A follow-up skeletal survey at 2 weeks increases the yield, with 21.5 % of subjects having new information noted on the follow-up skeletal survey (Fig. 2). Therefore, follow-up skeletal surveys have become customary in child abuse protocols (Harper et al. 2013). A 15-view limited follow-up survey found new information in 38 % of studies, with less radiation and without missing clinically significant new fractures (Harlan et al. 2009). Because of the high morbidity and mortality risk of missing child abuse, the follow-up skeletal survey should be routinely performed (Singh et al. 2012). Obtaining skeletal surveys on siblings of abused children has also been suggested. Although radiation exposure is a risk associated with obtaining a skeletal survey, there are no studies that specifically describe the actual risks from radiation exposure to children receiving a skeletal survey (Karmazyn et al. 2011). Radionuclide bone scans, which cost 6 times more than a skeletal survey, have been used more as a supplement to the radiographic skeletal survey, rather than as the primary imaging modality (Mandelstam et al. 2003). Magnetic resonance imaging (MRI) or ultrasound examinations may be used to supplement the skeletal survey in cases of suspected physeal or soft tissue injury. CT scans are used to evaluate for abdominal injury or abusive head trauma. MRI provides full assessment of the intracranial pathology in child abuse in order to detect asymptomatic, nonacute parenchymal brain lesions. Diffusion- and susceptibility-weighted imaging is very reliable for diagnosing hypoxic-ischemic brain injury and parenchymal hemorrhage (Parizel et al. 2003). MRI can also detect related spinal cord pathology in the child with AHT, such as ligamentous injury and intraspinal injuries such as spinal cord injury without radiographic abnormality or proximal brachial plexus injuries, especially at the root level.

A skeletal survey is useful to screen for occult fractures in children suspected to be at high risk for abuse. However, there are marked variations in the use of the survey across children’s hospitals. The controversy is in the interpretation of the AAP guidelines, what the physician determines to be a “suspected” case, the physician experience level, the standard of care for the institution, and the availability of child abuse specialists and programs. In a survey of 40 children’s hospitals, screening was performed for 83 % of children under the age of 2 years who were diagnosed with physical abuse (Wood et al. 2012). The likelihood of screening varied from 55 % to 93 % and was higher for those hospitals with a child abuse program. In particular, the use of a skeletal survey for children with traumatic brain injury varied from 38 % to 88 % (mean 68 %) and for infants with a femur fracture from 41 % to 94 % (mean 77 %). This variation in the interpretation and use of the skeletal survey has led to a multidiscipline expert panel RAND/UCLA appropriateness method development of recommendations for obtaining a skeletal survey in young children less than 2 years of age, presenting with fractures (Wood et al. 2014).

A skeletal survey is very useful for detecting occult fractures in infants less than 6 months of age (Duffy and Squires 2011). Children with abusive head trauma had a higher chance of an occult fracture compared to other types of abuse injuries (Day et al. 2006; Duffy and Squires 2011). For children with isolated skull fractures but no abusive head trauma, the skeletal survey found additional fractures in only 6 %, typically in premobile infants under 6 months of age (Laskey et al. 2013). The older child with a simple linear skull fracture is less likely to have a positive skeletal survey. Laskey et al. found that the majority of long falls had only a simple skull fracture with a negative skeletal survey (Laskey et al. 2013).

The skeletal survey should be thought of as a supplement to the history and physical examination in the infant and young child under the age of 2 years who cannot otherwise give a history or cooperate in a physical examination. A skeletal survey is recommended for all infants under 1 year of age with a fracture, unless the fracture is clearly in a mobile infant who is cruising and could accidentally sustain a fracture such as a distal radius and ulna fracture or a toddler tibia or fibula fracture and a simple skull fracture in an infant older than 6 months of age from a plausible mechanism or a plausible birth fracture such as the clavicle. For children 12–24 months of age, a skeletal survey is necessary if there is a history of abuse or domestic violence, there is a delay in seeking care for a painful injury, the history is not consistent or plausible for the fracture, the fracture was reportedly from a non-plausible mechanism such as being hit by a toy, or there are suspicious soft tissue injuries. For any child under the age of 2 years, all rib fractures, classic metaphyseal lesions, complex skull fractures, humeral fractures with physeal separation, or femoral diaphyseal fracture should receive a skeletal survey as part of their evaluation (Wood et al. 2014).

New periosteal bone is a classic finding of a healing fracture on a skeletal survey (Table 2). The follow-up skeletal survey at 2 weeks can give more precise information about dating of the fracture (Section on Radiology AAP 2009). Dating of a fracture remains an imprecise endeavor. A recent review of 82 accidental long bone fractures in 63 children younger than 72 months of age showed that the fracture could be dated as acute if less than 1 week from the fracture, recent if 8–35 days, or old if >36 days (Prosser et al. 2012). Soft tissue swelling, periosteal reaction, soft callus, hard callus, bridging, and remodeling were six key features that allowed dating (Table 2).

