Title: Hospital overcrowding and emergency admission mortality: A new imperative for patient safety
1Hospital overcrowding and emergency admission
mortalityA new imperative for patient safety
- Peter Sprivulis MBBS PhD, Julie-Ann Da Silva
BPsych, Ian Jacobs RN PhD, Amanda Frazer MBBS
LLB, George Jelinek MD FACEM - For the Emergency Care, Hospitalization Outcome
Study (ECHO)
2 The ECHO Study
- Links Perths prehospital care, emergency care,
hospital morbidity and death records 2000-2005 - ECHO Investigators include
- Roger Swift, Neil Banham, Simon Wood, Judith
Finn, Gary Geelhoed, Adrian Goudie, Tom
Hitchcock, Jack Hodge, Andrew Jan, Michelle
Johnston, Debra OBrien, Alan OConnor, Paul
Mark, David Mountain, Yusuf Nagree, Greg
Sweetman, Garry Wilkes - The ECHO investigators acknowledge
- The Western Australian Data Linking Unit
- ECHO is funded by
- Australian Health Ministers Advisory Council
- I also acknowledge
- The Commonwealth Fund
3The problem of overcrowding
- High hospital occupancy is associated with
emergency department boarding and emergency
department dysfunction - There is indirect evidence of harm from
overcrowding - Longer hospital lengths of stay
- More representations after not being seen in the
emergency department - Poorer hospital infection control/disaster
responsiveness - Hypothesis
- Overcrowding may be associated with increased
admission mortality
4Aim
- Examine the relationship between hospital
occupancy, emergency department Boarder occupancy
and emergency admission mortality
5Setting
- Three 400 550 bed hospitals
- 130 000 ED attendances
- 50 000 annual emergency admissions
- Over 80 of ambulance attendances
- Nearly 70 of ambulance diversion episodes
6Methods
- Analysis of probabilistically linked data for
first admission of adults between July 2000 and
June 2003 - Emergency department information system records
- Hospital admission records
- State death records
- Linkage conducted by the Western Australian
Department of Health Data linking unit as part of
the Western Australian Linked Data System
Holman CDJ, Bass AJ, Rouse IL, Hobbs MST.
Population-based linkage of health records in
Western Australia Development of a health
services research linked database. Aust N Z J
Public Health 199923453-59.
7Uncrowded Crowded Hospital
Hospital
BEDS Empty Occupied Boarder Outlier
- Low ward occupancy
- Good flow
- Empty beds
- Few ED boarders
- No medical outliers
- High ward occupancy
- Poor flow
- No empty beds
- Many ED boarders
- Medical outliers
8Analysis
- Multivariate analysis
- Survival and Death hazard by day 2, 7 and 30 for
adult admissions associated with - Hospital occupancy at 2359 hours on day of ED
attendance - Number of ED boarders present during hour of
attendance - Definitions
- Boarder admission with a total ED duration of
stay greater than 8 hours - Admission Hospital length of stay greater than
12 hours or died in ED - Controlling for
- Age
- Principal Diagnosis (medical illness vs traumatic
injury) - Transport to hospital (ambulance vs nonambulance)
- Referral source (physician vs nonphysician)
- Triage urgency (Australasian 5 level urgency
scale) - Also tested associations with
- Admission length of stay, ED duration of stay,
ED waiting time to be seen by Physician, ED
attendance rate per hour, Ambulance arrival rate
per hour, Hospital admission rate per hour,
Hospital bed turnover per day, Day of week, Time
of day, Month, Holiday periods, Winter-season,
Hospital attended
9Analysis
- Test for admission selection confounding
- Risk of admission vs overcrowding, multivariate
analysis - Deaths associated with overcrowding
- Characteristics of patients who died who
experienced overcrowded conditions versus those
who didnt experience overcrowding, multivariate
analysis
10Characteristics of study sample
Hospital occupancy group lt90 90-99 100 P
Sample characteristic or mean or mean or mean
Age ( 50 years or older) 62 64 66 lt0.001
Diagnosis ( Injury) 26 24 21 lt0.001
Day ( Mon-Fri) 64 75 80 lt0.001
Winter attendance ( Jun-Sep) 11 40 73 lt0.001
Gender ( Female) 46 47 47 0.05
Shift ( 0800-1559 hours) 44 45 46 0.12
Referral source ( physician referred) 34 34 34 0.36
