INTRODUCTION

Trauma may be defined as a group of unpredictable and undesirable events, involving varied levels of violence, that result in injuries on its victims.¹ Such events represent one of the largest causes of morbidity and mortality in the world, nowadays being the leading cause of death in the population below 45 years of age, mostly male.² The World Health Organization had predicted 5.3 million deaths due to trauma for 2015 alone, representing 9.2% of the total for that period. In 2030, 6.3 million deaths from trauma are expected.³

Trauma also presents significant impact when analyzed from the economic standpoint. In the United States, it is estimated that in 2013 alone, trauma injuries caused losses of more than 670 billion dollars when combined medical costs and work disability costs secondary to trauma.⁴ In the same country, in 2014, trauma was responsible for over 2 million years of potential life lost.⁵ In Brazil, in 2012, about 1 million hospitalizations occurred as a result of trauma, US$600.30 being the average cost of medical care for each victim. Approximately 150,000 deaths occur annually in Brazil due to trauma, most of them in the first hours after injury.^6,7

One of the most important challenges regarding trauma care is the fact that the victims' profiles differ significantly on nature and severity of injuries. This heterogeneity has been stimulating numerous scientific researches, especially with regard to the development of Trauma Scoring Systems (TSS), which involve the attribution of a certain score to trauma victims depending on specific variables.^8,9 Trauma scoring systems may guide decisions involving both prehospital care—for example, if a specialized transport is required—and inhospital care—such as, whether or not certain surgical procedures should be performed. In addition, TSS also provide important information concerning the anatomy of the lesions, physiological changes, and prognosis. These scoring systems also optimize communication among health professionals by standardizing the language being used and contribute to the quality control of the trauma response services. Furthermore, TSS provide epidemiological data that may guide campaigns on trauma prevention.^9,10

Over the past years, a wide array of TSS were developed, including RTS, which considers three physiological parameters: Respiratory rate, systolic blood pressure and GCS.¹¹ Despite remaining widely used and discussed, the calculation of RTS involves overly complex formulas, which could delay trauma care to the patient. This was one of the most important factors that inspired the development of two other TSS: GCS, age, and arterial pressure (GAP); and mechanism, GCS, age, and arterial pressure (MGAP).^12,13 The calculations of GAP and MGAP may be easily performed either at the trauma scene or in the hospital setting.⁹ In addition, both GAP and MGAP have the ability to predict mortality almost as precisely as the Trauma-Related-Injury Severity Score (TRISS), one of the most reliable TSS regarding prognosis.^12-16

In this context, the objective of the present study is to compare the mortality prediction capabilities of three physiological TSS: RTS, GAP, and MGAP.

MATERIALS AND METHODS

A descriptive, cross-sectional study was conducted involving trauma victims admitted to the emergency room of a Brazilian specialized trauma care center between December-2013 and February-2014. Data were uninterruptedly gathered by Medical students from a trauma academic league — Liga Acadêmica do Trauma do Hospital Universitário Cajuru (LATHUC). The students had previously received training on how to properly gather the data.

Three physiological TSS (RTS, GAP, and MGAP) were calculated at admission (Table 1). In order to perform these calculations, the following variables were analyzed: respiratory rate, GCS, systolic blood pressure, trauma mechanism, and age. Gender and trauma occurrence throughout the week were also analyzed at admission. Mortality and length of hospitalization were assessed during a follow-up period that lasted until August-2014.

Two groups were created — deaths (D) and survivals (S) — and compared in relation to the TSS, gender, age, GCS, and trauma mechanism. Student's t-test and the nonparametric Mannt—Whitney test were used to analyze quantitative variables, such as, means and standard deviations. Chi-square test was used for qualitative variables, like frequencies and percentages. In order to compare the predictive power of the three TSS—RTS, GAP, MGAP—an analysis of the ROC curve was performed, including cut-off values, sensibilities, and specificities. The adopted level of significance in this study is 5% (p < 0.05). STATA software (v. 13.1) was used to analyze the collected data. This study was approved by the Research Ethics Committee of the Pontifical Catholic University of Paraná, Brazil (decision 480.483 of April 12th 2013).

RESULTS

The data of 668 patients were collected between December 2013 and February 2014. The mean age was 37 ± 18 years and most patients (69.8%) were male. Trauma occurrences were most frequent on Saturday (20.1%), Monday (17.5%), and Sunday (14.8%). Additional epidemiology data is available in Table 2. Blunt trauma accounted for 90.6% of the cases, while penetrating trauma corresponded to 9.4%. The most frequently observed trauma mechanisms were motorcycle accident (30.1%), same-level fall (16.9%) and automobile accident (12.9%). Frequencies of other trauma mechanisms are available in Graph 1. As for the length of stay, 65.6% of the patients analyzed, remained in the hospital for less than 24 hours; 18.1% required hospitalization between 1 to 2 days; 9.6% between 3 to 7 days; and 6.7% were hospitalized for 8 days or longer.

Patients who survived (group S) amounted to 657 (98.4%) and 11 patients (1.6%) had death (group D) as an outcome. Group S had a mean age of 36.7 ± 17.8 years, whereas the mean age for group D was 55.2 ± 27.4 years (p = 0.049). The GCS scores presented asymmetrically, with 86.5% of the victims scoring 15 points. The GCS median, minimum, and maximum values were 15 (3—15) for Group S and 3 (3—15) to groups D (Table 3).

