Variations in Guidelines for Diagnosis of Child Physical Abuse in High-Income Countries

Key Points Question Are clinical guidelines for the early detection and diagnostic workup of child physical abuse complete, clear, and consistent across high-income countries? Findings In this systematic review that included 20 clinical guidelines issued in 15 countries, guidelines were clear but incomplete and discrepant, particularly in the definition of sentinel injuries and in recommendations for exploratory laboratory testing and advanced imaging. Meaning This systematic review found a lack of standardized guidelines for the identification and management of child physical abuse, which may contribute to practice variation.


Introduction
Physical abuse is estimated to occur in 4% to 16% of the population younger than 18 years in highincome countries, 1 and the World Health Organization has considered child physical abuse (CPA) an international priority and has developed a vast program of CPA prevention. 2 CPA is more frequent in infants 2 years or younger in whom, in the absence of criteria of certainty other than the rare confession of the perpetrator, diagnosis is complex and relies on a combination of social and clinical evaluations, imaging, and laboratory tests. [3][4][5] False-negative results expose infants to a risk of recurrence estimated at 35% to 50%, 6,7 with its associated morbidity and mortality in the short [8][9][10] and long term. 11,12 However, false-positive results may delay the diagnosis of severe underlying diseases, such as bone fragility (eg, osteogenesis imperfecta) or bleeding disorders (eg, hemophilia) and lead to an inappropriate child protection decision. 13 Thus, early detection of CPA based on sentinel injuries (ie, injuries in noncruising infants or with implausible explanations), 14 alone or interpreted with the help of clinical decision rules, such as the TEN-4 rule (ie, bruises on the torso, ears, and neck in children younger than 4 years may be indicative of CPA), 15 and accurate diagnostic workup with imaging and laboratory tests are of paramount importance.
To help physicians optimize the detection and diagnosis of CPA and consider differential diagnoses, clinical guidelines have been developed by academic societies and health agencies.
Despite these efforts at standardization, several studies reported suboptimal practices by health care practitioners. For example, in 2018, 36% of physicians in 4 European countries considered that an infant aged 10 weeks with bleeding from the mouth was not a child protection concern. 16 In a French national survey performed in 2015, only 28% of pediatricians would prescribe magnetic resonance imaging (MRI) of the head for the diagnostic workup of CPA in an infant aged 9 months with a fractured femur, numerous bruises, and head trauma. 17 Lack of completeness, clarity, and consistency are among the reasons why clinical guidelines fail to standardize practices and thus mislead health professional practices. 18,19 For example, we recently identified a between-guideline discrepancy for the imaging workup to be performed to detect skeletal injuries when CPA is suspected, notably the role of bone scintigraphy. 20,21 Identifying the specific fields for which guidelines lack completeness, clarity, or consistency for the early detection and diagnostic workup of CPA could help prioritize clinical questions requiring original diagnostic studies, systematic reviews (as performed in the aforementioned example of bone scintigraphy), 20 or an international consensus process. Our objective was to systematically investigate the completeness, clarity, and consistency of clinical guidelines for the early detection and diagnostic workup of CPA that were issued by academic societies and health agencies in high-income countries.

Methods
This systematic review was performed according to guidance from the Centre for Reviews and Dissemination 22 and its reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline. A protocol was registered (Prospero No. CRD42020203809). In this systematic review, we aimed to identify, analyze, and compare all guidelines that were issued after 2010 by academic societies or health agencies in high-income countries with a guidance on the early detection and/or diagnostic workup of CPA in infants aged 2 years or younger. 23 This age limit was selected because most nonclinically visible injuries are found before this age. 24,25 The database searches and data extraction and synthesis were performed by 2 of us (F.B. and Y.R.) independently. Disagreements were resolved by consensus or by consulting 1 or several other review authors (F.B., S.A., Y.R., and M.C.). We translated the non-English guidelines with the help of native-speaker physicians.

