Precise Disease Classification Optimizes Bundled Payments Session #99, March 7, 2018 Stuart Goldberg MD and Andrew Pecora MD
Total Page:16
File Type:pdf, Size:1020Kb
Precise Disease Classification Optimizes Bundled Payments Session #99, March 7, 2018 Stuart Goldberg MD and Andrew Pecora MD 1 Conflict of Interest Stuart L Goldberg MD Salary: Cota Inc. Ownership Interests: Cota Inc. Andrew L Pecora MD Salary: Hackensack Meridian Network Patents: Cota Inc. Ownership Interests: Cota Inc. 2 Agenda • Dr. Goldberg – Value Based Payments – Understanding Bundles (pros and pitfalls) – A better classification schema to organize data & detect variance • Dr. Pecora – Applying the schema to a large cancer practice – Building a cancer bundle program – Models of identifying variance – Early results from the program 3 Learning Objectives 1. Define the key features of the bundled payment reform model and the potential pitfalls 2. Recognize how a precise disease classification system can help health systems better understand their patient populations 3. Apply a precise disease classification system to identify and reduce care variances among similar patients treated within a bundled payment model 4. Create a data- focused team to track and better manage bundled payment programs 5. Show how data can be used to improve understanding of treatment patterns at scale 4 Suppose you were offered A bundled payment for all services relating to the care of your breast cancer patients for 1 year $100,000 Is it a good deal? What else do you need to know? 5 We all recognize the concept of precision medicine -give the right therapy to the right patient But what about precision payment? - match the payment to a more homogenous patient population If you don’t know much about your patients you cannot predict the therapy or the costs 6 Medicine is moving to Value Based Payment Models – a drive to reduce costs while maintaining quality. 7 CMS.gov website 25-01-2016 Bundled Payment Models • A single payment for all services related to a specific treatment or condition for a defined period of time • May span multiple providers in multiple settings. • Providers assume financial risk for the cost of services for a particular treatment or condition as well as costs associated with preventable complications. • Promotes decreased spending (unnecessary and necessary?) • Promotes coordination of care, finding of efficiencies • Requires data and tracking, especially regarding maintenance of quality • Must account for case mix – otherwise significant risks 8 If every patient/case was the same: the costs should be similar • – easy to figure a margin • -- easy to find areas to cut costs and increase return • -- easy to monitor when changes work or don’t work • -- easy to identify doctors or procedures that are outliers • -- easy to track outcomes (quality) •But medicine is often complex: • Bundles have moved furthest in repetitive homogenous diseases (joint replacements) but are hardest in variable diseases (cancer) 9 Would understanding our patient populations help us build better bundles? • The current ICD-10 billing codes are extremely broad • ICD-10 does not capture the specificity needed to understand complex diseases nor does it capture important co-morbidities that drive treatments and costs • Breast cancer ICD-10 is C50.x with the modifiers only capturing gender and location within the breast 10 The Grocery Uses “Barcodes” to Identify Items Accurately characterizes each item Allows tracking from “farm to table” Volumes and Inventory Quality (benchmarking) Price Links11 to other databases Cota Nodal Address (CNA): Precisely Classify for Precision Analytics 12 100 Number150 200 250 300 350 of400 Patients450 500 550 600 650 700 750 50 Pecora AL Pecora et al: J 0 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… data) (real N Clin Oncol Oncol 33, Clin 2015 (suppl; abstr e17699) 01.02.01.00… CNA “Barcodes” facilitate identification CNA identification facilitate “Barcodes” ≈ 01.02.01.00… 01.02.01.00… NJ throughout patients cancer breast 3,000 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… • • • • 01.02.01.00… Population science and Payer interest in most common types common most in interest Payer and science Population identification trial research/clinical for subtypes rare “tag” Can disequilibrium) (biologic utilized 200 approximately Only possible; cancer breast for CNA different 5000 Over 01.02.01.00… homogeneous of populations 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 13 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… 01.02.01.00… Breast Cancer Treatment Strategies by CNA: Identifying Variance • Some phenotypes are treated the same; • Others have great variance • Understanding the variance is key 14 Pecora AL et al: J Clin Oncol 33, 2015 (suppl; abstr e17699) 14 Understanding Variance is Key to Success in Participating in Value Based Payment Reform • Not all diseases are the same • Not all patients are the same • Not all doctors treat the same disease the same way • Not all doctors treat the same disease the same way each time • Not all etc. etc.