Trauma Scoring System and Validation of ICISS (ICD based Injury Severity Score) 2004 10 26
Outline Utility of trauma scoring system Trauma scoring systems Physiologic measure Anatomic measure Combination of physiologic & anatomic score ICISS validation studies Potential future works
Utility of Trauma Scoring System Measurement and documentation of injury severity Pre-requisite for the development, evaluation, and improvement of trauma care system and injury control Triage Augment clinical judgment of pre-hospital personnel with respect to on-scene treatment and transport Evaluation of outcome Compare the performance of a trauma care facility against a standard Assess the quality of care by a trauma care facility Preliminary outcome-based evaluation (PRE) chart: Identify patients with unexpected outcome Comparison of effectiveness of treatment modalities Epidemiologic data collection: Trauma registry
Risk Adjustment & Trauma Score Patient Factors Treatment Effectiveness + + = Random Event Patient Outcome,,,,,, /,
PRE Chart (Preliminary Outcome Based Evaluation)
Trauma Scoring Systems Physiologic score GCS (Glasgow Coma Scale) RTS (Revised Trauma Score) CRAMS (Circulation, Respiration, Abdominal/thoracic, Motor, Speech scale) APACHE (Acute Physiologic and Chronic Health Evaluation) Anatomic score AIS (Abbreviated Injury Scale) ISS (Injury Severity Score) ICISS (ICD based Injury Severity Score) Combination of physiologic & anatomic score TRISS (Trauma and Injury Severity Score) ASCOT (A Severity Characterization of Trauma) ICISS full model: ICISS+RTS
Glasgow Coma Scale Eye Opening Response Spontaneous--open with blinking at baseline 4 To verbal stimuli, command, speech 3 To pain only (not applied to face) 2 No response 1 Verbal Response Oriented 5 Confused conversation, but able to answer questions 4 Inappropriate words 3 Incomprehensible speech 2 No response 1 Motor Response Obeys commands for movement 6 Purposeful movement to painful stimulus 5 Withdraws in response to pain 4 Flexion in response to pain (decorticate posturing) 3 Extension response in response to pain 2 (decerebrate posturing) No response 1
Glasgow Coma Scale Score range: 3~15 Head Injury Classification: Severe Head Injury----GCS score of 8 or less Moderate Head Injury----GCS score of 9 to 12 Mild Head Injury----GCS score of 13 to 15 (Adapted from: Advanced Trauma Life Support: Course for Physicians, American College of Surgeons, 1993) Advantages Correlated w/ mortality Widely used for pre-hospital triage and determining level of consciousness Incorporated into the RTS
Revised Trauma Score (RTS) RTS =0.9368(GCS) + 0.7326(SBP) + 0.2908(RR) GCS: Glasgow Coma Scale SBP: systolic blood pressure RR: respiration rate
Revised Trauma Score (RTS) Coded Value GCS SBP RR 4 3 2 1 0 13-15 9-12 6-8 4-5 3 >89 76-89 50-75 1-49 0 10-29 >29 6-9 1-5 0 * GCS: Glasgow Coma Scale, SBP: Systolic Blood Pressure, RR: Respiration Rate
RTS (Revised Trauma Score) Score range: 0~7.84 Development Evolved from Triage Index and Trauma Score Subnormal value for any RTS variable: transport to trauma center Advantages Simple Widely used for prehospital triage More accurate prediction of outcome than Trauma Score Disadvantage Too many missing data for GCS in Korea
> RTS GCS 13, 70, 13, RTS? RTS = 0.9368(GCS) + 0.7326x( ) + 0.2908x( ) 6.3756 = (0.9368 x 4) + (0.7326 x 2) + (0.2908 x 4)
CRAMS (Circulation, Respiration, Abdominal/thoracic, Motor, Speech scale) Development Attempt to simplify the original trauma score for field triage Score for each 5 category 0 (severe physiologic/neurologic deficit, abdominal/thoracic injury) 2 (no deficit OR injury) Respiratory: Normal(2), Labored/Shallow(1), Absent(0) Major trauma: 8 or less Minor trauma: 9 or above
