Government Programs: Lessons from the CMS Physician Group Practice (PGP) Demonstration Project- Emphasis on Heart Failure

Transcription

1 Government Programs: Lessons from the CMS Physician Group Practice (PGP) Demonstration Project- Emphasis on Heart Failure 3 rd National Pay for Performance Summit February 28, 2008 F. Douglas Carr, MD, MMM Medical Director, Education & System Initiatives Billings Clinic

2 Outline Intro Billings Clinic PGP mechanics Focus on our approach to Heart Failure Performance Year 1 Results Observations Questions Barriers to Remote Monitoring

3 What is Billings Clinic? Group Practice w/ 225 Physicians, 65 midlevels, 29 (~50 sub-) specialties (Allergy to Urology) 10 clinic locations 272 (220) bed tertiary hospital Manage/support 7 CAHs Committed Employees 3 rd largest employer in Montana Integrated Delivery System/ Medical Foundation Board of Directors: community-based Leadership Council (Internal Board): physician majority + senior administrators

4 Lincoln Flathead Kalispell Clinic Locations & Number of Providers Lake Billings Clinic - Cody (7 MDs, 1 PA) Sanders Billings Clinic - Columbus (2 MDs, 1 NP) Mineral Billings Clinic - Forsyth (1 PA) Missoula Billings Clinic - Heights (4 MDs, 1 PA) Missoula Billings Clinic - Miles City (10 MDs, 3 PAs) Glacier Powell Lewis & Clark Pondera Teton Service Region Locations Toole Billings Clinic - Red Lodge (4 MDs) Granite Jefferson Billings Clinic- Main (200 MD, 47 PA) Deer Ravalli Lodge Butte Silver Billings Clinic - West (8 MDs, 1 PA) Bow Bozeman OB/Gyn (6 MDs, 2 NPs, 1 PA) Beaverhead Madison Affiliate Management Services Big Timber - Pioneer Medical Center Colstrip - Colstrip Clinic Columbus - Stillwater Hospital Forsyth - Rosebud Healthcare Livingston - Livingston Healthcare Dillon Cascade Broadwater Liberty Great Falls Meagher Gallatin Bozeman Chouteau Judith Basin Livingston Park Hill Havre Musselshell Wheatland Roundup Golden Valley Sweet Yellowstone Grass Big TimberColumbus Billings Stillwater Fergus Blaine HWY 2 Lewistown Petroleum Carbon Red Lodge Phillips Big Horn Powell Sheridan Lovell Sheridan I-15 Cody Crook I-90 Greybull Bighorn Gillette Buffalo Park Treasure Worland Hot Washakie Springs Thermopolis Valley Dawson Garfield Glendive Prairie Rosebud Wibaux I-94 Fallon Custer Forsyth Miles City Baker Hardin Glasgow Colstrip Johnson Daniels Scobey McCone Roosevelt Wolf Point Powder River Campbell Sheridan Richland Sidney Carter Weston Divide Williams Williston McKenzie Golden Valley Billings Slope Dunn Stark Dickinson Hettinger Bowman Adams Lovell - North Big Horn Hospital Red Lodge - Beartooth Hospital Scobey - Daniels Memorial Hospital Counties with Affiliate or Branch Clinic Other Service Area Counties Fremont Riverton Natrona Casper I-25 December 2007

5 Montana: 147,138 square miles and 922,002 people

6 CMS PGP Objectives Encourage coordination of Part A & Part B Coordinate care for chronically ill and high cost beneficiaries in an efficient manner Decrease the growth in Medicare spending over the next 3 years Health Care Education and Research

7 10 Organizations Physician Group Practices Everett, WA Everett Clinic Marshfield, WI Marshfield Clinic Integrated Delivery Systems Springfield, MO St Johns Danville, PA-Geisinger Billings, MT-Billings Clinic St. Louis Park, MN Park Nicollet Winston-Salem, NC-Novant- Forsyth Academic & Network Org. Middletown, CT Integrated Resources for Middlesex Area (IRMA) Ann Arbor, MI - University of Michigan Bedford, NH-Dartmouth Hitchcock

8 CMS PGP Project Timeline Base Year: Calendar year 2004 Performance Year 1: April 1, March 31, 2006 Performance Year 2: April 1, March 31, 2007 Performance Year 3: April 1, March 31, 2008 Performance Year 4: April 1, 2008 March 31, 2009

