From a collection to four observations

Follow each fictional measurement into its own row and keep a trace back to the input.

Each input observation yields exactly one output row. Two subjects with two observations each produce four rows, not a wide record with two measurement columns. Sequence numbers distinguish this example's rows within a subject, while the manifest retains the actual input IDs. Values stay as strings so the demonstration does not silently impose numeric interpretation.

01 / The collection

[
  {
    "id": "obs-a",
    "study": "CLD-SYN-001",
    "subject": "SYN-001",
    "measurement": "DEMO_A",
    "value": "72",
    "unit": "DEMO_UNIT",
    "date": "2025-02-03"
  },
  {
    "id": "obs-b",
    "study": "CLD-SYN-001",
    "subject": "SYN-001",
    "measurement": "DEMO_B",
    "value": "18",
    "unit": "DEMO_UNIT",
    "date": "2025-02-03"
  },
  {
    "id": "obs-c",
    "study": "CLD-SYN-001",
    "subject": "SYN-002",
    "measurement": "DEMO_A",
    "value": "75",
    "unit": "DEMO_UNIT",
    "date": "2025-02-04"
  },
  {
    "id": "obs-d",
    "study": "CLD-SYN-001",
    "subject": "SYN-002",
    "measurement": "DEMO_B",
    "value": "20",
    "unit": "DEMO_UNIT",
    "date": "2025-02-04"
  }
]

02 / The representation

VS synthetic output
STUDYID	DOMAIN	USUBJID	VSSEQ	VSTESTCD	VSORRES	VSORRESU	VSDTC
CLD-SYN-001	VS	SYN-001	1	DEMO_A	72	DEMO_UNIT	2025-02-03
CLD-SYN-001	VS	SYN-001	2	DEMO_B	18	DEMO_UNIT	2025-02-03
CLD-SYN-001	VS	SYN-002	1	DEMO_A	75	DEMO_UNIT	2025-02-04
CLD-SYN-001	VS	SYN-002	2	DEMO_B	20	DEMO_UNIT	2025-02-04

Inspect output JSON and CSV

[
  {
    "STUDYID": "CLD-SYN-001",
    "DOMAIN": "VS",
    "USUBJID": "SYN-001",
    "VSSEQ": 1,
    "VSTESTCD": "DEMO_A",
    "VSORRES": "72",
    "VSORRESU": "DEMO_UNIT",
    "VSDTC": "2025-02-03"
  },
  {
    "STUDYID": "CLD-SYN-001",
    "DOMAIN": "VS",
    "USUBJID": "SYN-001",
    "VSSEQ": 2,
    "VSTESTCD": "DEMO_B",
    "VSORRES": "18",
    "VSORRESU": "DEMO_UNIT",
    "VSDTC": "2025-02-03"
  },
  {
    "STUDYID": "CLD-SYN-001",
    "DOMAIN": "VS",
    "USUBJID": "SYN-002",
    "VSSEQ": 1,
    "VSTESTCD": "DEMO_A",
    "VSORRES": "75",
    "VSORRESU": "DEMO_UNIT",
    "VSDTC": "2025-02-04"
  },
  {
    "STUDYID": "CLD-SYN-001",
    "DOMAIN": "VS",
    "USUBJID": "SYN-002",
    "VSSEQ": 2,
    "VSTESTCD": "DEMO_B",
    "VSORRES": "20",
    "VSORRESU": "DEMO_UNIT",
    "VSDTC": "2025-02-04"
  }
]

STUDYID,DOMAIN,USUBJID,VSSEQ,VSTESTCD,VSORRES,VSORRESU,VSDTC
CLD-SYN-001,VS,SYN-001,1,DEMO_A,72,DEMO_UNIT,2025-02-03
CLD-SYN-001,VS,SYN-001,2,DEMO_B,18,DEMO_UNIT,2025-02-03
CLD-SYN-001,VS,SYN-002,1,DEMO_A,75,DEMO_UNIT,2025-02-04
CLD-SYN-001,VS,SYN-002,2,DEMO_B,20,DEMO_UNIT,2025-02-04

Download input.json Download output.json Download output.csv Download manifest.json

03 / The decisions between them

study → STUDYID: Copy study identity.
subject → USUBJID: Copy the authored subject key.
measurement → VSTESTCD: Copy the explicit teaching token; this is not terminology normalization.
value → VSORRES: Preserve the original string.
unit → VSORRESU: Preserve the fictional unit token.
date → VSDTC: Preserve recorded precision; do not add a timezone.
(authored) → DOMAIN: Set constant VS for this teaching example.
(authored) → VSSEQ: Assign fixed example sequence numbers within each subject; not a general sequencing rule.

Trace each row

obs-a → vital-signs-rows:1
obs-b → vital-signs-rows:2
obs-c → vital-signs-rows:3
obs-d → vital-signs-rows:4

JSON null becomes an empty CSV cell. Empty CSV cells cannot distinguish absent from empty text; use JSON for that distinction.

Read the variable lessons

STUDYID DOMAIN USUBJID VSSEQ VSTESTCD VSORRES VSORRESU VSDTC

Follow the source

Original ClinDevLab explanations. Publisher material is linked, not reproduced. Reviewed 8 October 2026.

SDTM 2.0https://www.cdisc.org/standards/foundational/sdtm/sdtm-v2-0 · CDISC · 2.0 · accessed 2026-10-08
SDTMIG 3.4https://www.cdisc.org/standards/foundational/sdtmig/sdtmig-v3-4 · CDISC · 3.4 · accessed 2026-10-08