* Vimal Kumar, Nitesh, M. Mohan Kumar, Prof. C. Subramanian

* Dept. of Orthopaedics, Sri Ramachandra Medical College and  Research Institute, Porur, Chennai, Tamil Nadu, India.

Address for Correspondence
Vimal Kumar,
Dept. of Orthopaedics, Sri Ramachandra Medical College and  Research Institute, Porur, Chennai, Tamil Nadu, India, 600 116.

                                                      TABLE 2

TRISS METHODOLOGY

Estimation of the probability of survival (Ps) was done by using the formula

Ps = 1/ (1 + e^-b)

Where b = b0 + b1 (RTS) + b2 (ISS) + b3 (AGE). The constant e is equal to 2.718282. b0,b1,b2 and b3 are coefficients derived from Walker – Duncan regression analysis applied to data from thousands of patients analyzed in the Major Trauma Outcome Study (MTOS) and are -1.2470, 0.9544, -0.0768 and -1.9052 respectively. RTS (Revised Trauma Score) the physiologic component of TRISS is

RTS = 0.9368 (GCS CODE) + 0.7326 (SYSTOLIC BP CODE)  + 0.2908 (RESPIRATORY RATE CODE).

                                                                        TABLE 3

ISS is the anatomic component of TRISS and is based on the Condensed Abbreviated Injury Scale (CAIS), 1985. Each of the six body regions was scored with the three highest AIS values given to any injury in that area. The AIS values for the three highest scoring body regions were squared and summed to form the ISS. NISS is similar to ISS except here squares of AIS of patients 3 most severe injuries regardless of body region is used (9).Age is coded as 1 if the patient is at least 55 years and 0 if otherwise.The results were Probability of Survival of TRISS using ISS (Ps TRISS (ISS)) and Probability of Survival of TRISS using NISS (Ps TRISS (NISS)).

ASCOT METHODOLOGY

Like TRISS, this includes description of anatomic injury of the patient, physiology on admission and patient’s age. ASCOT uses the three Anatomic Profile (AP) components (A, B and C) to describe anatomic injury and three coded value of RTS to describe physiology. Patients age is modeled more precisely than TRISS. AP component A is a summary score of all serious injuries (AIS>2) injuries to head, brain and spinal cord. B summarizes all serious injuries to the thorax and the front of neck. C summarizes all other serious injuries and D is the summary score of all non serious injuries (AIS 1 and 2). Each components value is computed using a generalization of Pythagorean Theorem for distance from its uninjured state equals the square root of the sum of the squares of the AIS in the component.

Ps = 1 / (1+ (e^-K))         e = 2.718282

K = k1 +(k2 * GCS CODE) + ( k3 * SYSTOLIC BP CODE ) + ( k4 * RESPIRATORY RATE) +
( k5 * A ) + ( k6 * B ) + ( k7* C ) + ( k8 * AGE CODE)

                                                TABLE 4

                                                TABLE 5

D component was omitted since they did not significantly change the outcome as per previous studies.

Results

Outcome

• Total number of patients studied – 245

• Patients survived                         – 224 (91.4%)

• Patients dead                               –   21  (8.6%)

By Wayne S. Copes in 1988, survival was 94.2% and non-survival was 5.8% with sample size of 11,195 patients from MTOS. By Howard R. Champion in 1996, survival was 90.3% and death was 9.7% with sample size of 15,374 patients from MTOS database.

Sex

• MALES      -    212 (86.5%)

• FEMALES -      33 (13.5%)

By Howard R. Champion in 1996, 77% were male and 23% were females with sample size of 15,374 patients from MTOS database. The mortality rate for females is 6.06% (2 patients) when compared to 8.9% (19 patients) in males.

Various parameters were compared with the outcome of the patient.

Age and Outcome

STATISTICALLY NOT SIGNIFICANT (P > 0.05)

Pediatric and elder patients accounted to about 7.8%. Pediatrics did not have any mortality. 84.5% was formed by patients of age group 16 – 54 yrs with mortality rate of 8.2%. Elderly patients had a mortality rate of 21%. By Howard R. Champion in 1996, from MTOS data base 6.2% were less than 15 yrs, 84.9% were between 15 and 54 and 8.6% were above 55 yrs.

