This article is a collaboration between The Atlantic and The Marshall Project.

“My baby is not breathing,” Josue Santiago told the Racine, Wisconsin, 911 operator. “Oh my God, man, please send somebody. Please. Elihu. Elihu,” he said, repeating his son’s name.

Until moments before the call, April 11, 2017, had been unremarkable in Santiago’s household. He got home from his night shift at a foundry and took over child care from his girlfriend, the baby’s mom, Glendalyz Galarza, as she went to work. He spent the day feeding and playing with Elihu, walking the dogs, drinking coffee, and texting with Galarza. Take the meat out of the freezer, she reminded him. Santiago worried about Elihu, though, texting Galarza that the baby was sleeping a lot and wasn’t hungry. In the afternoon, he took a photo of Elihu, and remembers thinking the four-month-old didn’t look like himself, looked too serious.

Then, just before 5 p.m., Santiago flopped down on his and Galarza’s bed to sleep and pulled Elihu’s bassinet close. “I lay down,” he told me when we spoke last year. “Something told me, Get up.” His voice slowed and lowered. “And when I get up and look at him, he’s pale. His lips were purple.” As he talked, Santiago clasped and unclasped his hands, his eyes wet. “He was shaking.”

Santiago picked him up, ran downstairs, and splashed water on the baby’s face: no response. He called 911, and an ambulance sped Elihu to a hospital.

Several months later and one state over, in a quiet Illinois suburb, Molly Hayes, a dentist, and Daniel Namie, an engineering salesman, brought their three-month-old son, Alex, home after four weeks in the hospital.

They’d thought they were being overly cautious when they first brought him to the ER, after he started wheezing and seemed to swallow milk wrong. Doctors didn’t know what the problem was, but it was getting worse.

Hayes and Namie were praying in a waiting room when doctors stepped out of Alex’s hospital room. The baby’s heart had stopped, they said. Hayes collapsed to the ground. Namie made himself enter the room, a priest accompanying him. “I thought that was it,” Namie told me. “I kissed him on the forehead, and I was so happy to feel that he was still warm when I kissed him.” The priest quietly baptized the baby with a syringe of water. Namie stumbled out and joined his wife on the ground in the waiting room, praying.

A few minutes later, the doctors came out. Hayes thought they were going to say the time of death. They didn’t. Alex’s heart was beating, slowly.

Doctors transferred Alex to Lurie Children’s Hospital in Chicago, where he underwent a battery of treatments. About a month later, on October 20, Alex was stable enough to go home, although with a feeding tube, an oxygen machine, and instructions for how to monitor his oxygen levels. Hayes had recently gone back to work at the dental practice, while Namie had quit his job to care for Alex. But on October 26, Alex’s oxygen levels plummeted, and he was rushed to a local hospital.

Within hours of bringing their children to the hospitals, each family’s life would change. Both sets of parents would eventually lose custody of their child. One parent would be jailed. One baby would live. One baby would die. And both children’s cases would turn on the diagnosis of a child-abuse pediatrician, an increasingly powerful medical specialty. These doctors are trained in diagnosing child abuse, in writing reports meant to hold up in court, and in providing testimony on behalf of state prosecutors. Many of their salaries are paid, in part, by the child-welfare departments charged with separating parents and children. The doctors’ opinions can be subjective and powerful, even overruling other specialists’. But none of the parents knew that child-abuse pediatricians existed—not even as they talked with them, unwittingly sharing information that became a part of a case against them.

“Every word we used showed up in court,” Hayes told me.

Child abuse is a pervasive and complex problem: Few children or abusers report harm themselves, so it’s up to other authorities, such as medical personnel and teachers, or bystanders, such as neighbors, to report suspected abuse. Some 3.5 million children in America were reported as being suspected victims of child abuse in 2018, the most recent year for which data are available; about 680,000 were ultimately determined by authorities to have been abused or neglected.

When suspected abuse has a medical component, like an injury or a bruise, child-abuse pediatricians step in to examine injuries, rule out causes other than abuse, such as a disease, and consider the family’s explanations. They are trained in what types of fractures generally stem from abuse, and what bleeding patterns in the brain can be caused by shaking. With an abuse expert on staff, the thinking goes, regular pediatricians don’t need to worry about overreacting to an innocent bruise, or missing warning signs because parents are convincing liars. Child-welfare workers also rely on this expertise as they consider removing children from their homes. The American Board of Pediatrics certified the first group of child-abuse pediatricians in 2009; there are now 344 such specialists nationally, stationed in all but three states.

