Burke Mamlin

1
AMPATH Medical Record System (AMRS)Collaborating
Toward An EMR for Developing Countries

• Burke Mamlin
• Paul Biondich

2
HIV Pandemic

• 20 Boeing 747s each day
• 10 additional Tsunamis in 2004
• 40 million living with HIV
• gt 95 in developing countries

3
Indiana University School of MedicineKenya
Program

• Established 1990
• Long-term commitment
• Exchange of ideas
• Building lasting relationships

4
(No Transcript)
5
Academic Model for thePrevention and Treatment
of HIV/AIDS(AMPATH)

• Increasing impact of HIV
• Established 2001
• Adopted MMRS
• Two clinics ? Nine clinics
• As of September 2005
• 15,528 patients enrolled (33/day)
• 7,437 adults and 338 children on ARVs

6
AMPATH Center (2004)
7
AMPATH Center (2005)
8
AMPATH Clinical Workflows

• Clinical Encounters are completely recorded on
  structured paper forms
• 80 patients per half day per clinician!
• Clinicians need basics clinical data abstracts,
  simple reminders
• Team of data entry personnel load these data into
  the database using Access forms

9
February, 2004

• Asked to be consultants
• How could existing HIV database be extended to
  more outpatient clinics?
• One week trip to Eldoret
• Immediate Observations
• No quick fix or retro-fit of existing system
• We were hooked and became very engaged after
  visiting Eldoret

10
Existing Efforts

• Multiple silos of disparate efforts

11
Existing Efforts

• Multiple silos of disparate efforts
• Non-coded data
• Flat (spreadsheet) data models
• Lack of standards
• Focus on user interface
• Minimal opportunity for collaboration or interface

12
Design Dogma

• Follow RMRS lead 30 years of practical
  experience in the design
• EAV model
• Enforce deep structure in the database to improve
  data integrity / quality
• Scale in size and connectivity
• Use of informatics standards and open
  source-based solutions where possible

13
Toward a Common Infrastructure
Data
Data
Data
Data
14
Toward a Common Infrastructure
Data
Data
Data
Data
Data
15
Toward a Common Infrastructure
Data
Data
Data
API
Data
Data
Partners in Health
16
Toward a Common Infrastructure
Data
Data
Data
Data
Data
Data
API
API
API
API
Data
Data
Partners in Health
17
AMRS Data Model v0.001
18
AMRS Data Model v0.10
19
AMRS Data Model v0.50
20
AMRS Data Model v0.52
21
AMRS Data Model v0.60
22
AMRS Data Model v0.70
23
Encounter
User
Patient
Form
Concept
Obs
Order
Business
AMRS Data Model v0.70
24

• Concept
• Dictionary of all medical concepts (terms)
  within our system..
• Concepts exist for both questions (urine ketone
  level?) and answers (large).
• Class (Drug) and Datatype (Numeric)
• Synonyms (Hgb)
• Sets (CBC)

Encounter
User
Patient
Form
Concept
Obs
Order
Business
AMRS Data Model v0.70
25
Encounter

• Patient (The Who)
• Comprehensive list of all known patient
  demographics.. matching!
• Tribe -gt Coded Equivalent of Ethnicity
• Archive of all addresses (patient_address)
• Archive of all patient names (patient_name)
• Archive of all identifying IDs MRNs, national
  IDs, etc. (patient_identifier)

User
Patient
Form
Concept
Obs
Order
Business
AMRS Data Model v0.70
26

• Observation (The What)
• All things clinically measured or observed
• Weight
• Platelet Count
• Cardiac Exam Findings
• Alcoholic Drinks / day
• Value Groups (complex answers)
• Obs Groups (complex questions)
• Complex Data?

Encounter
User
Patient
Form
Concept
Obs
Order
Business
AMRS Data Model v0.70
27
Encounter
User
Patient
Form
Concept

• Encounter (The Where)
• Tables that describe places within the
  healthcare setting, each observation or order is
  labeled with one of these.
• Encounter describes a health care provider
  lt-gt patient interaction
• Encounter Type describes various types of
  encounters (inpatient, outpatient initial visit,
  outpatient return visit, etc.)

Obs
Order
Business

• Location describes the physical location
  where observations and orders occur (clinics,
  hospitals, labs, etc.)

AMRS Data Model v0.70
28

• Order
• (The How)
• Tables that record care provider actions (Dr.
  Mamlin ordered a CXR), (Start Joe on AZT), etc
• Start vs. Completion vs. Discontinue
• Drug Orders vs. Orders still working
  through consensus..

Encounter
User
Patient
Form
Concept
Obs
Order
Business
AMRS Data Model v0.70
29
Vocabulary from Encounter Forms
30
Vocabulary from Encounter Forms
31
Creating a Form Schema
32
(No Transcript)
33
(No Transcript)
34
Architecture
FormEntry
Data
API
35
Architecture
Data
API
Java
ORM
36
Architecture

• Regenstrief building the core database layer,
  shared through GPL
• Regenstrief, Partners in Health others creating
  modules to interface with database

Partners in Health
37
Lessons

• Concept dictionary
• Stick to the practical
• Build as needed
• Fight ambiguity
• Easy / shareable tools
• Fractal of data modeling

38
Challenges

• Time and resources
• Moving targets
• Forms constantly changing
• Balancing ideal vs. pragmatic
• Choosing a platform
• Silos
• Collaboration
• Common infrastructure
• Flexibility
• Not a one-size-fits-all solution

39
Moving ForwardOpenMRS.org
Partners in Health
40
Acknowledgements

• Paul Biondich
• Bill Tierney
• Clem McDonald
• Terry Hannan
• Partners in Health
• The AMPATH Team
• World Health Organization
• Rockefeller Foundation, NIH

Partners in Health
41
AMPATH Medical Record System (AMRS)Collaborating
Toward An EMR for Developing Countries

• Burke Mamlin
• Paul Biondich

Partners in Health
42
(No Transcript)