ZEPRS The Experience of the Zambian Electronic Medical Record System 29 May 2007 RTI International is a trade name of Research Triangle Institute 1
ZEPRS The Experience of the Zambian Electronic Medical Record System 1. 2. 3. 4. 5. 6. 7. 8. Context Objectives Inputs Outputs Impact on Patient Care Making it Open Making it Scale Questions & Comments Chawama clinic MCH waiting room, Lusaka, Zambia 2
Context § Maternal mortality rate 940/1000 § Lifetime risk of death in pregnancy 1/25 § Disability Adjusted Life Years (DALY) at birth 30. 7 § Nearly 1/3 women infected with HIV § Virtually all modern health care for 2 million women in Lusaka provided by 23 clinics and the University Teaching Hospital (UTH) § 13 of 23 clinics provide antenatal care, 9 with labor wards § 47, 000 estimated total obstetric cases (2002) Children outside Chainda Clinic, Lusaka 3
Context § All medical records on paper § Protocol and records in reasonable order, but significant problems in operation § Frequent failure to follow up patients diagnosed with STDs § Failure in some cases to test for STDs § Patients may move among clinics, but limited sharing of patient records § No central database for monitoring patient population or quality of care Patient files at University Teaching Hospital, Lusaka 4
Context Telecommunications § Weak landline infrastructure § Congested 2. 4 GHz wireless segment Clinics § Dusty, congested § 7 of 23 clinics had telephone lines § All clinics with VHF radio for voice Personnel § No computer literacy § Very stressed due to patient load § High medical staff turn-over (40%) Electrical § Inadequate electrical systems at all points, no backup power, No lightening protection § Poorly regulated grid (Frequent power outages) Clinician at the door of the Prisons Kamwala Clinic in 2002. 5
Objectives 1. Improved health of patients 1. 1 Improved access to patient records 1. 2 Improved patient record quality 1. 3 Improved patient follow-up & drug adherence 1. 4 Improved information for research and analyzing interventions 1. 5 Useful information for Zambian health administrators Patient files at University Teaching Hospital, Lusaka, 2004 6
Key Requirements & Assumptions ü Electronic system to replace perinatal “Blue Book” ü Shared, centralized database of patient records ü Wireless network connecting clinics, hospital, administrative offices, data center ü Multiple workstations in each facility ü Ability to move workstations if necessary ü Ability to print summary cards in facilities ü Ability to function through 4 hour electrical grid failure ü Operating costs paid through research grants Pre-existing VHF radio tower at clinic. 7
Inputs Money § US$ 2. 7 million (Bill & Melinda Gates Foundation through University of Alabama at Birmingham) People § UAB executive team § RTI development team § Lusaka-based technical support team § Zambian medical advisory team Time § 4 years (1 Jan 2002 – 30 Sep 2006) ZEPRS and CIDRZ team at launch of electronic referral application in Lusaka. 8
Inputs: Money $ 2. 7 Million total over four years* § § § Subcontract costs include wireless network equipment ODC includes STTA travel Includes operating costs for the period of performance * RTI International accounting system. All costs include indirect costs. 9
Inputs: Money $ 1 Million in computer and networking equipment* § § § § 134 desktop computers, 5 laptop computers 29 Printers 130 computer carts Data center servers, storage, backup, UPS units, generator One 15 m radio tower Wireless high speed backbone network Wireless base stations and subscriber units Wireless repeaters Wireless access points Connections to multiple buildings in clinic compounds UTH fiber optic and wireless network Vo. IP network equipment and handsets UPS battery backup units Lightening protection * Costs based on RTI estimates 10
Inputs: People • • UAB executive team (5) CIDRZ management and advisory team (4) Zambian medical advisory team (10+) RTI development team (4) Consultant (1) Zambian technical support staff (3) Other Zambian testers and trainers 11
Inputs: Time 12
Outputs ü Wireless network ü Patient referral system ü Perinatal patient record system ü Clinicians trained ü Clinicians using system Cellular telephone tower with ZEPRS co-located equipment 13
Outputs: Wireless Network ü ü ü ü ü Line of sight Hub and spoke 45 Mbps backbone 20 Mbps subscriber links 10 Mbps within buildings Connects 24 clinics, 1 hospital, 2 administrative offices, 1 data center Internet access for all facilities Voice communication (Vo. IP) for all facilities Data center maintained by NGO with research funding Availability at 88. 3% (26 Nov – 2 Dec 2006) ZEPRS - Zambia Electronic Perinatal Record System. kmz 27 km diameter hub and spoke line of sight network Map of Lusaka showing wireless links 14
Outputs: Wireless Network ü Co-location on cell telephone towers, commercial building, ZNBC tower ü Repeaters used to reach some sights ü No redundant links ü LEAP, RADIUS, encryption for security ZEPRS - Zambia Electronic Perinatal Record System. kmz Map showing wireless links from TELECEL 1 hub 15
Outputs: Wireless Network ü Existing VHF radio towers at most clinics used to mount wireless network equipment. ü Multiple buildings in many clinic compounds connected using wireless and wired links. Satellite photo indicating wireless link to existing Matero Main clinic VHF radio tower. 16