A review of dozens of cases, including thousands of pages of medical records, child-welfare agencies’ records, and testimony, along with court decisions, contracts, and emails from child-abuse pediatricians, shows that these doctors can have near-unilateral power in labeling abuse—even though their conclusions are sometimes at odds with the opinions of specialists like orthopedists and hematologists. Their judgments are echoed, amplified, and often unblinkingly accepted by investigators. Indeed, instances in which medical professionals make reports to child-welfare departments are 40 percent more likely to be substantiated—meaning the agencies found that abuse occurred—than reports by nonmedical professionals, according to a Marshall Project analysis of the National Data Archive on Child Abuse and Neglect.

But child-abuse pediatricians may have a conflicted perspective. Many are paid in part by child-welfare departments and work directly with state lawyers in cases where the state is removing children from homes, and end up shaping arguments against parents, testifying in court, and working within a system that parents don’t understand is stacked against them. This is particularly problematic because child-removal cases play out in family court, where the state’s burden of proof is low and parents have limited legal rights.

“The first time I started to hear what parents are going through, I was shocked,” Maxine Eichner, a family-law professor at the University of North Carolina School of Law, told me. “There should be a role for child-abuse pediatricians,” she said, but many are “going well beyond their medical expertise and wielding the power of the diagnosis in ways that are really harmful for children and families.”

Molly Hayes outside her house. She and Daniel Namie lost custody of their children for more than six months after a child-abuse pediatrician found what he considered likely signs of abuse in their baby boy. (Lawrence Agyei)

After Alex’s oxygen levels plummeted and he was rushed to the hospital, Hayes and Namie were anxious for updates. His earlier condition had been diagnosed as acute respiratory distress syndrome; Lurie staff said that a number of things could have caused it, such as aspiration pneumonia, meaning that milk had built up in his lungs rather than going to his stomach. Hayes and Namie were prepared for another round of trading off shifts at the hospital with Alex and at home with their 19-month-old, Mara. Then an X-ray showed that Alex had clavicle and rib fractures. An earlier X-ray from Lurie, however, had shown no fractures. “I was seriously so unconcerned with this. I was like, ‘What is going on with his lungs?’” Hayes said. “I’m so naive. I’m a mandated reporter. I should have known.”

Mandated reporter: As a dentist, Hayes, like any medical professional, is required to call a child-welfare hotline if she has any suspicion of child abuse.

The doctor transferred Alex back to Lurie, where a social worker told Namie that a child-welfare case had been opened. When Namie told Hayes, “I was like, ‘Good. I want to talk to somebody about this,’” she said, still focused on Alex’s lungs and thinking maybe an incident of rough handling during the earlier hospital stay had caused the fractures.

According to Hayes and Namie, nurses, a hospital social worker, and a child-welfare caseworker told them that a pediatrician named Dr. Narang would get to the bottom of what was wrong with Alex. Soon after that, Narang spoke with Namie. Namie assumed that Narang was treating his son, and answered Narang’s questions about Alex’s history and their family.

That evening, the caseworker told Hayes and Namie that they couldn’t be at home unsupervised with Mara, because there were concerns about her safety. They hastily arranged for Mara to spend the night with a friend of theirs.

The next morning, Narang talked with Hayes alone. “We were so honest—I said something about Alex being unplanned, and I didn’t mean it like that,” Hayes told me. Soon after, Narang told Hayes and Namie that the hospital had found more fractures.

Sandeep Narang, a child-abuse pediatrician who started out as a lawyer, has given lectures nationwide on being an expert witness. (A Lurie spokesperson, as well as lawyers for Narang and the hospital, declined to comment for this article.) He received Alex’s case when the local doctor said “it was beyond his expertise to speculate on any cause of the fractures,” according to case notes, and Lurie ER doctors were concerned about abuse.

Other investigators would repeatedly defer to Narang’s knowledge, records show, interpreting his opinions, even when he hedged them, as hard facts. Before Narang had drawn a firm conclusion about the cause of Alex’s fractures, the child-welfare agency’s notes show a caseworker reporting that the doctor thought “the injuries were ‘indicative’ of abuse.” When Narang requested that the agency hold off on assuming protective custody of the children until more test results came in, the agency agreed—but then asked the parents to place Mara with the family friend, because the agency thought parental care was “unsafe,” per case notes.

Narang thought that the fractures looked fairly new—meaning they likely had occurred when Alex was at home. By October 30, he’d told child-welfare workers that “the team is still leaning towards non-accidental trauma,” a term for abuse, but he needed other tests. The caseworkers didn’t wait, asking that the parents agree not to see Alex and Mara without supervision, because the fractures “appear[ed] suscipious [sic] for abuse.”

