2016 Annual Conference Data Capture Challenges for Commercial Risk Adjustment “Data Capture” represents the successful gathering and transfer of complete and accurate data beginning with the member-provider encounter through to the EDGE Server submission in order to optimize commercial risk adjustment reimbursement. Presentation by: Robin Lemoine
61
Embed
Data Capture Challenges for Commercial Risk Adjustment · Data Capture Challenges for Commercial Risk Adjustment “Data Capture” represents the successful gathering and transfer
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2016 Annual Conference
Data Capture Challenges for Commercial Risk Adjustment
“Data Capture” represents the successful gathering and
transfer of complete and accurate data beginning with
the member-provider encounter through to the EDGE
Server submission in order to optimize commercial
risk adjustment reimbursement. Presentation by:
Robin Lemoine
2016
Annual
Conference
2
Definitions
Commercial RA: Affordable Care Act (ACA) Risk
Adjustment (RA) = Marketplace RA = HHS RA
CMS: Centers for Medicare & Medicaid Services (CMS) is
part of Dept. of Health and Human Services (HHS).
Issuer = Health Plan = HIOS (5-digit).
EDGE = External Data Gathering Environment. Issuers
submit data to EDGE for commercial risk adjustment.
TPA = Third Party Administrator.
REGTAP = Registration for Technical Assistance Portal
(REGTAP) serves as an information hub for CMS technical
assistance related to Marketplace and Premium Stabilization
programs.
EMR = Electronic Medical Records System.
2016
Annual
Conference
3
EDGE Data Flow
3
Member (MBR)
Provider (PRV)
Health Plan
TPA EDGE
Claim Submission
Submission
-Enrollment
-Med Claim
-RX Claim
-Plan Data*
Claim
Adjustment
Error Reporting &
Error Management Error Reporting
Submission
-Enrollment
-Med Claim
-RX Claim
-Plan Data*
*Plan Data can be corrected intermittently on the EDGE via a distinct update process.
2016
Annual
Conference
4
Member-Provider Encounter Never Occurs
4
May be difficult for member to visit provider.
Member is not aware of or ignores chronic condition.
2016
Annual
Conference
5
Challenges:
1) ROI for wide-net approach.
2) Identification of members suspected of chronic conditions.
• New population with limited data.
• Low member retention.
Member-Provider Encounter Never Occurs - Options
Prospective Member Assessment Options:
1) Member outreach initiatives.
2) Other issuer-provider coordinated initiatives.
2016
Annual
Conference
6
Medical Documentation Issues
6
Undiagnosed conditions or illegible medical notes.
Practitioner documentation insufficient to support and
substantiate coding for claims or encounter data. • Chronic conditions that potentially affect the treatment
considered are not documented.
• Condition descriptions not to the highest level of specificity
enough.
Does not qualify as a medical record. • Incomplete progress notes (for example,
unsigned, undated, insufficient detail);
• Unauthenticated medical records (appropriate
signature(s) missing).
2016
Annual
Conference
7
Medical Coding Issues
7
DX codes not at the highest degree of specificity.
Medical code omission.
Misinterpretation of notes.
Other errors.
2016
Annual
Conference
8
Electronic Medical Records (EMR) System Issues
8
Data entry/user operator errors.
EMR system limitations. • e.g. Max of 8 DX codes per claim.
EMR system failure. • e.g. Data corruption or loss.
2016
Annual
Conference
9
Provider Claims Billing System Issues
9
Incidental claims load failure
from EMR system.
Medical Billing System
limitations • e.g. DX code 6-digit max length.
Misconfigured system or
computer programming defect. • e.g. Invalid source-to-target data
mapping.
2016
Annual
Conference
10
Provider Issuer – Direct
Electronic Submission
10
Data Loss Risks: Extraction and transformation of data.
Claim file transfer.
Issuer claim intake.
Direct electronic submission to Issuer Fewer data loss risks (in theory).
Difficult to maintain as requirements change.
Less standardization than clearinghouse.
