Article, Traumatology

The Revised Trauma Score plus serum albumin level improves the prediction of mortality in trauma patients

a b s t r a c t

Introduction: The Revised Trauma Score (RTS) is used worldwide in prehospital practice and in the emergency department (ED) settings to triage trauma patients. The main purpose of this study was to evaluate the value of the RTS plus serum albumin (RTS-A) and to compare it with other existing trauma scores as well as to compare the predictive performance of the Trauma and Injury Severity Score with the RTS-A (TRISS-A) with the original TRISS.

Methods: This was a single center, trauma registry based observational cohort study. Data were collected from consecutive patients with blunt or Penetrating injuries who presented to the emergency department of a tertiary referral hospital, between January 2012 and June 2016. 3145 and 2447 patients were assigned to the derivation group and validation group, respectively. Main outcome was in-hospital mortality.

Results: Among patients in the derivation group, the median [interquartile range] age was 59 [43-73] years, and 66.7% were male. The area under the receiver operating characteristic curves of the RTS-A (0.948; 95% CI: 0.939-0.955) was higher than that of the RTS (0.919; 95% CI: 0.909-0.929). In patients with blunt trauma, the AUC of the TRISS-A (0.960; 95% CI: 0.952-0.967) was significantly higher than that of the original TRISS (0.949; 95% CI: 0.941-0.957).

Conclusion: The value of the RTS-A predicts the in-hospital mortality of trauma patients better than the RTS, and the TRISS-A is a better mortality predictor compared to the original TRISS in patients with blunt trauma.

(C) 2017

Introduction

The Revised Trauma Score (RTS) is used worldwide in prehospital practice and in the emergency department (ED) settings to triage trau- ma patients. The RTS is composed of physiological patient parameters: systolic blood pressure (SBP), respiratory rate (RR), and the Glasgow Coma Scale (GCS) [1]. While those parameters are easily obtainable dur- ing the early phase of trauma, the RTS almost always yields definite scores that guarantee clear decisions in triage settings. Whereas the physiological scoring system is readily influenced by many external and internal factors, it can produce many false positives. Because there was a need for a more accurate prediction of survival of trauma patients, the Trauma and Injury Severity Score was developed from the Major Trauma Outcome Study (MTOS). The TRISS consisted of the RTS, Injury Severity Score (ISS), mechanisms of injury and age [2]. It showed

* Corresponding author at: Department of Emergency Medicine, Gyeongsang National University School of Medicine, 15, Jinju-daero 816beon-gil, Jinju-si, Gyeongsangnam-do 52727, Republic of Korea.

E-mail address: [email protected] (D.H. Kim).

a better Predictive power of the survival of trauma patients than the RTS and remains the most prominent survival predictor in research for the quality control of trauma management and prevention.

The BIG score (Admission Base deficit, International normalized ratio, and Glasgow Coma Scale) was developed to predict trauma mor- tality in children [3]. The BIG showed a good predictive performance in the adult trauma population [4]. The emergency trauma score (EMTRAS) is a validated trauma score using age, GCS, and the initial re- sults of base excess and prothrombin time [5,6]. Both scoring systems were focused on quick triage in the early period of trauma management and included biochemical markers to improve the Predictive ability.

Hypoalbuminemia has been shown to be closely related to adverse outcomes in various medical and surgical settings [7-10]. The most con- vincing explanation is that hypoalbuminemia is caused by a condition of protein-energy malnutrition (PEM); PEM contributes to impaired wound healing, increased susceptibility to infection, multiorgan dys- function, Prolonged hospitalization, and in-hospital mortality [11]. Prac- tically, the patients with chronic underlying diseases, such as cancer, liver cirrhosis, or chronic renal failure, showed poorer clinical outcomes than previously healthy patients [12-15]. In other words, the serum

http://dx.doi.org/10.1016/j.ajem.2017.06.027

0735-6757/(C) 2017

Albumin level represents the status of a patient’s chronic illness, nutri- tion, and immunocompetence. We postulated that the Serum albumin level may provide an important clinical insight on the prognosis of trau- ma victims, and the additive value of serum albumin levels combined with the RTS may improve the predictive power of the RTS. Further- more, the predictive performance of the original TRISS may be improved when the singular value of the RTS was replaced with the value of the RTS and serum albumin.

