Introduction

The World Health Organization defines child maltreatment as abuse and neglect that is directed at children under 18 years old. Child maltreatment constitutes all forms of physical abuse, sexual abuse, emotional abuse, and neglect that results in actual or potential harm to a child's health or survival (World Health Organization, 2020). In cases of child abuse, only children with injuries or in a life-threatening situation are referred for medical treatment. However, child abuse does not necessarily present consistent symptoms and signs. Most abusers or caregivers tend to deny or refuse to provide a child's medical history, which overshadows the crucial point and misleads judgments on the presence of child abuse, making related diagnoses and treatments more difficult. Without immediate identification and intervention, the risk of repeated maltreatment in children experiencing abuse increases, leading to physical and mental trauma that may be life-threatening (Oral et al., 2008).

Healthcare providers such as community and hospital medical staffs often encounter maltreated children in their professional settings. The characteristics of abuse differ across cases. In some cases, children who have experienced abuse exhibit only mild or insignificant symptoms such as bruises (Mimasaka et al., 2010). Healthcare professionals often use medical histories, exhibited symptoms, and observations of interactions between the child and his or her caregivers to determine the presence of abuse.

Existing screening tools for child abuse may be used by a wide range of professionals in various settings. Some tools have been designed based on parents' or children's self-reported abuse (Saini et al., 2019), whereas others rely on the objective results of imaging examinations (Flom et al., 2016) or on physical and medical history analyses (Berger et al., 2016). Furthermore, community assessments of child abuse largely rely on parental statements, parental behavior, caring experiences, and direct observations to assess the state of a child and the home environment (van der Put et al., 2017). An efficient screening tool may assist healthcare providers to effectively identify potential cases of child abuse. Hoytema van Konijnenburg et al. (2013) reviewed the related literature and explored the use of physical examinations to screen child abuse in hospitals and communities. Their findings showed that 0.8%–13.5% of children are screened for child abuse. However, that study did not conduct a sensitivity and specificity analysis. Moreover, physical examination alone is insufficient for screening child abuse cases. Considering the difficulty in identifying child abuse, healthcare providers must pay greater attention to details and be more sensitive in detecting cases of potential or actual abuse to systematically and efficiently screen suspected cases in busy clinical settings. The use of screening tools that offer high sensitivity and cover common injuries and features of child abuse has been shown to increase the rate of detection of child abuse from less than 3% to 34% (Louwers et al., 2012).

An efficient screening tool that covers both risk factors and identification elements of child abuse will enable healthcare workers to identify child abuse effectively, while reducing the burden of judgment and evaluation and lowering the chance of overlooking cases. Furthermore, the early detection of child abuse will allow the provision of appropriate assistance, prevent further abuse, and reduce long-term negative effects (Salinas-Miranda et al., 2015). A recent systematic literature review by Saini et al. (2019) analyzed 52 screening tools for child abuse. This review included only instruments that measured any form of child abuse in articles published in English. The review found that most of the existing screening tools use self-reported and retrospective questionnaires and that they mainly explore child abuse cases occurring before 18 years old. The significance of this review is that it examined the quality of the screening tools using the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) checklist, wherein only eight of the screening tools were found to have a moderate or high quality of evidence. However, this systematic literature review is mainly applicable to assessing the abuse experience of victims and is less appropriate for infants and young children who are unable to express themselves or for assessments conducted in time-sensitive emergencies. Moreover, it did not include screening tools designed for use in healthcare settings. Hoft and Haddad (2017) also conducted a systematic literature review of existing screening tools and guidelines for identifying child abuse. The study analyzed nine screening tools, which included a questionnaire completed by parents on the risk of child abuse and sexual exploitation, a risk assessment questionnaire completed by the medical staff on physical and sexual abuse screening tools, and a questionnaire on the scale of potential child neglect evaluated by preschool professionals. Although the review by Hoft and Haddad examined different types of screening tools, it did not evaluate or compare the quality of each tool. Therefore, it was not possible to identify the effectiveness of each analyzed tool.

Healthcare covers a wide range of services from the management of acute medical units to health promotion and administration in communities. Thus, the suitability of a screening tool for child abuse varies depending on the context in which it is used. Healthcare workers—even those in acute care hospitals—face multiple difficulties when using screening tools for child abuse. For example, emergency rooms (ERs), pediatric intensive care units, outpatient departments, and community and homecare services may, respectively, use different screening tools based on their unique contexts. Rumball-Smith et al. (2018) used the Escape tool to construct a screening tool to detect child abuse in children under 13 years old using electronic medical records in the ERs of 13 hospitals. Their results showed that the reporting rate was considerably higher when the screening tool was used (1.3% vs. 0.4%; odds ratio [OR] = 2.90, 95% confidence interval [CI; 1.67, 5.02]) and that the reporting rate for positive cases was significantly higher than that for negative cases (50% vs. 0.3%; p < .0001).