Transport to emergency ( ambulance) 51 51 51 0.09
Triage urgency ( resuscitation cases) 3.8 3.6 3.7 0.27
Length of stay (mean) 6.6 6.7 6.7 gt0.2
Length of stay (mean), weighted for deaths 6.8 6.9 7.0 gt0.1
Total ( of 62,495 admissions) 26 64 10
For all between group tests
11Hospital occupancy vs 7-day mortality
P 0.002
12Overcrowding vs 7-day survival
Both hospital occupancy and ED boarder occupancy
are associated with increased 7-day mortality
13Overcrowding Hazard Scale
(ED Boarder occupancy at attendance) x (Hospital
occupancy at midnight)
Hospital occupancy () lt90 90-99 100
ED Boarder occupancy () OZS Weight 1 2 3
lt10 1 1 2 3
10-19 2 2 4 6
20 3 3 6 9
14Overcrowding Hazard Scale vs 7 Day survival
P lt 0.001
15Overcrowding Hazard Scale 7-day mortality hazard
P lt 0.001
16Deaths attributable to overcrowding
Censoring date Hazard ratio Hazard ratio Hazard ratio Hazard ratio Attributable deaths / 1000 Emergency hospital admissions Attributable deaths / 1000 Emergency hospital admissions Attributable deaths / 1000 Emergency hospital admissions Attributable deaths / 1000 Emergency hospital admissions P
HR 95CI 95CI 95CI Deaths 95CI 95CI 95CI
Day 2 1.3 1.1 - 1.6 1.0 0.4 - 1.4 0.001
Day 7 1.3 1.2 - 1.5 1.9 0.7 - 2.5 lt0.001
Day 30 1.2 1.1 - 1.3 2.3 1.2 - 3.2 lt0.001
120 deaths per year by Day 30 for Perth (pop.
1.8 million) Suggests over 1000 deaths per annum
for Australia (pop. 20 million)
17Tests for confounding Winter
Overcrowding hazard scale gt 2
Removing winter months data does not affect the
relationship between overcrowding and mortality
18Tests for admission selection confounding
Relative risk of admission associated with
hospital occupancy 1.0 (95CI 1.0-1.1, P
0.001) Relative risk of admission
associated with overcrowding 1.0 (95CI
1.0-1.1, P 0.06) - No evidence of adverse
selection confounding - There is a
very weak relationship between the perception of
bed availability and measured hospital occupancy
19Deaths associated withovercrowded conditions
- Undifferentiated with respect to
- Age
- Diagnosis
- Urgency
- Mode of transport
- Referral source
- Hospital length of stay
- Longer ED durations of stay
- (RR per hour of ED stay, 1.1, 95CI 1.1-1.1, P lt
0.001) - Slightly longer ED physician waiting time
- (RR per hour of ED wait, 1.2, 95CI 1.1-1.3, P
0.01)
20Conclusions
- Overcrowding is associated with increased
mortality - Can not be explained by adverse admission
selection - This suggests overcrowding is a patient safety
concern, not just a workflow issue - The Overcrowding Hazard Scale seems a useful
hazard assessment tool - Easy to measure in real time
- Count ED Boarders and the number of empty beds
- Provides an objective definition for overcrowded
conditions - OZS greater than 2 is associated with increased
mortality - Deaths during overcrowded conditions are
associated with longer ED physician waits and
longer durations of ED stay - Possibly a proxy for delays for time critical care
21Limitations and suggested further research
- No definitive assessment of mediators of the
relationship between overcrowding and harm - Further research is needed into the mediators of
this relationship - Adverse events, medical errors vs overcrowding
- Off usual service ward placement vs adverse
events and medical errors - Assessment of delays in time critical care vs
overcrowding
22Implications for policy
- Hospitals operating between at 90-100 occupancy
need to have near perfect management of patient
flow in and out of beds in order to operate
safely - Operating above 100 appears to be unsafe
- Prudence suggests hospitals should aim to operate
at no higher than 90 occupancy
23Understanding admission decision making
- Admission decisions may be affected by the
perception of poor bed availability - In reality however, there is a very weak
relationship between bed availability and this
perception, as illustrated by the Gap of Neglect
(next slide)
24The gap of neglect
- Over 300 bed variation in demand and supply at
our largest 3 hospitals. - Yet near perfect maintenance of a gap of neglect
of between 25 50 overnight beds - It always feels crowded!
Bed supply and demand 2004