9 PGP Demo Concepts Medicare Fee For Service continues as before If PGP is able to reduce the growth of Medicare spending for the cohort under its care compared to a regional comparison, CMS will share part of its savings with PGP Budget neutral project for CMS Meeting Financial Target= Gate Once Open, PGP s portion dependent on meeting Quality Measures

10 CMS PGP Beneficiary Assignment To be assigned to PGP: PGP must provide to beneficiary at least one E&M office or other OP service plurality of E&M office/op services provided in the year To be assigned to Comparison Group Beneficiary must have at least one E&M Service Cannot have had any E&M services at PGP during the current year or been assigned to the PGP previously. Must reside in a service area county. Service area counties must provide 1% of eligible beneficiaries.

11 PY-1 Billings Clinic versus Comparison Group Service Area Distribution Lincoln Flathead Glacier Pondera Toole Liberty Hill Havre Blaine Daniels Sheridan Valley Roosevelt Wolf Point Culbertson Sanders Lake Teton Chouteau Phillips 2 v 3 % McCone Richland Sidney Mineral Missoula Missoula Powell Granite Anaconda Deer Lodge Ravalli Beaverhead Dillon 16.4% of Assigned Beneficiaries in PY-1 are outside of the PGP Service area Lewis & Clark Silver Bow Jefferson Butte Madison Cascade Broadwater Meagher Gallatin Bozeman Judith Basin 1 v 7 % Park Wheatland Sweet Grass Stillwater 10 V 12 % Park Fergus Lewistown Golden Valley Yellowstone 61v 40 % 4 v 2 % Carbon 5 v 3 % Worland Hot Washakie Springs Thermopolis Fremont Petroleum Musselshell Rosebud 2 v 1 % 2 V 3 % Riverton Treasure Big Horn Garfield Hardin 1 V 16 % Sheridan 3 V 4 % Bighorn Buffalo Johnson Casper Custer Powder River Prairie 8 v 4 % Gillette Campbell Dawson 1 v 5 % Wibaux Fallon Baker Carter Crook

12 CMS PGP Demonstration Project Bonus Sharing Methodology PGP Financial Target Calculation: 1. Identify comparison group in same counties 2. Calculate rate of growth of per capita expenditures from base to performance year 3. Comparison group growth rate is applied to the PGP s base year 4. Individual risk adjustments (HCC) apply to both groups to account for case mix changes between years

13 Medicare Savings = CMS PGP Demonstration Project Bonus Sharing Methodology (Per Capita Target PGP s Performance Year Per Capita Expenditure) X (Assigned Beneficiaries) Bonus Sharing Formula 20% retained in Medicare Trust Fund 80 % eligible to PGP Percentage based on both financial and quality indicators and changes each performance year

14 PGP Project Financial Model SVNGS >2% 20% CMS Q: Quality E: Efficiency Y1 Y2 Y3 Y4 0.3 Q 0.4 Q 0.5 Q 0.5 Q 80% Performance Pay 0.7 E 0.6 E 0.5E 0.5 E

15 CMS PGP Quality Measures Outpatient Total 32 + flu and pneumonia vaccines Year 1: Diabetes Year 2: Year 1 plus HF and CAD Year 3: Year 2 plus Hypertension and colorectal and breast cancer screenings

16 Quality Measures: Diabetes % of patients in performance year with Claims Based Hgb-A1c tests LDL test Microalbumin testing or Dx/Tx for nephropathy Retinal exam by MD/OD: 1year/high risk, 2 years/low risk Chart/Hybrid Based HgbA1c 9.0% BP< 140/90 LDL<130 Complete foot exam documented Influenza vaccine Pneumoccocal vaccine

17 Quality Measures: HF % of patients in performance year with Claims Based LV-EF in same year if hospitalized for HF Chart/Hybrid Based Qualitative/quantitative LVF Assessment Visits Weight documented Visits BP documented HF Education documented in last 6 months LVSD on ß-blocker LVSD on ACEI HF/AF on Warfarin Influenza vaccine Pneumococcal vaccine

18 Quality Measures: CAD % of patients in performance year with Claims Based Lipid profile Chart/Hybrid Based LDL<130 Antiplatelet Rx Lipid Rx Hx MI, on ß-blockers last visit DM &/or LVSD on ACEI

