Cemil Browne Principal Consultant 20 May 2010 The Initiate EMPI Proposition Information Management
The Fundamental Problem In the healthcare ecosystem, data about the patient and provider is spread across multiple systems in an inconsistent and fragmented way Interoperability is hard to achieve Data will always live in multiple distributed systems Poor availability, quality, and trust in a common key to find or retrieve all the records associated with an entity Data quality problems will exist throughout all systems
Challenges to Achieving HIE Providers don t have access to complete Information.across continuum of care, impacts decision making Family History Exam Records Existing systems are not connected manually re-entering patient demographic and services information -- time consuming & error prone Costs continue to increase implementing new systems, but not achieving goals Limited information sharing across Boundaries cannot leverage patient or provider indexes Patient Symptoms Treatment Records Patient Healthcare Data Prescription History Ambulatory Admission History Ambulatory Data Providers are not recognised consistently cannot notify them in a timely & consistent manner or verify current credentials ensuring patient safety 3
What Initiate does Accurately find, link, index, maintain and provide secure access to patient and healthcare provider records within and across sources to promote collaboration and provide the foundation for electronic health records We enable quality of care and collaborative initiatives among local, regional and national healthcare organisations Without forcing you to do lots of programming, move large quantities of data, standardise database management systems or manage patient/provider identifiers Single solution for Patient, Healthcare Provider and Service Location identification
Links data across disparate source systems Patient / Provider Data Sources Source 1 Source 2 Source 3 Source n Name Local ID IHI DOB Sex Address 1 Postcode Home phone Mobile Dr Kath J. Jones 1:N4456 763543 15/06/1970 32 Sussex St 2000 92634622 0415266721 Dr Kate Lamb 2:2736 763543 Female 2000 02-9263-4622 Mrs. K. Jones 3:S7846 763543 15/06/1970 Level 1, 32 Sussex Rd 2000 9263-4622 +61415266721 Catherine Lamb n:97662 763543 15/06/2006 Female 92630-6000 0415-266-721 Dr Kath J. Jones 763543 15/06/1970 Female Level 1, 32 Sussex St 2000 02 9263 4622 0415 266 721 Consuming Systems System 1 System 2 System 3 System n
Accurate patient matching and linking Linking records electronically is one of the greatest challenges facing healthcare Healthcare organisations currently don t do a stellar job of it within their own systems Duplicate Medical Record Rates in single facility MPI: 8-20% The larger the database, the higher the error rate, often as high as 30-40% Over 30% of all Master Person Index (MPI) records have an invalid or blank value in Name (first/last), Date of Birth or Gender Jumps to over 60% if middle name counted Over 80% of all confirmed duplicate records have a data discrepancy in one or more key patient identifying fields i.e., Name, Date of Birth, Medicare # or Gender Nearly 40% of all duplicate records have a discrepancy in the first or last name Valid Medicare Number captured 50% of the time, sometimes worse 6
Patient identification it s not easy Misidentification is a widespread problem Many causes Patient Registration staff Technology Workflow Unique healthcare identifier is NOT THE SOLE ANSWER but rather another attribute than can be used to enhance matching A B C D Example Oz-Linkages Oz-Dups G-Canada Master Person Index (MPI) Size 2.8M 1.5M 790K 1.2M 4.4M 800K 1.7M Error Rate % 15 10 13 23 46 6 8 H-Canada 1.3M 5 7
National Patient Identifier & Client Registry: Not Mutually Exclusive National Patient Identifier Requires launch by government agency or organisation Backporting to existing records expensive and perhaps impossible May heighten consumer privacy & confidentiality concerns One (of many) data elements for patient ID Not silver bullet-- will have data quality errors just like existing data Compatible with EMPI technology to manage evolving strategy Client Registry/Federated Views national identifier as just another piece of data to facilitate patient matching Manages current environment with no identifier as well as potential future identifier Data maintained within firewalls of source system Readily deployed in short timeframe with standards, retrospective or prospective Requires EMPI technology 8 National Patient Identifier and Client Registry/Federated approaches are Complimentary and help advance patient matching, interoperability and EHR initiatives...in a collaborative, timely manner
The challenge facing the HI Adoption Today Medicare maintains a central index of 20m+ plus Medicare number records for Australia Downstream - the number of records increases continually due to births, unknowns, errors, inability to find existing records, etc. The demand for use of the HI numbers will increase significantly What is the expected adoption rate for HI numbers? To support a true interoperable national health record jurisdictions will need to achieve near complete HI number coverage across their systems Approx 95% is required before jurisdiction data can be migrated to the national data network as a safe foundation for a national shared EHR (sehr)
The challenge facing the HI Central Registry Adoption Every existing record at the jurisdiction, GP, community health level will have to be linked to a validated HI number and checked for incomplete data fields before jurisdictions, et al will accept it to be uploaded into their clinical systems Many records will be rejected as duplicates, untraceable or invalid Hospitals admit to 8-15% duplication it usually be higher and some have as many as 17 unique identifiers in use today The average time to resolve a duplicate record is approx 1 hour. Estimates for resolving current duplicates alone cost $$$$$$$$ Matching, linking and data quality is an ongoing issue, not a one-time project Existing technology at jurisdictions is many years old and lacks the algorithms, applications and performance to solve this problem. Robust search capability at point of service MUST provide <1% error in creating of duplicate records