Scrambling, the couple asked Namie’s parents to take the children. The child-welfare agency, without entering the couple’s house, had determined it “injurious,” so they rented an apartment for the children and their grandparents to stay in.

“You’re guilty before proven innocent in this stuff, and it’s awful,” Namie told me. Indeed, even before the child-welfare investigator got to the hospital, the agency opened a case against the parents with this note: “**ACTION NEEDED** Allegation 9; Child at hospital with perpetrators present.”

In March 2018, Narang, with all the requested tests back, gave his final opinion: “Most probably the result of trauma, with non-accidental trauma being the most probable subset.”

Alex and Mara’s room (Lawrence Agyei)

Worried about missing potential abuse, child-abuse pediatricians screen for a wide variety of physical issues, along with, of course, emotional and sexual abuse. At two Chicago hospitals, a contract requires child-abuse teams to screen any child under 3 who presents with a wide range of issues, from burns to bruises. In Florida, children 5 and younger who have “bruises anywhere” must be assessed. A presentation on the University of Utah School of Medicine’s website advises that children may be at risk for abuse if their parents have “young age, low education, single parenthood, large number of dependent children, low income.”

A family’s background is an important factor in whether a child is screened for abuse. To be certified, child-abuse pediatricians must “understand the influence of caregiver characteristics,” such as young parental age and military service, on abuse risk, along with “family poverty” and “family race and ethnicity,” according to an American Board of Pediatrics’ examination guide for the specialty obtained by The Marshall Project.

Doctors overdiagnose abuse in children they perceive as being lower-income or nonwhite. In a 2017 study, researchers gave child-abuse pediatricians cases of potential abuse with certain socioeconomic cues about the victims’ families, such as unemployed caregivers. When researchers reversed those cues—for example, by telling the doctors the caregivers were professionals—they found that diagnostic decisions changed in 40 percent of cases. A 2002 study showed that hospitals are more likely to report Black, Hispanic, and Native children for potentially abusive fractures, while other studies show that lower social class leads to more screening for abuse.

Cases referred to child-welfare agencies by medical professionals are much more likely to be substantiated than those referred by almost all other reporters, including teachers and social workers. Though the National Data Archive on Child Abuse and Neglect does not track referrals from child-abuse pediatricians specifically, an analysis by The Marshall Project found that from 2009, when the first specialists were certified, to 2018, the most recent year for which data are available, the number of reports by medical professionals grew by 55 percent, twice as fast as the growth of all reports during that time. In 2018, child-welfare agencies substantiated one in four reports by medical professionals, versus one in six reports by all other sources.

Once a child-abuse pediatrician diagnoses abuse, that assessment can take on a life of its own in family court.

In 1970, when Eli Newberger was a first-year pediatric resident at Children’s Hospital in Boston, child abuse had become a growing concern after an academic paper on the topic made waves, and every state passed a mandatory-reporter law. A hospital social worker asked Newberger to assess a 4-year-old boy who had been discharged from the hospital the week before with a fractured femur and referred to the Massachusetts child-welfare agency for possible abuse. The agency had relayed that the child might be at risk of being reinjured.

“I said, ‘Sure, it sounds serious. We could do this today if you like,’” Newberger told me. The social worker checked with the agency, which said the assessment could wait until the following week. Not knowing any better, Newberger said, he agreed to hold off. On Monday, the social worker paged him and, crying, told him that the child had died of widely disseminated scald burns over the weekend.

Later, Newberger would learn that those burn patterns likely meant a plunge into scalding bathwater. “If only I had insisted: ‘Call the police. Bring this child in.’ But, of course, I didn’t know what I didn’t know,” Newberger said.

That day, he retrieved hospital records for all the children whom the hospital had reported to the child-welfare agency in the past year. “To my horror,” he said, four of the 39 had returned with new injuries. Newberger alerted the hospital’s chief physician, who asked Newberger to find out how other hospitals handled suspected abuse cases.

Eli Newberger, a pioneer in the child-abuse-pediatrics field, worried that making child-abuse pediatrics a subspecialty might not ultimately help children and families. (Tony Luong)

Newberger learned that the few hospitals handling suspected abuse formally brought in pediatricians, social workers, or child-welfare representatives to review cases and make safe discharge plans. Newberger formed a team at Children’s to do so. In 1973, he published a study in the influential journal Pediatrics describing how the group had reduced the reinjury rate from 10 percent to just over 1 percent, while cutting hospital costs. Newberger was suddenly a pioneer in the child-abuse field, advising doctors and officials around the country.

By the 1990s, a group of pediatricians focused on child abuse was considering whether child abuse should be a medical subspecialty, requiring additional training and an exam.