19 Quality Measures: Preventive Care % of patients in performance year with Claims Based y.o. w/ mammogram in current or prior year Chart/Hybrid Based Screened for colon Cancer BP documented Last BP<140/90 If BP is >140/>90 documented plan of care

20 PGP Quality Thresholds: Absolute or Relative Targets The higher of 75% compliance, or the Medicare mean, or 70 th percentile of Medicare HEDIS 10 % reduction in gap between administrative baseline and 100% compliance, or

21 Base Year Data: Where does the $ go? Components of Medicare Expenditures For Billings Clinic Inpatient 40% Hospital OP 24% Part B 22% SNF 7% Home Health 3% DME 4%

22 CMS PGP Mantra Quality Measures are OUTPATIENT driven. Financial Savings are INPATIENT driven.

23 Base Year Data: What Diagnoses? HCC 80: CHF 1,945 discharges HCC 92: Heart Arrhythmias 1,898 HCC 15-19: Diabetes 1,683 HCC 108: COPD 1,887 HCC 79: Card-Resp Failure 1,305 HCC 105: Vascular Disease 969 HCC 131: Renal Failure 743

24 CMS PGP Patient Focus Areas METABOLIC DISEASE: DIABETES HTN CARDIOLOGY: HF CAD Needless Admissions: PREVENTIVE CARE: 5 Wishes Nursing Home Psychiatry Medication Reconciliation Colorectal & Breast Cancer Screening Immunizations

25 EMR: Vehicle for Process Improvement Chronic Disease Management Disease Registry Disease Management Modules (DM, HF, CAD, HTN) Clinical Guidelines Reports for organization and providers Patient information Health Maintenance Cancer, Osteoporosis, etc. Screening Immunizations Patient Safety Medication reconciliation during transitions of care Health Care Education and Research

26 Heart Failure Cost Savings Goal: Decrease All Cause Admissions 10-20% 50% Existing HF Clinic: patients Technology leverage Quality Improvements HF Clinic Re-design: Expand RNs, daily IP/OP coverage, Mid-level Adopt new Treatment Guidelines HF Management POC Modules in Cerner Feedback to providers with patient report cards HF Registry build > 3,000 Effective Patient Education CMS Quality Measures: Ejection Fraction Test Blood Pressure Screen Patient Education ACE/ARB Therapy Beta Blocker Therapy Smoking Status Health Care Education and Research

27 Heart Failure Disease Module: Alerts for Providers

28 E-form for Heart Failure Registry and Tel- Assurance Program Enrollment

29 E-form for documenting Smoking Status and Education

30 E-form for documenting Heart Failure Education

31 Heart Failure Point-of-Care Summary Screen

32 : HF Program Prior to PGP Outpatient tele-management program (POTs), within cardiac rehab HF inpatient pathway provider referral Physician-directed Nurse management 2.6 FTEs (RNs); M-F patients: discharged at 1 year if stable Reduced HF hospitalizations by about 10-20%? PGP/CMS-RTI stats 2004: 1800 HF patients, 1900 admissions How to have a greater impact?

33 Interactive Telephone System with Web-enabled Data Tracking Evanston Northwestern Hospital by Randy Williams, MD in late 1990 s Utilizes daily monitoring system for patients via Interactive Voice data collection Validated, proven system that manages by exception Allows for 1 RN to follow patients Has demonstrated reduction of all-cause hospitalization some hospitals

34 Interactive Telephone System with Web-enabled Data Tracking Patients call daily between 4 AM and Noon Data appears immediately on a web server HF Care Coaches (RNs) call outliers Manage per HF protocols (diuretic ) Refer to HF Clinic MD/NPP or PCP Goal: coordinate care w/ Tx Physician

35 Recorded Questions 1. Have you noticed more swelling in the last day? 2. Did you wake up short of breath last night? 3. Did you sleep in a chair or prop up with pillows more than usual last night? 4. Have you had any lightheadedness or dizziness in the last day? 5. Please enter this morning s weight. Have you felt more sort of breath in the last day?

