Medical examiners here can be a jury of one

Not even the chief of the medical examiner’s office, he said, can override him.

“It is of no concern to me whether he feels it is homicide or not,” he said in a 2014 e-mail to a colleague. “I am the medical examiner on the case and it is my opinion that gets recorded in the death certificate.”

A Globe review of Cummings’s changed decision, and two subsequent retractions of shaken-baby rulings by other medical examiners, found a highly decentralized system of ruling on suspicious deaths in Massachusetts, in which forensic pathologists are given extraordinary freedom to make — and change — their rulings, with little scrutiny of what factors, including personal ones, may have influenced them.

In this extraordinary series of revisions, which occurred over the course of 18 months, the examiners, all physicians, became powerful de-facto judges, wielding tremendous pretrial clout while operating out of public view and without fear of being overruled, according to a Globe review of e-mails, case notes, medical records, state ethics filings, court filings, and other documents. That made them prime targets for defense attorneys pursuing a novel strategy, focusing first on introducing doubt in the minds of these medical examiners — not jurors at trial — to help free their clients.

The lack of any kind of mandatory supervisory review of the examiners’ rulings also meant there was little opportunity to probe whether other factors, including possible conflicts of interest, could have played a role.

In Cummings’s case, for instance, a review might have uncovered the fact that, while he was analyzing the Malden infant’s death, he had months earlier finished work as a paid defense expert in a shaken-baby case on the West Coast as part of the private practice that he was operating on the side. Soon after changing his decision in the death of the Malden infant, Nathan Wilson, he would leave government work and expand the business, devoting himself almost exclusively to consulting work for defense lawyers on shaken-baby cases.

Nearly a year after Cummings’s revision, Dr. Katherine Lindstrom, another state pathologist, would retract her shaken-baby homicide ruling in the case of a 1-year-old Cambridge girl after an unusual level of pretrial lobbying by the defense, according to e-mails obtained by the Globe through a public records request. And Dr. Anna McDonald would follow suit on a third case soon after, a full year after she had stopped working as a Massachusetts medical examiner and begun working for a North Carolina doctor who frequently testifies for the defense in shaken-baby cases. Despite her job change and relocation, she retained control of the case, which the agency allows.

The consequences of these revisions have been profound: Two murder prosecutions were dropped, including the high-profile prosecution of a Cambridge nanny, and a third homicide case against a Burlington baby sitter has been seriously weakened.

Concentrating so much unchecked power in individual forensic pathologists is something many major medical examiner’s offices try hard to avoid, the Globe review found.

David Fowler, president of the National Association of Medical Examiners, said his accreditation organization calls for each autopsy report to undergo a mandatory “quality assurance” review process.

In Maryland, where he is the chief medical examiner, Fowler said he is required to personally review and sign off on all death certificates involving a homicide, a child under age 2, and any rulings that call the cause of death “undetermined.” In New York City, which has its own medical examiner’s office, each homicide determination must be initialed by a second examiner, according to a spokeswoman.

Cummings, 44, the only one of the three medical examiners who spoke to the Globe, defended his decision to retract his shaken-baby finding as an honorable response to fresh doubts based on new information from the defense. Lindstrom and McDonald each gave similar reasons when explaining their changed decisions to prosecutors, records show.

Cummings said he prides himself on being an independent thinker and hopes, as he pursues his private practice, to be a “reasonable voice” in the highly emotional debate around whether shaken-baby syndrome is wrongfully applied to explain some infant deaths. The major signs of this form of head trauma — marked by a combination of brain bleeding, brain swelling, and retinal damage — can have nonviolent causes, he said.

He said he has also started a nonprofit dedicated to helping indigent clients facing shaken-baby charges, and has recruited McDonald, whom he used to mentor while working for the state, to join him in this effort.

Some in the legal and medical community hailed these three retractions as a sign of professional courage, a willingness by doctors to admit mistakes. However, many in the state’s child-protection community sounded an alarm.

Officials from the Massachusetts chapter of the American Academy of Pediatrics called on the governor to investigate, saying these abrupt shifts in such critical cases make the state seem arbitrary and incompetent when it comes to identifying abusive head trauma, which they say kills or maims some 1,000 children a year nationwide.

They suggested a cottage industry of well-paid defense experts, promoting marginal medical theories, may be undermining valid prosecutions.

“Sadly these extraordinary and alarming events call into question both the capacity and independence of our medical examiner’s office,” they wrote in a letter to Governor Charlie Baker this spring.