That was a bad idea, Newberger thought. He’d become uncomfortable with child-abuse investigations, whose focus “transformed into a very much criminalized approach—less, primarily, on understanding and help than on the developing of an evidentiary base for prosecution and pursuing perpetrators,” he said. Newberger thought making child abuse a subspecialty would move the focus further in that direction, and wouldn’t necessarily help children or their families. He was outvoted.

Newberger, who is 79, closed his office at Children’s in 1999 but remained on staff until 2019. He now serves as an expert witness, analyzing cases for defense teams, “consulting in cases where families were badly treated and mistakes in diagnoses were made,” he said, and also for prosecutors and plaintiffs, when he believes abuse did occur. He said he gets a call from a parent’s lawyer requesting help about once a week.

The idea that the state should take custody of children is centuries old: The Elizabethan Poor Law of 1601 allowed English authorities to take poor children from their parents if they felt parents were unable to “keepe and maintaine theire Children.” This doctrine, called parens patriae—parent of the country—was established in the American legal system via an 1839 Pennsylvania Supreme Court decision. “May not the natural parents, when unequal to the task of education, or unworthy of it, be superseded by the parens patriae?” the court wrote.

Courts elsewhere delivered their own parens patriae decisions, while reformers created institutions for children whose parents were deemed inadequate. The institutions handled mainly poor and immigrant children removed from their homes, and inculcated Protestant, middle-class values. Those efforts became the basis of the modern child-welfare system, and of family courts.

The first juvenile court in the United States was established in 1899. The court setup was communal, based on the parens patriae idea. Through the 1970s, the “approach was to bring the family, the social worker into chambers and say, ‘We’re not here to accuse anybody of anything,’” says Richard Krugman, a pediatrician who treated abuse cases during that time. Many parents agreed to attend voluntary classes or accepted support from child-welfare services, and the child usually stayed with the family.

Under the Reagan administration, however, an attorney general directed prosecutors to charge and try abuse cases. And in 1997, under the Clinton administration, the Adoption and Safe Families Act shifted child-welfare agencies’ priorities from letting children stay in foster care—and ultimately reunifying families—to terminating parental rights and encouraging adoptions.

Today, parents in child-welfare cases don’t have the protections that defendants in criminal cases have, because their cases are heard in family courts, where the burden of proof is low: States must show “clear and convincing evidence” of abuse—or, in some jurisdictions, a preponderance of evidence, meaning, basically, that abuse or neglect is more likely than not. Some states have no requirement that low-income parents be provided with a free lawyer if they can’t afford one. It can therefore be hard to put on a vigorous defense case, says Judge John J. Romero Jr., the children’s-court division judge in Albuquerque, New Mexico, and a former president of the National Council of Juvenile and Family Court Judges. In a recent abuse case, Romero says, one defense expert charged $300 an hour, and court administration would cover only $90.

Even if a parent can mount an aggressive defense, doing so might antagonize the child-welfare worker deciding whether to reunite her with her child. “If you put up a fight, it’s going to look worse,” says Aaron Goldstein, the head of the civil division of the Cook County, Illinois, public-defender’s office.

Child-abuse pediatricians are not required to identify themselves as such, or to notify parents or caregivers that they’re being interviewed as part of a potential child-abuse case. Parents’ flustered conversations might become for-the-record statements without their realizing it. Child-abuse pediatricians consider it suggestive of abuse when parents can’t explain an injury, or when their explanation changes. Unaware that they’re being formally questioned, parents may offer ideas about the cause of the injury, and the pediatrician may see this as a shifting account.

Once a case is referred to a child-welfare agency, the caseworker tends to accept the pediatrician’s assessment without question. In a 2019 case in Brooklyn, a mother told an ER doctor that her daughter “spilled hot noodle soup on her stomach today,” causing a burn. An off-site child-abuse pediatrician reviewed photos and decided the child’s burns were “most consistent with a burn from a heated implement like an iron.”

The police were called. The mother told them that she didn’t own an iron. The girl told them “her mother made her noodles and the noodles fell down” and got on her flowered shirt. Police photographs from the apartment show a Cup Noodles container and noodles in the garbage, and a stained flowered tunic. No iron was found. Police did not pursue the case. But the child-welfare caseworker, guided by the opinion of the child-abuse pediatrician, did. After the girl’s father left his job in another city to be with her, and her mother took months of parenting and anger-management classes, the agency agreed to dismiss the case. “Everyone gets to back away quietly from the crisis situation of ‘Are we going to have a screaming match about how absurd this call that this doctor made was?’” says Emma Alpert of Brooklyn Defender Services, who specializes in medically complex cases and worked on this case.

