Doubling down on junk science to ensure shaken baby convictions
Updated: Nov 26, 2019
Earlier this month, Jesse Vinaccia was sentenced to 8.5 years jail for supposedly shaking to death his then girlfriend’s 17 week old son. The number of statements made by the expert witnesses during the trial that lacked any scientific foundation are too many to document in one blog post. Of course, at the heart of this junk science was the assertion that shaking can be diagnosed from the so called ‘triad’ of brain injuries. In her report, the Head of Forensic Pathology Services at the Victorian Institute of Forensic Medicine, Dr. Linda Iles, wrote that:
‘The triad of pathological change is thought to be produced through mechanical craniocervical trauma in infants. This is thought to be induced by shaking with or without a cranial impact.’
Dr. Jo Tully, Deputy Director of the Victorian Forensic Paediatric Medical Service at the Royal Children's and Monash Children's hospitals, wrote in her report that
‘The presence of sudden unexplained collapse, intracranial haemorrhages, retinal haemorrhages and cerebral hypoxic ischaemic injury in an infant indicates head trauma that was almost certainly caused by acceleration−deceleration and rotational forces.’
At trial, Tully stated that ‘When you have the pattern of subdural haemorrhage, retinal haemorrhage and encephalopathy seen in Kaleb, then I do not believe there is a medical controversy about that diagnosis, no.’
Never mind that one of the world’s premier medical evidence review organisations, the Swedish SBU, made an exhaustive two year review of all the literature and found that there was insufficient evidence to support such a diagnosis . Let’s just ignore that shall we? Not even mention it to the jury. No controversy here! No reasonable doubts! Those Swedes are being unreasonable! They don’t even constitute a controversy! Why mention the Swedes when we can draw on confession studies (see below) about which we have no expertise and know next to nothing! The Swedes are irrelevant and the jury does not need to hear about them! No mind that this was by far the most comprehensive review of the literature pertaining to Shaken Baby Syndrome made anywhere in the world!
But all this is covered in my previous blog post. It is just a re-run of the Joby Rowe case. Here I want to highlight another issue that demonstrates how far these experts went to ensure the conviction of Jesse Vinaccia, and how far it was from science. Because just asserting the junk science that the ’triad’ of brain injuries can be used to diagnose shaking was not enough to ensure a conviction. Even if there was evidence that shaking was the cause of the injuries to baby Kaleb (which there is not), one also needs evidence that his collapse occurred immediately following the shaking, i.e. that there was no delay between shaking and collapse. Because if there was a delay, one would not know when the infant was shaken and could not convict Jesse Vinaccia. The supposed shaking could have been done earlier by someone else.
So the experts also had to testify that the impacts of shaking are immediate, and that there is no ‘lucid period’ between shaking and collapse. This may seem strange to fans of contact sports like Aussie Rules, or anyone else who knows the slightest thing about concussion. When a person gets hit on the head, doctors will monitor them for at least 24 hours, because brain haemorrhage can be delayed. This period of delay is often called a “lucid period”, where the person feels fine for a while, before collapsing.
How do the experts know that a “lucid period” cannot also occur after shaking? What is the evidence base for that? Here is what Dr. Tully told the jury: “The most reliable evidence to date indicates that infants sustaining lethal inflicted craniocerebral trauma (severely injured infants with intracranial bleeding, brain swelling, loss of consciousness and respiratory distress) experience neurological deterioration and loss of consciousness very rapidly if not immediately following the traumatic event. In other words, the onset of the child's symptoms is almost always immediately after the episode of fatal abusive head trauma.” 
What literature and evidence base is she talking about? Let us turn to Dr. Iles, who testified that “based on the literature and based on predominantly confession evidence… there is a very short period of time between the event and the child collapsing, so we're not talking about a long period of time.” 
