Proposal To Address
Medications, Allergies And
Laboratory Disparities In Final
Rule MU Stage 2
Co-presented by
George Robinson, RPh FirstDataBank, Inc
James Shalaby, Pharm.D. PSMI Consulting, Inc
October 4th, 2012
1.Address key ambiguities found in review of the Final rules for
Stage 2 meaningful use and the referenced HL7 consolidated
CDA standard.
2.Recommend a practical, low effort, late breaking solution to
mitigate potential problems in CQM reporting.
3.Provide assistance if needed to help quickly resolve the
prioritized ambiguities.
Problems Found In Allergies #1
• Consolidated CDA standards for allergies to medications are at
the wrong term types in RxNorm:
• Fact: Most EHRs document allergies at the medication ingredient,
medication class or brand name level but not at the SCD level.
• Problem: The standard does not state a “shall” for RxNorm
ingredients level for medications.
1. There is ambiguity in cCDA and ONC documentation with respect to
ingredients in allergy documentation since it can be interpreted
that any ingredient allergies can be documented using UNII or
2. There is ambiguity in QDM vs. cCDA for non-drug allergies
standards (SNOMED CT for QDM vs. UNII in cCDA).
Problems Found In Allergies #2
• Consolidated CDA standards for allergies to medications are
inclusive of too many NDFRT classes:
• Fact: Most EHRs document allergy classes that span a very
narrow subset of NDFRT classes: The industry uses one of the 5
major vendors supplying drug terminologies in the U.S. which
collectively map to a very narrow subset of NDFRT.
• Problem: The standard reference a dynamic value set
(2.16.840.1.113883. which spans all NDFRT
Chemical Structure, Physiologic Effect and Mechanism Of Action
classes of which most do not map to any of the allergy classes
that are found in EHRs today.
Examples Of Problems Found In Allergies
Observed usage patterns for the documentation of medication allergens typically span
therapeutically active ingredients (e.g., amoxicillin, codeine), brand names (e.g., LIPITOR, PERCOCET)
and allergen groups (e.g., “Penicillins”)… named CDA Release 2.0 constraints for allergy observations
are inconsistent with observed usage patterns… specifically:
• Observed usage patterns for the documentation of medication allergens typically span
therapeutically active ingredients (e.g., amoxicillin, codeine), brand names (e.g., LIPITOR,
PERCOCET) and allergen groups (e.g., “Penicillins”)… named CDA Release 2.0 constraints for
allergy observations are inconsistent with observed usage patterns… specifically:
• Table 117 references “Medication Brand Name” examples that are specific to the Semantic Brand
Drug (TTY = SBD)… the “Brand Name” TTY would be more appropriate
• Table 118 references “Medication Clinical Drug” with examples specific to the Semantic Clinical
Drug (TTY = SCD)… it is unusual to have allergens reported at this level. More commonly,
“ingredient” (TTY = IN or PIN) would be observed for therapeutically active ingredient allergens
• Table 119 references “Medication Drug Class” examples that reference NDF-RT, this is correct, but
would suggest that the RxNorm RXCUI for the “integrated” NDF-RT external code into RxNorm
would be the constraint used
• Table 120 references “Ingredient Name Value Set” examples for FDA UNII; this is confusing… the
more practical use would be RxNorm IN/PIN for therapeutically active ingredients and FDA UNII
for “excipient” or “inactive” ingredients, using the RxNorm RXCUI for the “integrated” FDA UNII
external code within RxNorm
Proposed Solution For Allergies
• For Problem #1 we propose inclusion of a dynamic value set
for RxNorm ingredients in consolidated CDA to be added for
medication allergies (TTY of IN, PIN and MIN).
• For Problem #2 we propose adoption of a starter set of allergy
NDFRT classes (an allergy classes value set) to what was
recommended by NCPDP officially submitted on 5/24/2011 to
the HITSC Vocabulary Task Force (attached) which was a result
of an analysis of the most prevalent allergy classes used in the
industry (38 in total). This can be expanded over time.
Problem Found In Standard For Medication
• Consolidated CDA is too restrictive in value set definition for
• Fact: Patient medication lists in EHRs today are largely documented
not only at the SCD and GPCK levels but at the corresponding SBD
and BPCK levels. Additionally, during patient interview as well as in
the inpatient setting meds are frequently documented at more
general levels such as SCDF, SBDF and even BN or ingredients (but
rarely at the SCDC level especially for multi ingredient drugs).
• Problem: Consolidated CDA references a dynamic value set that is
only at the SCD and GPCK levels which make it very difficult (in many
cases impossible) to translate med list entries that are at the more
abstract levels (e.g., SCDF, SBDF, BN, ingredients) because of the oneto-many relationships seen in translating in that direction.
• Propose Solution Summary: Expand TTYs that can be use in the
Example Of Medication Documentation
Standard Problem
• Frequently in the inpatient setting orders are specified at a
broader level than what we see with prescriptions (outpatient
Inpatient Order
Outpatient Prescription Order
Digoxin .375mg po daily
Digoxin .125mg tablet QTY: 90
Sig: 3 tablets po qd
Zocor 30mg po qd
Zocor 10mg tablet QTY:90
Sig: 3 tablets po qd
Broader RxNorm
types: BN, ingredients
More specific RxNorm
types: SCD, SBD (note
strenghts vs. dose in
Proposed Solution For Medications Value Sets
• Expand the RxNorm TTYs for the dynamic value set to include
also include IN but generally that adds more noise than value.
