Traumatic posterior fossa extradural hematoma in children: a meta-analysis and institutional experience of its clinical course, treatment and outcomes Authors: Keng Siang Lee, Shi Hui Ong, Conor S Gillespie, Lee Ping Ng, Wan Tew Seow, Sharon YY Low Published: 2024 ABSTRACT AbstractPosterior fossa extradural hematoma (PFEDH) is rare but has a greater incidence amongst children. It is also associated with a rapid deterioration. The aim of this study was to present the management of PFEDH through our institutional experience and a meta-analysis. A retrospective single institution review of all children from 2004 to 2024 who underwent craniotomy for PFEDH was undertaken. The collected variables included: demographics, type of trauma, clinical findings, computed tomography findings, and clinical course. A systematic review using Ovid Medline, Ovid Embase, and Cochrane Central Register of Controlled Trials (CENTRAL), and meta-analysis were performed. Nineteen children with PFEDH who underwent surgery were identified. All 19 (100%) patients benefited from good Glasgow Outcome Scale (GOS) score 4–5, and there were no incidences of in-hospital mortality. From the systematic review, 391 patients, across twenty-four studies and our series, were included. A total of 308 were treated with surgery, whereas 83 patients were treated conservatively. A comparative meta-analysis was not performed as the two groups were deemed too heterogeneous in clinical characteristics. Instead, single-arm meta-analyses were performed. The pooled incidence of patients initially under conservative management requiring surgery was 9.90% (95%CI 1.61;22.21%, I2 = 35.2). The incidence of good functional outcomes in patients managed surgically and conservatively were 93.68% (95%CI: 88.69;97.57%, I2 = 0.0%), and 99.99% (95%CI: 96.53;100%, I2 = 0.0%), respectively. Overall pooled of mortality in patients managed surgically and conservatively were 0.57% (95%CI: 0.00;2.87%, I2 = 0.0%) and 0.00% (95%CI: 0.00;1.18%, I2 = 0.0%). Overall, our study reiterates that pediatric PFEDH is uncommon, and patients often present atypically. Based on our institutional experience and extrapolating data from our meta-analysis of the wider literature, neurosurgical intervention is a reliable therapeutic option with good clinical outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10143-024-03089-2.Keywords: Extradural, Hematoma, Pediatric, Posterior fossa, Traumatic Introduction Traumatic brain injury (TBI) in the pediatric population is a global cause of mortality and chronic disabilities; the latter imposing significant healthcare burden costs [1, 2]. Within the spectrum of TBI, posterior fossa extradural hematomas (PFEDHs) are rare and account for up to 3% of all extradural hematomas (EDH) [3–7]. PFEDH are more common in children [8, 9], and of significance, untreated cases are associated with rapid deterioration and demise [6]. However, PFEDH tends to present atypically – leading to the risks of dire consequences of a late or missed diagnosis [10]. Nonetheless, the widespread availability of computed tomography (CT) brain scans has led to better diagnostic capability, earlier neurosurgical intervention and overall, improved prognosis [5, 8, 11–14]. To date, most cases of traumatic PFEDH in children have been limited to case reports and small series [3–5, 7, 8, 11, 12, 15–31]. The aims of this study were hence twofold: (1) to present on the management of PFEDH through our institutional experience and (2) to corroborate our findings with the first meta-analysis of the published literature. Methods Data collection This is an ethics-approved, retrospective study conducted at the KK Women’s and Children’s Hospital (SingHealth CIRB Reference: 2020/2632) with data obtained from 2008 to 2024. Patients less than 18 years old, with a diagnosis of traumatic PFEDH confirmed on neuroimaging that required intervention by the Neurosurgical Service, KK Women’s and Children’s Hospital, were included. Patients aged above 18 years old, diagnosed with supratentorial EDH, non-traumatic posterior fossa bleeds, and those with incomplete medical information were excluded. In addition, patients with known bleeding disorders and/or on long-term anticoagulation medication were also excluded. Individual patient information was either obtained from electronic data or hardcopy notes using a standardized data collection form. Imaging details for each patient were obtained from the radiology archives and assessed for completeness. The following variables were collected: patient demographics, Glasgow Coma Scale (GCS) and symptoms on admission, details of injury, the Pediatric Emergency Care Applied Research Network (PECARN) rule [32], radiological features on CT brain scans, and perioperative details of the intervention of choice (Table 1). Table 1Summary characteristics, clinical and radiological features and outcomes of children with PFEDH at our institutionVariables of interestNumber of patients (%)DemographicsAge*, years5 (IQR 4–7)Male10 (52.6%)Mechanism of head injuryFall100 (100.0%)Road traffic accident0 (0.0%)Interval between trauma and admission≤ 24 h of injury11 (57.9%)> 24 to 48 h of injury3 (15.8%)> 48 to 72 h of injury5 (26.3%)Presenting Glasgow Coma Scale (GCS)GCS 13 to 1518 (94.7%)GCS 9 to 121 (5.3%)GCS 3 to 80 (0.0%)Symptoms reported upon admissionHeadache14 (73.7%)Nausea18 (94.7%)Vomiting18 (94.7%),Transient loss of consciousness0 (0.0%)Drowsiness4 (21.1%)Seizures0 (0.0%)Asymptomatic1 (5.3%)Hematoma dimensions on CT brain scanHematoma thickness > 15 ml5 (26.3%)Hematoma thickness 5 to 15 ml14 (73.7%)Hematoma thickness < 5 ml0 (0.0%)Hematoma volume (ml)8.5 (IQR 5.7–13.2)Other accompanying intracranial pathology on CT brain scanOccipital bone fracture18 (100.0%)Intraparenchymal contusion (ICH)6 (31.6%)Subdural hematoma (SDH)3 (15.8%)Subarachnoid hemorrhage (SAH)1 (5.3%)Supratentorial extension of primary hematoma2 (10.5%)Intraventricular hemorrhage (IVH)0 (0.0%)Pneumocephalus2 (10.5%)Mass effect and associated sequalae on CT brain scanCompression of fourth ventricle15 (78.9%)Obstructive hydrocephalus16 (84.2%)Obliteration of perimesencephalic cisterns14 (73.7%)OutcomesGood functional outcome GOS 4-519 (100.0%)Good functional outcome GOS-E Peds 7–819 (100.0%)Modified Rankins Score 0 (at 90 days post-surgery)19 (100.%)Death0 (0.0%)*Effect size reported as a median and interquartile rangeAbbreviations: CT = computed tomography, GCS = Glasgow Coma Scale, IVH = intraventricular hematoma, LOC = loss of consciousness, RTA = road traffic accident, SAH = subarachnoid hemorrhage, SDH = subdural hematoma Summary characteristics, clinical and radiological features and outcomes of children with PFEDH at our institution *Effect size reported as a median and interquartile range Abbreviations: CT = computed tomography, GCS = Glasgow Coma Scale, IVH = intraventricular hematoma, LOC = loss of consciousness, RTA = road traffic accident, SAH = subarachnoid hemorrhage, SDH = subdural hematoma Overview of neurosurgical workflow and approaches The decision for intervention is in accordance with a standardized workflow for PFEDH. Briefly, a correlation between the clinical condition of the patients and CT scan images is confirmed. The main radiological feature of concern is an expansile clot—that is, PFEDH volume > 10ml, > 15 mm in thickness, with a midline shift of > 5 mm. Associated findings include obliteration of peri-mesencephalic cisterns, fourth ventricular displacement and obstructive hydrocephalus (Fig. 1). Case example of a pediatric patient with PFEDH from our series. Representative pre-operative CT brain images in axial direction depicting: (A) a left-sided, biconvex hematoma with mixed density causing local mass effect on the underlying parenchyma; (B) the corresponding bone window of (A) showing an occipital bone fracture adjacent to the PFEDH; (C) another image slice that shows radiological evidence of obstructive hydrocephalus, fourth ventricular distortion and effacement of the peri-mesencephalic cisterns. (Abbreviations: PFEDH = posterior fossa extradural hematoma; CT = computed tomography.) The approach of choice is either a midline or paramedian suboccipital craniectomy or craniotomy [6]. For selected cases, the operating neurosurgeon may perform a single, large burrhole over the epicenter of the clot, especially if the CT scan demonstrates the clot to be more liquefied. Postoperatively, all patients are monitored overnight in either a high dependency or intensive care unit. We do not routinely perform postoperative CT scans if the patient has clinical improvement and adequate evacuation of the hematoma has been observed intraoperatively. Outcomes (case series) The primary outcome measure was functional outcome assessed using the Glasgow Outcome Scale (GOS), GOS-Extended (GOS-E Peds) Score and 90-day modified Rankin Scale (mRS). The secondary outcome was in-hospital mortality [33–35]. Statistical analysis Numerical variables were described as median (interquartile range [IQR]) for non-normal distributions. Due to the modest sample size of our case series, descriptive analysis rather than statistical analyses to investigate associations, were performed to avoid the risk of a Type 2 false negative error. Data were collated in Microsoft Excel (Microsoft, Redmond, WA, USA). All statistical analyses were performed using R software version 4.2.1 (R Foundation for Statistical Computing, 2022). P-values less than 0.05 were considered statistically significant. Systematic review and meta-analysis This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [36]. The protocol was registered on the PROSPERO international prospective register of systematic reviews (registration number CRD42024557754) [37, 38]. Outcomes (systematic review and meta analysis) The primary outcome was good functional outcome defined as GOS 4–5. In studies where other outcome measures were used, such as the modified Rankin scale (mRS) or GOS-E Peds Score, they were translated into good or poor outcomes as defined above. For example, mRS 0–2 or GOS-E Peds 7–8 were translated into good functional outcomes. The secondary outcome was in-hospital mortality. Search strategy Three electronic databases – Ovid Medline, Ovid Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) – were searched. Searches were performed in each electronic database from its inception until 4th June 2024. In addition to its