Anthem ICD-10 Adoption
Foundational Insights
Enterprise ICD-10 Adoption Program
Michael P. Fierro
Director, External Partner Readiness & Advocacy
What is ICD-10?
ICD-10 is the most profound and foundational change our industry will face in the next decade
Foundational change in moving from ICD-9 to ICD-10
Federal Mandate
‘
ICD-9-CM
(Diagnosis)
5 characters
≈14,000 codes
ICD-9-CM
(Procedure)
5 characters
≈ 4,000 codes
ICD-10-CM
7 characters
> 68,000 codes
ICD-10-PCS (In
patient)
7 characters
>
72,000 codes
Re-factors 40 years
of business and
technical
infrastructure
ICD-10-CM
is the diagnosis code
revision to ICD-9-CM. ICD-10 was
endorsed by the Forty-third World
Health Assembly in May 1990.
The Differences are as
Fundamental as the Differences
Between Speaking English and
French
ICD-10-PCS
(Procedure Coding
System) is currently designated to
replace Volume 3 of ICD-9-CM for
hospital inpatient use. It identifies
inpatient facility services in a way
not directly related to physician work
but directed towards allocation of
hospital services
The International Classification of Diseases version 10. The ICD is the international standard
diagnostic classification for general epidemiological, health management purposes and clinical use
The U.S. Department of Health and Human Services has finalized the ICD-10 compliance date as October 1, 2014
Foundational Assumptions
Numerous assumptions are relevant to Anthem’s adoption of ICD-10 and readiness
1.
Considering history, Provider adoption will be significantly less than 100% on the compliance date (e.g. hospital X is
non-compliant and is only able to send ICD-9 after the compliance date); Anthem is developing an active
engagement plan across our networks to influence higher adoption than with historical change
2.
There is not a consistent or clear industry perspective on treating the slow adopting provider. Historically, the federal
government has granted relief for slow adopting providers, and Anthem has developed solutions to accommodate
that potential outcome for ICD-10.
3.
Translation between ICD-9 and ICD-10 is foundational to refactoring all business and technology as well as ongoing
operations after the compliance date – we will be ‘bi-lingual before and after the compliance date and accommodate
the need for reference points between the code sets before the compliance date (e.g. recoding policies, guidelines,
and business rules) and after the compliance date (e.g. accommodating the need for longitudinal views of data)
4.
Dual Processing (the ability to accommodate both ICD-9 and ICD-10 end-to-end in all business and technical
processes without any data transformation on the record) is considered the only viable business and technical
solution to adopting the ICD-10 code set
5.
Operational Crosswalks (transforming the data on an in-bound or out-bound record from one code-set to another
code-set) should not be confused with translation between ranges of codes,
6.
Crosswalks represent significant risk to claim accuracy and efficiency , as well as reporting and analytics. Therefore
are considered non-viable solutions to adopting ICD-10 and should not be supported within the industry given the
inconsistency and associated inefficiency that may be added to the collective consumption of codes across
business and technical processes
Translating vs. Crosswalks
There are significant differences between a ‘dictionary’ that maps ranges of codes and a ‘crosswalk’ that
deterministically transforms data on transactions
Translation Maps between
ICD-9 and ICD-10 are the sources
of truth in translating between
the two code sets
They represent one to many,
many to many, and 1:1
relationships
Translation Maps should be
clinically sound, and given they
propagate through processes
and technologies, they should
be financially equivalent as
they should not produce
aberrations in reimbursement
Translation
Maps
Deterministic Crosswalk
Services
Translation
Services
Translation Services are
processes and automated utilities
that assist in refactoring business
processes and technologies as
well as assist in transaction
processing; examples include:
•
A library that houses Translation Maps
used to refactor all ICD-9 based
references into ICD-10 (e.g. med policy
and provider rate sheets)
•
An enterprise service, perhaps part of a
broader SOA, that assists in matching
records during the adjudication process
(e.g. ICD-10 claim after the compliance
date matched to an ICD-9 authorization
received before the compliance date)
Deterministic Crosswalks are a
subset of translation maps and
services that are used to
deterministically transform
ICD data on records
to allow
systems and processes to
operate in one code set only
•
Forward Crosswalks
deterministically transform
ICD-9 into ICD-10
•
Backward Crosswalks
deterministically transform
ICD-10 into ICD-9
Translation Maps and Services
Translation Maps and Services are a foundational aspect of adopting ICD-10
• Foundation for transitioning systems and processes to ICD-10
• Complicated by mapping general codes to very granular codes and by many to many relationships • Example: contract that pays for
bariatric surgery directly against ICD codes (find all 10s with clinical equivalence to the 9s currently used; update contracted pricing, system logic, etc.)
