Paying for Outcomes not Performance 1 3M. All Rights Reserved. Norbert Goldfield, M.D. Medical Director 3M Health Information Systems, Inc.
#Health Information Systems- Clinical Research Group Created original Medicare DRGs and CMS contractor for update and maintenance of the Medicare inpatient system since 1983 Developed for CMS the ICD-10-PCS, the General Equivalency Maps (GEMS). Now converting CMS payment, coverage determination and quality measures to ICD-10 3M APR Suite (APRs, PPRs, PPCs) is widely used in government and commercial payer quality outcomes & payment programs 3M HIS has designed and developed many Prospective Payment Systems in widespread use in the U.S. Operationally, we do not believe in Pilots. We believe in state wide (either private insurance and/or Medicaid) implementation followed by adoption, if interested by the Federal Government. We believe in adoption by other countries of our work. 2
An Outcomes Based Approach to Pay for Performance Pay for Performance links quality to payment Outcomes (our focus) Process- providers determine best processes and practices Focus on quality outcomes with largest payment impact Inpatient complications (PPCs) but more than HACs or Never Events Readmissions (PPRs) but related not all-cause Expand to Ambulatory Care: Admissions, ER Visits, Ancillary Services Fairly measure provider outcomes performance- deliver performance reports containing actionable information to foster improvement Adjust payment to provide direct financial incentives to decrease waste by reducing preventable events and increase efficiency 3 3M 2007. All rights reserved.
Key Elements of 3M Pay for Outcomes Methodology Population appropriate, clinically/financially meaningful classification and payment systems Potentially Preventable Events: classification of encounters/events in five major service areas that are potentially preventable and lead to unnecessary services or contribute to poor quality of care Measurement and scoring of provider PPE outcomes-comparison of risk adjusted PPE rates to empirically derived state norms-performance reports with actionable information Payment adjustment based on provider PPE performance compared to state best practice norms Payment bundling to incentivize responsible providers to increase efficiency and coordination ; will summarize our current CMS project 4
What Are Potentially Preventable Events A prime component of health care inefficiency and waste is the delivery of services that would be unnecessary if effective care was delivered Unnecessary services often lead to a increased payment In the context of a payer with a fixed expenditure budget, payments for unnecessary services result in lower payments to those providers who are delivering only necessary services. 3M has systematically identified and classified encounters/events in five major service areas that are potentially preventable and lead to unnecessary services or contribute to poor quality of care. Collectively known as Potentially Preventable Events (PPEs). PPEs will never be totally eliminated even with optimal care. Therefore, proper risk adjustment and scoring is required in order to use PPEs in provider profiling and payment systems 5 3M 2007. All rights reserved.
There are five types of health care encounters or events that are potentially preventable and lead to unnecessary services Potentially Preventable Complications (PPCs) Potentially Preventable Readmissions (PPRs) Potentially Preventable Admissions (PPAs) Potentially Preventable Emergency Room Visits (PPVs) Potentially Preventable Ancillary Services (PPSs) 6
PPE Controversies in General Categorical vs. Statistical Regression Model Never Pay vs. Rates Outcomes Measurement vs. Process Measurement National vs Locally Derived PPE s should not be an issue if you use a categorical as opposed to a regression model. 7
Potentially Preventable Readmissions Potentially Preventable Readmissions (PPR) Potentially Preventable Initial Admissions (PPIA) Potentially Preventable Visits (PPV) Potentially Preventable Complications (PPC) Potentially Preventable Services(PPS) 8
Readmissions Controversies a. Human vs. Hospital Centric Readmission rates b. All-Cause vs. Exclusions for readmissions that are not preventable c. Mental/Substance Abuse Services d. Palliative Care e. Discharge severity of illness vs. other approaches to risk adjustment f. 15 vs. 30 days readmission rates 9
