Editor’s Note: Four days after this story went to press, the medical journal Obstetrics & Gynecology released a report confirming, as this story suggested, that problems with computerized death certificates were at least partly responsible for the reported jump in maternal mortality rates in Texas.

A little after dark, Ebonie Chandler finally had a chance to relax. She had spent the past three days trying to enroll her five-year-old daughter, Blessn, in school, only to get the runaround. It was May 20, 2017, and Ebonie and her two youngest children had been living in Houston for a month. Ebonie had lived her entire life in southern California, but when the rent on her Long Beach apartment became too expensive, she moved with Blessn and her two-year-old son, Lyse, across the country. For now, she was living at her sister Sharonna’s home, in southwest Houston, having left her two older children back in the Los Angeles area with her mother. Ebonie spent this particular day picking up after Blessn and Lyse, making them waffles, keeping them entertained with different renditions of the alphabet song. Now that it was almost Blessn’s bedtime, the two lay in Ebonie’s bed, listening to R. Kelly on Pandora.

Ebonie was exhausted, not to mention a little nauseated, because on top of all she was dealing with, she was also nineteen weeks pregnant. Two of her previous six pregnancies had ended in a miscarriage, and Ebonie was worried she might be having another. Soon before she moved to Houston, she had started experiencing bleeding, especially when she sat down. In recent days, the blood had started to soak through the pads she was wearing, and through her clothes. Sharonna had taken her to the closest emergency room three times in the past month. At the most recent visit, a week earlier, doctors had kept her in the hospital overnight to give her a blood transfusion. Beyond recommending that she find an obstetrician, which she had not yet managed to do, they did little to ensure that Ebonie would receive good care going forward.

As she lay in bed, listening to music after another tiring day, Ebonie suddenly felt something wet beneath her. She made her way to the bathroom, and Blessn followed. When she reached to turn on the light, a cramp hit her right side, and she watched in alarm as the blood collected beneath her, pooling on the floor, much of it mixed with clots. Blessn looked at the blood and then at her mother. “Tell your auntie,” Ebonie instructed her.

Sharonna called 911. An ambulance arrived quickly, and when the driver loaded Ebonie into the back, he asked if she had health insurance. This was something she was familiar with from her recent hospital visits—in California, she was covered by Medi-Cal, the state’s version of Medicaid. But did that cover her in Texas? No one she had talked to during her ER visits had given her clear information, so she didn’t have a good answer for the ambulance driver. As is often the case in such situations, the driver decided to take Ebonie to Ben Taub Hospital, Houston’s largest medical facility that focuses on treating the uninsured.

The trip to Ben Taub was a short one, about fifteen minutes, but Ebonie worried the whole time that she was having another miscarriage. Sharp cramps in her belly left her gasping. “I’ve never felt that weak,” she later said. After they arrived at the hospital, a triage nurse directed Ebonie to the obstetrics intake ward, on the third floor. There, labor and delivery nurses settled her stretcher into one of the seven slots surrounding a large central desk that dominated the room like a command center. Large green curtains separated Ebonie from her neighbors.

Lab tests confirmed that Ebonie had lost a significant amount of blood. An ultrasound showed that her nineteen-week-old fetus looked healthy but also revealed something ominous. The placenta, the large network of blood vessels that nourishes the fetus during pregnancy, had become attached to the opening of Ebonie’s birth canal, a condition known as complete placenta previa. A normal birth can’t physically happen when something so large blocks the birth canal. To make matters worse, the location of Ebonie’s placenta put her at very high risk of losing dangerous amounts of blood throughout her pregnancy. The radiologist’s written comment reflected the urgency of her condition: “Continued surveillance is necessary.”

The obstetricians ordered two bags of blood and told Ebonie they intended to keep her in the hospital for as long as a week. They also told her that terminating the pregnancy would be the safest choice for her, because it would prevent any more life-threatening bleeding. By this point, Ebonie’s blood volume was half of what it should have been, and there was every reason to believe that the bleeding would continue. Her life was in danger.

But Ebonie had never had an abortion. She couldn’t imagine life without her four children. They gave her solace and stability, just as her own siblings had when she was a young girl. Ebonie had grown up poor and, at times, homeless. She remembered days when she would fall asleep on the beaches of California, wake up early to the sound of the surf, and use the public bathrooms to wash up. These were hardly happy memories, but the presence of her brother and sister beside her during those difficult times created a bond she held as sacred. And when Ebonie began to have her own children, this new family became the most important part of her life. She and her partner weren’t married, but they were committed to raising their children, even though he had stayed in California while she lived in Texas.

