Scottsdale Institute Fall 2009 Forum Case Study Observations
Scottsdale Institute Fall 2009 Forum Case Study: Observations and Lessons Learned from the Partners Health. Care Quality Data Warehouse/Report Central Project Jonathan S. Einbinder, MD, MPH Corporate Manager, Quality Data Management, Clinical Informatics Research & Development Partners Healthcare Presented by Aaron Abend, MBA Managing Director, Recombinant Data Corp 1 © 2009. J. Einbinder and Partners Health. Care
Fall 2009 Forum: Goals • Healthcare Reform & Achieving Meaningful Use: The Pivotal Role of Clinical Business Intelligence – Overall Reform Status – Implications for IT – Demonstrate how organizations can optimize existing data to improve care, manage populations and be paid for quality 2 © 2009. J. Einbinder and Partners Health. Care
This Presentation • • • The realities of measurement Improving data quality Determining and meeting user needs EHR versus Data Warehouse Reporting Approach 3 © 2009. J. Einbinder and Partners Health. Care
Acknowledgements • • • • Julie Greim Ben Orlowitz Mike Eskin Brian Horwitz Perry Mar Janet Cygielnik Alex Turchin Tara Flanagan Kerry Martin Venkat Macha Aaron Abend Dan Housman Jeremy Isikoff • • • Blackford Middleton Brian Hingston Eugene Breydo Qi Li Irene Galperin Eunice Jung Ruzhuo Li Anil Mangalampalli Maya Olsha-Yehiav Ed Mc. Donough Cindy Bero • And many others! 4 © 2009. J. Einbinder and Partners Health. Care
Case Study: Partners Health. Care 5 © 2009. J. Einbinder and Partners Health. Care
Partners is a large, multi-entity integrated delivery system • A federated delivery network of hospitals and clinics, plus affiliates • Information Systems centralized, but many heterogeneous systems • Teaching affiliate of Harvard Medical School 6 © 2009. J. Einbinder and Partners Health. Care
Quality Data Management (QDM) Overview • Clinical Data group within Clinical Informatics Research & Development • Manage data for reporting and analysis for clinical operations and quality. – Drivers: • • Pay-for-performance Public reporting Ad hoc data requests Certification Committee for Health Information Technology (CCHIT) Interest in registries and population management Clinical operations Initial focus on ambulatory electronic health record Increasing attention to inpatient data 7 © 2009. J. Einbinder and Partners Health. Care
The Longitudinal Medical Record (LMR) • Composite application • Mix of structured and unstructured data • Services-oriented architecture 8 © 2009. J. Einbinder and Partners Health. Care
The Realities of Measurement 9 © 2009. J. Einbinder and Partners Health. Care
Meaningful use Measurement • Meaningful use improves care • Improvement can be measured by comparing performance – Provider A to Provider B – Provider A to All Providers – Time A to Time B – Percentage improvements Competition Shows Progress Identifies problems 10 © 2009. J. Einbinder and Partners Health. Care
Selected Meaningful Use Measures Improve quality, safety, efficiency and reduce health disparities • 2011 • 2013 – % diabetics with A 1 c under control – % hypertensive patients with BP under control – % patients with LDL under control – % smokers offered smoking cessation counseling – % patients with recorded BMI – % eligible surgical patients who • received VTE prophylaxis – % orders entered directly by physicians through CPOE – Use of high-risk medications in the elderly – % patients over 50 with annual colorectal cancer screenings -- www. healthit. hhs. gov – Additional quality reports using HITenabled NQF-endorsed quality measures – Potentially preventable ED visits and hospitalizations – Inappropriate use of imaging, e. g. MRI for acute low back pain 2015 – Clinical outcome measures – Efficiency measures – Safety measures Over time, measures must improve © 2009. J. Einbinder and Partners Health. Care 11
