Discussion
In the present analysis, only one out of four trauma patients could be classified according to the PHTLS classification of hypovolaemic shock if a combination of all three parameters (HR, SBP and GCS) was assessed. Consequently, 73.5% of all trauma patients did not match the classification criteria given by PHTLS. Interestingly, subgroup analysis of the trauma mechanism showed no significant difference between blunt and penetrating trauma patients. Based on these observations, the PHTLS classification of hypovolaemic shock displays substantial deficits in adequate assessing and risk-stratifying trauma patients in the prehospital setting. This is consistent with our previous analysis in which 90.7% of all trauma patients could not be grouped according to the ATLS classification of hypovolaemic shock when a combination of the vital signs HR, SBP and GCS at ED arrival was assessed.
When patients were classified by HR, mean SBP decreased slightly through the groups I to IV if all trauma patients were analysed. On the other hand, in patients classified by SBP, there was no significant alteration of HR in any group and no distinct tachycardia was observed at all. Interestingly, this was independent of the trauma mechanism or the presence of TBI. These observations indicate that the proclaimed association between hypotension and tachycardia seems not to reflect reality. Victorino and colleagues demonstrated that in hypotensive patients (SBP<90 mm Hg) only 65% presented tachycardic (HR>90 beats/min), consequently 35% of those patients were normofrequent. In contrast, 39% of the tachycardic patients still presented with SBP greater than 120 mm Hg. Furthermore, the sensitivity and specificity of HR in predicting hypotension was very poor. These data are supported by a study based on records derived from the Froedtert Hospital trauma registry. HR at ED admission was inaccurate in predicting injury severity, the need for immediate surgical intervention, such as laparotomy or thoracotomy, or transfusion of packed red blood cells within the first 24 h. Furthermore, Ley and colleagues have demonstrated on 3727 trauma patients that 44% even presented with relative bradycardia, defined as HR less than 90 beats/min and SBP less than 90 mm Hg. This clinical state was associated with a higher mortality rate and occurred more often in older patients. Beside these considerations, pain, anxiety, parasympathetic and sympathetic stimuli, drugs or permanent medication as well as increased intracranial pressure or spinal cord injury can all effect HR in trauma patients.
In the prehospital setting, many patients may still be in a compensated state with SBP greater than 90 mm Hg. In these relatively normotensive patients, the evaluation of the prehospital shock index (SI) was suggested to be superior over HR or SBP alone. Patients with a prehospital SI of more than 0.9 had a 1.5-fold increased risk of massive transfusion compared to patients with a SI less than 0.9, indicating that these patients were in severe haemorrhagic shock, despite SBP still being within physiological ranges (>90 mm Hg). Furthermore, recent publications have questioned the definition of hypotension at a SBP of less than 90 mm Hg, which is also the trigger for trauma team activation. Both in blunt and penetrating trauma, a SBP or less than 110 mm Hg has frequently been associated with a significant increase in mortality. Furthermore, in patients with a SBP less than 115 mm Hg aggravating hypoperfusion has been reported with parallel worsening of base deficit. Also in the prehospital setting, Bruns and colleagues have demonstrated that a SBP of less than 110 mm Hg may also represent an important cut-off for mortality, injury severity and occult hypoperfusion. This observation is substantiated by Parks and colleagues who emphasised that systemic hypotension (mean SBP<90 mm Hg) is a very late marker of shock. Therefore, some authors recommend the measurement of blood lactate levels rather than the evaluation of vital signs in order to detect the presence of hypovolaemic shock. In the prehospital setting, lactate levels of more than 3.5 mmol/l have been shown to discriminate survivors from non-survivors while vital signs were unaltered. Prehospital lactate levels have also been reported as independent predictors of mortality while HR and SBP were not. In times of point-of-care testing, lactate levels can easily be obtained and therefore might be a valuable alternative in the initial assessment of the trauma patient, even in the prehospital setting.
In our analysis, PHTLS seems furthermore to underestimate mental disability in the presence of hypovolaemic shock. Although there are multiple reasons for a reduced GCS such as haemorrhage, inadequate oxygenation, hypoglycaemia or intoxication, TBI is the leading cause of impaired GCS in Germany. In the present analysis, 41.8% of all patients presented with a severe TBI (AIShead ≥3), which impaired GCS at scene. Out of these patients, only 12.2% could be classified according to PHTLS, but even when patients with severe TBI were excluded, the majority (63.3%) could not be classified according to PHTLS. This is most likely due to an already impaired GCS in groups I and II, as could be demonstrated when patients were classified by their HR or SBP alone. These observations are inconsistent with the PHTLS classification, in which group I patients are characterised as only 'slightly anxious'.
Certain limitations to this investigation have to be acknowledged. Our analysis is a retrospective study of trauma registry data with all the shortcomings associated with such an analysis. Furthermore, the TraumaRegister DGU® does not capture information about pre-injury medication such as antihypertensive therapies or previous intake of alcohol and/or drugs; all potentially influencing the parameters assessed. In addition, the use of beta-adrenergic blockers may have substantiated the misclassification of patients by suppressing HR as well as pronouncing drops in SBP following acute blood loss. Several components of the PHTLS classification of hypovolaemic shock such as urinary output or pulse pressure are not captured in the TraumaRegister DGU® and therefore could not be considered for the present analysis. However, based on our clinical experience, these variables only play a minor role in the prehospital assessment and treatment of multiple injured patients. A further validation on other trauma databases, especially on those that include higher numbers of penetrating injuries is warranted.