Screening and selection of articles

During our systematic literature search, a total of 1,903 articles was identified using a predefined search strategy. Following the removal of duplicate records, we screened 1,158 papers based on their title and abstract. This led to the exclusion of 1,110 articles, comprising 78 conference abstracts, 5 editorial comments, 100 case reports, 49 review articles, 176 original papers not involving pediatric populations, 84 original papers not focused on pediatric blunt trauma, and 618 papers unrelated to the topic of interest. Subsequently, the full text of the remaining 48 papers was retrieved and reviewed. After evaluation, 18 articles were excluded because they did not specify the type of trauma center(s) or any of the outcomes of interest. Ultimately, 30 articles that met the inclusion criteria were identified and incorporated into our analysis. The entire screening process and application of eligibility criteria were summarized using a flow diagram in accordance with PRISMA guidelines (Fig. 1).

Fig. 1

PRISMA flow diagram showing the review process. PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Study and participants characteristics

Table 1 provides a summary of the characteristics of included articles. Among the 30 studies examined, 17 reported CT rates in PTCs, 8 reported CT rates in non-PTCs, and 5 provided a comparison of CT rates between PTCs and non-PTCs. These studies were conducted in various countries, including 24 studies from the United States, three from Canada, one from the Netherlands, one from Australia, and one from Turkey. The study designs varied, with 23 retrospective cohorts, 5 prospective cohorts, 1 randomized controlled trial, and 1 case–control study. Trauma center types and levels also differed across the studies, with the majority conducted in level 1 trauma centers (N = 24). The aims of the studies encompassed evaluating CT rates for different types of traumas (N = 10), comparing CT rates among different trauma center types (N = 5),

Table 1 Characteristics of the included studies

developing prediction tools for specific injuries (N = 3), and assessing the effects of interventions and new protocols on CT rates (N = 12). Among the included studies, abdominopelvic CT rates were reported in the majority (N = 19), while rates for head (N = 14), chest (N = 12), cervical (N = 10), and other regions (N = 3) CT scans were less commonly reported.

Table 2 presents a comprehensive summary of the included studies focusing on pediatric patients with blunt trauma and their utilization of CT scans. The table provides information such as the number of patients, gender distribution, age, and ISS for each study. It also displays the proportions of patients who underwent CT scans for specific body regions. Additionally, the table includes reported outcomes such as the admission rate to intensive care units and the mortality rate. In cases where studies differentiated between PTCs and non-PTCs, this distinction is indicated in the table.

Table 2 Characteristic of pediatric blunt trauma patients and utilization of CT scans in included studiesQuality assessment

Table 1 of Supplementary File presents a summary of the responses to each question included in the JBI critical appraisal tool for analytic cross-sectional studies. The results indicate that the majority of the included studies demonstrated acceptable methodological quality.

The table highlights specific studies excluded from the quantitative synthesis. These exclusions were guided by the studies' particular inclusion criteria, which centered on cases with a strong clinical suspicion of certain injury types, including those showing positive signs during physical examination. Excluding these studies aimed to ensure that the quantitative synthesis findings truly represented the wider population of pediatric blunt trauma patients, without being skewed by a subset of cases with specific clinical suspicions.

Quantitative synthesisAbdominopelvic CT scans

We included 18 unique studies in this analysis. Out of these, 16 studies reported rates in PTCs, while 7 studies assessed rates in ATCs/MTCs. The proportion meta-analysis revealed that the overall rate of receiving abdominopelvic CT scans among pediatric blunt trauma patients was 42.5% (95% CI: 32.9%—52.7%) (Fig. 2). There was a high level of heterogeneity among the studies (I2 = 98.6%).

Fig. 2

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving abdominopelvic CT scans in pediatric trauma centers and adult/mixed trauma centers. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. ATC: Adult trauma center. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. MTC: Mixed Trauma Center. PL1&2: Pediatric Level 1 & 2 trauma centers. PTC: Pediatric Trauma Center

Through univariate metaregression analysis, we found a significant effect of trauma center types (PTC vs. ATC/MTC) on the reported rates (p < 0.01, delta-BIC = 2248.5). The subgroup meta- analysis showed that PTCs had a significantly lower abdominopelvic CT rate, with a pooled rate of 35.4% (95% CI: 26.8%—45.1%), compared to 44.9% (95% CI: 27.5%—63.6%) in ATCs/MTCs (Fig. 2). On the other hand, a univariate metaregression based on trauma center level (Level 1 vs. Level 2 or mixed) did not demonstrate a statistically significant effect of trauma center level on the reported CT rates (p = 0.07).

