December 1, 2015 CMS & HHS Risk Adjustment 101

RFP SUBMISSIONSERVICE

TECHNICAL COLLABORATIONS

LATEST NEWS & BLOGSJOB BOARD

ACCESSRAPS/EDS USER GROUPS

PREMIERNETWORKING

EDUCATIONALWEBINARS

EDUCATIONALWEBINARS

BENCHMARKS & BEST PRACTICES

EXCLUSIVE WEBCASTS

RISE (Resource Initiative and Society for Education) was built to put youin touch with other movers and doers in your field. We give you opportunities to meet these people and hear what has made them successful in the business. Best practices. Lessons learned - benefit from the RISE community today!

CONNECT WITH RISE:

• RISE is a virtual community of professionals like you

• RISE User Groups are tackling problems together through collaboration and

shared learning

• RISE is building educational tools and programs including free webinars that will

boost your professional IQ

• RISE has a job board so you can go to one place where positions are posted that

are relevant to you and professionals like you in the managed healthcare world

• RISE provides an RFP routing service so you can reach highly qualified vendors you

may not have known about that want to compete for your business

TAKE CHARGE. JOIN RISE NOW. IT’S FREE AND IT IS DESIGNED FOR YOU.

CMS&HHSRiskAdjustment101 Risk Adjustment Academy:

The cornerstones

of risk adjustment

December 1, 2015Doubletree by Hilton, North Miami, FL

CourseDevelopmentandContributors• Brian Boyce, BSHS, CPC, CPC‐I, Chief Executive Officer and Managing Consultant,

ionHealthcare, LLC

• Kameron Gifford, CPC and Chief Executive Officer, Empirical Risk Management

• Richard Lieberman, Chief Data Scientist (aka "Mad Scientist") at Mile High Healthcare Analytics

• John Murphy, Principal, Risk Adjustment Consulting

• Contributors Also Include:

• Tim Buxton, Episource

• Stacey Hernandez, SCAN Health Plan

• Tam Pham, Peak Health Solutions

• Jimmy Liu, Altegra Health

• Michele Miller, Centauri Health Solutions

FacultyforDecember1,2015:

John Murphy, Principal,

RISK ADJUSTMENT CONSULTINGJohn has worked on some of the most challenging issues in the business of health care. His 20 years of experienceincludes roles as Board Member of Bayside Community Health Center in RI; Manager of Analytics at New England QualityCare (NEQCA), Manager of Risk Operations at Tufts Medicare Preferred, and Director of Contract Settlement and Analysisat Blue Cross Blue Shield of Massachusetts. As a Director at Blue Cross Blue Shield MA, he provided the key analytics tosupport a new risk adjusted contracting model with academic medical centers. He also worked with the Senior Scientistat the leading risk adjustment and predictive modeling firm to educate key leaders and physicians on risk adjustmentfundamentals and applications. In 2008, he was recruited back to Tufts Health Plan to manage the risk operations at theirMedicare Advantage plan. He worked directly with senior executives in developing and implementing risk optimizationand mitigation strategies. His experience there included operational oversight for both a National & Targeted RADVAudit. His areas of expertise include government risk adjustment models, both Medicare and Exchange, and othercommercial risk adjustment models. Beginning in 2015, Risk Adjustment Consulting is also offering a suite of strategicand operational consulting services designed for provider organizations to successfully make the transition to risk andvalue based contract models.

FacultyforDecember1,2015:

Tara Russo, Vice President of Medicare Risk Adjustment and HEDIS Quality Initiatives ISLAND DOCTORSTara works for Island Doctors, a provider organization based in St. Augustine,Florida, that operates 14 wholly‐owned offices in Clay, Duval, Saint Johns, Flagler,Putnam and Volusia counties, as well as manage a network of 32 affiliate providersthroughout these six counties and around the Orlando area.Prior to her current role, Tara previously worked as a financial analyst for Humana,Florida Hospital and Visiting Nurse Service of New York. Earlier, she was theFinance Manager for the Department of Medicine at Memorial Sloan‐KetteringCancer Center in New York. She has a Master of Public Health from ColumbiaUniversity Mailman School of Public Health, and holds credentials as a CPC, CPMA,and CRC.

Who’sInTheRoom?Showofhandsplease…Organization Type:

• Plans, Providers, Vendors

Division/Area:

• Finance

• Coders

• Quality – HEDIS or otherwise

• Risk Adjustment

• Actuary

• Health Services

• Network Operations

• Compliance

• Other

Lines of Business:

• RA for HIX?

• RA for Medicare Advantage?

• RA for Medicaid?

RiskAdjustment:WhyDoWeDoIt?

WorkshopAgenda:December1,2015

Module 1 CMS risk adjustment models, history of payment model, CMS Bid

Module 2 – Risk Adjustment Organizational Structures and RADV

Lunch –

Module 3 – Health plan risk adjustment programs, build / buy services

Module 4 – HHS Risk Adjustment

Cocktail reception

RiskAdjustment&MedicareAdvantageAnHistoricalPerspective

TEFRA*RISKContracts:theBeginningofaLongRoadforPrivateMedicarePlans

• Healthplans paid at 95% of Average Adjusted Per Capita Cost (AAPC)

• Adjusted by Geography, Age, Sex & Original Reason for Entitlement

• Theoretically, Medicare sliced off 5% of costs automatically

• But no adjustment for healthcare conditions / burden of illness

• So, if healthplan enrolled members with less than average healthcare risk, CMS was

overpaying at 95% of average of all beneficiaries.

*Tax Equity and Fiscal Responsibility Act of 1982

PublicPolicyTowardsPrivateMedicarePlans• Original Medicare left with higher risk populations, having to pay 100% of their

healthcare costs

• These factors led to changes in federal legislation, resulting in shifting models of

payment to private Medicare plans

• Changes in plan payments yielded different levels of funding for benefits and

affected plan enrollments as a result

Medicare RiskProgramTEFRA

1976 - 1997

Medicare+Choice(M+C)

Balance Budget Act(BBA)

1997 - 2004

Medicare Advantage

MedicareModernization Act

(MMA)Passed 08/ 2003

Effective Jan 2005

Medicare AdvantageAffordable Care Act

(PPACA)Passed 03/2010

AAPCC= AverageAdjusted Per Capita

Cost95% of Medicare FFS

Costs

Frozen PaymentLevels - 2% per year

Risk AdjustedPayments

at 114% of MedicareFFS (weighted

average by county)

Risk AdjustedPayments Dropping to100% Medicare FFSweighted average bycounty 2012 to 2018

Original Medicare got bad risk?

Federal GovernmentMistrusting Health Plans.

Created New Product Types:PFFS, LPPOs, RPPOs, PACE, MSA, Demos

1999 PenetrationMedicare Population =

18%6.9 Million Enrolled







Created Private Part DPlans.

Revival of MedicareHealth Plan Movement

Recovered ExcessPayment.

Introduced Pay forPerformance

(Medicare Stars)

Private Medicare Plans History Illustrates Evolution

Payment to Plans

Payment year MA plans Evercare* SHMO* PACE and dual demonstrations*

risk /demographic risk /demographic risk /demographic risk /demographic 2000-2003 10% risk/90% 0% / 100% 0% / 100% 0% / 100%2004 30% risk/70% 10% risk/90% 2005 50% risk/50% 30% risk/70% 2006 75% risk/25% 50% risk/50% 2007 100% risk/0% 75% risk/25% 2008 and later 100% risk/0%

Risk Adjusted payments were determined by the relative health of individual members through the presence of certain health conditions (diagnoses)

* Demonstrations

Phase‐InofRisk‐AdjustedPayments

6.9 6.86.2 5.6 5.3 5.3 5.6

6.8

8.49.7

10.5 11.111.9

13.114.4

15.716.8

NOTE: Includes MSAs, cost plans, demonstration plans, and Special Needs Plans as well as other Medicare Advantage plans.SOURCE: MPR/Kaiser Family Foundation analysis of CMS Medicare Advantage enrollment files, 2008‐2013, and MPR, “Tracking Medicare Health and Prescription Drug Plans Monthly Report,” 2001‐2007; enrollment numbers from March of the respective year, with the exception of 2006, which is from April.

Total Medicare Private Health Plan Enrollment, 1999‐2015

Inmillions:

% of MedicareBeneficiaries

18% 17% 15% 14% 13% 13% 13% 16% 19% 22% 23% 24% 25% 27% 28%

MACodingAdjustment

Beginning in 2010, CMS instituted a separate adjustment to the Part C risk scores to account for differential coding patterns between MA and FFS.

Adjustment for MA Coding Pattern Differences: CMS will implement an MA coding pattern difference adjustment of 5.41 percent for payment year 2016 (the statutory minimum).

The goal of the MA coding adjustment is to maintain MA risk scores at the level theywould be if MA plans coded similarly to FFS providers (not necessarily a 1.0 average).

CMSPaymentModelOverview

So,WhatIsCMSRiskAdjustment?

As defined by CMS: • Risk adjustment predicts (or explains) the future healthcare expenditures of individuals based on diagnoses and demographics.

Age, Sex and

OREC*

Health Status

Adjusts Future Payment to Plan

*OREC = Original Reason for Entitlement Code ( old age and survivors insurance, disability insurance benefits, end stage renal disease, or combinations)

CMSRiskAdjustmentOverview

What is Risk Adjustment?