After her doctors suggested the possibility of an abortion, Ebonie began to cry. She called her partner in California, who told her not to end the pregnancy. She called her mother, and she talked with her sister, who was with her in the hospital. “What should I do?” she asked. Both of them told her the same thing: “Whatever you decide, we’ll support you. No matter what.” Ebonie envisioned her own children giving one another that same type of unconditional support, and she knew then and there what she would do: she would have the baby.     

Though Ebonie, who had just turned 33, wasn’t quite halfway through her pregnancy, doctors admitted her into the labor and delivery wing of the hospital, in case the bleeding became so significant that they would need to perform an emergency cesarean to save her life. Over the next six days, the staff checked on her constantly to make sure that her bleeding was kept under control. They asked her to change the pad beneath her every eight hours, but Ebonie could barely bring herself to do it. She was exhausted. All she wanted to do was rest.

But one statistic jumped out at the researchers and readers alike: according to the available data, from 2010 to 2012 the maternal mortality rate in Texas had doubled.

The vast majority of articles in academic medical journals are only read by doctors and researchers, but occasionally one will grab the attention of the wider public. That’s what happened in September 2016, when a report in Obstetrics & Gynecology made headlines around the world. The article in question, “Recent Increases in the U.S. Maternal Mortality Rate: Disentangling Trends From Measurement Issues,” attempted to grapple with a slow but steady increase in maternal mortality rates in the United States. Most of the article was a careful consideration of the data that proved there had been an incremental climb. But one statistic jumped out at the researchers and readers alike: according to the available data, from 2010 to 2012 the maternal mortality rate in Texas had doubled, from 18.6 deaths for every 100,000 live births to 38.7. (The national average, not including Texas and California, which provided data that didn’t fit the researchers’ statistical model, shifted from about 22.4 to 23.1.) How was it possible, the researchers wondered, that the number of mothers dying within six weeks of childbirth had jumped so high and so quickly?

The data seemed so implausible that the authors cast doubt on the figures. “[I]n the absence of war, natural disaster, or severe economic upheaval, the doubling of a mortality rate within a 2-year period in a state with almost 400,000 annual births seems unlikely,” they wrote. The numbers looked so off, in fact, that researchers decided not to include them in their calculation of a U.S. maternal mortality rate. Some journalists and activists were less cautious. Headlines like “Texas’ Maternal Mortality Rate: Worst in Developed World, Shrugged off by Lawmakers,” “Why Texas Is the Most Dangerous U.S. State to Have a Baby,” and “Maternal Mortality in Texas Is a ‘National Embarrassment’ ” spread across social media.

Some journalists and activists, however, raised important policy questions. In 2011, in the midst of the spike, state laws went into effect that required women seeking abortions to jump through a number of hoops, such as submitting to an ultrasound no more than 24 hours before the procedure. To many, it seemed plausible—likely, even—that these moves to decrease the number of abortions in the state had resulted in fewer women receiving consistent medical care. Many women, they noted, find their point of access into the health care system through an abortion. Decreased access to abortions can mean decreased access to health care, which means more births and more uncontrolled sickness, which eventually leads to more deaths. Marsha Jones, the executive director of the Afiya Center, a Dallas-based reproductive rights group that focuses on black women, noted, “When you do things like making access to abortions almost impossible, the impact that’s going to have on our state’s most vulnerable population is worse and worse.”

The public outcry inspired by the study was loud enough to attract the attention of the Texas Legislature. In 2013, it had created a task force to study these deaths, but because of Texas’s stringent medical privacy laws, the review was taking longer than expected, and the group was slated to work only until September 2019. It needed more time. Texas Democrats and Republicans bickered about many issues during last year’s special session, but after some prodding from Governor Greg Abbott, the bill to extend the Maternal Mortality and Morbidity Task Force passed by a wide margin.

Women’s health advocates breathed a sigh of relief. The bill not only extended the task force’s life, it expanded its scope. Researchers were now charged with investigating the impact of socioeconomics on maternal mortality. This was particularly important because of another startling statistic that surfaced in a different 2016 study: African American women were at much higher risk than any other ethnic group. “African American women make up eleven percent of births in Texas but thirty percent of maternal deaths,” noted state representative Shawn Thierry, one of the bill’s key supporters.

Perhaps the task force, it was hoped, would manage to answer the questions that current research has left hanging: If there really are that many mothers dying in Texas, why are they dying? And if they’re not dying in such large numbers, why do we think they are?

