Transforming validated clinical research into practice ...Transforming validated clinical research into practice using novel EHR integrations within EPIC and Cerner. Hi, I’m IH,
54
April 26 th , 2018 Iltifat Husain, MD Assistant Professor of Emergency Medicine Wake Forest School of Medicine Co-founder, IMPATHIQ Transforming validated clinical research into practice using novel EHR integrations within EPIC and Cerner
Iltifat Husain, MDAssistant Professor of Emergency Medicine
Wake Forest School of MedicineCo-founder, IMPATHIQ
Transforming validated clinical research into practice using novel EHR integrations within EPIC and
Cerner
Presenter
Presentation Notes
Hi, I’m IH, AP of EM at WF SOM. Co-founder of IMPATHIQ, we create health informatics tools that integrate with the EHR by utilizing the new Cerner and EPIC App Stores (who are based on FHIR). In this presentation, I’m going to show how we took a validated cardiovascular clinical decision tool and integrated it into the EHR by utilizing the new EPIC and Cerner App Stores. The focus of the talk though is going to be on how the launch of the new Cerner / EPIC App stores present a truly transformative opportunity for researchers and hospitals – mainly from a data extraction opportunity.
Acknowledgements
2
SIMON MAHLER, MDFounder
Associate Professor of Emergency Medicine
HEART Pathway Architect and Principle Investigator
SCOTT GILMORE, MD
Founder
Emergency Medicine
Software Design
Medical Informatics and Machine Learning
VINAY TANNAN, PHDFounding Partner
Medical Devices
Commercialization
Product and Business Development
Presenter
Presentation Notes
Before I start I wanted to acknowledge my colleagues at Wake Forest and co-founders at Impathiq – in particular, Simon Mahler, who is the PI and architect behind the Heart Pathway.
Overview
• Brief summary of our chest pain research– How we solved the problem of chest pain management in the Emergency Room
• Our implementation of CDS into the EHR (traditional method)– Lessons learned from going live to data extraction and the pain points
• Creating an iOS Application
• Learning about FHIR and the novel EHR App Store Model
3
Presenter
Presentation Notes
This talk will start off with a brief summary of…
Overview
• Implementation of CDS with the EHR App Store Model
• Data extraction opportunities with the new App Store Model
• Case study on the Cambridge Health Alliance Implementation
4
Learning objectives
• Understand differences between traditional EHR builds compared to the new App Store model
• Understand why FHIR alone is not a game changer, but addition of EHR App Stores is the transformative aspect
• Understand new research opportunities the EHR App Store Model presents
• Understand how to build a FHIR app and where to start
5
Presenter
Presentation Notes
There are 4 key learning objectives I want everyone to walk away knowing at the end of this talk.
EmergencyDepartments
5,025Chest Pain
Visits
8 MChest PainEvaluation
$13 B
PatientsAdmitted
> 50%Diagnosedwith ACS
< 10%
U.S. market per year; ACS = acute cardiac syndrome
MissedACS
2-4%
Chest Pain Problem
Presenter
Presentation Notes
Let’s start first with the problem we were trying to solve at Wake Forest – Chest Pain. Stems from the fact that majority of heart attacks don’t present in an “obvious manner” – you often can’t tell just by looking at the ECG that someone is having a heart attack, and sometimes the initial set of cardiac enzymes you get on a patient are negative. It is well known amongst physicians and administrators that patients who present to the Emergency Room with chest pain often receive a wide variation in treatment depending on who is caring for them. There are existing clinical decision tools, but there are two issues with them: 1) They under-triage patients, so 2 to 4% of patients who present to the Emergency Room are discharged and have bad cardiac outcomes. 2) They over-triage patients who could otherwise have been sent home -- this leads to ER crowding, hurts patient throughput times, and causes unnecessary and expensive cardiac testing in the ER patient population This was the problem we were trying to solve
7
Grant Funding
Over $2 million in funding supported the development and clinical validation of HEART Pathway
Sponsor directives to translate from research to product
T-32 Training Grant
Randomized Clinical trial
Study
Implementation Study
Multi-site implementation
study
Multi-site Essay Validation Study
Pre-hospital applications
study
Presenter
Presentation Notes
Simon Mahler, who I mentioned already, was the PI behind all of theses studies. Over a 2 year period, the Emergency Department at Wake Forest School of Medicine did a clinical study in order to create a chest pain tool that would provide a uniformity of patient care and address the issues I mentioned.