• CMS reimbursement methodology to Medicare Advantage Organizations

• A method used to adjust contract bidding and payment based on the health status and demographic characteristics of an enrollee

How is it determined?

• CMS calculates a risk factor for a member based on

• Demographics (age/sex)

• chronic conditions (diagnoses)

Why is it used?

• Reduces CMS financial exposure by paying based on the risk of healthcare required for the conditions of the enrollees

• Offers access, quality, protection for beneficiaries, reduces adverse selection, etc.

• Prospective‐Uses diagnosis as a measure of health status

GeneralMethodologyofCMSRiskAdjustment

Risk adjustment methodology

relies on enrollee diagnoses, as

specified by the ICD‐9‐CM*

guidelines to prospectively

adjust capitation payments for a

given enrollee based on the

health status of the enrollee.

Diagnosis codes submitted by

providers to MA organizations

Are used to determine

beneficiary risk scores,

Which in turn determine the risk‐adjusted

reimbursement.* International Classification of Disease, Ninth Revision Clinical Modification [shifting to ICD‐10 CM in October 2015]

PurposeofCMSRiskAdjustment

Allows healthplans to be paid for the actual risk of the beneficiaries they enroll

Accounts for differences in expected costs due to burden of illness (usually chronic conditions)

Adjusts annual MA contract bidding to CMS and CMS payment based on health status and demographics

Risk scores measure individual beneficiary’s relative risk, and risk scores adjust payments for expected expenditures

Allows CMS to use standardized bids as base payments to MA plans

• Each Medicare Advantage member is assigned a risk score based on their diagnoses and demographic criteria which calculates their costs/payments in a given year

• Conditions must be submitted annually, particularly chronic conditions

• ~ 3,000 ICD‐9 Diagnostic codes are grouped into ~70 Hierarchical

Condition Categories (HCCs)

• Model includes factors for age/sex, special status and HCC scores.

• HCCs are generally additive with hierarchies and disease interactions

Diagnoses‐>HCCs‐>CMSRiskScores

Example:

• A 1.0 risk score represents average annual Medicare costs for an

individual of $7,463, for illustration purposes.

• Higher than 1.0 means the individual is likely to incur costs higher

than $7,463, less than 1.0 means the individual will incur costs less

than $7,463.

• A risk score of 1.5 means expected costs of $11,195. (e.g. 50% more

expensive than the average score. $7,463*1.50)

CMSRiskScoreExample

PartCRiskAdjustmentModelforPaymentYear2016

First step is to calculate members’ scores . CMS has

fully shifted to the new version of the model known as V22 or

the 2014 model.

Each risk score is adjusted with the PY 2016

normalization factor for each payment year.

The risk score is also adjusted with the MA coding

adjustment factor.

Hierarchical Categories

• Families or hierarchical groups/categories are used in risk adjustment

• More severe or complicated illnesses (by ICD code) in the family or hierarchy will trump all others in the

category or family

• Sometimes codes which are trumped by others from a medical management perspective (Part C) may still

carry value from a prescription drug perspective (Part D)

• This leads to a strong need for coders to always code diagnoses to their highest specificity so that all

current diagnoses are accounted for each encounter

• ICD guidelines instruct coders to code for a principal diagnosis, but also all other comorbidities during

each encounter23

Hierarchical Categories in the CMS HCC Model

24

2014 Hierarchical Categories in the HCC Model

Infection BloodCerebrovascular

DiseaseComplications

Neoplasm Substance Abuse Vascular Transplant

Diabetes Psychiatric Lung Openings

Metabolic Spinal Eye Amputation

Liver Neurological Kidney Disease Interactions

Gastrointestinal Arrest Skin Disability Status

Musculoskeletal Heart Injury

CapturingCompleteDiagnosticData?

Ideal

Reality

• Industry billing has yet to fully evolve from a FFS model to a diagnosis model –provider payment based on Procedure Codes, not Diagnostic Codes

• Medical chart documentation often does not meet CMS requirements, which are more rigorous than standard practice

– > 30% of coded conditions are not supported in charts– > 40% of existing conditions are not reported through claims or

encounter data reporting

• Requirements for ICD‐9 codes to count for risk adjustment:• Face‐to‐face encounter • Only certain sources are allowable, e.g. radiology and labs are excluded• Documentation in the medical record must meet coding standards

RATheoryMeetsReality

• Physicians relying on FFS payment, bill services based on CPT procedure codes. ICD codes are minimally used only to agree with the procedures, not to be comprehensive or to highest level of accuracy

• However, CMS assigns plan payments for patients based on risk (not as reimbursement of services)

• Higher Specificity ICD-codes better define financial risk of a managed population

• 402.10 rather than 401.1

• 250.40 rather than 250.00

• ICD-9 Specificity allocates expected expenditures for the managed population

• However, CMS intends to pay only for those health conditions that are being managed, not ones that are present but not managed

CodingforRiskvs.FFSCodingforReimbursement

Documentation&CodingAffectReimbursement

CMSRiskAdjustmentModelVariations

InstitutionalHCCVersions

ESRDNew

Enrollees

RxHCCDisabled

TwoMedicareAdvantageHCCModels

Part C

CMS‐HCC

Demographic Risk Model

Part D

CMS‐RxHCC

Demographic Risk Model

TheCMSHCCModelsareEver‐Changing• The original DCG/HCC model in 2000 identified 804 costly diagnosis groups, mapped to 189 HCC codes

• Created a reporting model for reimbursement based on ICD codes within families of conditions. (Hierarchal Categories)• There are 2,944 ICD codes carrying Part C (i.e., Parts A and B) HCC value (over 3,000 in 2004)• There are 1,475 ICD codes carrying Part D (Rx) HCC value (over 3,000 in 2004)• Overlap: 978 ICD codes carry both Part C and Part D HCC value (~ 1500 in 2004)

• In 2014, the CMS‐HHC (Part C) model was heavily revised to create V22• Updated the data years used to recalibrate the model• Clinical revision of the diagnoses included in the model

• The CKD (Chronic Kidney Disease) story

• The industry reaction to this new version was so strong that CMS used a blended model in 2014 (mixing values from 2013 model and 2014 model)

• The CMS‐RxHCC (Part D) was also slightly revised in 2014

VariationswithinCMS‐HCC(PartC)Model

Established Members

Demographic Factors

Male / Female Age

Medicaid & Disabled Status

Originally Disabled

Risk Factor (HCC)

Community or Institutional

‘New Enrollees’

Demographic Factors

Male / Female Age

Medicaid Status

Disabled or Not

Risk Factor (HCC)

New enrollees are not included in the HCC model

as there is no history.

AMember’sRAF(RiskAdjustmentFactor)istheSumoftheirDemographicFactor+theCoefficientofAnyHCCs

Established Member

• RAF= Demographic Factor + HCCa+HCCb+ HCCc +…..

New Enrollee

• RAF = Demographic Factor only

NewEnrolleeExampleBen Beneficiary turned 65 in October of 2014 and became entitledto Medicare. Ben opted to enroll in a Part C plan.

He will be a “New Enrollee” until the year? risk score modelrun.

(see table on next slide)

Model Run Dates of Service Ben’s Status Reason

2015 Initial 7/1/14 – 6/30/14 New Enrollee Does not have 12 months of Medicare Part B entitlement

2015 Mid‐year 1/1/14 – 12/31/14 New Enrollee Does not have 12 months of Medicare Part B entitlement

2015 Final 1/1/14 – 12/31/14 New Enrollee Does not have 12 months of Medicare Part B entitlement

2016 Initial 7/1/14 – 6/30/15 New Enrollee Does not have 12 months of Medicare Part B entitlement

2016 Mid‐year 1/1/15 – 12/31/15 Full Risk Has 12 months of Medicare Part B entitlement

New Beneficiary to Full Risk Score Example

TheGoodNews!

All of these Demographic Factors and HCC Coefficients are in easy to use tables

CMS‐RxHCC– PrescriptionDrugRiskAdjustment• Part D (Prescription Drug Benefit) was created as the result of the 2003 MMA (Medicare Modernization Act)

• Basic “HCC” approach used to create the RxHCC Model; similar to Part C

• Model relies on certain conditions (diagnoses) to predict the prescription cost of those conditions.

• Example:

• Part C Model HCC 18 “Diabetes with Chronic Complications” has a value of 0.378

• Part D Model RxHCC 14 “Diabetes with Complications” has a value of 0.276

• If the Plan covers Parts C & D, a member with Diabetes with Complications would get an additional (0.378 * Part C bid rate + 0.276* Part D bid rate)

VariationswithinCMS‐RxHCC(PartD)Model

Established Members

Demographic Factors

Male / Female Age

Community – Low Income

Community – Non Low income

Institutional

Risk Factor (HCC)

Community – Low Income

Community – Non Low income

Institutional

‘New Enrollees’

Demographic Factors

Community –’Income’

Originally Disabled or not

ESRD or not

Institutional

ESRD or Not

Risk Factor (HCC)

New enrollees are not included in the HCC model as there is no

history.