Hospital scene
Ben Taub Hospital chief of obstetrics Dr. Carey Eppes (lower right, in white coat) and her team run a crash drill on a dummy to mitigate severe blood loss during pregnancy. Photograph by Brian Goldman

When news of Texas’s maternal mortality spike broke, I couldn’t help but think of my wife’s childbirth experience, right before Christmas 2015. While everyone in the room, including me, was doting on my newborn daughter, monitors began to show my wife’s blood pressure had dropped significantly and that her pulse was climbing. Nearly thirty minutes after having given birth, she grabbed her chest. She couldn’t breathe. Suddenly she could see only in black and white.

Even though I’m an internist and my father is an ob-gyn, I had resolved to check my medical impulses at the door during my child’s birth. I didn’t want to be the doctor dad looking over the obstetrician’s shoulder. But my wife’s cries unnerved me. So I yelled for the nurses, and then for the doctor. Once she arrived, I told her I thought my wife needed IV fluids, now. They rushed a bag of saline into her veins, and she regained color in her vision and began breathing normally again. After examining her, the doctor concluded that her blood pressure had dropped from the trauma of delivering an unusually adhesive placenta.

Thankfully, my wife recovered fully. But this close call left me with questions. What if I hadn’t been a doctor and quickly grasped the seriousness of what was happening? Could she have actually died? Are problems like this common?

With her case and the controversy over maternal mortality in mind, I decided to get a better sense of what was going on in Texas. As usually happens whenever I begin to research articles, I called one of my colleagues at Ben Taub, where I work. In this instance, I reached out to Dr. Carey Eppes, the chief of obstetrics. Eppes, who earned undergraduate degrees in molecular biology and psychology from UT-Austin and an MD from the UT Health Science Center at San Antonio, is the rare medical professional who is drawn to dilemmas that require both a sure-handed bedside manner and a fluency in statistics. Recently, she joined the publicly funded Texas Collaborative for Healthy Mothers and Babies, as a member of its executive committee and co-chair of its data subcommittee.

Eppes can, with complete authority, sound off about a whole range of topics, but she prefers to purse her lips, tilt her head forward, and home in on every word that comes at her in a way that says, You have my full attention. When I asked her what she thought about the maternal mortality numbers in the news, she paused. “I have a patient I’m really worried about right now,” she said. “Let me ask her if she’d be willing to talk with you.”

It was Ebonie’s sixth day at Ben Taub. Eppes planned to release her only if she could go 48 hours straight without any bleeding. And though Ebonie’s bleeding had slowed down considerably, she continued to experience spotting. Complicating the situation further, Ebonie was having early contractions, a sign that she could be close to delivering. At nineteen weeks into the pregnancy, though, the fetus had no chance of surviving outside her body. And yet, despite all that was going on in her life, Ebonie told Eppes she was willing to talk with me, in my sometime role as a journalist, and let me look at her chart.

I sat down beside her bed, introduced myself, and took note of her striking appearance. In addition to sporting a nose ring and numerous tattoos, Ebonie wore her hair parted straight down the middle, with one half colored red and pink and the other blue and green. Even wearing a standard-issue hospital gown, she managed to stand out.

I asked her about her life, focusing on her pregnancies. In 2002, when Ebonie was nineteen, her then boyfriend choked her until she lost consciousness, and her body began to shake uncontrollably. The next day she awoke in a hospital, and doctors told her that she had epilepsy, which was surprising to Ebonie, since she had never had a seizure before. A week later a urine test taken at a follow-up visit revealed she was two months pregnant.

Another seizure occurred a year later, after the birth of that child, when, in the midst of another argument, Ebonie’s boyfriend again grabbed her by the neck. Four years later, she had a second child with the same man and ended the relationship. Her third pregnancy, with her current boyfriend, ended in a miscarriage, as did the fifth. The fourth almost took a turn for the worse after the fetal monitor showed a dangerously low heart rate.

Ebonie felt so fortunate that both she and her new daughter had survived such a harrowing experience that she named her Blessn. She commemorated what she felt was a miracle with a tattoo on her neck that read “Blessn 4-26-12.”

About an hour into our conversation, when I asked her what it was like to be homeless as a child, Ebonie, who had been experiencing seizures throughout this pregnancy, told me she felt hot. I asked her if I could get her ice water. She closed her eyes and lost track of our conversation. “I feel another seizure coming on,” she said. Immediately, her eyes started to flutter. The heart monitor showed a rapid pulse, and the muscles of her face and neck twitched. I pushed the nurse call button and reported that Ebonie appeared to be having a seizure.

