Introduction

Severe traumatic brain injury (TBI) is an important cause of neurological disabilities and mortality in children [1–5], and continuous intracranial pressure (ICP) monitoring is critical to guide medical management in the pediatric intensive care unit (PICU), especially in those with the most severe injuries [5–9].

However, direct ICP measurement remains the reference method for intracranial hypertension (ICH) diagnosis and guided management [5–9], but represents an invasive surgical technique that cannot be initiated routinely at the bedside before cranial computed tomography (CCT) scan examination, and could be hazardous in patients with very critical conditions and coagulation disorders. Transcranial Doppler (TCD) examination, measuring blood flow velocity in both middle cerebral arteries, is a non-invasive method that requires training and experience to be efficient, but can be used at the bedside during the early phase of resuscitation. Its ability to detect altered cerebral hemodynamic profiles in patients with TBI and its usefulness in clinical decision making has been previously underlined [10–17]. However, clinical experience of early TCD monitoring in children with severe TBI has only been reported in short series [14, 17–19], and its reliability in predicting acutely increased ICP has not been clearly established. The aim of this study was to identify, in a series of children with severe TBI, the feasibility and reliability of TCD to predict ICH and resulting altered cerebral perfusion pressure (CPP) measured by invasive methods.

Material and methods

This is a descriptive and retrospective cross-sectional study based on the analysis of clinical data of children less than 15 years of age, presenting with a severe TBI defined by a scene Glasgow coma scale (GCS) ≤8. We reviewed the medical records of all children admitted to the emergency room of Hôpital Necker-Enfants Malades (Paris, France) between January 2000 and December 2005 (6 years). Children with non-accidental injuries and those whom TCD and invasive ICP monitoring was not performed at admission for any reason (like those with severe coagulation disorders and GCS 3 with absent brainstem reflexes or when those techniques were performed after 3 h of admission) were finally excluded from analysis.

Data were collected for age, gender, GCS, and pediatric trauma score (PTS), mechanism of accidents, and resulting injuries (multiple trauma or isolated head injury) at admission. All children were managed at the scene of the accident by a medical team of the emergency medical system (SAMU, Service d’Aide Médicale d’Urgence), transported to our trauma center, and benefited from standardized initial pre-hospital management including systematic venous access insertion, exclusive normal saline intra venous (IV) infusion, rapid sequence induction and tracheal intubation, assisted ventilation with associated continuous IV sedation, and volume expansion with colloids as needed, according to the French National Guidelines for emergency management of severe head injuries [5, 11, 20]. In emergency department and PICU, continuous IV sedation with benzodiazepine and sufentanil was maintained at least 48 h, mean arterial pressure (MAP) was adjusted within normal range for the age with fluid loading and/or vasopressor infusion as needed and assisted ventilation was carefully adapted to maintain a 35–40 mmHg end-tidal carbon dioxide.

All intended TCD examinations were performed during the resuscitation phase, and within the first 20 min following admission. All patients were, at the moment of the examination, stabilized with restored MAP, continuous IV sedation, and adapted assisted ventilation. Systolic (sV), end-diastolic (dV), and mean (mV) arterial blood flow velocities were measured in both middle cerebral arteries, using a 2-MHz continuous Doppler probe (Waki 1-TC, Atys Medical®, France), with a classical transtemporal window approach by a trained operator (senior anesthesiologist with >6 months experience in TCD). Pulsatility index (PI = sV − dV/mV) and resistivity index (RI = sV − dV/sV) were automatically calculated. TCD was considered altered in the following situations:

  • Evidence of bilateral backflow or no-flow despite careful examination with varying depth of insonation and maximum power settings.

  • End-diastolic velocity <25 cm/s or PI >1.31 as previously described [14]

Intraparenchymal pressure transducers (Codman®) were inserted as soon as possible, after CCT scan examination demonstrating cerebral injury, in the frontal region (opposite to the lesion when focal injury was present), by a trained neurosurgeon, and maintained at least 48 h. Cerebral perfusion pressure was calculated continuously as the difference between ICP and MAP measured by the means of an intra-arterial catheter. ICH was considered for any episode of ICP over 20 mmHg [21], lasting more than 10 min without concomitant external stimulation (tracheal suctioning, mobilization, physiotherapy, or nursing). Altered CPP was considered with measurements <50 mmHg with MAP in the normal range for the age [22]. MAP was considered as normal, according to age group, based on the American Heart Association criteria [23].

