Query+Health+-+Query+Format

include component="page" wikiName="siframework" page="Query Health Header"



This wiki page facilitates discussions of the Query Format specification.

Sample Queries:
The Sample Queries which have been created to help in documenting the implementation guidance and the translator work is located at [|Query Health Source Repository.]

Latest HQMF Models based on Query Health Discussions:
The Latest HQMF models and schemas with changes discussed below are posted in the [|Query Health Source Repository.]

Proposed Changes to HQMF:
This file captures the gaps identified in the HQMF standard for inclusion in a Scope Statement to start a project revising HQMF within the HL7 Structured Documents WG.

The above changes are further discussed in this presentation by Keith Boone to the Technical WG.

Current Query Format Work
at [|Query Health Source Repository.]. This is for review and comments so that it can be enhanced to be a more robust implementation guide that can be provided back to HL7.
 * The Draft HQMF Implementation Guide is based on the modified HQMF schema and models posted

Current implementation guidance includes examples drawn from HQMF work on NQF eMeasure 59. || **Click below to access the latest draft**
 * of the HQMF implementation guide:**

|| Latest Implementation Guide with Comments from Community || that the Query Health Technical Approach is designed to support. In general, the approach is designed to support the pilots who wish to implement the Query Health technical approach by supporting their implementation of a clinical data source(s) to support distributed queries.
 * The Query Health Sample Query Catalog is focused on capturing the list of questions and queries

The Sample Query Catalog leverages objects and elements defined in the Query Health CEDD documentation, including both the main CEDD document and the supporting data models that have been developed. The queries initially outlined in this catalog include the following quality measures:
 * 113 CMS eMeasures (as developed by CMS/NQF)
 * 65 ACO eMeasures (defined in the ACO Final Rule)
 * Additional New York PCIP Pilot measures (as outlined in Hub documentation) || **Click below to access the latest draft**
 * of the Sample Query Catalog:**

||  ||
 * HQMF to Procedural Translator Models/Schemas/Samples || [[file:siframework/HQMF To PRocedural Translator UML Model.docx|HQMF To PRocedural Translator UML Model.docx]]



||  ||

HQMF Issues To Be Resolved:
(Post on Google groups).
 * Issue ID || Issue Name || Issue Identifier (Point of Contact for Clarification of Issue) || Detailed Issue Description || Suggested Resolution || Final Resolution || IG Updated ||
 * 1 || Measure Period || Marc Hadley || The IG specifies the use of a 'measurePeriod' controlVariable to specify the measurement period. The NQF59 sample uses a different approach with two data criteria (StartDate and EndDate) with references to those in other data criteria expressions. || Pick an approach and make IG and samples align with chosen approach. || Final Resolution: Use ControlVariable "Measure Period". || v0.5 IG Section 3.2.1 and its sub-sections document the approach. ||
 * 2 || Patient Age || Marc Hadley || A patient's age is date dependent. Measure period has a start and end date and an effectiveTime can also specify both. What is the desired semantic when two dates are available: AND or OR or something else ? || Suggest: Analyze Use cases and then suggest a resolution.

[|Patient Age Discussion] || Final Resolution documented in the discussion post. ||  || (Post on Google groups).
 * 3 || Temporal relationships || Marc Hadley || The IG discusses use of a temporallyRelatedInformation element to constrain data criteria but the examples tend to use an embedded effectiveTime element instead. || For the most cases we can remove effective time from criteria elements.

[|Effective Time Element in Data Criteria] ||  ||   ||
 * 4 || Excerpting || Marc Hadley || The IG specifies a element to wrap the  element, the NQF59 example just has a subsetCode element. || Align the IG and examples || Completed. || Section 3.2.3.6 of the v.0.5 IG shows this. ||
 * 5 || Excerpt ordering || Marc Hadley || What is the order of operations when a data criteria includes an effectiveTime, temporallyRelatedInformation and subsetCodes. Different ordering leads to different results, e.g. most recent encounter between two dates vs most recent of all encounters intersected with a time period. We can't specify a simple static ordering since that would not allow us to express all possible queries. || Describe the difference between the last A1C measure > 9%, and the other being return the last a1c measure only if > 9%.

Need to create Examples for these and post it to the group.

Outcome: Document how to represent and how to process. ||  ||   ||
 * 6 || Single Query vs Multiple Queries || Srinivas Velamuri || Should we limit the HQMF for Query Health purposes to

1. One IPP, One Numerator, One Denominator along with their exclusions. 2. One IPP and multiple numerators, multiple denominators and their exclusions. 3. Multiple IPP's and multiple numerators and multiple denominators and their exclusions.

What does the Schema support currently ? || Make the PopulationCriteria Section Id mandatory.

Each Population Criteria Section ,has at most one numerator / denominator / IPP / Exceptions.

There can be multiple Population Criteria sections to address different groupings.