AnnualCMSBid

CMSPlanPaymentMethodology

County Level Base Payment

Relative Factors

Risk Adjustment

Factor

The Medicare Stars Payment

Factor

AnnualCMSCalendar

Published in the Annual Rate Announcement

The calendar provides important operational dates for all organizations such as the date CMS bids are due, the date that organizations must inform CMS of their contract non‐renewal, and dates for beneficiary mailings.

January 1, 2016Plan Benefit Period Begins

January 1 –February 14, 2016Annual 45‐Day Medicare Advantage Disenrollment Period (MADP).

Early January 2016Release of CY 2017 MAO/MA‐PD/PDP/SAE/EGWP applications.

Mid‐January, 2016Industry training on CY 2017 applications.

Late February 2016Applications due for CY 2017.

PartCBidandReviewProcess• By law, the Part C basic plan bid is the total revenue needed to offer original Medicare (Part A & Part B) benefits:

to enrollees who live in a specific service area (one or more counties)

who have a certain level of average risk expected by the MAO

& assuming the plan will charge cost sharing equivalent to FFS

• The law establishes rules for determining plan benchmarks – the upper limit on what the gov’t will pay for each enrollee.

• The law requires CMS to compare the plan basic bid to the plan benchmark to determine whether the plan must charge an enrollee premium or can offer supplemental benefits at a reduced price.

• For MA plans with bids below benchmarks, 75% of the difference (“rebate”) must fund coverage of supplemental benefits, e.g. reduction in FFS‐level cost sharing and/or coverage of additional non‐Medicare covered benefits.

PartCBidandReviewProcess

CMS reviews each bid for actuarial soundness

Ensures that each bid reflects costs of

providing proposed benefit package

Risk adjustment used to standardize bids to determine what CMS’ payment rate will be to the plan for each

enrollee.

Risk Adjustment allows direct

comparison of bids based on populations with different health status and other characteristics

Risk adjustment is also used to pay more

accurately by adjusting the monthly capitated bid‐based payments for enrollee health

status

SimplifiedExampleIllustratingUseofRiskAdjustmentinBidding

Plan derived costs for benefit package = $1,000

Plan estimated risk score for population = 1.25

Standardized plan bid = $800 ($1,000/1.25)

Plan actual risk score based on enrollment =

1.5

Risk adjusted plan payment = standardized plan bid * actual risk

score = $1,200 ($800*1.5)

Courtesy of Jeremy Walsleben, Kaiser Permanente, Colorado

Normalization

Normalization adjusts for growth in risk scores year

after year.

Reasons for this include population trends and

diagnostic coding between model estimation and

payment year.

Normalization Factor

CMS uses a standard of five years of data in the normalization trend.

Each year, CMS drops the earliest year and adds a new year of risk scores to the trend data to create the five‐year dataset.

By using a standard number of years, CMS calculates risk score trends based on recent trends in coding, while maintaining stability in the year to year trends.

Normalization factors are downward adjustments to risk scores and are applied to risk scores when they are calculated.

Risk scores on the MMR are always normalized.

Factors are announced in the Annual Rate Announcement

TypicalHealthPlanRiskAdjustmentProgramCMSRequirements,DataSubmission,Reports,Etc.

MAPlanSponsorRequirements

Ensure the accuracy and integrity of risk adjustment data submitted to CMS.

All diagnosis codes submitted must be documented in the medical record and must be documented as a result of a face to face visit.

The diagnosis must be coded according to ICD‐9‐CM / ICD‐10‐CM Guidelines for Coding and Reporting.

MAPlanSponsorRequirements

Part B only Beneficiaries

•Plan sponsor must submit ICD‐9‐CM Codes / ICD‐10‐CM under the same rules as a beneficiary with both Parts A and B.

• Submit all ICD‐9 –CM codes for Part A services provided under a non‐Medicare contract.

Risk Adjustment Reports

•Receive and reconcile CMS Risk Adjustment Reports in a timely manner.

•Plan sponsors must track their submission and deletion of ICD‐9‐CM Codes / ICD‐10‐CM

• diagnosis codes on an ongoing basis.

Recalculations

•Once CMS calculates the final risk scores for a payment year, plan sponsors may request a recalculation of payment upon discovering the submission of inaccurate ICD‐9‐CM Codes / ICD‐10‐CM

• diagnosis codes that CMS used to calculate a final risk score for a previous payment year and that had an impact on the final payment.

TheCMSMandate

The goal of complete and accurate documentation in progress notes is to help CMS evaluate the costs

of taking care of the patient and pay Medicare Advantage plans accordingly.

Thorough documentation promotes Continuity of Care!

Medicare’s guidelines state,” Code all documented conditions which coexist at the time of the visit that require or affect patient care or treatment”.

CMSGuidanceforMedicareRA

RAPS Participant Guide, section 6.4:

“Standard ICD‐9‐CM coding practices support the HCC model. In all

cases, the documentation must support the code selected and

substantiate that the proper coding guidelines were followed…. Upcoding

or changing diagnoses to obtain higher reimbursement is fraudulent.”

CMSGuidanceforMedicareRA

RAPS Participant Guide, section 6.4:

• “Physicians should code all documented conditions that

co‐exist at the time of the encounter/visit, and require

or affect patient care treatment or management. [sic]

• “Do not code conditions that were previously treated

and no longer exist.”

OneExampleofCMSRiskAdjustmentFunctionalOrganization

Operations Vendor Management Coding Program

Management Provider Engagement Outcomes Management

Analytics & Reporting

Data Infrastructure Quality Controls Reporting

Operations Network Internal

CMS Submissions (RAPS/EDPS) RAPS/EDPS reconciliation

Quality Assurance

PCP Claims Validation Vendor Coding Oversight Coding Education RADV Response Policy & Procedure

Oversight

Objective: Achieve accurate Risk Adjustment Factor (RAF) with complete documentation

Interfaces with Other Functional Departments:

• Quality• Care Mgt.

• Provider Relations• Stars

• IT• Pharmacy• Finance• Actuarial

WhatisRAPS?

• RAPS is the CMS Risk Adjustment Processing System through which risk adjustment data are processed.

After the data submitted by Medicare Advantage (MA) organizations passes the checks in the Front‐End Risk Adjustment System (FERAS), the data is sent to the CMS data center for RAPS processing.

RAPS performs complete editing of all detail records which are then stored in the RAPS database.

As a precautionary measure, RAPS performs balancing checks to ensure that the complete file was received from FERAS prior to editing data.

The RAPS system performs editing on the detail record transactions.

Data elements edited include HIC Number, Provider Type, From Date, Through Date, and Diagnosis Code.

If Date of Birth is submitted, RAPS also performs an edit on that field.

CMSRAPSModelRunPaymentperiods&datesofserviceincludedinmodelrun

• Contracts should recognize the connection between the model runs, the dates of service, and changes in risk score for initial, mid‐year, and final reconciliation.

PY2015 Initial Model Run {Payment 1/1/15 – 6/30/15}

DOS: 7/1/2013 – 6/30/2014

PY 2015 Mid‐year Model Run {Payment 7/1/15 – 12/31/15}

DOS: 1/1/2014 – 12/31/2014

PY2015 Final Model Run {Lump sum payment ~8/2016}

DOS: 1/1/2014 – 12/31/2014

Each year has 3 Model Runs:

In order for data to be included in the model run, MAorganizations must meet three submission deadlines eachyear: the first Friday in September, the first Friday in March,and January 31 after the payment year.

CMSSubmission1

MAO

The physician’s office or hospital codes claim and submits data to

MAO.

2

RAPSFormat

MAO sends diagnosis clusters in RAPS format to

Front‐End Risk Adjustment System

(FERAS) at least quarterly.

3

FERAS

Data goes to FERAS for processing where file‐level data, batch‐level data, and first and last detail records are

checked.

SubmittedMRADataElements

• MA organizations must collect certain data elements from the sources (providers/physicians) of risk adjustment data.

• The five (5) minimum data elements that must be collected and submitted are:

HIC (Health Insurance Claim) number,

Provider Type

From Date of Services,

Through Date of Services, and

Diagnosis Code

MRADataSubmissionTimeline

Payment Year

Model Run

Dates of Service Timeframe

2016 Initial 7/1/2014 – 6/30/2015 September 2015

2016 Mid‐year 1/1/2015 – 12/31/2015 March 2016

2016 Final 1/1/2015 – 12/31/2015 January 2017

RAPSEditsRules• The RAPS editing process takes place in four logical stages.

Stage 1‐ Field Validity and Integrity Edits

Stage 2 ‐ Field‐to‐Field Edits

Stage 3 ‐Eligibility Edits

Stage 4 ‐Diagnosis Code Edits

Risk Adjustment Processing System:

• Mechanism for submitting diagnoses to CMS

• Minimum Risk Adjustment Data Elements

• Diagnosis Code

• Health Insurance Claim (HIC) Number

• From Date

• Through Date

• Provider Type

RAPS Return File:

• Contains all submitted transactions to CMS

• CMS’s acknowledgement of what has been

accepted and what has been submitted in error

RAPSReconciliation

Filtering for RAPS SubmissionsMA organizations are required to filter risk adjustment data submitted to RAPS toensure it only comes from acceptable hospital inpatient, hospital outpatient, andphysician provider types.

Hospital inpatient data require admission and discharge dates of service from appropriatefacilities.

