How We Investigated Death Rates for Extremely Preterm Babies in This State’s Largest Maternity Hospitals
A New Mexico In Depth and ProPublica investigation found that the tiniest, most premature babies born at Lovelace Women’s Hospital in Albuquerque died at higher rates than they did at a hospital a few miles away, Presbyterian.
The for-profit Lovelace and nonprofit Presbyterian are New Mexico’s largest maternity centers.
The most comprehensive data on newborn hospital outcomes is collected by the Vermont Oxford Network, or VON, an international neonatal intensive care unit research collaborative. VON maintains patient-level intake and care data for member NICUs, including those at Lovelace, Presbyterian and the University of New Mexico Hospital. The data can be used to calculate death rates at individual hospitals. Maternal and newborn demographics, prenatal care, medical procedures and complications, neonatal hospital transfer history and babies’ outcomes are also captured in the data.
VON data, however, is not public; the network discloses outcomes only to member hospitals. The group declined the news organizations’ request for mortality rates of extremely preterm babies at Lovelace and Presbyterian. UNM, which runs the state’s highest-level NICU, shared its de-identified VON database and annual reports in response to a public records request, but neither Lovelace nor Presbyterian would do the same. The hospitals and VON also refused to share the network’s national average death rate for extremely preterm babies.
Since we did not have access to VON data for Lovelace and Presbyterian, the news organizations obtained aggregate birth and death certificate data for extremely premature babies from the state Health Department’s Bureau of Vital Records and Health Statistics. Hospitals submit data on each live birth and in-hospital newborn death for babies born to legal residents of New Mexico. We initially analyzed data for the most recent available five years (2015-2019), broken down by gestational age and birth weight. We used this data to calculate hospital death rates. To do a more detailed secondary analysis, the news organizations obtained data for babies born during the most recent available 10 years, 2010-2019, based on gestational age. Because the longer time period included more births, it allowed comparisons of additional patient subgroups and variables.
Hospitals also report diagnostic and billing code data to the state Health Department’s Hospital Inpatient Discharge Database, or HIDD, which collects diagnosis and treatment information about patients admitted to every hospital in the state, except those operated by the federal government. We obtained aggregate HIDD data from the Health Department.
In devising our analysis, we worked with national experts in neonatal intensive care, including Dr. Scott A. Lorch, a professor of pediatrics and associate chair of the Division of Neonatology at the Children’s Hospital of Philadelphia and a leading authority on NICU outcomes, and Dr. David C. Goodman, a professor at the Dartmouth Institute for Health Policy & Clinical Practice at Dartmouth College. Goodman helped analyze the data, and the analysis was reviewed by Lorch.
What We Found
Babies born weighing less than 1,000 grams (or about 2 pounds) are called extremely low birth-weight, while those born before 28 weeks of pregnancy are labeled extremely preterm. (A full-term pregnancy is 39–40 weeks.) While most babies born before 28 weeks are also extremely low birth-weight babies, that is not always the case. To account for that, the news organizations initially evaluated death rates using both birth weight and gestational age for the most recent five years for which Health Department data was available: 2015-2019.
Using the Health Department birth and death certificate data, we calculated a neonatal death rate at each hospital, which measured the percentage of live-born babies who died at their birth hospital within 28 days of delivery. We found that:
- Between 2015 and 2019, 34% of Lovelace’s 88 extremely low birth-weight infants died at the hospital, compared to 17% of Presbyterian’s 197.
- In those same years, 36% of Lovelace’s 84 extremely preterm babies died there, compared to 21% of Presbyterian’s 170.
- The extremely preterm neonatal death rates at each hospital didn’t change much when calculated for the entire 10 years between 2010 and 2019: 36% for Lovelace vs. 22% for Presbyterian when calculated by gestational age, and 33% vs. 20% using birth weights.
- The news organizations’ analysis found no disparity between the hospitals’ death rates for babies born at older gestational ages and higher birth weights.
Hospital-wide death rates were calculated, accounting for deaths regardless of whether babies were admitted to the hospitals’ NICUs. Goodman, who was the lead author of the September 2019 report Dartmouth Atlas of Neonatal Intensive Care, advised New Mexico In Depth and ProPublica that the most accurate comparisons include the deaths of all extremely preterm babies, whether or not they were admitted to a NICU.