The mean scores of GAP, MGAP, and RTS for group S were 22.8 ± 1.7, 27.4 ± 2.3, and 7.77 ± 0.33, whereas these scores for group D were 13 ± 7, 15.5 ± 7, and 4.57 ± 2.95 respectively (p < 0.001) (Table 4). There were significant statistical differences between the three scores in relation to both groups (p < 0.001).

Graph 1

Distribution of trauma mechanisms (%)

Graph 2

Analysis of the Receiver Operating Characteristic (ROC) curve. RTS: Revised Trauma Score; GAP: Glasgow Coma Scale, Age, and Arterial Pressure; MGAP: Mechanism, Glasgow Coma Scale, Age, and Arterial Pressure

The ROC curve (Graph 2) was analyzed so as to evaluate predictive power for death of the three TSS calculated at admission. The cut-off points indicated by the ROC curve adjustment were 7.7 for RTS, 19 for GAP, and 23 for MGAP (Table 5). The areas under the curve (AUC) were 0.926 for RTS (p < 0.001), 0.941 for GAP (p < 0.001), and 0.981 for MGAP (p < 0.001). Comparison of the three ROC curves revealed no statistically significant difference regarding the efficacy to predict deaths of the three TSS (p = 0.138).

Sensitivity and specificity were obtained considering a confidence interval of 95%. Regarding sensitivity, RTS achieved 90.9% (73.9—100%), whereas GAP and MGAP obtained 81.8% (59—100%) and 100% respectively. As of specificity, the performances were: 87.2% (84.7—89.8%) for RTS, 96.2% (94.7—97.7%) for GAP, and 91.6% (89.5—93.7%) for MGAP.

DISCUSSION

In this study, the majority of occurrences involved male subjects of working age¹⁷ and victims of blunt trauma. Most traumas were due to traffic accidents, i.e., motorcycle, automobile, and bicycle accidents; and run-overs. These epidemiologic findings corroborate the data found in the literature.^2,4,7,8,18

According to Trunkey,¹⁹ mortality due to trauma follows a trimodal distribution: Group I (50%) represents the often fatal injuries, usually resulting in death at the trauma scene; group II (30%) occurs within minutes to hours after the traumatic event and involves serious injuries that are potentially fatal in the absence of intensive primary care; and group III (20%) occurs from days to weeks after the event due to complications, such as, multiple organ failure and/or sepsis. Mortality related to groups II and III can be minimized by rapid and effective treatments. Therefore, a reliable and easily calculated TSS may help prevent mortality.^9,20

Currently, TRISS has been held as one of the most reliable TSS regarding prognosis, being widely discussed in the literature.^9,12,15 However, since it is an anatomical trauma score, meticulous examination of the lesions and even imaging exams, which are not always immediately available, must be performed. This could delay the initial assessment of trauma victims and their proper care, representing an important limitation to the use of anatomical scores.⁹ Thus, an alternative would be the use of TSS that consider physiological parameters, which may promptly indicate the existence of serious injuries and stratify death risk. In addition to using physiological variables that are quickly obtained, it is important that the calculations of TSS be easily performed.²¹ Regarding this aspect, GAP and MGAP stand out because they require simpler calculations, which could expedite the evaluation of trauma victims and improve the quality of both pre-hospital and hospital care.¹²

In this study, all three indexes—RTS, GAP and MGAP—demonstrated significant capability in predicting mortality in patients admitted to the emergency room, since the means of each TSS for the two groups — survivals and deaths — were significantly different (p < 0.001). This important prediction capability is also demonstrated visually through the extensive AUC of each ROC analysis. In this study, values below 7.7 (RTS), 19 (GAP), and 23 (MGAP) indicated a higher chance of death as an outcome.

When compared with each other regarding their respective ROC curves, the TSS demonstrated no statistically significant differences in mortality prediction. This finding differs from other published studies that demonstrate significant differences between these indexes, in which GAP and MGAP were better predictors of mortality when compared to RTS.^9,12,13 Moreover, according to Kondo et al,¹³ GAP is a better predictor of mortality than MGAP.

As for sensitivity, MGAP stood out in relation to RTS and GAP, the latter achieving the poorest performance in this aspect. Therefore, MGAP was associated with fewer false negatives for mortality, implying that a score above 23 drastically reduces the probability of death. Regarding specificity, GAP was superior compared to MGAP and RTS, the latter being the least specific of them all. Hence, GAP was associated with fewer false positives for mortality, meaning that a GAP value below 19 strongly suggests death as an outcome.

It has been demonstrated higher morbidity and mortality in elderly trauma patients when compared to younger patients victims of similar traumas.^22,23 Some studies even indicate an increase in mortality from the 4th decade of life onwards.^12,24,25 In this study, the mean age of patients who did not survive was significantly different than patients who survived (55.2 and 36.7 years respectively). This observation reinforces the importance of taking age into consideration and could partially explain the performance differences between GAP and MGAP, which are modified according to age; and RTS, which remains the same regardless of age. Another possible explanation for RTS's slightly worse performance in this study is the fact that, unlike MGAP, it does not take into consideration the trauma mechanism, even though penetrating traumas are more frequently associated with death—as already demonstrated in the literature.¹²

CONCLUSION

In the present study, most patients were young males and victims of blunt trauma. Lower RTS, GAP, and MGAP at admission involved higher mortality rates. Despite sample limitations, such as few deaths, the performances of both GAP and MGAP indicate their potential use as alternatives to RTS — especially considering that, when compared to the latter, GAP achieved a higher specificity, whereas MGAP demonstrated superiority in terms of sensitivity and specificity.