8
PUBLISHED AND VALIDATEDThe HEART Pathway randomized clinical trial: identifying emergency department patients with acute chest pain for early discharge, Circulation 2015 Link
Cost Analysis of the HEART Pathway Randomized Control Trial, AJEM Link
HEART Pathway accelerated diagnostic protocol implementation: prospective pre-post interrupted time series design and methods, JMIR 2016 Link
A multidisciplinary self-directed learning module improves knowledge of a quality improvement instrument: the HEART Pathway, J Healthcare Quality 2016 Link
Adherence to an accelerated diagnostic protocol for chest pain: secondary analysis of the HEART Pathway randomized trial, AJEM 2016 Link
Chest pain risk stratification: a comparison of the 2-hour accelerated diagnostic protocol (ADAPT) and the HEART Pathway, Critical Pathways in Cardiology 2016 Link
Use of the HEART Pathway with high sensitivity cardiac troponins: A secondary analysis, Clinical Biochemistry 2017 Link
Presenter
Presentation Notes
Form this grant funding, our group at Wake Forest was able to publish several studies – we are most known for the Circulation RCT that Simon Mahler published in 2015.
Most importantly – overall better health outcomes. We have a new manuscript we are publishing with over 8000 patients, 4000 who received care with the Heart Pathway algorithm. Yielded sig reduction in hospitalization/obs for low risk patients. MACE < 1 %, Heart Score is roughly 2%.
Transforming research into practice
• EHR integration of the tool
• iOS App
• FHIR EHR App Store integration
12
Presenter
Presentation Notes
In order to enroll more patients and publish an even larger study, we knew had to create an EHR app.
Development into EHR
• Utilized Implementation Grant
13
Presenter
Presentation Notes
We used an implementation grant to integrate Heart Pathway into the Wake Forest Baptist Health system (Main campus / Lexington / Davie Medical Center).
A patient presents to the emergency room with a chief complaint of chest pain.
A contextual alert is triggered in the EHR automatically or manually by the physician.
The physician is presented with the HEART Pathway within the EHR environment
The physician enters patient history, symptoms and ECG variables into the HEART Pathway.
Click here to run HEART Pathway
Based on the patient characteristics, consider running HEART Pathway for
clinical decision support
1 2
3 4
14
15
In seconds, HEART Pathway displays a risk assessment and clinical guidance to the physician and patient team.
HEART Pathway results are inserted into the EHR documentation.
The physician cares for the patient using best clinical judgment and HEART Pathway guidance.
5 6
7
Lessons learned from hardcoding into EHR
• Building initial iteration was surprisingly the easiest aspect
• What was difficult: – EHR Versioning – Updates, optimizations to workflow – Updates to actual protocol– Accessing data sets
16
Presenter
Presentation Notes
Initial build took hundreds of hours to do – due to significant versioning – but ended up being the easy part. EHR versioning – Frequent updates happen to Cerner and EPIC -- and when you have a tool that is hardcoded into your specific version of the EHR, you can have issues with updates that occur. This would continuously add on hours, and the maintenance for the actual pathway ended up being the bigger problem. When we wanted to have updates – that would take several hours as well – and we had to be placed in que for those updates Accessing our data sets was by far the TOUGHEST PART
Exporting Data
• The most difficult pain point – obtaining clean data sets
• Pain Points: – Data isn’t stored in a clean method– Clarity database – SQL– Significant FTE usage for Data cleaning and validation
17
Presenter
Presentation Notes
Data from EPIC is usually stored in Clarity, which is made up of either Oracle or Microsoft SQL database. Once data is put into the Clarity database, this is often the primary way that researchers and hospitals are able to mine the data their EHR is collecting. They do this by SQL – SQL is a programming language designed for managing data held in a relational database management system. So it’s not as simple as just getting data dumps – in order to get those data dumps, you have to do specific queries – and write PAGES of code. Further, time has to be spent on validation – this is for someone to go in and further confirm the data you are getting out actually is true // and then they have to troubleshoot when they run into issues.
