Post on 10-Feb-2017
Contribution to
Business Intelligence
H Johnson
Health Atlas Ireland
Health Atlas Ireland
What do we actually do?
Support the quest for better health for
patients, their families & the public
by exploiting the
quality assurance/improvement,
health mapping & research potential
of available data.
Home page
The moving parts
Datasets
User
management
Analyses Displays
DOH – Ministers Office
HSE divisions - all
Clinical programmes
Other agencies
Researchers
How does it begin?
Scale :
National,
Regional
Local
Complexity:
Large
Medium
Small
The main streams
Mapping
AeroMedical
Service location
Service coverage
Atlas Finder
Service Directory
Resource profiler
Tracker
Analytical
Medicine (AMP)
Surgery
Radiology
Histopathology
Endoscopy
Research
In-house
Collaboration
How do we start?
3 key questions:
• What’s the problem/question?
• What data is actually available?
• What’s to be displayed (that’s actionable)?
Systematised solution = software
development (OpenApp framework)
The Atlas “clinic”
Round table - brain storm/blue sky
3 key questions:
• What’s the problem/question?
• What data is actually available?
• What’s to be displayed (that’s actionable)?
Systematised solution - OpenApp
1) Health Intelligence Team
2) The interested party
3) Software engineers
The process
Specification
Funding CMOD, HSE ….
Time frame: weeks - months
Evolutionary = mutual trust
Meet as required X 3 teams
Develop
Test
Tweak
Governance
Hand over
Access roles – levels & depth
Deploy
Implement
Monitor
Support & maintain
Enhance
Project specification
2 – 10 pages
Business case
Variables
Formulae
Visuals
Governance
Look & feel
1) Attractive & simple
2) Fast
3) Interpreted at a glance
4) Easy to “explore”
5) Signal leaps out
6) Screen & paper format
Similarities?
AeroMedical
OSI maps
Service Directory
GeoDirectory
Powerlines
Landing sites
Phone location
Tracking
NQAIS
HIPE data
National view
Hospital group view
Hospital view
Team view
Record view
Now, trend, target
Irl
Field
National
Patient
Any challenges?
Data access - export
Encryption
Definitions
Timeliness
Data quality
Governance
Unique ID, address quality (Eircode)
Culture – respect for pattern analyses, sense of control …
Implementation
What’s the real problem?
Hospital data A
Hospital data B Prescribing B
Prescribing A
Registry A
Registry B
Mortality
Survey A
Survey B
Demography
Pr care data A
Pr care data B
Service A
Service B
E-Chart A
E-Chart B
Policy A
KPI Z
HR A € B
Guideline A
Unique ID DP FOI
QA QI
Other sector
KPI A
NQAIS Acute Medicine Programme
AMP set up in 2010 by RCPI & HSE to promote:
Swifter access to senior decision makers
Swifter access to necessary investigations &
interventions for acutely ill medical patients
Reduced overnight admissions
Shorter lengths of stay
AMAU
Design concepts
Numbers
Discharges (yr) 1.6 M
Diagnostic “specialties” 20
Diagnoses 5658
CCS 261
Consultant teams 1718
Hospitals 42
Admission types 5
User levels 4
XXXXXXXXX
Complex story – needed to be told simply
NQAIS Medicine views National
Hospital group
Hospital
Team
Clinically focused Easily learnt & understood
(user friendly)
Important signals visible at a
glance
Refresh monthly
Team AMP
Health Intelligence H&W HSE
NQAIS Surgery
OpenApp
NQAIS info governance
Communication – encrypted
Devel & support OpenApp
Governance policy
Data flow
Health Atlas Ireland server
NQAIS analysis & display
HPO HIPE national dataset
Data de-identified
MRN encrypted
Hospital Data collection
Inpatient & day case
Note – equivalent OPD & ED data not
collected
Agency NQAIS M
Section H Group
User group H
Data prep
Data
analysis
Display
via
browser
Health Atlas Ireland technical
& security infrastructure
Browser
NQAIS M controller
NQAIS Access requirements
Access to the internet
IP lockdown - register IP address with Health Atlas Ireland
Browser: Complex graphics require a modern browser
Chrome
Firefox
Internet 9 or higher
(If unsuitable – options, including Atlas browser)
Security cert
Diamond plot
National My
hospital
25th centile AvLOS 75th centile
On
target
Off
target
Short
LOS
Long
LOS
White
On-target
Yellow
Close to
target Red
Off-
target
Grid
Diamond creation, AvLOS,
trim point
25th
centile
75th
centile
Trim
point
AvLOS
Trim
75%ile + 3 X (25%ile –
75%ile)
LOS
Inter quartile
range
Target setting – for each CCS
Target setting AMAU only & same
day emergency On target = 6 hr
Off-target = 9 hr
Target yellow line
= top
25th Off-target
On-target
Calculate each team’s AvLOS (BDU/# cases)
Sort rows by AvLOS – shortest AvLOS at top
Calculate cumulative # & % cases - ranks teams by vol
On-target = AvLOS of team closest to top 25th centile
(top quartile or median of top 50%). Yellow line.