Meaningful reporting: Identify the issues and help find a solution Summary data by physician Measurement that identifies issues Drill-down to learn the source of the issues to identify meaningful interventions that drive improvement 12 © 2009. J. Einbinder and Partners Health. Care
Diabetes Metrics (provider and practice) 13 © 2009. J. Einbinder and Partners Health. Care
14 © 2009. J. Einbinder and Partners Health. Care
Diabetes report 15 © 2009. J. Einbinder and Partners Health. Care
More complex Measurement HARDER HARDEST Outcomes Process Measures EASIEST HARDER Descriptive Less complex Curiosity Operations QI Projects Less accountability © 2009. J. Einbinder and Partners Health. Care Public Reporting/P 4 P More accountability 16
More complex Outcomes Measurement example: CHF HARDER HARDEST • List of CHF patients with ACE/ARB prescriptions and ECHO/LVEF • Percentage of CHF patients prescribed ACE/ARB with all denominator exclusions. Process Measures EASIEST HARDER Descriptive • List of CHF patients with meds Less complex Curiosity Operations • Percentage of CHF patients prescribed ACE/ARB (with some or no denominator exclusions) QI Projects Less accountability © 2009. J. Einbinder and Partners Health. Care Public Reporting/P 4 P More accountability 17
Measure example: CHF Percentage of Heart Failure Patients with LVSD who were prescribed ACE Inhibitor or ARB therapy with all denominator exclusions applied. http: //www. ama-assn. org/ama/pub/category/15777. html © 2009. J. Einbinder and Partners Health. Care 18
Measure example: CHF Prescribed Not prescribed (medical reasons) Not prescribed (patient reasons) 19 © 2009. J. Einbinder and Partners Health. Care
Improving Data Quality 20 © 2009. J. Einbinder and Partners Health. Care
Measurement High Quality Data When you start measuring, the first thing you learn is that your data must be of high quality 21 © 2009. J. Einbinder and Partners Health. Care
Data Quality Example: “My patients” • PCPs – At Brigham, panels are supplied by BWH and loaded as a data source into the quality data warehouse. – MGH panels are defined using EMPI PCP field in combination with schedule data. • Non-academic centers and non-PCPs – Patients are included on a panel if the provider has written at least two notes for them in the past two years. (Note: for practices new to LMR, a one note-one year definition is used). • Patients known to be dead are excluded. • Some reports, e. g. Medication Frequency, include any patient for whom you have entered data 22 © 2009. J. Einbinder and Partners Health. Care
My Panel Report “Panel Definition: Panel provided by BWH. Provider listed in BICS as PCP, has had at least one visit with the patient in the past three years, and patient is not known to be dead. ” 23 © 2009. J. Einbinder and Partners Health. Care
Panel Data Sources • • Internal systems Claims data from payers Feedback from providers State social security registries 24 © 2009. J. Einbinder and Partners Health. Care
EHR versus Data Warehouse Reporting 25 © 2009. J. Einbinder and Partners Health. Care
Data Needed for Typical Report 1. 2. 3. 4. 5. 6. 7. 8. Period covered, etc. , Report request Patient data – EMPI Last visit – Registration LDL values – Lab System Vital Signs – EMR Who is diabetic? Definitions from HEDIS Comparison with peers – Computed summary data sets Patient procedures - Claims 26 © 2009. J. Einbinder and Partners Health. Care
High Quality Data Warehouse • Data from different sources can be compared and assessed 27 © 2009. J. Einbinder and Partners Health. Care
Sources for data presented to user 28 © 2009. J. Einbinder and Partners Health. Care
Sources for determining what data to show 29 © 2009. J. Einbinder and Partners Health. Care
What do you have (and how are you going to get it out)? What data do you need (numerators, denominators)? EHR 1. EHR reporting utility? 2. Vendor products? 3. Data warehouse? 4. Extract, transform, and load programs 5. Analysts and developers? + Other data 1. EHR • Data quality 2. Order Entry 3. Claims, billing 4. Lab 5. Schedule 6. Panels 7. Etc. © 2009. J. Einbinder and Partners Health. Care What do you want? Measures, Reports 1. What measures and reports do you want to produce? • Frequency? • For whom? • For what purpose? • How distribute? 2. Will you need to produce physician-level measures? • Patient attribution strategy? 30