Goals of This Investigation

The purposes of this study were to evaluate the predictive perfor- mance of a value of the RTS plus serum albumin (RTS-A) and the TRISS using the RTS-A (TRISS-A) as well as to compare these scores with the RTS, the BIG, the EMTRAS, and the TRISS.

Materials and methods

Study design and setting

This is a prospectively recorded registry-based observational study using data from the trauma registry of a tertiary hospital located in Jinju, Republic of Korea. Our hospital is the only tertiary referral hospital, not a specialized trauma facility, located in the southcentral region of the Republic of Korea covering a relatively small area of 70 km in radius. The annual ED census is approximately 33,000. The Gyeongsang Nation- al University Hospital Trauma Registry has been established as a part of the National Emergency Department-based Injury In-depth Surveil- lance funded by the Korea Centers for Disease Control and Prevention. The registry enrolls all injured patients who presented to our ED. The enrolled patients are categorized by mechanism of injury: blunt, pene- trating, thermal, asphyxia, poisoning, and others. A total of 240 items were prospectively recorded from the prehospital period to the hospital discharge. Items included demographic characteristics, initial physio- logical parameters, initial laboratory findings, the results from ED man- agement, and the results from hospital admission. Most of the data were abstracted from the electronic medical records in our hospital. Initial systolic and diastolic blood pressure, heart rate (HR), RR, body temper- ature, the GCS were collected from ED triage records after being entered by the Triage nurses. Abbreviated Injury Scales (AIS) were recorded ac- cording to clinical presentation, Imaging results, intervention findings and operative records. The RTSs and the ISSs were automatically calcu- lated and were recorded in the registry. Injury descriptions were pre- liminarily prepared by professional heath information managers and were reviewed by emergency physicians before final confirmation.

Participants and data collection

We used consecutive data from January 2012 to June 2014. Inclusion criteria were an age of 15 years or older, blunt or penetrating mecha- nism, ISS >= 1, and the results from ED management were either admis- sion or death during the ED stay. We did not include patients who were dead on arrival or those who were discharged or transferred from the ED. Demographic characteristics, Physiological parameters, internation- al normalized ratios of prothrombin time (PTINR) and percent of the ref- erence value of prothrombin time (PTPERCENT), base excess levels, and serum albumin levels (gram/deciliter) were collected. The EMTRAS, the BIG, and the TRISS were calculated for each patient. In-hospital mor- tality was the outcome measure. For validation of the final predictive model, we used the data from the same registry from July 2014 to June 2016. This study was approved by the Gyeongsang National Uni- versity hospital institutional review board.

Statistical analysis

Multivariate imputation with chained equations using predictive mean matching was used to control for the missing data [16]. The miss- ing values included the PTINR (4.6%), PTPERCENT (4.3%), BE (10.1%), albu- min (4.3%), and the GCS (0.1%). All variables which were needed for the trauma scores as well as sex, age, SBP, HR, RR, ISS, and the outcome var- iable, which was death, were included in the imputation. Ten imputa- tions were performed and the mean values of PTINR, PTPERCENT, BE, albumin, and the GCS replaced the missing values. Complete scores of the RTS, the EMTRAS, the BIG, the TRISS, and the RTS-A were calculated with the imputed data set. The same procedure was performed with the validation group.

To verify the adequacy of the RTS-A score in the prediction of in-hos- pital mortality, we compared the RTS-A with two Multivariable logistic regression models. First, after confirmation of the statistical significance of all the variables (age, sex, mechanism, PTINR, PTPERCENT, BE, albumin, and the RTS) with the univariable analysis, a multivariable analysis was performed with all significant parameters (model 1). The ISS was not included in the analysis due to its lack of availability during the early phase of treatment. Second, a multivariable analysis was conduct- ed using albumin and the RTS (model 2). Third, we added the RTS score and the serum albumin level (model 3), then compared the area under the receiver operating characteristic curves (AUC) of the three models using a nonparametric method [17]. While model 3 was far easier to cal- culate than either model 1 or model 2, because there were no statistical significances between the models, it was appropriate to choose model 3. The predictive properties of the value of the RTS-A were evaluated with a receiver operating characteristics (ROC) for discrimination and

Fig. 1. Derivation and validation cohort.

Table 1

Baseline characteristics of derivation and validation group.

