In this secondary analysis of a large prospectively populated pediatric minor head trauma database, we found that children with palpable signs of skull fractures on examination of the scalp represented a minority of the population, but they underwent head CT scans in approximately 70% of cases. CT scan rates were more than 90% when excluding palpable signs of skull fractures marked as “unclear exam” on the case report forms. Furthermore, although children with palpable skull fractures were found to have higher frequencies of depressed skull fractures, any skull fracture on CT, TBI on CT, and ciTBI than those without palpable signs, most did not have these outcomes. This shows that clinicians overestimate the presence of depressed skull fractures when examining children with minor head trauma. In children younger than 2 years of age approximately only one in five of those with “palpable skull fractures” will actually have a depressed skull fracture on CT, while 60% will have any skull fracture. At all ages, however, higher percentages of all study outcomes were found in children with palpable signs of skull fractures marked as “yes,” compared to those defined as having “unclear exam.” It is intuitive that higher clinicians’ confidence of their clinical examination would be associated with greater diagnostic discriminatory ability, thus justifying higher CT rates in children defined as having clear signs of palpable skull fractures.

The frequencies of our study outcomes were higher for children < 2 years old with palpable signs of skull fractures compared to children 2–18 years old, both overall, and with respect to perceived reliability of scalp physical examination findings (i.e., findings marked as “yes” versus “unclear exam”). These results are consistent with the parent study and are related with the different anatomical characteristics of the skull and scalp in younger children. The thinner calvarium, more vascularized scalp and looser subcutaneous tissues facilitate deformation and fractures of the skull and the development of large scalp hematomas as a result of direct trauma in younger children [12]. In both age groups, patients with palpable signs of skull fractures more often presented with other associated PECARN TBI rule predictors compared to children without these signs. This may reflect the higher head impact sustained by children who have skull fractures, thus explaining a higher frequency of associated symptoms.

Differently from our work, a recent secondary analysis of a multicentre prospective study including 1,018 children with blunt head trauma (of whom 85 (8.3%) had any skull fracture on CT, and 18 (1.8%) had depressed or basilar skull fractures) showed that physical examination was highly specific, but poorly sensitive in identifying skull fractures [13]. However, this study only included children who underwent head CT; the analysis grouped together depressed and basilar skull fractures and did not analyze separately younger and older children. Participants included children with GCS scores of any severity, with 72% reported to have GCS scores of 15. These reasons likely explain the lower clinician sensitivity found in that study compared to ours.

Other large multicenter prospective studies found that suspected depressed skull fractures on scalp examination was a predictor of TBIs in children, thus recommending a head CT scan in children with these physical examination findings [4, 5, 14]. However, the above studies grouped the findings of suspected depressed fractures either with penetrating skull injuries or tense fontanelle [5], or with suspected open fractures [4], and overall included fewer than 200 patients with these findings in each study. In addition, these studies did not separately analyze those patients with unclear scalp findings due to the presence of overlying large and boggy scalp hematomas as we did in this study. Large and/or boggy scalp hematomas complicate the identification of a step-off of the skull on physical examination, with the diagnosis more challenging and less reliable.

Skull ultrasound has proven to be accurate in identifying skull fractures underlying scalp hematomas [8, 9, 12, 15, 16] and, most importantly, can define their characteristics (depressed or complex fractures) [7]. POCUS could thus be a valuable adjunct to refine clinical decision-making in children with palpable signs of skull fractures according to the PECARN TBI rule. In those children with unclear physical findings of skull fractures and no other age-based high-risk PECARN predictors, the absence of fractures detected on skull ultrasound may support clinicians’ decision to opt for close ED observation rather than for a head CT. In contrast, the detection of a depressed skull fracture on ultrasound would prompt the performance of a head CT scan.

Our data, however, do not provide assistance about the best clinical course when a linear uncomplicated skull fracture is detected on ultrasound in otherwise asymptomatic children. Older reports have demonstrated that skull fractures identified on plain skull radiographs are associated with significantly higher rates of TBIs both in children and adult head trauma patients, although up to 50% of TBIs can occur in the absence of skull fractures [17,18,19,20]. At the time of those studies, however, ultrasound was not routinely available or used as a bedside screening tool. Previous studies and guidelines thus recommended skull radiographs as a screening tool for TBIs in the assessment of young infants with large scalp hematomas as the sole manifestation of head trauma and recommended CT scans in infants with skull fractures identified on radiographs due to the higher risk of associated TBIs [21,22,23,24]. In the era of high-quality clinical prediction rules and patient-centered outcomes, better estimates of the risk of clinically important injuries are available for head-injured children. The most recent prospective multicenter prediction rule studies on pediatric head trauma, however, have not separately analyzed the association between linear uncomplicated skull fractures on CT scan and the presence of TBIs in otherwise asymptomatic children [2, 4, 5, 14]. On the other hand, a recent systematic review and meta-analysis has shown that children with isolated uncomplicated linear skull fractures on CT scan have essentially no risk of neurologic deterioration and can be safely discharged home if asymptomatic and in the absence of suspicion of non-accidental injury [25,26,27].

The results of our study must be interpreted in light of its limitations. First, other clinical findings on examination of the scalp, such as crepitus, and not only the palpation of a step-off of the skull, may have contributed to the classification of physical findings as palpable skull fractures by clinicians. Unfortunately, we could not explore this nuance in our dataset given how the data were collected. Second, not all children in the study received head CT scans. Our work, however, is based on a prospective observational study that was not meant to interfere with routine clinical decision-making, and mandating the performance of a CT scan in all participants with head trauma would have been unethical. Nonetheless, more than 70% of children classified as having palpable skull fractures on examination received a CT scan increasing the likelihood that our results accurately reflect those that would be obtained in the overall population of children with palpable signs of skull fractures in general practice. Third, although to our knowledge we used the largest available prospective pediatric head trauma database, our sample size of patients with scalp findings and skull fractures is still relatively small, as shown by the relatively wide 95% CI for some subgroup analyses. Fourth, nearly 20 years have elapsed since data collection and practice has changed over time. However, given the large rigorous prospectively collected database used for our secondary analysis, and the nature of the research question explored, this is highly unlikely to have influenced our findings. In addition, our study shows how publicly available datasets from high quality studies are useful to explore refinement in clinical management when new tools are introduced in clinical practice. Finally, our results do not apply to the definitions of suspected depressed skull fractures used in other clinical prediction rule studies and thus cannot be extended to clinical practice in settings using those rules [4, 5].

Despite these limitations, our findings have important implications for clinical practice. We demonstrated that children with palpable signs of skull fractures following minor head trauma have higher frequencies of depressed skull fractures and TBIs on CT than those without. However, the discriminatory ability of the scalp examination is suboptimal and could be enhanced by direct bedside visualization of fracture characteristics, such as through skull ultrasound.