Ensure Only one Data Criteria section in IG. Make it a constraint. ||  ||   ||
 * 7 || Patient Level Query || Nagesh Bashyam || How would we represent Patient Level Query in HQMF such that the expected result would be patient level data. || Add the new metadata element using "Control Variable" data type just like Measure Period and specify QRDA Cat II or III. ||  ||   ||
 * 8 || Narrative-only queries || Marc Hadley || Section 2.4 of the latest QH IG draft allows HQMF queries with no structured body (i.e. just narrative text). I can understand this from a pure HQMF perspective (wanting to be all things to everyone) but I don't think this makes sense for the QH IG. || Include a requirement that the HQMF MUST include a structured query body. ||  || No optionality in the IG anymore. Schema will allow for it. QH Conformance statements require a Population Criteria section. Do we need to do anything more to enforce structured content ? ||
 * 9 || Embedded value sets || Marc Hadley || The 0.2 draft of the IG includes two different formats for embedding value sets in HQMF. We only need one. || Pick a single format and remove the other. || Completed. || Value Sets are now referenced using OIDs and are explained in section 3.2.3.4 of v0.5 IG ||
 * 10 || Inversion Indicators || Marc Hadley || temporallyRelatedInformation allows an inversionInd attribute that reverses the relationship. Alternately it is possible to put the relationship on the other element instead. Having multiple ways of expressing the same thing obfuscates the query and complicates implementations. || Remove inversionInd support, require queries to use the alternate approach. ||  ||   ||
 * 11 || Stratifiers and Populations || Marc Hadley || If we are going to support multiple initial patient populations then that can be used to implement both different populations and stratification (stratification is a subset/specialization of multiple populations). We don't need two separate mechanisms to accomplish the same thing. || Remove stratifiers and just support multiple populations. ||  ||   ||
 * 12 || Expression Format || Marc Hadley || HQMF does not define a syntax for expressions. For interoperable queries we require a standard expression format.

When should expressions be used .. ? || Adopt JavaScript, it is widely available and easily integrated into a variety of database environments: JavaScript Execution Environments ||  ||   || Another common pattern is encounter SAS diagnosis DURING the measurement period where both the SAS and DURING apply to the encounters. How do we represent this in HQMF v2 ? || For the former we require some kind of "composite" data criteria that represents the UNION of a set of referenced criteria.
 * 13 || Temporal relationships to multiple targets || Marc Hadley || Its common in the MU stage 1 and stage 2 measures to have temporal relationships to multiple targets. E.g. encounter starts before start (medication order OR medication administered OR medication dispensed).

For the latter the schema allows multiple temporallyRelatedInformation elements. ||  ||   ||
 * 14 || Counting of encounters/procedures || Floyd Eisenberg || HQMF needs to allow counting of encounters / procedures etc and should not limit itself to Patients only. ||  ||   ||   ||
 * 15 || Add support for Rations/Continuous Variables || Floyd Eisenberg/Telligen || HQMF needs to support for ratios and continuous variables to represent CQMs more completely ||  ||   ||   ||
 * 16 || Consider using functions for ordering and timing constraints || Telligen / Floyd Eisenberg || HQMF standard may have to allow the use of functions for specifying timing and ordering constraints ||  ||   ||   ||
 * 17 || Binding stratifiers to population criteria elements || Telligen || HQMF standard needs to allow binding of stratifiers to population criteria elements ||  ||   ||   ||
 * 18 || Functions of multiple targets || Marc Hadley || Its common in the MU stage 1 and stage 2 measures to have functions of multiple targets. E.g. COUNT>1 (medication order OR medication administered OR medication dispensed). How do we represent this in HQMF v2 ? || See issue 13. Having a "composite" data criteria that represents the UNION of a set of referenced criteria would provide the necessary abstraction to allow specification of a subset on the composite. ||  ||   ||
 * 19 || Enabling parsing of HQMF into a simple, vendor neutral, flat file structure || Steve Beller || The XML representations of HQMF greatly and unnecessarily complicate the computational requirements of CQMs. || The HQMF-XML structure should enable a translator to parse the XML into a simple, vendor neutral, non-proprietary flat file format that clearly delimitates the query parameters required for calculating the requisite numeric results of each CQM. This would enable computational tools, such as spreadsheet models, to compute the necessary values (as we have demonstrated in our proof of concept). || The HQMF improvements and the Intermediate Transalator model reviewed by the TWG facilitates the translation to various procedural implementations including SQL, java script, XQuery. Converting to spreadsheet is feasible from the existing model. ||  ||
 * 20 || Represent Fixed Measure Periods in terms of months for queries. || Floyd Eisenberg || HQMF V2 should allow the representation of fixed measure periods such as "April 01 - June 30" for the year when the query is instantiated. This functionality is used by CQM's and should be added to the IG to show how this is done ||  ||   ||   ||
 * 21 || HQMF is too open and should be restricted further. || Jeff Klann || HQMF is too open and should be constrained to ensure it is completely computable, maybe the intermediate translator model should guide this and should be included in the IG. ||  ||   ||   ||
 * 22 || Should width field be allowed/used in measurePeriod when high and low are specified || Marc Hadley || Should width field be used for any calculations when high and low values are specified as part of the measure period. ||  ||   ||   ||
 * 23 || What are the subset codes related to in terms of timeframes, it is not well defined || Floyd Eisenberg || The relationship between subset codes and timereference should be explicit and explained and documented. ||  ||   ||   ||
 * 24 || Embedded value sets || Marc Hadley || The resolution to issue 9 has completely removed the embedded value set capability. The original issue was that there were two different embedded value set format and we should pick one, not remove both ! || Re-instate the ability to embed an IHE SVS value set. ||  ||   ||

include component="page" wikiName="siframework" page="space.template.inc_contentleft_end"