Only babies born at each hospital were counted toward the number of births. Only those who died at their birth hospital within 28 days were counted toward the number of deaths. (Including babies who died after being transferred to UNM’s NICU raised Lovelace’s extremely preterm neonatal death rate to 39%. Because Presbyterian transferred only one extremely preterm baby to UNM, in 2013, and they survived, that hospital’s death rate was unaffected.)
The analyses omitted babies considered “pre-viable” (those delivered before 21 weeks of pregnancy or weighing less than 350 grams), because they would likely die regardless of medical intervention, according to the state Health Department.
As discussed, some babies born at Lovelace and Presbyterian were transferred to UNM for care. Additionally, neonatal transport team logs suggest each hospital sent a handful of babies to out-of-state hospitals. The transport team logs were not detailed enough to allow us to remove all extremely preterm babies who were transferred out of state from our death-rate calculations. Therefore, our death rate treated all babies transferred out of Lovelace and Presbyterian as if they had received care at their birth hospital and lived, almost certainly making our death rates underestimates.
In addition to in-hospital deaths, we also examined how frequently Lovelace and Presbyterian transferred newborns to UNM Hospital, the state’s only top-tier, level-4 NICU hospital.
Having the UNM NICU’s VON data allowed us to analyze the frequency of transfers to UNM, and the condition of and outcomes for those transferred babies. We found:
- Lovelace transferred more than three times as many newborns as Presbyterian to UNM between 2015 and 2019. Lovelace sent 66 babies, both full term and preterm, while Presbyterian transferred 17 babies, none of them preterm.
- Close to half (46%) of Lovelace-born extremely preterm babies either died at the hospital or were transferred to UNM between 2015 and 2019.
- Of 18 babies with necrotizing enterocolitis, or NEC — an inflammatory intestinal disorder and leading cause of extremely preterm newborn hospital deaths — who were transferred to UNM since 2012, 15 came from Lovelace. Twelve required intestinal surgery, and two died within hours of their arrival. Only one Lovelace baby was transferred with NEC and survived without surgery.
We also found a discrepancy in the number of Lovelace babies diagnosed with NEC when we compared the HIDD data and UNM’s VON intake logs. The hospital reported four cases of neonatal NEC to the state between 2015 and 2019, but the UNM VON data documented the arrival of 11 Lovelace babies with NEC during those years. Lovelace declined to explain the discrepancy.
In addition to comparing Lovelace to Presbyterian, the news organizations sought a national comparison rate for extremely preterm babies’ death rates at level-3 neonatal hospitals, but found a dearth of publicly available data. Most published studies are based on data that is more than a decade old.
In response to questions from the news organizations, however, the Centers for Disease Control and Prevention provided a national death rate for extremely preterm infants. The CDC’s rate included all hospitals, regardless of the facility’s level of care, using the same gestational age and birth weight parameters used in New Mexico In Depth and ProPublica’s analysis. The CDC found a national, hospital-wide extremely preterm neonatal death rate for 2010-2018 of 28%, near the midpoint between Lovelace’s 36% and Presbyterian’s 22% for 2010-2019.
The CDC’s national death rate might be higher than a rate for only level 3 hospitals would be, because it includes hospitals with level 1 and 2 neonatal designations, which are less equipped to care for these babies.
Experts advised against comparing level 3 facilities like Lovelace and Presbyterian to level 4 neonatal hospitals like UNM, because level 4 hospitals treat the sickest newborns and have more medical subspecialists and surgical experts on staff, making it difficult to interpret differences. However, for completeness, we calculated death rates for UNM, using the Health Department data and the same methods described above for Lovelace and Presbyterian. The 2015-2019 death rates for extremely low birth-weight babies and extremely preterm babies born at UNM were 29% and 33%, according to the news organizations’ analysis.
Scrutinizing Risk Factors
The cause of the disparity in Lovelace’s and Presbyterian’s extremely preterm neonatal death rates is not clear. Possibilities include one hospital having a sicker patient population, differences in patient care, or both.
The Health Department provided additional breakdowns of outcomes by patient demographics and treatment details, none of which alone explained the difference in death rates. The summary-level data did not allow us to examine several variables in combination.
Babies born at 21 to 23 weeks’ gestation are considered “peri-viable” and resuscitation practices for these babies vary between hospitals, experts and Albuquerque clinicians said, which may lead to differing outcomes.
After excluding babies born at 21-23 weeks’ gestation who died without being resuscitated at delivery from the analysis, Lovelace’s extremely preterm neonatal death rate was 29%, compared to 19% at Presbyterian. Excluding all babies who died within an hour of delivery, regardless of gestational age or resuscitation reports, Lovelace’s death rate was 26%, compared to 18% at Presbyterian and a CDC-provided national rate of 22%.