Although any aged child can sustain an injury from abuse, it is the child under age 2 years without the verbal skills to tell their story, who is the most vulnerable to repeat abusive episodes. Up to 9,000 infants and children under 2 years of age are hospitalized in the USA with a fracture (Leventhal and Martin 2010). Approximately 20 % of these fracture cases are caused by abuse, although the numbers are likely much higher due to underrecognition and underreporting (Ravichandiran et al. 2010).

Hobbs described six important patterns of fracture from physical abuse (Table 3) (Hobbs 1989). However, no fracture pattern or location is pathognomonic of child abuse. Several fractures such as metaphyseal corner fractures (classic metaphyseal lesion) (Fig. 3), posterior rib fractures, and fractures in various healing stages are so commonly seen in non-accidental trauma that a high suspicion for abuse is always warranted for these fracture patterns (Baldwin and Scherl 2013). Prediction of abuse is often compounded by methodology, small retrospective series, a wide variation in ages of children, inclusion of all types of fractures, and when the mechanism of motor vehicle accident (MVA) is included in the analysis. Although there may be an association with abuse for a certain type of fracture, that association does not prove that the injury was caused by abuse. For most studies that show this association for a particular fracture type, the association is usually less than 50 %. For fractures in general, features such as young age, Medicaid insurance, nonwhite race, male sex, and presence of additional injuries such as traumatic brain injury had associations with abuse (Leventhal and Thomas 1993; Leventhal and Martin 2010; Lane and Rubin 2002; Skellern and Wood 2000). For an active child who is capable of rough housing on furniture, the rate of suspected abuse was only 5.5 % (Hennrikus and Shaw 2003).

Age seems to be one of the strongest predictors of abusive fractures. For children under age 3 years who present with any type of fracture from any mechanism, the probability of abuse ranges from 10.8 % to 22.4 % (Leventhal et al. 2007, 2008; Leventhal and Martin 2010). For isolated long bone fractures in children less than 3 years of age, only 1 % of these fractures were diagnosed by the treating physician to be from child abuse (Taitz and Moran 2004). For younger-aged children, the probability for abuse dramatically increases. In infants (<1 year of age) with any type of fracture, 26.3 % were reported to child protective services (CPS), with abuse being confirmed in 15.2 % of the total study group (Skellern and Wood 2000). Leventhal found an even higher risk (37.5 %) of abuse in infants (Leventhal and Thomas 1993). Other studies have found age <4 months (Skellern and Wood 2000) and age <24 months (Leventhal et al. 2008; Leventhal and Martin 2010) to be associated with an increased risk of abuse. As a general principle, because of the higher association of abuse with age under 1 year and the difficulty in performing a reliable physical examination on an infant in pain, one cannot be faulted for obtaining a skeletal survey on an infant who presents with a fracture even if the suspicion for child abuse is low. Although fractures in infants are highly associated with abuse, this association is typically less than 50 %. While important to be mindful and even suspicious about abuse in an infant, it is equally important to not be judgmental.

Upper extremity fractures have been associated with abuse, typically in younger children. Fractures from abuse may be in unusual locations such as the distal clavicle, acromial tip, scapula, proximal humeral metaphysis, or distal humeral physis (Fig. 4). The mechanism of fracture is from violent blows or traction injuries and is suggestive of abuse in young children.

While it is essential not to miss child abuse as the underlying cause of a child’s injuries, it is just as important to remain objective, be open to alternative explanations, and make an accurate diagnosis. Inappropriately diagnosing child abuse is harmful to the child and family, risks the parents losing custody, and is a misuse of resources (Schwend et al. 2000). Be wary of false allegations, especially when there are custody and visitation disputes or if the charges are coming from the parent (Bernet 1993). Alternative diagnoses are also brought up in custodial and criminal cases, so diagnostic accuracy during the initial evaluation is essential.

Injury to a developing bone of an infant leads to subperiosteal hemorrhage, leading to subperiosteal healing bone by 10–14 days (Hobbs 1989). There is a normal physiologic diaphyseal reaction that is frequently present in infants younger than 6 months of age and is symmetric and diaphyseal only on the long bones of the extremities (Pergolizzi and Oestreich 1995). Initial radiographs can be normal, with a skeletal survey and follow-up survey being most important. Besides trauma and infection, other conditions such as Caffey’s disease (infantile cortical hyperostosis), vitamin C deficiency, vitamin D deficiency, metastatic neuroblastoma, Langerhans histiocytosis, vitamin A intoxication, leukemia, and various drugs are more rare causes of subperiosteal bone formation than child abuse (Hobbs 1989). Other causes of fracture include osteogenesis imperfecta, rickets of prematurity, disuse osteoporosis, copper deficiency, and osteomyelitis (Hobbs 1989). Congenital syphilis is rare but is increasing in prevalence. Congenital syphilis readily mimics the fractures of child abuse with diaphysitis, metaphysitis, and pathologic fractures in various stages of healing, mimicking child abuse (Lim et al. 1995). If the skeleton otherwise looks normal radiographically, this strongly excludes general conditions of genetic, metabolic, or bone disease (Hobbs 1989).