Neurologists arrived quickly and performed an EEG and administered a dose of magnesium to control the seizures. Over the next four days, the seizures stopped, as did the spotting. But there was still the problem of what Ebonie would do outside the hospital. If she had no health insurance, how would she see the right doctors?

Since Ben Taub is a safety net hospital, the staff has plenty of experience dealing with uninsured patients. A financial counselor helped Ebonie file her application for Medicaid, and Eppes and her team told Ebonie to return if she started bleeding again. They also scheduled an appointment for her eight days later with a high-risk obstetrician at one of Ben Taub’s clinics.

Ebonie went home on the last day of May, determined to stay vigilant and let the pregnancy progress. As it turned out, she didn’t have to wait long. On June 16 she awoke early to find that her bed was wet again. She awakened her son, who had been sleeping with her. “Get up,” she said. He jumped out of bed. His pajamas were soaked. Ebonie could see by the color, though, that it wasn’t blood.

Ebonie called for her sister. “I don’t know what the hell is going on, but I got fluids coming out,” she said. Her sister rushed into her room, touched the bed, and looked at her fingers.

“Do you think your water broke?” she asked.

“I don’t know,” Ebonie said.

Her lower belly was starting to cramp every five minutes. It felt like contractions to Ebonie, only stronger. The wait had ended. She had to go back to Ben Taub now. She was only a little more than 23 weeks pregnant, too early to deliver, but her life was in danger.

An ambulance arrived and the EMTs loaded Ebonie into the back. Though she told them she wanted to go to Ben Taub, the driver didn’t seem to be listening. “I’m high risk,” she told him, “and I want to deliver my baby there.”

Before Ebonie knew it, the ambulance had dropped her off at the emergency room of a different hospital in the Texas Medical Center. The driver had received notice that, due to overcrowding, Ben Taub’s ER was diverting drivers to other hospitals. It was an error the obstetrics team at Ben Taub had seen before. While the hospital’s main emergency room might reroute drivers at busy times, labor and delivery did not. “In the five years I have been here, we have never been on diversion,” said Eppes.

Patients say it’s not uncommon to end up at a different hospital from the one they’ve requested. Some of them feel that ambulance drivers make such decisions based on whether the patient is insured or not, or even the type of insurance they have. Because Texas Medicaid generally pays far less for obstetrical care than private insurance does, lower-income pregnant patients are often shuttled to hospitals that they may not have been to before. As a result, higher-risk Medicaid patients find it difficult to establish continuity with their doctors.

Ebonie’s new doctors determined that her water had broken and decided that the best course of action was to keep her hospitalized until it was safe to deliver the child. A neonatologist told Ebonie that a baby born at that point, less than 24 weeks into the pregnancy, would have only a 34 percent chance of surviving and a 90 percent chance of being born with a severe developmental problem. Ebonie’s C-section was scheduled for September 5, the start of her thirty-fourth week of pregnancy.

Unaware that she had the option to request a transfer to Ben Taub, Ebonie settled in. Over the next three weeks, nurses wrote daily goals on the whiteboard by her bed: “Stay pregnant! No seizures!” To chase away boredom, the hospital outfitted Ebonie’s room with DVDs to watch and books to read. But Ebonie’s mind was whirling so much she hardly noticed.

Once a week, Ebonie’s sister and brother (who also lived in Houston) brought Blessn and Lyse to visit. Her siblings’ schedules—one worked at Walmart, the other drove for Uber—and the expense of parking made more-frequent visits difficult. Still, Ebonie spoke with Blessn every night on the phone. Sometimes Blessn cried and pleaded for her mother to return home. “No,” Ebonie told her. “Mommy got to wait until the doctors take the baby out of her stomach.”

When Blessn’s father called from California, Ebonie explained that, even with a C-section, her health was at risk. “Are you serious?” he said. “Is this a life-and-death situation?”

“I’m dead serious.”

Some days, the solitude and the sporadic bleeding overwhelmed her, and Ebonie began to think about what might happen if things went wrong. She prayed constantly. But nothing seemed to comfort her. Her fear grew so intense that she began to question whether she should have made a different choice.

“A baby is precious, but I don’t want to die,” she said. “What will my family do if I die?”

Over the course of the twentieth century, the U.S. saw a consistent decline in the maternal mortality rate. In 1930 the rate was a startling 670 deaths per 100,000 live births, about double that of, say, the Netherlands. But a host of advances in medicine, including the development of antibiotics and better control of bleeding during labor, caused the number of deaths to drop dramatically over the following decades. By 1960 the U.S. had caught up to the Netherlands and other industrialized countries, like England and Sweden; by 1990 the number had dropped to just 12 deaths per 100,000 live births, an astonishing triumph of modern medicine. However, a series of studies in the late nineties suggested that the U.S.’s accounting methods were flawed and that we were missing a significant number of maternal deaths. Researchers worried that maybe American mothers weren’t as safe as they’d thought.