APACHE (Acute Physiologic and Chronic Health Evaluation) Development Classification system for ICU Variables used Values at ICU admission OR Worst values during 1 st 24 hours after ICU admission Score range: 0~299 Age Acute physiology: Temperature, Mean arterial pressure, Heart rate, Respiration rate, GCS, Oxygenation, Arterial ph, Serum sodium/ potassium/ creatinine, Hct, WBC, Chronic health : AIDS, Hepatic failure, Lymphoma, Metastatic cancer, Leukemia, Multiple myeloma, Immunosuppression, Chirrosis
APACHE (Acute Physiologic and Chronic Health Evaluation) APACHE II 12 variables Standardized coefficients for each patient group Treatment Outcome Admission eligibility Trauma patient group Post-operative/Non-operative Multiple trauma/head injury APACHE III Better prediction Commercial product
AIS (Abbreviated Injury Scale) Development Developed in 1969 for assessing severity of MVA List of several hundreds of injuries 6 Body region Head/Neck, Face, Thorax, Abdomen/Pelvic, Extremities, External 0(minor)~6(fatal) Revision: AIS-71 (original), AIS-85, AIS-90 Building block of ISS, standard injury severity score
AIS (Abbreviated Injury Scale) A.I.S SCOR ES HEAD & NECK 1 MINOR Headache/dizz iness 2ndary to head trauma Cervical spine strain with no fracture or dislocation 2 MODERATE Amnesia from accident Lethargic/ stuporous obtunded; can be aroused by verbal stimuli Unconsciousne ss 1 hr Simple vault fracture Thyroid contusion Branchial plexus injury Dislocation or fracture 3 SEVERE NOT LIFE THREATENING Unconsciousness 1-6 hrs Unconsciousness < 1 hr with neurological deficit Fracture base of skull Comminuted compound or Depressed vault fracture Cerebral contusion/ Subarachnoid hemorrhage Intimal tear/thrombosis carotid A. Contusion larynx, 4 SEVERE LIFE THREATENING Unconsciousne ss 1-6 hrs with neurological deficit Unconsciousne ss 6-24 hrs Appropriate response only to painful stimuli Fractured skull with depression > 2cm, lac dura or tissue loss Intracranial hematoma 5 CRITICAL SURVIVAL UNCERTAIN Unconsciousne ss with inappropriate movement Unconscious > 24 hrs Brain stem injury Intracranial hematoma > 100 cc Complete cervical cord lesion C4 or below
AIS to ICD Conversion AIS ICD-9CM Disadvantage 1:M mapping No appropriate AIS code for some ICD codes Not always correct assignment Coding quality: creeping Limited # of ICD codes in a discharge summary
ISS (Injury Severity Score) ISS = AIS(1) 2 +AIS(2) 2 +AIS(3) 2 Need a summary score based on AIS for multiple injuries Value from 1 to 75 Patients w/ AIS 6 injury assign ISS 75
Sample ISS Score Region Injury Description AIS Square Top Three Head & Neck Cerebral Contusion 3 9 Face No Injury 0 Chest Flail Chest 4 16 Abdomen Minor Contusion of Liver Complex Rupture Spleen 2 5 25 Extremity Fractured femur 3 External No Injury 0 Injury Severity Score: 50
ISS (Injury Severity Score): Disadvantages Discounting the importance of body region Mortality rates for subsets of ISS=16 cohort Head/neck 17.2% Face 0.0% Thorax 6.1% Abdomen 10.5% Limits in considering severity of multiple injuries Only consider 3 most severe injuries
MTOS (Major Trauma Outcome Study) Objective Establish national normative outcome for trauma Provide trauma care institutions with objective data for evaluation of their quality assurance and outcome results Methods Begun 1982 More than 170k seriously injured patients included About 160 North American hospitals submitted data to the MTOS Combination of physiologic and anatomic measure for evaluation of injury severity = TRISS method
TRISS (Trauma & Injury Severity Score) TRISS b=b0+b1(rts)+b2(iss)+b3(age) Ps=1/(1+e -b ) MTOS regression coefficients b0 b1 b2 b3 Blunt Penetrating -1.2470-0.6029 0.9544 1.1430-0.0768-0.1516-1.9052-2.6676
TRISS (Trauma & Injury Severity Score) Disadvantage Requirement for independent data collection system Inability to consider more than 3 injuries in deriving survival probability
