More Related Content Similar to Health IT Summit Atlanta 2014 - Keynote Presentation "Big Data, Value Analysis and Population Health Science at Mayo Clinic" (20) More from Health IT Conference – iHT2 (20) Health IT Summit Atlanta 2014 - Keynote Presentation "Big Data, Value Analysis and Population Health Science at Mayo Clinic"1. ©2014 MFMER | 3338355-1
Ryan Uitti, M.D.
Deputy Director, Kern Center for the Science of Health Care Delivery
IHT2– April 16, 2014
Big Data, Value Analysis and
Population Health Science at Mayo Clinic
6. ©2014 MFMER | 3338355-6
Use of Home Telemonitoring
in the Elderly to Prevent Readmissions
7. ©2014 MFMER | 3338355-7
Comparison:
Telemonitoring + Versus Usual Care
Telemonitoring Intervention
RN/MD team oversaw apx 100
patients and communicated
with them via phone or video-
conference if alerts arose
Daily telemonitoring sessions
(5-10 minutes) including
weekends and holidays
Collected weight, blood pressure,
blood sugar, pulse and peak flow data
Could arrange outpatient visits
8. ©2014 MFMER | 3338355-8
Results: Telemonitoring +
Versus Usual Care
Telemonitoring + Usual Care Statistics
Emergency Dept
Visits
35% 28% No difference
Hospitalization 52% 44% No difference
ED +
Hospitalization
64% 57% No difference
Note: Results are for a one-year period
9. ©2014 MFMER | 3338355-9
Results: Telemonitoring +
Versus Usual Care
Telemonitoring + Usual Care Statistics
Emergency Dept
Visits
35% 28% No difference
Hospitalization 52% 44% No difference
ED +
Hospitalization
64% 57% No difference
Deaths 15% 4%
Very
significant
Note: Results are for a one-year period
10. ©2014 MFMER | 3338355-10
Epilogue – What Next?
Not ready for prime-time
11. ©2014 MFMER | 3338355-11
Center for the Science
of Health Care Delivery
Improve patient health experience
Improve population health
Improve quality, control cost
Improve medical practice through
analysis and scientific rigor
12. ©2014 MFMER | 3338355-12
Value Framework
Patient
Provider Payer
Quality
Cost over time
(outcomes, safety, service)
13. ©2014 MFMER | 3338355-13
Quality
Measures
Patient
Satisfaction
Costs
Big Data
Health and
Quality of Life
14. ©2014 MFMER | 3338355-14
Value:
In the Eye of the Beholder
The importance
of reflecting and
respecting multiple
perspectives
Appreciating what
we don’t know about
the care experience
Embracing multiple
aims for improvement
concurrently
Source: Bellows J, Sullivan MP. Could a quality index help us navigate the chasm?
http://xnet.kp.org/ihp/publications/docs/ quality_background.pdf. Accessed July 11, 2012.
15. ©2014 MFMER | 3338355-15
Patient
Satisfaction
CostsQuality
Measures
Big Data
Health and
Quality of Life
Example: AWARE
Quality
Cost over time
(outcome, safety, service)
17. ©2014 MFMER | 3338355-17
Critical Care Unit
Situational AWAREness
DATA
Analytics
Metrics Outcomes
Thousands
of data points
About 200 actions/day
1.7 errors/day
29% potentially
serious injury
or death
18. ©2014 MFMER | 3338355-18
Paradigm Shifts
Clinical Management – EMR
Developed by intensivists
Organ/system based information organizer
Database
centered
Provider
centered
Patient
centered
19. ©2014 MFMER | 3338355-19
Provider Built
Field
observation
Surveys &
interviews
Workflow &
workshops
Simulated
tests
20. ©2014 MFMER | 3338355-20
AWARE Goals
Better Care
>90% adherence to best ICU practices
DVT prophylaxis
Stress ulcer prophylaxis
Lung protective mechanical ventilation
Daily assessment of – continuous
sedation; ventilator weaning; need
for intravascular devices and urinary
catheter; physical therapy goals
Better Health
50% reduction of preventable
ICU complications
5% increase in discharge home
vs other health care facility
Lower Cost
Cumulative $$ decrease up to 20%
– length of ICU stay; length of hospital
stay; resource utilization
Areas
Administrative
Patient
Clinical
22. Doe
John
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Doe
Jane
0-000-000
Home Screen Patient Box
After selecting a unit, the rooms
are shown with current patients.
When placing your pointer over
the patient (not clicking)
it will enlarge (as shown)
©2014 MFMER | 3338355-22
Level of care – Click on to select
ICU, PCU, or Floor
Indicators for ventilators,
vasopressors, dialysis, etc
Icons (left to right):
Discharge, Task List, Rounding Tool,
Problem List, Med List, Claim
Patient, and Room Number
Patient name
and MC#
Primary
service – Click
on the service
to text page
Service Pager
7 system review –
Placing pointer over a system will
reveal why it is yellow or red.