Initiate s accuracy dividends for the HI Initiate s accuracy delivers superior matching, enabling higher auto-merge of records and reducing manual intervention Saves money Saves time (NHS CR estimated saving of up to 250 man years) Act as a staging platform for data to be transmitted to / or from the HI registry HI data stewards would thus deal only with the most difficult matches making best use of their specialist skills Initiate s accuracy would speed up migration of local data within the jurisdiction, and to / from the national HI service and Jurisdictional HI registries Accurately matching records will result in: Higher levels of confidence amongst the public and clinicians, and the use in the Electronic Health Record / ICT projects
Interoperability that protects investment The design for the HI service must take into consideration the need to protect the investment in existing and new services This is NOT about replacing the existing Medicare Repositories Jurisdictions need to consider supplementing the HI Service with core EMPI functionality with a proven track record in healthcare The HI functionality must provide better match, search and index capability, together with a framework for managing issues such as data quality
CASE STUDY: PROVINCE OF ONTARIO 13
Case Study: Province of Ontario Hospitals: 100+ Physicians: 20K Records: 49M 62 data sources 250K Transactions/day Moving to SOA, web services, and HL7 integration Reducing wait times for key diagnoses and procedures key driver to EMPI initiative Accurate patient ID and linking of data required in order to support defining available medical personnel for key clinical conditions and procedures Within months benefits were being seen 81 hospitals, and over 1,700 surgeons are using the WTIS daily to help over 250,000 patients Clinician and clinic managers no longer working in the dark.. Fourteen months ago we had no idea about how long patients were waiting for cancer surgery in the province Multi-phased, quick deployment Phase One: Six (6) sources in 2 months, March 2006 Phase Two: 49 sources in 6 months, October 2006 Phase Three: 17 sources, June 2007
CASE STUDY: PROVINCE OF SASKATCHEWAN 15
Case Study: Province of Saskatchewan Health Authorities: 13 Physicians: 1,300 Records: 4M+ Sources: 13 regional and provincial including 70 source systems Transactions: 10K/day Integration: Uses Initiate s Bridge JAVA API with Enovation and WinCis Moving to SOA, web services, and HL7 integration Goal was to enable provincial EHR system per Infoway blueprints Infoway approved funding since in compliance with Client Registry blueprint and standards Supports PACS, Lab Repository, and pharmacy information program Creates common view of patient information across province Data linking achieved and patient care enhanced Conducted Initiate JumpStart SM to define existing data quality and data stewardship requirements Key in facilitating provincial level discussion regarding data quality and data governance and tie to patient safety Objective data quality assessment enabled solid, goal oriented business discussion Ongoing data quality initiatives supported at multiple levels through Initiate software maintaining duplication rate less than 1%
CASE STUDY: ACT HEALTH 17
ACT Health Early Adopter Project: Some of the challenges Consent Issues Patient Registration Staff issues Record consent in Index Verified HI Could contain duplicates Unverified HI When a match is not found but sufficient demographics are available Potential to generate duplicates if search is not exhaustive...or too specific (address) Provisional HI May need to be resolved/linked to verified ID at a later stage
Use as a Primary or Supplementary Identifier? Diagram source: NEHTA As a Primary Identifier Means full adoption of the IHI across the environment Has system impacts for all systems As a Supplementary Identifier Reduces the impact on existing systems Would be implemented as a cross reference within the PMI Registry
Identification put in perspective for NEHTA HI Medicare CDMS currently has 26M records 20M active (6M deceased/inactive) Largest State: NSW - Population 7.111 million* Records in NSW public hospital PMIs: ~ 22 million Average number of applications per Public Hospital with PMI data: 8 + EG. patient administration, pharmacy, radiology, pathology, emergency, operating theatres, diet management, UPI Total health records: > 176M Australia s population 22.344 million Extrapolation 526+ million public health records in Australia What is the transition strategy for linking existing healthcare records with NEHTA s new HI and HPI? How long will it take to adopt the HI? Can we wait 10+years? What are the rip and replace and/or modification costs?
Something to think about.. On a business trip, you wake up with strange, painful symptoms in the middle of the night. You take a taxi to the hospital emergency room where doctors try to help you. They need to know your medical history. And you don t know, or can t remember, or never knew the details. Although your airline ticket confirmation number, your car hire record, and even your mobile phone bills and calling history are available 24/7 on-line, your medical records are locked away in filing cabinets somewhere, partially hand-written and partially typed, stored in paper folders, and stacked alphabetically. At four in the morning, that person with the key to your medical information is fast asleep and, in this case, a thousand miles away. How would you reach him or her? Would you call your doctor s answering service and hope someone will go down to the office? Perhaps it can wait until morning but wouldn t it be better for the doctor treating you to have that information now? Meanwhile, that emergency room doctor is asking you to remember as much of your history as you can while your stomach is in a knot or your head pounds or the pain in your chest begins to creep into your jaw and down your arm.
Final Thoughts NEHTA is one piece of a larger health identification puzzle NEHTA HI s should be one component in identifying patients within current systems National Initiatives require local or regional support Improving local data quality will greatly assist national information sharing Replacing local or regional systems wholesale is unnecessary, risky and may not improve quality by itself Data quality initiatives must take into account front-end staff in order to achieve the best possible outcome Anything that makes data entry at the front-line more difficult must be carefully considered 22