(Goodman warned that resuscitation numbers are unreliable because hospitals frequently fail to report them.)
By 24 weeks’ gestation, resuscitation at delivery is always attempted, experts and clinicians said. When only babies born at 24-27 weeks’ gestation were included in the analysis, the death rate at Lovelace was 23% vs. 9% at Presbyterian for 2010-2019. The CDC-provided national rate for this group was 15%.
Another explanation for the death-rate disparity could have been if Lovelace had a disproportionate number of 21- or 22-week-gestational-age babies, who have poorer outcomes than babies born at 24-27 weeks. But we found that the distributions of births by week of gestation at the two hospitals were very similar.
The state Health Department provided demographic and medical treatment details for extremely preterm babies born at 24-27 weeks’ gestation. Lovelace’s higher death rate persisted in analyses of that data. For example, extremely preterm twins and triplets often fare worse than single-born babies. But this did not explain the hospitals’ death-rate disparity. Nor did maternal race or ethnicity, differences in the proportion of extremely preterm babies who were boys (boys tend to have worse outcomes than girls), or differences in prenatal therapies. Medically induced labor and other risk factors for extremely preterm babies, such as the proportion of teen mothers, mothers who underwent infertility treatment or mothers who had cesarean-section deliveries also did not account for the disparity.
Discrepancies in numbers for NICU admissions and birth defects precluded analysis of that data. All extremely preterm babies who survive delivery should be admitted to the hospitals’ NICUs, clinicians and experts agreed, but there were babies at both hospitals for whom there were neither NICU admissions records nor death certificates. Neither Lovelace nor Presbyterian reported birth defects with extremely preterm babies’ birth certificate data, but a handful of death certificate reports note birth defects as a cause of death. There was only one record of a Lovelace extremely preterm baby being transferred to UNM in the first 24 hours after delivery, so transfers should not explain the missing NICU admissions data. When asked for comment, the hospitals did not acknowledge or explain the discrepancies.
Identifying the causes of the hospitals’ extremely preterm neonatal death rate disparities would require access to patient records or to completely reported, patient-level records, experts said.
Lovelace presented several objections to our analyses. The hospital objected to comparing its extremely preterm newborns’ death rates to those of Presbyterian or any other individual hospital, arguing that comparisons should only be made to national benchmarks.
Goodman defended comparisons of outcomes between nearby level 3 neonatal hospitals: “A comparison within the same community, serving similar populations, operating under similar state policies and regulations, is of the highest value.”
And the news organizations obtained a national rate from the CDC, as described above, which was lower than Lovelace’s death rate.
In addition, the hospital contested the decision to look at hospital-wide outcomes for babies, regardless of whether or not they were admitted to the hospitals’ NICUs. Lovelace asserted that only NICU deaths should be measured because hospital-wide rates include deaths in the delivery room and could include nonviable babies. Presbyterian officials did not object to a hospital-wide comparison.
Lovelace said its NICU-only 2015-2019 death rate for extremely low birth-weight newborns was 22%. They would not say how the 22% NICU-only death rate compared to national rates in the VON. Presbyterian refused to disclose its NICU-only death rate.
Goodman rejected Lovelace’s objection to hospital-wide comparisons, noting that NICU staff should be present at the delivery of extremely preterm babies and that his research team includes “every extremely preterm baby who dies … in our research. They are cared for by the NICU team and are the responsibility of the NICU team whether or not they are administratively admitted to a hospital’s NICU.”
Another objection from Lovelace was the decision to calculate death rates for extremely low birth-weight newborns in the first place, noting that babies born weighing less than 1,000 grams represent only 2% of the hospital’s NICU-admitted babies. “This sample size is too small from which to derive any conclusions about the overall quality of care being provided,” Lovelace Vice President for Marketing Serena Pettes wrote in an email.
But extremely preterm babies are a formally recognized and widely studied patient population at neonatal hospitals. The American College of Obstetricians and Gynecologists defines as extremely preterm babies born before 28 weeks of gestation, the definition we employed. While extremely preterm babies do make up a small portion of all NICU admission, between 2010 and 2019, more than 500 such babies were born at Lovelace and Presbyterian.
ProPublica’s deputy data editor, Hannah Fresques, and a member of ProPublica’s data science advisory board reviewed the analysis