In a 2013 Brooklyn case, a child-abuse pediatrician conducted a “cursory” record review, in a judge’s words, to diagnose child abuse of a four-month-old who presented with fractures. The specialist didn’t speak with the baby’s parents, her grandmother, her foster-care parent, her pediatrician, or doctors who had examined and treated her. At hearings, a child-welfare worker testified that the child’s guardians were “loving” and “have done all that has been asked of them,” as the judge later wrote. But the agency removed the child from her guardians’ custody based on the child-abuse pediatrician’s assessment. The judge ordered the girl home after she’d been separated from her family for a year. (The child-abuse pediatrician could not be reached for comment.)

A power imbalance between child-welfare caseworkers and child-abuse pediatricians makes it difficult for caseworkers to question the doctors. An entry-level caseworker job typically requires just a two- or four-year degree, and can pay less than $40,000 a year. Someone who second-guesses a doctor’s opinion, and a specialist working regularly with her department at that, could derail her career.

This means that a second independent investigation of the case often doesn’t take place, even though child-abuse pediatricians believe it does. “Our partners, law enforcement and child-protective services, are the ones who actually make that final decision: They’re the ones who do the investigation and talk with everyone in the family, go into the home,” Suzanne B. Haney, the chair of the American Academy of Pediatrics Council on Child Abuse and Neglect, told me.

But in reality, it doesn’t always happen that way. As a child-welfare worker testified in a 2015 Pennsylvania hearing, “We have to go based upon the statement from the medical professional.”  

I

n November 2017, a family court held a custody hearing in Alex and Mara Namie’s case. A family-court judge decided that at least for the moment, the children weren’t safe with their parents.

Family court is where states can file to take temporary or permanent custody of children from parents if abuse is suspected; prosecutors can also pursue a separate criminal case. With a judge’s approval, a child-welfare caseworker makes arrangements for foster care, parental visitation, or parenting classes. But a state or child-welfare lawyer might take it further and argue for separation, culminating in a termination-of-parental-rights proceeding, where a judge makes the final decision.

The Illinois child-welfare agency decided to pursue separation; meanwhile, Hayes and Namie tried to make life as normal as possible for their kids. They burned through their savings, paying for lawyers and the new apartment. They saw the children daily, with Namie’s parents present. Hayes felt ripped with jealousy hearing other parents’ unthinking complaints. “Friends were saying, ‘I just want a break,’ and I welled up.”

In court, Hayes watched as her statements were repeated on the stand—that Alex had been “unplanned,” that she was back at work while Alex was cared for by his father and Mara by a nanny. Then there was a caseworker’s testimony that when the department took custody of Alex and Mara, Hayes’s “affect seemed a little flat. She didn’t seem upset. She wasn’t crying.” (A spokesperson for Illinois’ child-welfare agency did not respond to requests for comment.)

Sandeep Narang, a child-abuse pediatrician (Janice B. Terry)

Throughout a deposition and a court appearance, Narang maintained that he’d ruled out explanations for Alex’s fractures aside from abuse. But the parents’ lawyers focused on a specific report in the hospital records. A kidney specialist had examined Alex to see if he might have an underlying bone disease (irregular kidney function can signal problems with bones) and noted that Alex’s earlier treatments at Lurie had put him at increased risks of fractures, and his bones “appear thin throughout.” Narang, who is not a kidney or bone specialist, had ordered follow-up tests, which ruled out most genetic bone diseases.

When Steven Peskind, Hayes’s lawyer, asked Narang about the kidney specialist’s assessment that Alex’s bones were thin, he replied, “I wasn’t convinced that that was true,” adding that he didn’t know why the specialist “felt like those [earlier treatments] were correlated to the thinning bones.”

At a later hearing, Narang adjusted a key fact. Doctors can backdate fractures based on X-rays showing when bone healing began. Narang had posited that Alex’s fractures had occurred seven to 10 days before his second admission to Lurie. But when a lawyer asked Narang if it was more common for healing to start 10 to 14 days after a fracture, Narang said, “Correct.” Where was Alex in that time frame? the lawyer asked. “He would have been back at Lurie,” Narang replied.

By March, the parents’ lawyers had hired a pediatric orthopedist to review Alex’s records. Alex did have thin bones from the earlier Lurie treatments, according to the orthopedist, so even mild handling, such as putting in an IV, could have caused the fractures. In April, that doctor testified that he believed Narang had been wrong. The same month, Hayes and Namie gave away Alex’s now too-small baby clothing, much of it unworn.          