Mode of Injury and Outcome

STATISTICALLY NOT SIGNIFICANT (P > 0.05)

Two wheeler accidents were more common accounting for >50% of mode of injury with a mortality of 8.7%. Mortality from Four wheeler and pedestrians’ injuries were 6.6% and 7.4% respectively. Highest mortality rates were noted in patients who fell from height accounting to 16.6%. As per MTOS study in 1988 pedestrian mortality was highest accounting to about 10.8% followed by two wheeler injuries (7.9%), four wheeler injuries (6.5%) and fall (4.6%)(4). The high mortality rate in patients injured due to fall was probably due to the inadequate anatomical scoring.

Arrival and Outcome

STATISTICALLY NOT SIGNIFICANT (P > 0.05)

Most of the patients (46.1%) were brought to emergency between 1 to 6 hours of injury and only 38.4% were brought within the golden hour of first 60 minutes. Mortality was highest in those who attended emergency in 1 – 6 hrs with rate of 9.7%, followed by patients brought in golden hour (8.5%). Patients admitted between 6 to 24 hours and more than 1 day had a mortality rate of 4.1% and 7.1% respectively.

RTS and Outcome

STATISTICALLY SIGNIFICANT  (P < 0.001)

The nearest possible value of mean is taken as a cut off below which the survival probability is very less and is statistical significant. The mortality rate of patients with RTS less than 5.96 is 80%. Sensitivity of 38.0% and Specificity of 99.10%.

ISS and Outcome

STATISTICALLY SIGNIFICANT (P < 0.001)

Major trauma is defined as ISS equal to or more than 16 (19.6%). In our study the mortality rate for patients with ISS more than 16 is 37.5%. Mortality rate is 1.5% if ISS is less than 16.The sample study consisted of consecutive admissions and therefore included patients with relatively minor injuries (ISS < 9) accounting to 56.3% (138 patients). Sensitivity is 85.71% and Specificity of 86.60%. 

NISS and Outcome

 STATISTICALLY SIGNIFICANT (P < 0.001)

Major trauma is defined as NISS equal to or more than 16 (28.6%). In our study the mortality rate for patients with NISS more than 16 is 27.1%. Mortality rate is 1.1% if NISS is less than 16. Minor injury patients (NISS < 9) were 48.1% (118 patients). Sensitivity of 90.47% and Specificity of 77.23%. 

TRISS .ISS and Outcome

STATISTICALLY SIGNIFICANT (P < 0.001)

The average Ps of TRISS.ISS for survivors was 0.993 as compared to the 0.976 (MTOS study in 1996 with sample size of 9,178 patients). The average Ps of TRISS.ISS for non-survivors was 0.784 in our study as compared to the 0.362 in MTOS study.  In our study the mortality rate, if Ps value is less than 0.94, is 84.61%. Sensitivity is 52.38% and specificity 99.10%.

TRISS .NISS and Outcome

STATISTICALLY SIGNIFICANT (P < 0.001)

The average Ps of TRISS.NISS for survivors and non-survivors were 0.99 and 0.729 respectively.  In our study the mortality rate, if Ps value is less than 0.94, is 73.33%. Sensitivity is 52.38% and specificity of 98.21%.

Ascot and Outcome

STATISTICALLY SIGNIFICANT (P < 0.001)

The average Ps of ASCOT for survivors was 0.988 as compared to the 0.976 (MTOS study in 1996 with sample size of 9,178 patients). The average Ps of ASCOT for non-survivors was 0.817 in our study as compared to the 0.336 in MTOS study. In our study the mortality rate, if Ps value is less than 0.94, is 83.33%. Sensitivity is 47.61% and specificity is 99.10%. 

MODEL EVALUATION MEASURES :

• Disparity – difference between the average survival probabilities for survivors and non-survivors

• Sensitivity – percentage of  survivors with estimated Ps > 0.50

• Specificity – percentage of  non-survivors with estimated Ps < 0.50

• Misclassifications – number (percentage) of patients misclassified by a rule that predicts survival for patients with Ps < 0.50 and non-survival for patients with Ps > 0.50.

                                                            Ps of 0.5 was taken to be the demarcating point. Those with Ps value above were predicted to survive and those below were predicted to die. By TRISS.ISS in our study our predicted deaths were 3 but observed were 21 including the 3 cases which were predicted. Similarly by TRISS.NISS the predicted deaths were 5 and by ASCOT the predicted deaths were 3 but observed were 21 including the 5 and 3 cases which were predicted. NISS, as an independent entity had highest sensitivity (Table 6).