…. • Understanding these differences and planning is essential to minimize variances achieve the Triple Aim (plus provider stability) 15 Complex Diseases can be part of Value Programs • At Hackensack Meridian Health we have participated in several Bundled Programs in Cancer. • The CMMI Oncology Care Model – 6-month episodes triggered by the initiation of chemotherapy. – paid traditional Medicare fee-for-service (FFS) payments. – payments are aggregated and retrospectively compared with episode- specific target prices. – If requisite quality thresholds are met, and aggregate payments fall below the target, OCM practices receive a performance-based payment (PBP). – HOWEVER minimal case mix adjustments are made (even stage of cancer) 16 CNA Guided Care Oncology Program • Hackensack Meridian Health has developed an innovative cancer bundled care program with 1 year total cost of care episodes • Pilot with Horizon Blue Cross Blue Shield of NJ in 2018 (based on pilots conducted over past two years) • Proposal approved by PTAC (CMMI) 9 to 1 – (awaiting review by Secretary of Health and Human Services) 17 Development of a precision payment VBP 18 Program Starts by Analyzing Past Performances • 1503 patients with breast cancer treated during the past 3 yrs. • Understand type of patients (oncologic case mix) • In Oncology Care Model all would be just a few ICD codes • But in our CNA based program each patient is classified based on 18 different prognostic elements (to accurately understand differences and allow analytics) 19 Distribution of “Clinically Identical” Breast Cancer Patients at Hackensack Meridian as Sorted by the CNA classification system 1503 distinct breast cancer patients 468 different CNA cohorts 33% of all patients were in the top 10 CNAs 20 Breast Cancer Treatment Bundles Designed by Our Physicians • BUNDLES – A broad treatment strategy for which a projected price will be paid – Designed based on critical disease elements that dictate therapy – Examples: • Adjuvant her2neu oncogene negative • Adjuvant her2neu oncogene positive • Adjuvant patient observation only • Metastatic her2neu oncogene negative • Metastatic her2neu oncogene positive • Metastatic immunotherapy • Metastatic patient observation only 21 • LANES – More specified treatments within a bundle strategy – Payment is at the Bundle NOT Lane level – Examples in Bundle 1 (adjuvant her2neu oncogene negative) • Hormone only • Non-anthracycline based chemotherapy • Non-anthracycline based chemotherapy with radiotherapy • Anthracycline based chemotherapy • Anthracycline based chemotherapy with radiotherapy • Only radiotherapy 22 • SUBLANES – Drug level specificity of treatment – Examples (in Bundle 1, lane 1 – adjuvant, her2neu negative, hormone only) • Tamoxifen • Anastrazole • Letrozole • Anastrazole with leuprolide • Anastrazole with goserelin • Etc. 23 24 25 26 . 27 28 29 30 31 32 33 34 35 36 Applying this Precision Classification to our Bundled Program • Variation in treatment was observed by CNA. • Patterns differing by physicians more easily identified (example: echo vs MUGA saving hundreds per case) • The median reimbursement was $122,000 (which is all ICD-10 would provide). • By splitting the population into CNA cohorts, median reimbursement for the top 10 subgroups ranged from $27,500 (early stage, postmenopausal, hormone positive, her2neu negative, etc.) to $153,000 (metastatic, postmenopausal, hormone positive, etc.). Allows more tailored discussion of bundle prices 37 Using the precise classification of each patient’s cancer: • Able to understand the types of patients seen at our cancer center • Identify costs for the different types • Identify where the variances are occurring • Quickly assess outcomes and toxicities • Build a bundled program where the provider understands the average costs, and where savings can be made • Promote a collaboration between physicians based on data to improve care and reduce costs 38 Thank you -- Questions For more information: [email protected] [email protected] Also please complete the online session evaluation 39.