This wiki has undergone a migration to Confluence found Here

PQXF An extended form of the PQ data type for simple factors.

From HL7Wiki
Jump to navigation Jump to search

NOTE: Harmonization proposal on public display here for the purpose of commenting and collaborative editing. All your edits are tracked and nothing gets lost. FEEL FREE to improve the proposal and to add any question you want to raise in the discussion. Thanks!

Recommendation for HL7 RIM Change RECOMMENDATION ID:
Submitted by: Gunther Schadow Revision (# and date): 2
Date submitted: 20050212 Committee status: open
Submitted by: Gunther Schadow  
NAME: Individualized dosage with factor (e.g., for body mass)  

Stewards Position

REQUIRED - This table should contain one row for each Steward Committee affected by the recommendation.

(responsibility level: S=Steward; I=Interested)
O&O, Pharmacy SIG Approved S
RCRIM Unknown I
PC Unknown I

WHEREAS the PQ data type exists to represent constant physical quantities

WHEREAS the ActRelationship type derived-from exists and an attribute ActRelationship.localVariableName exists to define input data for expressing the input to computing attribute values

WHEREAS the Observation.derivationExpression was once considered for representing the formulas to compute other attribute values

WHEREAS such formula in one attribute would require an expression and assignment language

WHEREAS such language cannot presently be agreed upon

WHEREAS in the U.S. the Physician's Labeling Rule (PLR) has been signed and will take effect by the end of the year, and HL7 SPL is chosen by the FDA to be the standard format for electronic prescription drug information, which include critical safety data encoded in HL7 v3 form

WHEREAS the full dosage recommendations will probably not go into this guidance in all detail, because it is deemed too complicated

WHEREAS some statement of a maximum dose is really very important for prescribing safety and could be accomodated without being viewed as too difficult. But when maximum dosage really matters, it is often dependent on the patient's individual pharmacokinetic characteristics, which is approximated by body mass or BSA. For example, any maximum dose specification in pediatrics would be useless if not depending on body mass or BSA.

WHEREAS this requirement existed in HL7 v2 as well, and was addressed in a not further specified way simply by changing the unit of the dosage to mg/kg (or something of that kind.) However, the meaning of this is not specified and not standardized

WHEREAS if we do not do anything, we will (a) loose the oportunity to put critical safety information into the public domain (through the SPL process), and (b) the requirement will be addressed by implementers in a variety of unspecified and ways

WHEREAS eventually other individualized dosage formulas should be supported (e.g., for "sliding scale" dosing of insulin by blood glucose level, gentamicin by BSA and creatinine, etc.), this need not be immediately resolved. The immediate objective is only: specifying the maxDoseQuantity by body mass or body surface area

NOW THEREFORE let it be recommended to the InM committee that a PQXF data typeit be defined, which has a minimally necessary structure to support a body-weight or BSA factor in the specification of maxDoseQuantity of the detailed structure specified below.

PQXF would extend PQ by a factorLocalVariableName property. The meaning of such property is defined by multiplying the localVariable such referred when referring to the PQ value of PQXF.

Abstract specification:

 type PhysicalQuantityExpressionWithFactor alias PQXF specializes PQ {
   ST factorLocalVariableName;

The factorLocalVariableName contains the value of a localVariableName in an ActRelationship whose source is the Act containing the PQXF value and whose target is an Observation specification (in DEF or EVN.CRT mood) whose Observation.code or would identify the parameter definition in terms of the clinical coding conventions used by the communicating systems (e.g., LOINC, SNOMED). The factor is then the observation value of an observation so described.

Can something be said about the ActRelationships where this would be applicable in order to constrain it as fasr as possible? If it can't be constrained, can the reason for not doing so be included in the proposal? Rene spronk 01:37, 17 Feb 2006 (CST)
Well, the localVariableName attribute already exists in the RIM, it is not new. It would basically work with any relationship type. However, this would most often be implemented using a derivedFrom relationship. Gunther Schadow
If it can be reasonable limited to that relationship type, can this constrained be documented as part of the standard? Either as part of the PQXF data type specification or as part of the RIM? Such a constraint can be relaxed at a later point in time if there are use-cases that require the constraint to be relaxed. localVariableName is a powerful mechanism that may be (too) hard to implement if it isn't sufficiently constrained. Rene spronk 02:23, 23 Feb 2006 (CST)


   <maxDoseQuantity xsi:type="PQXF" value="57" unit="mg/kg">
     <factor localVariableName="bodyMass"/>  
     <activeIngredient ... />
       <code code="29463-7" codeSystem="&LOINC;" displayName="BODY WEIGHT:MASS:PT:^PATIENT:QN"/>

Note in the above example, everything you see, except the one element <factor .../> is what we can already do without any harmonization or data type action.

Add a data type, PQXF PhysicalQuantityWithFactor which can be used to specify dosage individualized by body mass or body surface area.

In addition, clarification is required on the ActRelationship.localVariableName definition. Existing definition: "A character string name for the input parameter from which the source Act of this ActRelationship derives some of its attributes. The local variable name is bound in the scope of the Act.derivationExpr with its value being an Act selected based on the input parameter specification."

Proposed definition: "A character string name for the input parameter from which the source Act of this ActRelationship derives some of its attributes. The target act must be an Observation, and the content bound to the variable is the Observation.value (if any). The local variable name is bound in the scope of the Act.derivationExpr or Act containing a PQR expression with its value being an Observation.value selected based on the input parameter specification."

Recommended Action Items

  • Implement the proposed solution