Session #2 Optimizing e-health using Clinical Decision Support: Real World Examples 2015 Minnesota e-health Summit 1:15 2:30 pm Tuesday, June 16, 2015 1
Today s Agenda Hear about two effective clinical decision support and optimizing examples. Understand how adoption of EHRs and other health information technologies helps improve clinical care and population health. Learn from a successful regional implementation of an EHR linked web-based clinical decision support system at multiple delivery systems and Minnesota Immunization Information Connection s (MIIC) utilization of immunization history and forecasting for clinical decision support. Paul Kleeberg, Stratis Health (Moderator) Aaron Bieringer, Minnesota Department of Health Gerald H. Amundson, HealthPartners Institute for Education and Research Patrick J. O Connor, HealthPartners Institute for Education and Research 2
Aaron Bieringer Minnesota Department of Health 3
OPTIMIZING E-HEALTH USING CLINICAL DECISION SUPPORT Aaron Bieringer MIIC Interoperability Coordinator Minnesota Department of Health June 4 th, 2015 Real World Examples
5 Overview What is MIIC? Where do we get our data from? What do we offer? How are people using our CDS? Next steps?
What is MIIC? 6 Minnesota Immunization Information Connection (MIIC) is our statewide immunization information system (IIS) or immunization registry Most complete source of immunization data in the state One of the more well-established registries in the nation: In existence for over 12 years One of six IIS sentinel sites
What is MIIC? 7 Immunization Data Sharing Statute: Data sharing allowed under MN Statute 144.3351 Currently no mandates for use or reporting Widely accepted and used by providers Holds data for almost all children and most adults 7.5 million clients and 72 million+ vaccinations Highly utilized by thousands of users
Where do we get our data from? 8 100% Percent of Known Providers Participating in MIIC 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% participation by provider type Pharmacy Specialty- OB/GYN Public Health Primary Care Clinic Hospital
9 Where do we get our data from? 9 Out of 4,734 active organizations in MIIC... 1,200 1,000 800 600 400 200 0 Primary Care Clinic School/School-Based Clinic Childcare, Headstart, Preschool Pharmacy Nursing Home, Long Term Care, Home Care Specialty Provider Hospital Public Health College/University Health Plan Other
Where do we get our data from? 10 Out of 6,823 active MIIC users... 13% 9% 12% 18% Reports Only School User Typical User Administrator Other 48%
11 What do we offer? Clinical decision support for immunizations (CDSi) Comprehensive vaccine history Immunization recommendations based on current ACIP guidelines Exchange data in 10 different electronic formats Vaccine ordering Inventory management Reporting functions Vaccination rate assessment Vaccine usage Improbable shots Reminder/Recall functions Track and report on lists of clients
12 What do we offer? Immunization history screen
13 What do we offer? Vaccine forecasting
What do we offer? 14 Clinical decision support for immunizations (CDSi): 40+ childhood, adolescent, and adult series Standard Advisory Committee on Immunization Practice (ACIP) recommendations MIIC staff maintained
15 What do we offer? Forecast management
16 What do we offer? Forecast management
17 What do we offer? Forecast management
18 How are people using our CDSi? Via the MIIC application Within EHRs/source systems Alternate access : XML and HTML-based access HL7-based query and response Key informant interviews: Collaborating with the U of M 10-12 active provider sites Metro and out-state Local public health (LPH), smaller clinics, and larger provider groups Getting feedback from users on awareness, use, and value Looking for suggestions for improvement Compile and compare the results for future consideration
19 How are people using our CDSi? Sample questions: How do you integrate this into your workflow? What happens when there are discrepancies in the data? Do you find the data valuable? What did you do before/what would you do without MIIC CDSi? What improvements would you find helpful?
How are people using our CDSi? 20 Interviews so far: Completed three Have two more scheduled Discussing dates for five more Notes of interest: Typically used for all groups Still logging into MIIC directly Find a lot of value in the data No CDSi from any other source
21 Next steps? Interviews: Complete Interviews Compile Results Discuss Internally Publish Results MIIC program: Discover barriers of use Increase adoption Increase the value of our CDSi
Contact information 22 Aaron Bieringer MIIC Interoperability Coordinator MIIC Operations Vaccine Preventable Disease Section 651-201-4071 aaron.bieringer@state.mn.us www.health.state.mn.us/immunize
Gerald H. Amundson Patrick J. O Connor HealthPartners Institute for Education and Research 23
Using Clinical Decision Support to Improve Diabetes Care Patrick O Connor MD MA MPH Center for Chronic Care Innovation Senior Clinical Investigator HealthPartners Institute for Education and Research 24
Disclosures: Patrick J. O Connor NIH Research grants (9 active grants) No industry funding Visiting Professor or Consultant: Peking University, University of Vienna, Mayo Clinic, Joslin Clinic, CDC, CMS, NIH, AHRQ, IHS 25
Cardiovascular (CV) Wizard Funded by a series of NIH (NHLBI) research projects Live for about 800,000 patients at 50 clinics in 2 large medical groups in ND and MN; go live by 2016 in 4 states with 4 large medical groups and 1.6 million patients Gilmer TG, O Connor PJ, Sperl-Hillen JM, Rush WA, Johnson PE, Amundson GH, Asche SE, Ekstrom HL. Cost Effectiveness of an Electronic Medical Record Based Clinical Decision Support System. Health Serv Res. 2012 Dec;47(6):2137-58. PMCID: PMC3459233. O Connor PJ, Sperl-Hillen JM, Rush WA, Johnson PE, Amundson GH, Asche SE, Ekstrom HL, Gilmer TP. Impact of Electronic Health Record Clinical Decision Support on Diabetes Care: A Randomized Trial. Ann Fam Med; 2011; 9(1) 12-21. PMCID: PMC3022040. 26
Algorithms Risk Reduction and Recommendations Lipids Blood Pressure Glucose Weight Smoking Aspirin
Description of CV Wizard REAL-TIME CLINICAL DECISICION SUPPORT (CDS) CV Wizard provides CDS for both the patient and provider through the EHR at patient encounters. It fires prompts to display and use it only for those with the greatest need (poorly controlled risk factors). PRIORITIZES RISK FACTORS NOT AT GOAL CV Wizard calculates 10-year ASCVD risk and potential absolute reversible risk (RR) for each clinical domain (lipids, BP, glycemic control, weight, tobacco use, and aspirin), and then prioritizes them. Goals are individualized and priorities dampened if already on maximal treatment. TREATMENT SUGGESTIONS CV Wizard displays treatment options for the six CV risk factors based on current therapy, most recent lab and BP values, distance from goal, comorbidities and allergies. Also displays safety alerts and prompts tests.