2016
Annual
Conference
11
Provider Issuer –
Electronic Submission to Clearinghouse
11
Data Loss Risks: Provider extraction/transformation of data.
File transfer to the clearinghouse.
Clearinghouse claim intake.
Clearinghouse transformations/extractions.
File transfer to Issuer.
Issuer claim intake.
Most standardization.
Difficult to maintain as requirements change.
Data loss risks are more numerous.
2016
Annual
Conference
12
Provider Issuer – Paper
Claim Submission to Vendor
12
Data Loss Risks: Provider extraction/transformation of data.
Claim transfer to the paper claims processor (vendor).
• Was the presentation helpful in for health plans? for providers?
• What information could have been added to improve the
presentation?
• Can you offer additional insight?
2016
Annual
Conference
34 *Slide 10 from “2016 PLAN DATA UPLOAD”, REGTAP Presentation (March 1, 2016).
Appendix A – Plan Data Upload
2016 Annual Conference
The New Era of Risk Adjustment Operations: 2017 and Beyond
Duke Owen, Mile High Healthcare Analytics
Richard Lieberman, Mile High Healthcare Analytics
2016
Annual
Conference Today’s Agenda
Risk-adjusted Marketplace products may be on the chopping block. Or not!
Medicaid risk adjustment is all the rage
Changes to Medicare-Advantage risk adjustment that take effect in 2017
Impact of the ongoing migration from RAPS to EDPS data submission
2
2016
Annual
Conference
What Will Happen in 2017?
3
There definitely is a new “sheriff” in town
The ACA will change, but what we do not know is by how much and how soon!
Converting political rhetoric into public policy is harder than it appears
Medicare-Advantage is likely to continue to grow in importance
Even if Medicaid is turned back to the States, that is where the innovation and strong oversight are!
2016
Annual
Conference
Medicaid is a Rapidly Growing Program
• Over the past half-century, Medicaid has transformed from a niche program to become a linchpin of the U.S. health care system.
• It is today the largest single insurer, serving nearly 73 million low-income and medically vulnerable individuals
• Nearly 16 million people have gained Medicaid coverage under the Affordable Care Act’s expansions; most had previously been uninsured
• 88 percent of adults are satisfied with their new Medicaid coverage: 77 percent rate it as either good, very good, or excellent
Source: M. Z. Gunja and S. R. Collins, "Five Facts About Medicaid," To the Point, The Commonwealth Fund, Dec. 2, 2016
2016
Annual
Conference
What the Nation Spends on Medicaid
Total Medicaid expenditures (Federal and State combined) for medical assistance payments and administration are estimated to have grown 12.1 percent in 2015 to $554.3 billion
◦ The Federal government share of total spending on Medicaid is $347.5 billion in 2015, representing 63 percent of total Medicaid benefit expenditures
◦ State Medicaid expenditures for benefits and administration are estimated to have increased to $206.8 billion in 2015, a growth rate of 5.9 percent
2016
Annual
Conference What the Nation Spends on Medicaid
Under current law, by 2024 Medicaid expenditures are projected to:
◦ Reach $920.5 billion, increasing at an average rate of 6.4 percent per year
Federal spending on Medicaid is projected to reach $563.2 billion, or 61 percent of total spending
State spending is projected to reach $357.3 billion by 2024
2016
Annual
Conference Risk Adjustment Tools in Current Use for Medicaid
The Chronic Illness and Disability Payment System (CDPS) and the MedicaidRx system -developed by Richard Kronick and Todd Gilmer at the UC-SD
Adjusted Clinical Groups (ACGs) - developed by Jonathan Weiner and Barbara Starfield and other researchers at the Johns Hopkins University.
Diagnostic Cost Groups (DxCG) – developed by Arlene Ash and Randall Ellis of Boston University
Clinical Risk Groups - developed by DRG team at 3M
Episode Risk Groups (ERGs) – developed by Symmetry, now owned by Optum
7
2016
Annual
Conference How Can I Find Out What a Particular State Is Using?
There is no single source or location to look at that is guaranteed to be up to date!
