Transcript Slides - RARE Campaign

RARE Conversations
October 30, 2012
Hosted by RARE Operations Partners:
Institute for Clinical Systems Improvement, Minnesota Hospital Association, Stratis Health
Our host today will be…
Kathy Cummings
Kathy Cummings is an ICSI Project Manager for the
Reducing Avoidable Readmissions Effectively (RARE)
Campaign, a collaborative effort led by ICSI, the
Minnesota Hospital Association and Stratis Health.
These organizations have joined together to engage
more than 80 hospitals and other partners across the
continuum of care to prevent avoidable hospital
readmissions in Minnesota.
Kathy holds a bachelor’s degree in nursing from the
University of Iowa and a master’s degree in human
resource development from the University of St.
Thomas.
Why RARE Conversations?
Share
Networking
opportunities
Engage
Learn
Conversation
October’s Conversation…
Risk Stratification
Sharing their work:
Hennepin County Medical Center &
Park Nicollet
More about the presenters…
Sandy Hilliker
Sandy is a Director of Case Management at
Hennepin County Hospital. She is an energetic,
responsive, trusted self-starter with extensive
experience in leadership. She has
demonstrated success in process
improvements and positive patient outcomes.
More about the presenters…
Scott Shimotsu
Scott Shimotsu is a healthcare analyst in the
Performance Measurement and Improvement
Department at Hennepin County Medical
Center.
Last year, he graduated from the University
Minnesota PhD program in Epidemiology and
Community Health. With over 12 years of
healthcare experience, Scott brings expertise
in advanced healthcare analytics, obesity
prevention and biostatistics. His research
areas include obesity prevention, diet and
alcohol use, and social determinants of
chronic disease.
Towards the Development of a
Readmissions Risk Tool
ICSI RARE Conversations
October 30,2012
Sandy Hilliker RN,DNP and
Scott Shimotsu, PhD MPH CPHQ
Case Management/Performance Measurement and Improvement
What did we do ?
• Created a adult high risk assessment tool
High Risk Criteria Score
– Two or more Admissions in the last 30 days
– Two or more ED/APS visits in the last 30 days
– Presence of:
•
•
•
•
•
Drug Use
Depression
Renal Failure
Heart Failure
Asthma
– Race
Low Risk Patients
• Low Risk Criteria
– No Admission, Readmission, or ED/APS visit in the last 30 days
– High Confidence in patient and family to give self-care, based on Teach
Back
• Interventions
–
–
–
–
Phone number to call if needed
Follow-up appointment made
Medication Reconciliation Prior to DC
Initiate any additional services as needed
Moderate Risk Patients
•
Moderate Risk Criteria
– One Admission last 30 days
– One ED/APS Visit in the last 30 days
– Regarding self-care, moderate confidence that patient or family, based on Teach
Back, can carry out the care needed.
– Presenting Illness (Cardiovascular, Pulmonary, Renal, or Infectious)
•
Interventions
–
–
–
–
–
–
–
Follow-up phone call post discharge within 48 hours
Medication Reconciliation Prior to DC
Follow-up Clinic Appointment within 5 days
Home care visit within 72 hours
Warm Hand off to clinic
Identify who patient calls with questions /concerns
Social Service assess within 24 hours of admission and implement discharge plan
• Identify community resources
• Identify transportation needs
• Identify tele-monitoring as needed (CHF, COPD, Diabetes) FUTURE
High Risk Patients
• Interventions
– Follow-up phone call post discharge within 24
hours
– Medication Reconciliation Prior to Discharge
Follow-up Clinic Appointment within 72 hours
– Home care visit within 48 hours ( Minnesota
Visiting Nurses Association )
•
Risk Tool Preliminary Evaluation
1. Retrospective Readmissions Factor
Study
– Social/Personal Risk Factors Among A
Diverse Racial/Ethnic Minority and Immigrant
Patient Population: A Multivariate Analysis
– May 1, 2011-April 30, 2012
2. Preliminary Metrics Evaluation Study
– ROC Curve Analysis
– Timeframe: July 2012-September 2012
Results
• N=2508 Cases with a Risk Criteria Score
• Low Risk
• Moderate Risk
• High Risk
60%
13%
27%
• Overall Readmission Rate
9%
• C stat (95% CI)
0.60 (0.56,0.64)
Risk Tool Preliminary Evaluation:
Results
Risk
N
Category vs.
Readmit
(yes)
% READMIT
LOW
105
7%
MODERATE
24
7%
HIGH
93
14%
Next Steps
• Assess Measures to capture interventions and
processes
• Year-to-Date Risk Tool Evaluation on
Readmissions and Process Measures
(January 2013)
• Reconsider New Risk Factors:
Socio-demographic, Environmental, Social
Support, Substance abuse
More about the presenters…
Eva Gallagher
Eva Gallagher is the Senior Director of
Quality, Innovation and Population Health at
Park Nicollet Health Services in Minneapolis,
MN. Eva completed the adult nurse
practitioner program at the College of St.
Catherine and earned a PhD in nursing from
the University of Minnesota.
More about the presenters…
Gregg Teeter
Gregg has worked at Park Nicollet Health
Services for the past nine years leading various
analytic and reporting departments (Demand
Planning & Analysis, Clinical Reporting &
Analytics, and Business Intelligence).
He is currently working in a Lead Analytic
Advisor role in support of enterprise level
initiatives. His primary focus in this role is to
support of Park Nicollet’s Population Health
and Pioneer ACO activities.
Identifying Patients At Risk For
Readmission At Methodist Hospital
Eva Gallagher
Gregg Teeter
For everything you love.
