• Child Abuse
• Paediatrics

Current guidance for investigating suspected non-accidental injury recommends a repeat skeletal survey after 11–14 days to detect any occult fractures missed on the original X-rays1 .

We investigated the yield when the current guidance is applied in routine practice in Barts Health NHS Trust, which includes three paediatric units and serves a child population of 236 000.

We reviewed 198 cases from January 2016 to September 2020, 174 (88%) patients were less than 2 years old and 133 (67%) were less than 1 year of age. 131 (66%) had a second survey.

The flow chart in figure 1 shows the diagnostic yield at the first and second skeletal survey.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Skeletal survey outcomes.

Of 131 children who had a second survey two further fractures were found (1.5%). As a proportion of the total who had a first survey, this was 1%. Both patients already had injuries identified; the impact on the safeguarding process is unclear. In the subgroup of 45 patients who had a second survey following an initial survey with no bony injury identified, no fractures were found in the second.

Why are our results different from the data used in formulating the current Royal College guidance?

A detailed look at the studies referenced in the guidance shows that these range from cohort sizes of 22–1470 children.2 ,3 4 They give yields of additional positive findings of 8.5–61%. However, all these studies are problematic as the second survey was done only on a small cohort of the total—most likely a selected population in whom there was the greatest clinical suspicion of abusive injury. Indeed; in Kleinman et al 2, which quotes yields of 61% it is stated that second surveys were done specifically because of high suspicion of abuse.

The largest study quoted by the guidance is Singh et al 3 in 2012. This study recognises the limitations of the previous studies because of the selected populations. However, even in this population, a small proportion of children who had a first survey had a second one. Out of 1470 children who had a first survey, only 169 had a second. We are not told what criteria were used to determine which children had a second survey. It is reasonable to assume that they were the children with whom there was the greatest concern about possible abuse. This study quotes a figure of 14% (24/169) of children having an additional injury. However, this is less than 2% (24/1470) of the original study population.

This proportion is not significantly different from our yield of 2/198 (χ² p value 0.48).

We would like to see the publication of the experiences of other units and would welcome debate among not only paediatricians and radiologists, but also practitioners from social care and family law, about what is the acceptable yield of additional injuries found to justify the intervention. We question whether the existing data really justifies a second skeletal survey in all children or whether a more nuanced approach can be taken.