In response, in 2003 the National Center for Health Statistics advised states to amend their death certificates for women to include a question regarding pregnancy, in hopes that doing so would uncover unreported deaths. The new question asked doctors to slot female decedents into one of five categories: not pregnant within the past year; pregnant at time of death; not pregnant, but pregnant within 42 days of death; not pregnant, but pregnant 43 days to 1 year before death; or unknown if pregnant within the past year.

Over the past fifteen years, most states have added this question to their death certificates (Texas added it in 2006). By 2014, the 1990 national rate of 12 deaths per 100,000 births had doubled, to 24—an apparent confirmation that we had, indeed, been underreporting the number of mothers who were dying.

But no state showed anything like the increase that was seen in Texas between 2010 and 2012—a doubling in the space of two years. And the numbers don’t seem to have been a statistical blip. Though the state’s rate dipped in 2014, it wasn’t by much: that year there were a total of 135 maternal deaths, which reduced the rate from 38.7 to 33.8. Even the most generous statistical analysis placed Texas forty-fourth in the country in maternal deaths between 2004 and 2014; most experts put us at or near the very bottom. (It’s important, though, to remember that even though the risk of a mother dying in Texas within six weeks of childbirth is higher than in virtually any other state, in absolute terms it remains an exceedingly low 0.034 percent.)

This data prompted three questions: Why did Texas’s rate jump so much higher so fast? Why did it happen in 2011 and 2012, even though Texas had introduced the death certificate pregnancy question in 2006? And was it possible that the numbers were wrong?

One answer, I suspect, has to do with the questionable nature of the raw data researchers have been working with. One might think that a death certificate is the gold standard for recording the cause of death. In fact, it’s often a very problematic document.

A decade ago, the Texas Department of State Health Services (TDSHS) began asking doctors to file death certificates electronically. The process is cumbersome, to say the least. (Anyone who has had to sit on an examination table while the doctor clumsily types on a computer keyboard—two index fingers going up and down, up and down—can attest to how difficult it has been for the medical field to incorporate electronic reporting into its practices.) It’s also fraught with roadblocks that may deter doctors from giving these certificates the time and attention they require.

As a doctor who is often required to determine cause of death, my responsibilities begin when I receive an email from the TDSHS entitled “Death Certification.” This arrives in my inbox anywhere from the day after a patient dies to the following month, depending on how long it has taken the funeral home to fill out the demographics portion of the certificate. The patient in question might be one I knew well during a prolonged stay at Ben Taub or someone I met briefly right before he died. After answering questions about the patient, such as whether the death occurred “naturally” or as a result of suicide or homicide, I determine if tobacco contributed to the death. Yes? No? Probably? Unknown? Chronically ill patients often have multiple ongoing problems, so determining a cause of death can be challenging. Did progressive kidney problems result in a patient’s death more than heart disease did? Two highly qualified doctors might come to different conclusions, and it may not be clear that either of them is wrong.

But in Texas, the most challenging portion of filing a death certificate is logging in to the system. The Texas Electronic Registrar has restrictive password rules—one uppercase letter, one lower, two non-alphas, between eight and sixteen characters, can’t start or end with a non-alpha. The system is complicated enough—or, I suspect, buggy enough—that I’ve had to reset my password each time I log in to the system. At first I attributed this to my own technological incompetence. But when I asked my colleagues how they filled out death certificates, they all related the same frustration. Some gave up and handed off the responsibility for filling out the certificates to an administrative coordinator. “It’s not worth it,” said one colleague. Perhaps not. But if doctors, who ostensibly knew the dead patient and know the medical lingo, aren’t reliable reporters of cause of death, how can we expect administrative coordinators, who often have little or no contact with the patient, to be?

The TDSHS introduced the electronic death registrar in 2007, which was well before the big leap in Texas maternal mortality occurred. But once again, the rate of adoption may have delayed the effects. Texas doctors were slow to use the new system, preferring the old paper certificates. According to a study in Texas Medicine, as of 2010, only 63 percent of death certificates were filed electronically. But by 2012, the year we saw the jump in maternal mortality, the number had increased to 91 percent. And it seems plausible that this shift was responsible for at least some of the rise in the reported maternal mortality rate that year. Any test or question that is designed to look for something—cancer, blocked arteries, or maternal deaths—will find more of what it’s looking for, sometimes more than is actually there. A division of the Centers for Disease Control that looked closely at the deaths of women age forty and older from 2006 to 2010 nationwide found that death certificates reported four times as many pregnancy-related deaths as the evidence showed actually occurred—most of the deaths weren’t pregnancy-related at all.     

