To retrieve the relevant cases from the database of Judicial Yuan, we conducted a comprehensive keyword search from 2008 to 2022 (Fig. 1). The initial search yielded 62,736 cases related to domestic violence. Further filtering was conducted by incorporating the keyword “child abuse,” which resulted in 911 cases. Subsequent exclusions were made based on case types, which led to the omission of 539 civil and administrative cases. The remaining 372 cases were then filtered by cases heard by district courts, which produced a subset of 192 cases. Additional exclusion criteria were applied, thus excluding cases including civil protection orders (72), forcible abduction [1], larceny [1], forging instruments or seals [1], offenses against personal freedom [3], and offenses against marriage and family [1]. Cases unrelated to children’s welfare [11] were also excluded. The refinement process led to a final selection of 101 cases. Finally, cases involving hired caregivers [5] and sexual abuse (66) were excluded. Ultimately, a total of 30 cases were included in the analysis.

Flow chart of the study

From 2008 to 2022 in Taiwan, initial search identified 62,736 domestic violence cases, which expanded to 911 upon inclusion of “child abuse” keyword. Following exclusions, 372 cases remained, of which 192 were from district courts. Further refinement removed cases with civil protection orders (72), abduction, (1) larceny, (1) forging, (1) personal freedom offenses, (3) and marriage/family offenses, (1) along with 12 unrelated to children’s welfare. This resulted in a final selection of 101 cases. Subsequent exclusions of hired caregiver (5) and sexual abuse (66) cases yielded a total of 30 for analysis

Table 1 presents the results of analysis of demographic factors in relation to child abuse after a specific examination of the types of maltreatment, perpetrators, age range of victims, and prognoses. Among the various types of maltreatment, physical abuse (50%) emerged as the most prevalent form followed by physical and emotional abuse (30%). In terms of perpetrators, the study identified the biological father as the primary offender (30%) followed by the biological mother (27.6%). The majority of victims fell within the age range of 1 to 5 years (56.7%). Furthermore, the prognosis for these victims was predominantly severe with outcomes ranging from death (56.7%) to a severe disability or a vegetative state (26.7%). This result highlights their vulnerability and emphasizes the critical nature of child abuse, which demands increased attention.

Table 2 presents an in-depth analysis of injury patterns and related conditions observed in the 30 judgments. The primary pattern identified as the most relevant was that of TEN-4-FACESp (83.3%), which pertains to bruises in the trunk (T), ears (E), and neck (N) in children under the age of 4 years or any bruise in children less than 4 months old [4], facial injuries, including the frenulum (F), angle of jaw (A), cheeks (C), eyelids (E), subconjunctivae (S), and patterned brushing (P) [7]. The second most prevalent finding (80%) was the coexistence of multiple abuse patterns alongside the primary one. This encompassed a spectrum of abusive injuries coupled with indicators of neglect, including malnutrition, failure to thrive, poor hygiene, inadequate clothing, and untreated medical conditions. Additionally, emotional abuse manifestations, such as withdrawal, anxiety, or developmental delays, were often observed concurrently. This finding suggested an association between physical abuse and neglectful behaviors, which potentially exacerbates the overall harm inflicted on victims. Intracranial injuries (63.3%), such as subdural hemorrhage (SDH), subarachnoid hemorrhage (SAH), or intracranial hemorrhage, rank as the third most frequent injury pattern. Another prominent pattern involves fractures (53.3%), which highlight the vulnerability of this age group to physical abuse. These fractures can occur in various body parts and are significant indicators of abuse. AHT (36.7%) accounts for more than half of intracranial injuries, which underscores its prevalence in child abuse cases. However, factors, such as cigarette burns (4%) and oral injuries (4%), contribute to a smaller percentage of the overall results.

A comparative analysis of expert decisions revealed variations in assessments and opinions among experts (Table 3). Out the 27 guilty judgments, clinical physicians reported 18 as child abuse cases (67%), while their opinions were not mentioned for the remaining 9 (33%). Strikingly, for the three cases with a judgment of not guilty, clinical physicians deemed that all of them were child abuse cases (100%). In the case of court-appointed physicians, they considered 13 out of the 27 judgments of guilty as child abuse cases (48%), while their opinions were absent for the other 14 (52%). Regarding cases with a judgment of not guilty, court-appointed physicians perceived them as cases of child abuse (100%). Finally, forensic medical examiners confirmed 15 of the guilty judgments as guilty (56%), while one guilty judgment was reversed to ‘not guilty’ (4%), and 11 judgments were not assessed by them (41%). Among the three cases with a judgment of not guilty, forensic medical examiners considered one as not guilty (33%), while their opinions were not mentioned for the two other cases (67%).

Table 4 delineates four cases exhibiting significant inter-expert discrepancies. It encompasses three acquitted cases, elucidating the rationale behind each acquittal, and one conviction where forensic medical examiners diverged from the guilty verdict. The first two cases involve victims that exhibited signs of AHT. However, the court declared both cases as not guilty because the evidence to conclusively identify a perpetrator, including any potential witnesses, was insufficient. Furthermore, due to the uncertain timeline implied by the available medical evidence, a judgment of not guilty was reached, even in the presence of a definitive AHT diagnosis.

In the third case, a one-year-old child who experienced hypoxic brain injury briefly after birth presented with evidence of multiple fractures involving metaphyseal and old and new SDH. Although the clinical and court-appointed physicians concluded that the injuries were consistent with child abuse, the forensic medical examiners determined a non-abusive cause due to the child’s prolonged bedridden conditions, which rendered the child susceptible to fracture and bleeding, determining the case to be non-abusive. The forensic medical examiner inferred that the metaphyseal was the result of an accidental injury during the rehabilitation of the patient. Furthermore, the presence of old and new SDH could be associated with the abovementioned susceptibility to bleeding.

In light of the prior history of the child of choking and the resultant hypoxic ischemic brain injury, the clinical and court-appointed physicians harbored significant suspicions regarding the possibility of a high-risk child abuse scenario. Upon revisiting the history of the case, retinal hemorrhage was initially diagnosed. Nevertheless, the analysis of forensic medical examiners suggested that the retinal hemorrhage may have been linked to prolonged hypoxia events and associated acidosis, which led to an increased tendency of bleeding.

The fourth case involved an 11-month-old female infant, ultimately judged guilty of AHT inflicted by her biological mother. The victim succumbed to neurogenic shock and recurrent sepsis two weeks post-admission to the intensive care unit. Notably, the forensic medical examiners faced challenges in identifying AHT evidence after this prolonged interval, leading them to conclude a non-abusive etiology.

These examples underscore the inherent complexity and challenge associated with the identification and prosecution of child abuse cases, particularly in instances involving AHT. It also reveals discrepancies among the opinions of various experts. The forensic medical examiners’ hypothesis—that spontaneous SDH can occur in bedridden children and that RH may result from hypoxia—appears inconsistent with current evidence [8].