Adherence
• In order to achieve adherence best practices reporting is critical
– Achieved > 90% adherence by performing clinical practice feedback
– Custom EPIC report writing• Dedicated FTE
18
Presenter
Presentation Notes
It doesn’t matter how great your tool is if it isn’t being used. In order to achieve high adherence numbers, we did not only education on the pathway – but more importunately, we provided provider feedback. We found out who was using the pathway and who wasn’t. This was absolutely critical and the main thing that increased our overall adherence rates. But in order to do this, we had to work with EPIC to have a custom report written for us that had to be run manually. The data we could get out of this took hours of time go to through on a weekly basis. We were dedicating FTE just for this.
Transforming research into practice
• EHR integration of the tool
• iOS App
• FHIR EHR App Store integration
19
Presenter
Presentation Notes
The next phase for us was the iOS App – we wanted to get our CDS tool into the hands of as many people as possible.
Lessons learned from creating an iOS App
• SWIFT
• Analytics tracking
• Great model for testing demand
21
Presenter
Presentation Notes
SWIFT isn’t THAT hard to learn – in order to make a basic application – you don’t need a high level coder. – but it all depends on how resource intense your application is. Especially if it doesn’t have any backend connections. Minimal maintenance once launched. Can track granular analytics – we used software called Parse. We had tens of thousands of downloads – and were getting requests by physicians to build into their EHR via the contact form we had in the app. -- But we had no interest in doing this b/c of the difficultly in doing custom builds within the EHR. As mentioned before, we found out that versioning was becoming an issue, and the maintenance would take more time than the actual build.
Transforming research into practice
• EHR integration of the tool
• iOS App
• FHIR EHR App Store integration
22
Presenter
Presentation Notes
Like I mentioned before – we didn’t have an interest in going further than our iOS app until the EPIC and Cerner App Store market place was announced – that’s when we saw there was potential to actually scale what we were doing in a way that was never possible before.
Presenter
Presentation Notes
SMART on FHIR is an open, standards-based platform for medical and health apps to integrate within the EHR. It is designed to lower the barriers for EHR systems to participate in an app ecosystem. SMART: Substitutable Medical Applications & Reusable Technologies. In 2011, Health Level 7 (HL7®) began developing a new clinical data standard called Fast Health Information Resources (FHIR®). FHIR developed its data models and API in a manner very similar to SMART: translating medical concepts into resource definitions and providing for granular data access of data through a REST-based API (think web services and website based development). In addition, FHIR provided API support for population queries and write-back capability. Smart launch is the authentication // and the FHIR is the actual read and write. -- EPIC and Cerner have embraced these concepts – and created App Stores out of them.
Provider Authentication* EHR bridges single sign-on with FHIR token* Reduces passwords user needs to remember* Speeds up the launch sequence
Read Access* Query discrete observations (labs, PMH, vitals, demographics, etc)* Incorporate into run sequence* Improves runtime and accuracy of inputs
Write Access* Create record in EHR after runtime is complete* Store results directly in the patient medical record* Log record of use
FHIR
Presenter
Presentation Notes
For the purposes of this presentation – you just need to know the following 3 things that FHIR can do.
25
Presenter
Presentation Notes
Example of how the EHR app stores work – EPIC’s app orchard store functions in a similar way. Hospitals are able to create apps individually, or download apps already available in the App Gallery.
