Why the shaken baby controversy is now stalking death row

While his lawyers argue his conviction is based on faulty evidence, the fate of Robert Roberson hangs in the balance

Robert Roberson should have died two weeks ago. He was scheduled to be executed on October 17 for murdering his two-year-old daughter Nikki in 2002 – the day before, the Texas Board of Parole had unanimously voted against recommending clemency.
But following a dramatic last minute intervention the execution was put on hold. Now, a complicated legal wrangle is playing out, with campaigners claiming an innocent, 57-year-old man risks becoming the first person to be put to death in the US for a murder conviction tied to the diagnosis of shaken baby syndrome.
Roberson, an autistic man, has always maintained his innocence and his lawyers claim his 2003 conviction was based on faulty and now outdated medical evidence. They claim new facts have since come to light that might suggest Nikki died from complications related to severe pneumonia. However, prosecutors maintain that the toddler’s illness does not in itself disprove that she suffered injuries caused by her father.
There are few more contentious or emotive debates in medical or legal circles than that highlighted by Roberson’s case.
The US American paediatric radiologist John Caffey first coined the term “whiplash shaken infant syndrome” in the 1970s to describe a form of abuse that caused brain injury and death in infants. He had started to notice recurring patterns on the X-rays he was taking of young children – including repeated bone fracturing and bleeding in the protective layer between the brain and the skull called the dura.
Over time, doctors and the police have come to rely on a so-called “triad” of “tell-tale” symptoms to provide evidence that a child may have been abused by shaking. These are brain bleeding, brain swelling and retinal bleeding.
However, sceptics believe that prosecutors became overly-reliant on science that was inconclusive at best. Some have even questioned whether the supposedly tell-tale injuries could actually be caused by shaking.
Quite aside from the disagreement over whether shaken baby diagnoses could be flawed and lead to wrongful convictions, medical experts worry that complicated scientific theories can be oversimplified and treated as fact by lawyers attempting to persuade a jury of a defendant’s guilt or innocence.
Medical experts point out that the only way to scientifically prove whether shaking a baby causes the triad of symptoms is to conduct experiments in which babies are shaken. And that, of course, is never going to happen.
There is no doubt that child abuse occurs and that shaking a small baby can cause severe injuries. But nowadays, many healthcare professionals prefer the terms “non-accidental head injury” or “abusive head trauma” – partly because it is less emotive and partly because it encompasses a wider range of diagnoses.
“Abusive head trauma” refers to a serious brain injury suffered when a child is injured either through shaking or some other violent impact – such as being slammed against a wall or thrown to the floor. There are around 1,300 such cases reported in the US each year, according to the National Center on Shaken Baby Syndrome.
A report by the University of the West of England that was released last year found that Covid lockdown had likely contributed to a marked increase in the number of babies in the UK suffering from abusive head trauma. There was a similar increase in cases around the global financial crisis when many people lost their jobs and society was similarly under stress.
Roberson’s case dates back over two decades from when the science and the issues with seeking a conviction based on a diagnosis of shaken baby syndrome solely from the triad of symptoms were less well understood.
Such cases usually seek to explain why a previously well child has become suddenly and catastrophically ill or died. However, Nikki had pre-existing medical issues and Roberson had taken her to see doctors for treatment in the days leading up to her death.
Roberson’s lawyers and supporters – who include the crime novelist John Grisham and are campaigning to halt his execution – say this should have been taken into account at his trial and point out that there was no evidence that he had previously abused his daughter.
Scans taken at the hospital revealed the triad of symptoms. But relying just on those signs is no longer accepted as proof of abuse, according to Kate Judson of the Center for Integrity in Forensic Sciences, a nonprofit that has supported Roberson. “That’s debunked,” she told The New York Times.
At Roberson’s trial the prosecution relied on testimony from police officers about Roberson’s apparent lack of emotion after his daughter’s death, which they argued was evidence of guilt. At the time Roberson’s autism had not yet been diagnosed.
Years before his case played out, shaken baby syndrome gained huge media attention in the UK following the 1997 trial of 18-year-old British nanny Louise Woodward who was accused of causing the death of eight-month-old Matthew Eappen in Massachusetts. It was a perfect illustration of how contentious such court cases could be.
Eappen’s autopsy revealed the classic triad and Woodward admitted under cross-examination that she had been “not as gentle as I might have been”. She was convicted of second degree murder.
