This wiki has undergone a migration to Confluence found Here
<meta name="googlebot" content="noindex">

Allergy & Intolerance Drug Sub-project

From HL7Wiki
Jump to navigation Jump to search

Back to Allergy & Intolerance

Purpose

Produce a list of unique substances and multiple substance medications used in allergy & intolerance lists, ordered by frequency of incidence. This list will support the use of common elements for data capture and validation of data exchange. This list is not intended to prevent the recording of unusual substances where necessary, whether by code or text.

We would like to provide a set of standard coded identifiers for these substances. The codes for such a list would need to be free, readily available, and accepted for use by international stakeholders.

Note that the criterion is clinical use, not chemical specificity. I.e., if all we know is that the patient reports an allergy to "fish," that's what we can record. Similarly, if a patient has a reaction to percocet, and the clinician doesn't have evidence that supports identification of a specific ingredient, then "percocet" is what we know.

Goal

  • A list of substances ordered by frequency of incidence, with frequency ratios
  • Identifiers for these substances, using freely available standard identifiers

Plan

  • Confirm goals
  • Assess assets
  • Evaluate gaps
  • Fill gaps

Questions

  1. Include substances only, or also null and negative values?
    1. Use is the criterion: include what is used. Agreed 10/19
    2. Specific negatives are rare; we anticipate two (nka & nkda).
  2. How do we confirm quality?
    1. Process
      1. Acquire maps.
      2. If count(maps) > 1 and they agree, status is ok.
      3. If count(maps) < 2, acquire more maps.
      4. If count(maps) > 1 and they disagree, review.
  3. Encode and then combine, or combine and then encode?
    1. Encoding is required to combine
  4. How do we weight lists?
    1. Use filtered rankings to assess divergence, but no weighting in frequency list.
  5. Rank all substances from contributed lists, or only those to a chosen level (97%, 99%, etc.)?
    1. Identified substances with counts > 500 (individually ~0.0017%; aggregate 0.71%)
    2. Include frequency ratios in resulting list; users may choose their own thresholds.

Open questions

  1. Salts: We assume that salt forms are irrelevant, and that incidences recorded as salt forms should be recorded as general forms (e.g., codeine vs codeine sulfate). Is this always true?
    1. Is it true for any other chemical modifications, e.g., esters? (Which would make aspirin an instance of a salicylate).
  2. Classes
    1. For classes that are commonly asserted but do not actually designate cross-reactive substances, how do we proceed?
  3. What system(s) should be used for encoding?
    1. Assumption: do we need to choose, or can we provide a list of substances with all pertinent code assignments?
    2. Criteria
      1. Maximal coverage of identified requirements
      2. Ability to add missing items
      3. Freely available
      4. International
    3. Candidates
      1. SNOMED CT: substances, classes; mixtures only as products. Licensing issue.
      2. RxNorm: substances & mixtures. No license issue, but US realm.
      3. NDF-RT: classes only. Class definitions problematic.
      4. UNII: substances only. US realm. no relationships (e.g., of salts)
      5. ATC: classes only. Class definitions problematic.
      6. INN: no access to list; tbd
      7. Proposal to use whatever G-SRS chooses to use. Will evaluate when available.

Minutes

28 December 2016

14 December 2016

16 November 2016

2 November 2016

19 October 2016