App Store Model[Apple / Google Model]
App Orchard Store
Physician or Patient Centric
App App Store
Presenter
Presentation Notes
The App stores went live at the end of last year – they follow the Google / Apple Model.
FHIR + App Stores = Transformative
27
Presenter
Presentation Notes
Individually they aren’t a big deal – but together – they are transformative. Scalability that was never possible before. Again, it’s the “Apple” / “Google” App store model – but even better now. No longer have to do “custom” builds, and as the next slide will show – you can now get clean data sets!
EHR Read Access EHR Write Access
FHIR
Database Architecture
Physician or Patient Centric
App
Presenter
Presentation Notes
Important slide to understand here – because you have “read” access – AND write access – you then have the ability to take the data and then export it HOW YOU WANT. Remember SQL // clarity reports – you don’t need those anymore! This is the aha momment. CLEAN DATA SETS – so for example – in the past, when we were doing SQL queries, spending hundreds of hours of time cleaning data sets, spending hundreds of hours of time figuring our adherence tracking – we wouldn’t have to do that anymore.
29
After validation, EPIC APP Orchard StoreHEART PATHWAY
Presenter
Presentation Notes
When you create an app, you have to submit it to the EHR app stores – similar to how you submit an app to the Google or Apple App Stores. Discuss Validation process – and once validated, get published in their “App Gallery”
EHR Apps function “Within” the Chart
• Same iFrame
• Native experience
• Within the workflow
30
Presenter
Presentation Notes
Native experience --> but note, ability to make the experience look MUCH better. // have opportunities to make it look much better. Will show you how it looks within the workflow.
31
VALIDATED IN EMR APP STORES HEART PATHWAY
Presenter
Presentation Notes
PowerChart experience // the Hyperspace environment w/ EPIC functions the same way.
32
VALIDATED IN EMR APP STORES HEART PATHWAY
Presenter
Presentation Notes
Remember, this functions very similar to a having a web browser within your EHR – so you can attach handouts, PDFs, and other tools.
33
VALIDATED IN EMR APP STORES HEART PATHWAY
Presenter
Presentation Notes
Writeback to the chart – in this case, the clinical notes area.
IMPATHIQ SYSTEM ARCHITECTURE
34
Continuously updated and Validated Pathways and Apps
HEART PathwayTM
Sepsis
Patient Safety
Code Stroke
Checklist #1
Compliance Reporting
Custom Protocols
Actionable health informatics insights pulled from EMR
Mobile phone connectivity
One click reporting forAuditing and Safety
Accreditation compliance
Auditing and optimizing EMR pathways
Interoperable with multiple EMRs
Plug and Play (FHIR): minimal work required by Hospital’s
clinical informatics team
IQ Engine
HospitalSystem EMR
Optimization, Custom Reports,
Research Opportunities,
MachineLearning
Clinical Decision Support
IQ Engine Provides the “handshake” between EMR and Pathways
Presenter
Presentation Notes
We created a database backend called the IQ engine for all of our pathways so that we could great real time clean data sets.
Data Analytics and Insights
Presenter
Presentation Notes
Because of how tight the FHIR App Store integration is with the EHR – we are able to get clean data sets out of the heart pathway app
Cambridge Health Alliance Build
• Process
• Time to build
• Transforming validated clinical research into practice
• Adherence tracking
• Clean Data sets
36
Presenter
Presentation Notes
Here is a case study of one of the hospital systems we are working with – Cambridge Health Alliance. Process: Health systems have the ability to log into the App Gallery and then “request” an app – and they get instructions on how to download the app into their health system. Time to build: less than 20 hours Truly taking validated clinical research and turning it into practice – CHA doesn’t have a massive informatics dept, so for them to have applied this tool to their EHR and maintained the versioning and maintenance wasn’t feasible. Further, they wouldn’t have been able to spend the resources to clean the data sets and track overall adherence.
