Introduction

Traumatic brain injury (TBI) remains a major cause of mortality and long-term disability in children worldwide, affecting more than 3 million children every year with a conservative incidence estimate of 50 per 100,000 persons [1]. Approximately, 1 out of 7 children with TBI seeking emergency medical care from hospitals suffer from moderate-to-severe TBI, with mortality rates in severe TBI up to 24% [2,3,4]. Many factors ultimately influence the outcome of paediatric TBI (pTBI) patients. The non-modifiable primary brain injury resulting from an external force is subsequently followed by a pathophysiologic cascade of events, including cerebral swelling, metabolic disturbances, seizures, cerebral vasospasms, and neuroinflammation which all contribute to secondary injury. The resulting sustained elevated intracranial pressure (ICP) is a key variable, as high ICP may further impair blood flow, thereby causing brain ischemia [5,6,7].

Clinical management of pTBI patients admitted to the paediatric intensive care unit (PICU) is targeted at limiting this secondary injury. However, there is little high-quality scientific evidence that supports the current management of pTBI. Randomized controlled trials evaluating management of moderate and severe TBI in children are very scarce [5, 8]. This lack of good evidence is reflected in the international guidelines for severe pTBI, at best providing level II recommendations [6]. However, it has also been reported that patients with moderate pTBI (i.e. those with Glasgow Coma Scale [GCS] 8–12) may suffer from physical and cognitive impairments and reduced adaptive functioning [2, 3, 9]. Also, the initial assessment of GCS is not always accurate and may lead to over- or underestimation of neurological severity [10]. As such, therapeutic interventions targeted for patients with severe TBI may potentially also offer outcome benefits in patients with moderate TBI. We therefore sought to characterize the outcomes of a cohort of children with moderate-to-severe TBI by using the Paediatric Cerebral Performance Category (PCPC) as we did not have data on more extensive psychometric testing after PICU admission [11, 12].

Materials and methods

Study population

This study was designed as a descriptive, retrospective study obtained from written and electronical health records of children < 18 years with moderate to severe TBI (GCS ≤ 12) admitted between July 2010 and June 2020 to the PICU of the Beatrix Children’s Hospital/University Medical Center Groningen (UMCG), a level 1 trauma centre. Patients with (suspected) abusive head trauma were excluded. The Institutional Review Board approved the study and waived the need for informed consent (Institutional Review Board UMCG, METc2021/534).

PICU management of children with TBI is guided by local practice algorithms based on the international paediatric guidelines [6]. Updates of these international guidelines were also implemented in our local practice algorithm. Depending on severity of injury, patients were mechanically ventilated in a time-cycled, pressure-limited mode of ventilation. All ventilated patients received analgesia and sedation by continuous intravenous infusion with benzodiazepines, opioids, and/or propofol. Wake-up calls were done at the discretion of the physician in charge if interruption of sedatives was not contraindicated, e.g. stable ICP < 20 mmHg and stable cerebral perfusion pressure or no clinical signs of raised ICP. After our local clinical algorithm was revised in 2017, amplitude integrated electroencephalography was used in all patients for detection of subclinical epileptic activity. Insertion of an ICP manometer was ultimately at the discretion of the attending neurosurgeon. A stepwise approach was used when there were signs of acute intracranial hypertension (for patients with ICP monitoring ICP > 20 mmHg for at least 5 min or bradycardia ± relative hypertension in patients without ICP monitoring), including checking for optimal baseline care and administration of hyperosmolar agents, with a shift in time from the use of mannitol towards the use of hypertonic saline. Cranial computed tomography scans were evaluated by one (paediatric) radiologist at the time of PICU admission and not reanalysed for this study.

Data collection

We collected baseline patient characteristics, trauma mechanism, the presence of associated lesions, data on (pre-) hospital treatment, radiological features on the first cranial computed tomography, and clinical, physiological and laboratory data from the first 7 days of PICU admission (definitions are described in Supplemental Table 1). The Paediatric RISk of Mortality II (PRISM-II) — 24-h score was calculated to assess patient acuity [13]. The GCS and paediatric trauma score were calculated to assess TBI and trauma severity [14, 15].

