Close

I Received Tips to Look Into How a Hospital Treated Premature Babies. Getting Data Was Nearly Impossible.

New Mexico limits the information it collects on neonatal centers. That makes it incredibly challenging to get reliable data, sort out what’s wrong and figure out how to fix it.

Creative Commons
(Photo illustration by Shoshana Gordon/ProPublica, source images: Shaun Griswold and Marjorie Childress/New Mexico in Depth, and Jennifer Pack via Flickr)

This article was produced in partnership with New Mexico In Depth, which is a member of the ProPublica Local Reporting Network.

It’s not unusual for health care reporters to get out-of-the-blue calls or emails from people in the industry. But when three clinicians from Albuquerque hospitals reached out to me to share concerns about the state’s largest for-profit maternity hospital, Lovelace Women’s, I took note.

Two of the tipsters worked at Lovelace. None knew the others had contacted me, but all three had concerns about how Lovelace cared for its most premature babies.

These delicate preemies are frequently born weighing less than 2 pounds. Their immature lungs are not ready for life outside the womb. Their intestines cannot handle normal feedings and are prone to inflammation, tears and life-threatening infections.

In recent years, most preemies transferred to the University of New Mexico Hospital for emergency intestinal surgery had come from Lovelace, one tipster said. (UNM is home to the state’s only level 4 regional referral NICU.) These babies had advanced cases of necrotizing enterocolitis, or NEC, that clinician said. An inflammatory disorder of the intestines, NEC is a leading cause of death among the smallest newborns.

Throughout 2020, I interviewed dozens of people: Albuquerque NICU clinicians, newborn transport team paramedics and nurses, hospital officials, state Health Department and Medicaid program staff, and nationally recognized NICU experts. I contacted other current and former Lovelace clinicians, some of whom echoed the tipsters’ concerns. I sought New Mexico Health Department hospital inspection reports and regulators’ emails about neonatal hospital care.

But to evaluate the clinicians’ concerns, I also needed data. Hospital data can be used to assess trends and details that no one health care worker is likely to detect. Data can help a reporter determine if a patient’s experience is typical or an aberration.

In this case, hospital transfers would show whether a disproportionate number of NEC transfers to UNM really came from Lovelace. Diagnostic data would, I hoped, reveal how common those infections were at Lovelace. And if I could get data on death rates, I could compare it to the numbers for a similar hospital in the area.

My problem: Getting reliable data about individual hospitals’ patients and outcomes in New Mexico is incredibly challenging.

First, I obtained data on hospital transfers: the UNM neonatal medical transport team’s logs and the state Health Department’s emergency medical services tracking system database. UNM and the Health Department are each supposed to document hospital patient transfers, like babies sent from Lovelace to UNM for surgery. I hoped these would provide a window into how frequently Lovelace and Presbyterian Hospital, Albuquerque’s only level 3 NICU hospitals, had medical emergencies demanding a higher level of care. (Level 3 NICUs are supposed to be able to provide sustained, expert care for even the smallest preemies and medically complex babies.)

But the state’s EMS database turned out to be incomplete. For years, air and ground ambulance services simply hadn’t filed reports and the state hadn’t forced the issue. And the UNM neonatal transport team’s service logs only went back a couple of years.

I requested a copy of the intake log from UNM’s NICU, which was available because UNM is a public institution subject to the state’s public records act. The log is reported to the Vermont Oxford Network, an international NICU research collaborative to which UNM, Lovelace and Presbyterian all belong. The network would not share member hospitals’ newborn outcomes. But having UNM’s intake logs meant I could look for patterns in emergency newborn transfers from Lovelace and Presbyterian.

I also requested autopsy records, hoping to identify patterns in newborn deaths at Lovelace and Presbyterian. But I learned that few autopsies are done for newborns at either hospital. Autopsies cost thousands of dollars, and grieving parents, rather than hospitals, pay for them.

For several months in the spring and summer, I was occupied with COVID-19 reporting. But by August and September I was looking into Health Department databases that contain information about births and each patient’s diagnosis, treatment and outcome at all of the state’s hospitals except for those operated by the federal government.

Reporters in some states can obtain patient-level hospital records that have been stripped of patient-identifying details but that name individual hospitals. Such records link each patient with details like sex, year of care, diagnoses, treatments, the hospital where they were treated, and whether the patient was discharged home, transferred to another hospital or died. That kind of data allows for complex statistical analyses of hospitals’ patient populations and outcomes, and it could help determine if one hospital’s patient population was higher risk than another’s, which could lead to worse outcomes even if both hospitals delivered the same level of care.