So it comes down to ‘confession studies’ from the literature. What does that mean? Quoting from my own peer reviewed article, ‘Confession studies draw conclusions from the fact that some people accused of shaking an infant on the basis of the SBS associated brain injuries have confessed.’  Take note that not many of the accused have confessed. The vast majority deny shaking. But a few have confessed under interrogation. So firstly, what about all the other cases where the person did not confess? Even if we believe all the confessions as valid, doesn’t that mean there could be other causes as well?
But secondly, how reliable are the confessions? This is a serious question because we know that up to 30% of people who were wrongly convicted and later found to be innocent by DNA testing had given a false confession.
To understand the value of a confession, one needs to look closely at the way it was given, at the nature of the interrogation. In my peer reviewed article, I analysed the shaken baby confession studies and, to cut a long story short, concluded that ‘current confession studies are not fit for purpose, if their purpose is to provide an evidentiary basis for SBS.’  This is because ‘none of the studies made analysis of the quality of the confessions in terms of how they were attained’ and because shaken baby cases are particularly susceptible to the three errors of interrogation methods, as outlined by interrogation expert Richard Leo in his study co-authored by Steven Drizen: ‘the misclassification error’, ‘the coercion error’ and the ‘contamination error’. Read my article for details, but the upshot is that the manner in which Shaken Baby cases are interrogated essentially makes them a perfect storm for eliciting false confessions. So we can say with a degree of certainty that some fraction of people will confess to shaking even if they are innocent. And that is all that is found by Shaken Baby confession studies: some people confessed.
But if you do not believe me (or the referees who accepted my paper), let us look at what the author of a very commonly cited confession based study on shaken baby syndrome, Matthieu Vinchon, said about such studies: ‘their scientific value should not be pushed too far however, regarding in particular the dating of trauma, since child abusers are generally unreliable record keepers, and many cases are confessed as a single episode whereas evidence suggests repeated insults.’ He also states that ‘our results were far from controlled scientific data’. And that is from a researcher who is an advocate of Shaken Baby Syndrome.
Another researcher who authored a peer reviewed confession based study, Jan Leestma, wrote that “case selection methodology employed in most [confession based] articles is inconsistent, often arbitrary, and, individually or collectively of insufficient numbers to permit robust statistical analysis, much less declarative supportable statements of how certain injuries are supposed to have occurred.” Regarding another often cited confession study, used to both justify the ‘triad’ and as evidence for immediate onset of symptoms, Leestma wrote that “The Biron and Shelton paper has many serious flaws that include selection bias, observer bias, lack of controls, failure to evaluate causal possibilities beyond shaking, and circular reasoning, to name a few.”
None of this stopped the experts from invoking confession studies as though they provide reliable evidence that collapse follows immediately after shaking, evidence beyond a reasonable doubt.
Even worse (if that is possible) is that even if the confession studies were the slightest bit reliable, they contain a number of cases where there was a delay between shaking and collapse, according to the testimony of the confessor [15,16,18,19,20]. So these particular confessors who mention delays need to be disbelieved, whilst confessors that say it was immediate are taken to be reliable beyond reproach and provide a sufficient evidence base to convict Jesse Vinaccia, and others. That is according to medical doctors who clearly have no expertise in interrogation techniques, nor the associated issues surrounding suggestibility and false confessions.
This is confirmation bias on steroids.
Added to this is a more technical scientific point: the analysis of confession based studies as pertaining to delay suffer from what is called a ‘selection effect’ or ‘selection bias’. This comes from the fact that confessions occur in so few supposed shaken baby cases. When trying to make conclusions about a whole population of cases in science, one often uses a sub-sample. The idea is that the sub-sample is chosen randomly so that it is representative of the whole population. When a sub-sample is not chosen randomly, it can result in selection bias, in which case that sub-sample cannot be used to draw conclusions about the whole population. In the confession studies, the studied sub-population consists of people who confessed. Let us assume that collapse is immediate in some, but not all, cases of shaking. The police generally accuse the person who was last alone with the child, so in cases where there was a lucid period, it is more likely that the police are accusing the wrong person and so it is less likely that the person will confess.
