Skip to end of metadata
Go to start of metadata

Context

As a second phase of the project, we want to expand the scope of the IG to include the consideration of additional relevant clinical information (beyond just the immunization history) for the patient by the CDS engine when making a personalized forcast. This would enhance the ability to make risk-based recommendations in addition to the core age-based recommendations. Types of clinical data could include:

  • Immunity 
  • Allergy
  • Adverse Reaction  
  • Medication Administration 
    • Immune Globulin
    • Medications
      • anti-virals
      • antibiotics
  • Problem list and diagnoses
  • Pregnancy Status 
  • Occupational Data
  • Social History
    • Smoking, drug use, etc 
  • Family History 
    • history of altered immunocompetence, etc
  • Procedure 
    • splenectomy, dialysis, radiation therapy etc
  • Living Arrangements 
    • dormitory, correctional facility, homelessness, etc
  • Travel plans

Options

A number of approaches could be used to make relevant clinical data available to the CDS engine. 

  • Clinical data could be passed in as part of the defined operation
    • A diverse set of FHIR resources (eg Observation, Condition, Procedure) could be identified by the IG and passed from the calling system to the CDS engine
    • The calling system could collect data on the patient and format it in the form of a QuestionnaireResponse before passing to the CDS engine
  • The CDS engine could perform RESTful queries of the calling system (or any other available clinical system (eg. the local HIE)) to gather additional relevant clinical data

Each of these approaches has pros and cons:


ProsConsComments
Passing in diverse resources
  • The calling system would not have to reformat clinical data into a new format, and can likely reuse existing FHIR support
  • The IG will still need to include definitions of the types of resources that are relevant (eg. not all Procedure resources are relevant, only those that may impact a forecast such as a splenectomy or a bone marrow transplant)

Passing in QuestionnaireResponse
  • The clinical data would be easier for the CDS engine to consume because the potential content provided by the calling system would be more constrained
  • This format lends itself better to gathering data directly from the patient which may not be regularly collected by an EHR or IIS (eg. international travel plans, occupation)
  • The calling system will need to extract the relevant data from their system and format it into a QuestionnaireResponse (initial implementation burden)
  • Only data included in the Questionnaire would be accessible to the CDS engine

Engine initiating queries
  • The CDS engine could limit clinical data to only those concepts supported by the engine 
  • Much more robust patient demographics would have to be passed to the CDS engine for subsequent querying
  • If querying a third system for clinical data, the opportunity for patient mismatching is possible

Including Clinical Content in Relevant Resources

Possible Resources

  • Immunity - Observations (lab results) or Condition (history of disease)
  • Allergy - AllergyIntolerance
  • Adverse Reaction  - Observation linked to an Immunization
  • Medication Administration - MedicationAdministration or MedicationStatement or possibly MedicationRequest
  • Problem list and diagnoses - Condition
  • Pregnancy Status - Observation (a profile may already exist)
  • Occupational Data - Observation (a useable profile (ODH) may already exist) 
  • Social History - ??
  • Family History - ??
  • Procedure - Procedure or possibly ServiceRequest
  • Living Arrangements - Observation?
  • Travel plans - Observation?

Possible Existing Profiles

  • US Core
  • QI Core
  • ECR Pregnancy Status
    • Profile on Observation which includes an extensible answer set of:

      102874004Possible pregnancy (finding)
      60001007Not pregnant (finding)
      77386006Pregnant (finding)
      UNKUnknown
    • .subject points back to the ECR Patient profile
      • Not clear if this would be a major problem
    • Requires support for two components, neither of which is required (but may be helpful for recommending Tdap in the right point in the pregnancy)

      • Estimated Gestational Age of Pregnancy
      • Estimated Date of Delivery
    • Also requires support for the method of determination of status (not useful for our use case)
    • If we can't use the profile as it is, we should at least try to adopt the answer value set and the LOINC code (82810-3 - Pregnancy Status)

  • ECR Travel History
    • A Zulip thread recommends using Observation (with a future effective date) to capture future travel plans
    • We might be able to use the ECR profile as-is even though it's intended for past travel history
      • .subject is not constrained which seems odd
    • If we can't use the profile, we should try to use the same value sets and codes
      • Would we want to include both coded and free text options for the destination? Would free text be helpful to a CDS engine?
  • ODH
    • Profiles exist for:
      • Employment Status (Observation)
        • Includes answers of Employed, Not in Labor Force and Unemployed
        • Probably not relevant
      • Past or Present Job (Observation)
      • Retirement Date (Observation)
        • Probably not relevant
      • Usual Work (Observation)
        • Typical work, not the current work (which should drive the immunization requirements
        • Probably not relevant
      • Combat Zone Period (Observation)
        • Probably not relevant
      • Occupational Data for Health (Composition)
        • Probably not relevant

Include Clinical Content as QuestionnaireResponse

There are several examples where questionnaires have been created to capture core data about individual's HALO information to inform vaccination recommendations.

