS (Disposition) domain to tabulate reasons for discontinuation by treatment arm.** If discontinuation rates or reasons differ between arms, missing data is likely informative (MNAR) and standard MMRM-MAR is questionable.\n\n```python\nds = pd.read_csv('DS.csv')\ndropouts = ds[ds['DSDECOD'] != 'COMPLETED']\ndropouts_by_arm = dropouts.merge(adsl[['USUBJID', 'ARM']], on='USUBJID')\ndiscontinuation_reasons = dropouts_by_arm.groupby(['ARM', 'DSDECOD']).size().unstack(fill_value=0)\n```\n\nThis is the data-quality precursor to choosing the estimand strategy in trial-reporting (see ICH E9(R1)) — missing patterns informed by DS drive the choice between treatment-policy, hypothetical, or composite ICE strategies.\n\n## Common Pitfalls\n\n| Pitfall | Symptom | Solution |\n|---------|---------|----------|\n| Event-level merged onto subject-level without aggregation | Row count inflates after merge | Aggregate first, then merge |\n| First chronological record used as baseline | Misclassified baseline | Filter on VSBLFL='Y' \u002F LBBLFL='Y'; derive only if missing |\n| Character (xxORRES) used for analysis | Inconsistent numeric coercion | Use xxSTRESN (numeric standardised); missing xxSTRESN with present xxORRES means 'NOT DONE' or '\u003CLLOQ' |\n| ARM used in safety analysis | Crossover or actual-treatment differs | Use ACTARM for safety; ARM for ITT efficacy |\n| RFSTDTC used as TEAE reference | Includes screening AEs | Use RFXSTDTC (first treatment); cite ICH E2A |\n| ADTTE CNSR confused with stat-pkg convention | Wrong event\u002Fcensoring assignment | CDISC: CNSR=0 means event; convert: `event = (CNSR == 0).astype(int)` |\n| Partial date parsing error | NaT in date column | `pd.to_datetime(..., errors='coerce')` |\n| SUPPQUAL granularity confusion | Wrong rows merged | Check IDVAR before choosing subject vs record-level merge |\n| Non-standard column names treated as standard SDTM | Missing variables | Inspect actual columns; map to semantic roles |\n| Pinnacle 21 not run before submission | FDA reject | Always validate against current SDTMIG and FDA Validation Rules before submission |\n\n### Common non-standard column mappings\n\n| Standard SDTM | Common alternatives | Role |\n|---------------|--------------------|------|\n| ARM \u002F ARMCD | TRTGRP, TRT01P, treatment, group | Treatment assignment |\n| AEDECOD | AEPT, ae_term, preferred_term | AE preferred term |\n| AESEV (text) | AESEV (numeric 1-4), severity, AETOXGR | Severity \u002F toxicity grade |\n| USUBJID | SUBJID, subject_id, patient_id | Subject identifier |\n| RFXSTDTC | trt_start, first_dose_date | First treatment date |\n| LBSTRESN | lab_value_num, result_numeric | Lab numeric result |\n\n## Quantitative Thresholds and Conventions\n\n| Threshold\u002FConvention | Source | Rationale |\n|----------------------|--------|-----------|\n| TEAE window: 28-30 days post-treatment for small molecules | ICH E2A; sponsor convention | Mode of action, half-life inform window |\n| RFXSTDTC for TEAE reference (not RFSTDTC) | ICH E2A | RFSTDTC includes screening; TEAE is post-treatment |\n| ABLFL='Y' for last non-missing pre-dose | CDISC ADaM IG v1.3 | Standard baseline definition; trust SDTM flag |\n| ARM (planned) for ITT efficacy; ACTARM for safety | ICH E9 | Crossover\u002FPP-violation handling |\n| Pinnacle 21 validation before submission | FDA Study Data Technical Conformance Guide | Standard quality gate; reject errors block acceptance |\n| Define-XML 2.1 for studies starting >=March 15, 2023 | FDA Study Data Standards Catalog | Older 2.0 still accepted for prior studies |\n| Dataset-JSON v1.1 (Dec 2024; FDA notice April 2025) | CDISC + FDA Federal Register | Modern replacement for XPT v5; timeline pending |\n| CNSR=0 for events, positive integers for censoring | ADaM BDS-for-TTE v1.0 | OPPOSITE of R `survival` and most stat packages |\n\n## Anticipated Reviewer Pushback\n\n| Pushback | Response |\n|----------|----------|\n| \"RFXSTDTC or RFSTDTC for TEAE?\" | RFXSTDTC per ICH E2A; RFSTDTC would include screening AEs |\n| \"Baseline from VSBLFL or derived?\" | VSBLFL when present; documented derivation rule when missing |\n| \"Pinnacle 21 errors?\" | All errors resolved or justified; warnings reviewed and documented |\n| \"Define-XML 2.1 ADRG provided?\" | Yes — analysis dataset traceability documented to variable level |\n| \"ITT vs FAS reconciliation?\" | Pre-specified in protocol with explicit FAS exclusion criteria |\n| \"OCCDS v1.1 multi-period flags?\" | TRTEM01FL...TRTEM##FL pre-specified for crossover periods |\n| \"ADTTE CNSR convention conversion documented?\" | Explicit: CDISC CNSR=0 means event; convert to event=1 for downstream R\u002FPython |\n\n## References\n\n- CDISC. 