Data Quality & Integrity
Point-of-Entry Validation
Summary
Data is validated as close to its point of entry as possible, and invalid data is rejected or explicitly flagged.
Reasoning
Validation at the point of entry prevents incorrect, incomplete, or malformed data from propagating into downstream systems. Explicitly flagging invalid data preserves its status for downstream handling. Silent acceptance or correction conceals the failure and can change the data's meaning without the knowledge of its producer or consumers.
Implemented By These Standards
Structural & Referential Integrity
Summary
Structural and referential integrity is enforced at the data layer where the data store supports it, and by the responsible service where it does not.
Reasoning
Schema constraints define which stored states are valid. Enforcing them in the data store applies them consistently to every write. Where the store cannot enforce a constraint, equivalent protection in the service responsible for writing the data prevents structurally invalid or inconsistent data.
Schema and data model changes can invalidate records that previously satisfied their constraints. Identifying and resolving these conflicts before applying a change prevents the change itself from introducing a structural or referential integrity failure.
Implemented By These Standards
Duplicate & Conflict Resolution
Summary
Uniqueness constraints prevent duplicate representations, and defined resolution rules reconcile records that conflict.
Reasoning
Integrated systems can represent the same real-world entity in multiple records, allowing consumers to receive contradictory values. Uniqueness constraints and defined matching or conflict-resolution rules identify the authoritative representation and prevent unresolved conflicts from persisting.
Implemented By These Standards
Automated Rule Testing
Summary
Validation, constraint, and data quality rules are verified through automated tests.
Reasoning
These rules define which data states are accepted and rejected. Automated tests make their verification repeatable and expose regressions when a rule changes. Testing with representative data verifies that a rule continues to distinguish valid and invalid data across the conditions it is expected to handle.
Implemented By These Standards
Change Traceability
Summary
A change to decision-informing data remains traceable to its source, timing, and responsible actor or process.
Reasoning
Operational, financial, and clinical decisions depend on being able to explain the data that informed them. Recording a change's source, time, and responsible actor or process establishes who or what changed the data and when.
Tracing the transformations between a data item's origin and current state explains how its value was derived without relying on informal knowledge. Retaining this information for an appropriate period preserves the evidence needed for audit and incident investigation.