But due to New Mexico laws like the Health Information System Act, the state publicly releases only aggregate data to protect patient privacy — counts, basically, of patients whose hospital records contained mention of a particular diagnosis or medical procedure. Patient-level data is exempt from disclosure under the state’s public records law. That limited my ability to do the sophisticated analyses that patient-level data sets allow. (Patient-level data can be purchased for thousands of dollars from the federal government, but data for recent years is not included.) And because of a change in diagnostic and billing codes in 2015, data from that year is difficult to interpret and to compare to data from other years.

When I submitted my initial data request, Health Department officials said the state had a longstanding practice of not disclosing individual hospitals’ data.

But when I read the state law, I could find no specific prohibition against disclosing hospital names; the state has discretion over when to release aggregate data for research purposes. So I asked state health officials about their practice of not disclosing data for named hospitals, emphasizing that I wanted the data to report on a pressing matter of public interest. I promised I wouldn’t use it to try to identify individual patients.

To my surprise, the Health Department reviewed and reversed its practice. It provided detailed aggregate data from two databases — a hospital patient database and a birth- and death-certificate database — for Albuquerque’s NICU hospitals. In a Zoom meeting, epidemiologists patiently fielded my questions.

The Health Department data was very useful but far from perfect. Because the state doesn’t monitor neonatal admissions or outcomes, hospitals’ NICU and newborn data is not audited for quality. The hospital patient database has gaps that complicated any analysis of extremely preterm babies’ outcomes. For example, it includes newborns’ birth weights, but not how premature they were at birth. And hospital patient data on babies’ NICU admissions isn’t linked to other details about babies’ health or medical procedures. Hospital-wide counts of babies with particular diseases were provided, but hospitals don’t always report them properly. For example, UNM’s intake log showed Lovelace transferring even more babies to UNM for NEC surgery than it had reported diagnosing with NEC in the first place. That meant I couldn’t rely on the data Lovelace had reported to the patient database.

Lovelace and Presbyterian officials initially said they would share their newborn NEC and transfer statistics but did not do so. Lovelace did not respond to many of my questions but defended its track record of neonatal intensive care. Lovelace Vice President for Marketing Serena Pettes claimed that for all NICU-admitted newborns — including both lower-risk full-term babies and premature babies — Lovelace’s neonatal death rate is “significantly lower than the national average” and has declined over time. “As a whole, our mortality rate is less than half of the national NICU average.”

Birth and death certificate submissions to the Health Department’s vital records bureau are supposed to include details of maternal medical history, labor and delivery, and medical care for the newborn, including whether they were admitted to the NICU. But reviewing data for extremely preterm babies, New Mexico In Depth and ProPublica found evidence that data is also incomplete for key variables, like congenital birth defects and NICU admissions. This meant I could not determine if the number of babies with major birth defects at each hospital, for example, might explain their different death rates.

But the Health Department’s vital statistics data — the information reported to the state by the hospitals for each live birth and newborn death — did allow me to calculate Lovelace’s and Presbyterian’s hospital-wide death rates for extremely preterm and extremely low birth-weight newborns. As the clinicians had suspected, Lovelace’s death rates for these babies were markedly higher than Presbyterian’s. (There was no such disparity in the death rates of babies born at later gestational ages or higher birth weights.) And UNM’s intake records confirmed that more Lovelace babies were transferred to UNM with NEC, including extremely preterm babies.

On March 12, weeks before New Mexico In Depth and ProPublica published stories describing the analysis, Lovelace distributed a memo to employees giving talking points about my reporting. It described the findings as “using statistics to create a narrative that is not supported by accurate data.” They’d said the exact same thing to me earlier in the month. I’d asked the Lovelace spokesperson then if that meant the hospital had submitted inaccurate information to the state. She did not reply.

While the data bolstered the tipsters’ concerns, the reasons why Lovelace’s extremely preterm babies die and are transferred to UNM at a higher rate than Presbyterian’s babies remains elusive.

Such disparities can reflect differences in clinical practices or in how sick patients are, or both. Without more complete patient-level data — or an independent, on-site review of the hospital’s patient records — understanding the causes of the disparity is impossible, experts said.

This points to another key finding from my investigation: New Mexico does not have the regulatory tools to sort out what’s wrong or how to fix it.

Filed under:

Latest Stories from ProPublica

Current site Current page