Endpoints

Favourable outcome was defined by PCPC ≤ 3, reflecting normal outcome, mild and moderate disability. Poor outcome was defined by PCPC scale ≥ 4, reflecting severe disability, coma or vegetative state and death or brain death [11]. PCPC score was determined at PICU discharge using discharge notes and letters. Secondary outcomes included mortality < 24 h of admission, PICU-free days at day 28 (defined as the number of days alive and out of the PICU during the first 28 days), PICU and hospital length of stay (LOS), ventilator days, GCS at PICU discharge and discharge status (home, other hospital or rehabilitation facility). Patients who died before PICU discharge within 28 days were assigned zero PICU-free days.

Statistical analysis

We stratified patients by PCPC (i.e. ≤ 3 and ≥ 4). Categorical variables were described as absolute number and percentage (%) of total and continuous variables as median and interquartile range (IQR). Continuous variables were analysed with the Mann–Whitney U-test, and categorical data was analysed using the χ2 test (Fisher exact test if the value of any cell was < 5). The Wilcoxon signed-rank test was used for repeated measurements comparing weighted means from days 0–1 to 2–3 of PICU admission. All analyses were performed with SPSS v23.0 (IBM Statistical Package for the Social Sciences [SPSS] for Windows, Armonk, NY: IBM Corp.). P-values < 0.05 were accepted as statistically significant.

Results

During the study period, 434 patients with the admission diagnosis “trauma” were identified from the PICU database. After exclusion of patients without TBI (N = 201), mild TBI [i.e. GCS > 12] (N = 97) or (suspected) abusive head trauma (N = 6), data from 130 patients was eligible for analysis (Supplemental Fig. 1), of whom N = 56 (43.1%) had moderate TBI and N = 74 (56.9%) had severe TBI. One-hundred and six patients (81.5%) had favourable neurological outcome (i.e. PCPC ≤ 3) at PICU discharge (Fig. 1). Of those patients, N = 55 (51.9%) had moderate TBI and N = 51 (48.1%) severe TBI.

Fig. 1
figure 1

Paediatric Cerebral Performance Category Scale, percentage of patients per category split by severity of neurotrauma

Full size image

Patients with accidental fall as the cause of injury significantly more often had favourable neurological outcome, N = 44 (97.8%). Regarding patients with PCPC ≥ 4, significantly more patients had lower GCS, lower initial motor score, bilateral dilated and fixed pupils, higher pre-hospital intubation rate, more associated lesions (including more thoracic trauma), more hypotension upon hospital arrival and more often pre-hospital cardiac arrest. Paediatric trauma score and PRISM II were significantly higher in patients with PCPC ≥ 4 (Table 1). All patients with PCPC ≥ 4 had cranial computed tomography lesions, with significantly more subdural hematomas, intraparenchymal and intraventricular haemorrhages, the presence of midline shift and absent cisterns, whereas 19% of patients with PCPC ≤ 3 had no cranial computed tomography lesions (Supplemental Table 2).

Table 1 Demographic, injury and early resuscitation characteristics
Full size table

PICU-free days at day 28 were significantly lower in patients with PCPC ≥ 4 (median 0 days, IQR 0.0–16.7) compared to PCPC ≤ 3 (median 26.3 days, IQR 23.4–27.2). No significant difference in PICU LOS or hospital LOS between patients with PCPC ≤ 3 and PCPC ≥ 4 was found (Table 2). Patients with PCPC ≥ 4 had significantly more ventilator days and lower GCS at PICU discharge. Discharge to a rehabilitation facility was more common among surviving patients with PCPC ≥ 4 (N = 11 [91.7%]) compared with patients with PCPC ≤ 3 (N = 22 [20.8%]) (Table 2). Twelve patients (9.2%) died, of whom 8 (66.7%) within 24 h of PICU admission. Mortality was the highest among patients with severe TBI (N = 11 [14.9%]).