Pay physicians for completing notes on time Money tied to quality metrics • $500, $1250 or $2500 bonus available for physicians who complete 80% of their visit notes within 120 hours of the visit. (amount of money each physician qualifies for is based on the number of clinic hours) 31 © 2009. J. Einbinder and Partners Health. Care
It took about 10 months to set up this metric 1 m 2 m Design Testing New Metric 3 m 4 m 5 m Benchmarking 6 m 7 m 8 m 9 m 10 m Measurement Begins Groups Select Metrics Disputes Incentive Payment Key phases ensure physician buy-in • Choice of measurement (design/testing) • Benchmarking • Ability to dispute • Actual measurement © 2009. J. Einbinder and Partners Health. Care 32
Long Term Results Incentive program p 4 p measure established Incentive p 4 p measure program established • Incentives directly affected “time to finalization” • Increased performance targets met through incentives 33 © 2009. J. Einbinder and Partners Health. Care
QDM Data Warehouse 34 34 © 2009. J. Einbinder and Partners Health. Care
What is a data warehouse? • • Database Multiple sources - integrated Optimized for population-based queries Verified & cleansed Key idea is to separate analytic systems from source systems. 35 © 2009. J. Einbinder and Partners Health. Care
Quality Data Warehouse • Externalize information from Partners’ ambulatory electronic health record • Make that information available in a variety of ways Other sources Patients Colonoscopy Schedule Smoking data Census data Providers Point-of-care labs LMR Ad hoc queries Clinics Vital signs Lab values Billing codes Allergies Med Rec Problems Meds Notes Health Maintenance Flow Sheets Advanced Directives Reminders DNR/DNI Asthma Action Plan End of visit Reports Quality DWH Quality Dashboard Population Mgmt Disease Registries © 2009. J. Einbinder and Partners Health. Care 36
Quality Data Warehouse • Problems • Inpatient Med Rec • BWH Endoscopy • Medications • POCT Inpatient Orders (Percipio) • Narratives Glucose (BWH) • Colonoscopy • Note Headers • 400 GB • Daily inpatient census (BWH) • Vital Signs • 4. 7 • Million Asthma Action Plans • Smoking – Patients • Health DNR/DNR • 5. 2 • Million Problems inpatient, LMR, Maintenance • End-of-Visit On. Call • 5. 4 • Millions Meds • CPM LMR Reminders • Tumor Staging • 257 • Advanced Million Lab Results • EMPI Directives • Inpatient • Clinic schedules • Restricted Notes admission • BICS ADO (Headers) diagnoses • Lab results • Allergies • ECHO (for • BWH POLr inpatients) • LMR Alerts © 2009. J. Einbinder and Partners Health. Care 37
Why have a data warehouse? • Transaction data is poorly suited for analysis • Running reports and queries in Production database may compromise system performance – Production databases not designed/intended to support aggregate queries. May take many hours to run “simple” queries – Requires dedicated programmer resources, and no one programmer knows all relevant data – Hard to query across systems • e. g. Patients with diabetes and A 1 C>8 and no visit in past 12 months 38 © 2009. J. Einbinder and Partners Health. Care
Determining and meeting user needs 39 © 2009. J. Einbinder and Partners Health. Care
Ad hoc requests © 2009. J. Einbinder and Partners Health. Care 40
