CISummit 2013: Bruce Landon, Clinically Integrated Networks and ACOs: Preparing for Risk and Reward

Clinically Integrated Networks and ACOs: Preparing for Risk and Reward

Bruce E. Landon, M.D., M.B.A.

Connected Insight Summit 2013October 8, 2013

Agenda

• Background—Policy context• Defining ACOs• Déjà vu• Identifying Organizations• Early Evidence and Major challenges

* Estimate is statistically different from estimate for the previous year shown (p<.05).

SOURCE: Kaiser/HRET Survey of Employer-Sponsored Health Benefits, 1999-2013.

Average Annual Premiums for Single and Family Coverage, 1999-2013

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

$0 $2,000 $4,000 $6,000 $8,000 $10,000 $12,000 $14,000 $16,000 $18,000

$16,351*

$15,745*

$15,073*

$13,770*

$13,375*

$12,680*

$12,106*

$11,480*

$10,880*

$9,950*

$9,068*

$8,003*

$7,061*

$6,438*

$5,791*

$5,884*

$5,615*

$5,429*

$5,049*

$4,824*

$4,704*

$4,479*

$4,242*

$4,024*

$3,695*

$3,383*

$3,083*

$2,689*

$2,471*

$2,196*

Single CoverageFamily Coverage

Cumulative Increases in Health Insurance Premiums, Workers’ Contributions to Premiums, Inflation, and Workers’ Earnings, 1999-2013

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 20130%

50%

100%

150%

200%

250%

57%

119%

182%

56%

117%

196%

14%

34%

50%

11%

29%40%

Health Insurance PremiumsWorkers' Contribution to PremiumsWorkers' EarningsOverall Inflation

SOURCE: Kaiser/HRET Survey of Employer-Sponsored Health Benefits, 1999-2013. Bureau of Labor Statistics, Consumer Price Index, U.S. City Average of Annual Inflation (April to April), 1999-2013; Bureau of Labor Statistics, Seasonally Adjusted Data from the Current Employment Statistics Survey, 1999-2013 (April to April).

Health Insurance Coverage in MA

Motivation

UnsustainableSpending

Sub-optimalQuality

“Accountable” CareOrganizations (ACOs)

Defining Accountable Care Organizations (ACOs)

• A group of providers (can include hospitals) that accepts joint accountability for health care spending and quality for a defined population of patients.– Spending compared to a target “budget”– Quality measured separately– Patients either elect in or are assigned

(prospectively or retrospectively)

The Evolution of Payment Systems:From Quantity to Value

Fee-for-ServicePay for performance

Bundled/Global payment (capitation)

• Time spent improving quality is time away from revenue generating visits

• Rewards quantity over quality

• Adds incentives for quality performance

• May support infrastructure costs

• Rewarded if downstream spending avoided

• Robust quality incentives needed

Accountability for SpendingKey: Spending is measured relative to a target.

1-sided ACOSpending

Target

This yearLast year Next year

Savings

Excess

2-sided ACO

-- Penalty

Reward Reward

Determined by FFS

Spending

ACOs in MedicarePioneer– 32 “Advanced” orgs, 15,000+

benes– 1 sided2 sided riskglobal

payment – Prospective assignment– 669,000 benes

• Shared Savings Program– Section 3022 of ACA– 220 selected, 5,000+ benes– Mostly one-sided (215/220)– Retrospective assignment– 3.2 million benes

Déjà vu?

• Why is today different?• What did we learn from the 90s?

Back in the 90s….

• What happened if an organization performed really well?

LOWER BUDGETS!

Multiyear Agreements

• Global payment is now a multi-period game– Success is not (necessarily) rewarded with a

budget cut!• Sufficiently long to learn and then accrue the

benefits of improved care management• Changing practice (and culture takes time)• Growth rates trend down over time

“I wish I could help you. The problem is that you’re too sick for managed care.”

Selection

• If risk plans siphon the healthiest patients, savings might be illusory

• Many examples from the past– Medicare plans with offices on the second floor– Offering free gym memberships, sneakers, etc.

Risk Adjustment

• Used to be age/sex • Now age/sex/diagnoses

– uses concurrent to adjust prospectively set budgets

– Best models (DxCG)– Diagnoses (in), procedures/drugs/hospitalizations

(out)• R2 20.6% in commercial, 15-17% in Medicare

• Prior spending for your population

Flying Blind

Infrastructure

• Better, more available data• Widespread adoption of EMRs (particularly

here)• Increasing EMR functionality• E-prescribing• Care management systems, enhanced ability

to coordinate and manage care

Robust Quality Incentives

• Pay for performance and other infrastructure payments

• Robust sets of measures (AQC and Medicare)• Higher amounts at stake

– AQC: originally up to 10% of budget, now determines shared savings

– Medicare: determines shared savings

Accountability for QualityKey: Quality is measured and reported, with incentives tied to performance.