It may be easier to pass clinical data in as questionnaire responses (either filled out by the patient or including data gathered by the EHR out of the patient record).

The Healthcare Associated Infection (HAI) FHIR IG uses Questionnaire and QuestionnaireResponse to report  to the CDC. The IG includes profiles on both resources (for both single patient and populate based reports) and then provides instances of the Questionnaire and example responses.

We could do something similar for the patient's clinical data. We could define a set of questions and associated answer types (in some cases, even with values sets for things like a list of chronic diseases). It may be more difficult for the requester to fill out the questionnaire but it would make the interoperability easier but there's be more certainty about what data would be passed in the operation. We also wouldn't have to come up with a set of codes (SNOMED, etc) for specific risk conditions (eg. chronic liver disease, immunocompromised) that could potentially be incomplete. 

Because the Questionnaire instance can included text for the question, we'd have to think about how we phrase the text (eg Pregnant? vs Are you pregnant? vs Is the patient pregnant?)

The CDSi project has created a list of coded observations about a patient that can serve as either an indication or contraindiction for vaccination. These coded observations are more granular than the HALO categories but most of them (at least the indications) should be able to be classified as one of the HALO concepts. The google sheet of the coded observations mapped to the HALO concepts is available.

Based on HALO and harmonized schedules (child and adult), the list of questions could be something like:

  • Pregnant? (females only)
    • Pregnant
    • Not Pregnant
    • Possibly Pregnant
    • Unknown
  • Relevant Chronic Disease (select all that apply)
    • Chronic Liver Disease
    • Chronic Heart Disease
    • Chronic Kidney Disease
    • Diabetes
  • Immunosuppressed (including HIV)?
    • Yes
      • HIV Severely Immunocompromised (CD4+ count <15% and total CD4 cell count of <200/mm3)
      • HIV Not Severely Immunocompromised (CD4+ count ≥15% and total CD4 cell count of ≥200/mm3)
      • Severe Combined Immunodeficiency
      • etc
    • No
  • History of STD?
    • Yes
    • No
  • Medical Conditions (select all that apply)
    • Asplenia
    • Persistent complement component deficiencies
    • Cochlear Implant
    • Organ Transplant
    • CSF Leak
  • Born outside the US
    • Yes
    • No
  • Behavior (select all that apply)
    • Alcoholism
    • Injection drug user
    • Non-injection drug user
    • Men who have sex with men
    • Not in a long-term mutually monogamous relationship
    • Smoker
  • Occupation (select all that apply)
    • Healthcare worker
    • Lab worker
    • Veterinarian
    • Other
    • n/a
  • Residential situation (select all that apply)
    • Homeless
    • First year college student in dormitory
    • Resident of institution
    • Close contact with an international adoptee
  • Future travel destinations (select all that apply)
    • geographic location value set


Possible Contraindication related questions (not from HALO)

  • Allergies/Hypersensitivities (select all that apply)
    • Gelatin
    • Arginine
    • Latex
    • Chicken Protein
    • Gentamycin
    • Neomycin
    • Streptomycin
    • Polymyxin B
    • Yeast
    • 2-phenoxyphenol (preservative)
    • Alum
    • Protamine Sulfate
  • Received blood products in the previous x months?
    • Yes
    • No
  • Medical Treatments
    • Chemotherapy
    • Radiation therapy
    • Immunosuppressive therapy
    • Long-term aspirin therapy
    • Antimicrobial or antiviral medication in the last 72 hours

HALO organized data (Grey).csv

The set of questions from the CDC harmonized schedules overlap pretty closely (although a few may require slightly different wording):

QuestionAnswerCDC ChildCDC Adult
PregnantY/NXX
Immunocompromised (excluding HIV)Y/NXX
HIV Infection CD4 Count<200
>=200
XX
AspleniaY/NXX
Complement deficienciesY/NXX
End-stage renal disease; or on hemodialysisY/NXX
Heart DiseaseY/NXX
Lung DiseaseY/NXX
AlcoholismY/N
X
Chronic liver diseaseY/NXX
DiabetesY/NXX
Health care personnelY/N
X
Men who have sex with menY/N
X
CSF LeaksY/NX
Cochlear Implants Y/NX
  • No labels