The chief medical examiner, Dr. Henry Nields, 62, who oversees 12 full-time forensic pathologists, declined the Globe’s request for an interview about the string of retractions and his management of the office.

During his decadelong tenure, his office has been plagued with low staffing and huge case backlogs, two factors that have led the National Association of Medical Examiners to give it only a “provisional” accreditation.

Public Safety Secretary Daniel Bennett, who oversees the medical examiner’s office, acknowledged that three retractions in such a short time is unprecedented, but he said he trusts that these medical examiners acted professionally. He said the office also allows its medical staff to pursue consulting practices on the side, provided the work does not conflict with Massachusetts cases.

Medical examiners typically discuss cases with their supervisors and colleagues, he said, but he does not believe additional oversight is necessary.

“The medical examiner who does the autopsy is the one who makes the final decision in every case,” Bennett said.

Cummings had just started his $150,000-a-year full-time state job when the body of 6-month-old Nathan Wilson was brought to the Boston morgue.

The child’s father, Geoffrey Wilson, an administrator in the MIT Media Lab, had been caring for him the morning of March 7, 2010, a Sunday, while his wife was at church. When his wife returned to their Malden home, she found the baby in his crib, struggling to breathe.

Nathan’s parents then drove their only child to the emergency room at Boston Children’s Hospital, where doctors found the infant unresponsive. He had multiple bruises on his head and body, bleeding on the left side of his brain, and bleeding in both eyes, records show.

Two months later, after conducting an autopsy and receiving test results, Cummings testified that Nathan had died of a violent act, not natural causes.

“There’s an injury to the head that is both blunt, which means something hit it, and shaken,” Cummings told a grand jury. The baby’s father was charged with murder, even as he insisted he was innocent and would never harm the baby he cherished.

But after issuing his homicide ruling, Cummings apparently began wondering if this child abuse diagnosis, while valid in many cases, was over-applied in some.

In a blog posting in December 2010, he referred to the “dogma” of this diagnosis and questioned whether the typical triad of symptoms always points to abuse.

“What once was the truth became a little foggy to me!” he wrote.

In the months to come, Cummings’s doubts would grow as Wilson’s defense lawyers, led by J.W. Carney Jr., recruited medical experts to review Nathan’s death.

One was Dr. Elizabeth Laposata, a former chief medical examiner in Rhode Island. She met with Cummings and suggested Nathan may have had Ehlers-Danlos syndrome, a rare genetic disorder that in its most severe form can cause damage to blood vessels and be deadly.

Although the baby was never confirmed to have this disorder, the child’s mother and maternal grandmother did — and this fact appeared to play a role in Cummings’s thinking as he cited the family’s medical history in his revision.

But hospital records obtained by the Globe show that the mother after extensive testing was found to have a less serious form of the illness that is commonly marked by excessively flexible joints and loose skin, but is almost never fatal.

Cummings — who agreed to one interview with the Globe, then did not respond to subsequent requests for comment — said that he had an “extended dialogue over months” with the defense in 2013, over the phone and in personal meetings, prior to his new ruling.

Carney told the Globe that such open exchange of information is a positive thing — and needed. He said medical examiners, who are supposed to be independent, are typically too close to prosecutors, and defense attorneys often are “just trying to balance the access.”

Cummings said Carney is a forceful advocate — one, he indicated in the interview, he dreaded facing on the witness stand. But his changed ruling was based entirely on medical evidence, he maintained, adding that he now sees himself as part of a more open-minded younger generation of forensic pathologists.

On Sept, 24, 2013, shortly after one of his meetings with the defense, Cummings e-mailed the prosecutor on the case, Katharine Folger.

“Hey, I’ve been thinking about this case. I want to change it to undetermined. I am interested to hear what you think tomorrow. I won’t do anything until we talk,” he wrote.

Around this time, as part of a private business that he disclosed on state ethics forms, Cummings had recently finished serving as a paid consultant on a shaken-baby case in the San Francisco area, involving a father facing homicide charges.

In a turn that what would foreshadow events in Massachusetts, prosecutors there dropped the case when a California medical examiner revised the manner of death to “undetermined” after hearing from Cummings and five other experts, said defense attorney Stuart Hanlon.

In the late summer of 2014, about a year after first sharing his doubts with prosecutors, Cummings revised Nathan’s death certificate. Though the reason for the delay is unclear, some e-mails suggest Folger thought Nields, the chief medical examiner, was sympathetic to the prosecution’s homicide stance and was urging him to intervene.

