Using Data for Quality Improvement: Reporting and Payment
The Maryland Experience
AHRQ Conference
Using Administrative Data to Answer State Policy Questions
Robert Murray, Executive Director
Maryland Health Services Cost Review Commission
BMurray@HSCRC.State.MD.US
Slide 2
Overview of Presentation
• Context: A Self-Contained Data Collection and Reimbursement System
• Data Bases established for Rate System
• Data Considerations
• Quality of Care Example/Application
• Reporting
• Link to Payment and Financial Incentives
Slide 3
Context: Maryland All-Payer Hospital Rate Setting System
• Last State to Control Hospital Charges (All-Payer)
• System made possible by Waiver from Medicare
• Primary Statutory Responsibilities:
• Very strong data collection authority
• Rate setting authority
• Data are the Foundation & Building Blocks
• Many Positive Externalities from Data Collection
• Comparative analyses
• Basis for rate system
• Use of data by consumers and public
• Evaluation of disparities and inequity
• Pay for Performance and Quality Assessment
Slide 4
Policy Objectives & Use of Data
• Cost Containment (cost data à payment)
• Access to Care (data on uninsured à UC Pools)
• Equity in Payment (data on payment levels)
• Financial Stability (data on operating performance)
• Accountability/Transparency (System performance vs. Targets; Community Benefit Performance)
• Now a focus on Quality Improvement
Slide 5
Maryland Data Bases & Applications
• Service Volumes, Cost and Financial Data à Payment
• Medical Record Discharge Data à Structuring Payment DRGs
• Extensive data on the uninsured receiving care à UC Pools
• Wage and salary data by facility à Adjust Payment (LMA)
• Residents and Interns Survey à Adjust Payment (GME)
• Financial and Operating Data à Monitor Financial Stability
• Community Benefit Data à Hold Hospitals Accountable
• Present on Admission à Lower Complication Rates
• Admissions and Readmissions à Lower Re-Admission Rates
Slide 6
Importance of "Data Efficacy"
• How Complete?
• Sampling less desirable and less defensible
• How Accurate?
• Audits, Cross-checks & Reconciliations
• Benchmarks vs. Other States
• Uses of the data (for payment?)
• How Timely?
• Health Care Market changes rapidly
• Most effective policy decisions require timely data (<2 years old)
• How Robust?
• Availability of other data for adjustments/correlations
• Policy Decisions more powerful when data bases are combined
• Thresholds for being able to use data for reporting or payment
• How Fair?
• Adjust for factors beyond the control of providers
• Adjust for certain factors you don't want providers to influence
Slide 7
Characteristics of Data Use in Maryland
• Very direct link: Data à Policy Decisions
• Entire system built from bottom up using granular data
• Many positive externalities to comprehensive data collection effort (research, public health)
• Large role for public agency to make data available for the Market and Public
Slide 8
Example: Using Administrative Data to Lower Complication & Re-Admission Rates
Slide 9
Re-Admission Rates &Diagnosis Present on Admission (POA) - Context/Rationale:
• Next logical step after process measure P4P
• CMS taken first step: Hospital Acquired Conditions
• States can go further - tailor concept to local conditions
• Goal: To Reduce Complication and Re-admission rates
• Focus attention on poor performers (reporting) and correct payment incentives
• Reward hospitals who are doing the best job - lowest complication rates and re-admission rates (risk-adjusted)
Slide 10
Key Elements in the Exercise
• Goal: Improve Quality of care (and reduce cost) by lowering complication and re-admission rates
• Data use: Administrative Discharge Data Set
• Key Data Elements:
• Present on Admission indicator (POA) for complications
• Probabilistic match of patients in data set across hospitals for re-admissions
• Other tool required: Use of Severity Adjusted DRGs
• Mechanisms to create behavioral change by hospitals:
• Private or Public reporting of performance
• Link to payment (Medicaid and/or Large private payer in state)
Slide 11
PPCs and PPRs
• Potentially Preventable Complications (PPCs)
• Harmful events (accidental laceration during a procedure) or negative outcomes (hospital acquired pneumonia) that may result from the process of care and treatment rather than from a natural progression of underlying disease
• Potentially Preventable Readmissions (PPRs)
• Return hospitalizations that may result from deficiencies in the process of care and treatment (readmission for a surgical wound infection) or lack of post discharge follow-up (prescription not filled) rather than unrelated events that occur post discharge (broken leg due to trauma).