23. ©2014 MFMER | 3338355-23
AWARE overview
Information organized by organ and systems
B) Historical
contextual data
E) Provider
actions/support
A) Organ identifier
and status
C) Current organ
physiological status
D) Status of relevant
investigation
Red
values are
Critical
Yellow
values are
Abnormal
24. Patient View
Every value is clickable with trending options
Click on Bedside Monitor to see
live or click back to this main
AWARE patient view.
White Board is a community area for communication.
Each service can use it for different things. In 10-3/4
we use it for patient updated each shift.
Problem List,
Procedures,
Operative Notes,
& Hospital
Admission
Everything is
clickable linking to
the note it originated
from.
Cardiac – Shows cardiac
labs, meds, access dates,
ECG, and ECHOs.
Cardiac is RED here
because the Lactate is 9.1.
Click on them!
Renal – Shows renal labs,
and meds. Renal YELLOW
because UO is low,
noticeable change in wt, &
electrolytes are abnormal.
Hem – Shows labs,
meds, blood products
received & transfusion
review (gives
suggestions for
transfusing or not to
transfuse.
Neuro – Shows meds,
nursing assessment,
neurology notes, pain,
imaging, & EEG.
Click on them!
Respiratory – Shows
meds, nursing assessment,
airway grade, pulmonary
notes, vent/O2 settings,
imaging, ABG, & PF ratio.
Click them!
GI – Shows meds, nursing
assessment, imaging, labs,
& GI notes.
ID – shows antibiotics given
in the last 24 hrs, labs,
micro, temp, & meds. Also
Braden score is listed here.
25. ©2014 MFMER | 3338355-25
Outcome:
Everyone’s Happy
Reduced Cognitive Load
(Happy Clinicians)
80
70
60
50
40
30
20
10
0
Application
NASA-TLX
Standard
Novel
Reduced Errors
(Happy Patients)
Application
Errors (no.)
70
60
50
40
30
20
10
0
Standard
Novel
Reduced Time
(Happy Administrators)
Standard interface
Novel interface
Task attempt
Time (sec)
250
200
150
100
50
0
1 2 3 4
27. ©2014 MFMER | 3338355-27
3 months to collect data
to answer 2 questions
Seconds to collect and
answer the same questions
20 Years Ago Today
28. ©2014 MFMER | 3338355-28
2003
First Human Genome
Time: 10 Years
Cost: $1 Billion
TODAY
Genome Sequencing
Time: 1 Week
Cost: $1,500
29. ©2014 MFMER | 3338355-29
Cost of Whole Genome Sequencing
?
$1,000 to sequence
one human genome
31. ©2014 MFMER | 3338355-31
Types of questions that may be pursued
Comparative
Effectiveness
Behavioral and
Policy Research
Variation in Care
Research
Heterogeneity of
Treatment Response
Optum Labs
H E A LT H
C A R E
R E S E A R C H A N D
I N N O VAT I O N
Provider
Academic
Professional/
Consumer
Organization
Government
Payer
Pharma/
Life
Sciences
An open, collaborative center for research
and innovation for health care stakeholders
interested in improving patient care.
Projects must be primarily to improve
patient care and lower the cost of improved
care, and be transparent to the entire
collaborative.
32. ©2014 MFMER | 3338355-32
Optum Labs — Data and Tools
Advanced Analytics and Data Visualization Data Growth Through Partnership
>149M
“Administrative”
>30M
Clinical
315M
US Population
Mayo Health
System
2
Health
Plan 1
Health
Plan 1
Health
System
3
Clinical
Research
33. ©2014 MFMER | 3338355-33
Optum Labs — Research Process
Data sets and resources are
integrated into a separate
“sandbox.” Data contributions
are tagged and valued.
Contributor data is
de-identified and stored in
standardized data sets, on
secure, private environments.
Project research is done in
the “sandbox” environment
only according to the
Research Proposal.
Upon work completion,
the “sandbox” is dissolved.
Publications and clinical
translation proceed as
appropriate.
Integration Research & Analytics OutputsData
Health
Economics Biostatistics
ActuarialEpidemiology
Innovative
Health Care Insight
Clinical
Data
Admin
Data
Pharmacy
Data
Population
Data
Data Sets
Project
“Sandbox”
Researchers
Real Estate
34. ©2014 MFMER | 3338355-34
Focuses on understanding the underlying behaviors driving patient and
provider behaviors, as well as the evaluation of alternative policy initiatives
Example: Can the application of economic theory to the analysis of claims data improve our
understanding of patient medication adherence? Does the use of copays alter conclusions about
the effects of benefit design on initial prescription fills and refills?