“Many people believe that science, including medical science, is a binary process,” the family-court judge said as she gave her decision in May 2018. “It’s not that easy,” she said. The case boiled down to the opinions of Narang and the orthopedist, the judge said, and she found the orthopedist more convincing, since he was a bone expert and Narang wasn’t. The judge dismissed the state’s petitions. That day, the parents brought their children home, and bought champagne to celebrate. But Hayes didn’t want it; she took her kids on a routine outing instead. “It was the first time I’d ever pushed my two kids in a Target cart,” she said.

These days, Mara, red curls in a bun, climbs in and out of her parents’ arms. Alex moves around so much, he seems motor-powered. A third child, a girl named Ruthie, was born last August. The kids’ fluorescent scribbles cover the fridge. Hayes and Namie filed a lawsuit last year against Narang, Lurie, and a child-welfare worker; on Tuesday, a judge dismissed the case against Narang and Lurie, but kept the lawsuit against the caseworker active.

“There has to be reform,” Namie said.  

Namie and Hayes with their children in suburban Illinois (Lawrence Agyei)

It was jalapeño coffee that brought Elihu’s parents together. In 2016, Josue Santiago went to a friend’s house one evening. Glendalyz Galarza, whom he’d known when they were teenagers in Chicago, was there, and they talked all night. The next morning, Santiago tried to impress her with his signature cinnamon coffee, but he accidentally used ground jalapeño. Galarza choked it down anyway. They began dating.  

Santiago had moved around a lot as a kid, living in Puerto Rico, Connecticut, Indiana, Wisconsin, and Chicago. When he eventually moved out on his own, to sleep on flour sacks at the bakery where he worked, his dad said, “‘Oh, you’re not gonna be nobody,’” Santiago recalled. In 2012, Santiago had a son, followed by two daughters. After splitting with their mother, he took the children on weekends. He got a solid job at the foundry and made extra money by tattooing friends and acquaintances.  

When Galarza got pregnant, the couple was shocked but happy. In December 2016, they had Elihu. He looked just like his mother, they thought.

On April 11, 2017, after Santiago called 911 for Elihu, Galarza came home from her bank-teller job just as an ambulance arrived. At a local hospital, doctors assessed Elihu and thought his convulsions had occurred naturally, but he needed a higher level of care; they transferred him to Children’s Hospital in Milwaukee. There, the doctors said the baby showed retinal and brain hemorrhaging—and they didn’t seem to think it had occurred naturally. Their questions seemed to focus on Santiago’s role, “pretty much like they saw him and they automatically assumed he did something to the baby,” Galarza told me. Santiago thought a lot about how he looked, as a Puerto Rican man with tattoos in a state whose population is almost entirely white; later, when a photo of him was posted online, people commented: “‘He’s MS-13, and look at the teardrops in his eyes,’” he told me. “I don’t even have fucking teardrops in my eyes. These are stars.”

Behind the scenes at Children’s, a nurse in the intensive-care unit requested an abuse consultation. Angela Rabbitt, a child-abuse pediatrician, questioned the parents and examined Elihu. She didn’t tell them she was a child-abuse pediatrician, Galarza and Santiago said. “The most common cause of this constellation of injuries … can be seen when a child is violently slammed, shaken and/or thrown,” Rabbitt wrote in a medical report, and “in the absence of a severe bleeding disorder or plausible accidental mechanism these findings are diagnostic for abusive head trauma.” She documented an unusual finding in Elihu’s blood labs, and suggested further testing to rule out congenital problems that could cause excessive bleeding.

Elihu Santiago was born in December 2016. His parents say he was a combination of serious and happy, with a strong resemblance to his mother. (Lawrence Agyei)

The night after he entered Children’s, on April 12, child-welfare workers took custody of Elihu. His parents couldn’t be alone with him anymore. When the police arrived at the hospital, instead of interviewing Galarza and Santiago, they talked to Rabbitt. She “specifically stated that this type of bleeding in the brain is not from any type of fall” and “would be consistent with a shaking motion,” the police wrote.

The morning of April 13, doctors told Galarza and Santiago that Elihu had become unresponsive. That afternoon, police questioned the couple separately for six hours at a police station. Santiago told police repeatedly that he hadn’t hurt the baby, asked if running down the stairs with him could’ve done it, said he wanted to help. Left alone in the room, Santiago prayed aloud in Spanish. “Padre, sé que somos inocentes, Padre …. ¿Qué haces?” (“Father, I know that we are innocent, Father … What are you doing?”).

