During the study period, 454 children were admitted in our ICU for traumatic head injury. Two hundred and seventy-six patients had an isolated traumatic head injury; they were all included in this study. Victims were rescued and brought to our hospital either by fire fighters or by a pre-hospital emergency medical services team in 32% of cases. Sixty-eight percent of our patients were brought by the own facilities of their family members. One hundred and twelve patients (40%) came from Sfax city or its neighborhood; however, 60% came from other cities of south Tunisia.
There were 196 male (71%) and 80 female patients (29%) with a mean age of 6.7 ± 3.8 years (range = 0.3–15). Fifteen percent of children were aged less than 2 years, 17.8% were 3–5 years old and 41% were 6–10 years old. The demographic and clinical parameters on admission are shown in . Head traumas were mainly from traffic (58.3%) and home (39.1%) accidents. However, head injuries from home accidents were commoner among younger children than older children compared to traffic accidents and vice versa . The mean GCS score on admission was 8 ± 2 points. The head trauma was complicated by mild brain injury (GCS, 13–15) in 17.8%, moderate brain injury (GCS, 9–12) in 33.3%, and severe brain injury (GCS, 3–8) in 48.9%.
Demographic and clinical parameters on admission of all study population
Cause of head trauma in each age groups
In our study, brain CT-scan was performed on admission for 270 patients. For the remaining six patients, MRI was performed instead because of logistic difficulties. At admission, 259 patients needed intubation, sedation, and mechanical ventilation with a mean duration of 4 ± 6 days. On admission, 69 (25%) patients needed craniotomy. The most neurosurgical intervention were evacuation of a subdural hematoma (N = 6), evacuation of an extradural hematoma (N = 30), lobectomy (N = 3), elevation of depressed skull fracture (N = 10), and decompressive craniectomy in one case.
The results of brain CT-scan are presented in . According to Marshall tomographic grading “Traumatic Coma Data Bank classification” there were 29% type I, 37.7% type II, 9.8% type III, 1.1% type IV, 17.4% type V, and 5% type VI. In this study, normal cerebral CT scans were observed in 40 patients (14.5%). During the ICU stay, 178 patients (64.5%) had complications: nosocomial infections 46 cases (16.7%), pneumonia 27 (9.7%), tract urinary infection 9 (3.2%), meningitis 6 (2.1%), septicemia 6 (2.1%), and inner ear infection or sinusitis 6 (2.1%). During the ICU stay, 101 patients (36.6%) had required fluid resuscitation. Catecholamines were used in 14 patients (5.1%) (dopamine in 12 patients (4.3%), epinephrine 5 (1.8%), and dobutamine 1 (0. 3%)).
Cerebral CT-scan findings among patients
A total of 95 patients (34.4%) had rhabdomyolysis (CPK > 500 IU/l).[25
] Hyponatremia (<130 mmol/L) was present in 76 (27.5%), hypernatremia (>145 mmol/L) in 19 (6.9%), diabetes insipidus in 5 (1.8%), stage III or IV pressure ulcer[26
] in 10 (3.6%), and neurogenic pulmonary edema in 6 (2.1%).
During the ICU stay, 222 patients (80.4%) developed SSIs. shows the frequency of each SSI. Finally, 260 (94.2%) patients developed during their ICU stay one or more organ failure.
Frequency of secondary systemic insults among patients
Mean ICU stay was 5.8 ± 102 days. Forty-eight patients (17.4%) died. Regarding the time of death, the mortality percentage was 60.4% in the first 24–48 h, 23% between 3 and 7 days, and only 16.6% thereafter. Brain herniation diagnosed clinically was the main cause of mortality (62.5%), whereas the other cases of mortality were: acute respiratory distress in 8.3% and sepsis with multiorgan failure in 22%.
Among the 228 survivors, 19 patients (8.3%) had a functional motor deficit, 15 (6.6%) had subjective symptoms, and 16 (7%) had posttraumatic seizures. The Glasgow Outcome Scale performed within a mean delay at 12.7 months after hospital discharge (range = 0.5–96 months) were as follows: 48 deaths (17.4%), 2 vegetative state (0.7%), and 176 good recovery (63.8%).
Univariate analysis showed that low PTS on admission, high ISS, high PRISM, presence of shock, menigeal hemorrhage, a serum glucose > 11 mmol/L and bilateral mydriasis are associated with mortality . Finally, in our study, a Na+ > 145 mmol/L on ICU admission (maximal level encountered 154 mmol/L) was associated with a poorer outcome (P < 0.0001).
Factors associated with death in univariate analysis
The multivariate analysis showed that factors associated with a poor prognosis were PRISM > 24 (P = 0.001; OR = 10.98), the presence of neurovegetative disorders (P = 0.004; OR = 7.1), type IV lesion. According to Marshall tomographic grading (P = 0.02; OR = 13.2) and the presence of meningeal hemorrhage (P = 0.03; OR = 2.74).
A significant association was found between PRISM score and mortality rate. This model had a high discriminative power. In fact, a PRISM score > 24 was associated with death with a sensitivity of 37%, a specificity of 99.95%, and with an area under the ROC curve at 0.85 .
In addition, as shown in , a low value of GCS score on admission was associated with a poor outcome. In fact, GCS score ≤ 8 was associated with death with a sensitivity of 62%, a specificity of 74%, and an area under the ROC curve at 0.70.
Mortality rate correlated with Glasgow Coma Scale Score
In our study, we founded a good correlation between the PRISM score and the outcome. In fact, a PRISM score > 24 was associated with a poor outcome with a sensitivity at 37% and specificity at 99%, and an area below the ROC curve at 0.85 . However, ISS and PTS are not enough discriminating with areas below the ROC curve at 0.0.62 and 0.72, respectively.
Roc curve for ability of PRISM score to predict mortality
According to “Traumatic Coma Data Bank” classification, mortality rate was at 6.25% in type I group, 7.7% in type II, 37% in type III, 33% in type IV, 23% in type V, and 93% in type VI (P < 0.001) . Moreover, as shown in , the mortality rate was narrowly related with the number of developed SSIs (P < 0.0001). In fact, mortality rate was increased from 11.2% in patients with only one SSI to 60% in those having more than five SSI.
Mortality rate according to “Traumatic Coma Data Bank” classification
Association between mortality and the number of developed secondary systemic insults (SSI)
Finally, the development of organ failure was associated with mortality (P < 0.001). In fact, mortality rate was increased from 0% in patients without organ failure to 64% in those having more than four organ failures. shows the association between of each organ failure and the mortality.
Association between organ failure and prognosis