Table 2 Outcome categorized by Paediatric Cerebral Performance Category Scale ≤ 3 and ≥ 4
Full size table

We found no difference in the number of patients with decompressive craniectomy between the two groups. All decompressive craniectomies were performed on the day of admission for evacuation of extracerebral hematomas. ICP monitoring was performed in N = 12 (50%) patients with PCPC ≥ 4 and in N = 15 (14.2%) patients with PCPC ≤ 3. The median age of these patients was 11.8 years, with an IQR of 8.4–14.9 year. Use of vasopressors, neuromuscular blocking agents and bolus(es) of hypertonic saline or mannitol was significantly more common in patients with PCPC ≥ 4 during the first 72 h of PICU admission. We also observed that more patients with PCPC ≥ 4 were on benzodiazepines, opioids or propofol during days 2–3 (Table 3). Also, initiation of enteral feeding in the first 72 h of admission was significantly lower in this category.

Table 3 Inhospital treatment, monitoring and administration of continuous medication on days 0–3 categorized by Paediatric Cerebral Performance Scale ≤ 3 and Paediatric Cerebral Performance Scale ≥ 4
Full size table

Except for significant differences between patients with PCPC ≤ 3 and PCPC ≥ 4 in weighted averaged serum sodium concentrations and temperature on days 0–1 (140 mmol/l vs. 144 mmol/l and 36.9 °C vs 36.5 °C), days 2–3 (141 mmol/l vs. 145 mmol/l and 37.2 °C vs. 36.8 °C), and serum glucose on days 0–1 (6.0 mmol/l vs. 7.2 mmol/l) (Fig. 2), no other differences were observed (Supplemental Table 3). In approximately one-third of patients in both outcome groups, we observed mean arterial blood pressure [(MAP)] between p5-and p50 (N = 32 [34.4%] vs. N = 8 [36.4%]) on days 0–1 and N = 20 (41.7%) vs. N = 6 (37.5%) on days 2–3 (Fig. 2).

Fig. 2
figure 2

Inhospital measurements on days 0–1 and days 2–3 categorized by Paediatric Cerebral Performance Category Scale ≤ 3 and Paediatric Cerebral Performance Category Scale ≥ 4

Full size image

Discussion

In the present study, we observed significant morbidity and mortality in severe pTBI, whereas subjects with moderate TBI had a favourable outcome. Overall mortality rate was high and in two-thirds of the patients within the first 24 h of PICU admission, which is comparable to previously reported [16,17,18,19]. We observed similar differences in injury characteristics such as trauma mechanism [20], fixed pupils [16, 21,22,23,24] and low motor score [16, 25]. Significantly, more associated lesions were seen in the poor outcome group, including more thoracic injuries, which make patients prone to secondary injury due to hypoxia or hypovolemia [18, 21, 26]. These findings underscore the need for preventive measures despite advances in management and neuromonitoring in pTBI as it remains subject of debate if outcomes can significantly be improved by PICU management [7]. Thus, preventive measures, such as speed limits, traffic education and the use of bicycle helmets [27, 28], are key in reducing early mortality and long-term morbidity.

Similar to observations by others, fall and road traffic accident (either as pedestrian, cyclist or passenger) are the most common cause of trauma [1, 18, 29], with fall being more present in younger subjects and traffic accidents in older subjects [3, 30, 31]. We, like others, found that subjects with a fall had more often a favourable outcome despite the initial injury severity [20]. This may be explained by a different trauma impact compared with the acceleration-deceleration forces that account for immediate shearing of connective nerve fibres in traffic-related injuries [7].