Note: PPRs/PPCs definitions and methodology developed by 3M Health Information Systems
Slide 12
Major PPCs (Twenty-nine of the Most Significant PPCs)
Major Cardiac and Pulmonary Complications
• Stroke & Intracranial Hemorrhage
• Extreme CNS Complications
• Acute Lung Edema & Respiratory Failure
• Pneumonia, Lung Infection
• Aspiration Pneumonia
• Pulmonary Embolism
• Shock
• Congestive Heart Failure
• Acute Myocardial Infarct
• V Fibrillation, Cardiac Arrest
• Pulmonary Vascular Complications
Other Major Medical Complications
• Major GI Complications w transfusion
• Major Liver Complications
• Other Major GI Complications
• Renal Failure with Dialysis
• Post-Hem & Other Acute Anemia w transfusion
• Decubitus Ulcer
• Septicemia & Severe Infection
• Other Major Complications of Medical Care
Major Peri-Operative Complications
• Post-Op Wound Infection & Deep Wound Disruption w Procedure
• Reopening or Revision of Surgical Site
• Post-Op Hemorrhage & Hematoma w Hemorrhage Control Proc or I&D Proc
• Post-Op Foreign Body & Inappropriate Op
• Post-Op Respiratory Failure with Tracheostomy
Major Complications of Devices, Grafts, Etc.
• Malfunction of Device, Prosthesis, Graft
• Infection, Inflammation, & Other Comp of Devices and Grafts Excluding Vascular Infection
• Complications of Central Venous & Other Vascular Catheters & Devices
Major Obstetrical Complications
• Obstetrical Hemorrhage w Transfusion
• Major Obstetrical Complications
3M Health Information Systems
Slide 13
Redesigning Incentives - PPCs
• Using Administrative data (and POA) - can calculate rates of PPCs by hospital
• Rates of Complications are specific to each facility but risk adjusted to account for its patient population
• Identify where there is statistically significant variation from an "expected" rate of complications
• The Expected rate - Policy decision
• Best practice?
• Statewide average?
• Potential Applications:
• Provide Reports back to the Hospital (private reporting - NY state)
• Publish performance (PPRs - Florida )
• Link to payment (Medicaid and/or Private Payers)
Slide 14
NY Hospital Example 2003 Major PPCs - All Service Lines
Stroke & Intracranial Hemorrhage
• Discharges at risk for PPCs: 39509
• Discharges with major PPC
• Actual: 79
• Expected: 89.4
• Major PPC/1000
• Actual: 2.00
• Expected: 2.26
• Percent difference: -11.7
Extreme CNS Complications
• Discharges at risk for PPCs: 37958
• Discharges with major PPC
• Actual: 18
• Expected: 26.7
• Major PPC/1000
• Actual: 0.47
• Expected: 0.70
• Percent difference: -32.7
Acute Lung Edema & Respiratory Failure
• Discharges at risk for PPCs: 39078
• Discharges with major PPC
• Actual: 398
• Expected: 460.6
• Major PPC/1000
• Actual: 10.18
• Expected: 11.79
• Percent difference: -13.6
Pneumonia, Lung Infection
• Discharges at risk for PPCs: 36506
• Discharges with major PPC
• Actual: 292
• Expected: 261.2
• Major PPC/1000
• Actual: 8.00
• Expected: 7.16
• Percent difference: 11.8
Aspiration Pneumonia
• Discharges at risk for PPCs: 38055
• Discharges with major PPC
• Actual: 101
• Expected: 101.5
• Major PPC/1000
• Actual: 2.65
• Expected: 2.67
• Percent difference: -0.5
Pulmonary Embolism
• Discharges at risk for PPCs: 40076
• Discharges with major PPC
• Actual: 34
• Expected: 36.7
• Major PPC/1000
• Actual: 0.85
• Expected: 0.92
• Percent difference: -7.4
Shock
• Discharges at risk for PPCs: 39761
• Discharges with major PPC
• Actual: 68
• Expected: 97.4
• Major PPC/1000
• Actual: 1.71
• Expected: 2.45
• Percent difference: -30.2
Congestive Heart Failure
• Discharges at risk for PPCs: 35732
• Discharges with major PPC
• Actual: 189
• Expected: 109.5
• Major PPC/1000
• Actual: 5.29
• Expected: 3.06
• Percent difference: 72.9
Acute Myocardial Infarct
• Discharges at risk for PPCs: 38813
• Discharges with major PPC
• Actual: 146
• Expected: 154.3
• Major PPC/1000
• Actual: 3.76
• Expected: 3.98
• Percent difference: -5.4
Ventricular Fibrillation/Cardiac Arrest
• Discharges at risk for PPCs: 40291
• Discharges with major PPC
• Actual: 133
• Expected: 133.2
• Major PPC/1000
• Actual: 3.30
• Expected: 3.31
• Percent difference: -0.2
PV Complications Except DVT