36 The Challenges of Patient Adherence Diet Noncompliance 24% Rx Noncompliance 24% 16% Inappropriate Rx 19% Failure to Seek Care 17% Other Vinson J Am Geriatr Soc 1990;38:1290-5

37 Existing Approach to Chronic Population Care Management # of At Risk Individuals Unmet Need and Unmanaged Risk Care Mgmt Targets Top 5-10% of population Disease Severity / Future Financial Risk

38 The Target Population # of At Risk Individuals Telephonic Enhanced Adherence Monitoring Pharos Target Population Disease Severity / Future Financial Risk

39 Current model of HF Program 6 RNs providing 7day/week coverage for IP education/enrollment and OP care management. IP Care: Focused on ADHF patients (Case ID: referral, registry, admit Dx, BNP, etc.) Core Measures documentation assisted by new Cerner Powerforms Discharge planning: Euvolemia & Early follow-up visit (5 days) Opportunity for enrollment for other patients with HF OP care: RN triage and intervention (pre-approved protocols for diuretics/electrolyte management) ~500 patients w/ HF use TelAssurance Daily telephone call in, IVR system 5 questions + weight variances precipitates RN follow-up Minimum # (~50) called at least monthly, unable to use TA Midlevel : Available for post-hospital, emergent, and Rx Titration per physician discretion

40 Billings Clinic Service Area Lincoln Sanders Flathead Lake Glacier Pondera Teton Toole Liberty Hill Havre Chouteau Blaine Phillips Daniels Sheridan Valley Roosevelt Wolf Point Culbertson Glasgow Richland Sidney McCone Mineral Primary Missoula Secondary Tertiary Missoula Powell Granite Anaconda Deer Lodge Ravalli Beaverhead Dillon Population Sq. Miles Yellowstone 136,029 2,635 Secondary 54,006 26,101 Tertiary 341,927 92,630 TOTAL: 531, ,366 Lewis & Clark Silver Bow Jefferson Butte Madison Cascade Broadwater Meagher Gallatin Bozeman Judith Basin Livingston Park Wheatland Sweet Grass Stillwater Park Fergus Lewistown Golden Valley Big Timber Columbus Carbon Cody Red Lodge Powell Worland Hot Washakie Springs Thermopolis Fremont Petroleum Musselshell Roundup Yellowstone Billings Lovell Greybull Big Horn Bighorn Treasure Garfield Hardin Rosebud Forsyth Sheridan Sheridan Buffalo Johnson Custer Miles City Powder River Prairie Gillette Dawson Glendive Campbell Wibaux Fallon Baker Carter Crook Source: Claritas, Inc Population Estimate US Census Bureau Land Area Sizes Riverton Casper

41 All Cause Hospitalizations in HF Patients - Per 1000 (using hospital and clinic for determining HF Population (HCC 80)) Residents of Yellowstone and Contiguous Counties only Mar-04 Jun-04 Sep-04 Dec-04 Mar-05 Jun-05 Sep-05 Dec-05 Mar-06 Jun-06 Sep-06 Dec-06 Mar-07 Jun-07 Sep-07 Dec-07 Combined Intervention Non-intervention

42 Heart Failure Primary Dx Hospitalizations Per 1000 (using hospital and clinic data for determining HF Population (HCC 80)) Mar-04 Jun-04 Sep-04 Dec-04 Mar-05 Jun-05 Sep-05 Dec-05 M ar-06 Jun-06 Sep-06 Dec-06 M ar-07 Jun-07 Sep-07 Dec-07 Combined Intervention Non-intervention

43 >50% reduction in hospitalizations or ~ 6/100/month enrolled in TA Averted Admissions Jan-March 06 April-June 06 Total: # 516 Medicare: ~80% Averted Admissions - All Payers July-Sept 06 Oct-Dec 06 Jan-March 07 April-June 07 Quarters July-Sept 07 Oct-Dec 07 Averted Admissions vs RTI Rate^^ Averted Admissions vs NonIntervention Group^^

44 Estimated Cost Savings to Medicare Upper Est.: $2,770,000 Lower Est.: $1,736,000 Medicare Program Savings $500, $400, $300, $200, $100, $- Averted Admissions - Medicare Program Savings April-June 07 Jan-March 06 April-June 06 July-Sept 06 Oct-Dec 06 Jan-March 07 Quarters July-Sept 07 Oct-Dec 07 Estimated program Savings vs RTI Rate Estimated program Savings vs NonIntervention Group