Potentially Preventable Events Potentially Preventable Readmissions (PPR) Potentially Preventable Initial Admissions (PPIA) Potentially Preventable Visits (PPV) Potentially Preventable Complications (PPC) Potentially Preventable Services(PPS) 10
Controversy: PPCs vs. CMS Implementation of HACs Post admission HACs are excluded from the diagnoses used to assign the MS-DRG that determines Medicare payment Exclusion of the HACs means that any payment increase associated with an HAC is entirely eliminated The payment reduction is done for individual patients on a case-by-case basis 11
Limitations of the CMS HAC Payment Policy Elimination of the entire HAC payment increase on a case-by-case basis implies that HACs are always preventable Limits the number of HACs to near never event complications Initial set of FY2009 HACs reduce Medicare hospital payments by only two one-hundredths of one percent (0.0002) The CMS HAC payment policy is reasonable for near never event complications, but precludes significant expansion of HACs No new HACs added for FY2010. High volume, high impact complications like pneumonia and septicemia not included Provides no financial incentive for hospitals to code HAC diagnoses CMS HAC Medicaid rule mandates Medicaid programs to not pay for Medicare HACs-and also permits targeting other provider preventable conditions as Maryland has done 12
Potentially Preventable Events Potentially Preventable Readmissions (PPR) Potentially Preventable Admissions (PPA) Potentially Preventable Visits (PPV) Potentially Preventable Complications (PPC) Potentially Preventable Services(PPS) 13
Potentially Preventable Admissions (PPAs) The PPAs are more comprehensive than the U.S. DHHS Agency for Healthcare Research and Quality (AHRQ) list of ambulatory care sensitive conditions as initially defined in the 1980s (Prevention Quality Indicators or PQIs). Yet they are fairer to hospitals and accountable care organizations because they exclude those admissions that are not preventable except for years of prior preventive care. Two types of PPAs : those that can be implemented immediately vs 3-5 years down the road. 14
Importance of Risk Adjustment in PPA Measurement: Clinical Risk Groups CRGs are a categorical clinical model which uses standard claims data to assign each patient to a single mutually exclusive risk category. Each CRG is clinically meaningful and can be used to predict future health care utilization and cost (prospective) and to explain past health care utilization and cost (retrospective). Clinically meaningful groups of individuals who require similar amounts and type of resources DRGs: Resources used during a hospitalization CRGs: Total resources used in the future or in the past over an extended period of time A relative payment weight is associated with each group Payment weights: Reflects practice patterns Clinical groups: Describes type of individuals 15
ACRG3 Payment Weights for the Medicare Population Severity Level CRG Status 1 2 3 4 5 6 Healthy 0.2009 History of Significant Acute Disease Single Minor Chronic Disease Minor Chronic Disease in Multiple Organ Systems Single Dominant or Moderate Chronic Disease Disease in Chronic Multiple Organ Systems Dominant Chronic Disease in Three or More Organ Systems Dominant and Metastatic Malignancies 0.4993 0.4266 0.5867 0.4666 0.5640 0.6411 0.8663 0.5256 0.7189 0.9370 1.1841 2.0850 3.7962 0.8857 1.4277 2.1845 2.9002 3.6478 6.1852 1.3768 1.8098 2.5294 3.6102 4.9347 6.6154 1.4912 1.9160 2.9433 3.9762 5.1218 Catastrophic Conditions 1.5661 2.7608 5.3801 9.0080 10.8938 13.2945 16
Medicare Mortality Rate and Relative Resource Weights by Selected CRG for Subsequent Year Severity Level CRG 1 2 3 4 5 6 DM Pmt 0.5953 0.7797 0.9246 1.3985 Mort 2.8% 3.0% 4.0% 5.4% CHF Pmt 0.8950 0.9782 1.1783 1.7863 Mort 8.6% 9.1% 9.8% 13.6% COPD Pmt 0.8426 1.0144 1.3077 2.2961 Mort 4.7% 5.8% 7.5% 15.9% COPD & DM Pmt 0.9925 1.1082 1.4112 1.7560 2.2504 3.3735 Mort 5.3% 5.5% 6.4% 8.3% 8.8% 11.9% DM & CHF Pmt 1.0632 1.2664 1.6494 2.0645 2.6528 3.6650 Mort 7.1% 8.1% 10.6% 12.4% 15.7% 20.0% COPD & CHF Pmt 1.0956 1.4792 1.7433 2.2875 2.8244 3.8638 Mort 8.2% 12.4% 11.6% 14.6% 19.8% 23.9% DM & COPD & CHF Pmt 1.4588 2.1968 2.5539 3.2849 4.2358 5.7845 Mort 12.5% 12.9% 15.3% 17.7% 21.6% 30.7% 17
Potentially Preventable Events Potentially Preventable Readmissions (PPR) Potentially Preventable Initial Admissions (PPIA) Potentially Preventable Visits (PPV) Potentially Preventable Complications (PPC) Potentially Preventable Services (PPS) 18