Derivation group

Validation group

p value

Value

Available %

Value

Available %

Number of patients

3145

2447

Age (yrs)

59 [45-73]

100

60 [46-73]

100

0.1652

>= 65 years, n (%)

1279 (40.7)

100

997 (40.7)

100

0.9578

Male (%)

66.7

100

66.5

100

0.8750

Mechanism, blunt (%)

95.7

100

96.2

100

0.3485

ED physiological parameters

SBP (mm Hg)

135 [117-150]

100

130 [110-140]

100

b0.0001

RR (breath/min)

20 [18-20]

100

20 [18-20]

100

0.8174

HR (beats/min)

80 [72-92]

100

80 [73-90]

100

0.5088

Glasgow Coma Scale

15 [15-15]

99.9

15 [15-15]

99.9

0.2084

Laboratory results

PTINR

1.04 [0.98-1.12]

95.4

1.04 [0.98-1.12]

98.2

0.3309

PTPERCENT (%)

93 [83-103]

95.7

93 [83-101]

96.6

0.2427

Base excess

-2.0 [-4.2-0.3]

89.9

-1.2 [-3.4-0.6]

89.4

b0.0001

Albumin (gram/deciliter)

4.1 [3.7-4.4]

95.7

4.1 [3.8-4.4]

98.9

0.0044

ISS

9 [4-13]

100

9 [4-13]

100

0.4464

Mortality (%)

9.8

100

9.0

100

0.3104

SBP systolic blood pressure. RR respiratory rate.

HR heart rate.

PTINR international normalized ratio of prothrombin time. PTPERCENT percent of the reference value of prothrombin time. ISS Injury Severity Score.

All continuous variables showed skewed distribution and were expressed in median value [interquartile range].

Hosmer-Lemeshow statistics for calibration. The AUC of the value was compared with those values of the RTS, the EMTRAS, the BIG, and the TRISS. We replaced the RTS in the TRISS with the value of the RTS-A (TRISS with albumin, TRISS-A) and the AUC was compared with the original TRISS. Then, we calculated the regression coefficients of the TRISS-A in a subgroup of patients who had a blunt injury mechanism (the number of patients with a penetrating mechanism of injury were too small), and then we compared those results with the original TRISS in the patients with a blunt injury mechanism.

Independent t-tests or the Mann-Whitney U tests were used for con- tinuous variables, and chi-square tests were used for categorical vari- ables. All p values were two-sided, and a value of p b 0.05 was considered statistically significant. Analyses were performed using Stata 12 software (StataCorp, LP, College Station, TX).

Results

Characteristics of study subject

During the study period from January 2012 to June 2014, a total of 22,254 patients were entered into the Gyeongsang National University Hospital Trauma Registry. Among 3187 eligible patients, 42 patients

were excluded because they were confirmed dead on ED arrival, and 3145 patients were included in the study. For validation, 19,652 patients were entered into the registry. Among 2462 eligible patients, 2447 pa- tients were ultimately included (15 were confirmed dead on arrival) (Fig. 1). In-hospital mortality was 9.79% (derivation group) and 9.0% (validation group). Table 1 summarizes the baseline characteristics of both groups.

Verification of the Revised Trauma Score plus serum albumin model

Univariable logistic regressions were performed for age, sex, mecha- nism, PTINR, PTPERCENT, BE, and albumin. Age, sex, and all the biomarkers showed statistical significance in predicting in-hospital mortality. We chose PTINR instead of PTPERCENT because it is a more standardized meth- od. multivariable analyses were performed using age, sex, PTINR, BE, al- bumin, and RTS (model 1). The Area Under the Receiver Operating Characteristic Curve was 0.963 (95% confidence interval (CI): 0.956-0.970). The AUC of the multivariable analysis with the RTS and serum albumin level was 0.949 (95%CI: 0.941-0.956) (model 2). The difference between the two models was 0.014, which was statistically non-significant (p = 0.148; 95%CI: 0.004-0.024). The regression coeffi- cients of the RTS and serum albumin level in model 2 were similar

Table 2

Area under receiver operating characteristic curves (AUC) of the RTS plus albumin compared with other trauma scores.

Variable

Derivation group

H-L ?2 (p value)

Validation group

AUC 95% CI

p valuea

AUC 95% CI

p valuea

RTS-A

0.948 0.939-0.955

5.6 (0.70)

0.953 0.944-0.961

RTS

0.919 0.909-0.929

0.0001

6.9 (0.14)

0.902 0.890-0.914

b0.0001

EMTRAS

0.959 0.952-0.966

0.0690

4.3 (0.51)

0.946 0.936-0.954

0.3283

BIG

0.932 0.923-0.941

0.0639

8.3 (0.40)

0.925 0.913-0.935

0.0030

TRISS

0.952 0.944-0.959

0.5433

3.4 (0.95)

0.954 0.945-0.962

0.8281

H-L ?2 Hosmer-Lemeshow chi-sqaure.

