Transcript Probability of survival - PS14

Probability of Survival (Ps)
&
Hospital Survival Rate (Ws)
The Trauma Audit & Research Network (TARN)
Probability of Survival
Once ISS is assigned
Probability of Survival (Ps) calculated each
submission
Why calculate PS?
• Need to assign weight to deaths and survivors.
• Some deaths more statistically significant than others.
• Case mix adjustment.
• Performances measurement: hospital and networks.
History of PS
 TARN developed first Ps model in 2004
 Remodelled in 2007, 2009, 2012
 4 components used in Ps since 2004:
 ISS
 Age
 Gender
 GCS
PS 12 model
Age
Gender
ISS
GCS/
intubation
i i i i
PS 12 calculation
$
63.0%
What is Ps?
PS% is retrospective measure of pts with same profile on TARN database.
TARN database: past 4 years (Approx 200,000 cases)
Ps = 63%, then 63 out of every 100 patients with that profile have
previously survived.
37 out of every 100 patients have previously died.
Probability of Survival
Ps is calculated using:
GCS taken on arrival in ED at first receiving hospital
where unavailable
Pre Hospital GCS
where unavailable
Presence of Intubation/ventilation
where unavailable
Impute a “probable” GCS (equivalent weighting)
PS14 developments
Launched December 2014
1. Pre-Existing Medical Conditions (PMC) added
Charlson comorbidity index (CCI) adds ‘weighting’ PMC
2. True 30 day outcome model introduced
ONS (Office of National Statistics) data linkage using NHS No.
PS 14 –PMC and true 30 day outcome
Launched December 2014
Age
Gender
ISS
GCS/
intubation
i i i i
PS 14 calculation
$
Patient PS
ONS outcome linkage
*Charlson index (1984, revised).
PMC
i
*
PS14 Weighting groups
Pre-Existing Medical Conditions
Weight
PMC group
0
Bone conditions
Connective tissue disorder
Diabetes
GU disease
HIV
Mental health
Neurological disorders
Nil
Paraplegia
Pulmonary disease
1-5
Alcohol
Blood conditions
Cancer
CVA
Congestive heart failure
Dementia
MI
Other conditions
Peripheral vascular disease
6-10
Metastatic cancer/Haematological
malignancy
>10
Liver disease
Renal disease
PS 14
Weightings for Age, GCS, Gender & PMC
Age: 25
Gender: Male
ISS: 25
GCS: 15
PMC:
i
i
i
i
i
PS 14 calculation
$
Ps: 98.7%
Nil
PS 14
Weightings for Age, GCS, Gender & PMC
Age: 25
Gender: Male
ISS: 25
GCS: 15
i
i
i
i
PS 14 calculation
$
Ps: 97.8%
PMC:
Alcohol abuse
i
PS 14
Weightings for Age, GCS, Gender & PMC
PMC:
Age: 25
Gender: Male
ISS: 25
GCS: 15
Alcohol Abuse
Liver Disease
i
i
i
i
i
PS 14 calculation
$
Ps: 93.7%
PS 14
Weightings for Age, GCS, Gender & PMC
PMC:
Age: 25
Gender: Male
ISS: 25
GCS: 3
Alcohol Abuse
Liver Disease
i
i
i
i
i
PS 14 calculation
$
Ps: 25.3%
PS 14
Weightings for Age, GCS, Gender & PMC
PMC:
Age: 65
Gender: Male
ISS: 25
GCS: 3
Alcohol Abuse
Liver Disease
i
i
i
i
i
PS 14 calculation
$
Ps: 6.5%
PS 14
Weightings for Age, GCS, Gender & PMC
PMC:
Age: 65
Gender:
Female
ISS: 25
GCS: 3
Alcohol Abuse
Liver Disease
i
i
i
i
i
PS 14 calculation
$
Ps: 7.06%
PS 14
Importance of accurate injury detail
Full injury detail
Code
Subdural haematoma bilateral & 2cm thick on both sides
140655.5
Base of skull fracture
150200.3
5 ribs fractured on left
450203.3
Full thickness Rectal laceration
543624.3
Spiral Fracture of left Shaft of Femur
853251.3
Vertical Shear fracture to pelvis with blood loss >20%
856173.5
Open Comminuted fracture to Tibial Shaft
854272.3
Incomplete injury detail
Code
Subdural haematoma
140650.4
Base of skull fracture
150200.3
Ribs fractured on left
450210.2
Rectal laceration
543620.2
Fracture of left Shaft of Femur
853221.3
Pelvic Fracture
856151.2
Tibial Shaft Fracture
854221.2
Accurate ISS
59
Incomplete ISS
29
Accurate Ps
49%
Ps 14
Importance of accurate injury detail
Age: 40
Gender: Male
ISS: 29
GCS: 5
PMC:
i
i
i
i
i
PS 14 calculation
$
Ps: 79%
Nil
PS breakdown: 4 years Calendar data: Shown on Website
•
•
•
•
•
•
Ps Bandings: Expanded for PS14
No. of patients in each band
No. of expected survivors for each band
Actual number of survivors
Difference = Actual – Expected x 100
No. in group
Adjusted Difference = Difference x fraction of pts on TARN database
Hospital Survival Rate: 4 years Calendar data
+1.2 Survivors
Statistically significant outcome
(+0.39 to +1.97)
+1.2 Survivors
Not Statistically significant
(-1.83 to +4.2)
• Total Ws shown
• Yearly Ws shown
Comparative Outcome Analysis (Ws) graph
• Compares outcomes between all submitting Hospitals
• 4 Comparative Outcome graphs included in Clinical Reports:
2 x Caterpillar plots (showing outcomes by Survival Rate)
1.
Outcome at 30 days or discharge (whichever is sooner)
2.
True 30 day outcome (linked to ONS data)
CATERPILLAR PLOT: Ascending Survival rate
Your hospital
95% confidence
intervals
All hospitals
Comparative Outcome Analysis (Ws graphs)
• Compare Outcomes between all submitting Hospitals
• 4 Comparative Outcome graphs included in Clinical Reports:
2 x Caterpillar plots (showing outcomes by Survival Rate)
1. Outcome at 30 days or discharge (whichever is sooner)
2. True 30 day outcome (linked to ONS data)
2 x Funnel plots (showing outcomes by Precision –no. of cases)
1.
Outcome at 30 days or discharge (whichever is sooner)
2.
True 30 day outcome (linked to ONS data)
FUNNEL PLOT: Precision (no. of cases)
Your hospital
All hospitals
Lower Precision: Fewer cases
(not as reliable)
Greater Precision: More cases
(more reliable)