ACO Benchmarks and Financial Success SOA Sponsored Research

Transcription

1 ACO Benchmarks and Financial Success SOA Sponsored Research Presented by: Rong Yi, PhD Milliman, New York City 6 th National Predictive Modeling Summit December 6, 2012

2 DISCLAIMER The research project is sponsored by the Society of Actuaries and overseen by the Project Oversight Group Results included in this presentation are preliminary and subject to change as work is still in progress Tools used in this project are for research purposes only. Milliman does not intend to benefit any third party or endorse any commercial tools. 2

3 Project Structure and Acknowledgement Authors: Rong Yi, Bill O Brien, Chun Yau SOA Project Oversight Group Louise Anderson Dewayne Ullsperger Dan Bailey Greger Vigen John Bertko Jim Whisler Mark Bethke Rina Vertes Kristi Bohn SOA: Steve Siegel, Sara Teppema, and Barbara Scott Optum provided the ETG grouper for this research 3

4 Paradigm Shift Weak economy Federal and state deficits Uninsured Healthcare reform Consumer Service expectations based on iphone/droid, not post office Paradigm shift: permanent pressure on cost 4

5 New Payment Systems that Cut Revenue Are Forcing Organizations to Look at Data Payment reform schemes all cut spending Public sector: CMMI bundled payments, Pioneer ACOs, MSSP, State programs contracting with ACOs directly Private sector: % of premium, PCMH, ACOs, bundles Not all payment reform involves providers New capitated managers: radiology benefit managers, oncology benefit managers, post-acute benefit managers, etc. What makes sense? What s the budget? What s in it? How much risk? What are critical success factors? 5

6 Example of Commercial Ambulatory Care Sensitive Admits National Well Managed National Loosely Managed Ambulatory Care Sensitive Admissions (ACSAs) (Admits/1000 Commercial) Client Congestive Heart Failure Bacterial Pneumonia COPD Urinary Infection Dehydration Diabetes Long Term Complications Adult Asthma Hypertension Angina Lower Extremity Amputation Diabetes Uncontrolled Diabetes Short Term Complication Total ACSAs/ ACSAs As Portion Of Total Non-Mat Ad 14% 15% 18% 6 Client is a Hospital Employee and Dependent Population

7 Cost of People with Diabetes - PPPM (per patient per month) 7 Service Low Utilization Regions High Utilization Regions Inpatient Facility $ $ Outpatient Facility $ $ Professional $ $ Prescription Drugs $ $ Other $37.52 $47.65 Additional Benefits $8.17 $2.89 Grand Total $1, $1, Allowed amounts before cost sharing. Adjusted to common fee levels Source: Milliman analysis of Marketscan 2010 database

8 Variation in Professional Cost Medicare Population Nationwide Avg Well Managed Benefit Total Util Per 1,000 Members Average Paid per Service Paid PMPM Total Util Per 1,000 Members Average Paid per Service Paid PMPM Inpatient Surgery Primary Surgeon 431 procs $ $ procs $ $9.35 Inpatient Surgery Asst. Surgeon 48 procs $ $ procs $ $0.39 Inpatient Anesthesia 150 procs $ $ procs $ $1.50 Outpatient Surgery 686 procs $ $ procs $ $8.37 Hosp Visits 3,501 visits $61.57 $ ,926 visits $90.60 $14.54 Office/Home Visits 7,716 visits $49.63 $ ,466 visits $63.26 $39.36 All Other Professional $ $67.85 Sources, Medicare 5% Sample from 2009, Milliman Health Cost Guidelines 8 Oct 25, Milliman, New York

9 Questions from the ACO Community What works and what don t? What segment of the population and what service components can be managed effectively to generate shared savings? How am I doing currently? Where can I be going forward? What s the benchmark and how do I compare to benchmarks? or? 9

10 SOA ACO Project s Main Objectives Establish a methodology to develop cost and utilization benchmarks within an episode of care Control for benefit design and reimbursement Control for health status Develop a financial model to estimate potential savings within episodes of care, as an ACO s care efficiency improves Bring population payment methodology and bundled payments together under one analytic structure 10