synonyms and related terms, concepts of “pediatric”, “posterior fossa” and “extradural hemorrhage” were used. Supplementary Table 1 presents the full search strategy used for each database. Eligibility criteria Articles were selected for inclusion if they were either a primary interventional or observational study evaluating the management of PFEDH in an exclusively pediatric population (aged less than 18 years old). Additionally, we included mixed age population studies that provided analyses exclusive to the pediatric cohort. Supplementary Table 2 provides the full list of inclusion and exclusion criteria. Study selection Titles and abstracts were independently screened against the pre-defined eligibility criteria developed by two reviewers (KSL and CSG). Any disagreements were resolved through discussion between the reviewers or further adjudication by a third reviewer (SYYL). When encountered with multiple publications analyzing the same cohort over overlapping study periods, the publication that reported the largest patient data of relevant outcomes was used for evaluation. Data extraction To ensure standardization and consistency, a proforma was developed and piloted to extract data on the following variables: (1) study details, (2) study design, (3) country and dataset, (4) selection criteria, (5) patient demographics, (6) treatment/control, (7) indication for treatment, and (8) results. Risk of bias assessment The quality of included studies was assessed using the Joanna Briggs Institute (JBI) checklist for cohort studies and case series [39]. Data synthesis Meta-analyses were performed via the random effects model due to heterogeneity within and between individual studies as well as sampling variabilities across studies [39]. To obtain risk ratios (RRs) from reported binary outcomes, pairwise meta-analysis was conducted using the Mantel-Haenszel method without continuity correction, with the Paule-Mandel estimator. Overall pooled proportions of demographic comorbidities of included patients were computed with the generalized linear mixed model (GLMM) method using a random intercept logistic regression model via logit transformation [39]. Knapp-Hartung adjustments were used to reduce the chance of a false positive and to control the estimate uncertainties of between-study heterogeneity [39]. For pooling of means of numerical variables, we computed missing means and standard deviations (SDs) from medians, ranges (minimum to maximum) and interquartile ranges (IQRs) using the methods proposed by Hozo et al. and Wan et al. [40, 41]. The I2 statistic was employed to assess inter-study heterogeneity. I2 provides an estimate of the percentage of variability in results across studies that is due to real differences and not due to chance. I2 ≤ 30%, 30–50%, 50–75%, and ≥ 75% indicated low, moderate, substantial, and considerable heterogeneity, respectively. The quality of evidence for each outcome was evaluated using the GRADE framework [39, 42]. Results Overview of study population and patient characteristics A total of 157 pediatric head injury cases underwent neurosurgical intervention at our institution. Within this heterogeneous group, there were 19 (12.1%) cases of PFEDH (63.2% left side). There was no gender predilection (52.3% males) and the median age was 5 years (IQR 4–7 years). A fall from standing height was the most common mechanism of injury (57.9%). Twelve (63.2%) patients presented to hospital only after 24 h of injury. Most patients had full GCS scores (84.2%) at the time of admission. The most common symptoms presented were nausea and vomiting (94.7%), headache (73.7%) and drowsiness (21.1%). No patients experienced a lucid interval. The median volume of the hematoma was 8.5 ml (IQR 5.7–13.2 ml). Occipital bone fractures were observed in all 19 (100.0%) patients. Compression of the fourth ventricles, obliteration of the peri-mesencephalic cisterns and ventriculomegaly were observed in 15 (78.9%), 14 (73.7%), and 16 (84.2%) cases respectively. Five (26.3%) and 14 (73.7%) patients underwent a burrhole or craniotomy respectively. At the time of surgery, all patients (100.0%) were seen to have bleeding from the occipital fracture lines. No mortality was observed in our series. All 19 (100.0%) patients had good recovery after a median hospital stay of 4 days (IQR 3.5–5 days), with good functional outcome at discharge. There were no surgical complications. The length of follow-up ranged from 3 months to 8 years, and the median follow-up duration was 3 months (IQR 1.5–6.75 months). All patients reported good functional outcomes (GOS-E Peds8 and mRS0) on their last documented follow-up. Systematic review and meta-analysis Characteristics of included studies The systematic review search yielded 496 unique publications. Twenty-four studies including 372 pediatric patients met the eligibility criteria for inclusion in our meta-analysis (Fig. 2) [3–5, 7, 8, 11, 12, 15–31]. Including our institutional experience, a total of 391 pediatric patients were included. PRISMA flow diagram for studies included and excluded from the systematic review and meta-analysis All included studies were retrospective observational studies – five cohort studies