Another significant problem with death certificates in Texas is that sometimes the people who fill them out have no connection to the medical profession. In 2016, Supreme Court justice Antonin Scalia died at the Cibolo Creek Ranch, in West Texas’s Presidio County. When the sheriff was shown the justice’s body, he couldn’t declare him dead; only a doctor or medical examiner could do that—or, in Texas, a justice of the peace. Since Presidio County has no medical examiner, and no doctor or justice of the peace was available, the sheriff called the county judge, who declared Scalia dead and declined to order an autopsy. This judge, who was also responsible for filling out the death certificate, decided that Scalia had died due to chronic illness.

That was a good guess, given Scalia’s history of high blood pressure and smoking. But it was only a guess. Without an autopsy, nobody knows for sure what killed him. The judge did an admirable job of reaching out to Scalia’s doctor to fine-tune her guess, but would other officials have done the same had the decedent been less prominent? If even a Supreme Court justice receives a half-baked death certificate, how can anyone else count on getting better treatment? It’s not difficult to see how allowing people with no medical training to issue death certificates could lead to unreliable data.

Is it possible that at least part of the spike in maternal mortality rates in Texas resulted not from an increase in actual maternal deaths but from a perfect data storm brought on by the increased reportage ushered in by the pregnancy check box, the widespread adoption of an unwieldy electronic filing system, and a crazy quilt of people responsible for reporting? This seems possible, especially since we’re talking about such a small number of deaths. Because maternal mortality is such a rare occurrence, an increase of a few dozen reported deaths can make a huge difference in the maternal mortality rate. At the very least, we should be skeptical of these numbers.

Of course, it’s also possible that Texas is experiencing a genuine and alarming upsurge in maternal mortality. Even after a year and a half of debate and investigation, no one knows for sure. Such are the limitations of the death certificate data we have to work with.

If death certificates are so prone to error, then why do researchers keep referring to them? In part because they’re the only direct data we have on maternal mortality. But some researchers in Houston are dissatisfied with those limitations and came up with a work-around. Using the electronic charts and data collection that track virtually every hospital patient in the country, they decided to look at expectant mothers who were experiencing high-risk pregnancies. Even if the vast majority of these women survived, figuring out what was going on with women who could have died during their pregnancies or deliveries might be extremely valuable. And in Houston, there is no shortage of such women. 

On July 10, three weeks after doctors admitted Ebonie into her new hospital’s labor and delivery ward, she began to feel cramps in her lower belly. She called the nurse and told her she thought she was having contractions.

“Are you in labor?” the nurse asked. She lifted up the bedsheet and quickly covered Ebonie back up. “You’re in labor,” she said. “We need to call the doctor.”

Ebonie was now 25 weeks and 6 days into her pregnancy. Her C-section wasn’t scheduled for another two months, but now it looked as though the calendar was eluding her control.

The nurse went to let the doctors know. When she returned, she checked Ebonie again. She could hardly believe what she was seeing: the baby’s head was emerging. Ebonie felt a burst of pressure in her rectum.

“I need to push,” she said.

“Don’t push!” cautioned the nurse.

“Bitch, my baby is pushing his ownself out!” Ebonie later recalled saying.

What the hospital had tried desperately to prevent was happening: Ebonie was having her baby in one of the most dangerous manners possible, through a birth canal blocked by the placenta.

“Don’t push! Don’t push!” yelled the nurse.

With so much pressure, Ebonie couldn’t help it. Within minutes, the labor was over, and her newborn boy was immediately taken to the neonatal ICU, since he couldn’t breathe on his own. Ebonie had only glanced at him before he was whisked away.

As Ebonie lay there, she started to bleed again. Doctors attempted to deliver the placenta by pushing up and down on her belly for 45 minutes, to no avail. Ebonie, who was already drowsy from the pain of childbirth, began to cry. The bleeding wouldn’t stop.

The documentation in Ebonie’s chart is unclear about what occurred next. Doctors rushed her to emergency surgery to remove the placenta. “Breathe some of this air,” she remembers the anesthesiologist telling her, before everything went black.

