Neonate With Choking

Pharyngeal trauma: When to suspect physical abuse

“…In most cases the injury is witnessed, or the child is able to provide a history accounting for such an injury. Non abusive mechanisms can also involve trauma during intubation, swallowing a foreign body or a fall with a penetrating object in the mouth such as a toy or straw (Chauhan et al, 2006; Morrison & Pashley, 1988; Nolte, 1993; Shaughnessy et al, 2015; Tostevin et al, 1995).…”

“…Caretaker frustrations during infant feeding and calming efforts may exacerbate tensions resulting in excessive forces in bottle feeding, insertion of fingers, or other objects about the infant's mouth (Ablin & Reinhart, 1990; McDowell & Fielding, 1984). While several cases of oropharyngeal trauma have been previously documented within the literature, most discuss accidental trauma, as well as trauma caused by foreign bodies, or trauma caused by medical intervention, such as intubation (Chauhan et al, 2006; Morrison & Pashley, 1988; Nolte, 1993; Shaughnessy et al, 2015; Tostevin et al, 1995). Apart from a recent report which highlighted abuse as part of the differential diagnosis and included appropriate clinical management, most published cases either do not include abuse within the differential diagnosis or fail to mention other workup considered, leaving a sizeable gap in knowledge regarding the importance of diagnostic examinations (Bishop et al, 2022; Brietzke & Jones, 2005; Chauhan et al, 2006; Hennelly et al, 2010; Narchi, 2003; Park & Witt, 2004; Printz et al, 2017; Shaughnessy et al, 2015).…”

Further Reflections on the Use of Large Language Models in Pediatrics

“…This metric is on par with past performance of human raters . The authors also exclude essential figures, videos, and tables from the LLM input that are usually available to humans. Clinical information stored in visuals (eg, rash photographs, fluoroscopy videos, and laboratory results) are often necessary for diagnostics.…”

“…The authors also exclude essential figures, videos, and tables from the LLM input that are usually available to humans. Clinical information stored in visuals (eg, rash photographs, fluoroscopy videos, and laboratory results) are often necessary for diagnostics. Future studies that include a clinician accuracy benchmark and identical tasks will better contextualize error rates.…”