Illustration courtesy of Academic Forensic Pathology Incorporated.
When I wrote about shaken baby syndrome last winter for the magazine, a Canadian pathologist named Evan Matshes was working on intriguing new research that hadn’t yet been published. Matshes’s paper is just out, in Academic Forensic Pathology, the journal of the National Association of Medical Examiners. It opens a new avenue of inquiry for this difficult and fraught diagnosis.
In the most contested cases, shaken baby syndrome is diagnosed based on a triad of internal injuries in the brain: subdural bleeding, retinal bleeding and brain swelling. There is no evidence of impact, like a skull fracture. And there’s also no obvious sign of the kind of neck injury that comes with severe whiplash. In the absence of such external injuries, biomechanical engineers have raised doubts about whether it’s even possible to shake a baby to death. Many doctors who treat abused children, on the other hand, say that clinical observations make it clear that this does happen.
Matshes’s research shows how death from shaking could in fact occur — but not because of the traditional triad of injuries to the brain. As NPR notes: “The new findings split a lot of the difference between the warring camps on shaken baby syndrome. For supporters, there’s evidence that shaking alone can lead to a baby’s death. But it also says skeptics were right to suggest it’s not the head injury that causes death and that shaking deaths are likely rare.”
In investigating the deaths of 35 babies, Matshes did autopsies in a new way. The usual practice is to dissect only part of the spinal column. Matshes dissected the spine down through the neck and into the nerve roots. What he found was striking. He looked at the spinal columns of 12 babies whose history showed evidence of injury from hyperflexion — in other words, severe whiplash, from shaking or, for example, from a car accident. In all 12, he found bleeding in the nerve roots of the part of the spinal column called C3, C4 and C5. Matshes also dissected the spinal columns of 23 babies for whom there was not solid evidence of an injury from whiplash. (Most of the babies in this group died of SIDS, or from being smothered by an adult who was sleeping with them.) Only one baby in this group of 23 had bleeding in the same C3, C4, C5 region, and that child’s history, while inconclusive, made shaking a distinct possibility.
The C3, C4 and C5 are the part of the spinal column that controls the diaphragm. Babies depend on their diaphragms to breathe more than older children or adults. So Matshes thinks that damage to these nerve roots is fatal because by paralyzing a baby’s diaphragm, it stops the baby from breathing. This internal neck injury, previously undetected, could be the missing piece of the puzzle: the causal mechanism that the biomechanical experiments haven’t accounted for.
If Matshes proves correct about this, then his work indicates that pathologists should look beyond the traditional triad of injuries in the brain, and into the spinal column, to determine whether a baby was shaken. In other words, the current standard method of establishing that a child had been killed via shaking may not be reliable. Matshes says he’s no longer comfortable relying on subdural and retinal bleeding alone — no matter how extensive — to rule a baby’s death a homicide from shaking, given other explanations that need to be ruled out. “I just don’t know if I don’t look at the neck,” he says.
I sent Matshes’s paper to four doctors, two supporters of the traditional shaken baby diagnosis and two critics. They all said the paper, while based on a small sample, pointed to a new area worthy of more research. “It’s terribly important to direct us to look at the neck,” said Waney Squier, a pediatric neuropathologist in Britain who frequently testifies for the defense in shaken baby cases. “In terms of ideas, it’s a really interesting paper,” agreed Desmond Runyan, a professor of pediatrics on the other side of the debate, who will soon move to the University of Colorado to direct the Kempe Center for the Prevention and Treatment of Child Abuse and Neglect.
At the same time, Squier and Runyan both pointed to a methodological weakness in Matshes study: he knew the histories of the babies he autopsied when he examined their necks. His study was not “double blind” — the scientific gold standard for ensuring that a researcher doesn’t skew his findings in the direction that will confirm his hypothesis. Matshes says that pathologists doing autopsies are ethically bound to know the subject’s history. “In forensic pathology, since we can’t do randomized controlled double-blinded studies, we have to make certain concessions,” he said.
This week, Frontline, NPR and Pro Publica aired a joint investigation into prosecutions for child deaths blamed on abuse that raised questions much like the ones I raised in my article. (Here’s an online chat with the reporters that I participated in.) Runyan brought up this latest wave of coverage while we were talking about Matshes’ new paper. The latest wave of coverage, he said, “just goes to show we need better research,” he said. Amen to that.