Results

Our study had 8.6% mortality compared to 9.7% in Champions series in 1996. Males were most commonly injured. Patients between 16 – 54 years of age had 8.2% mortality which is comparatively less than mortality of 21% for patients more than 55 years. Two wheeler injuries is the commonest accounting to 51.4% and mode of injury did not significantly affect the outcome. Highest mortality was noticed in patients who fell from height. Time of arrival of the patient in emergency after injury did not significantly affect outcome even if patients were brought in golden hour. RTS, ISS and NISS each were independently statistically significant with outcome. RTS score of less than 5.96 has a mortality rate of 80%. ISS and NISS of more than 16 are considered as Major Trauma with mortality of 37.5% and 27.1% respectively. Sensitivity is highest for NISS whereas Specificity is highest for TRISS.ISS, TRISS.NISS and ASCOT. All combined scoring systems have accurately predicted death with Ps < 0.5 but with low sensitivity and were statistically insignificant (Table 6).

Discussion

Among the more than 50 scoring systems available for quantitative evaluation of injury severity, only a few have proved effective in clinical practice. In particular, the Revised Trauma Score (RTS) is referring to physiological variable and the Injury Severity Score (ISS) referring to anatomic data. There is a tendency in the development of new scoring systems to aim at higher predictive accuracy, forfeiting practicability. The initial purpose of scoring an early assessment of the risks is being pushed into the background. The TRISS method, which includes the RTS, ISS and Patients age, is regarded as the international standard. However, in our study it has the disadvantage of a low sensitivity of 14% for blunt trauma, resulting in a high rate of unexpected deaths.

In an attempt to increase the sensitivity of the TRISS we combined the physiologic scoring (RTS) with New Injury Severity Score (NISS) instead of the standard Injury Severity Score (ISS). This would be opt from the results, as RTS had the highest individual specificity and NISS had the highest individual sensitivity. The results of which were shown in the table 6. The sensitivity improved only to 24%.

In a novel way to still improve the sensitivity of TRISS we shifted the midline value to a point where atleast 50% sensitivity could be attained without much alteration in specificity.

If Ps of 0.94 is taken to be a demarcating point then those with Ps value above 0.94 were predicted to live and those below are predicted to die. The results were given in table 7.

                                                            TABLE 7

If Ps of 0.94 is kept as demarcating point Sensitivity of TRISS and ASCOT improves to around 50%. This also made the combined scoring results statistically significant. In our study we found TRISS does relatively better than ASCOT in terms of sensitivity. The ASCOT method, in which the ISS is replaced by the Anatomic Profile and the age of the patient is given more consideration, hardly brings better results in spite of quiet time consuming methods. Nevertheless, they are an important scientific instrument for comparative examinations and indispensable for quality assurance and economic analyses.

Conclusion

Based on our study, TRISS based on NISS can be used as a dependable method for predicting survival of Indian blunt injury victims with a modification of midline to 0.94.

Our study documented some potential benefits of using Trauma Scoring systems but however with a caution as we used a relatively small sample size from one institution over a short time interval. General recommendations cannot be made until further results from other emergency units of India are available.

FUTURE

Preexisting conditions, time elapsed before starting treatment and male gender are related to survival of patients and should be included along with age and the various physiologic and anatomical measures currently being used to predict survival for those patients. Patients who had undergone initial treatment in form of fluid therapy may show altered physiological parameters which should also be taken into consideration. To improve the predictive accuracy, biochemical parameters and chronic diseases should be considered in addition to existing scores.

Present injury severity instruments using Western population derived coefficients though accurately predict death, were not satisfactorily predicting survival of trauma patients.

Another limitation of these scoring systems is its focus on mortality as the primary means of assessing the quality of trauma and prediction of mortality as the only outcome measure in the care of an injured patient. Its application has to be also extended to measure the morbidity, disability, cost and length of stay in hospital. The newer development in scoring by using ICD classification and calculating ISS (ICISS) marks a absolute necessity of Trauma Registries to be maintained in segments of India, to keep in pace with emerging trends.

References

1. Rating the Severity of Tissue Damage I. The Abbreviated Scale ; Committee on Medical Aspects of Automotive Safety ; JAMA, Jan 11, 1971, Volume 215; No 2.

2. The Abbreviated Injury Scale, 1985 Revision : A condensed chart for clinical use ; Ian D. Civil and C.William Schwab ; The Journal of Trauma; Volume 28, No 1; Page 87 – 90.