By that time, Rabbitt’s analysis had become a hard fact. “It’s medical records telling us this, you know,” an investigator told Santiago. “I just talked to Dr. Rabbitt up at Children’s again, and there’s only a couple ways this happens, right?” the investigator said: car crashes, and “in head-trauma cases, where sometimes parents lose their shit.” The couple handed over their cellphones. Police photographed their house. They had Santiago reenact, with a doll, how he’d found Elihu. “My son’s dying in the hospital,” he said, his voice sounding charred. Elihu died on April 16.

Angela Rabbitt, a child-abuse pediatrician (Medical College of Wisconsin)

On April 20, Rabbitt got back tests showing that Elihu had had a Factor VII blood deficiency, a bleeding issue “not uncommon in head trauma,” she wrote, adding that she would discuss additional tests with the hematology department. No further mentions of bleeding issues or tests occur in the hospital or child-welfare records. (Rabbitt declined to comment; a Children’s spokesperson said that “medical evaluations are an important factor in the process, but the role of law enforcement and state agencies is to decide when and how to proceed on cases.”)

Santiago told various people—including investigators—that there was a history of “bad blood,” as he put it, in his family.  But the investigators “would shove it to the side,” Galarza said.

The police and child-welfare officials seemed to accept Rabbitt’s assessment without question.

On April 27, Santiago, whose record amounted to two driving-without-a-license violations, was arrested and charged with murdering his son. He would stay in jail for more than a year and a half.

Rabbitt’s opinion formed the legal basis for the arrest. “Dr. RABBITT stated that this type of bleeding is the result of a traumatic brain injury,” an investigator wrote. One of the public defenders representing Santiago questioned this: “The conclusion, then, that you made that Mr. Santiago was somehow responsible for this injury was because of Dr. Rabbitt’s opinion, correct?” Mindy Nolan asked a police investigator at a preliminary hearing.

“The information that she gave us is what led us to the conclusion that we came to,” the investigator answered.

A similar exchange occurred with child-welfare services. A caseworker met with a hospital social worker and Rabbitt, and wrote in her report: “They believe that CPS needs to get involved and figure out what actually happened to Elihu.” But instead of doing that, the child-welfare worker removed Elihu from Galarza and Santiago’s custody “per the report completed by Dr. Rabbitt.” Though the worker filed follow-up reports and conducted interviews, nothing in the records suggests she did anything with that information. (A Wisconsin child-welfare spokesperson said state law prohibited the department from commenting on specific cases.)

The medical examiner, too, relied on Rabbitt’s assessment when he ruled Elihu’s death the result of “non-accidental” head injuries.

In document after document, Wisconsin officials essentially cut-and-pasted Rabbitt’s technical findings as evidence of abuse, without context or analysis. As a detective told Santiago, “It’s science.”

Glendalyz Galarza and Josue Santiago at Galarza’s parents' house in Chicago (Lawrence Agyei)

A review of contracts and correspondence from several states shows a close relationship between child-welfare agencies and child-abuse pediatricians.

Child-abuse pediatricians can have the majority of their salary paid by child welfare. Take Houston’s UTHealth: Last year, the Texas child-welfare department covered 62 percent of the lead child-abuse pediatrician’s salary, or about $113,000, plus $24,000 in benefits, $13,000 in travel for her team, and supplies such as computer disks. Contractually, the child-abuse-pediatrics team must assess cases, testify, and send the child-welfare agency monthly progress reports.

Child-abuse pediatricians are financially dependent on contracts like these, along with academic institutions or grants, since little of their work with patients is billable. In a 2019 contract, a child-abuse pediatrician at Comer Children’s Hospital, in Chicago, got 75 percent of her salary paid by Illinois’ child-welfare agency, via an organization coordinating child-abuse pediatricians’ work. It also paid for most of her team, a total of more than half a million dollars a year. Narang, the Lurie doctor, received about $45,000 of his $233,000 salary from Illinois’ child-welfare agency via that coordinating organization in 2018.

Once they’ve diagnosed abuse, child-abuse pediatricians basically become prosecution witnesses rather than independent investigators. This is clear in e-mails obtained by The Marshall Project between Florida child-abuse pediatricians and the state’s child-welfare department: A child-welfare lawyer emails a Miami child-abuse pediatrician to get his thoughts on “medical records that I received from the parents’ attorneys.” A St. Petersburg pediatrician warns a child-welfare lawyer to look out for a doctor who’s been providing “irresponsible testimony” in child-abuse cases.

The doctors’ legal sophistication adds to their authority. In recordings of trainings for child-abuse pediatricians reviewed by The Marshall Project, doctors learn about law-school topics such as Frye and Daubert evidentiary hearings, and how certain doctors need to be to use the term reasonable degree of medical certainty in court—“all you really mean is you’re pretty sure you’re right,” according to the law professor instructing the doctors.