Off-target” value = AvLOS of team at 50th percentile
Red line.
Cases & bed days “lost”
Cases lost & bed days lost indicate the “theoretical”
potential for increased patient throughput if the target
AvLOS was to be consistently achieved
Bed days lost = Bed days used - (# cases X target AvLOS)
Cases lost = Bed days lost
Target AvLOS
Cases/bed days lost value can help identify & prioritise
processes within the hospital most likely to reveal the
potential for AvLOS reduction
NQAIS design
Gold buttons
1) Hospital
2) Period (month/year)
3) Diagnosis (+proc)
4) Specialty
5) Age
6) Display
Silver buttons
1) Summary 1
2) Summary 2
3) Grid
4) Plots
5) Trends
Other
1) Report
2) PDF
3) Excel
Selection
box
Summary views Summary 1
Summary 2
Grid view
Key
values
Diagnoses
National
Group
Hospital
Admission type
Plot view
Expand ribbon
Key values
Enlarge
Values
Axes
Diamond plot
Histogram LOS
Cumulative LOS
Readmission
trend
Explorer Flexible interaction of
parameters of interest
Charlson Dx 2-30 risk Score
Acute myocardial infarction 5
Cancer 8
Cancer metastatic 14
Cerebral vascular accident 11
Congestive heart failure 13
Connective tissue disorder 4
Dementia 14
Diabetes 3
Diabetes complications -1
HIV 2
Liver disease 8
Liver disease - severe 18
Paraplegia 1
Peptic ulcer 9
Peripheral vascular disease 6
Pulmonary disease 4
Renal disease 10
Trends No. cases
trend AvLOS
trend
Record review
On-screen review
List options
Excel export
(beware DP)
E-MRN
Record review – MRN decryption
Nominated user - click lock icon on
Record display screen to begin
MRN decryption if:
1) Special password
2) “See identifier” role
3) Viewing a single hospital view
4) Physically present in hospital -
IP lockdown
5) State reason
All MRN decryption requests
logged for audit purposes
Comment – user feedback
Enter your comment
•What you find works well
or not so well
•Make a suggestion
•Upload a screen shot if
required
Comment is forwarded to
OpenApp for moderation &
feedback as appropriate
NQAIS – implementation
1. Identify “patterns of interest” - PLOs
2. Identify options - root cause analysis
3. Implementation plan
4. Outcome
5. Document & share learning (simply)
6. Celebrate success & apply elsewhere
7. Monitor AvLOS trends
Display tiers
Clinical team view
Hospital view
Hospital group view
National view
E-Chart
KQIs
Unusual
patterns
On tgt
Close to tgt
Off-tgt
PLOs
QA/QI
The future?
“Data unification”
Record linkage/association
A) Patient care – patient centric data
B) Research – de-identified data
Infrastructures to enable both safely & properly