: TRISS RTS 6, ISS 18, 35, TRISS? 1 TRISS. 2 b = b0 + b1 x (RTS) + b2 x (ISS) + b3 x ( ) = -1.2470 + (0.9544x6) + (-0.0768x18) + (-1.9052x0) 3.097 = -1.2470 + 5.7264-1.3824 + 0 3 : (Ps) = 1/(1+e-b) e = 2.718 e-3.097 = 0.0452 (Ps) = 1/(1+e-b) = 1/(1+0.0452) = 0.9568
ICISS (ICD based Injury Severity Score) SRR (Survivor Risk Ratio) SRR ICDj = Number of patients that survived with ICD j Number of patients with ICD j ICISS survival probability ICISS = SRR ICD(1) x SRR ICD(2)... x SRR ICD (10) Liver Laceration (0.9), EDH (0.8) ICISS=0.9*0.8=0.72 ICISS full model b=b0+b1(rts)+b2(iciss)+b3(age) Advantage Use of multiple injuries in deriving survival probability Much less effort required for data collection than ISS and TRISS
SRR Table: Example ICD code ICD code description Total patients Live Dead SRR S00 Superficial injury of head 11 11-1.000 S000 Superficial injury of scalp 1,640 1,588 52 0.968 S001 Contusion of eyelid and periocular area 425 422 3 0.993 S002 Other superficial injuries of eyelid and periocular area 167 167-1.000 S003 Superficial injury of nose 100 100-1.000 S004 Superficial injury of ear 108 108-1.000 S005 Superficial injury of lip and oral cavity 171 169 2 0.988
ICISS (ICD based Injury Severity Score) ICD (Trauma registry) ISS ISS, TRISS
Risk-adjusted Outcome Risk(severity) + Quality = Outcome Z-score Z = (A-E)/S S = Pi*(1-Pi) +1.96 < or > -1.96 W-score W=(A-E) / (N/100)
Study(1): Objective To evaluate the predictive validity of ICD-10 based ICISS as a predictor of mortality for patients with blunt injuries By comparing with ISS, TRISS, and ICD- 9CM based ICISS
Study(1): Framework North North Carolina Trauma Database (89,827 patients) TRISS TRISS Database of of Two Two EMCs EMCs (367 (367 patients) Korean Korean EMC EMC Trauma Database (47,750 patients) ICD-9CM SRR SRR Survey Survey of of ICD-9CM & ICD-10 ICD-10 codes codes ICD-10 ICD-10 SRR SRR Predictive Validity Validity of of ICD-9CM based based ICISS ICISS Predictive Validity Validity of of ICD-10 ICD-10 based based ICISS ICISS Predictive Validity Validity of of ISS ISS and and TRISS TRISS
Study(1): Measure of Performance Measures of Discrimination Disparity, Sensitivity(%), Specificity(%) Misclassification rate(%) Area under the ROC (receiver operating characteristic curve) Goodness-of-fit statistics Hosmer-Lemeshow statistic
ICISS vs. ISS: All Blunt Injury ISS ICD-9CM based ICISS ICD-10 based ICISS Disparity 0.245 0.378 0.194 Sensitivity(%) 43.6 56.4 38.5 Specificity(%) 97.2 97.9 94.1 Misclassification rate(%) 14.2 10.9 17.7 ROC analysis 1) 0.892 0.909 0.843 H-L Statistic 2) 9.381 (p=0.226) 12.891 (p=0.116) 1) ROC analysis : Receiver Operating Characteristic analysis 2) H-L Statistic: Hosmer- Lemeshow Statistic 5.147 (p=0.742)
ICISS Full Model vs. TRISS: All Blunt Injury TRISS ICD-9CM based ICISS Full Model ICD-10 based ICISS Full Model Disparity 0.644 0.737 0.627 Sensitivity(%) 75.6 82.1 73.1 Specificity(%) 96.9 98.3 96.2 Misclassification rate(%) 7.6 5.2 8.7 ROC analysis 1) 0.958 0.976 0.956 H-L Statistic 2) 3.406 (p=0.906) 7.738 (p=0.460) 1) ROC analysis : Receiver Operating Characteristic analysis 2) H-L Statistic: Hosmer- Lemeshow Statistic 7.294 (p=0.505)
ICISS Full Model vs. TRISS: Intracranial Injury TRISS ICD-9CM based ICISS Full Model ICD-10 based ICISS Full Model Disparity 0.684 0.769 0.629 Sensitivity(%) 79.7 84.7 76.3 Specificity(%) 94.4 96.0 93.6 Misclassification rate(%) 10.3 7.6 12.0 ROC analysis 1) 0.829 0.882 0.791 H-L Statistic 2) 4.948 (p=0.763) 9.053 (p=0.338) 1) ROC analysis : Receiver Operating Characteristic analysis 2) H-L Statistic: Hosmer- Lemeshow Statistic 3.417 (p=0.906)
