linking incomes to outcomes did you really get what they said you got? michael pine, m.d., m.b.a....

Linking Incomes to Linking Incomes to OutcomesOutcomes

Did You Really Get Did You Really Get What They Said You Got?What They Said You Got?

Michael Pine, M.D., M.B.A.

Michael Pine and Associates, Inc.(773) 643-1700

[email protected]

© 2009 Michael Pine and Associates, Inc.

OverviewOverview

Creating a Value-Driven Health Care Market

Affordable Data We Can Believe In

Comparative Performance of Alternative Data Sets

From Information to Understanding to Action

Accountability in a Value-Driven Market

Creating a Value-DrivenCreating a Value-DrivenHealth Care MarketHealth Care Market


Creating a Value-Driven Health Care MarketCreating a Value-Driven Health Care MarketManaging Seven Essential RsManaging Seven Essential Rs

REWARDSREWARDS

contract

RESPONSIBILITIESRESPONSIBILITIES

STATICSTATICSTRUCTURESSTRUCTURES

DYNAMICDYNAMICPROCESSESPROCESSES

contract

RESOURCES, RESOURCES, ROLESROLES

& & RELATIONSHIPSRELATIONSHIPSRESULTSRESULTS

Hospital 21

Hospital 20

Hospital 19

Hospital 18

Hospital 17

Hospital 16

Hospital 15

Hospital 14

Hospital 13

Hospital 12

Hospital 11

Hospital 10

Hospital 9

Hospital 8

Hospital 7

Hospital 6

Hospital 5

Hospital 4

Hospital 3

Hospital 2

Hospital 1

ADVERSE OUTCOMES IN PERCENT5% 10% 15% 20% 25% 30% 35% 40%

P > 0.05 P = 0.01 to 0.05 P = 0.001 to 0.01 Predicted Observed

Hospital 21

Hospital 20

Hospital 19

Hospital 18

Hospital 17

Hospital 16

Hospital 15

Hospital 14

Hospital 13

Hospital 12

Hospital 11

Hospital 10

Hospital 9

Hospital 8

Hospital 7

Hospital 6

Hospital 5

Hospital 4

Hospital 3

Hospital 2

Hospital 1

ADVERSE OUTCOMES IN PERCENT5% 10% 15% 20% 25% 30% 35% 40%


RISKSRISKS


Aligns Risks and Responsibilities

Links Results and Rewards

Balances Quality and Cost

Combines Individual Choice and Market Discipline

Provides Accurate, Relevant Information

Holds All Participants Accountable

Characteristics of a Value-Driven Characteristics of a Value-Driven MarketMarket


Assign

Responsibility

Defines

Roles

Convey

Authority

Invites

AccountabilityAccountability

Requires

Accountability and Performance Accountability and Performance MeasuresMeasures

Measures


DecisionsDecisions

DataDataProvide

Information

Produces

Understanding

Supports

Invite

EvaluationEvaluation

Requires

Risk-Risk-AdjustmentAdjustment

Mutes Mutes ExtraneousExtraneousInfluencesInfluences

Guide

Actions

Linking Data, Decisions, and Linking Data, Decisions, and AccountabilityAccountability


Risk-Adjustment and Performance Assessment

Mortality in CABG Surgery

Hospital 4

Hospital 2

Hospital 8

Hospital 9

Hospital 15

Hospital 16

Hospital 17

Hospital 11

Hospital 12

Hospital 3

Hospital 5

Hospital 14

Hospital 10

Hospital 1

Hospital 13

Hospital 18

Hospital 7

Hospital 6

0% 2% 4% 6% 8%MORTALITY IN PERCENT


Affordable DataAffordable DataWe Can Believe InWe Can Believe In


Data for Monitoring Clinical Data for Monitoring Clinical PerformancePerformance

Claims Data HCFA Mortality ReportsHealthGrades.com HCUP Inpatient Quality and Patient Safety