Furthermore, in the univariable metaregression analysis, we found that mean age (p < 0.01, delta BIC = 40.4) and mean ISS scores (p < 0.01, delta BIC = 28.3) significantly affected the observed rates, while publication year did not have a significant effect (p = 0.7, delta BIC = -3.0). Upon visual inspection of forest plots sorted by mean ISS scores and mean age, a trend was observed indicating higher CT rates in higher ISS scores and younger ages (Figs. 1.1 and 1.2 of Supplementary File). It should be noted that mean ISS scores and mean age were reported in only a proportion of the included studies.

Subsequently, we conducted bivariate metaregression analyses to combine the observed significant effects. In the bivariable metaregression of trauma center type and mean ISS scores and mean age, we found that the effect of trauma center type remained significant even after controlling for mean ISS score (p < 0.01, delta BIC = 230.6) and mean age (p < 0.01, delta BIC = 218.9). Furthermore, the effects of mean age and mean ISS scores also remained significant in the bivariate model (p < 0.01, delta BIC = 42.3 and p < 0.01, delta BIC = 29, respectively).

Among the studies included in our analysis, a total of 5 studies comparatively reported abdominopelvic CT rates in ATCs/MTCs compared to PTCs, allowing for an OR meta-analysis. The findings from the comparative meta-analysis revealed a significant difference, indicating that abdominopelvic CT scans are significantly more common in ATCs/MTCs compared to PTCs, with a pooled OR of 1.8 (95% CI: 1.34—2.43) (Fig. 3).

Fig. 3

Forest plot of the random effects odds ratio meta-analysis of five studies that comparatively report the rates of abdominopelvic CT scans in pediatric blunt trauma patients in adult or mixed trauma centers compared to exclusively pediatric trauma centers. ATC: Adult Trauma Center. CI: Confidence Interval. MTC: Mixed Trauma Center. PTC: Pediatric Trauma Center

Cranial CT scans

In this analysis, we included 13 studies conducted in PTCs and 6 studies in ATCs/MTCs that reported rates of cranial CT scans. The overall pooled rate of CT scans across all studies was 38.5%

(95% CI: 27.7%—50.6%), with a high level of heterogeneity (I2 = 99.1%) (Fig. 4).

Fig. 4 

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving cranial CT scans in pediatric trauma centers and adult/mixed trauma centers. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. ATC: Adult trauma center. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. MTC: Mixed Trauma Center. PL1&2: Pediatric Level 1 & 2 trauma centers. PTC: Pediatric Trauma Center

Metaregression analysis examining the effect of trauma center type showed a significant impact on the observed rates (p < 0.01, delta BIC = 1125.3). In subgroup analysis, the pooled rate of cranial CT scans in PTCs was 36.9% (95% CI: 25.3%—50.3%), while in ATCs/MTCs, it was 42.9% (95% CI: 31.4%—55.3%) (Fig. 4). Similarly, metaregression analysis considering trauma center level revealed a significant effect (p < 0.001, delta BIC = 103.6), with level 2 or mixed trauma centers demonstrating a higher pooled rate of 48.7% (95% CI: 28.5%—69.3%) compared to level 1 trauma centers, which had a pooled rate of 34.4% (95% CI: 23.5%—47.3%) (Fig. 5).

Fig. 5 

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving cranial CT scans, stratified by trauma center level. The analysis includes studies conducted in level 1 trauma centers and level 2 trauma centers, as well as studies reporting rates in a combination of both levels. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. PL1&2: Pediatric Level 1 & 2 trauma centers

Furthermore, metaregression analysis of mean ISS scores indicated a significant effect on the reported rates (p < 0.01, delta BIC = 176.2), with studies having higher mean ISS scores showing higher rates of cranial CT scans. Similarly, metaregression analysis of mean age demonstrated a significant effect (p < 0.01, delta BIC = 220.1), with studies having higher mean ages showing lower CT rates (Figs. 2.1 and 2.2 of Supplementary File). However, metaregression did not reveal a significant effect of publication year on the observed rates (p = 0.4).