That possibility upset Cummings. He insisted to a colleague that he alone was in charge of the case — and even complained about being “bullied” by prosecutors.

Soon after, Middlesex District Attorney Marian Ryan dropped murder charges against Geoffrey Wilson. A finding of an “undetermined” cause of death doesn’t rule out the possibility of homicide, but she didn’t believe jurors, after hearing testimony from a wavering medical examiner, could be persuaded to reach such a verdict.

Within the year, Cummings would leave his state job and expand his private consulting practice, a field in which experts typically receive anywhere from $200 to $500 an hour.

Cummings told the Globe that he is committed to advancing accurate knowledge about abusive head trauma. When he reviews cases for the defense, he said, he often sees evidence of abuse and tells them so. But he also lets them know when a shaken-baby diagnosis is not clear-cut.

“I don’t do this for the money,” he said.

While Cummings was embroiled in his case, his colleague, Dr. Katherine Lindstrom, 38, whom he had helped recruit to the office, was in the middle of a controversy of her own.

Her case, which drew international media attention, involved a nanny from Ireland: Aisling Brady McCarthy, who was accused of killing a 1-year-old girl, Rehma Sabir, in her family’s Harvard Square apartment.

On Jan. 14, 2013, Rehma was taken by ambulance to Boston Children’s Hospital, unconscious and suffering from brain swelling, brain bleeding, and retinal damage. Doctors also found head bruises and healing bone fractures in her spine and limbs.

The nanny was the sole caretaker when Rehma was found unconscious in her crib. The child’s grandmother, arriving for a visit, was the first to discover something was wrong.

Police also found bloody wipes in a wastebasket and blood on the child’s pillow and clothes. There was also a chunk of drywall missing where the wall met the corner of the changing table.

In April 2013, after completing her autopsy report, Lindstrom told a grand jury that Rehma was the victim of a homicide from a combination of blunt force trauma and excessive shaking. In these cases, doctors are generally required to testify that they believe their findings “to a reasonable degree of medical certainty.”

Meanwhile, the baby’s nanny consistently asserted her innocence, saying she loved Rehma and would never harm her. Her lawyers also cited reports indicating that the baby’s bone fractures probably occurred when the child was abroad with her mother, not when McCarthy was caring for the baby.

One of the first on the defense team to reach out to Lindstrom was Dr. Shaku Teas, a forensic pathologist from Illinois who would later conclude the baby may have succumbed to some type of infectious event that triggered catastrophic brain bleeding.

“I would love to talk to you about the case but I understand if you cannot as it apparently is a high profile case in your area,” she wrote in a November 2013 e-mail, signing it “Shaku.”

The two would ultimately exchange two dozen e-mails, often related to obtaining medical information or figuring out how to meet or speak.

“Katie, thanks again for taking the time to talk to me today,” Teas wrote a month later.

By Jan. 30, 2015, one of the nanny’s defense attorneys, David Meier, a former veteran homicide prosecutor, began asking for doctor-to-doctor meetings that would be attended only by Lindstrom and defense medical experts.

“As you know, having lawyers in the middle of things sometimes prevents meaningful and productive conversations between the actual medical doctors. . .” he wrote in an e-mail to one of the lawyers in the medical examiner’s office, who had questioned his request. “I hope this helps. Again there is no hidden agenda here.”

Meier did not respond to the Globe’s requests for comment.

By May 2015, the defense team had succeeded in getting the court to approve a trial delay that would enable Lindstrom to fully review the information they had provided.

The defense ultimately offered reports from at least nine medical experts who provided a variety of alternative explanations for Rehma’s injuries, including a rare infectious event, an immune system problem, and a metabolic bone disorder. They also cited the possibility of a bleeding disorder, von Willebrand disease, which Rehma had been tested for once as a baby when she showed signs of easy bruising. The initial results were in the normal range for her blood type, however, records show, and doctors suggested retesting if concerns persisted.

At least two defense medical experts also said it was impossible to date Rehma’s prior bone fractures with any precision. Although some defense experts hedged their conclusions, a forensic pathologist from North Carolina was unequivocal.

“Rehma Sabir did not die from abusive head trauma or complications from blunt force head injuries,” wrote Dr. Patrick Lantz, who at the time oversaw a small autopsy service affiliated with the Wake Forest Baptist Medical Center in Winston-Salem.

Lantz — who had recently hired McDonald out of the Massachusetts medical examiner’s office — said Rehma probably died of complications from an infection that triggered catastrophic brain swelling.