Indicators

Clinical DataAPACHE Pennsylvania Health Care Cost Containment

CouncilCleveland Health Quality ChoiceSpecialty Society Registries (e.g., STS, ACC)


Claims Data Versus Clinical DataClaims Data Versus Clinical Data

Data Is the Foundation for:Public ReportingPerformance-Based ReimbursementQuality Improvement Initiatives

Must Balance the Need for:Accurate Measurement of Clinical

PerformanceEase and Cost of Data Collection


Relative Ease of Data CollectionRelative Ease of Data Collection

Standard ClaimsStandard Claims

Numerical LaboratoryNumerical Laboratory

Vital SignsVital Signs

Other Clinical DataOther Clinical DataManual

Automated

Dat

a C

oll

ecti

on

Claims DataClaims Data

Clinical DataClinical Data


Efficient Use of Clinical DataEfficient Use of Clinical Data

AlbuminMental Status

Hemoglobin FEV1

Cost to Cost to CollectCollect Hig

hLow

AnalytiAnalytic c

PowerPower

High

Low


Enhancing Claims DataEnhancing Claims Data

Present-on-Admission CodingMayo ClinicNew York State’s SPARCS DatabaseCalifornia’s OSHPD DatabaseUB-04 CMS’s New Coding Requirements

Numerical Laboratory DataMichael Pine and Associates Agency for Healthcare Research and Quality

(AHRQ)

AHRQ’s New Hybrid Database Demonstrations


Creating a Hybrid DatabaseCreating a Hybrid Database

Standard ClaimsStandard Claims

Numerical LaboratoryNumerical Laboratory

Vital SignsVital Signs

Other Clinical DataOther Clinical Data

Present-on-AdmissionPresent-on-Admission

Claims DataClaims Data

Clinical DataClinical Data

Hybrid DataHybrid Data


Potential Benefits of a Hybrid Potential Benefits of a Hybrid DatabaseDatabase

Explicitly Distinguish BetweenComorbidities That Are Present on

AdmissionComplications That Occur During

Hospitalization

Provide Objective Clinical Data Validate the Subjective Assignment of

DiagnosesAid in Defining the Severity of Diagnosed

ConditionsAid in Delineating Underlying

Pathophysiology

Comparative Performance of Comparative Performance of Alternative Data SetsAlternative Data Sets


Sources of Data for AnalysisSources of Data for Analysis

188 Pennsylvania Hospitals for Primary AnalysesClaims Data for Discharges from 7/00 to 6/03Corresponding AtlasTM Clinical Data

Abstracted from Medical RecordsHospital Day Recorded for Each Data Element

New York and California Claims DataIdentify Potentially Problematic Risk FactorsAssess Effect of Improperly Designated

Complications


Inpatient Quality Indicators (Mortality)Inpatient Quality Indicators (Mortality)

Medical ConditionsAcute Myocardial InfarctionCerebrovascular AccidentCongestive Heart FailureGastrointestinal HemorrhagePneumonia

Surgical ProceduresAbdominal Aortic Aneurysm RepairCoronary Artery Bypass Graft SurgeryCraniotomy


Patient Safety Indicators Patient Safety Indicators (Complications)(Complications)

Elective Surgical Procedures

ComplicationsPhysiologic / Metabolic AbnormalitiesPulmonary Embolus / Deep Vein ThrombosisSepsisRespiratory Failure


