LAUSR

Thank you for your interest in our study and the interesting questions you raise. In this study, our aim was to validate and if possible sharpen the predictive value of clinical findings for the diagnostic of abusive head trauma (AHT) and to escape the Bcircularity trap^ (basing the diagnosis on clinical findings then evaluating the diagnostic value of the same clinical finding). We had not intent to replace the classical triad with another (and even less to christen it). The reason why we replaced the general notion of Bretinal hemorrhage^ (RH) by diffuse (or severe) RH (question 2) was that several studies have shown the presence of mild RH in accidental trauma [1, 2], and we found that severe RH, extending to the periphery of the retina were found only in the context of AHT. In our study, we also found that cerebral edema (or encephalopathy), although a strong predictor of AHT, was present only in a minority of cases (the most severe ones), and was thus affected by a high rate of false negative. We found it important to insist on diagnosing the less severe cases, without encephalopathy, because the literature shows that missing this diagnosis puts the child at high risk of recurrent abuse, with catastrophic consequences [3]. We were also attracted to the fact that most shaken babies lacked signs of impact, and we recorded systematically these findings in our database. The issue of Babsence of signs of impact^ may be contentious, however. It presupposes that all children are carefully examined on admission, even when no notion of trauma is given by the caretakers (as is regularly the case in child abuse), and that these negative findings are recorded systematically; hence the importance of the prospective collection of data. It also demarcates shaken baby syndrome (SBS) from beaten child syndrome, and even from Bshaken-impact baby syndrome^, a more severe variant of AHT. Whatsoever, our study showed a strong correlation of this negative finding with confessed AHT, compared with witnessed accidental trauma. Thus, we did not replace encephalopathy with scalp swelling (question 1), but we evaluated both variables for their diagnostic value. Regarding the recalculation carefully performed by the authors of the letter, we must correct the latter: not all accidental traumas had scalp swelling, it was enough, not to fulfill the Btriad^ criteria, that they lacked subdural hemorrhage, or severe retinal hemorrhage (which they usually did). We admit that we were a bit disturbed to find a 100% positive predictive value for the association of severe RH with subdural hematoma (SDH) and absence of signs of impact, because this figure does not look like a scientific result; however, from a legal perspective, we think that this is precisely what a judge hopes for. Since 2010, our registry of head injuries in infants has increased to 1146 cases, including 321 abusive cases, and 281 corroborated cases (accidental or abusive); we have repeatedly recalculated the figures with the newly accrued cases, and the results have just confirmed the very strong predictive value of the association: SDH, severe RH, and absence of scalp swelling. In response to questions 3–5, in almost all cases, confession of abuse was obtained during the judiciary process. Information regarding the confession was obtained from legist physicians enlisted as expert by the judge. It has been argued earlier that confession may be biased by a plea-bargain between the accusation and the accused [4], however pleabargain is not practiced in the French judiciary system. From our medical perspective we have no competence to decide whether these confessions were genuine or not, but since no * Matthieu Vinchon [email protected]