On Aug. 31, 2015, after reviewing all the reports, Lindstrom publicly issued her retraction. She cited other possible causes, including von Willebrand disease, for Rehma’s death.

When news of her changed ruling — and the dropped murder charge — hit the Internet, Lindstrom received an e-mail from a familiar name: “Katie, Just heard — or read online. You were very courageous. I admire you. Shaku.”

Months later, while working under Lantz in North Carolina, Dr. Anna McDonald became plagued with self-doubt about one of her old Massachusetts cases, one that had traveled with her to her new job.

At one time, she had had little question about how 6-month-old Ridhima Dhekane died.

Records showed the Burlington baby arrived at Boston Children’s Hospital with serious bleeding on both sides of the brain, and retinal damage. An MRI found blood in her ventricles and compression fractures in her spine.

Ridhima had become unconscious while in the care of Pallavi Macharla, a former doctor from India who was working as a baby sitter.

Macharla told police that Ridhima had vomited up some applesauce, then began “having trouble breathing.” She later contacted the baby’s mother, who raced back from work. The mother called 911.

McDonald, then a training fellow, conducted the autopsy under the supervision of her chief, Nields. She took the job at Wake Forest before completing her report, in which she ruled that Ridhima was killed by “blunt force and shaking injuries of the head.”

The baby sitter, who has insisted she is innocent, was arrested for murder.

About nine months after her ruling, McDonald received an update from Middlesex prosecutors, saying the baby sitter’s attorney had a medical expert who said the baby had died of a heart attack and that the brain injuries could have been caused by aggressive medical efforts to revive her. The attorney and expert were Carney and Laposata, the same pair that helped persuade Cummings to retract his homicide finding.

It was around this time that McDonald began to openly re-review the case.

McDonald, 36, has declined to speak to the Globe about the case, saying she must refrain while the criminal proceeding is pending. The Globe was unable to obtain her e-mails — and possible communications with prosecutors and the defense during this time — because communications at Wake Forest are not subject to public records laws.

Her Wake Forest supervisor, Lantz, told the Globe that he played no role in influencing McDonald, one of three forensic pathologists that he oversees.

Lantz said McDonald knew he was a defense expert in the Massachusetts case involving Rehma, the Cambridge infant, only after he issued his report, and she did not consult him as she weighed her new doubts about Ridhima’s death.

“I didn’t know she was re-reviewing that case,” he said.

In March, McDonald changed the cause of Ridhima’s death to “undetermined.” She said the baby may have died of a heart attack of unknown cause and other injuries may have been caused by aggressive efforts to revive the child, all references to Laposata’s report.

Despite McDonald’s shift, Ryan, the district attorney, is still pursuing a murder prosecution in the case, confident her office can refute a heart attack as the cause of death. Among the evidence: Ridhima’s heart was “successfully transplanted into a baby who is thriving,” Ryan said.

She said the chief examiner, Nields, in an unprecedented move, has agreed to testify at trial against his former underling, saying he still believes Ridhima was killed.

The trial of the baby sitter has yet to be set, but the anticipated testimony of Nields and McDonald will probably make for riveting drama, a reflection of how medicine and law have collided in profound ways in these shaken-baby cases.

And this child abuse diagnosis is likely to be contested with even more intensity in Massachusetts courtrooms in the years to come.

In the past year, the state’s highest court vacated shaken-baby convictions of two men, saying they were denied adequate legal representation when their trial lawyers failed to offer jurors any medical testimony to challenge the government’s case.

The justices did not declare these men innocent, but determined that there was enough controversy around this diagnosis that prosecutors should not be able to secure convictions without some pushback by the defense.

Meanwhile, as the legal controversies around shaken-baby cases swirl, onetime defendants Geoffrey Wilson and Aisling Brady McCarthy are trying to move on, forever plagued with the stigma of having once been accused of murdering a baby.

And the parents of Rehma, Sameer Sabir and Nada Siddiqui, are still seeking justice for their daughter. In February, they filed a civil wrongful-death suit against McCarthy, serving her legal papers in Ireland, where she was deported because she had been working here illegally. A clerk this month issued a default ruling against McCarthy because she has not responded.

The state’s top pediatricians’ group still wants the governor to investigate the handling of shaken-baby cases within the medical examiner’s office, though Bennett, the public safety secretary, sees no reason to do so.

“I’m satisfied it’s not something about the medical examiner’s office itself that has caused those three cases to be changed from homicide to undetermined,” he said.

Patricia Wen can be reached at patricia.wen@globe.com. Follow her on Twitter @GlobePatty.