Search Strategy and Selection Criteria
We searched MEDLINE (via PubMed), Web of Science, Google Scholar, websites reporting guidelines (eg, SUMSearch 2, 26 Guidelines International Network, 27 and Trip Database 28 ) from inception to June 15, 2020 (last update), with no language restrictions, as well as the websites of academic societies and health agencies in the 24 countries with the highest incomes (eTable 1 in Supplement 1). 23 The search strategy for databases combined groups of keywords pertaining to child abuse, diagnosis, and guidelines (eTable 2 in Supplement 1). We assessed potential guidelines for inclusion by screening titles, abstracts, and, eventually, full texts of all search results. When several guidelines had been published by the same academic society or health agency since 2010, we included only the most recent one. We also contacted the national chairs of the countries covered by the European Confederation of Primary Care Paediatricians (ECPCP) (eTable 1 in Supplement 1) and asked them about current guidelines in their countries. Finally, we screened reference lists of included guidelines. When guidelines were published in several parts, 3 we considered them as a single guideline. When guidelines endorsed another guideline, 29,30 we chose the most recent one. 4 We included all the guidelines aimed at providing general guidance for the early detection and diagnostic workup of CPA. For some specific review questions, we also included guidelines with a narrower scope, notably those regarding the recommended diagnostic workup to detect abusive head trauma or skeletal injuries.

Data Extraction and Synthesis
For each included guideline, we extracted its characteristics, including country, year of dissemination, development process with the report of the group membership involved, search methods (eg, systematic review, in particular the description of the scope and grading or rating of the recommendations), [31][32][33] and specific content. We classified the scope of each guideline as a general guidance for the early detection and diagnostic workup of CPA or a narrower one. We compared the presence and the detailed content of a definition of sentinel injuries. We listed the recommended diagnostic workup (imaging and laboratory tests) for CPA.

Data Analysis
Given the numerous tests proposed, we restricted the subsequent analyses to those suggested in more than 2 guidelines. From the guidelines' text, we classified whether the test was recommended or not, and whether it was recommended systematically or on a case-by-case basis according to the clinical context or if the recommendation was unclear. We calculated the overall proportion of missing statements and the proportion of unclear statements among the nonmissing statements for all tests.
To compare the guidelines' contents, we grouped the tests according to the 4 domains they dealt with: detection of skeletal, head and spine, or thoracoabdominal injuries and exploration of differential diagnoses. For each diagnostic test, we calculated the proportion of guidelines providing a statement for it if it was expected given their scope. Analyses were conducted from July 2020 to February 2021.   (Table 3). All guidelines recommended radiological skeletal survey up to age 2 years, except 1 guideline that did not mention any age limit, 40 and 1 "Systematic," "systematically," "should," "is required," "in all children"

JAMA Network Open | Pediatrics
The test has to be performed systematically "Could," "may," "might," "consider," "in case of," "is often used," "if the child is at risk for," "possibly," "based on findings" The test should be performed on a case-by-case basis, according to the clinical context "Consider neuroimaging," 46 "additional imaging studies may be indicated" 46 The recommendation is unclear (regarding the neuroimaging and additional imaging to be performed)       recommended systematic performance of liver and pancreatic enzymes, and 1 guideline 38 recommended testing liver enzymes "in case of bruises or muscle injuries" (without specifying their location). Six guidelines recommended renal function testing systematically, 36,40,46 or according to the clinical context without giving more details. 37,41,44 Four guidelines recommended troponin and/or creatine kinase testing to detect cardiac injury systematically 4,36,40 or according to the clinical context 41 without giving more details, but the other guidelines did not mention these tests.

Discussion
In this systematic review of 20 clinical guidelines for the early detection and diagnostic workup of CPA in infants, we identified a few unclear statements but a frequent lack of completeness of guidelines and numerous between-guideline discrepancies. Guidelines agreed with recommending radiological skeletal survey, head CT, head MRI, and eye fundus examination but disagreed on whether these should be systematically performed or not. Other main discrepancies dealt with defining sentinel injuries and performing bone scintigraphy, follow-up skeletal survey, spinal MRI,   Performed for the detection of glutaric aciduria type 1. cranial ultrasonography, chest CT, and abdominal ultrasonography and CT systematically, on case-bycase basis, or not. For ruling out differential diagnoses, guidelines agreed on blood tests to explore primary hemostasis and coagulation, but with some discrepancies in the tests to be performed, and only half of the included guidelines mentioned the need to investigate bone metabolism.