There are State Medicaid & CHIP Profiles at: https://www.medicaid.gov/medicaid/index.html
◦ But many of these are outdated!
To understand what is happening in most states requires exhaustive, state-by-state research
◦ MCO contracts have to be reviewed
◦ EQRO reports must be read
◦ Multiple state web pages must be perused
Mile High Healthcare Analytics has compiled a comprehensive database
Use of Risk Adjustment in Medicaid is Continually Evolving
• 15 states use the combined CDPS/MedicaidRx model
• 5 states use the diagnosis-based CDPS Model
• 4 states use MedicaidRx alone
• 3 states use the Johns Hopkins ACG System
• MA uses DxCG
• NY uses CRGs
• AZ uses ERGs
2016
Annual
Conference Medicaid States without Risk Adjustment
2016
Annual
Conference How Medicaid Risk Adjustment Works
Risk scores are calculated for a group of members last year. These are called “Plan-level” risk scores
The group-level average risk score from the prior period is applied to a different group of enrollees in some future fiscal year
For example, risk scores determined in 2015 using 2014 claims or pharmacy history will be used to set health plan rates in 2016
The risk score gets baked in to the rates prospectively paid to the plans
13
2016
Annual
Conference The Historical Health Plan Risk Score
14
0.95
0.89
1.07
1.40
1.8
0
1.35
1.01
2014
1.21
2016
2016
Annual
Conference
Medicaid Risk Adjustment: Value Proposition Medicaid risk adjustment is a zero-sum game
o Risk scores are calculated for each MCO and compared to the overall risk score for all MCOs within the same aid category (e.g., TANF, ABD)
o CMS requires states to ensure budget-neutrality
o MCOs “win” by submitting complete and accurate encounter data
Return on investment has two components:
o What the MCO prevents in redistribution to its competitors
o Truly incremental premium, obtained by closing diagnosis coding gaps, capturing withholds & bonuses tied to quality improvement, & managing MLR
$100 in Premium From State to Cover
Medical Costs across 3 MCOs
If Relative Risk Scores Differ Across the
MCOs, then the $100 is Divided
Proportionately
Total
Return
from Risk
Adjustment
Prevented
Payment to
other
MCOs
(unobserve
d)
Received
Payment
(Increment
al
Premium)
MCO takes no action
All-MCO average risk score by aid
category
Risk Score after Revenue
Management
& Encounter Data Interventions
2016
Annual
Conference
Diagnosis Code Collection History for MA
Since the inception of comprehensive risk
adjustment in 2004, through the end of 2011, the
sole source of diagnosis codes were RAPS files
◦ Back in 2002-2003, RAPS files were conceived,
bowing to Congressional and industry pressure
◦ RAPS are a very limited extract from claims data:
five fields, with filtering the responsibility of the plan
(or its vendor)
◦ CMS has been intending to sunset the RAPS
submission process for several years
2016
Annual
Conference
RAPS Data Isn’t Good for Your Health
Historically, Medicare Advantage Organizations (MAOs) have done their own filtering and submitted to CMS risk adjustment eligible diagnoses in a minimum RAPS data file
◦ The filtering process is fraught with many deficiencies– CMS’ specifications (based solely on provider specialty) will often filter incorrectly
◦ Plans have struggled with getting the filtering right
◦ The diagnosis code filtering (editing) “guidance,” which requires plans to determine which diagnosis qualify for RAPS, practically sets plans up for compliance and RADV problems
2016
Annual
Conference Migration from RAPS to EDS data
18
For PY 2015, diagnoses (2014 dates of service) submitted on encounter data records served as an additional source of diagnoses in the calculation of the risk scores
For PY 2016 (2015 dates of service), risk scores used for payment were a blend of two risk scores:
◦ 10% of the risk score calculated using diagnoses from encounter data records and FFS (FFS) claims added to 90% of the risk score calculated using diagnoses submitted to the Risk Adjustment Processing System (RAPS) and FFS claims
• CMS began collecting encounter data from MA plans in 2012
2016
Annual
Conference
RAPS Will (Finally) Sunset
19
The 2017 Final Notice establishes the phase-out
schedule for RAPS data submission:
◦ PY2016: 10% EDS and 90% RAPS
◦ PY2017: 25% EDS and 75% RAPS
◦ PY2018: 50% EDS and 50% RAPS
◦ PY2019: 75% EDS and 25% RAPS
◦ PY2020: 100% EDS data
Sun-setting RAPS transfers the responsibility for
diagnosis code filtering from plans and vendors to
CMS
2016
Annual
Conference
EDS Data Will Be A Positive Development
Yes, migrating to the (sort-of) 837 claims format is a pain in the…..