Clinical Performance Analytics
Presenter
Presentation Notes
We went live with them last month, and this is the data we’ve collected so far.
OpportunitiesWhat you can do with FHIR
38
Patient Portal: Patient Feedback / Following up with patients in trials
Targeting specific patients for enrollment in trials
Multi-Center studies (One build crosses platforms, and store in a central database)
Creating pathology specific data sets (for machine learning,
research, AI)
Patient Portal: Collect survey data or create patient centric
Scalable (custom builds for each hospital are no longer required)
Real time clinical metrics and patient data exportation, dramatically reduce time for cleaning data sets
Compared to traditional EHR builds, much more cost effective
Rapidly change and optimize clinical protocols without going through hospital informatics team
Crosses over EHRs (EPIC, Cerner, Allscripts): Perfect for Research
Perform integrations remotely and access clinical data remotely
Ability to rethink the EHR experience
40
Presenter
Presentation Notes
With FHIR you have the ability to rethink the EHR experience.
41
Sepsis Huddle App
Team based sepsis approach
Checklist driven care
Dedicated EHR sepsis management
Instant outcomes reporting and Auditing
Presenter
Presentation Notes
We wanted to create a unique Sepsis app – one that approached sepsis on a team based approach and via checklist driven care, and that could do automatic auditing / instant outcomes reporting.
42
Sepsis Huddle App
Presenter
Presentation Notes
You can reimagine the UI completely – you’re not constrained to the traditional EPIC / Cerner UIs
43
Sepsis Huddle App
Presenter
Presentation Notes
Nurse screen – you can see each others screens – critical – often you have very different screens that people are seeing. Here you have a dedicated area to manage sepsis
44
Sepsis Huddle App
Presenter
Presentation Notes
Keep track of vitals – did pt become hypotensive or not? // severe sepsis / shock, etc – can create a VIRTUAL space to manage Sepsis patients.
FHIR and Clinical Performance Analytics
Adherence Tracking
Clinical throughput tracking Incentive driven care
Rapidly optimize clinical protocols
Presenter
Presentation Notes
FHIR allows for clinical performance analytics that were not possible before.
Building a FHIR App in the EHR App Store
• Figure out your scope • Patient Facing or Physician Facing• What are the resources you will need to access from the EHR?
– Labs, vitals, patient demographics, past medical history, social history, medication lists
• What do you want to write back to the EHR? – Calculations, decision support, patient feedback
• Is a backend database required? – Tracking behavior, results, creating data sets
46
Building a FHIR App in the EHR App Store
• Basic Skills:– HTML / CSS / JS– REST API Experience– OAUTH Experience
• Advanced Skills:– Server-side language (Golang / Node.js / Python / Rust / PHP / Ruby)– MS / Linux Server Administration / Architecture Experience– Database Architecture / Modeling / Security– "Full Stack” developer
47
Presenter
Presentation Notes
Resources: Your comp sci department! Ask them if they can partner up! For basic skills – if someone can design a website or has overall programming / development experience and knows how to work with API – then they should be able to build a FHIR app (that’s what the REST API is) For Advanced skills – such as database structure, backend data collection being automated, you need someone with more skill. You would need a “Full Stack” developer. This is key buzzword that people use to describe someone who can do frontend / backend / database builds / and work on server infrastructure.