Outputs: Wireless Network Connecting multiple buildings at clinics ü Wired and wireless connections used to extend network to multiple buildings. 17
Outputs: Wireless Networking wards at UTH ü Fiber optic cable, copper cable, and wireless equipment connects wards at UTH 18
Outputs: Patient Referral System ü Notifies referral clinic or hospital of incoming patients ü Provides critical information to prepare for patients ü Updates referring clinic on patient status ü Maintains records of all referrals ü Used by 24 clinics and hospital successfully for 2 years ü Replaced by ZEPRS patient record system in February 2006 Clinicians launch referral system in July, 2004 19
Outputs: Patient Referral System Number of Referrals by Week 20
Outputs: Patient Record System ü Provides clinicians access to patient records 24 hours a day, seven days a week ü Alerts clinicians automatically to complete procedures, improve follow-up, and make sure critical-care issues are addressed ü Data used to coach medical teams ü Longitudinal data used to prioritize, design, monitor, and evaluate interventions 21
Outputs: Patient Record System Perinatal, VCT, PMTCT, Delivery ü Pregnancy Dating - captured at first visit and automatically updated during pregnancy with EGA/EDD ü Pregnancies linked, with critical problems from one pregnancy pre-populating for the next pregnancy ü Safe Motherhood - request labs (CD 4, RPR, HB) and view lab results, ARV counseling visits ü Ability to add problems, populate automated system problems, link problems from mother to child and vice versa ü Convert visits can be flagged as problem visits ü Role based access control - clerks have no access to medical data, only demographics ü Graphical partograph matches WHO partograph ü Infants linked for each pregnancy to mother ü Patient Referral system for easy tracking ü Data warehouse and XML data export facilitate reporting in SAS, SPSS, etc. ü Standalone mode for remote clinics with occasional connectivity - can sync records automatically with the main system. ZEPRS helps remind and coach clinicians through an extensible rules system. 22
Outputs: Clinicians Trained ü All training conducted by Zambians ü Basic computer skills training conducted by IT specialists ü Referral and patient records training conducted by senior clinicians ü 800 nurses, midwives and other health workers trained ü ü Basics skills (Windows, Word) ZEPRS mail system Referral system Patient record system Initial training for referral application. 23
Outputs: Clinicians Using System Average 584 new patients each week* * 5 Feb 2006 – 16 May 2007. Weeks include one-day weeks at beginning and end of year, resulting in 54 weeks for 2006. 24
Outputs: Clinicians Using System More than 39, 000 patients * 5 Feb 2006 – 16 May 2007. Weeks include one-day weeks at beginning and end of year, resulting in 54 weeks for 2006. 25
Outputs: Clinicians Using System 19 Clinics actively using system * Feb 2006 – 16 May 2007. Kanyama and State Lodge with only a few patients to date. 26
Outputs: Clinicians Using System 7, 657 babies delivered * Feb 2006 – 16 May 2007. 27
Major Correction Points 1 Problem Correction Congested 2. 4 GHz spectrum; 802. 11 b wireless bandwidth (4 -6 Mbps) may be inadequate for applications and Vo. IP Higher speed (40 Mbps, 20 Mbps) 5. 8 GHz wireless and Vo. IP equipment added cost of USD 300, 716 Time spent developing detailed medical record structure Discharged consultant and initiated agile development for patient record system Quality failure of South African contractor developing referral software Cancelled subcontract and completed software in-house Selection of client computer type (desktop PC, thin client, laptop PC, tablet PC) debated Industry standard desktop PCs selected for usability, maintenance, sustainability Custom-designed prototype PC carts are expensive (USD 600) and attract rats Switched to lower cost wire carts for full deployment 28
Major Correction Points 2 Problem Correction Lead software developer diverted to ART patient management system No-cost extensions Training of clinicians by IT specialists may not be effective Redesigned training to be done by senior clinicians Lightening protection for wireless equipment proves inadequate Installed best available protective devices at all points. Increased on-site stock of spares. Electrical outages at a few clinics much longer than predicted Tested off-line version of software with “opportunistic” synchronization with central database (not fully operationalized) Failure of application and database servers in close succession 2 years after hand-over Used temporary PC application server as quick fix, and then replaced failed servers 29
Major Correction Points 3 Problem Correction Rand/USD exchange rate places software development in South Africa out of reach RTI assumes responsibility for developing software in-house Extended power outages (some caused by illegal tapping of transformers) at some clinics Adjusted UPS battery backup for minimum 4 hour up-time. Set all UPS units to log grid power data. Installed automated network monitoring and alert system Limited Internet bandwidth and increased use of Internet by clinicians Controlled and limited Internet access by clinics Limited availability of clinicians for testing Enlisted medical students for testing Lusaka-based technical support staff lack capacity for efficient and effective support Strengthened capacity through close mentoring 30