So by ‘selecting’ a sub-sample of only those that confess, there is a bias against cases that involve a ‘lucid period’ from being included in your sub-sample. This biased selection means that the sub-sample is not representative of the whole population. So even if all confessions were 100% reliable (which they are not), and even if 100% of those that confessed stated that there was no lucid period (which they did not), then this small sub-sample of cases is biased and it would still not provide evidence regarding whether or not lucid periods exist, nor how common they are in comparison to immediate collapse.
I am really not sure why I raised a technical science point against arguments that are so fundamentally non-scientific in a far broader sense. The experts are not concerned about science, they are concerned about convictions. The term ‘noble cause corruption’ describes situations where police officers are convinced that a suspect is guilty, but do not have enough evidence to prove it, so they plant some evidence. The evidence of the experts against Jesse Vinaccia is not corrupt. But the fact that the experts are convinced of guilt, and the fact that they are convinced that they are fighting for helpless infants, leads to what I like to call ‘noble cause bias’, which leads to confirmation bias whereby they use very, very low quality evidence to support their views on shaken baby syndrome, and ignore all evidence that contradicts their view. Tunnel vision.
In summary, claiming that there cannot be a delay between shaking and collapse has no scientifically sound evidentiary base. It is junk science. The claims rely on totally disregarding what research has shown us about interrogations and confessions, disregarding cases that do report lucid intervals, and disregarding some very basic principles of statistics when relying of a sub-set of data.
 Jesse Vinaccia case report by Dr. Linda Iles at p 26.
 Jesse Vinaccia case report by Dr. Joanne Tully at p 19.
 R. vs Vinaccia  VSC 683 at p. 306
 Lynøe N, Elinder G, Hallberg B, Roseen M, Sundgren P, Eriksson A. Insufficient evidence for ‘shaken baby syndrome’ – a systematic review. Acta Paediatr. 2017;106:1021–1027.
 e.g. Sonntag V. Acute Subdural Hematoma With a Lucid Interval JAMA 1978; 240 (21): 2284-2285
 Jesse Vinaccia case report by Dr. Joanne Tully at p 25.
 R. vs Vinaccia  VSC 683 at p. 360
 Brook, C. Is there an evidentiary basis for shaken baby syndrome? The conviction of Joby Rowe, 2019, Australian Journal of Forensic Sciences
 see e.g. https://www.innocenceproject.org/dna-exonerations-in-the-united-states/
 see ref.  at p6
 see ref.  at p5
 Leo. R. & Drizin, S. The three errors: pathways to false confession and wrongful conviction, University of San Francisco Law Research Paper No. 2012–04
 Vinchon, M. Response to Lyn∅e: questions about isolated trauma shaking and confessions, July 2017, Child s Nervous System 33(9412)
 see ref  at p2
 Leestma, J. E. Case analysis of brain-injured admittedly shaken infants: 54 cases, 1969-2001.Am J Forensic Med Pathol. 2005 Sep; 26(3): 199–212.
 Biron D, Shelton D. Perpetrator accounts in infant abusive head trauma brought about by a shaking event. Child Abuse Negl. 2005;29:1347–1358.
 Leestma J. E. Shaken baby syndrome”: do confessions by alleged perpetrators validate the concept? J Am Physicians Surg 2006;11 (1) 14- 16
 Gilliland, M. Interval duration between injury and severe symptoms in nonaccidental head trauma in infants and young children. Journal of Forensic Sciences, 1998 43, 723–725.
 Vinchon M, de Foort-Dhellemmens S, Desurmont M, Delestret I. Confessed abuse versus witnessed accidents in infants: comparison of clinical, radiological, and ophthalmological data in corroborated cases. Childs Nerv Syst. 2010;26:637–645.
 Adamsbaum C, Grabar S, Mejean N, Rey-Salmon C. Abusive head trauma: juridical admissions highlight violent and repetitive shaking. Pediatrics. 2008;126:546–555.