Data Content, Structure, and Standards
This domain tests the data a health record contains, the standard to use for a given exchange need, and the ability to recognize a data quality failure. Items are applied: identify the applicable data set, choose the terminology that can hold a concept, or name the characteristic that has been violated.
Data sets and definitions
- UHDDS supplies standard inpatient elements - principal diagnosis, other diagnoses, principal procedure - that underpin ICD-10-CM/PCS reporting and Medicare inpatient payment.
- Setting-specific sets: OASIS (home health), MDS (nursing facility), IRF-PAI (inpatient rehabilitation), DEEDS (emergency department).
- Data are raw facts; information is data processed into meaning; knowledge applies information. Data elements are the smallest named units; a data dictionary defines each field's format, allowed values, and source.
Data quality characteristics
AHIMA's model includes accuracy, accessibility, comprehensiveness, consistency, currency, definition, granularity, precision, relevancy, and timeliness. Two systems showing different admission dates for the same patient is a consistency and accuracy failure, not a formatting issue.
Standards and terminologies
- Classifications (ICD-10-CM/PCS, CPT, HCPCS) group concepts for reporting and payment; clinical terminologies such as SNOMED CT capture granular clinical meaning at the point of care.
- LOINC identifies laboratory tests and observations; RxNorm normalizes medications; NDC identifies drug packages; UCUM specifies units.
- Exchange standards: HL7 v2 messaging, CDA documents, FHIR resources and APIs, DICOM imaging; X12 837 claim, 835 remittance, 270/271 eligibility, 276/277 claim status.
- Interoperability depends on shared semantics, not just connectivity; information blocking is a separate legal and operational issue.
Worked example
Two hospitals want to pool hemoglobin A1c results on a quality dashboard. The correct build uses LOINC to identify the test, the numeric result with its UCUM unit, and standardized patient and date fields. ICD-10-CM cannot carry the result: it is a classification, many patients with abnormal results have no diabetes code at all, and a code such as E11.9 would collapse controlled and uncontrolled patients into one group. Choosing the classification here would produce a biased denominator.
Traps that cost marks
Treating classifications and clinical terminologies as interchangeable; assuming SNOMED CT is a billing vocabulary; reporting CPT on inpatient facility claims (inpatient procedures are ICD-10-PCS); mixing UHDDS and OASIS elements; and answering "data quality" when the item describes a data definition or metadata problem.