3. Assessment of Severity of Injury; D.W.Yates; Scientific Foundation of Trauma

4. The Injury Severity Score Revisited; Wayne S. Copes, Howard R. Champion, William J. Sacco; The Journal of Trauma; Volume 28, No 1; Page 69 - 77.

5. Evaluating Trauma Care: The TRISS Method; Carl R. Boyd, Mary Ann     Tolson and Wayne S. Copes; The Journal of Trauma; Volume 27, No 4; Page 370 – 378.

6. Progress in Characterizing Anatomic Injury; Wayne S. Copes, Howard R. Champion, Mary Lawnick and William J. Sacco; The Journal of Trauma; Volume 30, No 10; Page 1200- 1207.

7. Assessment of Penetrating Injury Severity; Ronald F. Bellamy, Joshua S. Vayer; Advances in Trauma; 1988; Volume 3; 163 – 261

8. Differential Performance of TRISS- Like in early and Late Blunt Trauma Deaths; Bryan G. Garber, Paul C. Hebert, George Wells; The Journal of Trauma; Volume 43, No 1; Page 1 – 7.

9. A Modification Of the Injury Severity Score that both improves accuracy and simplifies Scoring; Turner Osler, Susan Baker and William Long; The Journal of Trauma; Volume 43, No 6; Page 922 – 926.

10. A New Method for Estimating Probability of Survival in Pediatric Patients Using Revised TRISS Methodology Based on Age Adjusted Weights; Laura C. Schall, Douglas A. Potoka and Henri R. Ford; The Journal of Trauma; Volume 52, No 2;Feb 2002; Page 235 – 241.
11. TRISS Unexpected Survivors: An Outdated Standard; Rob Norris, Randy Woods; The Journal of Trauma; Volume 52, No 2; Page 229 – 235.
12. Outcome after Major Trauma: 12 Month and 18 Month Follow up results from the Trauma Recovery Project; Troy L.Holbrook, John . Anderson; The Journal of Trauma; Volume 46, No 5; Page 765 – 773.
13. Measuring Injury Severity: Time for a Change? ; Frederick D. Brenneman; The Journal of Trauma; Volume 44, No 4; Page 580 – 582.
14. A simple Mathematical Modification of TRISS Markedly Improves Calibration; Turner M. Osler; The Journal of Trauma; Volume 53, No 4;oct 2002; Page 630 – 634.
15. Injury Severity Grading in Trauma Patients: A Simplified Technique Based upon ICD-9 coding. Robert Rutlege, Samir Fakhry, Christopher Baker and Dale Oller; The Journal of Trauma; Oct 1993, Page 497 – 507.
16. The End of the Injury Severity Score (ISS) and the Trauma and Injury Severity Score (TRISS); Robert Rutledge and Turner Osler; The Journal of Trauma; Volume 44, No 1; Page 41 – 49.
17. Comparision of the Injury Severity Score and ICD-9 Diagnosis codes as Predictors of Outcome in Injury: Analysis of 44,032 Patients; Robert Rutledge and Turner Osler; The Journal of Trauma; Volume 42, No 3; Page 477 – 489.
18. Trauma Registry Injury Coding Is Superfluous; Robert Rutledge and Turner Osler; The Journal of Trauma; Volume 43, No 2; Page 253 – 257.
19. Improved Predictions from A Severity Characterization of Trauma (ASCOT) over Trauma and Injury Severity Score (TRISS): Results of an Independent Evaluation; Wayne S. Copes, Howard R. Champion, William J. Sacco; The Journal of Trauma; Volume 40, No 1; Page 42 - 49.
20. A Comparison of Abbreviated Injury Scale 1980 and 1985 Versions; Wayne S. Copes, Mary Lawnick, Howard R. Champion, William J. Sacco; The Journal of Trauma; Volume 28, No 1; Page 78 - 85.
21. A new characterization of Injury severity; Champion HR, Copes WS, Sacco WJ; The Journal of Trauma; 1990; May; 30; Page 539 – 547.
22. Trauma Score Systems: Cologne Validation Study; Bouillon, Lefering, Vorweg; The Journal of Trauma; Volume 42, No 4; Page 652 - 657.

This is a peer reviewed paper 

Please cite as : Vimal Kumar:Trauma Scores As A Prognostic Indicator For Blunt Injury Patients In India – A Prospective Study.

J.Orthopaedics 2006;3(4)e2

URL: http://www.jortho.org/2006/3/4/e2

ANNOUNCEMENTS

Powered by
VirtualMedOnline