and 19 case series. Table 2 presents a detailed summary of the characteristics and findings of the included studies. (Supplementary Tables 3 and 4 detail the risk of bias assessment.) Table 2Summary of included studiesFirst author and yearCountryStudy designStudy periodIndications for surgeryTotal patients included, nSurgical, nConservative, nImputed mean age at surgery (SD), year*Male, n (%)Good function outcomes GOS4-5, n (%)Death, n (%)Ammirati M et al. 1984USACase series1976 to 1981NR44NR5.62 (3.49)1 (25%)4 (100%)0 (0.0%)Bellotti C et al. 1987ItalyCase seriesNRLarge hematomas and patient is symptomatic54110.00 (6.40)4 (80%)5 (100%)0 (0.0%)Berker M et al. 2003TurkeyCase series1971 to 2000When the diagnosis is established, surgery should be performed as soon as possible even if the neurological examination is normal.1616NR7.74 (4.29)8 (50%)14 (87.5%)1 (6.25%)Bozbuga M et al. 1999TurkeyCase series1982 to 1992Obliteration of the perimesencephalic cisterns, compression and/or displacement of the fourth ventricle, and the presence of hydrocephalus99NRNR9 (100%)0 (0.0%)Brambilla G et al. 1986ItalyCase series1979 to 1985NR44NR10.25 (5.31)NR3 (75%)1 (25.0%)Chaoguo Y et al. 2019ChinaCohort study2001 to 2005Clot volume of ≥15 ml or presence of fourth ventricle compression or displacement and/or obstructive ventriculomegaly4817316.86 (4.55)24 (50%)46 (95.83%)0 (0.0%)Ciurea AV et al. 1993RomaniaCase series1987 to 1992Surgical treatment must be carried out immediately when the diagnosis has been established99NR7.31 (2.89)6 (66.67%)8 (88.89%)0 (0.0%)Costa Clara JM et al. 1996SpainCase series1989 to 1994Once the diagnosis of a PFEDH is made, surgery is mandatory33NR12.66 (5.77)1 (33.33%)3 (100%)0 (0.0%)Echara M et al. 2023IndiaCase series2016 to 2021PFEDH ≥15 ml, or the presence of mass effect (compression of the fourth ventricle and/or hydrocephalus)4040NR15.23 (3.96)32 (80%)35 (87.5%)1 (2.5%)Ersahin Y et al. 1993TurkeyCase series1979 to 1991Once the diagnosis of a PFEDH is made, the symptomatic hematomas are evacuated immediately99NR8.46 (4.33)5 (55.56%)8 (88.89%)1 (11.1%)Garza-Mercado R et al. 1983MexicoCase seriesNRNR66NR8.79 (6.14)4 (66.67%)5 (83.33%)0 (0.0%)Gupta PK et al. 2002OmanCase series1992 to 2000Hematoma > 10 ml in volume, > 15 mm in thickness, with a midline shift of > 5 mm, medullary compression and associated intracranial lesions1818NRNR14 (77.78%)15 (83.33%)0 (0.0%)Han K et al. 2018ChinaCase series2012 to 2015Hematoma > 15 ml in volume8534.63 (1.06)4 (50%)8 (100%)0 (0.0%)Jamous MA et al. 2021JordanCohort study2010 to 2021Large hematoma thickness (≥ 15 mm) or the presence of fourth ventricle compression or obstructive ventriculomegaly or a GCS of < 1416797.70 (6.00)12 (75%)14 (87.5%)1 (6.25%)Jang JW et al. 2010KoreaCohort study2001 to 2008Large hematomas with mass effects, as well as associated intracranial lesions199107.94 (3.90)10 (52.63%)18 (94.74%)0 (0.0%)Koç RK et al. 1998TurkeyCase series1983 to 1994NR99NR9.44 (4.87)5 (55.56%)7 (77.78%)2 (22.22%)Lui TN et al. 1993TaiwanCase series1977 to 1989Once the diagnosis of a PFEDH is made, all are evacuated immediately2929NRNRNRNR3 (10.34%)Miao Z et al. 2023ChinaCase series2016 to 2021Treatment with surgical evacuation depended on the clinical status of the patients as well as the presentation of CT image4040NR5.83 (1.08)24 (60%)32 (80%)NRMori K et al. 1983JapanCase seriesNRHematomas > 20 ml and patient is symptomatic33NR5.66 (1.15)1 (33.33%)3 (100%)0 (0.0%)Pang D et al. 1983USACase series1978 to 1982Clinical signs of local brain compression or herniation (such as increasing drowsiness, pupillary abnormality, hemiplegia, decerebrate posturing), or cardiorespiratory abnormalities3NR37.33 (5.85)NR3 (100%)0 (0.0%)Peter JC et al. 1990South AfricaCase series1979 to 1989NR55NR7.00 (1.58)2 (40%)5 (100%)NRPrasad G et al. 2015IndiaCohort study2008 to 2014Hematoma with a volume of > 20 ml or radiological evidence of posterior fossa mass effect in the form of fourth ventricle distortion or compression and/or obstructive ventriculomegaly, irrespective of the GCS score2216610.81 (4.64)17 (77.27%)20 (90.91%)0 (0.0%)Sencer A et al. 2012TurkeyCohort study1995 to 2011Hematoma thickness > 15 mm or hematoma thickness between 5–15 mm with injuries/ conditions (contusion, pneumocephalus, or SAH) causing additional mass effect or a GCS < 154029117.88 (4.67)22 (55%)40 (100%)0 (0.0%)Sheng HS et al. 2017ChinaCase series2010 to 2015Criteria for standard craniectomy: hematoma volume > 30 ml or presence of severe mass effects like cerebellar tonsillar herniationCriteria for trephination mini-craniectomy: hematoma volume 10-30 ml, with mild mass effects (e.g. fourth ventricle compression or displacement and/or obstructive ventriculomegaly) and stable neurological conditions77NR5.70 (2.32)3 (42.86%)7 (100%)0 (0.0%)Abbreviations: NR = Not reported, UK = United Kingdom, USA = United States Summary of included studies Criteria for standard craniectomy: hematoma volume > 30 ml or presence of severe mass effects like cerebellar tonsillar herniation Criteria for trephination mini-craniectomy: hematoma volume 10-30 ml, with mild mass effects (e.g. fourth ventricle compression or displacement and/or obstructive ventriculomegaly) and