Ebonie’s blood level, the doctors discovered, was very low. The surgeons transfused her with two bags of blood as they worked, and even as they did so, she continued to bleed. They removed the placenta and sent it to the pathology lab for analysis. The bleeding had finally stopped.

Ebonie awoke from the anesthesia with a sore throat and pain in her belly. Doctors transferred her to a recovery room, where another blood test revealed that the two bags weren’t enough. She received another two bags of blood, for a total of more than two liters post-delivery.

This was, by any measure, a traumatic delivery and surgery. Yet 48 hours later, Ebonie was discharged. The documentation, again, is unclear, but it appears that once doctors felt sure that she was no longer bleeding, they decided it was safe to send her home.

Though Ebonie’s doctors weren’t sure how she managed to deliver her child with the placenta blocking the birth canal, her attending obstetrician speculated that, over the preceding weeks, the placenta had migrated into a position that allowed a premature child to squeeze out. During her three weeks in the hospital, no ultrasound was performed to confirm the position of the placenta. Analysis of the placenta in the pathology lab offered no clues. Her doctors seemed to have sent her home with many questions hanging in the air.

One thing was clear: Ebonie had come perilously close to becoming a maternal mortality statistic.

Ebonie Chandler with her son King and his sister Blessn. Photograph by Brian Goldman

After the 2016 Obstetrics & Gynecology study focused the world’s attention on Texas’s problems, Houston Endowment, a foundation that grants over $60 million annually to local nonprofits, decided not to wait for the state’s Maternal Mortality and Morbidity Task Force to inch its way through more medical charts and another set of recommendations. Maternal mortality, the endowment decided, was a major and urgent local priority, so the group would work in parallel with the state. “We don’t know of an issue like this that’s as connected to ten other issues,” said Ann Stern, the endowment’s president and CEO.

The endowment reached out to experts in women’s health throughout the city—doctors, but also those who felt the tendrils of this issue on a daily basis, such as social workers, crisis hotline operators, elected officials, and mothers, especially those most at risk. The endowment’s Steering Committee to Reduce Maternal Mortality and affiliated groups included some of the state’s task force members, as well as one particularly forward-thinking doctor, Cecilia Ganduglia Cazaban, a health services researcher at the UT School of Public Health.

Cazaban, a native of Argentina with a master’s in public health from Harvard and a DrPH in health services research from UT, decided that, because our information on cause of death can be so unreliable, she would focus not on deaths but on near misses, which are known as severe maternal morbidity. If researchers could study why pregnant women had been admitted to the ICU or received a blood transfusion, that might give a clearer picture of why certain mothers had a higher risk of death, and how the community might be able to target their efforts to help them.

If a woman received Medicaid, her risk of being a near miss increased by 14 percent. If she was African American, it increased by 28 percent.

Last October I asked Cazaban to take me through some of the data she had on the mothers who had almost died. In her office on the ninth floor of the Reuel A. Stallones Building, she squinted briefly at the computer screen, then clicked open her spreadsheets for me. As the numbers appeared, she gave them a confident nod. The data we were looking at contained diagnoses that had been input by doctors at a patient’s discharge. These were the same diagnoses that doctors submitted for billing—a reason, perhaps, they might be entered relatively accurately. Cazaban was trying to match deliveries with diagnoses known to cause an increased risk of death: sepsis, shock, or blood transfusion, for example.

She used this method to analyze three million deliveries in Texas, including half a million in Harris County, from 2008 to 2015. What she found surprised her. The rate of near misses was climbing in both Texas and Houston, but they were much higher in the city—Texas’s near-miss rate was 2 percent; Harris County’s was 2.4 percent. A lot of that had to do with the number of chronic illnesses seen in Houston women. Cazaban identified five illnesses—hypertension, diabetes, obesity, depression, and mental illnesses other than depression—as risk factors. If a woman had just one of these illnesses, her risk of becoming a near miss shot up. And 22 percent of Houston women who delivered a child in 2015 had at least one of these conditions.

When Cazaban looked deeper into the numbers, she found other variables that were linked to near-miss deliveries. If a woman received Medicaid, her risk of being a near miss increased by 14 percent. If she was African American, it increased by 28 percent.

Cazaban and the members of the working group decided that the best way to wield this data into results was to pinpoint where the near misses lived in Houston. She constructed a map of the city showing the zip codes where the women most at risk resided—the deeper the red, the higher the risk. One of the areas that appeared crimson red on her map was the zip code inside the Loop that includes NRG Stadium, which she could see from her office. It’s an area that is largely African American. Twenty-three percent of its residents live below the poverty line.