How CV Wizard Supports STEEEP IOM Principle SAFE TIMELY EFFECTIVE EFFICIENT EQUITABLE PATIENT CENTERED CV Wizard Clinical Decision Support Addresses over-treatment, under-treatment, and important safety issues Promotes timely attention to reversible risk factors at targeted high-risk patient encounters (many types of encounters) Consistent with the latest evidence ICSI, JNC8, 2013 ACC/AHA, USPSTF ASA, and 2015 ADA. Proven effectiveness (A1c, BP) Collects, interprets, and displays all EHR data relevant to six major CV clinical domains within a second (one click) Is standardized yet personalized to individuals (e.g. race is used for ASCVD risk calculation and BP treatment recommendations). CV Wizard use is prompted for high risk patients independent of race/ethnicity/geography. Designed to show the clinical priorities, quickly elicit patient treatment preferences, and to be used for shared decision making.
CV Wizard Risk Calculators ACC/AHA pooled risk equation to calculate 10-year and/or 30-year ASCVD risk, and can be used for people age 20-75. Used to estimate benefit for lipids, BP, tobacco. UKPDS Risk equation is used to estimate benefits of A1c reduction, floor A1c 8%. USPSTF published data to calculate benefits from aspirin (stroke risk for women and CHD risk for men). Framingham BMI equation calibrated to the ACC/AHA score to calculate benefits of weight reduction 30
Behind the scenes technology and workflow process Automated process Nurse Enters BP EMR Displays Best Practice Alert (BPA) with link to CDS Data to Web Data to EMR Web Service Assess total & reversible ASCVD risk Runs Prioritization Algorithms Runs Treatment Algorithms Runs Safety Algorithms Data to Web Data to EMR Manual process User clicks on CV Wizard navigator section EMR displays prompt with link to CDS Nurse clicks on link to CDS User clicks on link to CDS Web Site delivers CDS Browser Displays Patient and Provider Interfaces PRINT Provider Interface on Door Patient Interface to Patient
Automatically triggers for high risk patients, or manual trigger for any patient age 20-75 Double Click Click on link to open url
Provider version 33
Patient version 34
Web Display - Provider
Web Display - Patient
Provide feedback and/or report errors 37
Technical Components EMR Interface Trigger process Data extraction Web communications Enable CDS display Web Service Algorithm processing CDS content build Web Site Display CDS 38
Automated BPA Trigger Flow Epic Patient Visit BP Entered Yes BPA Criteria Evaluation: - Age 40-75 - Visit BP filed - non Hospice visit - Custom Extension evaluation Extension: Chronicles Routine Extract Pt Data Master Files Send SOAP Request,Parse Response via Cache API: Invoke^CSIWEBSERVICE Interconnect CV Wizard 2 CDS Web Service File Risks, CDS link ID via AddFlowsheetValue Web Service Interconnect Doc Flowsheet CDSLink ID Display CDS Yes BPA: Suggest Wizard display with CDS Link Click CDS Link CDS Web Site, Display CDS
Implementation considerations for other sites Implementation at Altru, Site Specific requirements Data verification Interface build Algorithm verification Treatment recommendation content Security agreements, limited system access Site analysts to carry out build, test, pilot, golive Ongoing maintenance
Summary: CDS Can Improve Diabetes Care Proven in RCT to Improve A1c and BP in Diabetes High use rates (around 75%) at targeted encounters Provider satisfaction 94-95% Results stored and populate flow sheets, dot phrases, After Visit Summaries, and other tools Efficiently elicits patient treatment preferences Powerful care coordination & case management tool Transparent, evidence-based risk estimates Uses real-time BP, avoiding re-work No data retained in web service
Outpatient CDS: Future Vision Patient-Centered (NOT Disease-Specific) CDS One integrated CDS System Identify high risk patients and provide personalized care recommendations Prioritize treatment recommendations based on QALY or DALY or What???? (benefits often small) Flexible to support the quality agenda, interests providers and affordable and efficient care
Wizard Development Serious Mental Illness (SMI) Wizard Funded Adolescent HT and obesity recognition and management Funded & Implemented Adolescent Abdominal Pain in ED Funded Prediabetes Identification & Management Funded Preventing A1c relapse using SMBG data Funding Decision Pending Personalized CA Screening and Primary Prevention Funding Decision Pending
Questions & comments? Project Team- primary contacts: Patrick O Connor, MD MPH patrick.j.oconnor@healthpartners.com JoAnn Sperl-Hillen, MD joann.m.sperlhillen@healthpartners.com Heidi Ekstrom, MA- Project Manager [GO TO PERSON] heidi.l.ekstrom@healthpartners.com 45