JAMA Network Open | Pediatrics
The guidelines were based on a limited number of well-designed primary studies, particularly for the definition and diagnosis of sentinel injuries, 14,50 the use of bone scintigraphy, 20 and laboratory tests 29,51 to perform in case of suspected CPA. The lack of primary studies may have led guidelines developers to opt for expert consensus rather than evidence-based guidelines, and this may explain the substantial heterogeneity among guidelines. Also, the methods used by the developers of these guidelines often did not follow international recommendations. 32

Interpretation and Implications
The lack of completeness of guidelines and between-guideline inconsistencies may mislead physicians' decisions. Efforts at the national level to standardize practices by producing guidance to help physicians optimize the detection of inflicted injuries and consider differential diagnoses may be jeopardized by the heterogeneity observed among guidelines at the international level, given their high online accessibility. Between-country variability of clinical guidance may be explained by regional variations in the epidemiological characteristics of diseases or accessibility of diagnostic tests, 53 but the detection and the diagnostic workup of CPA in high-income countries should be standardized. Other factors, such as clinical recommendations published in journals with high impact factor, 9 could also influence physicians' decisions, but we believe this between-guideline heterogeneity explains in part the variability and the suboptimality of observed practices for the detection and diagnostic workup of CPA. 16,17,54 Our systematic review could aid in drawing the research agenda to optimize the detection and diagnostic workup of CPA. First, priority clinical questions for which guidelines lacked consistency were bone scintigraphy, follow-up skeletal survey, spine MRI, cranial ultrasonography, chest CT, abdominal CT and ultrasonography, and laboratory tests for abdominal injuries or for differential diagnoses. The methods needed to reach international consensus may vary depending on the clinical question. For example, in the past decade, well-designed original studies and systematic reviews showed the important role of head and spine MRI to detect additional and extracranial injuries. [55][56][57][58][59] Thus, an update of the oldest included guidelines would probably lead to more between-guideline consistency. Other systematic reviews have shown the complete lack of well-designed original studies, such as for bone scintigraphy, 20 pointing to the need to conduct such studies. There is also a clear need to define sentinel injuries, to agree not just on their location, size, patterns, and number but also on the term sentinel injury because it is not shared by all experts in the field of CPA. The TEN-4 rule could help in the definition of sentinel injuries by providing a simple tool to help clinicians classify bruises as sentinel injuries. 15 Finally, our results suggest that developers of guidelines for CPA detection and diagnosis should follow international recommendations for their development process to notably rate the strength of recommendations based on the available evidence. Given the constant production of knowledge in the detection and diagnostic workup of CPA, an international consensus should be actualized on a regular basis to incorporate all the available evidence, as has

Limitations
This study has some limitations. First, we could not find guidelines in more than one-third of the highincome countries included in the search, even though we performed a systematic search of several databases and relevant websites, with no language restriction. We could only identify 4 additional guidelines by asking European experts in pediatrics what guidelines were in force in their country. We may have missed existing guidelines, in particular in countries not covered by the ECPCP. However, an exhaustive search would probably have increased the between-guideline variability. Second, we subjectively decided the specific guidance expected for each guideline according to the title and scope, by a consensus of coauthors, to compare them. Third, by removing the tests suggested in 2 guidelines or fewer, we risked not analyzing promising new tests, such as whole-body MRI for the detection of skeletal and muscular injuries. 61

Conclusions
This systematic review identified flaws in guidelines' completeness and between-guideline discrepancies that could contribute to the observed variations in clinical practices. Primary care health practitioners and hospital-based physicians are the first-line and key actors for the early detection and diagnosis of CPA in infants, and their decisions should be based on complete, clear, and consistent guidelines. Our findings may help identify priorities for well-designed original diagnostic accuracy studies, systematic reviews or an international consensus process to produce clear and standardized guidelines to optimize practices and infants' outcomes.