Ad hoc requests and queries Find out how many lots of In preparation for EMR release, count List patients 18 -28 who have each vaccine Randomized our clinic trial of centralized Provide pediatric BMI data for P 4 P to Individual ZZZ was found number of preliminary notes created not received initial Gardisil used last year (for next Department ofand Public smoking counselor nicotine Measure for documentation of body have written fraudulent at least 21 days ago, which are vaccination. As part of a QI year’s order) replacement. Health is auditing the Need counts of mass for kids. prescriptions. Find all who subject to index auto-finalize functionality. project, will call patients charts all patients that smokers and of payer to choose prescriptions written by that are eligible for Gardisil. fulfill criteria NNN. pilot sites. individual. © 2009. J. Einbinder and Partners Health. Care 41
Report Central 42 © 2009. J. Einbinder and Partners Health. Care
Partners Health. Care: Report Central • What is Report Central? – Report Central is a reporting module that allows users to display descriptive reports and quality reports. – Data for the reports comes from the Quality Data Warehouse. • Who may use Report Central? – More than 7000 users at 300 clinics have access to Report Central. – Academic Medical Centers, Community Practices – PCPs, Specialists, Adult/Pediatrics, Physicians, non. Physicians 43 © 2009. J. Einbinder and Partners Health. Care
Quality Dashboards • • • Performance Summary Comparison with peers and benchmarks Drill down to patient detail Ad hoc query/filter criteria Create/save/export patient lists 44 © 2009. J. Einbinder and Partners Health. Care
Reports can be used for different purposes • Externally focused (judgment) – Get paid, hired/fired/promoted, publicly reported – Inpatient focus, administrative data, manual abstraction – Increasing ambulatory focus (pay-for-performance) • Claims +/- lab results (and EHR) • Internally focused (learning) – Quality improvement, clinical understanding, efficiency – Inpatient and outpatient, automated and manual abstraction – More use of clinical data • Population decision support (patient care) • Research 45 © 2009. J. Einbinder and Partners Health. Care
Accessing Report Central from LMR 46 © 2009. J. Einbinder and Partners Health. Care
Report Central – Provider Menu 47 © 2009. J. Einbinder and Partners Health. Care
Prescribed medications – a descriptive report 48 © 2009. J. Einbinder and Partners Health. Care
Pediatric body mass index (BMI) 49 © 2009. J. Einbinder and Partners Health. Care
Pediatric Immunizations (2008 CDC Guidelines) 50 © 2009. J. Einbinder and Partners Health. Care
Dashboard – Quick identification of issues 51 © 2009. J. Einbinder and Partners Health. Care
52 © 2009. J. Einbinder and Partners Health. Care
53 © 2009. J. Einbinder and Partners Health. Care
Disparities of care – Diabetes • Business case: Create a race disparities report for diabetes, which can serve as a baseline view of differences sorted by patients’ race. Race disparities in chronic disease management may influence a patient's chance of receiving specific procedures and treatments. The report presents data on LDL, (Low density lipoprotein) Ophthalmology Examination, Body Mass Index, Microalbumin, and Blood Pressure. • Solution: List a breakdown of patients by key diabetes indicators by race to identify opportunities to address care disparities. 54 © 2009. J. Einbinder and Partners Health. Care
55 © 2009. J. Einbinder and Partners Health. Care
Meaningful data for providers drives adoption 56 © 2009. J. Einbinder and Partners Health. Care
Summary • Meaningful use comes from measurement • Measures used for accountability will take longer and be harder to develop than ones used for learning • Even descriptive reports may have value and are much easier to produce than more complex measures • Clear patient attribution methods are essential to acceptance of reports • Measurement requires accurate data • Accurate data comes from active use • A data resource that is verified is essential • A Quality Data Warehouse provides that data resource for Partners Health. Care Systems 57 © 2009. J. Einbinder and Partners Health. Care
Thank you! • Jonathan S. Einbinder, MD, MPH • jseinbinder@partners. org • Aaron H. Abend, MBA • aabend@recomdata. com 58 © 2009. J. Einbinder and Partners Health. Care
- Slides: 58