11 SOA ACO Project s Main Objectives (cont.) Develop an analytic framework to evaluate different delivery systems Hospital based vs. physician based High cost vs. low cost area Financial stability in relation to the size of the ACO 11

12 Step 1 - Selecting Episodes 10 Base ETGs Essential part of the commercial population Highly relevant to population health management Included in published literature such as PROMETHEUS High cost variation at the episode level, perhaps resulting from Medical comorbidities and health status Risk adjustment Various treatment options Repricing claims Practice pattern differences referrals, care setting, prescriptions Patient socioeconomic status, preferences, compliance Other factors 12

13 10 Base ETGs Hyperlipidemia, other Hypertension Joint degeneration, localized back Diabetes Ischemic heart disease Cerebral vascular disease Asthma Pregnancy, with delivery Congestive heart failure Chronic obstructive pulmonary disease 13

14 DRAFT Table 1 ACO Episode Risk Adjustment Study Aggregate Costs by Episode Type Entire 3 Million Population At Population Level Episode Type: Included Episodes per Thousand Member per Year Normalized Allowed Cost per Included Episode PMPM Asthma # No Complication and No Surgery $ 2.18 with Complication Only ,415.8 $ 0.95 with Surgery Only 2 $ with Complication and Surgery 3 $ Total Asthma $ Ischemic Heart Disease # No Complication and No Surgery ,290.4 $ 3.17 with Complication Only ,231.2 $ 0.39 with Surgery Only ,800.3 $ 2.35 with Complication and Surgery ,241.1 $ 1.77 Total Ischemic Heart Disease , $ 7.68

15 Variations in Cost & Utilization (draft exhibit) Aggregate Costs by Episode Type 1 Million Population with Area Factors High Cost Area Medium Cost Area Low Cost Area Episode Type: PMPM PMPM PMPM Asthma No Complication and No Surgery $ 2.29 $ 2.09 $ 2.17 with Complication Only $ 1.06 $ 0.92 $ 0.87 with Surgery Only $ $ $ with Complication and Surgery $ $ $ Total Asthma $ 3.36 $ 3.02 $ 3.03 Ischemic Heart Disease No Complication and No Surgery $ 2.91 $ 3.33 $ 3.28 with Complication Only $ 0.36 $ 0.42 $ 0.38 with Surgery Only $ 1.98 $ 2.54 $ 2.54 with Complication and Surgery $ 1.53 $ 1.86 $ 1.91 Total Ischemic Heart Disease $ 6.78 $ 8.16 $ 8.12 * Areas defined using Milliman s area factors for total cost 15

16 Table 4 SubPop Utilization per 1,000 Episode Base ETG Ischemic Heart Disease All Service Type High Medium Low Draft Exhibit Inpatient Med/Surg Admissions Other Inpatient Total Inpatient Outpatient Avoidable_ER Advanced Imaging ER Urgent OP Surgery Radiology General Pathology_FOP Therapies Other Outpatient High/Medium/Low areas are based on Milliman s area factors for total cost 16

17 Table 4 continued Table 4 SubPop Utilization per 1,000 Episode Base ETG Ischemic Heart Disease All High Medium Low Service Type Professional/OthOffice Visits 1, , ,621.2 ER Visits Consults Inpatient Surgery Outpatient Surgery Inpatient Visits Preventive Services Pathology_PROF 2, , ,562.1 Radiology 1, , ,017.2 Physical Therapy Cardiovascular 3, , ,007.4 Durable Medical Equip Home Health Other Pro/Other Total Pro/Other Draft Exhibit 17 Pharmacy Branded Drugs 3, , ,496.7 Non Branded Drugs 4, , ,442.3 Total Pharmacy 7, , ,939.0

