In the Boston Globe, columnist Yvonne Abraham writes about how prosecutors in the Middlesex County, Mass., district attorney’s office withheld exculpatory evidence in the Shaken Baby Syndrome case against Irish nanny Aisling Brady McCarthy.

McCarthy was accused of killing Rehma Sabir in January 2013. She was charged after Alice Newton, a prosecution medical expert, concluded (as she did in the Wilson case) that the 1-year-old had suffered injuries, including severe bleeding in the back of the eyes, which indicated abusive head trauma, also known as shaken baby syndrome.

After McCarthy was jailed without bail, prosecutors sought the opinion of Dr. Alex Levin, an eye specialist, on whether the injuries to the baby’s eyes indicated abuse. In a series of phone calls starting in August 2013, Levin expressed hesitation about coming to that conclusion. He told prosecutors he had found less severe retinal hemorrhaging, and repeatedly raised the possibility that the baby’s injuries might have been caused by something other than abuse — an immune disorder called Job Syndrome — according to a court document . . .

Not only must exculpatory evidence be shared, it must be shared promptly.

“Immediacy is not required, but the law demands some level of promptness,” said Daniel Medwed, professor of law at Northeastern University. “More to the point . . . we should expect and demand immediate compliance.”

The best prosecutors would have shared that information right after the first phone call. But, prosecutors on the McCarthy case kept it to themselves: not only after that first phone call with Levin, but through several more over the course of a year. This even after defense attorneys, who learned of Levin’s work by happenstance, asked for it repeatedly . . .

In a September 2013 e-mail, the medical examiner told prosecutor Katharine Folger that he wanted to change his homicide finding about the Wilson baby. (He did not do so officially until Aug. 1, 2014, and complained to Folger that the Middlesex District Attorney’s office had attempted to pressure him into sticking with his original homicide finding, according to the examiner’s case notes, obtained by the Globe.)

Folger had been in on some of the calls with Levin, too. Yet, even as the Wilson case was unraveling, Folger and other prosecutors didn’t share Levin’s speculation about the immune disorder with McCarthy’s defense attorneys.

A judge has since found that they should have turned all of that information over much, much sooner. McCarthy spent more than a year in jail before she was released earlier this year.

As Abraham points out, this is the second botched “shaken baby” case in Middlesex County. Geoffrey Wilson was accused of killing his son Nathan by shaking him to death. He was cleared only after his attorneys produced evidence that the child had a genetic condition that causes veins and arteries to rupture, one of the alleged symptoms of Shaken Baby Syndrome. Here’s Abraham writing on that case last week:

Initially, forensic pathologist Peter Cummings ruled that the 2010 death of the baby, Nathan Wilson, was a homicide, the result of abusive head trauma, or shaken baby syndrome. His father, Geoffrey Wilson, was charged by Ryan’s office with the baby’s murder . . .

Cummings decided to file an amended death certificate changing the cause of death from “homicide” to “could not be determined.”

In supplemental case notes, handwritten by Cummings and placed in the baby’s closed case file the day before the murder charges were dropped, Cummings said that the DA and her staff had not wanted him to change his finding. He called the way they dealt with his office on the case “unethical and unprofessional.” He accused them of “M.E. shopping” in the hopes of getting a different opinion.

“I told them I felt bullied and at times as though I was being forced to sign the case out in a way I did not think was honest,” Cummings wrote in the case notes, obtained by the Globe.

Let’s also not forget that it was Middlesex County that put the now much disputed Shaken Baby Syndrome on the map in the 1997 prosecution of Louise Woodward, a British au pair accused of killing 8-month-old Matthew Eappen. The state’s expert witness in that case has since said that he would not give today the testimony that he gave then, explaining that we now know that a number of medical conditions can produce the symptoms that experts once claimed could only have come from shaking. It’s just another example of the problems with asking judges to be the gatekeepers of what science does and doesn’t get into the courtroom. Dozens, perhaps hundreds, of people were convicted based on expert testimony that we now know was at best grossly overstated, and at worst was simply false.

But these prosecutors didn’t just rely on bad science; they actively suppressed evidence that not only should have informed that their theories about these cases were flawed, but was ultimately the evidence that led to the accused getting freed. A just system would sanction them. If they aren’t punished, there’s little disincentive to do it again, or for other prosecutors who might be tempted to shortchange a suspect’s rights.