Transcript Document

How to prove that you are
being useful - a guide to
system level metrics in urgent
and emergency care
Dr Ian Sturgess
Partner,
NHS Interim Management and Support
Senior Clinical Lead,
Emergency Care Intensive Support Team
A whole system perspective
Focus on CDM and more effective responses to urgent care
needs – ACS condition management
Clear operational performance framework and integrated in to primary
care
Improved integration with primary care
responders
Front load senior decision process
incl primary care
General
Practice &
GP OOH
Communit
y Support
Ambulance
Service &
GP OOH
Discharge Process
Health
Promotion
A+E
MAU/SAU/
Short Stay
Optimise
ambulatory
emergency care
Inpatient
Wards
Redesign
to left
shift LOS
Information flow converting the unheralded to the heralded
Preventative/
Predictive care
Disease management
Managed populations
Alternatives
to acute
admission
settings
Alternative
access for
diagnosis
Alternative
settings for
therapy
Alternative
sites for
discharge
Alternative
sites for
readmission
“Quality begins with
intent, which is fixed by
management.”
W. E. Deming, Out of the Crisis, p.5
3
System Level Improvement? How
Do You Know?
• Be clear about your aim statement – with a
definable system level improvement metric(s) –
how much by when and how measured.
• Measure some key ‘Process’ metrics.
• Ignore balancing metrics at your peril!
• Use Statistical Process Control effectively
Clarify what your aim is.
1. Promote healthy living and independence.
2. Attendance and admission avoidance at times of
acute care need.
3. Reducing length of stay and readmissions by
effective early supported discharge
4. End of Life care
All are amenable to having a definable ‘aim statement’ with a
linked system level impact metric, a process metric and a
balancing metric.
The Three Faces of Performance Measurement
Improvement
Accountability
Research
Aim
Improvement of care
Comparison, choice,
reassurance, spur for
change
New knowledge
Methods:
Tests are observable
No test; merely
evaluate current
performance
Test blinded or
controlled tests
Accept consistent bias
Measure and adjust to
reduce bias
Design to eliminate bias
• Sample Size
“Just enough” data,
small sequential
samples
Obtain 100% of
available, relevant data
“Just in case” data
• Flexibility of
Hypothesis flexible,
changes as learning
takes place
No hypothesis
Fixed hypothesis
• Testing Strategy
Sequential tests
No tests
One large test
• Determining if a
Change is an
Improvement
Run charts or Shewhart
control charts
No change focus
Hypothesis, statistical
tests (t-test, F-test, chi
square),
p-vlaues
• Confidentiality of
the Data
Data used only by those
involved with
improvement
Data available for public
consumption and
review
Research subjects’
identities protected
Aspect
• Test
Observability
• Bias
Hypothesis
6
Focus on the Vital Few!
There are many things in life that are interesting to
know. It far more important, however, to work on
those things that are essential to quality than to
spend time working on what is merely interesting!
The challenge, therefore, is to be disciplined
enough to focus on the essential (or vital few)
things and set aside those things that might be
interesting but trivial!
7
System Levels
Example
Nursing
Services
Macrosystem
Mesosystem
Nursing
Divisions
Microsystem
Frontline
Nursing
Units
Source: Bojestig, Jonkoping CC Sweden
8
Building a Cascade of Measures
Outcome - system level eg admissions,
death, harm, Institutionalisation etc
L1
System
L2
Board & CEO
Process + Outcome
L3
Service Line
L4
Process (+ Outcome)
Microsystems: Units, Depts
L 5
Physician & Patient
Individual
Process Metrics
Adapted from Lloyd & Caldwell
Drivers
Secondary Drivers
Appropriate use of intensive
hospital services (ICU care)
Hospital Care
SWVMC
CAMC
Memorial Hermann
(Sepsis + CHF)
Identification of patient
severity and wishes with
respect to end of life care
Timely referral to palliative
care / hospice options
Identification of provider
responsible for coordination
Appropriate Utilization
of Resources at the
End-of-Life
Coordination of Care
Execution of a shared
treatment plan (all providers
and patient and family)
Utilization Measures
(last six months of life)
•Hospital days
•ICU days
•Physician visits
Lehigh
Handoff management
Assist patient and family to
establish goals and intention
Patient and Family
Support
Preparation of family
caregivers to cope with
exacerbation
24 hour access to appropriate
services
Provider Supply
Availability of providers
Availability of resources
Managing the Streams
Identify the stream
– Short stay
Sick specialty
– Allocate early to teams skilled in that stream
Sick frail
Complex
Number of patients
250
Short stay – manage to the hour
Maximise ambulatory care
200
150
100
50
Clarity of specialty criteria
Specialty case management plan at
Handover – no delays
Green bed days vs red bed days
Minimise handover
Decompensation risk
Early assertive management
Green bed days vs red bed days
Complex needs – how
much is decompensation?
Detect early and design
simple rules for discharge
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59
Length of stay (days)
The road to ruin:
Capacity plans and contracts
based on average past activity
Fail to account for
variation in demand +
Fail to deliver
required activity
to meet demand
Income less
than expected
Fail to account for
variation in capacity
Guarantee waiting
times beyond emergency
and elective targets
Increased
variations
in capacity
Reduces
effective
capacity
Increased staff overtime
& waiting list initiatives
Increased costs
Cost cutting
initiatives
Impact – Beds occupied – Total
Objective – Hard Red Lines
Aim – Reduce Acute beds occupied to SPC mean of 600 or less + reduce crude in-hospital mortality
rate by 10% + a fall in SCHMI by 31st March 2012
Process measure – The whole system action plan etc etc ie holding the system to account not just the
acute sector.
