This February, Slate’s Emily Bazelon reported on the troubling state of shaken-baby syndrome prosecutions and investigations. In 50 to 75 percent of charged cases, according to prosecutors, the only medical evidence that a child was actually shaken is a “triad” of symptoms: subdural hemorrhage, retinal hemorrhage, and brain swelling. There is growing evidence, however, that this traditional diagnostic trio is flawed, which is particularly worrisome when it comes to criminal convictions for caretakers with no history of child abuse or violence.

In a blog post for the New York Times today, Emily describes a just-published piece of research that presents a new theory for how shaking can kill a child and raises additional diagnostic questions. In autopsies of babies believed to have been shaken, pathologist Evan Matshes found something no other researchers have previously noticed: bleeding in the nerve roots of spinal column segments in the neck. Those segments control the diaphragm, which babies use to breathe. When the spinal column is damaged in that area, the diaphragm can’t function, which explains how shaking causes brain damage and can be fatal. In the same study, Matshes looked at children who died from accidental head injuries. He found that, in the period before death, they’d developed brain swelling and bleeding disorders that led to retinal hemorrhage—in other words, there are other explanations for the symptoms that are traditionally used to conclude a child was shaken. Taken together, these findings suggest that shaking can, indeed, kill a child. They also suggest, though, that medical examiners may no longer conclude or testify that children were shaken by only looking for the traditional symptoms. Matshes' study is small and preliminary; if further research shows that his findings are correct, then the nerve roots in the neck, it seems, are key.

A recent joint investigation by Frontline, NPR, and ProPublica buttresses doubts about the medical evidence used in shaken-baby trials. The report, aired on PBS last week, found disturbing trends: a drastically underfunded death investigation system; medical examiners and coroners who fail to consult specialists in child injury and ailments; a chronic lack of standards and board certifications; and forensic pathologists who are effectively operating as law enforcement agents, rather than objective arbiters. This leads, the investigation concluded, to flawed medical testimony, too many inaccurate autopsy reports, and likely wrongful convictions.

The notion that a child could be shaken to death arose in 1971. Four decades later, these reports suggest we need better research. Even more, they suggest we should take a hard look at the faulty forensic evidence that pervades our criminal justice system. (The joint investigation goes far beyond child deaths—it suggests that the entire field of death investigation is in a woeful state.)  Perhaps this implicates our funding priorities: While prosecution offices are flush, grants and funding to study and investigate child abuse and death are lacking. Child abuse is, indeed, a horrific thing. But so are false charges and prison convictions.