Risk
Adjustment
in
the
Medicare
ACO
Shared
Savings
Program
Presented by:
John Kautter
Presented at:
AcademyHealth Conference Baltimore, MD June 23-25, 2013
Introduction
Role of risk adjustment in the ACO Shared Savings
Program
CMS-HCC risk scores
Risk Adjustment in Historical Benchmark Period
Risk Adjustment: Overview
Risk adjustment is a method for adjusting expenditures
to account for differences in expected health costs of
individuals
Adjustment can take into account demographic
information (age, sex, eligibility) and health status
(diagnoses)
Why adjust?
To account for changes in severity and case mix over time and
to more accurately set ACO performance targets.
Recognize ACOs that care for complex patients, and not
CMS-HCC Risk Scores
CMS-HCC (Hierarchical Condition Categories) risk scores
The CMS-HCC model uses beneficiary demographic
characteristics and prior year diagnoses to predict relative Part A and Part B Medicare fee-for-service program payments
The CMS-HCC model does not incorporate Medicare Part D
costs
The CMS-HCC model is prospective, meaning it uses prior year
information to predict costs
Same risk adjustment model used for Medicare Advantage
Separate CMS-HCC models for
Aged-disabled community and institutional
ESRD
Demographic Factors Used in CMS-HCC
Risk Scores
24 age-sex cells
E.g., male age 80-84
Medicaid dual eligible status
By sex and aged vs. disabled entitlement
Disabled status
Current disabled:
• Separate age/sex and Medicaid factors
• Selected diagnoses have different risk weights
Currently aged, originally entitled to Medicare by disability
Diagnoses Used in CMS-HCC Risk Scores
The diagnoses used to calculate risk scores for fee-for-service
(FFS) beneficiaries are from FFS claims
Use International Classification of Disease, Version 9, Clinical
Modification diagnosis codes (ICD9-CM)
Diagnoses from the following settings/providers are used
Hospital inpatient
Hospital outpatient
Physician
Clinically-trained non-physician (e.g., clinical psychologist)
The CMS-HCC model counts only the most severe manifestation
Hypothetical Illustrative Example of
Beneficiary’s Individual Risk Score
Calculation
Beneficiary is male, age 77, with the chronic conditions: congestive
heart failure, diabetes with complications, and chronic obstructive pulmonary disease
Risk adjustment model coefficients
Male age 77 = $5,100
CHF = $3,900
Diabetes w/ comp = $3,300
COPD = $3,700
Beneficiary’s predicted expenditures are $16,000
Average expenditures for all beneficiaries are $10,000
Risk Adjustment in Historical Benchmark
Period
The CMS-HCC prospective risk adjustment models will be used to
calculate the ACO’s assigned beneficiary population’s risk scores for the benchmark years, which are used in calculating the historical
benchmark.
Changes in the ACO’s risk score between benchmark years 1 and 3
will be used to trend forward benchmark year 1 expenditures. Similarly, changes in the ACO’s risk score between benchmark years 2 and 3 will be used to trend forward benchmark year 2 expenditures.
Risk Adjustment in Historical Benchmark
Period (cont’d)
BY1 BY2 BY3
[C] Assigned Beneficiary HCC Risk Scores
ESRD 1.148 1.152 1.178
Disabled 1.002 1.004 1.007
Aged/dual 1.282 1.386 1.463
Aged/non‐dual 1.039 1.042 1.062
[D] Risk Ratios to BY3
ESRD 1.026 1.022 1.000
Disabled 1.005 1.002 1.000
Aged/dual 1.141 1.055 1.000
Aged/non‐dual 1.022 1.019 1.000
BY1 expenditures will be adjusted by the BY1 to BY3 risk ratio (BY3 risk score
÷ BY1 risk score). Similarly, BY2 expenditures will be adjusted by the BY2 to
Risk Adjustment in Performance Years
The benchmark will be annually adjusted for changes in health
status and demographic factors during the performance year.
Newly assigned beneficiaries vs. continuously assigned
beneficiaries
1) Newly assigned beneficiaries in a given year = Beneficiaries
assigned to the ACO in that year but not assigned to the ACO in the prior year and not receiving primary care services from the ACO in the prior year.