stable neurological conditions Abbreviations: NR = Not reported, UK = United Kingdom, USA = United States Patient baseline characteristics, treatment approach and radiological features Pooled prevalence of baseline characteristics, stratified according to treatment arm, is summarized in Table 3. Of the 391 patients, 308 were treated with surgery, whereas 83 patients were treated conservatively. Table 3Summary of pooled baseline demographics, etiologies, clinical findings of included patients, CT brain findings and associated pathology between the two groups (surgical versus conservative)FactorSurgical cohort (n = 308)Conservative cohort (n = 83)Number of studies reporting this variableNumber of patients/ analysedPooled proportion [95% CI]I2 (%)p-value of I2 (from χ2 test)Quality of Evidence (GRADE)Number of studies reporting this variableNumber of patients/ analysedPooled proportion [95% CI]I2 (%)p-value of I2 (from χ2 test)Quality of Evidence (GRADE)Demographics Age*161837.78 [6.15; 9.41]94.8< 0.001Low3166.61 [1.98; 11.24]53.40.117Low Gender1720461.51 [52.61; 70.11]0.00.518Low31954.13 [22.29; 84.56]33.70.221LowMechanism of injury Fall1721575.44 [61.51; 87.48]75.0< 0.001Low44992.70 [81.25; 99.61]0.00.898Low RTA1418231.73 [19.79; 44.74]60.90.002Low4497.30 [0.39; 18.75]0.00.898LowTiming Within 24 h of injury712579.66 [69.68; 88.31]24.30.244Low42672.23 [50.65; 90.30]11.50.335Low 24–48 h of injury46028.69 [0.00; 78.98]85.10.001Low22019.48 [3.83; 41.05]0.00.420Low 48–72 h of injury35512.01 [0.00; 43.09]84.00.002Low2209.94 [0.01; 28.81]0.00.870LowPreoperative GCS GCS13-151419665.84 [50.76; 79.64]69.9< 0.001Low670100.00 [98.38; 100.00]0.00.987Low GCS3-8813217.97 [8.61; 29.31]49.10.056Low4560.00 [0.00; 2.67]0.00.946LowSymptoms Headache2026663.55 [46.96; 78.89]84.5< 0.001Low77357.02 [34.27; 78.60]58.60.025Low Vomiting2026677.69 [67.76; 86.51]54.70.002Low77346.59 [29.04; 64.50]33.20.174Low Transient LOC1218536.85 [19.59; 55.73]78.2< 0.001Low56012.70 [0.00; 42.51]78.60.001Low Drowsiness1213129.49 [16.72; 43.71]51.90.018Low5587.24 [0.00; 35.66]77.80.001Low Seizures71117.95 [0.89; 18.91]50.30.060Low298.25 [0.00; 40.90]0.00.519Low Asymptomatic61285.26 [0.00; 16.38]73.90.002Low45722.04 [11.19; 34.79]0.00.979LowSigns Cerebellar signs77439.88 [19.87; 61.42]55.30.037Low21419.42 [0.83; 47.88]0.00.550Low Babinski’s sign42240.64 [18.57; 64.37]0.00.909LowNANANANANANA Papilloedema42524.81 [2.42; 55.81]46.10.135LowNANANANANANA Aniscoria3747.75 [2.14; 15.65]0.00.861LowNANANANANANA Abducens palsy53124.27 [7.19; 45.54]10.50.346LowNANANANANANA Occipital swelling56044.22 [23.39; 65.97]46.90.110LowNANANANANANAAssociated pathology Occipital bone fracture2025776.30 [64.20; 86.89]72.5< 0.001Low56766.88 [52.61; 79.93]0.00.474Low Contusion1020117.64 [10.62; 25.72]34.60.131Low5673.80 [0.01; 11.39]0.00.723Low SDH81766.85 [2.72; 12.20]5.00.391Low3268.80 [0.00; 31.51]41.80.180Low SAH51149.01 [3.85; 15.61]0.00.905Low3524.49 [0.01; 13.26]0.00.626Low IVH3662.51 [0.00; 9.18]19.40.289LowNANANANANANA Pneumocephalus61189.41 [3.86; 16.42]1.30.408Low2213.18 [0.00; 17.75]0.00.340LowCT findings Hematoma thickness > 15 ml68552.84 [32.12; 73.14]65.10.014Low4273.47 [0.00; 32.19]57.80.069Low Hematoma thickness 5-15 ml68512.95 [0.00; 51.01]88.3< 0.001Low42719.33 [0.00; 65.78]81.50.001Low Hematoma thickness < 5 ml68556.94 [36.54; 76.33]64.50.015Low32449.83 [0.00; 100.00]91.0< 0.001Low Mean hematoma volume (ml)*1016923.74 [16.19; 31.30]93.5< 0.001Low66210.46 [6.16; 14.76]87.9< 0.001LowMass effect Compression of fourth ventricle1112249.95 [34.77; 65.13]52.10.022Low3540.00 [0.00; 1.60]0.00.900Low Ventriculomegaly1011636.10 [15.57; 59.06]79.7< 0.001Low4540.00 [0.00; 2.93]25.70.257Low Obliteration of persimencephalic cisterns911043.42 [21.53; 66.48]77.0< 0.001Low4540.00 [0.00; 2.93]25.70.257Low*Effect size is reported as a mean. Abbreviations: CT = computed tomography, GCS = Glasgow Coma Scale, IVH = intraventricular hematoma; LOC = loss of consciousness, RTA = road traffic accident, SAH = subarachnoid hemorrhage, SDH = subdural hematoma Summary of pooled baseline demographics, etiologies, clinical findings of included patients, CT brain findings and associated pathology between the two groups (surgical versus conservative) *Effect size is reported as a mean. Abbreviations: CT = computed tomography, GCS = Glasgow Coma Scale, IVH = intraventricular hematoma; LOC = loss of consciousness, RTA = road traffic accident, SAH = subarachnoid hemorrhage, SDH = subdural hematoma The pooled proportion of males were 61.51% (95%CI: 52.61; 70.11%, I2 = 0.0% [p = 0.518]) % and 54.13% (95%CI: 22.29; 84.56%, I2 = 33.7% [p = 0.221]) in the surgical and conservative groups, respectively. Overall pooled mean age across the surgical and conservative groups were 7.78 years (95%CI: 6.15;9.41 years, I2 = 94.8%, p < 0.001) and 6.61 years (95%CI: 1.98;11.24 years, I2 = 53.4%, p = 0.117), respectively. The most common symptoms encountered in the surgical group were headache (63.55% (95%CI: 46.96;78.89%, I2 = 84.5% [p < 0.001])) and vomiting (77.69% (95%CI: 67.76;86.51%, I2 = 54.7% [p = 0.002])). The most common symptoms encountered in the conservatively managed group were headache (57.02% (95%CI: 34.27;78.60%, I2 = 58.6% [p = 