Rigorous screening tools appropriate for use in the healthcare environment may assist healthcare providers to detect child abuse early as well as reduce their workload, improve work efficiency, and increase confidence and job satisfaction (Carson, 2018). To this end, this study was designed to identify the current screening tools used by healthcare providers to detect child abuse, identify the assessment content used in these tools, and evaluate the reliability and quality of these tools using a systematic literature review. The findings will be used to propose the most appropriate, reliable, and validated screening tools for child abuse that may be used in various segments of the healthcare industry. The research questions of this study were as follows: (a) What screening tools are used by healthcare providers to detect child abuse? (b) How should screening tools for detecting child abuse be evaluated? and (c) What are the psychometric properties of child abuse screening tools?

Methods

This systematic review was constructed based on Preferred Reporting Items for Systematic Review and Meta-Analysis (Moher et al., 2009). In addition, the COSMIN checklist (Prinsen et al., 2018) was used to conduct the literature review, and the Grading of Recommendation, Assessment, Development, and Evaluation (GRADE; Schünemann et al., 2017) was adopted to grade the quality of evidence to evaluate the measurement properties and formulate results and recommendations.

Search Strategy

This systematic review included an extensive search of relevant domestic and international publication databases, including Airiti Library, PubMed, MEDLINE, CINAHL, Education Resources Information Center, Cochrane Library, Embase, and OpenGray. An extensive literature search was conducted for all full-text articles published before October 2019 to canvass the most comprehensive range possible. The Boolean logic operator “OR” was used for joint sets of synonyms and “AND” for keyword conjugations. These operators were used in combination and separately during the keyword search. The search keywords included the following: (“child*abuse [MeSH],” “child*maltreatment,” “child neglect,” or “abusive head trauma”) and (“instrument,” “screening,” “measurement,” “scale,” or “questionnaire”) and (“health care”).

Inclusion Criteria

The screening tools identified in this systematic review were required to meet the following three inclusion criteria: (a) have as their main objective the evaluation of victims who had been abused either physically (including abusive head trauma), sexually, or emotionally or had been subjected to neglect; (b) be applicable to children less than 18 years old; and (c) be designed for use by healthcare professionals such as medical staff in ERs, pediatric wards, and community healthcare units. Furthermore, the reviewed articles were required to have been published in either Chinese or English.

Exclusion Criteria

Review articles, commentaries, editorials, and expert opinion articles were excluded from consideration.

Search Outcome

Publications were first screened independently by two researchers based on their titles and abstracts, and duplicate publications were deleted. Publications that used screening tools for the objective evaluation of children younger than 18 years old and those used by healthcare professionals at hospitals and communities were selected. Finally, 23 publications met the inclusion criteria, and 15 screening tools were selected for further analysis. The literature search in this study was based on the Preferred Reporting Items for Systematic Review and Meta-Analysis statement (refer to the flow diagram presented in Figure 1).

Figure 1:

Systematic review: PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart

Quality Appraisal

Two researchers reviewed all selected publications independently using the COSMIN checklist (Prinsen et al., 2018), which includes nine measurement properties, including internal consistency, reliability, measurement error, content validity, construct validity, hypothesis testing, cross-cultural validity, criterion validity, and responsiveness. The quality of publications was classified as inadequate, doubtful, adequate, and very good. The GRADE approach was adopted for evaluating the quality of evidence (Schünemann et al., 2017). Publications were further graded as not serious, serious, very serious, and undetected with reference to risk of bias, inconsistency, imprecision, indirectness, and publication bias. Furthermore, the quality of evidence was classified as high, moderate, low, and very low. If there were divergent opinions in the process of quality assessment, a final decision was reached after holding discussions with a third reviewer.

Results Study Selection

Twenty-three publications, which used 15 different screening tools, met the inclusion criteria. The applicable subjects, assessment items, and reliability and validity of the identified tools are presented in Tables 1 and 2. The included tools assessed the following forms of child abuse: physical abuse (n = 6); abuse-related head trauma (n = 3); physical abuse and neglect (n = 2); child abuse and neglect (n = 1); physical and sexual abuse (n = 1); physical abuse, emotional abuse, and neglect (n = 1); and sexual abuse (n = 1). These tools were mainly developed in the United States or in European countries, with seven developed in the United States, four in the Netherlands, two in the United Kingdom, and one each in Spain and South Korea.