Forward Mapping
• Foundation for longitudinal analysis and for insulating systems from ICD-10 code set • Complicated by many to many
relationships
• Example: Diabetes Mellitus with underlying Coma in ICD-10 maps to Diabetes Mellitus and Coma in ICD-9 ICD-9 ICD-10 ICD-9 ICD-9 ICD-9 ICD-9
Backward Mapping
Usage
− Upgrade Provider Reimbursement /Contracts, MedicalPolicies, Clinical Protocols, and Business rules, etc. − Comparing records of
different code sets during transaction processing (e.g. ICD-10 claim and ICD-9 auth)
Usage
ICD-10 ICD-9 ICD-9 ICD-10 − Longitudinal analysis − Comparative analysis forICD-10 reimbursement modeling
− Comparing records of different code sets during transaction processing (e.g. ICD-10 claim and ICD-9 auth) ICD-10 ICD-9 ICD-10 ICD-9 ICD-10 ICD-10 ICD-10 ICD-10
Crosswalks
Crosswalks are a subset of translation Maps and Services supporting very defined
strategies...transforming data in order to insulate systems and processes
Forward Crosswalk
• Deterministically transforms ICD-9 data to ICD-10 data; purposely excludes translation options for purposes of getting as close a possible to 1:1 relationship
• Operates without supporting clinical evidence
• Mapping used in operations such as claim adjudication for systems that can only adjudicate ICD-10
• Example: inbound claim has ICD-9 code and crosswalk deterministically selects one of multiple ICD-10 choices
ICD-10 ICD-9 ICD-10 ICD-10 ICD-10 ICD-10
x
x
x
x
• Mapping used in operations such as claim adjudication for systems that can only adjudicate ICD-9 (e.g. insulation strategy) • Deterministic backward mapping eliminates
many ICD-9 codes that are currently in use, thereby further reducing the ICD-9’s already insubstantial granularity
• Less ambiguous than forward mapping
ICD-9 ICD-10 ICD-9 ICD-9 ICD-9 ICD-9
x
x
x
x
Backward Crosswalk
Usage
− Systems that only adjudicate ICD-10 natively − Transform claims with ICD-9 and date of
service/discharge after the compliance date (when system can accommodate only ICD-10)
Usage
− Systems that only adjudicate ICD-9 natively − Transform claims with ICD-9 and date of
service/discharge after the compliance date (when system can accommodate only ICD-10)
ICD-10 Claim
ICD-9
Claim ICD-10 Only Processing
ICD-9 Claim
ICD-10
Claim ICD-9 Only
Solution Options Evaluation Criteria
Given the practical reality of ICD-9 codes existing post compliance from non compliant providers (and
run out) multiple options have been considered
How clinically equivalent are the mappings between
code sets
How equivalent are payment outcomes with
today’s outcomes
What are the build, integration and maintenance implications
What are the conflicts with corporate policy and
strategy
Technology
1
2
3
4
Forward Crosswalk
Backward Crosswalk
Dual Processing
Make technology and business process operate in ICD-10 only; requires
transformation of any inbound ICD-9 to an equivalent ICD-10
Insulate existing systems and processes by transforming inbound ICD-10 to an
approximately equivalent ICD-9
Allow processes and technology to utilize the code set corresponding to the value provided on the incoming transaction (either ICD-9 or ICD-10); effectively making the end to end process bi-lingual; this does assume
translation references for refactoring as well as some in-line processing (e.g. match 10 claim w/ 9 auth)
A B C
Evaluation Criteria
Solution Options
SAMPLE Findings – Forward Crosswalk
Forward Crosswalk / ICD-10 Only Strategy is considered not clinically viable
Code Type*
Able to Choose a
Single Mapping?
Yes (% of total) No (% of total)9 10
Diagnosis
Codes
74
(81%)
17
(19%)
9 10
Procedure
Codes
12
(10%)
105
(90%)
Code mapping analysis has revealed significant challenges in building deterministic crosswalks
that would allow effective and accurate transformation of data on the in bound claim record from
ICD-9 to ICD-10…….these findings were applied to scenarios across the claim value chain (e.g.
intake, pricing, benefits, care management analytics)
1
*Codes sourced from three platforms representing commercial and government sponsored business; In-Patient Claims data were used from 1/1/2009 – 12/31/2009
Key Findings
•
Anthem Clinical coding team was unable to determine a single
mapping choice for 19% of Dx codes and 90% of Proc codes
•
Inability to make selection was mostly due to lack of supporting
clinical data that would justify selection of data that are
significantly more detailed (e.g. ICD-10 presents anatomical
region, approach, device, contrast, specific trimester, etc. that
are not consistently provided on an ICD-9 claim);
additional
data NOT necessarily accessible from the claim and
requesting such data from provider is not practical
Anthem Solution Assumption
Eliminate Froward Crosswalk / ICD-10 Only Strategy for claim
adjudication
While the % of codes without a ‘single mapping solution’ is not extremely high, there are
technical and business issues that negatively impact this solution scenario.
Key Findings
•
Anthem Clinical coding team was unable to determine a
single mapping choice for 4% of Dx codes and 8% of
Proc codes
•
Targeted codes (ICD-9) did not provide clinical
equivalence due to increased specificity @ source code
level (ICD-10) and required additional clinical information
(e.g. medical record operative notes, etc.)