• Discharges at risk for PPCs: 40056
• Discharges with major PPC
• Actual: 17
• Expected: 25.5
• Major PPC/1000
• Actual: 0.42
• Expected: 0.64
• Percent difference: -33.2
Major GI Complications with Transfusion
• Discharges at risk for PPCs: 34142
• Discharges with major PPC
• Actual: 29
• Expected: 26.6
• Major PPC/1000
• Actual: 0.85
• Expected: 0.78
• Percent difference: 9.0
Major Liver Complications
• Discharges at risk for PPCs: 39953
• Discharges with major PPC
• Actual: 10
• Expected: 16.1
• Major PPC/1000
• Actual: 0.25
• Expected: 0.40
• Percent difference: -37.7
Other GI Complications with Transfusion
• Discharges at risk for PPCs: 34197
• Discharges with major PPC
• Actual: 24
• Expected: 13.9
• Major PPC/1000
• Actual: 0.70
• Expected: 0.41
• Percent difference: 72.1
Renal Failure with Dialysis
• Discharges at risk for PPCs: 39033
• Discharges with major PPC
• Actual: 23
• Expected: 26.2
• Major PPC/1000
• Actual: 0.59
• Expected: 0.67
• Percent difference: -12.0
3M Health Information Systems
Slide 15
Data Considerations
Data Validity Issues for PPCs
• Present on Admission (POA) now required by Medicare
• Must Verify Accuracy of Present on Admission Statistic
• Error/Edit checks
• Bench mark vs. other States (California/Maryland)
• Verify accuracy of overall SDX and procedure coding
Data Validity Issues for PPRs
• Probabilistic matching to track patients across hospitals
Slide 16
Link to Payment - Rates of PPCs/PPRs
• Can Aggregate Results into overall Quality Scores and rank hospital performance on 2 dimensions
• Attainment (absolute level in a given year)
• Improvement (year-to-year performance)
• Hospital Attainment/Improvement scores can be calculated and arrayed on a distribution
• Medicaid/Private Payers can redistribute some proportion of payment (amount "at-risk") based on performance along this distribution
• Applies to both PPCs and PPRs
Slide 17
Translating a Distribution of Performers to Payment (Medicare Value based Purchasing)
This slide contains a graph regarding a Pay for Performance scheme. Up until the proportion of points earned is 0.25, the proportion of conditional payments earned is zero. When the proportion of points earned is between 0.25 and 0.75, the proportion of conditional payments earned rises from 0.15 until 1. The curve between 0.25 and 0.75 proportion of points earned is considered the Exchange Function. To determine where hospital performance belongs on this curve, the distribution of hospital performance (PPC rates vs. Expected) are calculated. Hospitals with higher levels of attainment or improvement scores will be placed at the top of the curve, close to the maximum reward of 100 percent payback.
Slide 18
Link to Payment - Payment Reductions
• For Complications that are "highly preventable" (like Medicare HACs) - DRG payments should be reduced
• Highly preventable PPCs are 100% or nearly 100% preventable
• They show very little variation across hospitals after adjusting for risk factors
• Payment reductions applicable to DRG-based payment systems
• Craft payment decrement commensurate with level of preventability (i.e., 90% decrement & 10% retention)
Slide 19
Flaw in Severity Adjusted Payment System that needs to be fixed
APR-DRG System
Developed for an "All-Patient" population
Clinical logic more appropriate for all types of care
314 DRG categories
4 Splits based on clinical factors for different levels of "severity" of illness (SOI)
The more complications, the higher the SOI
There is an incomplete table listing DRGs by number and associated costs based on the level of SOI. Cost rises for each DRG as SOIs increase.
Slide 20
Case Examples of Preventable Complications and how the current Payment System unfairly and inappropriately increases a Hospital's revenue when it makes a preventable mistake
This lower half of this slide contains a table of a case example of a preventable complication of DRG 221. The patient suffered from PPC 38, Post-Op wound infection and deep wound disruption with procedure, and was readmitted to the hospital, increasing the cost of the hospital bill from $16734 to $25938, $9204 of unintended revenue for the hospital.
|