18 Step 2 Data Standardization Exhibits in Step 1 show significant cost and utilization variation. control for the variations that we can control for: Contractual/pricing Health status Repriced claims to a uniform fee schedule Risk adjustment Applied risk adjustment to all cost and utilization outcomes, except for branded vs. generic drugs Factors include age/gender and HCCs Some cost and utilization outcomes are correlated with health status and some do not (see next slide) 18

19 Risk Adjustment Models for Cost Outcomes Base ETG 19 Highest R-Squared # Episodes in Model Dev Sample Average Episode Cost Model R Sq (%) Label Outcome Variables Pregnancy, with delivery 44,068 Professional and all other costs $5, % Pregnancy, with delivery 44,068 Inpatient facility cost $6, % Diabetes 233,263 Professional and all other costs $ % Ischemic heart disease 76,647 Professional and all other costs $1, % Pregnancy, with delivery 44,068 Outpatient facility cost $2, % Cerebral vascular disease 20,129 Professional and all other costs $1, % Ischemic heart disease 76,647 Inpatient facility cost $3, % Base ETG Lowest R-Squared # Episodes in Model Dev Sample Average Episode Cost Model R Sq (%) Label Outcome Variables Congestive heart failure 11,126 Outpatient prescription cost $ % Hyperlipidemia, Professional and all other costs 398,559 Professional and all other costs $ % Cerebral vascular disease 20,129 Outpatient prescription cost $ % Asthma 236,976 Outpatient prescription cost $ % Ischemic heart disease 76,647 Outpatient prescription cost $ % Hypertension 654,739 Outpatient prescription cost $ % Congestive heart failure 11,126 Outpatient facility cost $ %

20 Before and After Risk Adjustment Insert table 1A Ischemic Heart Disease All Population Entire 3 Million Population Mean Percentiles Raw Data (No Risk Adjustment) Service Type Measure Unit Raw th th th th th th Inpatient Med/Surg Admissions per Episodes Total Inpatient Costs Cost per Episo 3,447 11,045 27,507 56,306 Outpatient Avoidable ER Visits per Episodes Advanced Imaging per Episodes Total Outpatient Cost per Episo 1, ,406 5,622 5,052 6,178 Mean Percentiles Risk Adjusted Service Type Measure Unit Raw th th th th th th Inpatient Med/Surg Admissions per Episodes Total Inpatient Costs Cost per Episo 3, ,543 4,094 9,637 16,027 20,086 Outpatient Avoidable ER Visits per Episodes Advanced Imaging per Episodes Total Outpatient Cost per Episo 1, ,507 1,929 2,490 1,761 2,879 20

21 Before and After Risk Adjustment (cont.) Table 5B Episode Insert ranking before table and after 1Arisk adjustment Ischemic Heart Disease All Population Entire 3 Million Population Mean Percentiles Raw Data (No Risk Adjustment) Service Type Measure Unit Raw th th th th th th Professional/OthOffice Visits per Episodes Total Pro/Other Cost per Episo 1, ,337 3,503 5,342 11,310 Pharmacy Total Pharmacy Cost per Episo ,279 1,040 1,216 1,366 Total Cost per Episo 7, ,855 5,022 21,210 39,118 75,160 Mean Percentiles Risk Adjusted Service Type Measure Unit Raw th th th th th th Professional/OthOffice Visits per Episodes Total Pro/Other Cost per Episo 1, ,071 1,845 3,077 3,713 4,825 Pharmacy Total Pharmacy Cost per Episo ,004 1,090 Total Cost per Episo 7,007 2,951 4,874 8,645 16,057 22,506 28,879 21

22 Steps underway Finalize risk adjustment methodology Simulations - ACOs with different population size Develop the financial model for ACOs under different efficiency assumptions Report expected to be released in late 2012 or early

23 Some Takeaways (so far ) Risk adjustment can reduce some of the cost and utilization variations within an episode My patients are sicker cannot always be used to justify higher cost and utilization Professional cost has the highest correlation with health status industry s emphasis on standardize preventive care and population health management Prescription cost has the lowest correlation with health status Prescription patterns matters more Savings opportunity varies by episode Episodes conforming to standard or EBM seem to have less savings opportunity 23