Outpatient data require diagnoses from appropriate facilities and covered services

contained on the CMS covered outpatient listings.

Physician data require visits with a professional listed on the CMS specialty list.

Diagnoses must result from a face‐to‐face encounter with an acceptable provider.

MRAReports• Plans are required to comply with CMS requirements to submit accurate data in a timely manner, which includes submitting diagnoses, meeting the quarterly submission requirement, and not submitting duplicate diagnosis clusters.

• Plans attest when signing EDI Agreements that they will, to the best of their knowledge, information, and belief, submit risk adjustment data that are accurate, complete, and truthful.

• Non‐compliance may result in CMS restricting future risk adjustment submissions by an MA organization, so it is important for plans to understand the types of errors that are identified.

When a Medicare Advantage (MA) organization submits a RAPS file to FERAS, FERAS performs the format and integrity checks.

MAO Deletion Responsibilities

• MA organizations may submit corrections and deletions on the same record or in the same file.

Duplicate deletes in the same record on the same day cause system problems.

Remember:

Only accepted diagnosis clusters may be deleted.Erroneously submitted clusters must be deleted.

Incorrect clusters must be deletedfrom the system before correct

cluster information can be added.

DuplicateDiagnosisClusters

• What if a member goes to the doctor on separate occasions and receives the same diagnosis each time.

• Since the plan submits RAPS records every month, is the diagnosis stored each time a RAPS record is sent/received by CMS, or only the first time the diagnosis cluster was submitted?

Each of the clusters would be unique diagnosis clusters because they have different dates of service. (Duplicate diagnosis clusters are those that have the same HIC, from and through dates of service, diagnosis code, and provider type). Therefore, they will appear on the report in the counts for total stored. The diagnosis would be stored, but later de‐duped when the model was run.

Can diagnosis clusters be duplicated over time?

Plans may contact CSSC at 1‐877‐534‐CSSC for assistance locating the files that triggered the duplicate diagnosis cluster.

RAPS Management Reports

Report Details

RAPS Monthly Plan Activity Report

• Provides monthly summary of the status of submissions by submitter ID and plan number• Report layout• Available for download the second business day of the month, in months with activity

RAPS Cumulative Plan Activity Report

• Provides cumulative summary of the status of submissions by submitter ID and plan number• Report layout• Available for download the second business day of the month, in months with activity

RAPS Monthly Error Frequency Report

• Provides monthly summary of all errors associated with files submitted in test and production• Report layout• Available for download the second business day of the month

RAPS Quarterly Error Frequency Report

• Provides a quarterly summary of all errors on all file submissions within the 3‐month quarter• Report layout• Available for download the second business day of the month following each quarter

AnalyzingRAPSManagementReports

• When analyzing the monthly RAPS management reports, CMS urges MA organizations to consider the following questions:

“Is my organization collecting enough

data from physicians and providers?”

“Is my organization collecting the correct data from physicians and providers?”

“Are external issues affecting data collection?”

“Are internal processes supporting data submissions?”

Payment

• The Risk Adjustment Model is

lagged, meaning 2015 data

(diagnoses) drives 2016

revenue

• The graphic to the right

displays the submission

deadlines for risk adjustment

data to CMS

• Initial and Final Submission

Deadlines are commonly

referred to as “sweep dates”

CMSEDPSOverview

Encounter Data Processing System

• New system CMS is currently implementing.

• Captures more claims and demographic data than

the RAPS system and is designed to improve the

accuracy of risk adjustment payments and member

RAF scores

• Encounter data can be used to develop and calibrate

CMS‐HCC risk adjustment models

• CMS can use the data for calculating Medicare

Disproportionate Share Hospital (DSH) percentages,

Medicare coverage purposes, and quality review and

improvement activities

CMSEDPSvs.RAPS

• Major differences between the EDPS and the current RAPS System. Data collection changes from the 5 RAPS elements to ALL elements of a HIPPA standard 5010 format 837.

Timing of required data submission changes from quarterly intervals to monthly.

Increase in volume of data collected, edited, and stored.

For payment year 2016 (2015 dates of service) RAPS and EDPS will run in parallel with 90% of the payment calculated from RAPS and 10% calculated by EDPS.

This ratio is very likely to focus more on EDPS than on RAPS in the coming years

CMSEncounterData System Process Flow

MAO and Other EntitiesEncounter Data

Front-End System(EDFES)

Encounter DataProcessing System

(EDPS)

EDPPS

EDIPPS

EDDPPS

Subsystems of the EDPS

EODS

Risk Adjustment System (RAS)C M S Medicare Advantage

Prescript ion Drug System(MARx)

OR

OR

MAReports OverviewType Report

FERAS FERAS Response

RAPS Transaction RAPS Return File

RAPS Transaction RAPS Transaction Error Report

RAPS Transaction RAPS Transaction Summary Report

RAPS Transaction RAPS Duplicate Diagnosis Cluster Report

RAPS Management RAPS Monthly Plan Activity Report

RAPS Cumulative Plan Activity Report

RAPS Monthly Error Frequency Report

RAPS Quarterly Error Frequency Report

Analytics&Reporting

• CMS provides many, many reports. A typical, well run, program will actively monitor and use

these reports to:

• Reconcile payments

• Project future revenue (recall that diagnoses excepted in current year drives next year’s

revenue)

• Core data source for analytics and external reporting (provider groups)

• Critical Information to Analyze & Develop Useful Reporting

• What HCCs (typically chronic conditions) did each member have in previous years but have not

been submitted in the current year…Dropped HCCs

• Is there a group or practice that has HCC outliers?

• 60% of Plan’s diabetic population has complications but Group A’s is only 20%

QualityAssurance

• Does the plan have a system that proactively checks that diagnoses codes submitted are

consistent with the medical record?

• CMS requires that for medical records being reviewed, past HCCs must be validated

• Coding education and oversight should be the #1 priority

• How often do the Plans’ coders find errors in the medical records?

• How often are the Plans’ coders wrong? Is there independent monitoring of these coders?

• If a provider or provider group only has uncomplicated diabetics; do their medical records

confirm that? Are they documenting complications but using the ‘easy or familiar’ code? Is

there a process to identify these cases and provide interventions?

• Policies & Procedures should be developed, approved, disseminated, and followed.

RADVCMSRiskAdjustmentDataValidationProgram

Purpose

CMS’ RADV audit initiative is the Agency’s primary strategy to

address the national payment error rate for the MA program, which is was estimated to be 11 percent for

FY 2011.

In addition to recovery of overpayments through RADV

audits, CMS also expects that these contract‐level audits will have a

sentinel effect on the quality of risk adjustment data submitted for payment by MA organizations.

• RADV validates diagnoses submitted for payment.

• RADV is a corrective action to help reduce the Part C error rate. ‐ Each year CMS

reports a National Payment Error Estimate to comply with the Improper Payments

Elimination and Recovery Act (IPERA) of 2010

• CMS expects that RADV will have a sentinel effect on quality of risk

adjustment data submitted for payment going forward.

CMS Risk Adjustment Data Validation (RADV) Overview

MAO‐submitted risk adjustment diagnoses must be:

• Based on clinical medical record documentation from a face‐to‐face encounter

(patient and provider)

• Coded in accordance with the ICD‐9‐CM Guidelines for Coding and Reporting

• Assigned based on dates of service within the data collection period

• Submitted to the MA contracts by acceptable: RA provider type

• RA provider data source

RADVVerifiesDiagnosesSubmittedforPayment

CMSRADVMethodology

“Notice of Final Payment Error Calculation Methodology for Part C Medicare Advantage Risk Adjustment Data Validation Contract‐Level Audits” was published on February 24, 2012

On December 21, 2010 the Centers for Medicare and Medicaid Services (CMS) posted on its website the “Medicare Advantage Risk Adjustment Data Validation (RADV) Notice of Payment Error Calculation Methodology for Part C Organizations Selected for RADV Audit – Request for Comment”.

CMSRADVAudits

• Section 1853(a)(3) of the Social Security Act requires that CMS risk adjust payments to Medicare Advantage (MA) organizations.

RADV audits determine whether the diagnosis codes submitted by MA organizations can be validated

by supporting medical record documentation.

This medical record documentation must meet certain criteria and standards specified in

RADV materials that CMS provides to audited contracts.

Diagnoses that cannot be validated contribute to a payment

error rate.

EstimatingMAPaymentError

National RADV

• Estimates national (Part C) payment error

• Includes both continuously and noncontinuously enrolled beneficiaries in eligible contracts

• Small sample of members (5‐10)




Contract RADV

• Estimates contract level payment error

• Contracts randomly selected from among all active contracts

• Only continuously enrolled beneficiaries

• Includes 201 members





SamplingFrameforCMSRADV• First, CMS identifies all beneficiaries under each MA contract who are “RADV‐eligible” because they meet the following criteria:

Enrolled in an MA contract (H‐number, E‐number, or R‐number) in January of the payment year— based on CMS' monthly member enrollment files;

Continuously enrolled in the same MA contract (as identified in step (1) above) from January of the data collection year through January of the payment year;

Non‐End Stage Renal Disease (non‐ESRD) status from January of the data collection year through January of the payment year;

Non‐hospice status from January of the data collection year through January of the payment year;

Enrolled in Medicare Part B coverage for all 12 months during the data collection year (i.e., defined as full risk enrollees for risk adjusted payment); and

Had at least one risk adjustment diagnosis (ICD‐9‐CM code) submitted during the data collection year that led to at least one CMS‐Hierarchical Condition Category (HCC) assignment for the payment year.