Statistical analysis was performed using the Statistical Package for the Social Sciences for Windows. For data analysis all quantitative variables were expressed as mean and SD. We considered, according to medical literature, the invasive ICP as the gold standard method to measure ICP. Sensitivity, specificity, positive, and negative predictive values were verified. This descriptive single-center study was performed in agreement with French law requirements (institutional review board consultation, no informed consent required, anonymous computer data bank analysis).

Results

In the study period, 315 children presented to our emergency department with severe TBI. Among them, 48 with a GCS of 3 dying within the first hours and 143 with missing data, were excluded. This let 124 severely head-injured children who benefited from early TCD examination. Initial no-flow/backflow was present and confirmed after CCT scan examination depicting severe diffuse injuries in 6% (7/124). All of them died within the first hours and were excluded from final analysis. This let 117 severely head-injured children in our final sample considered for analysis.

Mean age was 7.6 ± 4.4 years, with male prevalence (71%). Median GCS was 6, and median PTS was 4, indicating the magnitude of injury in this group of patients. Excluding children with non-accidental injuries, these patients did not differ from those where data were missing. Demographic data are presented in Table 1.

Table 1 Demographic data verified at hospital admission of 117 children with severe head trauma—Hôpital Necker-Enfants Malades (Paris, France) from 2000 to 2005
Full size table

Cerebral hemodynamic profiles within the normal range were detected in 31.6% (37/117), and altered TCD profiles were identified in 68.4% (80/117) of the children. Intraparenchymal ICP transducers were inserted within the first 3 h after CCT scan examination, in this group of patients. There was no complication (infections, intracranial hemorrhage, or technical failure) related to invasive ICP monitoring. Increased ICP was noted upon transducer insertion in 27% (32/117) of the children. An alteration in initial TCD profiles was observed in 94% (30/32) of children with initially increased ICP, and 59% (50/85) of those with normal initial ICP values. Initial ICH was predicted by early TCD examination with a sensitivity of 94% and a 95% probability to identify those with normal ICP (Table 2). Regarding CPP, initial measurements were above the threshold value of 50 mmHg in 62% (72/117) of the cases. In children with decreased initial CPP, early TCD profile was considered as altered in 80% (36/45), compared to 61% (44/72) in children with normal initial CPP. Decreased initial CPP was so predicted by TCD examination with a sensitivity of 80% (Table 3).

Table 2 Comparison of intracranial pressure and transcranial Doppler at admission in 117 children admitted at the emergency room of Hôpital Necker-Enfants Malades, Paris, France (2000 to 2005)
Full size table
Table 3 Comparison of cranial perfusion pressure and transcranial Doppler profiles at admission in 117 children admitted to the emergency room of Hôpital Necker-Enfants Malades, Paris, France (2000 to 2005)
Full size table

Discussion

Previous studies have pointed out the usefulness of TCD examination to help medical decision in patients with cerebrovascular diseases [24–27] or hydrocephalus [28–30]. This non-invasive method, even though not measuring directly ICP, can easily detect alterations of cerebral blood flow velocities reflecting ICH. Its ability to predict acute ICH in children with TBI has been sparsely advocated, but experience was still limited in this particular indication [8, 17].

All children included in the present study had severe accidental TBI resulting in coma with a median GCS of 6. Their general characteristics were similar to those previously reported in our trauma population [31]. All of them benefited from a standardized initial medical to PICU management [31] in accordance with French National Guidelines for emergency management of TBI. These recommendations include continuous ICP and CPP monitoring as the most adapted means of detecting developing ICH requiring specific therapeutic interventions in comatose children. However, it remains an invasive and expensive method that could only be used routinely in specialized units [32]. In our unit, all children except those with no cerebral blood flow, were routinely submitted from an invasive ICP, to extensive experience could maintain a null incidence of complications with intraparenchymally inserted transducers as reported in the literature [33, 34]. Regarding CPP, a threshold value of 40–45 mmHg under which children with severe TBI could experience worse outcome was cited in the literature [35, 36]. Careful cerebral pressure monitoring appears of critical interest to detect episodes of cerebral pressure alteration that could be determinant in mortality and outcome. A major point of concern could be to identify as soon as possible those children at risk of ICH and decreased CPP, who carry an increased risk of complications and bad outcome.