Balancing – Deliver a decrease in Long term care ie more patients returning to live at home. No
increase in 30 day re-admission rate
Trust NEL Admissions and Discharges
Aim – Reduce emergency admissions – by 20 by 31st March 2012
Processes – RAT in A+E, 10 Care improvements, improved EoL care etc
Balancing - prevent any increase in institutional care
Zero LOS Discharges - Trust
Excl paediatrics, midwifery and obstetrics
Aim – Increase zero LOS
Process – deliver AEC
Balancing – Reduce overall NEL admissions
2 midnights or less LOS Discharges Trust
Aim – Increase short stay discharges
Process – deliver AEC + short stay review process
Balancing – Reduce overall NEL admissions
In-Patients with LOS 14 days or more Trust
Aim – Reduce I/P with LOS 14 + to 75 or less by 31st March 2012
Process – Early identification of at risk group, CGA, early supported discharge schemes
Balancing – no increase in institutional care – aim for a reduction in over 75s in Long term Care
Background to Statistical Process
Control (SPC)
• Introduced by Walter Shewhart (Bell Telephone
Laboratories 1924)
• The method was exported to Japan in the 1950s, where it
was successfully applied in industry.
• SPC techniques “demonstrate the simplicity and power of
control charts at guiding their users towards appropriate
action for improvement”. 1
1. Mohammed MA, Cheng KK, Rouse A, Marshall T. Bristol, Shipman, and clinical governance: Shewhart's forgotten lessons. Lancet
2001; 357(9254):463-467
Can you identify the flaws in the
following “dashboard?”
Measure
Acute MI Core
Measures
Congestive Heart
Failure Core Measures
Pneumonia Core
Measures
Press-Ganey Patient
Satisfaction
OR Turnover Time
Falls
Medication Errors
Total Knee and Hip
Infection Rates
Surgical Site Infection
Rates for Cardiac
Surgery
Time to answer nurse
call lights on all
Med/Surg Units
Current Performance
6 Decile National, 4h decile State
Goal for 2007
2 state decile or above
4th Decile National, 2nd decile State
2nd State decile or above
3rd Decile National, 1st Decile State
2nd State decile or above
57% Rate us “Excellent”
Statistically significant improvement
i.e 62% “Excellent” rating
15 minutes
Less than 5 per 1000 patient days
Less than 7 per 1000 patient days
h
22 minutes
7 per 1000 patient days
5.1 per 1000 patient days (from Nurse
Variance Reports)
1.2%
4.2%
nd
Less than 4.1 %
i.e. Better (lower) than 50th %tile for
NNIS
Less than 10.4% i.e. Better (lower)
than 50th %tile for NNIS
We are developing a standard measure, We are aiming to achieve significant
and will report in future meetings to
improvement in timeliness of
Board on this initiative
response to patients concerns.
Given two different numbers, one will always
be bigger than the other!
What action
is appropriate?
Something
very important!
Last
month
This
month
What does this data tell us?
Patients treated
650
600
550
500
450
400
350
300
Jan07
Feb07
Mar07
Apr07
May07
Jun07
Jul07
Aug- Sep07
07
Oct07
Nov- Dec- Jan07
07
08
Feb08
Mar08
Apr08
May08
Jun08
Jul08
Aug- Sep08
08
Oct08
Nov- Dec08
08
Variation in a system is normal 1
• The variation is caused by factors that are inherent
in the system over time
• They affect all outcomes
• This is ‘common cause’ variation or
• The causes are ‘unassignable’
• Common cause variation can be reduced by
tackling things that affect the process all the time
1. Roberts T. Understanding variation [Online] July 2005 [cited July 2009]; [17 Pages] Available from:
http://www.evidencebasedpractice.org.uk/documents/presentations/spc_TEBPCJune2005.ppt
Some variation may not be normal
1
• The factors are not present in the process all the time
• They do not affect everybody
• They arise because of specific circumstances
• This is ‘special’ or ‘assignable’ cause variation.
1. Roberts T. Understanding variation [Online] July 2005 [cited July 2009]; [17 Pages] Available from:
http://www.evidencebasedpractice.org.uk/documents/presentations/spc_TEBPCJune2005.ppt
Two types of SPC chart
• If you want to compare different individuals,
units or hospitals etc over a single time
period, a ‘funnel chart’ may be helpful
• If you want to compare a single individual,
unit or hospital over different time periods, a
‘time chart’ may be helpful
Anatomy of an SPC ‘Funnel Chart’
Chart title
Non Elective
COPD
admissions/ /100
100 COPD
Patients
Patients
admissions
Elective
Non
60.0
Likely Special Cause Variation
Practices with higher or lower than
average admissions may be explained
by a variety of factors
50.0
Upper
99.8%
40.0
Upper
95%
30.0
Overall
Mean
20.0
Likely Common Cause
Variation
Lower
95%
Lower
99.8%
10.0
Likely Special Cause Variation
0.0
0
50
100
150
200
250
Size
COPD
ListList
Size
Example data for illustrative purposes only
300
The Improvement Process
100
90
80
70
60
50
40
30
20
10
0
Special causes present unpredictable
Process predictable
Time
Process improvement
3 Dangers to Beware Of…
• Reacting to special cause variation by changing the
process
• Ignoring special cause variation by assuming “its part
of the
•
process”
• Do not compare more than one process