SampleSizeandStrataforCMSRADV

• Next, CMS selects a sample of beneficiaries from each contract’s cohort of RADV‐eligible enrollees. Enrollee‐based stratification will be used in the process of sampling enrollees.

In order to derive the strata, the RADV‐eligible enrollees in each contract will be ranked from lowest to highest based on their community risk score.

Highest

Middle

Lowest

20167

67

67

MAPaymentErrorCalculationEnrollee‐level Payment Error Calculation

CMS will calculate each contract’s payment error based on the validation results.

For each sampled enrollee, the RADV‐corrected risk score and corrected payment will be calculated based on the CMS‐ HCCs that are supported by RADV medical record review findings for the enrollee.

Enrollee‐level payment errors will be defined as the difference between the original payment and the RADV‐corrected payment (per member per month).

The payment error for each enrollee will be either positive(representing a net overpayment), or negative (representing a net underpayment).

An annual payment error amount will be calculated for each sampled enrollee based on the number of months the person was enrolled in the selected MA contract (and was not in ESRD or hospice status) during the payment year.

• Risk Adjustment Data Validation

• Purpose: to ensure risk adjusted payment integrity and accuracy

• Method: Review of hospital (inpatient & outpatient) and physician

medical records

• Objectives:

• Verify enrollee CMS‐HCCs by submitting “best medical record”

• Identify risk adjustment discrepancies

• Calculate enrollee‐level payment error

• Estimate national and contract‐level payment errors

• Implement contract‐level payment adjustments

CMSAudits

This checklist list was provided to plans involved in the calendar year (CY) 2009 and CY 2010 national RADV audits. This list may help to determine a record’s suitability for Risk Adjustment Data Validation (RADV). Any items checked “no” may indicate that the record will not support a CMS-HCC.

1. Is the record for the correct enrollee? 2. Is the record from the correct calendar year for the payment year being audited (i.e., for audits of 2011 payments,

validating records should be from calendar year 2010) 3. Is the date of service present for the face to face visit? 4. Is the record legible? 5. Is the record from a valid provider type? (Hospital inpatient, hospital outpatient/ physician) 6. Are there valid credentials and/or is there a valid physician specialty documented on the record? 7. Does the record contain a signature from an acceptable type of physician specialist? 8. If the outpatient/physician record does not contain a valid credential and/or signature, is there a completed CMS-

Generated Attestation for this date of service? 9. Is there a diagnosis on the record? 10. Does the diagnosis support an HCC? 11. Does the diagnosis support the requested HCC?

CMSRiskAdjustmentDataValidation(RADV)MedicalRecordChecklistandGuidance

When submitting a record for RADV, consider the following:

• If the condition warrants an inpatient hospitalization, the HCC may be supported by an inpatient record. Examples of such conditions may include septicemia, cerebral hemorrhage, cardio respiratory failure, and shock. For these conditions, an inpatient record, a stand‐alone inpatient consultation record, or a stand‐alone discharge summary may be appropriate for submission.

• When possible, obtain a record from the specialist treating the condition, e.g. an oncologist for a cancer diagnosis. These records may be more likely to sufficiently document the condition.

• Pay special attention to cancer diagnoses. A notation indicating “history of cancer,” without an indication of current cancer treatment, may not be sufficient documentation for validation. For example, if in an attempt to validate HCC 10 (Breast, Prostate, Colorectal and Other Cancers and Tumors) a Medicare Advantage contract submits a record that indicates a patient has a history of cancer that was last treated outside the data collection year, the HCC may be not be validated.

CMSRiskAdjustmentDataValidation(RADV)MedicalRecordChecklistandGuidance(cont’d)

When submitting a record for RADV, consider the following: • When reviewing medical records, pay special attention to the problem list on electronic medical records.

Often, in certain systems, a diagnosis never drops off the list, even if the patient is no longer suffering from the condition. Conversely, the problem list may not document the HCC your MA contract submitted for payment.

• Any problem list in submitted documentation should be included and not just referenced. • Records submitted to validate HCCs that encompass additional manifestations or complications related to the

disease (e.g. HCC 15, Diabetes with Renal Manifestations or Diabetes with Peripheral Circulatory Manifestations) should include language from an acceptable physician specialist which establishes a causal link between the disease and the complication. An acceptable record that clearly defines the complication or manifestation and expressly relates it to the disease may validate the HCC. A record that does not define and link this relationship may not validate the HCC.

• If a physician or outpatient record is missing a provider’s signature and/or credentials, consider using the CMS‐Generated Attestation that was provided with your data. CMS will only consider CMS‐Generated Attestations for RADV. Minimum requirements for inpatient records state that these must contain an admission and discharge date. In addition:• inpatient records must include the signed discharge summary,• stand‐alone consultations must contain the consultation date, and • stand‐alone discharge summaries submitted as physician provider type must contain the discharge date

CMSRiskAdjustmentDataValidation(RADV)MedicalRecordChecklistandGuidance(cont’d)

FinalMAPaymentErrorCalculationMethodology

• Methodology will be applied to next round of contract level audits conducted on payment year 2011

• Extrapolation will begin with payment year 2011 • Approximately 30 contracts will be selected for audit • Contracts will be able to submit multiple medical

records per CMS‐HCC • Fee‐for‐Service Adjuster will be applied to payment

recovery amounts

CMSRiskAdjustmentDataValidation

National RADV







Contract RADV









• Effective with the CY 2011 RADV audits, CMS will allow audited MA contracts to

submit up to five medical records for each audited CMS‐HCC per enrollee.

• A CDAT management function will allow MA contracts to add, re‐order, and delete

medical records for each enrollee up until the submission deadline.

• CMS will consider individual hardship requests if an MA contract identifies the need

to submit more than five medical records per sampled CMS ‐ HCC.

CMSRADVProcess–MedicalRecordReview

ProgramManagement:OptimizingResults

TheCMSAnnual“MiracleCure”Member on December 31st Same Member on January 1st

CMS requires all HCC diagnoses to submitted each and every year the condition is present.It is of critical importance that plans ensure that members with HCC diagnoses be seen by a qualified provider and all current HCC diagnoses be evaluated and reported each year.

• Provider education and engagement

• CMS also allows plans to collect data from “alternate sources” which can

be looked at as being either Prospective or Retrospective

WhatCanbeDonetoEnsurethatAllHCCDiagnosesareCapturedEachYear?

• Initiatives taken prior to an encounter to ensure valid diagnosis codes are documented at the time of the encounter

• These include Patient Assessment Forms (PAFS) and Comprehensive Health Assessments (CHAs)

•Pros: Aids in capturing complete and accurate diagnoses at the time of the encounter

•Cons: Slow adoption

• Initiatives taken prior to an encounter to ensure valid diagnosis codes are documented at the time of the encounter

• These include Patient Assessment Forms (PAFS) and Comprehensive Health Assessments (CHAs)

•Pros: Aids in capturing complete and accurate diagnoses at the time of the encounter

•Cons: Slow adoption

Prospective

• Actions taken after the encounter has already taken place in order to ensure complete data collection

• These include Chart Reviews

• Pros: Charts relatively easy to access

• Cons: Incomplete diagnostic profile

• Actions taken after the encounter has already taken place in order to ensure complete data collection

• These include Chart Reviews

• Pros: Charts relatively easy to access

• Cons: Incomplete diagnostic profile

Retrospective

Allows plan to get a full picture of “New Enrollees” for following year payment

Provides early identification for case and disease management programs

Fairly costly to vend – Some plans will pay providers directly for a comprehensive review early in the year

CMS has tried to eliminate these over the past few years. They feel that it is used to gather HCC diagnoses only; no real health value. They are allowing it to continue provided that the results are sent to the member’s PCP.

ProspectiveAssessments

Targeting / Suspect Generation

Which members and providers to review?

Record Retrieval

How to get tens of thousands of records from providers?

Provider Engagement

Why should overworked providers help?

RetrospectiveReviews:TheCoreofRiskAdjustmentProgramManagement

RetrospectiveReviews:Targeting/SuspectGeneration

Perfect World Real World

In a perfect world, providers would correctly document,code, and submit all conditions all the time.

In the real world, documentation is flawed, incorrectdiagnosis codes are used, and many chronic conditions do not get submitted each year

RetrospectiveReviewTargeting/SuspectGeneration

Dropped HCCs – Member has been diabetic for years but no current diagnoses found

Co‐Morbidity – Member is Morbidly Obese; strong likelihood of Diabetes

Pharmacy – Member takes a drug typically used for vascular disease but no diagnosis is found

High # of HCCs – Members with many reported HCCs typically have even more that go unreported

Very Low # of HCCs – This is a population that typically have some HCCs, especially as they age. Others – Deviations of actual cost vs predicted costs; use of other risk adjustment models; outliers in frequency comparisons;

Stratification1. The ‘value’ of HCCs vary widely. Some have a weight of 2.546 and others have a

weight of 0.046. This difference represents ~$22,000 per year.

