Session 38PD, Use of Big Data to Optimize Plan Design. Moderator/Presenter: David V. Axene, FSA, CERA, FCA, MAAA

Transcription

1 Session 38PD, Use of Big Data to Optimize Plan Design Moderator/Presenter: David V. Axene, FSA, CERA, FCA, MAAA Presenters: Jordan Armstrong David V. Axene, FSA, CERA, FCA, MAAA Timothy W. Smith, ASA, MAAA SOA Antitrust Disclaimer SOA Presentation Disclaimer

2 2018 SOA Spring Health Meeting Use of Big Data To Optimize Plan Design Integrating Risk Adjustment David V. Axene, FSA, FCA, CERA, MAAA President, Axene Health Partners, LLC

3 Overview Introduction Use of personas Use of 3M CRGs Stratification of care 3 Legged Stool 2017 Axene Health Partners, LLC Slide 2

4 Introduction Partnered with SCIO on a major benefit optimization project where client was pursuing $100M reduction in health care costs over a period of 3 years SCIO provided persona and specialized healthcare analytics expertise AHP provided actuarial modeling, risk analysis and network/pharmacy optimization expertise AHP utilized 3M s CRG (Clinical Risk Groups) technology for its risk analysis End product for the client was a set of recommendations which accomplished their goal for each of their major lines of business 2017 Axene Health Partners, LLC Slide 3

5 Use of Personas SCIO defined 6 personas High risk, prefer some choice, prefer wider choice of providers Highest risk, older members who prefer no restrictions in choice Medium risk, prefer some choice, cost conscious Medium risk, price sensitive, cost impact decision making Low risk, price sensitive, younger members, minimal understanding of health care system Low risk, inefficient users, visit ER instead of PCP 2017 Axene Health Partners, LLC Slide 4

6 Use of Personas continued Key statistics by personas Personas Total Distribution 29% 24% 25% 6% 12% 3% 100% Allowed PMPM $ $ $ $ $ $ $ Admits/1, Days/1, ALOS ER Visits/1, Office Visits/1, Scripts/1, Axene Health Partners, LLC Slide 5

7 Use of 3M CRGs 3M developed Clinical Risk Groups (CRGs) as a tool to understand risk of various populations categories of patients ranging from healthy non-users to catastrophic conditions Developed by physicians grouping patients into clinically intuitive categories, followed up by cost differences Individuals are assigned to a single CRG category Not widely used at this time, but in our professional opinion is one of the best risk stratification processes in the marketplace Effective methodology for dealing with co-morbidities and disease progression Example: There are 134 CRG categories including diabetes in the description 2017 Axene Health Partners, LLC Slide 6

8 Use of 3M CRGs continued CRG Information PMPM Avg Monthly Allowed Code Description Status Severity AHP Members PMPM ,392 $ Healthy 1) Healthy ,447 $ Healthy, Non-User 1) Healthy ,626 $ Delivery with Complications without Other Significant Illness 2) One or More Significant Acute Diseases $3, Delivery without Complications without Other Significant Illness 2) One or More Significant Acute Diseases $3, Pregnancy without Delivery without Other Significant Illness 2) One or More Significant Acute Diseases $ Major Gynecological Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $2, Diagnosis of Major Neonatal, Chromosomal, or Developmental Condition without Other Significant Illness 2) One or More Significant Acute Diseases $2, Newborn without Major Diangosis without Other Significant Illness 2) One or More Significant Acute Diseases $1, Major Trauma or Infection Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Catastrophic Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $2, Malignancy Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Significant Neurological Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Significant Cardiovascular, Pulmonary or Other Vascular Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Major Mental Illness or Substance Abuse Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Significant Connective Tissue or Orthopedic Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ Significant Gastrointestinal, Hepatic, Renal, Hernia, Blood, or Other Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases ,389 $ Diabetic Diagnosis without Other Significant Illness 2) One or More Significant Acute Diseases $ History of Significant Prescription Medication NEC without Other Significant Illness 2) One or More Significant Acute Diseases ,291 $ or More Significant Acute Illnesses from Different MDCs Excluding ENT 2) One or More Significant Acute Diseases $ Significant Acute Illness - Span 90 Excluding ENT 2) One or More Significant Acute Diseases $ Significant Acute Illness Excluding ENT 2) One or More Significant Acute Diseases ,558 $ Significant Acute ENT Illness - Span 90 2) One or More Significant Acute Diseases $ Significant Acute ENT Illness 2) One or More Significant Acute Diseases $ Significant Acute Procedure 2) One or More Significant Acute Diseases ,386 $ Delivery with Complications with Other Significant Illness 2) One or More Significant Acute Diseases $5, Delivery without Complications with Other Significant Illness 2) One or More Significant Acute Diseases $4, Pregnancy without Delivery with Other Significant Illness 2) One or More Significant Acute Diseases $ Major Gynecological Diagnosis with Other Significant Illness 2) One or More Significant Acute Diseases $ Axene Health Partners, LLC Slide 7

