Circular reasoning, confessions and abusive head trauma : A critical analysis of Edwards et al. (2020)

A major issue in current clinical research regarding shaken baby syndrome (SBS) and abusive head trauma (AHT) has been described as circular reasoning. Circular reasoning in this context means that a child protection team makes the SBS/AHT diagnosis, and that this diagnosis is also used as the reference test (gold standard) when researchers classify SBS/AHT cases as true positive cases. Circular reasoning results in a high risk of bias. In order to avoid this, videotaped, witnessed and spontaneously confessed shaking has been suggested as the currently least problematic means to avoid circular reasoning. Against this backdrop it is welcomed that Edwards et al. (2020) have conducted a systematic literature review in order to identify studies where someone has confessed to have shaken a baby with or without signs of impact. Even though Edwards et al. dismissed the risk of bias due to false confessions, false confessions can occur, e.g. if an expert witness convinces police and/or prosecutor that, in a case of isolated retino-dural haemorrhages (i.e. without signs of impact), violent shaking is the only possible mechanism. In such cases circular reasoning can hardly be avoided. In the best world, circumstances under which confessions are provided should be documented by video recording to allow critical assessment. Edwards et al., and Adamsbaum et al. (2010), explicitly make an important distinction between two major stated reasons for shaking a baby: a) shaking after a baby's collapse as a revival attempt, and b) shaking due to anger, frustration due to colicky crying, etc. Based on this distinction, we propose that these two major stated reasons for shaking actually should be separated, and that a caregiver who claims that he/she did shake the infant but as a revival attempt can be telling the truth. This hypothesis could be tested, e.g. by analysing the effect of AHT preventive interventions after separating the two categories of stated reason for shaking. Edwards et al. criticised the Swedish SBU report for having included only two (2) confession studies and claimed that they themselves, with their systematic literature review, ‘clearly refute the argument that there are insufficient data within the published literature to support shaking as an important cause of AHT’. The correctness and relevance of this criticism depends on the research question, the difference between ‘data’ and ‘evidence’, as well as if the inclusion criteria, the exclusion criteria and the applied quality assessment are in accordance with the accepted principles of systematic literature reviews and evidence-based medicine