Potentially Preventable Ancillary Services The research literature shows significant variation in the use of ancillary services. As the Medicare Payment Advisory Commission recently pointed out in a January 2011 report on regional variation: Areas that are high use in one sector (such as inpatient, ambulatory, and post-acute) tend to be high use overall, and all three sectors contribute to overall variation. We also find that areas with high service use among Medicare decedents (those who died during the year) tend to have high service use for non-decedents as well. In short, the pattern of high use often extends across different services and different groups of beneficiaries. 1 19
RISK ADJUSTMENT IS KEY Cannot overemphasize the importance of comprehensive and detailed risk adjustment Clinical Risk Groups used as underlying classification tool in PPS (and PPA and PPV) risk adjustment 20
Potentially Preventable Events Potentially Preventable Readmissions (PPR) Potentially Preventable Initial Admissions (PPIA) Potentially Preventable Visits (PPV) Potentially Preventable Complications (PPC) Potentially Preventable Services(PPS) 21
Classification Systems to identify preventable ER visits There have been several methods developed to identify potentially preventable emergency visits with the goal of reducing their frequency. Of greatest relevance: New York University Emergency Department Visit (NYU ED) severity algorithm The 3M Health Information Systems Enhanced Ambulatory Patient Group (EAPG) based classification system. 22
Differences between the EAPG based PPV system and other approaches: The PPVs identify potentially preventable emergency visits, and also identify associated potentially preventable ancillary services including radiology and pharmaceuticals. A visit may not have been preventable but the associated ancillary might have been, or both visit and the associated ancillary could have been preventable (e.g. CT Scan of the head for headaches that are not traumatic in nature). The 3M PPVs thereby enable focus on overuse of services provided in the ED. The PPV algorithm is flexible in that the end user can change the algorithm to exclude some services and include others. Ancillary services can be bundled or packaged into the PPV. PPVs can be used as the unit of comparison across all outpatient settings Using EAPGs as the unit of measure, the PPVs are risk adjusted for the individual visit. The PPVs can be further risk adjusted on a population basis using CRGs. 23
States and Private Insurers Using 3M Potentially Preventable Events Maryland has implemented Potentially Preventable Complications based pay for performance (see below). NY implementing Q4 2011 NY and MD now implementing PPR based pay for performance Three more states planning to do so in 2011-2012. Six states have established PPR comparative reporting (FL, TX, UT, WA,HI,NY). Five more are doing so in 2011 Texas and two other states evaluating the full Potentially Preventable Event suite in performance reporting in 2011-foundation for quality based payment CO, NC and NY developing medical home/aco programs with 3M tools Two states will adjust Medicaid MCP rates based on PPE performance Ten commercial payers analyzing PPE prevalence and costs in their claims and determining how to address them LTC and PPEs being explored in several states. 24
New Texas Quality Outcomes Based Payment Legislation Texas Senate Bill 7 passed June 2011 and effective Oct 2001 mandates ambitious Medicaid quality based payment program covering hospitals, managed care plans-also medical homes, ACOs and nursing homes Requires use of outcomes measures and specifically potentially preventable events. Mandates adjustment to hospital inpatient payment based on PPR and PPC performance Mandates adjustment of Medicaid managed care plan premiums based on plans performance with respect to quality measures including PPEs Encourages Commission to develop quality based payment for health homes which include measures for reductions for potentially preventable events Encourages Commission to develop P4P programs for nursing homes which may include reduction of preventable emergency visits Separate section mandates all public payers to establish PPR and PPC reporting 25