2021. Analysis Data Model Implementation Guide (ADaMIG) v1.3.\n- CDISC. 2021. Occurrence Data Structure (OCCDS) v1.1.\n- CDISC. 2012. ADaM Basic Data Structure for Time-to-Event Analyses v1.0.\n- CDISC. 2024. Dataset-JSON v1.1.\n- FDA. 2024. Study Data Technical Conformance Guide.\n- FDA Federal Register Notice. April 2025. Dataset-JSON Pilot Comment Request.\n- ICH. 1995. E2A: Clinical Safety Data Management -- Definitions and Standards for Expedited Reporting.\n- ICH. 1998. E9: Statistical Principles for Clinical Trials.\n- ICH. 2019. E9(R1) Addendum on Estimands and Sensitivity Analysis.\n- Pinnacle 21 (Certara). 2024. Community Edition Validation Rules.\n- PHUSE\u002FCDISC. 2024. Dataset-JSON Pilot Reports.\n\n## Related Skills\n\n- clinical-biostatistics\u002Flogistic-regression - Model binary outcomes from prepared ADaM\u002FSDTM data\n- clinical-biostatistics\u002Ftrial-reporting - Use prepared analysis datasets for ICH E9(R1) estimands and CONSORT 2025 reporting\n- clinical-biostatistics\u002Fmissing-data-sensitivity - DS-domain reasoning informs estimand choice\n- clinical-biostatistics\u002Fsurvival-analysis - ADTTE CNSR convention; time-to-event preparation\n- expression-matrix\u002Fmetadata-joins - General metadata joining patterns\n",{"name":5,"source":24,"license":15,"tool_type":25,"description":7,"imported_at":26,"source_path":27,"primary_tool":28,"source_repository":29,"goal_approach_exempt":30},"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills","python","2026-09-20","clinical-biostatistics\u002Fcdisc-data-handling\u002FSKILL.md","pyreadstat","GPTomics\u002FbioSkills",true,"main","2d1750717747097b602acf30bd09a7cbd4b4c6ea","awesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-cdisc-data","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous","2026-11-09 11:45:36",{"source":8,"skillName":5,"commands":38,"prompt":48,"zipAvailable":30},[39,42,45],{"tool":40,"command":41},"npx","npx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-cdisc-data",{"tool":43,"command":44},"bunx","bunx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-cdisc-data",{"tool":46,"command":47},"pnpm","pnpm dlx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-clinical-biostatistics-cdisc-data","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-cdisc-data\u002FSKILL.md",[50,56],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"examples\u002Fcdisc_data_preparation.py","SCRIPT","脚本",2061,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-cdisc-data\u002Fexamples\u002Fcdisc_data_preparation.py",{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"usage-guide.md","REFERENCE","参考文档",7035,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-clinical-biostatistics-cdisc-data\u002Fusage-guide.md",[63,71,78,85,92,99],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":70},"01M7FFFZPC2025JQQHPHDX154J","dei-assessment","Assess diversity, equity, and inclusion with metrics. TRIGGERS - Use when user needs help with dei-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-assessment\u002FSKILL.md",[],[],"2026-11-16 10:05:34",{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":70},"01M7FFFWN8H4WA7A6Y398QKB6J","dei-action-plan","Create DEI action plans with goals and initiatives. TRIGGERS - Use when user needs help with dei-action-plan related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-action-plan\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":70},"01M7FFFSHTM0QZRVZ2NRJTJ0PY","culture-fit-assessment","Design culture add assessment processes. TRIGGERS - Use when user needs help with culture-fit-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fculture-fit-assessment\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":70},"01M7FFFMA12RKEQA38KHJSXXPS","cowork-to-code-bridge","Use an already-installed, independently verified cowork-to-code bridge to run narrowly approved actions on the user's own macOS, Linux, or WSL2 machine through a local file queue.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcowork-to-code-bridge\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":70},"01M7FFFH1F6M8YRGW5V487GH1K","counter-offer-response","Draft counter-offer responses for retention. TRIGGERS - Use when user needs help with counter-offer-response related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcounter-offer-response\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":70},"01M7FFFDGEHV8Y7J9N40HZXW52","continuous-feedback-system","Design continuous feedback systems. TRIGGERS - Use when user needs help with continuous-feedback-system related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcontinuous-feedback-system\u002FSKILL.md",[],[]]