Data Used in CLAIMS ModelsData Used in CLAIMS Models

Age and Sex

Principal Diagnosis

Secondary DiagnosesChronic ConditionsConditions Generally Present on Admission

Selected Surgical Procedures


Data Used in POA and HYBRID ModelsData Used in POA and HYBRID Models

POA ModelsAll Data Used in CLAIMS ModelsAdditional Secondary Diagnoses

Frequently Hospital-AcquiredUsed When Clinical Data Establish Presence on

Admission

HYBRID ModelsAll Data Used in POA ModelsNumerical Laboratory Data

Routine Chemistry, Hematology, and Blood Gas Analyses

Available in Electronic Form from Most Hospitals


Data Used in CLINICAL ModelsData Used in CLINICAL Models

All Data Used in HYBRID Models

Vital Signs

Laboratory Data Not in HYBRID Models e.g., bacteriological analyses, cardiac ejection

fraction

Key Clinical Findings from Medical Recordse.g., immunocompromised, lethargic

Composite Clinical Scores e.g., ASA Classification, Glasgow Coma Score


Bias Due to Suboptimal DataBias Due to Suboptimal Data

+ 2 Std Dev

Good Average Poor

- 2 Std Dev

Measured Performance

+ 0.5 Std Dev

Problematic Problematic

- 0.5 Std Dev

Bias

OK


Bias Due to Suboptimal Data Bias Due to Suboptimal Data (Mortality)(Mortality)

0%

10%

20%

30%

40%

50%

60%

70%

0.5 1.0 1.5 2.0Upper Threshold for Bias in Standard Deviations

Per

cen

t E

xcee

din

g U

pp

erT

hre

sho

ld

RAW CLAIMS POA HYBRID


Bias Due to Suboptimal Data Bias Due to Suboptimal Data (Complications)(Complications)

0%

10%

20%

30%

40%

50%

60%

70%


Per

cen

t E

xcee

din

g U

pp

erT

hre

sho

ld

RAW CLAIMS POA HYBRID


Bias in Measurement of ComplicationsBias in Measurement of Complications

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%


Per

cen

t E

xcee

din

g U

pp

er T

hre

sho

ld

Observed vs Predicted Rates of True Complications

Bias Due to Failure to Risk-Adjust True Complication Rates

Bias Due to Misclassifying Comorbidities As Complications


Numerical Laboratory DataNumerical Laboratory Data

22 Tests Enter At Least 1 Model

14 of These Tests Enter 4 or More Models• pH (11) • SGOT (7)

• Prothrombin Time (10)

• Platelet Count (7)

• Sodium (9) • Albumin (5)

• White Blood Count (9) • pCO2 (4)

• Blood Urea Nitrogen (8)

• Glucose (4)

• pO2 (8) • Creatinine (4)

• Potassium (7) • CPK-MB (4)


Vital Signs, Other Lab Data, Composite Scores

All Vital Signs Enter 4 or More Models

Culture Results Enter 2 Models

Ejection Fraction Enters 2 Models

Both Composite Scores Enter 4 or More Models

•ASA Classification (6)•Glasgow Coma Score (4)

•Pulse (8) •Blood Pressure (6)

•Temperature (6) •Respirations (5)


Abstracted Key Clinical FindingsAbstracted Key Clinical Findings

35 Clinical Findings Enter At Least 1 Model

Only 3 of These Enter More Than 2 ModelsComa (6)Severe Malnutrition (4)Immunosuppressed (4)

14 Have Corresponding ICD-9-CM Codese.g., coma, severe malnutrition

Coding Regulations Limit Utility of Claims Data

The Bottom LineThe Bottom Line

Claims Data Enhanced with Present-on-Admission Modifiers and

Numerical Lab Data Can SupportAccurate Performance Assessment

From Information to From Information to Understanding to ActionUnderstanding to Action


Understanding

Motivation

Action

Information

Knowledge

Explanation

From Information to Understanding to Action


Three Barriers to Effective Decision Three Barriers to Effective Decision MakingMaking

Inconsistent Reporting of Complications

Dissociation of Services and Clinical Benefits

Inability to Relate Outcomes to Processes of Care


Coding Hospital-Acquired Coding Hospital-Acquired ComplicationsComplications

Potential Barriers to Accurate CodingExpertise and Teamwork Required for Accurate