Child-abuse pediatricians defend their work as important and nuanced. Shalon Nienow, a child-abuse pediatrician in San Diego, told me that every consultation takes a minimum of one to two hours, and she often rules out abuse when families’ descriptions are “inconsistent” with the injury: “Sometimes people assume that a history is inconsistent, and it’s because they haven’t taken the time to ask the right questions,” she said. Haney, of the American Academy of Pediatrics Council on Child Abuse and Neglect, who is an Omaha child-abuse pediatrician, said that “we really understand the ramifications of erring either way, whether it means a child who’s returned to a risky situation, or a child who is removed from a loving home.”

Spurred by press coverage of questionable child removals, some lawmakers are thinking about ways to fix the system. In Texas, after an NBC News/Houston Chronicle series, legislators are considering asking child-welfare services to get a second medical opinion before removing children, or asking courts to appoint independent experts to evaluate medical assessments.

Another solution starts with analysis of outcomes in previous cases. In instances of missed abuse, who missed it and how? In cases where a pediatrician diagnosed abuse, what happened later? Krugman, a professor at the University of Colorado School of Medicine’s Kempe Center, a pediatric-abuse division, recently co-founded an organization to fund research on and raise awareness of child abuse.

“I can’t think of any other field,” he says, where “they practice without having any idea what the outcomes of their practice are.”

Eli Newberger also urges data-based efforts. “Doctors make mistakes all over the place, but in this area of practice, there’s no review,” he said.

In jail, his baby dead, Santiago bought sleeping pills: a few at a time, collecting enough to kill himself.

Galarza, on the outside, wasn’t faring much better. “I didn’t go upstairs at all,” where Elihu’s room was, she said. She’d drive by railroad tracks and imagine driving into an oncoming train.

One of Santiago’s older brothers, a truck driver, who lived with the couple, insisted that Galarza come with him on the road so he could keep an eye on her. “He would make a little bed in the middle and I would sleep there ’til the next day,” she said.

In jail, Santiago fell out of contact with his older children; their mother brought them to visit just once, he said. He lost joint custody and missed thousands of dollars in child-support payments, which he’d never been late on before. Unable to afford the Racine house, Galarza and Santiago’s brother gave it up, sold furniture, gave away Santiago’s dogs, and moved into an RV. Santiago’s brother asked him what to keep from his house. “I just told him, save at least my son’s clothes and his crib,” Santiago said.

In July, Santiago held Elihu’s baby blanket as Galarza looked on. He hadn’t seen the blanket since he had been in jail. (Lawrence Agyei)

After more than a year and a half in jail, Santiago was released on bond in December 2018. From the start, Mindy Nolan and his other lawyer, Erin Preston, had raised the possibility of genetic issues. But testing Galarza’s and Santiago’s blood took months. In the spring of 2019, Nolan and Preston received the results. Both parents had markers to indicate a Factor VII abnormality: They lacked a clotting agent that Elihu appeared to have lacked too. The lawyers consulted a bleeding expert. He said that Elihu had likely inherited the Factor VII deficiency, which can cause excessive bleeding and strokes.

The prosecutor took these results to the medical examiner and Rabbitt. The medical examiner changed his mind, saying he believed Elihu had had a congenital bleeding disorder that could have caused his death. Rabbitt altered her position only slightly, now saying it was “likely” that Elihu had been abused, “rather than by a reasonable degree of medical certainty,” the prosecutor wrote in a letter to the judge.

In June 2019, three weeks before trial, the prosecutor dropped the case. Judge Wynne P. Laufenberg, in granting the prosecution’s motion, advised Santiago to undergo genetic counseling or “parental counseling and education before you are to bring another child into this world” with Galarza. She advised Santiago to “express” his “gratitude” for his public defenders’ work “by never causing yourself to be charged with another felony.”

A few weeks later, Santiago and Galarza sat in her parents’ house in Humboldt Park, Chicago, where the couple now live. Her mom was in the kitchen, making chiles rellenos, and oil sizzled and popped as they talked. Santiago was unemployed; he’d been working since he was 16, and had nothing to show for it. He was living with his girlfriend’s parents. His son was gone. “I lost everything,” he said.

The year improved. Galarza went to work as a dental assistant. Santiago got an overnight shift printing labels for yogurt. They saved up for their own place. And, in December, they had a healthy baby girl. They named her Glendalyz Santiago. So she’ll have something from her mother, her father, and her brother, they gave her the middle name Elih.


Weihua Li contributed reporting.