RTS-A Revised Trauma Score plus serum albumin. RTS Revised Trauma Score.

EMTRAS Emergency trauma score.

BIG Admission base deficit, International normalized ratio, and Glasgow Coma Scale. TRISS Trauma and Injury Severity Score.

a p value represents the statistical significance in the difference of the AUC between RTS-A and those of other trauma scores.

Table 3 Regression coefficients of original TRISS and the reestimated regression coefficients of the TRISS with albumin (TRISS-A) in patients with blunt trauma.

Coefficient

TRISS

TRISS-A

b0

-0.4499

-9.3882

b1

0.8085

1.2764

b2

-0.0835

-0.0899

b3

-1.7430

-0.8140

Trauma and Injury Severity Score = Probability of survival = 1/(1 + e-b). b = b0 + (b1 x RTS) + (b2 x ISS) + (b3 x Ageindex).

Trauma and Injury Severity Score with albumin = Probability of survival = 1/(1 + e-b). b = b0 + (b1 x RTS-A) + (b2 x ISS) + (b3 x Ageindex).

(-1.556 and -1.002). For this reason, we simply added the RTS and the serum albumin level (model 3), then compared the results of the two multivariable regression models. The AUC of model 3 (AUC = 0.948; 95%CI: 0.939-0.955) showed no significant difference compared to those from models 1 and 2 (AUC differences = 0.014 and 0.001; p = 0.148 and 0.930; 95%CI: 0.004-0.024 and -0.010-0.012, respectively).

Model 3 was easier to calculate and was not statistically different from model 1 or model 2. Therefore, model 3 was chosen as our final model.

Comparison between RTS-A and other trauma scores

The RTS-A ranged from 0.3000 to 13.0408. The AUC of the RTS-A was 0.948 and showed feasible calibration (H-L ?2 = 5.6, p = 0.70). The AUC of the RTS-A was significantly higher than that of the RTS (0.919) and showed a non-significant difference compared to the EMTRAS (0.959), the BIG (0.932), and the TRISS (0.952) (Table 2).

The RTS-A was added into the TRISS, replacing the RTS, and a multi- variable logistic regression was performed to re-estimate the regression coefficients. The sample of patients with a penetrating injury mecha- nism was too small (n = 135, 4.3%); therefore, we analyzed the patients with blunt trauma only (n = 3010). The re-estimated regression coeffi- cients are exhibited in Table 3. The AUC of the TRISS-A was 0.960 and was significantly higher than those of the RTS-A, the RTS, the BIG, and the TRISS models. Although the AUC of the TRISS-A was higher than the EMTRAS, it failed to show significance (Table 4). In the validation group, the AUC of the RTS-A was significantly higher than those of the RTS and the BIG (Table 2). The AUC of TRISS-A was significantly higher than those of other scores (Table 4).

Discussion

In our study, the RTS-A showed a superior performance when com- pared to the RTS and the TRISS-A that used the RTS-A instead of the RTS

Table 4

Area under receiver operating characteristic curves (AUC) of the TRISS with albumin (TRISS-A) compared with other trauma scores.

Variable Derivation group Validation group

AUC

95% CI

p valuea

AUC

95% CI

p valuea

TRISS-A

0.960

0.952-0.967

0.968

0.960-0.975

RTS-A

0.946

0.938-0.954

0.0011

0.953

0.943-0.961

0.0001

BIG

0.931

0.922-0.940

0.0004

0.925

0.914-0.936

b 0.0001

EMTRAS

0.959

0.951-0.966

0.8427

0.946

0.936-0.954

0.0012

RTS

0.918

0.907-0.927

b 0.0001

0.903

0.891-0.915

b 0.0001

TRISS

0.949

0.941-0.957

0.0109

0.955

0.946-0.963

0.0038

TRISS-A TRISS with albumin.

RTS-A Revised Trauma Score plus serum albumin. RTS Revised Trauma Score.

EMTRAS Emergency trauma score.

BIG Admission base deficit, International normalized ratio, and Glasgow Coma Scale. TRISS Trauma and Injury Severity Score.

a p value represents the statistical significance in the difference of the AUC between the TRISS-A and those of other trauma scores.

alone predicted mortality better than the original TRISS and other trau- ma scoring systems with physiological and Biochemical parameters.