0.025])) and vomiting (46.59% (95%CI: 29.04;64.50%, I2 = 33.2% [p = 0.174])). Symptoms of transient loss of consciousness (LOC) (36.85% (95%CI: 19.59; 55.73%, I2 = 78.2% [p < 0.001]) versus 12.70% (95%CI: 0.00; 42.51%, I2 = 78.6% [p = 0.001])), drowsiness (29.49% (95%CI: 16.72; 43.71%, I2 = 51.9% [p = 0.018]) versus 7.24% (95%CI: 0.00;35.66%, I2 = 77.8% [p = 0.001])), and seizures (7.95% (95%CI: 0.89;18.91%, I2 = 50.3% [p = 0.060]) versus 8.25% (95%CI: 0.00;40.90%, I2 = 0.0% [p = 0.519])) were more common in the surgical group than the conservative group. The pooled proportion of mass effective was greater in the surgical group than the conservative group – compression of the fourth ventricle (49.95% (95%CI: 34.77;65.13%, I2 = 52.1% [p = 0.022]) versus 8.25% (95%CI: 0.00;40.90%, I2 = 0.0% [p = 0.519])), ventriculomegaly (36.10% (95%CI: 15.57;59.06%, I2 = 79.7% [p < 0.001]) versus 8.25% (95%CI: 0.00;40.90%, I2 = 0.0% [p = 0.519])), obliteration of the perimesencephalic cisterns (43.42% (95%CI: 21.53;66.48%, I2 = 77.0% [p < 0.001]) versus 8.25% (95%CI: 0.00;40.90%, I2 = 0.0% [p = 0.519])). Following that, the pooled proportion of associated pathology was greater in the surgical group – occipital bone fracture (76.30% (95%CI: 64.20;86.89%, I2 = 72.5% [p < 0.001]) versus 66.88% (95%CI: 52.61;79.93%, I2 = 0.0% [p = 0.474])), contusion (17.64% (95%CI: 10.62;25.72%, I2 = 34.6% [p = 0.131]) versus 3.80% (95%CI: 0.01;11.39%, I2 = 0.0% [p = 0.723])) and pneumocephalus (9.41% (95%CI: 3.86;16.42%, I2 = 1.3% [p = 0.408]) versus 3.18% (95%CI: 0.00;17.75%, I2 = 0.0% [p = 0.340])). The mean hematoma volume was greater in the surgical group (23.74 ml (95%CI: 16.19;31.30 ml, I2 = 93.5%, p < 0.001)) compared with the conservatively managed group (10.46 ml (95%CI: 6.16;14.76 ml, I2 = 87.9%, p < 0.001)) (Table 3). Conversion to surgery when initially managed conservatively Conversion to surgery from conservative management was reported in five studies across 77 patients initially managed conservatively. Reasons cited were enlargement of the hematoma and/or clinical deterioration. The overall pooled incidence of conversion was 9.90% (95%CI: 1.61;22.21%, I2 = 35.2%, p = 0.187). Rates of good functional outcomes and mortality A comparative meta-analysis of outcomes was not performed as the two groups were deemed too heterogeneous in terms of baseline clinical characteristics. Instead, single-arm meta-analyses were performed (Table 4). The overall pooled proportion of good functional outcomes across 279 patients between these studies, and our series, managed surgically was 93.68% (95%CI: 88.69;97.57%, I2 = 0.0% [p = 0.46]) (Fig. 3). The incidence of good functional outcomes was reported in eight studies assessing 82 patients managed conservatively. Overall pooled incidence was 99.99% (95%CI: 96.53,100%, I2 = 0.0% [p = 0.62]) (Supplementary Fig. 1). The overall pooled incidence of mortality across 263 patients between these studies, and our series, managed surgically was 0.57% (95%CI: 0.00;2.87%, I2 = 0.0% [p = 0.71]). The incidence of mortality was reported in eight studies assessing 82 patients managed conservatively. Overall pooled incidence was 0.00% (95%CI: 0.00,1.18%, I2 = 0.0% [p = 0.99]). Table 4Pooled outcomes of included patients between the two groups (surgical versus conservative)OutcomesSurgical (n = 308)Conservative (n = 83)No. of studies reporting variableNo. of patients/ analyzedPooled proportion [95% confidence interval]I2 (%)P value of I2 (from χ2 test)Quality of Evidence (GRADE)No. of studies reporting variableNo. of patients/ analyzedPooled proportion [95% confidence interval]I2 (%)P value of I2 (from χ2 test)Quality of Evidence (GRADE)Good functional outcome2227993.68 [88.69; 97.57]0.00.461Low88299.99 [96.53; 100.00]0.00.624LowDeath212630.57 [0.00; 2.87]0.00.706Low8820.00 [0.00; 1.18]0.00.997Low Pooled outcomes of included patients between the two groups (surgical versus conservative) Forest plots, with random-effects model, of pooled percentage of (A) good functional outcome (GOS 4–5), and (B), in-hospital mortality, in patients PFEDH who underwent surgery, respectively Discussion Overview of PFEDH in the pediatric population Amidst the spectrum of head injuries treated by neurosurgeons, the entity of PFEDHs is relatively uncommon [11, 14, 15, 17]. Nonetheless, a late diagnosis is associated with a high mortality rate [14]. Despite published studies reporting a higher incidence of PFEDH in children in comparison to adults, children tend to fare better with prompt intervention [8, 9, 30]. PFEDHs are usually associated with occipital bone fractures. Here, the anatomical disadvantage is that children tend to have larger occipital sinuses; hence, torrential venous bleeding may be encountered in fractures crossing the midline with tearing of adjacent venous sinuses. Under such circumstances, bleeding leads to progressive stripping of the dura from the inner table of the skull [43]. This expanding PFEDH could be rapidly fatal due to the compromise of a smaller posterior cranial fossa, leading to brainstem compression, tonsillar herniation and obstructive hydrocephalus [3, 8, 44]. Specifically for PFEDH, one of the main challenges is that affected patients