Exploring a New Mechanism Increasing Emergency Department Visits

Exploring a New Mechanism
Increasing Emergency Department
Visits
Katrina Hull
University at Albany
April 17, 2015
Contents
•
•
•
•
•
•
Background
Current Theories
Dynamic Hypothesis
Model Scenario
Model Results
Discussion
2
Emergency Department Crowding
• Definition
– Wait times
– Ambulance diverting
• Historical trend
– Increase in per capita use, twice what would be
predicted by population growth
– This research focuses on the increased per capita
use
3
Source: Kaiser Family Foundation
<http://www.statehealthfacts.org/comparetrend.jsp?sub=217&sortc=1&o=a&ind=388&typ=1&sort=a&cat=
8&yr=138&srgn=1>
4
Research Goal
• Propose an endogenous dynamic mechanism
increasing emergency department visits
• Develop a model of this hypothesis
• Use the model to determine useful empirical
data to support this hypothesis
5
Health of ED Patients
• Ambulance diversions correlated to poorer
outcomes for heart attack patients
• Overcrowded EDs may result in lower quality
of care
– Patient boarding
– Stressed physicians less effective
• Stakeholders
– ED Patients
6
Internal Process Variables
• Premise: A better structured ED and Hospital
could handle the increased load
• Prior focus of system dynamics work
• Models examine patient flow through the ED
to discharge (appropriate or not) or admission
• ED as backdoor to admissions
• Stakeholders
– ED staff
– Hospitals
7
General Population Health
• Increase ED use as symptom of poorer health
in the population
• Ambulatory sensitive care conditions
• Failure of system to care for vulnerable
populations
• ED visits are urgent but should be avoidable
• Stakeholders
– Society
– Patients
8
Literature Summary
•
•
•
•
Lot of exogenous theories
Multiple actors with disparate motivations
Conflicting policy approaches
Perfect space for a model
– Unify multiple stakeholder perspectives
– Highlight interaction of their activities
– Examine outcomes of suggested policies
9
Dynamic Hypothesis
• Where would ED patients come from within
the healthcare system?
10
The Story
• Costs of hospital admissions rose due to
factors such as improved medical technology
• Payers became alarmed at the cost of hospital
admissions
• Policies were created to reduce the cost of
individual admissions and reduce total
admissions
• The unintended consequence was more
emergency department visits
11
Reference Mode
440
420
ED Vistis per 1000
400
380
Increase Outside
Model Boundary
360
Physicians Redirecting Patients
340
320
300
1990
1995
2000
Model
Historic
2005
2010
2015
No Change
Historical Data from Avalere Health analysis of American Hospital Association Annual Survey data
12
Feedback Loops
percent of gp admits
redirected to ED
ED referrals
by gp
Expected wait
time for gp
referral
Redirecting
GP referrals
to ED
total ED visits
wait time to admit
gp referral
pressure from payer
to reduce hospital use
total costs to
payers
Adjustment
to Increasing
Costs
Expected
costs
13
Payer Sector
<total admissions>
Time adjusted cost
per admission
<Time>
<Total ED Visits>
time adjusted
cost per ED visit
Total inpatient cost
ED costs
total costs
<Time>
chng expected
costs
expected
costs
pressure from payer to
reduce hospital use
-
time to change
expectations
14
Time Adjusted Costs
Time Adjusted Cost per ED Visit
Time Adjusted Cost per Hospital Admission
$1,400
$14,000
$1,200
$12,000
$1,000
$10,000
$800
$8,000
$600
$6,000
$400
$4,000
$200
$2,000
$0
$0
1990
1995
2000
2011
2015
1990
1995
2000
2006
2012
2015
Based on time series data from 1990, 2000 and 2010
(2015 value extrapolated based on exponential growth)
15
General Practitioner Sector
<pressure from payer to
reduce hospital use>
normal wait time to
admit gp referral
effect of costs on
artificial barriers to direct
referrals
actual wait time to
admit gp referral
chg expected wait
for gp referral
time to adjust
expectation
Expected wait time
fo gp referral
expected over
normal wait time
effect of wait time to
admit on gp referrals to
ed
gp sensitivity to
wait
percent of gp admits
redirected to ed
ED referrals by
gp
direct admits
from gp
<Population>
payer sensitivity
to pressure
desired admits
from gp
total gp visits
per year
gp visits per
capita
percent of gp visits
requiring hospital
admission
16
Unknown Parameters
•
•
•
•
Normal wait times
Actual wait times
Percent of patients referred to ED
BUT, these don’t matter because the model is
normalized, the key unknowns are:
– Payer sensitivity to pressure
– GP sensitivity to pressure
• GP and Payer time to adjust expectations also
unknown
17
Sensitivity Testing
• GP Sensitivity to wait times is the slope of the
Effect of Wait Times on Referrals to ED
• This effect was formulated using a Gompertz
function rather than as a lookup to allow for
easier sensitivity testing
• Effect of Wait Time=
−15𝑒 −(𝐺𝑃 𝑆𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦)( 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑜𝑣𝑒𝑟 𝑛𝑜𝑟𝑚𝑎𝑙 𝑤𝑎𝑖𝑡 𝑡𝑖𝑚𝑒)
𝑒
• Similar for Payer sensitivity
18
Effect of Wait Times on GP Redirects
1.00
0.90
Percent of Patient Redirected
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0
1
1.5
2
2.5
3
3.5
4
4.5
5
5.6
6
6.5
7
7.5
8
Expected over Normal Wait Time
19
Final Value of Per Capita ED Visits
20
In 2 Dimensions
Surface Base Run
50%
75%
95%
100%
ED visits per 1000
400
375
350
325
300
1990
1996
2002
Date
2008
2015
21
22
23
Estimating Sensitive Parameters
• Next step is find data on these sensitive
parameters
• Where? I don’t know yet.
24