Behavioral and
policy research
Explores the well-documented extensive variations in treatment patterns
by geography and other dimensions
Example: How are measures of geographic variation in care affected by the definition
of geographic region?
Variations in care
Seeks to understand what patient subpopulations are most likely to
respond to a particular treatment
Example: Is a drug equally safe among all patient subpopulations? How could such
information be used to design more efficient trials for future clinical development?
Heterogeneity of
treatment response
Improves the quality of research from observational studies more
generally through fundamental research on data infrastructure and
statistical methodologies
Example: What is the potential value of multiple imputation methods to fill gaps in the data?
Methodology research
Research Themes: Areas of Focus
35. ©2014 MFMER | 3338355-35
Use of new anticoagulants in atrial fibrillation
Longitudinal variation in care analysis of hip and knee surgery
• National trends in the screening, diagnosis, and treatment of localized prostate
cancer
• Unplanned hospital readmission and emergency department care for acute
diabetes complications
• Utilization and variations in uses of proton beam therapy
Step-down protocols in asthma medication
• Diagnosis, treatment, and service utilization for spine-related problems
• GLP-based anti-hyperglycemic medications and risk of acute pancreatitis
and pancreatic cancer
Currently underway or awaiting publication
Likely candidate for clinical translation project
Sample Research Projects
36. ©2014 MFMER | 3338355-36
• American Medical Group Association, Alexandria, Va.
• Boston University School of Public Health, Boston, Mass.
• Lehigh Valley Health Network, Allentown, Pa.
• Pfizer Inc. (NYSE: PFE), New York, N.Y.
• Rensselaer Polytechnic Institute (RPI), Troy, N.Y.
• Tufts Medical Center, Boston, Mass.
• University of Minnesota School of Nursing, Minneapolis, Minn.
Seven Leading Health Care
Organizations Join Optum Labs
37. ©2014 MFMER | 3338355-37
Patients are seen by outside providers/physicians.
Optum Labs data
Patients call and are given an appointment at Mayo.
Example in Action
38. ©2014 MFMER | 3338355-38
Patients are seen by initial Mayo team.
Document patient expectations – “Pt Exp’n”
Patients indicate their expectations.
39. ©2014 MFMER | 3338355-39
Patients are presented medical vs. surgery information
Document education
Patients make a decision about their care: medical/surgery
Shared decision making – SDM
40. ©2014 MFMER | 3338355-40
Patients receive care
… some being treated medically, others with surgery
Collect risk factors and other data
Patients see medical/pre-operative Mayo team
Collect treatment data
Mean length of stay
for primary TKA
OPTUM (x age = 56.6)
3.0 days
MAYO CLINIC (x age = 70)
2.85 days
41. ©2014 MFMER | 3338355-41
Patients complete care at Mayo
Collect discharge disposition data
Patients might be seen by outside providers
Post-Mayo – Optum Labs data
Patients later report their outcomes from medical care/surgery
Patient-reported outcomes – PRO
Discharge to home
OPTUM (x age = 56.6)
81.4%
MAYO CLINIC (x age = 70)
63%
30-day readmissions
OPTUM (x age = 56.6)
4.4%
MAYO CLINIC (x age = 70)
1.6%
42. ©2014 MFMER | 3338355-42
Surgical Process Flow
for Costing
- TDABC method
C (Circulator Nurse)
Surgical Assistant
Scrubs Technician
RN Anesthetist-NA
Radiology Technician
S (Surgeon)
A (Anesthesiologist)
AR (Anesthesiologist Resident)
R (Resident/Fellow)
Inpatient Space
Operating Room
Surgery
Process
Post
Surgery
E22
Patient Prep
for Surgery
C AR
20
A
5 20
20 R
20
C
20 E22
Operation
(Incision
to Closure)
CAR
91
A
46
S
73
91 R
86
C
91
91
E28
Operation
(Incision
to Closure)
CAR
88
A
44
S
71
88 R
83
C
88
88 10
E30
EMR
documentation
and contact family,
supervision time,
post procedure
note, order tests
S
10
R
5
Hip or
Knee?
Hip
Knee
FLOW 1
43. ©2014 MFMER | 3338355-43
The Value Equation Comes to Life
Quality outcome data:
Patient-centric outcomes
Practice performance outcomes
Cost:
Outside Mayo
At Mayo
“Cost avoidance”
44. ©2014 MFMER | 3338355-44
Data are collected from all Mayo Clinic sites
Comparing and adopting best practice
helps improve value for all
THA +22 +120% TKA +14 +110% PHM +0.96 +5.36% HD +3.34 +5.23% DHI +0.81 +3.
45. ©2014 MFMER | 3338355-45
Age BMI Strength Exercise
85% probability
of going home 3 days postop
AND being able to stand/walk
without pain for 30-min 3 months postop
Knee Replacement Value Proposition