• For the “shall” constraint for coded product name (generics) :
• For the “may” constraint for brands: BN, BPCK, SBD, SBDF, SBDG
Problems Found In CQM Value Sets For
• NQF CQMs reference RxNorm based value sets that are too restrictive for practical
and accurate reporting.
• Fact:
Most certified commerical EMRs use vendor drug data from one of the 5 NLM sources
(FDB, Medispan, Multum, Gold Standard, Micromedex
The value sets referenced for quality measure are frequently specified to be more
detailed than their intended practical definition.
• Problem: Reporting entities that know how to use the advanced services/features
of their drug vendor products to identify patients that meet the intent of the
quality measure but will be penalized for submitting QRDA reports that reference
RXNORM codes that are not stated in the intentional value set definitions.
• Examples: Proprietary drug databases (NLM sanctioned sources) have enhanced drug
classifications that can identify drug concepts intended for specific therapeutic uses and
thereby potentially not only identify drugs missed in RxNorm based extensional value
sets but also suppress reporting on patients that are on same ingredient drugs that do
not meet the therapeutic intent (such as heparin flushes for anticoagulation therapy).
Proposed Solution For Medications Value Sets
• The Consolidated CDA use of dynamic value sets will certainly help
but the problem still remains for reporting entities that wish to
leverage enhanced attributes to identify drugs that meet the value
set intent.
• Proposed solution is one of flexibility rather than major revision:
1. Publish additionally “in plain English” a intentional definition of the
value set (e.g., anticoagulants including LMWH, heparins, oral
aspirin excluding heparin flushes). The extensional value set can be
used by reporters either as is or as a representative example of
drugs to be used.
2. Allow the option to either use the specified value sets or
alternatively report using RxNorm TTYs in (SCDF, SBDF, SCD, SBD,
3. Regardless of whether the reporter uses the referenced extensional
value set or the broader RxNorm value set (#2) they would still
need to submit the QRDA reports.
Justification For Proposed Medications Value Set Solution
We’re giving the reporting entities the option of doing more sophisticated
/accurate reporting (because the vendor databases have more precise
attributes for tuning to get back the right drugs that meet the intent of the
value set stated in “English”). They’re not being penalized for discovering
drugs that meet intent but were missed by the value set.
We’re still allowing the current NQF value sets to be used as per the original
rule design for those reporting entities that just want to use the provided
extensional value set because it’s easier for them.
CMS would now be able to easily analyze the QRDAs that were submitted but
had new RxNorm concepts that were not in the original value sets to see what
should be added vs. what was inappropriately reported during this field test
period of MU 2. These reports can help identify what term types would be
more appropriate for intentional definitions as well as what extensional value
set gaps existed based on what was submitted vs. what was submitted that
does not meet the intent. These statistics can be posted as evidence/guidance
for adding more constraints but based on real world observations during that
Intentional value sets can then be stated citing observed evidence / statistics
from #3 with more confidence to reporting entities.
Summary Of Semantic Types Use In Real World Settings
Use Contexts
Medications Semantic Types
formulations Medication
abstract IN,MIN,PIN
on/decision support
Medication functional
classifications: NDFRT
Mechnism Of Action,
Structural, Pharmacologic
Medication therapeutic
classifications: (Not part of
standards today)
xxx (really needs a valueset
since not all classes are
applicable in each axis)
Generally documented at
that higher level and not at
the formulation level.
Medlist documentation
Generally documented at
the routed drug + dosing or
the formulation level in
EMRs. The latter commonly
done during patient
interview of med history.
Medication ordering
Mostly at the formulation
level for outpatient setting
but in inpatient, commonly
ordered as a routed
medication with dose
information as well , e.g.,
“captopril oral” + 50mg TID
vs. “captopril 25mg oral
tab” + 2 tabs TID.
Quality measures
inclusion criteria
Quality measures
The number of x’s reflects relative level of use for each context.
Generally not drug strength
specific but sometimes can
Generally similar to allergies
and intolerances and rarely
at the drug formulation
Routes Problem
• HL7 CDA constraints reference FDA Medication Routes
• Draft eMeasures referenced Snomed CT clinical routes
• Problem – expressed eMeasures and eMeasure value set
inconsistent with HL7 constraint
• Problem – Drug Knowledge Base providers and 2014 CEHRT
developers do not have clear direction on which route of
administration code set to use within clinical information
exchange and CQM applications
• Recommended Solution – eMeasure value sets should be
constrained to FDA Routes
Future Lab orders and scope
• Comments for "Test Data for 170.314(e)(2)" document
• ONC definition of the Common MU Data set item (11) requires
LOINC (170.207(c)(2)) for laboratory test results. It is our
understanding that there is no requirement for lab orders to
be coded in LOINC. In the Care Plan section (item L.b.ii, page 5
in the Test Data document), lab tests scheduled for the future
would not need to be coded in LOINC since they are "orders"
(have not been performed yet). Also, the code mentioned in
the test data (30313-1) is to be used for a laboratory result of
hemoglobin not for an order.

similar documents