In the bivariable metaregression considering trauma center type and level, both factors showed significant effects (trauma center type: p < 0.01, delta BIC = 1106.6; trauma center level: p < 0.01, delta BIC = 85). The effect of trauma center type remained significant after controlling for mean age (p < 0.01, delta BIC = 129.1) and mean ISS (p < 0.01, delta BIC = 164). The effect of trauma center level remained marginally significant after controlling for ISS score (p = 0.03, delta BIC = 2.3). However, this effect did not remain significant after controlling for mean age (p = 0.3). It is important to note that the bivariate metaregression controlling for mean ISS included 15 studies, while the bivariate metaregression for mean age included only 13 studies.

Consistent with the metaregression results in the proportion meta-analysis, the comparative OR meta-analysis of studies reporting rates in pediatric and other trauma centers showed that ATCs/MTCs had a significantly higher cranial CT rate compared to PTCs, with a pooled OR of 1.69 (95% CI: 1.2—2.36) (Fig. 6).

Fig. 6

Forest plot of the random effects odds ratio meta-analysis of five studies that comparatively report the rates of cranial CT scans in pediatric blunt trauma patients in adult or mixed trauma centers compared to exclusively pediatric trauma centers. ATC: Adult Trauma Center. CI: Confidence Interval. MTC: Mixed Trauma Center. PTC: Pediatric Trauma Center

Chest CT scans

A total of 15 studies were included in our analysis: 10 studies conducted in PTCs and 5 studies in ATCs/MTCs. Through proportion meta-analysis, we found that the pooled rate of chest CT scans was 19% (95% CI: 14.2%—24.9%) with significant heterogeneity (I2 = 95.4%) (Fig. 7).

Fig. 7

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving chest CT scans in pediatric trauma centers and adult/mixed trauma centers. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. ATC: Adult trauma center. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. MTC: Mixed Trauma Center. PL1&2: Pediatric Level 1 & 2 trauma centers. PTC: Pediatric Trauma Center

When exploring the effect of trauma center type through metaregression, we observed a significant impact on the reported rates (p < 0.01, delta BIC = 3150.2). In subgroup analysis, the pooled rate of chest CT scans in PTCs was 14.5% (95% CI: 9%—22.5%), while in ATCs/MTCs, the pooled rate was 25.4% (95% CI: 19.6%—32.2%) (Fig. 7).

Similarly, metaregression and subgroup analysis based on trauma center level revealed a significantly lower chest CT rate in level 1 trauma centers (pooled rate: 17.5%, 95% CI: 12.1%—24.6%) compared to level 2 or mixed centers (pooled rate: 23.3%, 95% CI: 15.8%—33.1%) (p < 0.01, delta BIC = 54.1) (Fig. 8).

Fig. 8 

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving chest CT scans, stratified by trauma center level. The analysis includes studies conducted in level 1 trauma centers and level 2 trauma centers, as well as studies reporting rates in a combination of both levels. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. PL1&2: Pediatric Level 1 & 2 trauma centers

Furthermore, metaregression analysis of the mean ISS demonstrated a significant effect on the observed chest CT rates, indicating a trend toward higher rates in studies with higher mean ISS scores (p < 0.02, delta BIC = 3.5) (Fig. 3 of Supplementary File). However, univariable metaregression did not show any significant correlation between mean age or publication year and chest CT rates (p = 0.3 and p = 0.8, respectively).

In bivariable metaregression with trauma center types and levels as covariates, both trauma center type (p < 0.01, delta BIC = 3156.1) and trauma center level (p < 0.01, delta BIC = 60) showed significant effects. It is important to note that the number of studies included in this bivariate analysis was relatively small (N = 13).

Lastly, in the OR meta-analysis of 4 comparative studies reporting chest CT rates, we found that ATCs/MTCs had a significantly higher chest CT rate compared to pediatric centers, with a pooled OR of 2.7 (95% CI: 1.19—6.14) (Fig. 9).

Fig. 9

Forest plot of the random effects odds ratio meta-analysis of four studies that comparatively report the rates of chest CT scans in pediatric blunt trauma patients in adult or mixed trauma centers compared to exclusively pediatric trauma centers. ATC: Adult Trauma Center. CI: Confidence Interval. MTC: Mixed Trauma Center. PTC: Pediatric Trauma Center

Cervical spine CT scans

We included a total of 12 studies that reported rates of cervical CT scans in our analysis, comprising 5 studies conducted in PTCs and 7 studies in ATCs/MTCs. The proportion meta-analysis of these reported rates revealed a pooled rate of 28.8% (95% CI: 14.5%—49.2%) with a high level of heterogeneity (I2 = 98.5%) (Fig. 10).