CMS will extract (i.e., filter) diagnoses submitted to EDS that are eligible for risk adjustment
◦ In the long-run, this will be good for plans. But in the short-run……
CMS will eventually recalibrate the risk adjustment model with MA data, instead of FFS
◦ Be careful what you wish for!
Coding Pattern Adjustment will end when EDPS is fully phased in
2016
Annual
Conference
Practical Significance of RAPS Phase-Out
21
Beginning with 2016 payments, deficiencies in the EDPS data stream will alter a contract’s aggregate risk score
For the first several years of the EDPS mandate, MA Plans viewed EDPS submissions as solely a compliance function (something you “gotta” do)
◦ Data must be used operationally in order to find errors and omissions
The linkage of disease to treatment (or lack thereof) will now be apparent in data
Merely achieving technical compliance with EDPS specifications is not sufficient
2016
Annual
Conference Collecting and Storing Comprehensive
Data
22
Plans must ascertain if their “data lakes” contain all of the data
◦ Multivariate statistical models can be used
◦ Tie-back to the aggregate financials
◦ Provider-level surveys
◦ Use retrospective medical record review
A data lake is a storage repository that holds a vast amount of raw data in its native format until it is needed
◦ Enterprises with lots of silos often create multiple data warehouses or multiple data feeds
2016
Annual
Conference New Filtering Method Yields Different Results
23
The RAPS-based method (relying on physician specialty) often results in too many diagnosis codes being selected
◦ Physician specialty is a poor proxy for a face-to-face visit
The EDS method relying on procedure codes and UB-04 bill types is much more accurate
◦ The procedure codes do a better job of identifying face-to-face clinician visits
◦ Yet in the short run, most plans are going to experience a negative impact to their risk scores
◦ In the long-run, compliance will pose fewer risks for health plans
2016
Annual
Conference
Remaining Revenue Neutral
24
To remain revenue neutral, plans must ensure that they have clean claims/encounters for every service rendered by the plan
◦ Fixating on the CMS’ filtering methodology is a waste of valuable plan time and effort– focus on collecting (or not losing) all of the data!
Plans lose far more diagnosis codes to missing data than they will lose to the filtering paradigm
◦ Does anyone look at claims triangles by delegated entity?
◦ Some plans and provider groups look at counts of encounters, but often don’t review specialist claims in sufficient detail
2016
Annual
Conference
Processing Data So It Passes the Filtering Logic
25
On institutional claims, the UB-04 bill type field serves as the gate to further processing of the claims for risk adjustment
◦ All diagnosis codes from “bona fide” inpatient claims with bill types of ‘11X” (hospital inpatient) and “41X” (religious nonmedical) are accepted without the need to apply further edit logic
But, inpatient “observation stays,” where the bill type may be “12X” require the processing of the associated CPT-4/HCPCS code on the UB-04 line items
No CPT-4 or HCPCS code was needed for these diagnoses to pass through to RAPS
2016
Annual
Conference
EDS provides a surveillance opportunity for CMS
Medicare Part C is the last Medicare “provider” to
have to submit full encounter data
◦ CPT codes now matter!
CMS will be using EDS data for much more than computing risk scores
◦ Recalibration of the CMS-HCC models, using EDS instead of FFS data
◦ MedPAC has used EDS data to examine trends in in-home assessments
◦ If the industry doesn’t accept the fact that encounter data quality is a concern of every player, they may be in for a tremendous surprise!