The TRISS, as the gold standard predictor of trauma mortality, con- tains physiological (RTS), anatomical (ISS), and demographic (age) var- iables. Recent studies have shown that combinations of physiological and biochemical information might be more feasible tools to estimate the mortality risk of patients with trauma [3,5]. In this study, the EMTRAS was the best method among previously developed trauma scores. However, the EMTRAS cannot be included into the TRISS where- as the RTS-A can. Our intentions were to add a simple biochemical marker that is easily and quickly obtainable in early phase of trauma management to the RTS and to improve the predictive ability of the gold-standard TRISS.

Serum lactate level is the most rigorously cited biomarker in trauma mortality [18-20]. While very few studies (if any) have addressed the al- bumin issue in trauma mortality, those studies were confined to specific injuries, such as traumatic brain injury, hip injury, or burns [21-24]. Al- bumin has several advantages over other biomarkers: it is one of the most available and low-cost biomarkers worldwide, and is almost al- ways included in initial routine work-up for major trauma; and it pro- vides clinical insight that represents the baseline patient status that is not comprehensible from physiological parameters.

We enrolled patients regardless of trauma severity. The ISS of the en- rolled data was >= 1, while most of previous studies enrolled patients with severe trauma, such as those with an ISS >= 16. Additionally, our study was based on an emergency department, not a trauma center. In the reasons, our results seem more applicable in Developing countries or regions where the trauma care system is not fully established.

The median age of our study population was 59 and 60 years in the derivation and validation group respectively, and the proportion of patients >= 65 years of age was 40.7% in both groups. This age distri- bution was much higher compared to the MTOS data (8.2%) in which the patient data were submitted from 1982 through 1986 [2,25]. About twenty years later, American College of Surgeons reported the proportion of patients >= 65 among adult trauma population (>= 15 years of age) was 17.3% in 2004. It has been rapidly increasing by about 1.5% per year, and reached to 34.4% in 2016 [26]. This trend of increase in the incidence of elderly trauma was also ob- served in European population. Kehoe et al. showed the average age of the patients from the Trauma Audit Research Network in UK was 36.1 in 1990 and increased up to 53.8 in 2013. Annual increase in the study was 1.43 years, which was faster twice than that of the previous decade [27]. These previous findings consistently suggest that global trauma population is rapidly getting older. Therefore, we believe our model could be applicable to other trauma popula- tion, despite our result that serum albumin provides additional dis- criminative power to the RTS might be exaggerated by frequent comorbidities in old patients.

This study has some limitations. First, selection bias was concerning. In total, there were 1729 patients among the eligible 22,254 patients in the derivation group, and 1782 patients among the 19,652 in the valida- tion group who were transferred to other facilities. We did not enroll these patients because the final outcome could not be measured. Sec- ond, this was a single-center study. External validation in a multicenter study is warranted. Third, we enrolled patients who were >= 15 years of age, so this model may not be used for pediatric patients. Fourth, the TRISS-A was not evaluated in patients with penetrating trauma due to the small number of patients.

Conclusion

The value of the RTS plus serum albumin (RTS-A) predicts the in- hospital mortality of trauma patients better than the RTS, and the TRISS using the RTS-A (TRISS-A) is a better mortality predictor com- pared with the original TRISS in patients with blunt trauma.

Funding

The authors have no financial relationships relevant to this article to disclose.

Competing interests

The authors have no conflicts of interest to disclose.

