Minimising bias in the forensic evaluation of suspicious paediatric injury

• Forensic experts must safely manage uncertainties when interpreting medical evidence.
• Cognitive and contextual biases can influence reliability of expert opinion.
• Forensic paediatricians can provide opinion for prosecution or defense if qualified.
• Forensic assessment must be differentiated from advocacy in medical child protection.
• Medical opinion should be restricted to biomechanics and injury plausibility.

In the rules of evidence in all legal jurisdictions, medical experts are required to maintain objectivity when providing opinions. When interpreting medical evidence, doctors must recognise, acknowledge and manage uncertainties to ensure their evidence is reliable to legal decision-makers. Even in the forensic sciences such as DNA analysis, implicit bias has been shown to influence how results are interpreted from cognitive and contextual biases unconsciously operating. In cases involving allegations of child abuse there has been significant exposure in the media, popular magazines, legal journals and in the published medical literature debating the reliability of medical evidence given in these proceedings. In these cases judges have historically been critical of experts they perceived had sacrificed objectivity for advocacy by having an investment in a ‘side’. This paper firstly discusses the issue of bias then describes types of cognitive biases identified from psychological research applied to forensic evidence including adversarial bias, context bias, confirmation bias and explains how terminology can influence the communication of opinion. It follows with previously published guidelines of how to reduce the risk of bias compromising objectivity in forensic practices then concludes with my own recommendations of practices that can be used by child protection paediatricians and within an organisation when conducting forensic evaluations of suspicious childhood injury to improve objectivity in formulation of opinion evidence.