Table 1 - Description of Selected Studies That Examined Child Abuse Screening Instruments Study Instrument Inclusion Criteria (Children) Form of Child Abuse Sample Size No. of Items Scoring/Cutoff Point Sensitivity Specificity AUC % 95% CI % 95% CI % 95% CI 1. Berger et al. (2016; United States) PIBIS < 1 y in the emergency department (ED) Abused head trauma 1,040 4 A 5-point scale that assessed (a) abnormality on dermatologic examination (2 points), (b) age ≥ 3.0 months (1 point), (c) head circumference > 85th percentile (1 point), and (d) serum hemoglobin level < 11.2 g/dl (1 point)/cutoff point: total score of 2 points 93.3 [89.0, 96.3] 53.0 [49.3, 57.1] 83.0 [80.0, 86.0] 2. Chang et al. (2004; United States) DIPCA < 3 y identified by External Injury Codes (E-codes) in the range of 967.0–967.9 Physical abuse 11,919 6 A 15-point scale with (a) 1 point for fracture of base or vault of skull; (b) 2 points each for contusion of eye, rib fracture, intracranial bleeding, multiple burns, or age of 1–3 y; and (c) 6 points for age of 0–1 y/cutoff point: total score of 3 points 72.5 89.1 86.0 3. Chang et al., (2005; United States) SIPCA < 14 y identified based on ICD-9, Clinical Modification codes 800–959 Physical abuse 58,558 6 A 15-point scale with (a) 1 point for fracture of base or vault of skull; (b) 2 points each for contusion of eye, rib fracture, intracranial bleeding, multiple burns, or age of 1–3 y; and (c) 6 points for age of 0–1 y/cutoff point: total score of 3 points 86.6 80.5 89.0 4. Cowley et al. (2015; United Kingdom) PredAHT < 3 y with an intracranial injury in the pediatric intensive care unit (PICU) Abused head trauma 198 6 Yes/no questions/cutoff point of 3 points 72.3 [60.4, 81.7] 85.7 [78.8, 90.7] 88.0 [82.3, 92.6] 5. Ezpeleta et al. (2017; Spain) INTOVIAN < 3 y in public health centers Physical abuse, emotional abuse, neglect 219 9 Yes/no questions/cutoff point of at least 1 point – – 6. Hymel et al. (2014; United States) Four-variable CPR < 3 y for intensive care of head injuries Abused head trauma 291 4 Yes/no questions/cutoff point of at least 1 point 96.0 [90.0, 99.0] 43.0 [35.0, 50.0] 78.0 7. Kemp et al. (2018; United Kingdom) BuRN-Tool < 16 y with a burn in the pediatric ED Physical abuse, burn 1,327 7 Integer scores ranging from 0 to 3 points/cutoff point: total score of 3 points 87.5 [61.7, 98.4] 81.5 [77.1, 85.4] 87.0 [83.0, 90.0] 8. Louwers et al. (2014; Netherlands) Escape < 18 y who visited the ED Physical abuse 38,136 6 Yes/no questions/cutoff point of at least 1 point 80.0 [67.0, 89.0] 98.0 9. Paek et al. (2018; South Korea) FIND < 14 y who visited the ED with injuries Physical abuse, neglect 3,855 8 Yes/no questions/cutoff point of at least 1 point – – 10. Pierce et al. (2010; United States) TEN-4 BCDR < 4 y with abusive or accidental trauma in the PICU Physical abuse, bruises 95 1 Bruising on the torso, ear, or neck for a child aged less than 4 y, and bruising in any region for an infant aged less than 4 months 97.0 84.0 11. Schols et al. (2019; Netherlands) ERPANS < 1 y assessed during a home visit of families Physical abuse, neglect 1,257 31 4-point response format ranging from 0 (never observed or reported) to 3 (very often observed or reported)/cutoff point: total score of 1 point – – 12. Shakil et al. (2018; United States) PedHITSS <12 y in clinic completed by parents Physical abuse, sexual abuse 422 5 5-point Likert scale (0 = never, 1 = rarely, 2 = sometimes, 3 = fairly often, or 4 = frequently)/cutoff point: total score of 1 point 85.0 [81.0, 89.0] 13. Sittig et al. (2011; Netherlands) SPUTOVAMO-R < 7 y with physical injury in the ED Physical abuse 5,000 6 Yes/no questions/cutoff point of at least 1 point – – 14. van der Put et al. (2017; Netherlands) IPARAN < 1 y assessed during a home visit of families Child abuse, neglect 4,692 16 4-point response scale (always, often, sometimes, never) or a yes/no option
Each item is assigned a score between 0 and 2/cutoff point of at least 1 point 66.7 77.4 72.0 [59.3, 84.7] 15. Wells et al. (1997; United States) SASA < 15 y with risk of sexual abuse based on the tool completed by parents in a clinic Sexual abuse 121 12 Yes/no questions/cutoff point: total score of 3 points or more 90.9 88.5