2. Aside from Dropped HCCs, where there is prior evidence, targeting new member‐conditions have varying probabilities.

• A member taking a drug that is only associated with diabetes has a high probability of having that condition.

• A member taking a drug that is used for many different conditions, including diabetes, has a lower probability of having diabetes,

3. Using information from 1 & 2 can be very helpful in selecting targets for retrospective reviews. Note that the probabilities in 2 are mostly anecdotal so while useful, it should be one of many factors in target selection.

RetrospectiveReviewRecordRetrieval

Targets associated with 2‐3 ‘best’ providers & DOS

How many targets are ‘worth’ pursuing? Declining ROI…what is the optimal target list? Provider PITA factor?

Logistically complex – poor data on provider location; should provider fax the records or will someone physically collect them; scheduling and timing; who will code all these charts; etc.

Results are a gold mine of data and information if managed well. PDFs should be easily accessible, data (typically excel) should be analyzed and reports generated

Another process CMS is not fond of. Same reason as prospective reviews…just getting HCCs. CMS now requires that retrospective reviews be done to validate submitted HCCs; if unreported HCCs are found then they can be submitted. Plans must have a process in place to meet this requirement.

RetrospectiveReviewProviderEnragementEngagement

Providers are increasingly subjected to more and more record requests:

• HEDIS

• STARS Program

• Medicare Advantage Retrospective Reviews

• HHS Risk Adjustment Retrospective reviews

• Others

There are still many small provider groups who are not using an EMR; or using one in a meaningful way.

• The staffing at many of these groups is very tight; asking them to take time to collect specific records, make copies, and fax or mail them to the plan may not be met with a positive response

Even some larger groups with integrated EMRs may rely on HIPAA to make access to the EMR difficult.

Significant planning and communication should be done to optimize providers’ response

RetrospectiveReviewProviderEngagement

• Most MA risk contracts are percent of premium deals• Group gets a majority of the CMS revenue and must cover whatever medical costs are incurred.

• They are self motivated to ensure that all HCCs are accounted for.

• That motivation does not always trickle down to the individual provider.

• The risk entities infrastructure can take on many of the critical tasks in the retrieval

• The “What’s in it for me?” Barrier• Many ways to address this; plans may try a variety of approaches to overcome this.

• Clinical – Missing conditions is not good medicine

• Competitive – Answers the whose patients are sicker question

• Fear – Almost all provider payments will be based on risk adjustment soon.

• Financial – Reimbursement for time and material

• Contractual – Providers typically have a duty to provide these records

Provider Group in a Risk Contract Provider Group with no Risk Contract

ProviderTrainingandReporting

Education on Risk Adjustment

Develop group, provider, & member level reporting on RAF

scores

Send un‐blinded, comparative reports to

provider groups

Manage the inevitable push back from the providers

This is actually a positive. Once providers accept the analytics,

the reports become a competition

Address individual questions/accusations such as “I know I have at least 15 Morbidly Obese patients” by showing the

absence of the diagnosis codes for those patients

ProviderEducationandReporting• Providers have focused on CPT codes since Hippocrates. Diagnoses codes were a distant second.

• CPT codes equaled payment

• In a perfect world, see prior slide, risk adjustment departments could be staffed by 1 FTE or less in the IT department.

• Moving toward that perfect world is an ongoing challenge but must be done. It won’t happen quickly but improvements can be made.

• Reporting that is specific to a provider, provider group, and system is most effective.

• Two insights on providers:

• They are extremely skeptical about any report from a health plan

• However, once satisfied with the validity; they tend to embrace these reports…especially those that show comparative performance

ProgramOperations• Almost all of the functions within a risk adjustment department revolve around

getting diagnoses codes that, for some reason, have not been submitted to the plan.

• A large part of this is also ensuring that a submitted diagnosis code can be validated

by the medical record. The reverse is true as well – does the documentation and

diagnosis code show the full complexity of a condition?

• If a diagnosis code is submitted that links to HCC 108 “Vascular Disease”; does

the medical record documentation fully support that? Or is it a RO or a

suspected condition?

• Does the documentation appear to support two linked conditions but is worded

in a way that the coder cannot assign the proper code?

• The next slides will show what actions are taken at specific times of the year to

optimize the correct HCCs

Operations– November&December2015• Finalizing 2014 retrospective reviews

• January 2016 is the very last time these can be submitted. Payment will be in a lump sum in

July/August 2016

• Members with dropped HCCs in 2015 must see a provider by 12/31/15

• Monthly revenue for first six months of 2016 are known, caeteris paribus.

• Based on results of the September 2015 Model run (7/1/14 – 6/30/15 DOS)

• CMS report on results of open enrollment period

• Established members staying with the plan

• Established members coming from another plan

• New enrollees coming from another plan

Operations– January&February2016

• Last push for CY2014 submission

• Selection process for New Enrollees to the plan for Prospective Assessments

• Begin Prospective Assessments

• Implement RAF (Risk Adjustment Factor) build up process

• Dropped HCCs are the biggest source of missed revenue. This process looks at all the HCCs

for each member in 2015 with the expectation that, if chronic, should be seen in 2016.

• In the early part of the year, the build up will naturally be low as members may not have

had a chance to see their provider yet

• Recall that everyone is completely health on January 1st according to CMS

Operations–March&April2016

• Complete March submission – DOS 1/1/15 – 12/31/15

• Drives monthly payment from 7/1/16 – 12/31/16

• Finalize any new interventions for 2016

• Continue Prospective Assessments

• Start analyses to develop Retrospective Review target list

• Continue tracking RAF build up

Operations–May,June,&July2016• Begin Retrospective Reviews

• Develop materials to send to targeted providers

• Logistics should be in place

• Continue RAF build up and begin intervention

• By June and July, the plan should know which members have not had a prior year HCC submitted yet.

• Contact providers who submitted members’ HCCs in 2015 as well as members’ PCPs to encourage them to get the member in to document the HCCs that are still active.

• Identify members without an E&M visit in 2016 and get them seen

• Closely monitor Retrospective Reviews until January 2017 submission!

Operations– August&September

• Continue RAF build up monitoring and interventions

• * Reconcile lump sum payment (* may be done in different area)

• Adjustments based on final 2014 submission

• Adjustments to January – June monthly payments

• These payments are based on the September submission of 7/1/14 – 6/30/15 DOS

• Adjustments to member enrollment

• Any other final adjustment to 2014

• Complete September submission – DOS 7/1/15 – 6/30/16

• Drives monthly payment for 1/1/17 – 6/30/17

• Closely monitor Retrospective Reviews until January 2017 submission!

CommonMisconceptions

Trusting clinicians (or software) to code correctly

Ignoring ICD‐10‐CM as a resource for ICD‐9‐CM coding

Expecting non‐RA coders to code accurately for RA

Delaying the implementation of a CDI program

Underestimating the importance of the Official Guidelines to Coding and Reporting

Overlooking the need to audit coders regularly

Coding diseases and conditions without support

Keeping references and resources current

Assuming Coding Clinic guidance is only for inpatient coders

Believing your education is complete

TipsforClosingtheGaps1

See Each Patient at Least Once Each YearThe health status of a Medicare Advantage patient needs to be re‐determined each year. Diagnoses from a prior year do not “carry over” for CMS.

2Evaluate and Document All Chronic ConditionsAll conditions that constitute the “composite health picture” of the senior patient should be evaluated and documented clearly and legibly in the progress note of the medical record.This is not limited to what brought the patient to the doctor today. What other conditions is thepatient dealingwithevery day?

3Code All DiagnosesThe codermust be careful to capture all diagnoses thathavebeenproperly documented. Does the coderhaveaccess to the latest ICD‐9‐CMcodes? Does the coder code to the highest levelof specific toaccurately report the levelofdiseaseseverity?

4

Use an Accurate, Up-to-date Superbill (or Favorites List in EMR)If a superbill is used, does it contain a wide variety of ICD‐9‐CM codes to allow the specificity of the disease to be coded accurately? Is it up to date? Are providers trained towrite in additional diagnoses if they apply or do they use the closest match on the superbill instead? Is the superbill evaluated each year to ensure it meets the needs of thepractice?

5Make Sure the Data is CapturedThe provider must be aware of the limitations of their practice management system. Howmany diagnosis codes does the system allow? Is there potential for any codes to be dropped?Is the provider correctly sequencing the diagnoses?

6

The Claim or Encounter Format or Form Must Contain All the DataWhen the data is extracted for claims or encounter reporting, are all diagnosis codes extracted to be sent to the health plan? Does the claim process limit the number of diagnosesthat can be submitted? Is the practice in the habit of only sending one or two diagnosis codes to support the CPT code on the claim?

7

Verify that Clearinghouse or Submission Vendor Can Send and Receive All Recorded CodesHow many codes can the vendor support for data submission? Are valid codes being dropped because the provider has not updated the number of codes that can be submitted?Many claims systems and practice management systems are being enhanced to capturemore data due to HIPAA data requirements. Has the vendor’s submission been expandedto accept additionaldataaswell?

8Verify that Health Plans Can Send and Receive All Recorded CodesNot all health plans have expanded their systems to accept large numbers of diagnosis codes. Howmany codes can your payer accept? What happens to any codes submitted beyondthe accepted number? Is there an alternative submissionmethod (ASM) available?

