During the study period, the medical records of 4943 patients were retrospectively analyzed. We considered 3832 patients eligible for the study (mean age 5.3 years, SD 4.8). 2613 patients were ≥2 years, while 1219 were younger. 551 patients (14.4%) underwent CT examination, of which 96 were younger than 2 years (17.4%) and 455 (82.6%) were age ≥2 years (median age 10, 99 years; 25th percentile = 5.78; 75th percentile = 14.35). (Figure 2). All the characteristics of the study population are summarized in Table 1.
We first examined the group of children ≥2 years, with a median age of 6,015 (IQR 3.44-10.73).
907 patients received frontal trauma (34.71%), 293 parietal (11.21%), 161 temporal (6.16%), 623 occipital (23.84%), 161 facial trauma (6.16%) and in approximately 689 cases it was not possible to identify a site of the trauma.
Applying the PR, we observed that for 30 children (1.15%) the CT scan was recommended and 29 of these performed it (96.7%). The PR recommended clinical observation for 653 of these patients (25%). Of these 286 performed a head CT scan (43.8%). CT scan was instead to be avoided according to the PR in 1930 patients (73.85%), but 140 patients underwent head CT scan in this risk category (7.26%).
In this age group, 10 patients presented ciTBI of which 7 underwent neurosurgery and 3 hospitalized over two nights. (Table 1)
The PR assigned 2 of the patients undergoing surgery to the group "CT scan recommended”, while the other 5 were classified in the "CT scan versus observation" group and all presented clinical deterioration within the first hours of observation.
Regarding the 3 hospitalized patients with ciTBI, according PR 2 were classified as a "CT scan recommended" group, and 1 was part of the “CT scan versus observation” group but, within the first hours of observation, he developed marked sleepiness.
Therefore, applying the PR, no patient with ciTBI would have been discharged without a diagnosis, with a sensitivity of 100%.
455 patients (17.41%) underwent head CT scans of which 40 were abnormal.
Among the 34 patients with CT scan alterations considered by the rule belonging to the "CT scan versus observation" group, 28 had a fracture (of which 10 were multiple), 4 had subarachnoid haemorrhage and 2 subdural hematomas. 5 patients with CT scan alterations were considered by the PR to be “at high risk”, 4 had skull fractures and 1 had a left temporal parenchymal hematoma with intraventricular haemorrhage from post-traumatic rupture of arteriovenous malformation.
Only 1 patient with CT scan changes, considered by the PR “at low risk”, had a subarachnoid haemorrhage in the left frontal area, that did not require neurosurgery.
Besides the statistical correlation between the PECARN recommendation categories and the presence of CT scan changes was statistically significant (p = 0.000).
We then carried out in this population a detailed analysis of the variables considered by the PR, and we analysed them individually and after multivariate analysis.
We confirmed that presenting episodes of repeated vomiting (OR: 6.0, CI: 1.2–6.3), a severe mechanism of trauma (OR: 3.4, CI: 1.6-7.1), receiving trauma in the parietal site (OR: 2.8, CI: 1.2 –6.3) and in the occipital site (OR: 2.1, CI: 1.0–4.3) remain to be independently associated with the presence of alterations at head CT scan.
The analysis of the ROC curves of the PR showed a sensitivity of 97.5% (CI 86.8% - 99.9%) in identifying patients with CT scan alterations and a specificity of 33, 5% (CI 29% - 38.3). (Figure 3)
In the subpopulation of 1219 patients <2 years enrolled in the study the median age was 0.99 years (IQR 0.63-1.39).
477 patients received frontal trauma (39.1%), 126 parietal (10.3%), 56 temporal (4.6%), 151 occipital (12.4%), 36 facial trauma (3%) and in approximately 441 cases it was not possible to identify a site of the trauma.
According to the PR, for 17 children (1.4%) CT scan was recommended and 16 performed it (94%). The PR recommended clinical observation for 358 patients (29.4%), of these 56 performed a head CT scan (15.6%). CT scan was instead to be avoided in 844 patients (69.20%) but 24 underwent head ct scan in this risk category (2.84%).
In the group of children <2 years only 3 patients presented ciTBI, 2 underwent neurosurgery and 1 hospitalized more than two nights. The PR assigned one of the patients undergoing surgery to the group "CT scan recommended” and the other in the "CT scan versus observation" group. The latter, few minutes after the start of clinical observation, developed profound drowsiness and CT showed an epidural hematoma. The last child with ciTBI in this age group, classified in the "CT scan versus observation" group, presented clinical worsening and irritability for which he performed an intensive observation in PICU and prolonged hospitalization. Therefore, applying the PR also in this age group, no ciTBI would have been discharged without diagnosis with a sensitivity of 100%.
96 patients (7.86%) underwent head CT scans of which 49 were abnormal.
Among the 40 patients with CT scan alterations considered by the rule belonging to the "CT scan versus observation" group, 39 had a fracture (of which 12 were multiple), while the only 1 presented an isolated left frontal subarachnoid hemorrhage.
All the 8 patients with CT alterations, considered by the PR to be “at high risk”, presented skull fractures of which 5 were parietal, 1 occipital and 2 multiples.
Also in this age group the statistical correlation between the PECARN recommendation categories and the presence of CT scan changes was statistically significant (p = 0.000).
Instead, only the younger age at the moment of trauma is independently associated with CT scan alterations at the multivariate analysis.
In children under 2 years, the analysis of the ROC curves of the PR showed a sensitivity of 97.96% (CI 89.1% - 99.9%) in identifying patients with CT scan alterations and a specificity of 48.94% (CI 34.1% - 63.9%). (Figure 4)