CodingDifficulty Achieving Consistency in ReportingBenefits to Hospitals of Not Coding Complications

Consequences of Inconsistent CodingAffects Comparative Assessments of Clinical QualityAffects Reimbursement

Detection of Coding Errors Chart Reviews Are Inefficient and CostlyWell-Designed Screens Can Detect Problems

Efficiently


Screens for Correct Coding of Screens for Correct Coding of ComplicationsComplications

Types of Admissions ScreenedAdmissions for High-Risk Medical ConditionsAdmissions for Elective Surgical ProceduresAdmissions for Childbirth

Nature of ScreensCoding of Chronic Conditions

Without Acute ComponentWith Acute Component

Coding of Conditions That Often Are Hospital-Acquired

Relation of Mortality Rates to When Condition Occurred

Relation of Coded Complications to Lengths of StayInternal Consistency of Obstetrical Coding


Risk-Adjusted Post-Operative Lengths of Stay

All Live Discharges

-10

0

10

20

30

40

50

60

1 31 61 91 121 151 181 211 241 271 301 331 361 391 421 451 481 511Sequence Identifier

O

BS L

OS m

inus

PR

ED

LO

S (d

ays)

Average 3 Std Dev Normal LOS Long LOS w Cpl Long LOS w/o Cpl


Risk-Adjusted Post-Operative Lengths of Stay

Live Discharges without Reported Complications

-10

0

10

20

30

40

50

60

1 38 82 129 180 225 270 310 354 398 448 491Sequence Identifier

O

BS L

OS m

inus

PR

ED

LO

S (d

ays)

Average 3 Std Dev Normal LOS Long LOS w/o Cpl


Distribution of Hospital POA Coding Distribution of Hospital POA Coding ScoresScores

ScoreHospitals

(#)Hospitals

(%)

>90% 65 39.4%

>80% to 90% 41 24.8%

>70% to 80% 26 15.8%

>60% to 70% 19 11.5%

60% or lower 14 8.5%

Total Scored 165 100%

>10% Unknown 22 n/a


Linking Results and RewardsLinking Results and Rewards

Patient(Clinical

Risk)

Benefits

(Results)

HealthcareServices

Structure

Payment

(Reward)$

Costs(Results)

$


Pricing Fragmented Components of Pricing Fragmented Components of CareCare

Fragmented Services

Process A

Structure

Process B

Process C$ A

$ B

$ C

Payment

(Reward)

Patient(Clinical

Risk)

Benefits

(Results)

Costs(Results)

$


Components of an Episode of CareComponents of an Episode of Care

Precipitating Event(Clinical Risk)

Care

Outcome(Benefit)


Reimbursement for Episodes of CareReimbursement for Episodes of Care

Benefits

(Results)

HealthcareServices

Structure Costs(Results)

Payment

(Reward)

New Health Event

ClinicalRisk

Premium

(Reward)

Risk of Occurren

ce

$

$

$Episode of Care


Services Associated with an Episode of Services Associated with an Episode of CareCare

Individualized Services

Required Services


Alternative Practice Patterns

ServicesRendered

ServicesRendered

ServicesRendered

Optimum Care Inefficient Care

Ineffective Care Ineffective, Inefficient Care

ServicesRendered


Costs of Alternative Practice PatternsCosts of Alternative Practice Patterns

Care of Associated

Adverse Outcomes

OptimumRoutine

Care

TOTAL

Cost

OptimumCost-Effectiveness

InefficientIneffective

Care

= Actual Cost

InefficientCare

Ineffective

Care


Payment in a Value-Driven MarketPayment in a Value-Driven Market

Insurance for Risk of Occurrence:

Capitation By Beneficiary

Evidence-Based Care Required by Population:Fee-for Service By Encounter

Individualized Health Care Services:Global Fee By Episode of Care

Care of Potentially-Avoidable Complications:Warranty For Episode of Care


Use Fair Empirically-Derived StandardsUse Fair Empirically-Derived StandardsTo Set Global Fees and WarrantiesTo Set Global Fees and Warranties

I would be agood boy if only

you loweredyour standards!