Note. “INTOVIAN” was a European Commission-funded project name. AUC = area under curve; y = years; PIBIS = Pittsburgh Infant Brain Injury Score; DIPCA = Diagnostic Index for Physical Child Abuse; SIPCA = Screening Index for Physical Child Abuse; PredAHT = Predicting Abusive Head Trauma; CPR = Clinical Prediction Rule; BuRN-Tool - Burns Risk assessment for Neglect or abuse Tool; FIND = Finding Instrument for Nonaccidental Deeds; TEN-4 BCDR = Torso, Ear, and Neck Bruising Clinical Decision Rule; ERPANS = Early Risks of Physical Abuse and Neglect Scale; PedHITSS = Pediatric Hurt-Insult-Threaten Scream-Sex screening tool; SPUTOVAMO-R = acronym consisting of the first letters of the question in Dutch; IPARAN = Identification of Parents At Risk for child Abuse and Neglect; SASA = Symptoms Associated with Sexual Abuse; ICD-9 = International Classification of Diseases-9th Edition.

Table 2 - COSMIN Checklist for Evaluating the Methodological Quality of Individual Studies That Utilized Child Abuse Screening Instruments Study Instrument Internal Consistency Reliability Measurement Error Content Validity Structural Validity Hypotheses Testing Criterion Validity Responsive-ness Cross-Culture Validity Berger et al. (2016) PIBIS Very good NA Adequate Adequate NA Adequate Adequate Very good NA Chang et al. (2004) DIPCA Adequate NA NA NA NA Adequate Adequate Very good NA Chang et al. (2005) SIPCA Very good NA NA Adequate NA Adequate Adequate Very good NA Cowley et al. (2015) PredAHT Adequate NA Adequate NA NA Adequate Adequate Very good NA Ezpeleta et al. (2017) INTOVIAN Very good Adequate NA Adequate NA Adequate NA NA NA Hymel et al. (2014) 4-variable CPR Adequate NA Adequate Adequate NA Adequate NA Doubtful a NA Kemp et al. (2018) BuRN-Tool Very good NA Adequate Adequate NA Adequate NA Very good NA Louwers et al. (2014) Escape Very good NA Adequate Adequate NA Adequate Adequate Very good NA Paek et al. (2018) FIND Adequate NA Doubtful b Adequate NA Adequate Doubtful c NA NA Pierce et al. (2010) TEN-4 BCDR Adequate NA Adequate NA NA Adequate Adequate Adequate NA Schols et al. (2019) ERPANS Very good Adequate Adequate Adequate Adequate Adequate NA NA NA Shakil et al. (2018) PedHITSS Very good Very good Adequate Adequate Adequate Adequate Adequate Adequate NA Sittig et al. (2011) SPUTOVAMO-R Very good NA NA Adequate NA Adequate Adequate NA NA van der Put et al. (2017) IPARAN Very good NA NA Adequate NA Adequate Very good Doubtful a NA Wells et al. (1997) SASA Adequate Adequate NA Na NA Adequate Very good Adequate NA

Note. Methodological quality: inadequate, doubtful, adequate, very good, and not applicable (NA). “INTOVIAN” was a European Commission-funded project name. PIBIS = Pittsburgh Infant Brain Injury Score; DIPCA = Diagnostic Index for Physical Child Abuse; SIPCA = Screening Index for Physical Child Abuse; PredAHT = Predicting Abusive Head Trauma; CPR = Clinical Prediction Rule; BuRN-Tool = Burns Risk assessment for Neglect or abuse Tool; FIND = Finding Instrument for Nonaccidental Deeds; TEN-4 BCDR = Torso, Ear, and Neck Bruising Clinical Decision Rule; ERPANS = Early Risks of Physical Abuse and Neglect Scale; PedHITSS = Pediatric Hurt-Insult-Threaten Scream-Sex screening tool; SPUTOVAMO-R = acronym consisting of the first letters of the question in Dutch; IPARAN = Identification of Parents At Risk for child Abuse and Neglect; SASA = Symptoms Associated with Sexual Abuse.

aLower specificity and AUC. b Unexplained or missing data. c Evaluation by emergency medicine board-certified physician without gold standard.

