Uploaded image for project: 'FHIR Specification Feedback'
  1. FHIR Specification Feedback
  2. FHIR-38899

Inconsistent representation of negation across resources



    • Icon: Change Request Change Request
    • Resolution: Unresolved
    • Icon: Highest Highest
    • FHIR Core (FHIR)
    • R5
    • Orders & Observations
    • STU
    • AllergyIntolerance


      Need a way other than terminology (which is extensible- i.e., not required) to indicate that a condition is absent, but has a clinical significance to the patient.  This is common to record if there are conditions that are absent in the patient which may directly inform the clinical workflow (i.e.,  the assessment and treatment planning for the patient).

      For example:

      • Absence of Hormone Receptors (e.g., Estrogen, Progesterone)
      • Absence of Genetic Mutations (e.g., BRCA1, BRCA2)
      • Absence of Allergies - esp. in context of known allergies (e.g., Allergy to Tree Nuts, but No Allergy to Peanuts)


      Additional Context and Examples:

      Pertinent Positive:

      • Diabetic retinopathy

      Pertinent Negatives important in association with diabetic retinopathy:

      • No Vitreous hemorrhage
      • No glaucoma
      • No vision loss
      • No neuropathy
      • No nephropathy
      • No cardiovascular conditions
      • Employment: No employment is an observation that could also be a problem.

      These are examples of pertinent negatives.  They are pertinent negatives to exclude in the differential diagnosis to rule out an acute situation requiring immediate intervention.  Having negation included in the terminology as one statement indicates it to be a condition or status, such as 'unemployed', which it should not. Employment is an observation that could also be a problem.

      Having negation represented with consistency is key for interoperability as complexity (reduced pre-combinatorial explosion) is minimized and simplicity optimized.  This mitigates the potential for errors and is computable for machine processing (& learning), allowing for modularity, reusability, inferencing, CDS and research.  Regarding research, iso-semantics, or equivalence, needs to be determined with near 100% certainty for CDS and other use cases.


      In FHIR, there is inconsistent representation of negation.  As discussed above, the increases complexity.  For example:

      • In the conditions resource, negation is represented by Condition.clinicalStatus:  active | recurrence | relapse | inactive | remission | resolved
      • In the observations resource, negation is represented by Observation.component.valueCodeableConcept:  'Negative'
      • In the familyMemberHistory resource, negation is represented by FamilyMemberHistory.condition.code:   "No Known Problems" or negated conditions, such as "No history of malignant tumor of breast", for an individual family member.
      • In the AllergyIntolerance resource, negation is represented by AllergyIntolerance.verificationStatus:  unconfirmed | confirmed | refuted | entered-in-error

      As can be seen, while similar to some degree, these FHIR Resources are inconsistent in their representation and subject to the risks noted above.




            Unassigned Unassigned
            marti_velezis Marti Velezis
            Marti Velezis, Raja Cholan
            4 Start watching this issue