Abstract
The problem of circular reasoning in studies of abusive head trauma (AHT) has been known for decades.1-3 Yet circular reasoning continues to be a central part of the methodologies adopted by researchers in the field. The present study by Snelling et al.,4 published in Emergency Medicine Australasia, is a case in point. Snelling et al. aim to compare characteristics of subdural hematomas (SDH) in cases of AHT (n = 20) to cases with non-inflicted aetiologies (n = 17). Categorisation as AHT or non-inflicted aetiology was ‘based on child protection interagency outcome’, with ‘the final determination of SDH being due to AHT was made via interagency meetings involving representatives from the government statutory authority, CPU (Child Protection Unit) paediatrician, police service and department of education’. This tells us who made the determination. It does not tell us how the determination was made. Science is not concerned with such ‘who’ information. What is important, and what is omitted from the study, is detailed information as to the evidentiary basis by which the ‘multi-disciplinary team’ made the categorisations. What is known is that certain medical findings are believed by CPU paediatricians to be associated with AHT, and that these findings (to quote from Snelling) ‘provide medical evidence for interagency consideration’. What is also known is that ‘child protection social workers and police officers are heavily reliant on clinicians to guide their decision-making’.5 Categorisation of the cases therefore relies very heavily on the interpretations of medical findings made by CPU paediatricians. Yet these medical findings are then listed in Snelling et al. as outcomes of the categorisation process, and as being associated with either AHT or with non-inflicted aetiologies. This is circular reasoning. Let us look at a concrete example. Certain retinal haemorrhages (RHs) are widely believed to be indicative of AHT, and their existence therefore forms a significant part of the CPU paediatricians' determinations, and hence the categorisations made by the multi-disciplinary team. In Snelling et al., this is reflected in their table 2, where 14 (74%) AHT cases have RHs, whereas zero non-inflicted cases have RHs. However, RHs are not presented in Snelling as a key input of the categorisation process. RHs are instead presented as an output of the study. Data omitted from the study, and presented here in Table 1, strongly suggests that the existence of RHs was indeed a key reason that these cases were categorised as AHT. From Table 1, there were 17 cases where an injury was actually witnessed. Seven of these had ophthalmology examination and three were found to have RHs. Reference to table 2 of Snelling et al. shows that all three were categorised as AHT, despite having been witnessed to not be AHT. Eighteen cases had reported histories of injury or a fall. Of these, 10 had ophthalmology examinations, with eight having RHs. All eight of these cases were classified as AHT in Snelling et al., despite the reported history of injury or fall. The result of categorising cases with RHs as being AHT, even if they are witnessed or reported to not be AHT, is obvious: all ‘AHT’ cases in the study will have RHs, which ‘confirms’ the belief of CPU paediatricians that RHs are associated with AHT but not with accidents. Processing a belief through circular reasoning and presenting it as an outcome of a study, as done in Snelling et al., does not transform that belief into science. Circular reasoning is bad science. It is time for Australasian researchers, referees and journals to put a stop to the use of this methodology. None declared.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call