Population Characteristics

Of the 15 child abuse screening tools identified in the included articles, eight targeted children who had sought medical attention with injury and three targeted all children who had sought medical attention. The tools used in these studies included the Pittsburgh Infant Brain Injury Score (PIBIS), INTOVIAN (a European Commission-funded project name), and Escape (Berger et al., 2016; Ezpeleta et al., 2017; Louwers et al., 2014). The remaining four studies that utilized self-report tools to assess risk of child abuse were the Early Risks of Physical Abuse and Neglect Scale (ERPANS), Pediatric Hurt-Insult-Threaten Scream-Sex (PedHITSS) screening tool, Identification of Parents At Risk for child Abuse and Neglect (IPARAN), and Symptoms Associated with Sexual Abuse (SASA; Schols et al., 2019; Shakil et al., 2018; van der Put et al., 2017; Wells et al., 1997). These self-report tools were designed to be used by parents to provide descriptions of either their parental behavior or their observations of their child's behavioral problems. Eleven studies provided descriptions of child characteristics, including mean ages of less than 1 year (n = 7), 1–4 years (n = 2), and 5–8 years (n = 2).

Screening Tools

Healthcare providers mainly provide care in hospital and community settings. When an abused child is sent to the hospital, healthcare providers will conduct physical examinations and tests using medical devices to evaluate whether that child's symptoms are related to abuse. In addition, when conducting family visits in the community, there is also the opportunity to assess the caregiver's parenting and environment to facilitate the early detection of abused children. To identify the key factors of child abuse, the assessment items and methods of the child abuse screening tools include questions on the symptoms of abuse and on whether the child is in a high-risk environment. On the basis of the assessment items, the screening tools of the selected studies were classified into three major categories. Those in the first category involved an objective assessment of the consistency of the mechanism and severity of injury as determined through a physical examination and review of the child's medical history as well as its appropriateness with the child's development and abilities. In this category, interviews and physical examinations such as visual inspections to check for bruising and burns/scalds were used as the basis for assessing whether the mechanism of injury was consistent with the child's developmental stage. Of the 15 tools identified in this review, five were in this category, including INTOVIAN for assessing physical abuse, emotional abuse, and neglect (Ezpeleta et al., 2017); the Burns Risk assessment for Neglect or abuse Tool (BuRN-Tool) for assessing burns (Kemp et al., 2018); the Torso, Ear, and Neck Bruising Clinical Decision Rule (TEN-4 BCDR) for assessing bruising (Pierce et al., 2010); and Escape and SPUTOVAMO-R (acronym consisting of the first letters of the question in Dutch) for assessing physical abuse (Louwers et al., 2014; Sittig et al., 2011).

Tools in the second category involved the use of biochemical tests and precision imaging in addition to interviews and physical examinations for determining the mechanism of injury. Examinations included x-ray imaging for detecting skull or long bone fractures, computed tomography imaging for detecting intracranial hemorrhaging, fundoscopic examination for detecting retinal hemorrhaging, the measurement of head circumference, and the measurement of serum hemoglobin level. Among the 15 tools identified in this review, six belonged to this category, including PIBIS, Predicting Abusive Head Trauma (PredAHT), and the four-variable Clinical Prediction Rule (CPR) for assessing abusive head trauma (Berger et al., 2016; Cowley et al., 2015; Hymel et al., 2014); the Screening Index for Physical Child Abuse (SIPCA) and Diagnostic Index for Physical Child Abuse (DIPCA) for assessing physical abuse (Chang et al., 2004, 2005); and Finding Instrument for Nonaccidental Deeds (FIND) for assessing physical abuse and neglect (Paek et al., 2018).

Tools in the third category involved assessing the risk of child abuse through parental self-reporting on physical and mental health issues, parenting and disciplining methods, and child-related emotional and behavioral issues. Among the 15 tools identified in this review, four were in this category, including ERPANS for assessing physical abuse and neglect (Schols et al., 2019), PedHITSS for assessing physical and sexual abuse (Shakil et al., 2018), IPARAN for assessing child abuse and neglect (van der Put et al., 2017), and SASA for assessing sexual abuse (Wells et al., 1997).

Except for ERPANS, which comprises 31 items (Schols et al., 2019), and the TEN-4 BCDR, which comprises one item (Pierce et al., 2010), the remaining 13 tools comprise between four and 16 items. Yes/no questions were used in eight tools, a 4- or 5-point Likert scale was used in three tools, and a weighted scoring system was used in four tools.

Most of the reviewed publications (n = 11) did not state whether training was required before using the associated screening tool. However, basic medical and nursing knowledge was clearly a necessary although unstated prerequisite, as professional knowledge is required to determine the level of consistency between physical examination and medical history results. Screening tool training methods and content were elucidated in four publications, as follows: SPUTOVAMO-R: description of