I would be agood boy if only

you loweredyour standards!


Standardized Hospital Costs and Adverse Standardized Hospital Costs and Adverse OutcomesOutcomes

350 High Performing & 113 Suboptimally Performing 350 High Performing & 113 Suboptimally Performing HospitalsHospitals

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

$4 $8 $12 $16 $20 $24 $28

Total Hospital Cost (thousands $)

% A

dv

ers

e O

utc

om

e R

ate

Good Performance Problematic Performance Average Good Average Problematic

6.91%$13,259

3.99%$11,192


Aligning Risks, Responsibilities, and Aligning Risks, Responsibilities, and RewardsRewards

In a Virtual PartnershipIn a Virtual Partnership

Payer Bears Risk of Occurrence

Managing Organization (e.g., Physician-in-Charge)Receives Standard Negotiated Payment Minus

WithholdOverruns in Total Cost of Episode Covered by WithholdTotal Savings Shared with Payers

Participating CaregiverReceives Standard Negotiated Payment Minus

WithholdAchievement of Intermediate Milestones Determines:

Return of WithholdPayment of Bonus

Consultant / Supplier Receives Standard Payment


• Provider Selection•Network Formation•Reimbursement•Accountability•Strategic Planning•Marketing

External MonitoringMonitoring

Internal Internal MonitoringMonitoring

• Quality Control•Quality Improvement•Cost Management

Assesses Performance

Links Processes to Outcomes

External and Internal MonitoringExternal and Internal Monitoring


Traditional Mortality and Morbidity Traditional Mortality and Morbidity ReviewReview

Analyses of Single Cases with Adverse Outcomes

Peer Review Aided by Medical Literature

Objectives VaryIdentify and Correct Substandard PracticeEducate ParticipantsImprove Processes of Care

Problems AboundRarely Affects Individual PracticeDivorced from Organizational Decision MakingLacks Scientific Credibility


Fallacy of Generalizing from Single Fallacy of Generalizing from Single CasesCases

One Tree

Does Not a Forest Make


Designing Robust Observation StudiesDesigning Robust Observation Studies

Strengths of Randomized Controlled Clinical TrialsRandomization Is Performed Prior to InterventionTreatment and Control Groups Are Similar

Overcoming Weaknesses of Observational StudiesTreatments Usually Are Not Randomly

AdministeredSelect Controls with Same Likelihood of

Treatment Propensity Analyses Match Important

Characteristics


Relating Clinical Processes to Relating Clinical Processes to OutcomesOutcomes

Clinical Care Often Is Individualized

Risk Profiles Affect Outcomes and Routine CareComplications Often Are Related to Higher Initial RiskTreatment May Vary with Initial RiskDifferences in Risk Profiles Confound Comparisons

Matching by Predicted Outcome Reduces BiasMatch Cases with and without ComplicationsCompare Potentially Important Elements of CareDifferences Suggest Opportunities for ImprovementChart Abstraction Often Required to Assess Processes


Identify Opportuniti

es to Improve

AlterProcess

es

A Cycle of Continuous Quality A Cycle of Continuous Quality ImprovementImprovement

Adverse Outcom

es

Analyze Process

es

Select Control

s

Plan for Improveme

nt

Matched Matched ControlsControlsReduce Reduce

Confounding Confounding BiasBias


Accountability in a Value-Driven Accountability in a Value-Driven MarketMarket

Information about risks and results guides:purchasing decisions and reimbursementperformance improvement initiatives

Evaluation focuses on episodes of care, not on individual cost centers

Margin and market share accurately reflect:quality of careclinical efficiency

Yes We Can!

linking incomes to outcomes did you really get what they said you got? michael pine, m.d., m.b.a....

Documents