Fig. 10

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving cervical CT scans in pediatric trauma centers and adult/mixed trauma centers. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. ATC: Adult trauma center. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. MTC: Mixed Trauma Center. PL1&2: Pediatric Level 1 & 2 trauma centers. PTC: Pediatric Trauma Center

When conducting metaregression analysis with trauma center type as a covariate, we found a significant effect of trauma center type on the observed rates (p < 0.01, delta BIC = 7). In the subgroup meta-analysis, the pooled rate of cervical spine CT scans in PTCs was estimated to be 23% (95% CI: 9.5%—45.8%), while in ATCs/MTCs, the pooled rate was 45% (95% CI: 31.2%—59.6%) (Fig. 10). However, due to the limited availability of studies reporting cervical CT rates in level 2 or mixed-level trauma centers, it was not possible to perform metaregression or subgroup meta-analysis based on trauma center level.

Furthermore, in the univariable metaregression analysis, we considered mean ISS, mean age, and publication year as covariates. Among these, only mean ISS showed a significant effect (p < 0.05, delta BIC = 2.7), suggesting a trend towards higher cervical spine CT rates in studies with higher mean ISS upon visual inspection (Fig. 4 of Supplementary File). Unfortunately, the limited number of studies prevented the conduction of a multivariable meta-regression analysis and OR meta-analysis for this outcome.

Receiving at least one CT scan

Among the studies included in our analysis, 10 studies reported rates of receiving at least one CT scan (of any type). The proportion meta-analysis of these reported rates yielded a pooled rate of 59.1% (95% CI: 46.5%—70.6%) (Fig. 11). Similar to other reported rates, there was considerable heterogeneity among the included studies (I2 = 98.5%).

Fig. 11

Forest plot of the random effects meta-analysis of the proportion of pediatric blunt trauma cases receiving at least one CT scan (of any type) in pediatric trauma centers and adult/mixed trauma centers. The studies within each subgroup are arranged based on the mean Injury Severity Score. AL1&2: Adult Level 1 & 2 trauma centers. ATC: Adult trauma center. CI: Confidence Interval. ISS: Injury Severity Score. ML1&2: Mixed Level 1 and 2 trauma centers. MTC: Mixed Trauma Center. PL1&2: Pediatric Level 1 & 2 trauma centers. PTC: Pediatric Trauma Center

Upon conducting a meta-regression analysis, we found a significant effect of trauma center type on the rates (p < 0.05, delta BIC = 799.9). Subsequently, the subgroup meta-analysis demonstrated a pooled rate of 54% (95% CI: 42.1%—65.5%) for PTCs, while ATCs/MTCs had a pooled rate of 69.3% (95% CI: 52.9%—82%) (Fig. 11). However, the meta-regression analysis for the effect of trauma center level did not reveal any significant effect (p = 0.5).

Furthermore, we performed univariable meta-regression analyses to examine the effects of mean age, mean ISS, and publication year. Among these variables, only ISS showed a significant association with the heterogeneity of the observed rates (p < 0.01, delta BIC = 6.3). Visual inspection of the forest plots, sorted by mean ISS, indicated a trend towards higher rates in studies with higher ISS scores (Fig. 5 of Supplementary File). However, due to the limited number of studies reporting ISS scores along with the rates of CT use, a multivariable meta-regression analysis was not conducted for this outcome.

Publication bias

The Doi plots representing the proportion meta-analyses for each outcome are presented in Figs. 6.1–6.7 of Supplementary File. Additionally, funnel plots illustrating the OR meta-analysis can be found in Figs. 7.1–7.3 of Supplementary File. When assessing publication bias using the LFK index for the proportion outcomes, minor asymmetry was observed in the cervical CT rates and the rates of receiving at least one CT scan (Figs. 6.4 and 6.5 of Supplementary File). However, significant asymmetry was noted in the Doi plot for the rates of abdominopelvic CT scans (Supplementary Fig. 6.1). Similarly, the funnel plot for the OR meta-analysis of abdominopelvic CT rates showed asymmetry. To address this, we conducted a nonparametric trim-and-fill analysis specifically for OR meta-analysis of abdominopelvic CT rates (Fig. 7.1 of Supplementary File). However, the inclusion of additional studies did not alter the significance level of the observed outcome, and the pooled OR remained at 2.1 (95% CI: 1.6—2.9) following imputation.