are likely to present in an insidious fashion and may not manifest the classic textbook description of an initial loss of consciousness followed by a lucid interval, subsequently developing an ipsilateral pupillary dilation and progressive obtundation [10, 45]. Clinical assessment in young children is often difficult due to their variable, non-specific complaints, especially in the preverbal age group [46–48]. Owing to their smaller posterior fossa, precipitous drops in the neurological status may occur without forewarning [6, 43]. Congruently, we observed that most patients had full GCS and no focal neurological deficit in our cohort, at the time of admission. Following that, headache, and or vomiting are the most common presenting complaints in our study, aligning with other publications [3, 5, 11, 15, 23]. Therefore, a diagnosis of PFEDH should be considered if there is a history of headache, vomiting and trauma to the occipital region, even for relatively stable GCS on admission. In corroboration with existing studies, we practiced a low threshold for neurosurgical intervention. This is to mitigate the risk of sudden and rapid clinical deterioration, especially for symptomatic patients with existing radiological evidence of mass effect [7, 17]. Imaging characteristics in PFEDH The current gold standard for neuroimaging in TBI is a CT brain to reliably exclude any life-threatening intracranial hemorrhage [11, 49]. The impact of an early CT upon the treatment of PFEDH is easily recognized—that is, diagnosis can be established quickly to initiate intervention; hence, affording better prognosis [5]. As mentioned previously, occipital linear fractures are common in up to 80% of cases of PFEDH. Occasionally, the hematoma may be undetected on the initial CT scan if the scan is performed too early because PFEDHs tend to appear less dense [13]. This is due to the fact that the bleed tends to be venous and liquefies earlier than supratentorial hematomas [50]. Extension of the PFEDH to the supratentorial compartment is another common radiological finding [4]. In our cohort, we noted two (10.5%) had such extensions, comparable to other reports [4, 23, 51]. The presence of obstructive hydrocephalus is a poor prognosticator [3, 8, 14, 15, 23, 26, 52–54]. Presence of other accompanying lesions such as contusions, subdural hematomas, subarachnoid hemorrhage, intraventricular bleeds and pneumocephalus have been reported [3–5, 7, 10, 11, 23]. In our study, 16 patients (84.2%) had radiological features of early obstructive hydrocephalus and six (31.6%) had other accompanying bleeds and/or pneumocephalus. Nonetheless, we did not observe a poorer outcome for the subgroup of patients. Towards an evidence-based management strategy As reflected in our population, time lapse from injury may be delayed up to a few days from initial injury [17]. Early diagnosis relies on astute clinical observations and failure to recognize the more subtle clinical presentation in children is a genuine concern [10]. To mitigate this, some authors have suggested serial control scans at various time-points post-TBI, for cases where the clinical history is suspicious of PFEDH even if no bleed is obvious on earlier imaging [5, 11]. We are cognizant that most clinicians are mindful of exposing young patients to excessive ionizing radiation in view of the risks of future malignancies [55–58]. Other pertinent challenges associated with CT scans include the possible need for sedation during the scan and parental preferences [58]. Levelling on these opposing points, we advocate that an early CT scan is a reasonable option for selected patients versus a more conservative approach, considering the potential for rapid deterioration. Interestingly, we also observe the limitations of the PECARN rule in our series. This is a prediction score used to assess the need for CT scans children with a history of minor head trauma that has been validated in large, multi-centred studies – the PECARN rule provides the highest level of internationally valid guidance available [59, 60]. However, here, only four patients (21%) of our children underwent initial CT scan based on PECARN predictions. We hypothesize that the role of PECARN may have a limited negative predictive value in this specific cohort of patients with PFEDH, due to the initial subtle nature of PFEDH presentation. This limited negative predictive value may be attributed to the fact that the PECARN rule was derived from a wider population that included a spectrum of traumatic head injuries of varying severity [59]. With the limited utility of the PECARN in PFEDH, the infrequent occurrence of PFEDH and its extended history of evolution from comparatively low-impact fall mechanisms, affected children may initially be cleared with false reassurance – leading to false negative cases with dire consequences. A revised sub-criteria to recognize their diagnosis early may be more suitable for them. However, this requires input from international workgroups and external validation in future studies. The traditional treatment of PFEDH is surgical evacuation. Balancing the of risk of a rapid deterioration with conservative management versus performing a