AnzarSeha

References

By Dr Rida Akodad · Publication director · Updated 21 August 2026

This page gathers the standards cited across this site, with what each actually normalises and the address of the body that publishes it. It serves as the single source: other pages link here rather than copying links.

What "following a standard" means

A distinction first, because it is regularly blurred in this field: following a standard means adopting its definitions, names or structure. Being certified means an authorised body has verified that adoption and attested to it.

Good practice and general framework

StandardWhat it normalisesPublished by
ICH E6(R3)Good clinical practice: data integrity, traceability to source, responsibilities of actorsICH
EQUATOR NetworkThe directory of reporting guidelines by study typeEQUATOR Network
STROBEReporting of observational studies: what a paper must state about population, variables and missing dataSTROBE Initiative

Data structure and capture

StandardWhat it normalisesPublished by
CDISCThe family of clinical research data standardsCDISC
CDASHCapture: which variables to collect, under which name, with which definitionCDISC
SDTMThe tabulation structure for transmission and submissionCDISC
HL7 FHIRExchange of health data between systemsHL7 International

Terminologies and units

StandardWhat it normalisesPublished by
SNOMED CTA structured clinical terminology: concepts, relationships, hierarchiesSNOMED International
LOINCIdentification of laboratory observations and clinical measurementsRegenstrief Institute
ICD-10The international classification of diseasesWorld Health Organization
UCUMUnambiguous writing of units of measureRegenstrief Institute

The difference between these four is often misunderstood: ICD-10 classifies in order to count, SNOMED CT describes in order to reason, LOINC identifies what is measured, UCUM says in which unit. One datum can carry all four, each answering a different question.

Personal data protection

FrameworkScopeAuthority
GDPR — Regulation (EU) 2016/679Processing under European Union lawNational supervisory authorities
Law 09-08Personal data processing in Morocco, health data includedCNDP

These two frameworks define, among other things, pseudonymisation and its status: pseudonymised data remains personal data. That is why this site never uses the word "anonymised" in its place.

Reuse and reproducibility

StandardWhat it establishesSource
FAIR principlesThat research data be findable, accessible, interoperable and reusable — metadata being the conditionWilkinson et al., *Scientific Data*, 2016

The FAIR principles are what justify the delivered dataset comprising six pieces rather than a single table: without a dictionary and metadata, a deposited dataset is reusable by nobody, including the team that produced it.

This site's citation policy

  • No citation is invented. Every standard listed here is named with the address of the body that publishes it, verifiable in one click.
  • No performance figure is advanced without measurement. This site publishes no adoption rate, no unsourced quantitative comparison, and no clinical result.
  • No claim about a competitor without a source. The comparisons address only observable properties of a tool's model.
  • Screenshots come from the demonstration account, on fictional data. No figure shown is a result.

Frequent questions

Are these standards mandatory for academic research?

No, none of them. Following them is not a regulatory obligation but a methodological choice: it makes the dataset interoperable, comparable, and legible to someone who does not know the team that produced it.

Is a licence needed to use SNOMED CT?

Access to SNOMED CT depends on the country and the affiliate status of the organisation using it; the terms are published by SNOMED International and should be checked with them rather than assumed.

Where to start if none of these standards is familiar?

With UCUM and CDASH. UCUM because declaring units costs almost nothing and avoids the most expensive errors; CDASH because reusing existing variable names removes the need to invent any, and makes the dataset comparable from day one.

Sources and standards