While we found in our cohort that subjects with moderate pTBI almost all had favourable outcome at PICU discharge, it has been reported that these patients may suffer from long-term physical and cognitive impairments and reduced adaptive functioning [2, 3, 7, 9]. Also, initial assessment of GCS may not always be adequately representative for the injury severity. For example, the neurological status may have changed from the initial score assessed at the trauma scene to the score calculated in the emergency department, or the score may have been simply miscalculated. This may thus lead to over- or underestimation of the true neurological severity [10]. With this in mind, we therefore decided to include subjects with moderate pTBI as well in our analysis. Furthermore, it may be speculated that therapeutic interventions targeted for patients with severe TBI may also offer outcome benefits in patients with moderate TBI, although this remains to be further studied.

We acknowledge that our study is affected by the fact that we arbitrarily defined neurologic good outcome by PCPC ≤ 3 since data on Glasgow Outcome Scale-extended (GOS-E) or GOS-E Peds were unavailable [11, 12, 32, 33]. PCPC ≤ 3 includes patients with mild-to-moderate disability at PICU discharge, as reflected in the number of subjects in our cohort discharged to a rehabilitation facility. It can be argued that PCPC may not be granular enough to discriminate between good and poor outcome, and our study was not designed to explore long-term patient outcome. It is known that lifelong disability in physical, cognitive, behavioural and social function is common in patients with TBI [3, 34], although further neurological improvement from poor to favourable ≥ 6 months after PICU discharge may still occur [20]. One group of investigators reported that 61% of children with moderate-to-severe TBI required specialized medical and educational services 1 year after trauma [9]. Others found that almost two-third of patients with moderate-to-severe TBI functioned similar to normative age peers 10-year post-trauma [31]. This signifies the need for long-term follow-up to evaluate functional status over time and to intervene with appropriate rehabilitation to get the full potential for (complete) recovery [28, 35].

In principle, patients with especially severe pTBI were managed according to the international pTBI guidelines [6]. Nonetheless, we observed a low use of ICP monitoring. Due to the retrospective nature of the study and the limited number of data points, only weighted average analysis of ICP and CPP was done; a more detailed analysis considering minute-by-minute fluctuations in these parameters was not feasible. We did find statistically significant, but clinically irrelevant differences in serum sodium between subjects with PCPC ≤ 3 and ≥ 4, but this must be seen in the context of injury severity and more subjects with PCPC ≥ 4 receiving hypertonic saline or mannitol. As such, we could not confirm an association between serum sodium > 150 mmol/l and mortality [36], but this is probably due to the fact that we included subjects with moderate pTBI in our analyses. Nonetheless, it appears justified to implement advanced neuromonitoring or measurement of biomarkers for individualized post-trauma care to optimize outcome [37,38,39,40].

There are several limitations to our study that need to be discussed. First, our study was designed as a retrospective observational, single-centre study. This limits generalizability of our findings. Furthermore, data was obtained from the patient’s medical record; thus, inherently, clinical and outcome data may have been missing, leading to selection bias. Missing clinical data were not imputed. Second, although our unit implemented the international paediatric TBI guidelines, practice variability leading to confounding by indication (i.e. the sickest patient is the most likely to get a specific intervention) may have significantly impacted our study findings. This type of bias can only be overcome by randomization. Third, the period studied includes revisions from the international paediatric TBI guidelines in 2012 and 2019 [6]. Our study was not designed as a quality study, so we could not ascertain how changes in clinical management were implemented and thereby affected the primary outcome.

Conclusion

We report significant morbidity and mortality in subjects with severe pTBI, whereas subjects with moderate TBI had a favourable outcome. Overall mortality rate was high and in two-thirds of the patients within the first 24 h of PICU admission. Significantly more associated lesions were seen subjects with poor outcome. These findings underscore the need for preventive measures such as speed limits, traffic education and the use of bicycle helmets to reduce early mortality and long-term morbidity in pTBI. Follow up after discharge is essential to optimize outcome on physical, cognitive, behavioural and social functioning.