But after hearing the defence plea for the charge to be changed to manslaughter, the judge reduced Woodward’s sentence from 15 years to life to just 279 days, which was time served and meant that she walked free. What’s more, one of the main prosecution witnesses later claimed that the triad could be caused accidentally and should not be considered conclusive proof of abuse.
A report in the US which analysed 151 cases in which defendants were exonerated between 1989 and 2023 after a false conviction found that a significant minority were in cases related to “shaken baby syndrome” or “abusive head trauma”. Some have concluded that this means shaken baby syndrome is a now-discredited theory.
However, the reality is far more complex. According to the National Center on Shaken Baby Syndrome in the US, 97 per cent of more than 1,400 convictions related to abusive head trauma from 2008 to 2018 have been upheld.
There have certainly been numerous horror stories about parents being wrongly-convicted of killing their children. But there are also examples of justice failing in the opposite direction – including in the UK.
In 2007, Ellie Butler was rushed to hospital after going white, floppy and struggling to breathe. She survived but scans revealed the triad of symptoms. Her father Ben was arrested, tried and convicted of causing grievous bodily harm.
He later appealed and was acquitted despite a paediatric neuroradiologist arguing that Ellie displayed signs of abusive shaking. In 2013, Ellie died. Her father was subsequently convicted of her murder and is now serving a life sentence.
For many years there were a number of cases in which there was no evidence of trauma other than the “triad” of injuries. Often it was asserted that the triad indicated trauma by shaking; such shaking had to be extremely violent and sustained (and therefore was evidence of malevolent intent); and, finally, that if the child became ill while in the sole charge of a carer, that carer was responsible.
However, each of these assertions has since been contested. And a landmark Court of Appeal judgement in 2005 found that “the mere presence of the ‘triad’ does not automatically or necessarily lead to a diagnosis of [non-accidental head injury] and/or a conclusion of unlawful killing”.
Medical and legal experts say that rather than relying on the triad of symptoms to prove or disprove that abuse has occurred, courts should take each case on its merits. Prosecutors now try to paint a holistic picture of why a previously healthy child has become suddenly and catastrophically ill.
“No one individual feature is diagnostic in and of itself,” says one consultant neuropathologist based in the UK. Additional evidence might, for example, include the child’s antenatal history, GP visits, X-rays and the text messages and search history of their carers.
Critics of past convictions claim that prosecutors and the expert witnesses they rely on have failed to consider whether a child’s injuries might have been caused by other incidents – such as falls from a sofa or a baby seat – or illnesses like heart disease, pneumonia or sepsis. It might also be the result of re-bleeding from birth injuries.
In 2003, a scientific paper published by the neuropathologist Dr Jennian Geddes suggested a lack of oxygen, possibly caused by accidental choking, could cause all the symptoms of the triad.
For a while this paper was seized upon by defence lawyers. But under cross-examination during an appeal hearing against three shaken-baby convictions in 2005, Dr Geddes was forced to admit her theory was a hypothesis and not a fact. However, she also pointed out that the long-held view of the triad was also merely a hypothesis.
In Roberson’s case, lawyers say their client took his daughter to hospital after she fell out of bed having been ill all week. Evidence gathered since his conviction suggests Nikki might have been suffering from pneumonia, which developed into sepsis.
They also claim it’s possible she was erroneously prescribed medication that made it harder for her to breath. At a 2022 hearing to consider new evidence in the case, a judge rejected these theories.
The Republican-dominated Texas House of Representatives is now calling for Roberson’s execution to be halted. The detective who worked on the original murder case says he now believes Roberson is an innocent man.
A number of courts in the US have overturned convictions and district attorneys have dropped charges in other cases relating to shaken baby syndrome in recent years but so far those in Texas have been unmoved.
Roberson’s execution was halted once before in 2016 when his case went before the Texas Court of Criminal Appeals – the state’s highest criminal court. However, that hearing ultimately did not lead to a reversal of his conviction or an altering of his sentence.
His most recent execution date was only avoided when politicians stepped in and subpoenaed Roberson, demanding that the condemned man testify before a committee at the Texas Capitol at a date set for after his execution.
The state’s attorney general has said the case has already been litigated through the justice system. The Supreme Court of Texas must now decide whether that hearing goes ahead. A man’s life hangs in the balance.