• HEDIS

• Contractual Quality Measures

• STARS Program

• Very large domain of quality measures for different entities

• Osteoporosis Management

• Rheumatoid Arthritis

• BMI Assessment

• Asthma Management (potential)

• Depression Screenings (potential)

RiskAdjustmentandClinicalDepartments

Buy,Build,orBoth?

Optimal Performance

In‐house department with some

key elements vended

Majority of functions vended out

All functions done in‐house

TypicalVendorFunctions

Prospective & Retrospective programs

Dash boards and reporting

Chart retrieval

Coding

VendorManagement

RFP/I

• When considering a vendor, a well thought out RFP/I must be developed

• Review the replies carefully

Monitor Deliverables

• Have a clear SOW in the contract and ensure that all portions are being met

• Prepare for systems integration issues.

Problems

• Identify any issues with vendor; determine origin of problem. Did the plan contribute to the issue; if so, how. Work collaboratively as partners

• Develop a CAP (Corrective Action Plan) addressing all contributing factors

CodingQuality– FoundationofRiskAdjustment• Monitor coders on a regular basis and track results

• Use external coders to audit your internal coders

• Spend the extra money and do blind coding

• Use third party coders to monitor vendor coding

• Voldemort

• Develop internal coding guidelines to address grey areas

• Select a random set of records and have internal, vendor, and 3rd party code the same records

• Identify all areas where there are differences• Determine the ‘correct answer’; where there is one

• Carefully review differences that fall into the grey areas

• Decide where your plan falls on the Conservative – Aggressive spectrum• Conservative approach sacrifices some dollars now to avoid penalties later

• Aggressive (not fraudulent) approach accepts more of the grey areas for dollars now

• Working with the audit results, identify and develop guidelines for coding grey areas that fall within your spectrum

HHSRiskAdjustment

Building Blocks of HHS Risk AdjustmentHHS Model HCC Model:

Compare & Contrast

HHS Model Premium

Stabilization: The Marketplace

HHS Model Payment

Integrity: RADV

HHS Model

HCC Model:Compare & Contrast

HHS Risk Adjustment: Same Heritage as CMS Risk Adjustment Model

While conditions and weights vary, the main determinants of risk scores are the same: Age, Gender, Diagnosis codes HCCs

Diagnosis‐based and hierarchical

Same sites of services and provider types

Diagnoses addressed yearly in “face‐to‐face” encounters

Key Differences:

121

CMS (Medicare) ACA (Indiv/Small Group)

Prospective Concurrent65+ Population 0‐64+

Single Model (Part C) 3 Models (Infant, Child, Adult)

All plans can receive payment Plans send or receive payment transfers based on relative risk score to market.

RAPS & EDPS (full encounters) EDGE Server (de‐identified data)

Key Characteristics of Risk Adjustment Models

• Most models are additive:

• Each category (groups of diagnosis codes) is

assigned a weight, which when added together

comprise the risk score for a particular member

CMS‐HCCs (Medicare), HHS‐HCCs (small‐

group/individual) and most Medicaid risk

adjustment models (CDPS, DCGs,

MedicaidRx, etc.) are all additive linear

models

(The Johns Hopkins ACG System, used by several

Medicaid programs, is not additive (tree‐based). CRGs,

used by New York State Medicaid is also a non‐additive

model)

Courtesy of Richard Lieberman

HCC 1 wt. + HCC 2 wt. + HCC wt. 3= Risk

Adjustment Factor

Key Characteristics of Risk Adjustment Models (cont’d.)

• Prospective models use categories derived from prior period data

to predict cost/utilization in a future period and there is a lag

between risk assessment period and payment year

• Concurrent models use categories from a period to explain

cost/utilization in the same period

• Risk scores reset every year [Really? Yes]

• Risk scores are always assigned to individual members, however…

123Courtesy of Richard Lieberman

Key Characteristics of Risk Adjustment Models (cont’d.)

Different product line sponsors designed their methodologies very

differently:

• Medicare Advantage: Assigns risk scores to individual members

and pays for each member individually

• Medicaid: Most states pay at the plan level with a multi-year lag;

others for each member individually

• Health Benefit Exchanges: Will pay at the plan level, with no

payment lag


Exchange Risk Scores Are Normalized to the State-Wide Average

• Medicare Advantage: risk scores are purely individual, only

related to the entire Medicare population of 49 million beneficiaries

• Health Benefit Exchanges: plan-level risk scores are compared to

other plans in the same state, within metal level

• Medicaid: risk adjustment tends to work like small-group/individual

market risk adjustment, with standardization to a plan-wide average

or typically at sub-state regions


Risk Score Normalization


Risk Adjustment Model & Methodology(Medicare & Medicaid)

127

Risk Adjustment Administrator (CMS or State

Medicaid)Pharmacy

Data

Medical Claim Data

Eligibility Data


Risk Adjustment Model & Methodology (Exchanges)

128

Risk Adjustment

Administrator (Exchange)

Issuers’ Edge Server

Pharmacy Data

Medical Claim Data

Eligibility Data De-identified Risk Assessment

Scores/Prevalence Data


Data Required for Risk Score Calculations

• ICD-9-CM diagnosis codes

• CPT-4/HCPCS

• Bill types

• Acceptable source of data

• Claims

• Encounters

• Paid claims only? “…diagnoses reported on institutional and medical claims that result in final payment action or encounters that result in final accepted status.”


Diagnosis Overlap Is SignificantStillabout1/3newinHHSmodel;numberingofHHS‐HCCsnotrelatedtoCMS‐HCCs;HCCswithinagrouphavesamecoefficient

HHS vs Old CMS Model HHS vs New CMS Model

130

Three Models – Adult, Child, Infant, each with some unique differences

Makes clinician training and data mining complex:

Antidote is to completely and accurately capture diagnoses

Adult Model (Age 21+)

Child Model (Age 2 – 20)

Infant Model (Age 0‐1)

Applicable model is determined by the age at the end of the benefit year (Exception: Infants born at the end of one benefit year and discharged in the second benefit year are considered age 0 for both years)

131

Adult Model (Age 21+) –someconsistencywithchangestoCMSmodel(Diabetes,CKD),thoughsomerecentlyaddeddiagnosesfromtheMedicaremodeldidn’tmakethecut

What’s new and/or different?

Cancer HCCs expand from 4 in the old CMS model (5 in the new CMS model) to 6 in the HHS model

>650 Pregnancy diagnoses (more than ½ of new diagnosis codes)

Some HCCs age‐dependent

Breast cancer appears in 2 HCCs dependent on the age of the member. (< 50 or 50+)

Pregnancy spans both the adult and child models (ages 12 to 55)

In CMS, not in HHS (examples)

Morbid obesity

Stable Angina

Old MI not in either model

Alcohol dependence

Complications from alcohol dependence are in the model

Cannabis dependence

Other drug dependencies are in the model

Exudative Macular Degeneration

132

In HHS Model Diagnoses Differentphysicianspecialtiesneedtobetrained;alsocongenitaldefectsnottypicallydocumentedandreportedannually

CNS Infections hierarchy

Hydrocephalus

Heart Infection/Inflammation

Acute Pancreatitis and other Pancreatic

Disorders (including celiac disease)

Autism and Pervasive Developmental

Disorders

Transplant status codes have been spread

across the model in multiple HCCs

Anorexia, Bulimia

Personality Disorders (e.g.,

Obsessive‐compulsive disorder,

antisocial personality)

Cleft Lip and Palate

Congenital Disorders (including

heart for infant and child models)

Chromosomal Abnormalities (e.g.

Down’s syndrome)

133

Not All Interactions Are Created Equal: SomeHCCshavesignificantimpactonriskscores.Perhapsobvious:Highseverityillnessesrequireappropriatetreatment,care,etc.

Interaction Types

High Cost Interactions

Medium Cost Interactions

Only one interaction may be assigned to a

beneficiary

If both high and medium interactions are

present, only the high cost interaction will

be applied.

Variables used in determining interaction level

and coefficient

Severe Illness Indicator

Interaction Factor (designated HCCs)

Adults with a severe illness may have high cost

interactions

E.g. Adult with seizure disorder and convulsions

(HCC 120 – is a “severe illness”) and

opportunistic infection (is a high‐cost

interaction), so coefficient would be = 12.427

(silver tier)134

“I’m trying to think, don’t confuse me with facts” – PlatoInfantmodelhasinteractions,butchildmodeldoesnot

Child (Ages 2‐20)

Largely same diagnoses as adult model (82 diagnosis

codes are in child but not in adult model)

Coefficients are different

Age parameters for some HCCs

Pregnancy diagnoses (age 12+)

Other diagnoses e.g. alcoholic cirrhosis of the liver

(age 15+)

Others end at age 17 (e.g. SIDS, Reye’s syndrome)

Infant (Ages 0‐1)

Similar to adult interactions

Only most severe diagnoses count (in conjunction with

maturity/age factor)

Diagnoses/HCCs collapse into severity levels that

include virtually all of the adult and child HCCs

E.g. a full‐term infant with an intestinal transplant,

coefficient = 131.511, extremely immature infant with

same condition = 391.399 (both silver tier)

135

Interactions = No Interactions = Yes

Combining Maturity Level and Severity Level to Calculate HCC Coefficients

Group Infant Silver Metal

Term* Severity Level 5 130.511

Term * Severity Level 4 18.560

Term * Severity Levels 3 5.765



Age 1 * Severity Level 5 61.217





Age 0 Male Add‐on 0.574

Age 1 Male Add‐on 0.094

136

HHS Model

HCC Payment Model

Basic Form of the Payment Transfer Calculation


Risk Adjustment at the Plan Level

Plan 1Average risk score

= 0.9Exchange

Plan 2Average risk score = 1.1


HHS Risk Adjustment Payment Transfers Are Simple……onlyifyou’reanactuary,oryoulikemath!