References

  1. Champion HR, Sacco WJ, Copes WS, et al. A revision of the trauma score. J Trauma 1989;29(5):623-9.
  2. Champion HR, Copes WS, Sacco WJ, et al. The major trauma outcome study: estab- lishing national norms for trauma care. J Trauma 1990;30(11):1356-65.
  3. Borgman MA, Maegele M, Wade CE, et al. Pediatric trauma BIG score: predicting mortality in children after military and civilian trauma. Pediatrics 2011;127(4): e892-7.
  4. Brockamp T, Maegele M, Gaarder C, et al. Comparison of the predictive performance of the BIG, TRISS, and PS09 score in an adult trauma population derived from multi- ple international trauma registries. Crit Care 2013;17(4):R134.
  5. Raum MR, Nijsten MW, Vogelzang M, et al. Emergency trauma score: an instrument for early estimation of trauma severity. Crit Care Med 2009;37(6):1972-7.
  6. Joosse P, de Jong WJ, Reitsma JB, et al. External validation of the emergency trauma score for Early prediction of mortality in trauma patients. Crit Care Med 2014;42(1): 83-9.
  7. Jellinge ME, Henriksen DP, Hallas P, et al. Hypoalbuminemia is a strong predictor of 30-day all-cause mortality in acutely admitted Medical patients: a prospective, ob- servational, cohort study. PLoS One 2014;9(8):e105983.
  8. Jin S, Bochicchio GV, Joshi M, et al. Admission serum albumin is predictive of out- come in critically ill trauma patients. Am Surg 2004;70(12):1099.
  9. Jin G-X, Li L, Cui S-Q, et al. Persistent hypoalbuminemia is a predictor of outcome in cervical spinal cord injury. Spine J 2014;14(9):1902-8.
  10. Garwe T, Albrecht RM, Stoner JA, et al. Hypoalbuminemia at admission is associated with increased incidence of in-hospital complications in geriatric trauma patients. Am J Surg 2016;212(1):109-15.
  11. Hoffer LJ. Clinical nutrition: 1. Protein-energy malnutrition in the inpatient. Can Med

    Assoc J 2001;165(10):1345-9.

    Franch-Arcas G. The meaning of hypoalbuminaemia in clinical practice. Clin Nutr 2001;20(3):265-9.

  12. Huisman EJ, Trip EJ, Siersema PD, et al. Protein energy malnutrition predicts compli- cations in liver cirrhosis. Eur J Gastroenterol Hepatol 2011;23(11):982-9.
  13. Hu W-H, Chen H-H, Lee K-C, et al. Assessment of the addition of hypoalbuminemia to ACS-NSQIP surgical risk calculator in Colorectal cancer. Medicine 2016;95(10).
  14. Gracia-Iguacel C, Gonzalez-Parra E, Barril-Cuadrado, et al. Defining protein-energy wasting syndrome in chronic kidney disease: prevalence and clinical implications. Nefrologia 2014;34(4):507-19.
  15. Royston P, White IR. Multiple imputation by chained equations (MICE): implemen- tation in Stata. J Stat Softw 2011;45(4):1-20.
  16. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Bio- metrics 1988;44(3):837-45.
  17. Cerovic O, Golubovic V, Spec-Marn A, et al. Relationship between injury severity and lactate levels in severely injured patients. Intensive Care Med 2003;29(8):1300-5.
  18. Callaway DW, Shapiro NI, Donnino MW, et al. Serum lactate and base deficit as pre- dictors of mortality in normotensive elderly blunt trauma patients. J Trauma Acute Care Surg 2009;66(4):1040-4.
  19. Vandromme MJ, Griffin RL, Weinberg JA, et al. Lactate is a better predictor than sys- tolic blood pressure for determining blood requirement and mortality: could prehospital measures improve trauma triage? J Am Coll Surg 2010;210(5):861-7.
  20. Yukl RL, Bar-Or D, Harris L, et al. Low albumin level in the emergency department: a potential independent predictor of delayed mortality in blunt trauma. J Emerg Med 2003;25(1):1-6.
  21. Caleman G, Morais JFd, Puga MEdS, et al. Use of albumin as a risk factor for hospital mortality among burn patients in Brazil: non-concurrent cohort study. Sao Paulo Med J 2010;128(5):289-95.
  22. Baltazar GA, Pate AJ, Panigrahi B, et al. Malnutrition as measured by albumin and prealbumin on admission is associated with poor outcomes after severe traumatic brain injury. Am Surg 2015;81(2):E61.
  23. Kieffer W, Rennie C, Gandhe A. Preoperative albumin as a predictor of one-year mor- tality in patients with fractured neck of femur. Ann R Coll Surg Engl 2013;95(1): 26-8.
  24. Champion HR, Copes WS, Buyer D, et al. Major trauma in Geriatric patients. Am J Public Health 1989;79(9):1278-82.
  25. American College of Surgeons. National Trauma Data Bank. https://www.facs.org/ quality-programs/trauma/ntdb/docpub. [accessed 17.06.06].
  26. Kehoe A, Smith JE, Edwards A, et al. The changing face of major trauma in the UK. Emerg Med J 2015;32(12):911-5.

Leave a Reply

Your email address will not be published. Required fields are marked *