Transcript Ramsey_RoleofInformatics - South Carolina Hospital Association

The Role of Informatics
in CIN’s, ACO’s and the Triple Aim
October 13, 2016
Dave Ramsey
Director of Informatics
CCI Labs
A bit about CCI
• CCI is an LLC created to consult and support organizations seeking to form CIN’s and ACO’s
• It has capability to deliver:
• Data aggregation
• Data analytics
• Organizational consulting
• Care management tools
• CCI Labs is a not for profit 501(c)3 aligned with CCI and the University of South Carolina School
of Medicine
• CCI Labs conducts applied research in population health
• The two companies have approximately 60 staff members in aggregate
What’s a CIN?
• A Clinically Integrated Network
• A collection of practices and their associated physicians who agree to share patient
data for the purposes of quality improvement on behalf of the patients served
• Employed and affiliated physicians may negotiate collectively with payers regarding
reimbursement arrangements
• They are not just a loose collaboration to “gain leverage” and are regulated by the FTC
What’s an ACO?
• An Accountable Care Organization
• A collection of practices, hospitals and their associated physicians who agree to share
coordinated care for a specific population
• Typically there is some reward (upside potential) and maybe some risk (downside
potential) for managing these patients against a defined set of criteria which include
• Medical quality metrics
• A threshold of expense
• Patient satisfaction
• The most widely known ACO is the MSSP (Medicare Shared Savings Program) of CMS
What’s a Population
•
A “population” is a group of patients who share some common persistent characteristics
• Geography
• Ethnicity
• Chronic disease
• Economic status
• Employees
• Age brackets
• The health of these populations leads to commonality of medical treatment or medical
policy
So for the MSSP, the population is those served by Medicare and hence generally of age 65+
What Role Does Informatics Play
• Data Collection
• Acquisition
• Normalization
• Quality assurance
• Metric Reporting
• Periodic calculation of standardized metrics for all participating physicians, practices and
hospitals
• Reporting of these same metrics to governmental agencies
• Support for Quality Improvement Initiatives
• Huddle reports
• Gaps in care
• Custom metrics (ex. for a PCMH quality initiative)
• Support for Care Coordination
• Care transitions
• Case management
• Price of care assessment (and how it relates to relevant payment models)
• Comparison to regional norms
• Understanding underlying costs
Data that is systematically used
• Patient data:
• Demographic
• Vitals
• Labs
• Meds
• Visit and scheduling
• Diagnosis
• Insurance
• Billing
• Provider data:
• Identification (NPI, address …)
• Association with a practice
• Title/Role (MD, DO, PA, NP, practice manager …)
• Membership (ACO, CIN, research projects …)
• Practice or Hospital:
• Addresses
• Sites and identifiers
• EMR’s used
• IT point of contact
• Billing data processors
Reporting
• Reports are program or institution specific
• MSSP for CMS
• DRP/HSRP for NCQA
• PCMH for NCQA
• MACRA for CMS
• Homegrown reports for employers
• Typically CCI is asked to keep statistics on providers accessing reports as a measure of provider
engagement
• Reports are, at the highest level, just statistical but patient data relative to each metric is
available from the reporting system with additional authentication and transmission techniques
Physician Report Cards
Your results
Choose
report
Choose who the
report is about
Choose who to
compare results
against
Comparator
results
Above metric
of success
Below metric
of success
Your missing
data
Metric name (click
for definition)
Email’s you patient data
used in this metric
Comparator
missing data
Huddle Reports
ABCS Daily Huddle Report (Planned Patient Visits Today)
Demographics
Patient
• Program Specific or
general purpose
• Who a provider is
seeing today
Age
ABCS Patient Metrics
Data
Visits MSSP Share
Ba l l a rd, Ja cki e
65
0
Y
Y
Cos tel l o, Jus ti n J
61
1
Y
YF
Ka kei , Ka thi
63
3
N
N/A
Hopki ns , Joe
68
0
Y
Y
Ma s on, Luke
61
1
N
N/A
Stei n, Jul i e
42
3
N
N/A
Turner, Sa ra
58
0
Y
YF
Wes ton, Fra nk
61
1
Y
YF
Young, Greg
52
3
Y
Y
A1
A2
B1
C2
CVA
CHD
CVD
Dm
Htn
Y
4
13
13
Y
3
N
N
M
Ces
N
2
10
12
N
16
N
Y
N
Unk
N
8
2
3
N
1
N
Y
128/85 120
M
Non
N
4
8
9
N
3
N
N
170/96
56
H
Ces
Y
4
13
12
Y
16
N
Y
155/86
93
N
Unk
N
4
7
6
N
1
N
Y
132/85 140
M
Non
Y
4
13
13
Y
3
N
N
Y
170/92
56
N
Ces
N
4
9
11
N
16
N
Y
N
148/86
93
N
Unk
N
11
10
10
N
1
Y
Y
128/85 130
H
N
170/92
56
N
145/88
93
Y
N
Y
Y
S
IVD
GI
Actions to consider
Non
Y
C1
Risks
Ni ck Ri veri a , MD
Cros s Creek Pri ma ry Ca re
Wednes da y, Apri l 2, 2014
CKD
Increase statin dosage, confirm MSSP
data share
Control BP: Aspirin
Smoking
Increase statin dosage
Control BP, Aspirin
Control BP;Smk Status
Increase statin dosage
Control BP
Control BP;Smk Status