Potentially Preventable Events: Savings Direct PPR savings estimated at 2.5-5% of inpatient expenditures. Based on MedPAC 2007 Report to Congress NY Medicaid reducing inpatient expenditures by 1% using PPR and PPC performance adjustment to base rate Direct PPC savings estimated at 1-2 % of inpatient expenditures. Based on Maryland experience with PPC rate adjustment Direct PPA savings estimated at 4-8 % of inpatient expenditures. Direct savings for PPVs estimated at 1-2% hospital expenditures These estimates describe sustainable, as opposed one time, savings-as providers change behavior with respect to potentially preventable events 26
Prometheus - Impossible to Implement on a Large Scale Basis According to Recent Publication Tailoring difficult to replicate on a national scale. Payment rates tied to specific protocols place too much weight on judgments of a panel of experts and be too rigid to allow providers discretion to deliver appropriate care to individual patients. Providers with costs that were already below the rate would be attracted to the pilot and rewarded even if their behavior is unchanged. Keeping evidence-based case rates up-to-date enormous administrative challenge given rapidity of change in medical practice and technology. Tried before for hospital care and failed (PMCs) 27
Disease vs Patient Centered Episodes; Types of Events that can Meaningfully Trigger an Episode Disease centered episodes Must separate services into those related and those unrelated to the specific disease Patient centered, event based Focus on the individual patient s total disease burden No need to associate services with specific diseases Focus on all services before and after a health care event such as Hospitalization Significant procedure in a same day surgery unit or physician s office 28 8/9/2011 3M 2007. All rights reserved.
Components of a Patient-Centered, Event-based Episode Episode trigger: The event that precipitates the episode (e.g., a hospitalization for coronary bypass surgery) Episode acuity: The acuteness of the patient s conditions at the time of the episode trigger event (i.e. the severity of illness of the patient during hospitalization) Episode window: The number of days of pre-event and postevent encompassed by the episode Episode service scope: the services included in the episode (e.g., physician office visits, pharmaceutical usage) Chronic disease burden: The extent of the patient s comorbid chronic diseases at the beginning of the event 29 8/9/2011 3M 2007. All rights reserved.
Implementation Issues The hospital (or physician group) will receive the episode payment and be responsible for paying for all services in the episode scope that are not directly delivered by the hospital Hospital must act as a payer or Virtual bundle in which claims are tracked into bundles with retrospective adjustment based on virtual bundle performance prospectively applied Need access to full claims history at the time of admission for the purpose of determining chronic disease burden (i.e. assign the CRG) Negotiation of outlier/stop loss provisions Exclusion of completely unrelated services (e.g., readmission due to car accident) Need to monitor under use of services 30 8/9/2011 3M 2007. All rights reserved.
PPE Summary PPEs provide the sum total of the vast majority of immediately actionable events that are potentially preventable PPEs can be drilled down to the individual PPEs need to be population risk adjusted PPEs can be used to identify the effectiveness and/or payment incentives of any accountable care effort 31 3M 2009. All rights reserved 3M Confidential
Payment Adjustment Principles P4O should initially focus on those outcomes for which a quality failure results in an increase in payment. P4O financial incentives should be substantial enough to induce provider behavior change. P4O should focus on outcomes that are amenable to quality improvement efforts. P4O outcome standards should be empirically derived based on performance levels that are being achieved by the best-performing health care organizations 32
Payment Adjustment Principles (cont) P4O should not mandate the specific care processes that hospitals use to achieve the P4O outcome standards. P4O financial rewards and penalties should be determined based on a hospital s overall relative outcome performance and applied as an overall hospital payment adjustment, rather than as a patient-specific payment adjustment. The determination of the relative outcome performance of a provider must be risk adjusted to take into account patient severity of illness. P4O methodologies must be transparent, clinically precise, and comprehensive, with a uniform and consistent structure. 33