Domain # Example

Patient Experience 7 Patient’s rating of doctor

Care Coordination 6 Rates of readmissions

Preventive health 8 Tobacco screening

At-risk populations 12 Hemoglobin A1c < 8%

Medicare “Shared Savings Program”

Year 1 – pay for reportingYears 2-3 – pay for reporting and performance

Who Should Become an ACO?

Potential Winners and Losers?

McWilliams, Chernew, Zaslavsky, Landon. Delivery System Integration and Health Care Spending and Quality for Medicare Beneficiaries. JAMA Intern Med. 2013;173(15):1447-1456. doi:10.1001/jamainternmed.2013.6886

Profligate Spenders v. Organized Groups?

• Profligate Spenders– Loosely connected– Poorly integrated– Culture of excess– But…..budgets will be

generous

• Organized Groups– Tightly integrated– Tightly managed– Culture of value– But…budgets are already

constrained!

Identifying ACOs

• Organic networks could form the rational basis for ACOs– To identify organizations ready to become ACOs– To identify markets ready to transition to global payment

• Monitoring performance– Measuring cohesiveness over time using a variety of

measures– Measuring leakage

Barabasi A. N Engl J Med 2007;357:404-407

Complex Networks of Relevance to Network Medicine

Building Physician Networks

Date of download: 9/21/2012Copyright © 2012 American Medical Association.

All rights reserved.

From: Variation in Patient-Sharing Networks of Physicians Across the United States

JAMA. 2012;308(3):265-273. doi:10.1001/jama.2012.7615

Methods: Community Detection

• Network communities are associated with functional networks

• Identify sets of nodes that are more connected than expected—optimize assignment across communities

• Straw man—compare properties with hospital affiliation networks

Community Detection Algorithm

is number of shared patients between two MDs is the degree (or strength) of node is the number of edges in the network (or their weight) is the community assignment of node is the Kronecker delta which is equal to 1 if the arguments are identical, otherwise it is zero

𝑄= 12𝑚∑

𝑖∑𝑗

[𝐴𝑖𝑗−𝑘𝑖𝑘 𝑗

2𝑚 ]𝛿(𝑠𝑖 , 𝑠 𝑗)

Tallahassee FL and Norfolk VA

Using Administrative Data to Identify Naturally Occurring Networks of Physicians.Landon, Bruce; MD, MBA; Onnela, Jukka-Pekka; Keating, Nancy; MD, MPH; Barnett, Michael; Paul, Sudeshna; OMalley, Alistair; Keegan, Thomas; Christakis, Nicholas; MD, PhD Medical Care. 51(8):715-721, August 2013.

Communities (n=273) Hospitals (n=416)

Percent with at least 1:

Orthopedist 97 97

Ophthalmologist 9 92

Cardiologist 96 87***

Neurologist 91 82**

Psychiatrist 84 76*

Dermatologist 85 75*

Gastroenterologist 86 82

Network Characteristics of Community and Hospital Networks

Percentage of Care in Potential ACOs, at Least 5 PCPs and 3,000+ Patients

0

10

20

30

40

50

Hospital

Community (1 hospital per community)

Emer

gen

cy R

oo

m V

isit

s (%

)

6264666870727476

Adm

issi

ons

(%)

Percentage of Care in Potential ACOs, at Least 5 PCPs and 3,000+ Patients

020406080

100

Phys

icia

n Vi

sits

(%

)

020406080

100

Hospital

Community (1 hospital per community)

PCP

Visi

ts (%

)

Overall Standardized to Median Sized Hospital

Standardized to Median Sized Community

020406080

100

Spec

ialis

t Vis

its

(%)

Early Evidence

The Alternative Quality Contract (BCBSMA)

The “Halo” Effect(Spillover to Medicare Patients)

McWilliams, Landon, Chernew. JAMA. 2013;310(8):829-836. doi:10.1001/jama.2013.276302

1

Challenges/Issues

• Alignment of Incentives• “Keeping Score”*• Investing to reorganize care delivery• Disincentives for advanced organizations• ACO model versus Medicare Advantage

*Perspective. Keeping Score under a Global Payment System. Bruce E. Landon, M.D., M.B.A.N Engl J Med 2012; 366:393-395

Viewpoint July 24, 2013. Reenvisioning Specialty Care and Payment Under Global Payment SystemsBruce E. Landon, MD, MBA; David H. Roberts, MD

Conclusions

• Maintaining the status quo is no longer tenable• ACOs are at the vanguard of a larger movement

towards payment reform• Time of great change…with great opportunity• To succeed under these arrangements will take

significant efforts to reorganize how care is delivered

• Network Science might provide useful tools for identifying and tracking ACO performance

Thank you!

• landon@hcp.med.harvard.edu• blandon@bidmc.harvard.edu