Recent Emergency Room/ Hospital Admission & Discharge (patients you have seen in the last 2 years)
Patient
ER Admit. Date
Reason ER Discharge Date
Hosp. Admit Date Reason Hosp. Discharge Date Notes
• Red shows area
needing attention
• Right column are
recommendations
for action at this
visit
Ballard, Jackie
2/15/2014
585
Costello, Justin J
1/18/2014
401.9
2/15/2014
1/19/2014
Kakei, Kathi
786.5
3/4/2013
R07.9
Medicare Share Savings Plan (MSSP)
Yes or No
Data Sare
Yes
YF is face to face confirmed
Asprin
A1 or A2, not both
A1 is primary
for men, is chd risk> gi risk; women cva risk>gi risk, then asa
A2 is secondary with requirted IVD
Blood Pressure
B1 is control to 140/90 if Htn
Cholesterol
C1 will be LDL controlled (ATP 3 guidelines)
C2 will be moderate dose statin or high dose statin based on risk (2013 guidelines) [CCI recommended]
Smoking
Non, Cessation or Unknown
3/6/2014
Monitored for chest pain
Inflammatory vascular disease
Gastrointestinal
CVA Cerebrovascular Accident (stroke)
CHD Coronary Heart Disease
CKD Chronic Kidney Disease
CVD Cardiovascular Diasease
Dm Diabetes Mellitus
Htn Hypertension
IVD
GI
Other data and metrics
• Some institutions have local quality measures and these may require additional data
• Time to administer antibiotic from time of diagnosis of sepsis
• Percent of new mothers given breast feeding education
• Percent of generic drugs relative to brand drugs
• CCI does calculate these institution specific metrics
• They may require chart reviews if the data isn’t discrete or well recorded
• When chart reviews are conducted, they are often sampled to contain the cost of
collecting the data and therefore are estimates which are institutionally interesting
but not valid as measures of individual provider quality
Quality Improvement
•
•
•
•
Coding and data collection for improved analytics and metric achievement
Addressing gaps in care
Evaluating patient health outcomes over time
Evaluating population health outcomes over time
How does Informatics support Payment Models
• Fee for service
• Cost/Price
• Measures to determine if the service lead to a positive outcome or to a readmission
• Patient Satisfaction
• Bundled payments
• Annotating the team members
• Evaluating the members contribution to cost
• Measures to determine if the service lead to a positive outcome or to a readmission
• MACRA MIPS AAPM
• Quality measures
• EMR utilization measures
• Quality Improvement activities
• Prediction of costs (run rate)
• Risk assessment
What CCI Labs (research) is studying
•
•
•
Advanced protocols for the treatment of chronic diseases or conditions
• Hypertension (AMA MAP)
• Heart attack and stroke reduction (CMS Million Hearts study)
• Smoking cessation (Tip Top protocol)
Rapid learning systems for understanding populations
• New data sources
• Psychosocial data
• Environmental data
• Geospatial data
• Community data
• Public policy data
• Non-hypothesis driven analytics (e.g. machine learning)
• Continuous mining of data and evaluation of impact of changes
Primary Care Provider guidance for treatment of clusters of chronic diseases
A deeper look at data analysis behind our cluster work
•
•
•
•
•
•
•
Characterized just data from patients in our MSSP ACO
Agglomerative Approach (bottom-up)
Start with each patient in a separate group
Merges patients that are ‘close’
Keeps merging groups that are ‘close’
Continues until all groups are merged into one
Creates a tree or dendogram
11 most frequent conditions
> 50,000 patients
– 72% Hypertension
– 69% Lipid Metabolism Disorders
– 39% Vascular Complications
– 38% Obesity
– 35% Osteoarthritis
– 29% Mental Health (Depression, Abuse)
– 28% COPD
– 28% Diabetes
– 15% Cancer
– 14% Renal Failure
– 12% Congestive Heart Failure
Easy to calculate – just count
If we look at clusters we find …
Cluster 5
Cluster 6
Cluster 7
Cluster 8
5383 (10.6%)
6237 (12.3%)
5226 (10.3%)
6581 (13.0%)
Diagnosis 1
Cancer, 100%
CHF, 95%
COPD, 100%
DM, 100%
Diagnosis 2
HBP, 74%
HBP, 94%
HBP, 71%
HBP, 88%
Diagnosis 3
Lipid, 73%
Lipid, 83%
Lipid, 66%
Lipid, 85%
Diagnosis 4
Vascular, 39%
Vascular, 76%
Obesity, 37%
Obesity, 56%
Diagnosis 5
OA, 36%
COPD, 59%
Vascular, 32%
OA, 44%
Diagnosis 6
Obesity, 33%
DM, 50%
OA, 30%
Vascular, 37%
Diagnosis 7
COPD, 28%
Mental, 42%
Mental, 21%
Mental, 36%
Diagnosis 8
Mental, 27%
Renal, 42%
DM, 15%
COPD, 19%
Diagnosis 9
DM, 22%
Cancer, 20%
n
Diagnosis 10
Computationally difficult to define clusters but important results
Current work – guidance for primary care
•
•
•
•
Looking at all treatment regimes for each disease in each cluster
Evaluating medications that are in conflict among diseases in a cluster
Evaluating hierarchy of treatment urgency
Reducing the Cluster to a single protocol with just a few pages of recommendation rather than
hundreds of pages
• Giving guidance to physicians via a portable device
• Must do
• Should do
• Should not do
• Must not do
• Threshold for referral
Questions