TCD alone could not provide reliable information regarding cerebral blood flow/metabolism coupling and should be used dynamically to detect autoregulation impairment. It has been reported as an unreliable indicator of intracranial pressure in children [37]. However, the key issue for determining within the first minutes following admission which children should be considered at risks of developing ICH and decreased CPP, needing close monitoring could be the ability to perform a non-invasive, highly reproducible examination, that could be used at the bedside once the patient has been minimally stabilized [37]. TCD allows instant measurement of blood flow velocities in major cerebral vessels that could be affected by increased external strains resulting from increased ICP [38–40]. In a previous short series of 36 children with moderate to severe TBI, we could determine that decreased end-diastolic velocity and increased PI were the most determinant factors of bad outcome with threshold values of <25 cm/s for end-dV and >1.31 for PI [14]. Moreover, negative velocity appearing as a backflow on TCD profile could be clearly associated with brain death in children with severe TBI [41]. Using these definitions of altered TCD profiles, we verified that early TCD examination upon admission could predict ICH with a 94% sensitivity and 95% negative predictive value. A major limitation of this study could be that TCD examination was performed immediately after admission and could not be precisely correlated to ICP and CPP that were only measured after CCT scan examination. It could be suspected that, in children with severe TBI, in-hospital aggressive management could result in stabilized ICH before ICP monitor insertion. As mentioned in a previous study, mean arterial pressure, fever, impaired cerebral autoregulation, PaCO2, hematocrit, or sedation could be confounders of the relationship between cerebral perfusion pressure and cerebral blood flow velocities, mainly mean blood flow velocity [16]. In our patients, end-diastolic velocity and pulsatility index were chosen as reflects of significant TCD alterations that could accurately predict bad outcome and increased risks of ICH. These variables could be more directly affected by external arterial stains than mean flow velocity and could more accurately reflect increased ICP [14].

Regarding CPP, that could be a major factor of bad outcome, altered TCD predicted initial decreased CPP with an 80% sensitivity and 76% probability to identify those with normal CPP (negative predict value). More than trying to correlate TCD variables with measured ICP and CPP, the aim of this study was to identify upon admission before CCT scan examination can be performed, children at risk of increased ICP. Initial TCD might be considered as a valuable tool to identify children at risk of developing ICH and decreased CPP at admission. These results confirm the benefit of using TCD in children with increased ICP reported by others [10, 14, 17, 18]. It could be argued that TCD examination and ICP monitoring were not performed within the first 3 h in all children admitted to our trauma center during the study period. This may introduce a potential bias of selecting patients according to the severity of the initial presentation. As previously mentioned, the group of patients included in the study did not differ significantly from our general trauma population, invasive ICP monitoring was not used in this population in 20% of children with an admission GCS of 3 who died within the first 6 h in 90% of the cases. Early TCD examination was performed by physicians with extensive experience of this method, and moreover, in patients who could be minimally stabilized during the early phase of resuscitation, eliminating the confounding factor of altered cerebral blood flow velocity related to severe hypotension, profound anemia, or hypo/hypercapnia. Finally, all the intended TCD examination could be performed successfully within the first 20 min after admission without major technical difficulties and none depicted a “no-flow” profile that could be reported to the inability of the operator to identify middle cerebral artery. To our knowledge, the ability of first-line TCD, performed within minutes in the resuscitation room, before CT scan, and ICP transducer insertion, to identify patients at risk of initial increased ICP, has not been reported previously.

We can conclude that TCD examination is a safe, reproducible, and reliable method to identified children at increased risk of ICH and decreased CPP after severe TBI, and its use should be encouraged in PICU. With 94% sensitivity, TCD demonstrate the possibility of screening patients with high levels of ICP, and the negative predictive value of 95% demonstrates its ability to identify patients with normal ICP. It could be used as a first-line non-invasive method and unlike invasive ICP monitoring, it could be performed within the first minutes following admission before CCT scan examination. Other studies must be performed to verify the possibility of its use as an alternative to systematic invasive ICP monitoring.