Anthem Solution Assumptions
Reimbursement analysis influences viability of backward
crosswalk
Additional analysis on complex mappings (e.g. instances
where 1 ICD-10 results in multiple ICD-9s as well as
conflicts with corporate policy) complicates technical
solutioning as well as analytics
Code Type*
Able to Choose a Single
Mapping?
Yes (% of total) No (% of total)10 9
Diagnosis
Codes
279
(96%)
12
(4%)
10 9
Procedure
Codes
5638
(92%)
522
(8%)
* In-Patient Claims data 1/1/2009 – 12/31/2009 across all lines of business were used by clinical and coding teams to develop corresponding set of claims coded in ICD-10
SAMPLE Findings – Backward Crosswalk
Backward Crosswalk / ICD-10 Only Strategy passes the clinical equivalency test for claim
adjudication but not analytics
Evaluation Dimensions
Solution Complications
Dual Processing remains the most optimal solution which implies renovation across systems and business
functions (and further decisioning on renovate or retire)
1 2 3 4
Forward
Crosswalk
NOT
VIABLE
NOT
VIABLE
ABackward
Crosswalk
BVIABLE
Dual
Processing
C How clinically equivalent are the mappings betweencode sets
How equivalent are payment outcomes
with today’s outcomes
What are the build, integration and
maintenance implications
What are the conflicts with corporate policy
and strategy
Code Set Translation Effort Technology Decomposition Effort Business Decomposition Effort ICD-9 to ICD-10 ICD-10 to ICD-9
Both ICD-9 and ICD-10
Primary Solution Assumption
:
all systems and processes
should consume and adjudicate
the code-set natively, with no
data transformation.
Unable to make deterministic forward crosswalk 90% of time for procedure codes Risk - unknown integration complexity Cost - build, integration, reconciliation of crosswalk is not FREERisk - Inconsistent payments across platforms Issue - Inconsistent w/ provider contracting 77% of DRGs mismatched and 27% of reimbursement impacted influences admin cost and cost of
care misstatement IN S U L A TE R E N N OVA TE 58% of DRGs mismatched and 25% of reimbursement impacted influences
admin cost and cost of care misstatement (and this was only one of
many methodologies analyzed)
Amount / Type of payment outcome modeling is being
confirmed
Anthem
Business and IT
Renovation
I.
Renovate policies
/procedures / job
aids
II.
Renovate systems
a) structural changes
(e.g. expand field
length)
b) logic changes
(e.g. rules that
apply benefits)
Claims
Processing
and Surround
Clinical and
Medical Policy
Provider,
Actuary,
Underwriting
Reporting
and Analytics
External
Readiness
• Remediate 6 claim processing and over 200 other IT assets• Claim volume > 400 million per year
• Members > 30 million
• Remediate 9 systems • Update 2,000+ policies,
procedures, and job aids
• Remediate 21 systems • Update rate sheets for 1,500
facilities
• Engage and educate ~268,000 Providers
• Sunset 62 assets • Upgrade 14 systems
• Update standard and ad-hoc reports
• Prepare and monitor readiness of • 14,000 Data Trading
Partners
Imbedded
ICD-10 Multi-Year Timeline:
Leveraging the ICD-10 delay to improve post-compliance
outcomes by increasing focus on Business Continuity Testing with external partners
2011
2012
2013
Design
Build
End to End Internal Test
Business Continuity Testing
Outcomes
•
Fluency
in
ICD-10 (systems,
processes, and
people)
•
Corporate
Continuity
in
admin cost, cost
of care, MTM,
controls)
•
Quality
improvement in
service / care
delivery
Implement
Add Value to the Customer
Training and other Operational Readiness Activities
Vendor/Trading Partner Readiness Activities
Provider Engagement, Contracting and Readiness Activities
Monitor
Our Approach to Transitioning Anthem to ICD-10
The Code Set Competency Center will provide distinct services to the platform and functional remediation
teams throughout the remediation lifecycle
Context:
• There are hundreds of thousands of ICD-9 references across Anthem’s systems and processes
• Manually configuring against all ICD-9 references would equate to millions of hours of effort
• Anthem has developed a structured approach to renovating all assets.
1
Establish ICD-9-to-ICD-10
Translation Maps
2
Configure Systems/Renovate
Processes
3
Test and Refine
Configuration/Renovation
ICD9/10 WLP Reference Map Expose mappings ICD-10 Training ICD-10 QA Program ICD10 Subject Matter ExpertiseTest Scenarios and Data for High-Risk
Areas
Support for Refinement of
Maps
The Code Set Competency Center:
•trains the remediation teams on coding patterns and new concepts
•provides subject matter expertise and support manual review of lists and system configuration
•implements QA program
•identifies test scenarios/ data to help identify fall out from configuration Co d e S et Co mp eten cy Cent er A ss ets
The Code Set Competency Center: •creates WLP Reference Map (e.g. using
targeted list recoding as a starting point) •exposes WLP Reference Map and specific
list maps
•develops training materials for remediation teams
The Code Set Competency Center: •supports testing of high risk areas •supports refinement of code and list
maps