Disadvantages of App Stores
• New technology / unfamiliarity • Complete FHIR specs are not enabled at this time• Overpromises in the past hinder the present• Linking triggers in workflow • No background fetching
48
Presenter
Presentation Notes
Unfamiliarity – give example of a research team we were doing a build for recently and their CMIO didn’t realize they already had the Ignite / FHIR API. Also – lack of developers with experience in building FHIR app. Just because it’s a FHIR spec or can be mapped on FHIR, doesn’t mean it’s available on EPIC and Cerner – things are gradually being turned on. When SMART FHIR burst onto the scene more than 5 years ago, it had a ton of hype – and people thought it would solve the problem of EHR interoperability – but frankly – only with the launch of the EHR app stores have we seen EPIC / Cerner create a platform that truly does become a game changer. So when you tell people about FHIR – often times they don’t get excited – but they don’t understand how w/ the App Stores, things have finally become scalable. BPA still has to be customized to link to the FHIR App – FHIR apps can’t create their own “notifications” within the EHR= at this time (this is changing in the future). For FHIR apps to work, they have to be “on” – e.g. you can’t have an app “running in the background” using system resources or even fetching data (similar to what Apple / Android phones can do)
Data is stored for admin use to provide actionable insights.
PATIENT Saved in the EMR
Data is applied to patient care
51
Continuously updated and Validated Pathways and Apps
HEART PathwayTM
Sepsis
Patient Safety
Code Stroke
Checklist #1
Compliance reporting
Custom Protocols
Custom Pathways and apps can easily be developed for Hospitals by utilizing the iQ Engine
Presenter
Presentation Notes
IMPATHiQ’s first pathway is the HEART PATHWAY, which is built on the iQ engine. Heart pathway is the first clinically tested chest pain algorithm to be integrated with an EMR utilizing HL7 and FHIR protocols (Randomized Control Trial published in Circulation). Our Pathway was created to help manage the care of patients who experience chest pain and present to the Emergency Room. HEART Pathway is a clinical decision tool that was created at Wake Forest School of Medicine for patients who present to the ER with chest pain — our pathway reduces admissions for chest pain, while also reducing the risk associated with managing chest pain in the ER. The pathway has now been utilized on over 15,000 patients. It is well known amongst physicians and administrators that patients who present to the Emergency Room with chest pain often receive a wide variation in treatment depending on who is caring for them. There are existing clinical decision tools, but there are two issues with them – 1) They under-triage patients, so 2 to 4% of patients who present to the Emergency Room are discharged and have bad cardiac outcomes. 2) They over-triage patients who could otherwise have been sent home -- this leads to ER crowding, hurts patient throughput times, and causes unnecessary and expensive cardiac testing in the ER patient population (ER patient population consists of a large self pay population – up to 30% or more) that will result in care that will not be reimbursed for a hospital system. Further – chest pain admissions are almost always observational status visits – which generate significantly less revenue than inpatient status visits. There is also significant malpractice risk with chest pain. This is why anytime you tell a nurse or your doctor you are experiencing chest pain on the phone, they request you immediately go to the Emergency Room. Chest pain has the highest claims payouts for family physicians and emergency room physicians. Over a 2 year period, the Emergency Department at Wake Forest School of Medicine did a clinical study in order to create a chest pain tool that would provide a uniformity of patient care and address the two issues posed above. We were able to reduce the “miss rate” to less than 1%, while also reducing admissions by more than 20% and reducing hospital length of stay time by more than 12 hours per patient. These was a significant cost savings associated with these outcomes as well.
IQ Engine gives leadership new tools
IQ EngineUsage and Adherence tracking
Best practices training and compliance
Populations Health Metrics
Analytics across EHRs
Analytics across different hospital systems
Accreditation and reimbursement reporting
HOW IMPATHIQ EMPOWERS HOSPITAL SYSTEMS
53
Clinical Decision Support Pathways
IQ Engine Informatics
Clinical Decision Support Pathways
Optimizing Healthcare Administration
Meds/Tests Triage Admissions Length of Stay
Legal Risk Consistency Accreditation Financials
Evidence Based
Clinically Validated
Results Driven
EMR Interoperable
Cloud Enabled
HIPAA Compliant
Costs of App Stores
• Ranging from $5,000 to $15,000 for publishing apps
• Costs associated with data transactions
• 20% revenue from sale of apps
54
Presenter
Presentation Notes
In return for those costs though – you achieve scalability and the ability to create a backend structure for easily exporting clean data sets.