relatively safe prophylactic surgery, surgery has conventionally been recommended even if the child is initially asymptomatic [4, 43]. However, no surgery is risk-free and torrential venous bleeding may be encountered intraoperatively especially with associated fractures crossing the midline with tearing of adjacent venous sinuses. To date, there is no clear, defined criteria for deciding between conservative and surgical treatment [5, 13, 17]. We observe the following CT brain characteristics are often used: hematoma > 10 ml in volume or > 15 mm in thickness, with a midline shift of > 5 mm, or mass effect of peri-mesencephalic cisterns obliteration, fourth ventricle displacement, and or presence of hydrocephalus [5, 14, 21, 23, 52, 53, 61]. Generally, most studies emphasize that enlarging hematomas and neurological deterioration are urgent indications for surgery due to the risk of rapid compromise in the posterior fossa. In our meta-analysis, a fifth of patients were managed conservatively with close neuro-monitoring and serial imaging, whilst the rest were treated surgically. The prognosis of PFEDH is reported to be good [3–5, 7, 8, 11, 12, 15–31]. Regardless of either approach, studies generally report good prognosis in pediatric PFEDH [3–5, 7, 10, 11, 15, 19, 21, 23]. In addition, the rate of good postoperative outcomes imply that surgery does not further morbidity or mortality to the existing neurological state. For the purposes of our study, we did not perform a comparative meta-analysis between the conservative and surgical cohorts as the clinical characteristics were too heterogeneous between both groups. In addition, an established criteria for either of the treatments remain to be elucidated, which we have summarized in Table 2 [3–5, 7, 8, 11, 12, 15–31]. Future research could address this current paucity in evidence with direct head-to-head comparisons. Limitations and future directions Owing to its rare entity, the pediatric cases of traumatic PFEDH published in the literature are limited to case reports and small series. Henceforth, the limitations of our meta-analysis stem from inclusion of small retrospective observational studies and the heterogeneity across them [62–64]. Restricted by the modest number of suitable studies and small number of events, our ability to delineate risk factors for poor outcomes was also limited, largely due to the risk of Type 2 false negative errors. Nonetheless, congruent findings from our study and meta-analysis do demonstrate the effectiveness of neurosurgical intervention in children with PFEDH, as long as they are interpreted judiciously with the aforementioned limitations. Separately, we want to highlight that patients in this age group tend to live longer, and significant brain trauma leads to long-term consequences that may permanently compromise their motor development and cognitive abilities [2]. The acquired injury in a developing brain can disrupt subsequent its development, with delayed consequences emerging over time, causing significant short- and long-term alterations in a wide range of functional abilities [65–67]. Longitudinal studies have reported that most children experience persistent disability at one year after severe TBI, especially in the cognitive, behavioural, emotional and social domains [67, 68]. Building on this knowledge, we recommend future studies to consider the long-term quality of life and the major factors limiting the resumption of daily activities in PFEDH patients [14]. Conclusion Overall, our study reiterates that pediatric PFEDH is uncommon, and patients often present atypically. Based on our institutional experience and extrapolating data from our meta-analysis of the wider literature, neurosurgical intervention is a reliable therapeutic option with good clinical outcomes. Emphasis is on early recognition of their subtle symptoms by mindful clinicians to initiate early neuroimaging. Future multi-centre collaborations may be warranted to propose a standardized criterion for conservative management of PFEDH. As the way forward, we advocate continued efforts to educate the public on pediatric head injury prevention. Supplementary Information Below is the link to the electronic supplementary material. (DOCX 43.1 KB) Acknowledgements NIL. Author contributions All authors listed have made substantial, direct, and intellectual contribution to the work and approved it for publications. Keng Siang Lee: Conceptualization, Methodology, Formal analysis and investigation, Writing - original draft preparation, Writing - review and editing, Visualization, Funding acquisition. Shi Hui Ong: Data curation. Conor S Gillespie: Formal analysis and investigation. Lee Ping Ng: Data curation. Wan Tew Seow: Writing - review and editing, Supervision. Sharon YY Low: Conceptualization, Data curation, Writing - review and editing. Funding Open Access funding provided to Dr Keng Siang Lee by King's College London. Data availability No datasets were generated or analysed during the current study. Declarations Ethical approval Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials (DOCX 43.1 KB) Data Availability Statement No datasets were generated or analysed during the current study.