Discussion
• Aligning Resources
• Developing A Care Model
• Identifying High Risk Patients
For everything you love.
Reengineered Support for Patients
Care Integration Role Definition – RN Care Coordinators and Social Work
For everything you love.
Reengineered Support for Patients
Care Integration Focus
Before - LOS
After - Transitions
For everything you love.
Reengineered Support for Patients
RN Care Coordinators paired with Hospitalists
Pilot found improved
teamwork,
better able to prioritize
work, potential
discharge errors found
For everything you love.
Reengineered Support for Patients
RN Care Coordinators paired with Hospitalists
For everything you love.
Care Model Enhancements
• Inpatient
– Consults as needed: pharmacy, nutrition, CDE, PT,
OT, spiritual care
• Post-Discharge
– Post discharge phone calls
– Discharge appointments – 3-5 days for high risk
– Home visits to all high risk patients
– Transition call to NH, TCU
– Care consultant assigned as needed
For everything you love.
Predicting Which Patients Are At High Risk Of
Readmission
Vision:
What if we could predict which patients have a high probability of
being readmitted?
If we could, what could we do, while that patient is under our care, to
decrease that risk?
Challenge:
Which combination of variables are key drivers for risk of
readmission?
Probability
of
Readmission
A
B
C
D
For everything you love.
Readmission Model Details
• Model 1.0
– Developed in the Spring, 2012
– Design based on variables identified from literature review
– Subjective weighting and scoring of the variables added up to a
total score
– Aggregated and displayed results in Epic with a banner on the
inpatient record
– Most important: the tool became part of the process
• Model 2.0
– Developed concurrently
– Based on data in our enterprise data warehouse
– Identified the drivers of readmissions from an analysis of historical
data to develop a regression equation that has actual predictive
power
– Went live in October, 2012
For everything you love.
Model 2.0 Readmission Driver Variables Evaluated
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Patient demographic variables
Account type/subtype
Admit source
Admit status
Admit service
Discharge disposition
Length of stay
Infection control status
High risk diagnoses within past year
High risk diagnoses during index admission
HCC score
Admits in past 3yrs, 2yrs, 1yr, 6mo, 3mo, 1mo
# days since last admit
EC visits in past 3yrs, 2yrs, 1yr, 6mo, 3mo, 1mo
# days since last EC visit
UC visits in past 3yrs, 2yrs, 1yr, 6mo, 3mo, 1mo
# days since last UC visit
PC visits in past 3yrs, 2yrs, 1yr, 6mo, 3mo, 1mo
# days since last PC visit
CAM scores (# of positive scores, most recent
result during admission, most recent result prior
to index admission)
•
•
•
•
•
•
•
•
•
•
•
•
•
•
PHQ9 score (max score during admission,
most recent score prior to admission)
Systolic BP (highest, lowest, most recent during
admission)
Pulse (highest, lowest, most recent during
admission)
BMI
Bun/Creatinine lab values (count, min, max, std
dev, most recent)
Glucose values (count, min, max, std dev, most
recent)
Hemoglobin A1c values (count, min, max, std
dev, most recent)
Serum albumin values (count, min, max, std
dev, most recent)
Braden score
Falls risk score
Medications
Homecare in past 6-12 months
Assistive devices during index admission
Level of assist during index admission
For everything you love.
Model Differences
Previous Model Variables
Current Model Variables
Age
Age
Living arrangements
Race
Type of residence
Marital status
Readmit or ER visit w/in past 2 weeks
Gender
Multiple medical problems
HCC score
Falls risk score
Length of current stay
CAM score
# of admits in past 6 months
Braden score
# of ED visits in past 6 months
Patient type (medical or surgical)
Analysis suggested that the prior model’s predictive power was low, while Model 2.0’s predictive
power was significantly better (as good as anything that has been published)
Model Limitations:
•Variables in the prior model were dependent nurse input
•Model 2.0 dependent on patient having prior utilization data
For everything you love.
Next Steps
• Operations
– Visibility of the banner post discharge
– Automated communication back to PC regarding
acute events (EC, inpatient, obs)
• Measures & Models
– Analyze and track the impact of the change
– Expand model to Observation and EC patients
– Real time census updates and automating the
transfer of the score into Epic
– Evaluate condition specific predictive models
For everything you love.
Our host today…
Kathy Cummings
Kathy Cummings is an ICSI Project Manager for the
Reducing Avoidable Readmissions Effectively (RARE)
Campaign, a collaborative effort led by ICSI, the
Minnesota Hospital Association and Stratis Health.
These organizations have joined together to engage
more than 80 hospitals and other partners across the
continuum of care to prevent avoidable hospital
readmissions in Minnesota.
Kathy holds a bachelor’s degree in nursing from the
University of Iowa and a master’s degree in human
resource development from the University of St.
Thomas.
Questions
Question # 1
• How are you identifying patients at high-risk for
readmissions?
Question # 2
• How does it impact the care and services you provide
for these patients?
Now we will take questions from the field…
RARE Conversations
Upcoming RARE Events:
•
RARE Rapid Action Learning Day, Thursday November 8, 2012
Crown Plaza Conference Center, Plymouth, MN, 8:30am-3:30pm
•
RARE Webinar, Analyzing Your Portal Data, Friday December 7, 2012,
12 noon -1p.m.
RARE Conversations
To suggest future topics for this series, “RARE
Conversations” networking, contact Kathy Cummings,
[email protected]
Thank You for Your Participation!
A recording of this RARE Conversation will be
available within 3 days and posted on the RARE
website, www.rarereadmissions.org