9 Use of 3M CRGs continued Using detailed claims experience, a population is readily analyzed to assess its risk as you would with any risk adjustment model Characteristics of Health Care Costs Personas Total Risk Score % Healthy 59% 61% 60% 63% 59% 55% 60% % Acute 9% 8% 9% 8% 10% 10% 9% % Chronic 32% 31% 31% 29% 31% 36% 31% % Level 1 Severity 26% 25% 25% 23% 25% 28% 25% Claims Cost - % Healthy 10% 10% 10% 11% 13% 11% 11% Claims Cost - % Acute 16% 16% 17% 18% 15% 10% 16% Claims Cost - % Chronic 71% 71% 70% 68% 70% 76% 70% Claims Cost - % Level 1 Sev. 35% 35% 34% 35% 36% 36% 35% 2017 Axene Health Partners, LLC Slide 8

10 Stratification of Care Integrating type of care with Dartmouth Atlas categories of care Combining results from personas, with risk, and care stratification and care management opportunities 2017 Axene Health Partners, LLC Slide 9

11 3 Legged Stool For Benefit Design Success (Happy) Reimbursement and Provider Incentives Care Management Appropriate Benefit Design 2017 Axene Health Partners, LLC Slide 10

12 Provider Network & RX Optimization within Benefit Design

13 Tim Smith Leads the Pittsburgh & Washington DC practices of Axene Health Partners, since 2015 Worked at Highmark Health through the merger of Highmark BCBS and the hospital system Allegheny Health Network, Prior to Highmark, worked for Coventry from , through a time of significant growth, building their first Medical Economics department Has an MS in Information Systems, and has taught as an adjunct professor at Penn State University 2

14 Networks, RX & Personas Personas point to pricesensitive members, or those most willing to utilize lower cost networks and medications 3

15 Optimizing Networks & RX Benefits You may have a sense of which provider networks within your products are the most efficient, but do you know by how much, and the potential savings from Steering & Tiering? You may also have a sense of which medications you prefer your members to be taking (e.g. generics), but can you measure their impact on the total cost of care? 4

16 Is your data, Big Data? big da ta Noun Computing noun: big data 1.extremely large data sets that may be analyzed computationally to reveal patterns, trends, and associations, especially relating to human behavior and interactions. 2."much IT investment is going towards managing and maintaining big data" 5

17 Building Networks Finding common links between: Patients Hospitals Primary care physicians Specialists 6

18 Use of CRGs to Measure Efficiency Expected Cost / Normalizing Mix A Market Basket of Goods (Patients with CRGs) Average market basket cost across all Patients = $24. Average Cost at Walmart (low-cost network or RX) = $20 per basket Average Cost at Whole Foods (high-cost network or RX) = $30 Efficiency Score =.833 Efficiency Score =

19 8

20 National Health Expenditures by Category: Macroeconomic View Little progress shifting $ out of the hospital to less intensive settings National Health Expenditures, by Type of Expenditure, in the US, In Billions $ TOTAL $ 27.2 $ 3,031.3 Hospital Spend $ 9.0 $ as a % of total 33.1% 32.1% Physician & other clinical $ 6.0 $ as a % of total 22.1% 22.7% Retail Prescription $ 2.7 $ as a % of total 9.9% 9.8% Home/DME as a % of total 2.9% 4.3% Sub-total 68.0% 68.9% Other spend includes Admin/Insurance, Nursing/Residential, Investment, Dental Source: cdc.gov 9

21 Case Study: High-Efficiency Network Design While both utilization and unit cost play into the overall efficiency of a provider, often the price of the facility where key specialists perform heavily influences their efficiency 10

22 Case Study: High-Efficiency Network Design And the referral networks of primary care physicians to their preferred specialists tends to heavily influence their efficiency scores. PCP Efficiency Primary Ortho Referral Efficiency Primary Cardiac Referral A 1.20 Q 1.30 X 1.20 B 1.10 Q 1.30 Y 1.05 C 0.95 R 0.95 Y 1.05 D 0.85 S 0.95 Z 0.85 Efficiency 11

23 Medicare Advantage Case Study Now, unit costs are normalized, for the most part 2 networks reviewed, in the same region: Network A MLR 95% Efficiency score: 1.01 Network B MLR 78% Efficiency score: 0.99 Estimated Impact of Risk Coding: 15%! 12

24 Case Study: RX Impact on Total Cost of Care Helps to answer questions around whether a brand drug is effective enough to justify its higher cost Condition Efficiency Total Cost per Episode Expected Cost per Episode RX Costs Per Episode Condition A Brands 1.30 $13,000 $10,000 $5,000 Condition A Generics 1.05 $10,000 $9,500 $1,000 Condition A Brand Drug X Condition A Brand Drug Y 1.40 $15,400 $11,000 $5, $10,800 $9,000 $4,500 13

25 Networks & RX Benefits Other Potential Big Data Elements Financial data Where members work & live Where providers are working Provider social networks Outcomes following treatment OTC medications Competing formularies 14