ICISS Full Model vs. ISS: Non-intracranial Injury TRISS ICD-9CM based ICISS Full Model ICD-10 based ICISS Full Model Disparity 0.511 0.626 0.596 Sensitivity(%) 57.9 57.9 63.2 Specificity(%) 98.2 98.8 98.2 Misclassification rate(%) 6.0 5.5 5.5 ROC analysis 1) 0.938 0.979 0.97 H-L Statistic 2) 2.929 (p=0.939) 0.968 (p=0.998) 1) ROC analysis : Receiver Operating Characteristic analysis 2) H-L Statistic: Hosmer- Lemeshow Statistic 3.618 (p=0.890)
Through comparing the agreement between judgments derived from ICISS survival probability and those derived from a professional panel method Study(2): Objectives To evaluate the utility of ICISS To detect preventable deaths To compare the performance of trauma care facilities
Study(2): Framework 2 Tertiary 6 EMCs // 4 Non-tertiary Trauma Deaths (131) Trauma Inpatients (1,087) ICISS Survival Probability Professional Panel Review on on Preventability W-scores of of Each EMC using ICISS Agreement Rates Correlation Coefficient
Study(2): Data Collection Two professional panels Each panel: emergency physician(1), general surgeon(1), neurosurgeon(1) Review process Independent review by each panel member using structured review form Decision rule: Unanimous agreement rule Time for review per case: about 1 hour Preventability: Preventable (P): Ps 0.75 Potentially preventable (PP): Ps 0.25~0.75 Non-preventable (NP): Ps < 0.25 Data collection for ICISS RTS : GCS, SBP, RR Maximum of 10 ICD-10 codes
Study(2): Analysis Agreement of judgment on preventable deaths Overall agreement rates Kappa statistics Correlation between the preventable death rate and the W-score in each EMC W-statistic = (A-E)/(n/100) Represents excess survival or excess mortality per 100 patients after adjusting severity of trauma patients A: Actual Survivors E: Expected Survivors N : Number of Observations Spearman correlation coefficient
Agreement between the Preventability of the Panel and ICISS Survival Probability Prevent -ability ICISS 0.75 0.25-0.75 0.25< Total P 14 ( 35.0) [ 10.7] 3 ( 7.0) [ 2.3] 0 ( 0.0) [ 0.0] 17 ( 13.0) P/PP PP 16 ( 40.0) [ 12.2] 13 ( 30.2) [ 9.9] 7 ( 14.6) [ 5.3] 36 ( 27.5) Subtotal 30 ( 75.0) [ 22.9] 16 ( 37.2) [ 12.2] 7 ( 14.6) [ 5.3] 53 ( 40.5) NP 10 ( 25.0) [ 7.6] 27 ( 62.8) [ 20.6] 41 ( 85.4) [ 31.3] * P: Preventable, PP; Potentially preventable, NP; Nonpreventable 78 ( 59.5) Total 40 (100.0) [ 30.5] 43 (100.0) [ 32.8] 48 (100.0) [ 36.6] 131(100.0) [ 100.0]
Agreement between the Preventability of the Panel and ICISS Survival Probability 2-way classification: P/PP, NP Overall agreement: 66.4% Kappa : 0.36 McDermott(1996) : 65.6% (0.35) 3-way classification: P, PP, NP Overall agreement: 51.9% Kappa : 0.26
ICISS - - : 55.4% (46/83) Cayten (1991) : 21.5% Karmy-Jones (1992) :16.1% : 85.4% (41/48)
Correlation between W-score W and Preventable Death Rate by Emergency Medical Center 60.0 (-8.98, 47.8) (0.56, 55.0) (-0.11, 50.0) Preventability(%) 40.0 20.0 (1.54, 37.0) (2.66, 34.6) (4.44, 21.1) 0.0 Spearman=0.77(p=0.07) -9.00-6.00-3.00 0.00 3.00 6.00 9.00 W-score *Numbers in parentheses refer to W-score and preventable death rate of each emergency medical center.
Conclusion The ICISS methodology can be extended to ICD-10 horizon as a standard injury severity measure in the place of TRISS for blunt injury. The ICISS is useful in detecting preventable deaths and in comparing the performance of trauma care
Potential Future Work Development of standardized coefficients of a ICISS prediction model based on a large scale study Possible to evaluate quality of care of emergency medical centers based on national standards Similar to MTOS (major trauma outcome study) of North America Development of ICISS prediction models considering various mechanisms of injury which could affect validity of trauma scoring system Blunt injury: Motor vehicle accident, low fall, other blunt injury Penetrating injury: gunshot wound, stab wound
Questions?