140

• Reported risk score in state market

Plan Level Risk Score

• Weighted average of all health plans in state market

Market Risk Score

•Factors that are function of benefit design / membership

•Age, metal tier

Plan Allowed Rating

•Weighted average of all health plan factors in state market

Market Allowed Rating

•Weighted average of market premiums

State AvgPremium

HHS Risk Adjustment Is a Zero-Sum Game:

141

Health Plan #1 Health Plan #2 Average/Total

Member Months 600,000 360,000 960,000

Average Members 50,000 30,000 80,000

Premium PMPM 310.00$ 300.00$ 306.25$

Risk Score 1.000 0.950 0.981

PMPM Adjustment 5.85$ (9.75)$

Premium Adjustment 3,511,146$ (3,511,146)$ ‐

Net Premium PMPM

(after Adjustment)315.85$ 290.25$

‐ ) * $306.25 = $5.85

Reimbursements to health plans for higher risk occur by transferring funds between health plans within a state (or rating area)

Risk Score Payment Formula Is More Complex

Remember, data on Edge server is de‐identified

142

Reflected in Risk ScoreNot Reflected in Risk Score

Acute Diagnosis Other Diagnosis

Hypothetical Example

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Member Status

HCC Weight 4.215 1.120

Not a member Member in risk adjusted Small Group plan

Risk score does not reflect HCCs captured when patient is not member (unlike CMS model)

Summary of Risk Adjustment Process Timeline

2013

143

March 2013Final HHS Payment

Notice

Throughout 2013Technical Requirements Released/ Operations

Implementation Specifications for Data Storage Issued

2014

January 2014-April 30, 2015

Issuers populate Edge server with

data

2015

April 30, 2015 Reinsurance and Risk Adjustment data collection

deadline

June 30, 2015Payments and

charges for implementation for Benefit year 2014


Same Pattern Repeated in 2016

HealthCare Reform Premium Stabilization Programs: The Three R’s

144

Risk Adjustment is the only permanent feature

Forces insurers to compete based on quality, service, and price ‐ not their ability to draw good risk

Goes hand‐in‐hand with regulations that prevent insurers from denying care to those with pre‐existing conditions

Results in insurers being more equitably compensated for the health burden of the people they insure

Summary of 3 R’s By Market

• 145

Sold withinExchange

Sold Outside Exchange Who Administers

ACA Provision

IndividualSmallGroup

IndividualSmallGroup

Grand-Fathered

State Run Exchange

Federal Run Exchange

Risk Adjustment

Yes Yes Yes Yes NoState or

HHS1HHS

Reinsurance Yes No Yes No No StateState or

HHS1

Risk Corridor Yes Yes No No No HHS HHS

1State can decide to administer or allow HHS to administer. If HHS administers, all parameters will be federal.


Reinsurance Parameters

• Reinsurance available for the first three contract years

• $60,000 attachment point

• 20 percent co-insurance from $60,001 - $250,000 per member

per year

• National reinsurance cap of $250,000

• Issuers can purchase private reinsurance for liability above $250,000

• CMS will charge issuers $5.25 pmpm for reinsurance


Differences in Timeline Impact to Chart Review and Data Mining Operations

Payment transfers will occur six months after the benefit year

Risk scores based on diagnoses, de‐identified in health plan possession by April

30, 2015 for 2014 Plan year (2014 revenues)

Payment transfers based on differences in risk scores will be invoiced by HHS by

June 30, 2015

Other key difference is that percentages of members with HHS‐HCCs (based on

FFS data) is lower

Adult 19%

Child 9%

Infant 45% 147

Population is different and members have fewer chronic conditions (only 20% of adults based on FFS data) – maybe all hay, no needles!

HHS Model

Data Integrity and RADV Model

Rulemaking, Data “Submission” and RADV:

Markedly different–will keep you on your toes

149

Rulemaking Timelines Differ Between CMS and HHS Models

Benefits & Parameters (proposed rule) published in December (vs. mid‐February for CMS)

Comment period under 30 days (vs 14 days for CMS)

Final notice comes out in March (vs April for CMS)

Data Stored on Edge Server or Amazon Cloud

Unlike MA risk adjustment, data will reside within a health plan’s database (Edge server or Amazon cloud) and will not be “submitted” to HHS

Risk Adjustment Data Validation

All issuers (health plans) required to perform annual Risk Adjustment Data Validation Audits on their data under HHS oversight

RADV results impact future year “premiums”

HHS RADV Will Be Annual 100% for All Plans

CMS

Sample Stratification

Hi, med, low, based on the disease component of the member’s risk score

Auditor

Both initial and second validation conducted by CMS contract auditors

Extrapolation

Yes – Plan year being audited

HHS

Sample Stratification

Hi, med, low per model (adult, child, infant), based on age group, risk level, and presence of HCC

1 stratum for population with no HCC

Initial validation by auditor retained by plan, second level audit by HHS

Yes – Future year

150

Will it be more complex? Extrapolation? YES

HHS RADV: Informational for first two years, but will have “teeth!” – Focus on accuracy as well as completeness

CMS

• Acceptable Visits

• Face‐to‐face encounters

• Eligible “physician” provider

• Risk Score Error Types

• Unsupported HCCs

HHS

• Face‐to‐face and tele‐health (to be defined)

encounters

• Eligible “physician” provider

• Unsupported HCCs

• New HCCs (zero tranche)

• Demographic data errors

151

Maintaining Payment Integrity

• Comprehensive audits are necessary because of the relative nature of the risk scores across all issuers in a state

• Issuers will hire their own, Initial Validation Auditors (IVAs) to conduct the RADV audit

• CMS will audit the IVA auditors using Secondary Validation Auditors (SVAs)

• Financial impacts of audits will not be applied retroactively


Risk Adjustment Audits in the Small-Group Market

• Audit program to look like Medicare Advantage (RADV)

• BUT, every issuer get audited every year!

• RADV audits are state-specific

• Approximately 300 members will be audited each year per

issuer

• Audit results are extrapolated to all members and applied to a

future year’s revenue

• No financial sanctions for 2014 and 2015 contract years


Timeline for Implementation for 2014 Plan Year

2015

March 2015 Issuers Provide

Auditor Information to HHS

April – June 2015Selection of Audit Sample, Issuer/Auditor Training, & Distribution of Sample to

Issuers

July – November 2015

Initial Data Validation of Auditor Sample

2016

December 2015-March 2016

HHS Oversight of Data Validation Audit Sample

April – May 2016Announcement of HHS Findings and

Processing of Appeals

June 2016 Estimate Risk Scores and

Stimulate Payment Adjustment


Quality Improvement and Reporting:

FEDERAL GUIDANCE AND RULES: FEDERALLY- FACILITATED EXCHANGE

Certification Year

QHP issuers without existing accreditation

QHP Issuers with existing Commercial, Medicaid accreditation for the state

Year 1 (2013) Schedule accreditation reviewAttest that accredited policies and procedures are comparable to Exchange (not all the same)

Years 2 and 3 (2014-2015)

Be accredited for Exchange product(policies and procedures)

Attest that accredited policies and procedures comparable to Exchange

Year 4 (2016) Exchange product is accredited data, performance data submitted


Some Key Take-Aways

1. Risk score and payment calculation (more complex) is zero sum game – dependent

on metal mix of membership, market risk – there will be “winners” and “losers”

2. Members have fewer chronic conditions (only 20% of adults based on FFS data) ‐

need to be data ninjas. Also, clinical onboarding (required for Covered California) to

identify conditions early becomes increasingly important

3. Differences in timeline – concurrent model and 4 months after data year for HHS,

13‐25 months for CMS – Implications to chart review and data mining operations –

need to bring members in sooner, fully document and code conditions at first visit

156

Some Key Take-Aways

4. Diagnoses overlap, some diagnoses in HHS, not in CMS Different physician

specialties need to be trained (Pediatricians, Ob‐Gyns, etc.), place additional focus on

high impact HCCs (transplants, ESRD, hemophilia).

5. RADV for every plan, every year (informational in first two years, then it really

matters). This RADV has teeth. Initial validation audit, secondary validation audit.

Potential for other errors. Focus on accuracy as well as completeness.

157

WRAPUP

NOTES__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

NOTES

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

NOTES__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

__________________________________________________

December 1, 2015 CMS & HHS Risk Adjustment 101 · • So, if healthplan enrolled members with less...

Documents

Transcript of December 1, 2015 CMS & HHS Risk Adjustment 101 · • So, if healthplan enrolled members with less...