In Texas, a black mother is four times as likely to die within 42 days of childbirth as a white one. Is this a problem that doctors can do anything about? Or is the maternal mortality rate of African American mothers caused by larger societal problems that medicine ought to steer clear of? The answers to those questions are important; if Texas can reduce maternal mortality among black women, we’ll go a long way toward solving our problem.

Once Cazaban finished her map, Houston Endowment convened a series of focus groups to hear directly from the women in the reddest zip codes, those most likely to become near misses. Of the 83 women who participated, 63 percent were African American, and 73 percent either lacked health insurance or were on Medicaid. When asked about the quality of reproductive health care they received, most of them said that doctors and other providers tended to gloss over family planning conversations and discussions about birth control options. The Medicaid patients said that they often felt that their doctors didn’t spend sufficient time with them—perhaps because Medicaid’s low reimbursement rates discouraged doctors from doing so. Many African American mothers felt less confident about visiting their doctors than their white or Hispanic counterparts did. The researchers concluded that doctors’ implicit biases—their unconscious attitudes and behaviors—affected whether African American mothers received adequate postnatal care.

In March, Houston Endowment, along with the Episcopal Health Foundation and the Cullen Trust for Health Care, committed to exploring the development of a visionary new pilot project that will aim to streamline care so that mothers in high-risk areas can receive seamless medical care, including good postnatal and prenatal care, and birth control. The pilot will utilize a major hospital system and a federally qualified health care center as providers. In addition, the endowment said it would create a county-wide data bank of deaths and near misses. This will be complicated, since this new local data silo will have to communicate with the state’s existing data silos. But Jessica Pugil, who runs a strategic planning and evaluation firm in Houston and volunteers for the steering committee, thought there was no time to wait for Texas to fix its data problems. “If whatever we do here can serve as a framework taken up by the state, that would be ideal,” she said.

Change can also happen on a much smaller scale. At Ben Taub, Ebonie’s former obstetrician, Dr. Eppes, has spearheaded an effort to change the way doctors and nurses there care for hemorrhaging soon-to-be mothers.

Previously, when prospective mothers bled, nurses and doctors in the delivery ward merely glanced at each pad before disposing of it. When they documented the amount of blood loss, they did so based on a visual estimate—something, Eppes said, doctors are “horrible” at. Most hospitals in Texas do the same. Eppes and her team pushed for those pads to be weighed, so that doctors and nurses could keep precise track of blood loss. After some initial resistance on the part of the nursing staff—weighing these pads takes considerably more time than eyeing them—the practice was adopted at Ben Taub, as the staff recognized the efficacy of doing so. For patients like Ebonie, with severe bleeding episodes, accurate blood loss totals could mean the difference between life and death.

If we want to see Texas’s rate of maternal deaths drop, then leaders in the medical field, in our communities, and in our state legislature will have to place as much of a priority on mothers as on the care of newborns. We’ll also need better information. To that end, TDSHS has decided to revamp the way it collects vital statistics, including death certificates. The new Texas Electronic Vital Events Registrar, to be launched next January, will attempt to improve the quality of certificate data by linking birth records with deaths. It’s unclear if the password rules will change.

Ebonie was rocked by her traumatic delivery and post-delivery. She was no longer in physical danger, but she continued to fear for her life.

“Even when I went home, I was still scared,” she said.

But as is often the case with new mothers, she had no time to dwell on her own problems. For the next two weeks, Ebonie returned to the hospital every day to visit her new son, King, whose young organs continued to develop in the ICU. The neonatologists told her that they intended to keep him in the hospital until his original due date, in October, three months in all. I met with her in a children’s play area outside the ICU. Ebonie had brought Blessn to visit her baby brother. While we talked, Blessn relished pushing herself up and down a little seesaw.

When I asked Ebonie if she was still having seizures, she smiled. “No,” she said. One of the positive outcomes of her recent hospitalization was a thorough study of her seizures, which led the neurologists to conclude that they were the result of the emotional stress of having been the victim of abuse, an illness known as conversion disorder. Mothering seemed to have ameliorated the condition, at least for the moment. It was certainly a better remedy than the seizure medicines she had been taking.

“I’m very happy that you’re okay,” I told her, as we stood up to leave.

Ebonie nodded in the manner of someone who is multitasking. With all the things going through her mind—King’s lungs, Blessn’s school, finding her own place eventually so she wouldn’t have to impose on her sister—how could she worry about reciprocating a small courtesy? How could she worry about her own health, even?

As we walked toward the door, she quickly turned around.

“Come on, Blessn,” she said. “We got two buses to catch.”