45 PGP PY-1 Results Press Release by CMS July 2007 (Q2PY3!) 1 result is improved Diabetes quality measures across all org, most are > national benchmarks Process Improvements were applied to all patients/payers; no restriction of services Aggregate Savings to CMS of ~$21M over ~225,000 beneficiaries 2/10 orgs (Marshfield Clinic & U Michigan) achieved >2% $7+M in bonus payments

46 Billings Clinic PY-1 Results Achieved 8/10 Diabetes targets, 20/22 total points (91%) Did not exceed the 2% threshold, thus no bonus received in 1 st year Inpatient costs reduced } vs. Comparison Outpatient costs higher } (HF and many other programmatic efforts not functional until PY-2)

47 Challenges Our Issues Cultural change management EMR enhancements: Not off-the-shelf ( certification ); Registry build PC Model: Access, Documentation, Process Redesign Medication Reconciliation Care Management Robust Advance Directives implementation CMS/Methodology Investment in resources/infrastructure (cash flow, risk) Financial bar high, demo too short ( 5 years?) Data abstraction requirements Lack of real-time data from CMS Attribution of beneficiaries Comparison group selection Risk Adjustment (HCC): coding specificity,

48 Working with CMS

49 What have we learned from playing with CMS? Delay in receiving data fails to support the improvement process Initial results reflect the delay between process improvements and clinical outcomes. Quality improvement may also not inherently be more efficient, especially in the short-term. Coding specificity will play a significant role in national programs that attempt to track quality with claims data (e.g.. PQRI)

50 Observations on HCC risk adjustment scores Base Year Variation among the 10 PGP groups: to PY-1 results were significantly influenced by the Δ between attributed and comparison groups HCC scores tend to increase 2%/year in managed care markets ; demo range: 0.5% to 6.5% Future of CMS Value Based Purchasing initiatives (P4P) likely to hinge on greater specificity of charge data Future Medical Home management fee likely will be risk adjusted MS-DRG and HCC risk drivers are similar: specificity & comprehensive, emphasis on some usual suspects

51 Statistical Challenges for Rural Counties? Risk Adjustments for rural counties, especially with beneficiary numbers <1,000, can be very volatile. (>17%) The net financial effects for Billings Clinic are greatly related to the change in Risk Adjustment. (-3% relative to comp group)

52 Observations Look more broadly at interventions HF as 1 Dx is 13% of all admissions for HF patients PGP lessons may be hidden in the details 2/10 achieved financial targets in 1st year Success in individual programs may be lost in the overall analysis After PY1, all groups are more aware that HCC risk adjustment is a CSF Premier analysis of HF is limited to a more narrow hospital/episode of care perspective

53 Observations DM by providers, not intermediaries, allows for direct intervention and best integration of care and QI into the care delivery system Medical Home: Is the infrastructure sufficient to achieve the outcomes? Not all diseases have same monetary impact or ROI/timeline HF vs. Diabetes, COPD, CAD, HTN, Cancer Prevention The Tyranny of FFS

54 .Any Questions?

55 Barriers to wide-spread adoption of remote monitoring to chronic care Payment systems FFS promotes piece-work, lack of accountability of outcomes No FFS recognition of DM or care coordination activities 3 rd Party vs. Provider-based DM Providers can better integrate DM into clinical treatment Coordination of IP/OP Greater clinical accountability for the total care of patient EHR interface/ communication with all providers

56 Barriers to wide-spread adoption of remote monitoring to chronic care Provider integration Coordination among specialties Outpatient + Inpatient EHR is the vehicle for coordination and guideline use at POC Dissemination of guidelines POC alerting/ordering/documentation Other infrastructure (human resources) required for DM Registry build and maintenance Organizational change management

57 Barriers to wide-spread adoption of remote monitoring to chronic care Provider Acceptance Professional change management (Team Process) Workflow change Accountability/feedback (report cards) Patient Acceptance Passive Activated (Wagner Chronic Care Model) Selling is not skill set in Health Care Enrollment is always less than reported 3 rd party: 10% Provider: 20% Seller : 30+%

58 Barriers to wide-spread adoption of remote monitoring to chronic care HIT Interface/interoperability Registry build and maintenance Current EHR certification doesn t include supporting accepted quality measures Quality measures are not uniform Reporting capabilities lagging Geographic Ideal for the model/technology Increases provider interaction complexity