Major Correction Points Prototype custom $585 rugged mobile enclosed carts Custom enclosed cart prototype, made in South Africa. Advantages § Some protection against dust § Good protection against theft Disadvantages § Expensive § Requires antenna extension § Hiding place for rodents Prototype cart being used for launch of referral System. 31
Major Correction Points Commercial < $165 rugged mobile wire carts In examination room. Being used by clinicians. Advantages § Inexpensive § Readily available § No hiding places for rodents § No antenna extension required Disadvantages § No protection against dust § Weaker protection against theft 32
Sustaining It Network availability Availability October 1, 2005 - April 30, 2006 (16 months) Availability can be higher than 90% on an extended basis. . . Availability October 1, 2005 - April 30, 2006 (7 months) Availability suffers due to management inattention and contention for resources. 33
Sustaining It Network availability Causes § Lightening § Extended power failures at some clinics § Equipment failure Other Factors § Management of support staff § Inadequate onsite spares § Aging equipment 34
Making it Open ü Software developed using open source tools and best-of-breed web architecture ü Software adapted easily to other contexts and applications ü ZEPRS software released under ASL 2 open source license ü ZEPRS documentation published under a Creative Commons Attribution-Non. Commercial -Share. Alike 2. 5 license ü ZEPRS demonstration and development websites open to all www. idg-rti. org 35
Collaborating and Combining Open Source Efforts ZEPRS § Problem-based care § Electronic referrals § Web-based forms § Developed documentation § Experience in large scale deployment § Years of experience in operation § Off-line mode § Stand-alone version (Eclipse RCP) Open. MRS § Regenstrief data model and concept dictionary § Broader development and support community General records at Kalinga Clinic prior to ZEPRS. 36
Making it Scale Expanding beyond a metropolitan area Tested ü Offline version for clinics with multiple PCs ü Automated record synchronization with centralized database Proof-of-concept ü Stand-alone packaged version (Zephyr) Concept § “Organic” scaling § Aggregate data reporting via mobile phone network § Automatic SMS reminders and notifications to patients Safe Motherhood form from Zephyr stand-alone version of ZEPRS. 37
Making it Scale “We are balancing local resources and remote resources to provide a perception of continuous connection to a central data source. When we experienced some problems with ZEPRS network availability, we attempted to bring server resources closer to the clinics by building a more distributed model – placing mini-servers in each clinic. Then we implemented a system to enable these remote servers to “phone home” when the network links were back up to fetch updated records and synchronize with the master server. ” - Chris E. Kelley, RTI Senior Software Developer 38
Making it Scale ZEPRS can operate as an easy-to-install stand-alone application on a single PC. 39
Making it Scale Technologies such as Zeroconf can enable users with minimal technical support to create a shared, networked database within a facility. 40
Making it Scale Operating in “off-line” mode, installations of ZEPRS can report aggregate data and synchronize patient records with centralized databases when telecommunication links are available. 41
Questions & Comments RTI Development Team Eileen Reynolds, Project Manager Chris E. Kelley, Senior Software Developer Niamh Darcy, Senior Technical Advisor Pablo Destefanis, Senior Networking and Telecommunications Specialist Gordon M. Cressman, Senior Project Advisor Lusaka Technical Support Staff Dennis Nkula, Zambia Project Coordinator Francis Banda, Technical Support Specialist Jamie Mwanza, Technical Support Specialist Center for Infectious Disease Research in Zambia Dr. Jeffrey Stringer, MD, Director, Co. Investigator Dr. Perry Killam, MD, Project Advisor Harmony Fusco, Project Manager Chafye Siulata, Project Assistant UAB Team Dr. Robert L. Goldenberg, MD, Principal Investigator Dr. Dwight Rouse, MD, Co-Investigator Dr. Sten Vermund, MD, Co-Invesigator Dr. Francis Nuthalpathy, Health Informatics Advisor Dr. Alan Tita, Health Informatics Advisor Zambian Medical Advisors Dr. Moses Sinkala, MD Dr. Elwyn Chomba, MD Dr. Ben Chirwa, MD Dr. Henri Phiri, MD Dr. Christopher Ngandwe, MD Dr. Chipepo Kankasa, MD Dr. Macha, MD Dr. Mpundu Makasa, MD 42
Making it Open ZEPRS Open Source technologies and components Platform Component Selected Solution Operating System Red Hat Enterprise Server Backup Arkeia Backup Wireless Authentication AEGIS Premium Server Relational Database My. SQL Application Java, Struts, JDBC, Apache DBUtils, AJAX – DWR and script. aculo. us, Quartz, Junit, XStream Web Application Server Apache (httpd) Tomcat (servlet container) E-mail Cyrus IMAP, Squirrel Mail, Sendmail, Spam Assassin E-mail Server Anti-virus AMa. VIS 43
Inputs: People 44
Inputs: People 45
Inputs: People 46
Inputs: People 47
Inputs: People 48
Скачать презентацию ZEPRS The Experience of the Zambian Electronic Medical
067be63fb0936f071a97d7a7404ff4f7.ppt