2) Continuously assigned beneficiaries in a given year =
Beneficiaries assigned to the ACO in that year and assigned to the ACO in the prior year or receiving primary care services from the ACO in the prior year.
How to Determine Final Risk Scores to Use
in Adjusting Benchmark
For newly assigned beneficiaries, an ACO's CMS-HCC prospective
risk scores will be annually updated in each performance year to
adjust the historical benchmark for changes in severity and case mix relative to the newly assigned population from the historical
benchmark period (BY3).
For continuously assigned beneficiaries, patient demographic
factors will be used to adjust the benchmark to account for changes between the benchmark period and the performance year, unless the continuously assigned population shows a decline in its CMS-HCC risk scores, in which case health status changes for this
Example 1 –CMS-HCC Risk Score Used
for Continuously Assigned Beneficiaries
These risk ratios are multiplied by the historical benchmark dollars and added to the National growth increment to produce the ACO’s updated benchmark expenditures
Determine Final Risk Ratio
[R5] Step 1 ‐ Calculate Risk Ratios Demographic Ratio: HCC Ratio: Weights
ESRD 1.017 1.006 6%
Disabled 1.000 1.015 10%
Aged/dual 1.015 1.011 14%
Aged/non‐dual 1.000 0.990 69%
Average (dollar weighted) 0.997
[R6] Step 2 ‐ Final Adjusted Risk Scores HCC Score: Avg Adjusted Score:
ESRD 1.090 1.174
Disabled 1.050 1.044
Aged/dual 1.300 1.490
Aged/non‐dual 0.830 1.058
[R7] Step 3 ‐ Final Risk Ratios Risk Ratio:
ESRD 0.997
Disabled 1.037
Aged/dual 1.018
Aged/non‐dual 0.996
1.080 1.040
Risk Score to use:
1.200 1.540
Dollar weighted average risk ratios
Historical Benchmark $ x Continously Assigned Person Years
Weighted average of PY1 HCC score for newly assigned plus appropriate risk score for continously assigned
PY1 average adjusted score / BY3 average HCC score
Because dollar weighted average HCC risk ratio is less than one, use HCC risk ratio for continuously assigned
Example 2 – Demographic Factors Used
for Continuously Assigned Beneficiaries
These risk ratios are multiplied by the historical benchmark dollars and added to the National growth increment to produce the ACO’s updated benchmark expenditures
Determine Adjusted Benchmark
[R5] Step 1 ‐ Compare Risk Ratios Demographic Ratio: HCC Ratio: Weights
ESRD 1.017 1.065 6%
Disabled 1.000 1.021 10%
Aged/dual 1.015 1.031 14%
Aged/non‐dual 1.000 0.996 69%
Average (dollar weighted) 1.008
Newly Assigned
[R6] Step 2 ‐ Final Adjusted Risk Scores PY1 HCC Score: Avg Adjusted Score:
ESRD 1.090 1.183
Disabled 1.050 1.035
Aged/dual 1.300 1.494
Aged/non‐dual 0.830 1.067
[R7] Step 3 ‐ Benchmark Adjustment Risk Ratio:
ESRD 1.005
Disabled 1.028
Aged/dual 1.022
Aged/non‐dual 1.005
1.546 1.090 1.025 Risk Score to use:
1.212
Continously Assigned
Dollar weighted average risk ratios
Because dollar weighted average HCC risk ratio is greater than one, use demographic ratio for continuously assigned
Weighted average of PY1 HCC score for newly assigned appropriate risk score for continously assigned
PY1 average adjusted score / BY3 average HCC score
Historical Benchmark $ x Continously Assigned Person Years
Incorporated Risk Scores in Adjusted
Benchmark
Risk ratios are multiplied by the historical benchmark dollars and
added to the National growth increment to produce the ACO’s updated benchmark expenditures
Wrap-up – Risk Adjustment for the
Medicare Shared Saving Program
Risk adjustment helps to more accurately establish an ACO’s
performance target, that is used to measure the ACO’s financial performance.
Risk adjustment methodology reflects changes in health status of an
ACO’s assigned beneficiary population over time, both in
establishing the historical benchmark and during the performance year.
Risk adjustment is performed at the population level, taking into