How to deal with mortality in perinatal research: Part 1.

At the PAS-meeting that just finished, we had a very well received topic symposium which was all to do with how death and dying has changed in neonatology over the last decade. My part of the symposium was to discuss death and dying in perinatal research, how to deal with the occurrence of mortality when planning and interpreting research.

I divided the talk into 2 parts, how to deal with mortality in epidemiologic/observational type research, and what to do about prospective studies, including randomized trials. This posting is an edited version of the first part of that talk.

The main points I had to make were that, for epidemiologic research, we have to start earlier in the ‘supply chain’ than the NICU. Obstetric attitudes can affect survival, well before babies get to the NICU, if we make the denominator fetuses who were alive when the mother was admitted to obstetrics, then we get different figures, compared to fetuses born alive, or to babies admitted to NICU.

I gave some references which show both that this is a real effect, and that it is changing over time.

This data for example is from the NICU in Groningen, the Netherlands, presented at PAS in 2012, first author Koper JF.

2001 to 2003 (n=126) 2008 to 2010 (n=113)
Stillbirth, dead on arrival to the hospital 44 (35%) 37 (33%)
Stillbirth, withholding surgical intervention (KJB: mostly C-section) 29 (23%) 17 (15%)
TOP, congenital malformation 22 (17%) 33 (29%)*
Induction for risk of extreme preterm birth 1(1%) 7 (6%)
Withholding resuscitation (comfort care) 25 (20%) 13 (12%) *
Failed resuscitation in the DR 5 (4%) 6 (5%)

and data from our hospital presented at the same PAS by Amélie Dupont-Thibodeau (cohort 1 and cohort 2 are very similar periods to the Groningen data).

Cohort 1; n = 166 Cohort 2; n = 298 P value
IUFD (% of total n) 44/166 = 26% 51/298 = 17% 0.0036
IUFD, alive at admission (WH C-section) (%) 29/166 = 17% 21/298 = 7% 0.0004
TOP (%) 67/166 = 38% 167/298 = 56% 0.00045
comfort care: prematurity (%) 26/166 = 15% 36/298 = 12% NS
Comfort care: congenital anomaly (%) 9/166 = 5% 21/298 = 7% NS
Failed resuscitation (%) 0 2/298 = 1% NS

Changes in obstetric approaches, due to changes in obstetric attitudes and the results of parental counseling, have significant effects on who is born alive. Once born alive, infants will get to the NICU if active, curative, care is instituted, there are very few failed resuscitations, almost all deaths in the DR are due to a decision to withhold, or limit resuscitation.

Neonatal attitudes and training also affect survival, but not just in neonatal survival, also in who is classified as a live birth. Wally Carlo et al’s research on Essential Newborn Care showed a decrease in still birth rates, clearly this is a result of personnel in the delivery room recognizing that some infants they thought were stillborn were actually viable, if they received active care. This effect would not have been seen if the group had not collected stillbirth rates. Similarly, comparison of different hospitals in our large databases, even when we are comparing outcomes of very immature babies, not the term and late preterm babies in Carol et al’s study, are affected by who is considered a live-, or a still-birth.

Here are the data: before “Essential Newborn Care” 23 per 1000 stillbirths, afterward 15.9 per 1000. RR 0.69 (95% CI 0.54-0.88). (Carlo WA, et al. Newborn-care training and perinatal mortality in developing countries. The New England Journal of Medicine. 2010;362(7):614-23.)

Nick Evans made a point, during our question and answer session, that babies who deliver in peripheral hospitals, because a decision was made not to transfer the mother, may also be a significant contributor to mortality. The proportion of potential very preterm babies affected could well differ from one region to another, so collecting regional data, which includes mothers whose babies were alive at (for example) 20 weeks, and who deliver a live- or still- born baby at extremely preterm gestations is important in order to be able to make valid comparisons.

Once babies arrive in the NICU, our problems are not finished.

In order to be able to continue to improve outcomes of the NICU, we need to know what causes of death are most frequent, so that we can then direct our energies at attacking those causes. But in recent publications, (from the NICHD network, from Pediatrix and from the German Neonatal Network, for example) the causes of death are not always clear, by which I mean that CNS injury, as one example is often listed as a cause of death. But in the very preterm, brain injury very rarely leads to physiologic instability and death. The usual sequence, in contrast, is that a head ultrasound shows a serious bleed, which leads to a discussion about long term function, and then withdrawal of life-sustaining interventions. In contrast, sepsis may lead to profound physiologic disturbance and thus to death by a more direct route, and necrotising enterocolitis might lead either to a discussion about whether surgery is appropriate, or to critical illness (or both). The ways to address these complications of prematurity change depending on the pathway to death.

These pathways may also be very different between one hospital and another. So in one center they may almost never have ‘death due to IVH’, not necessarily because they have discovered how to prevent it, but because they are skeptical about the predictive value of early head ultrasound. Another center might have less death due to NEC, as they routinely go to surgery, but have more deaths due to sepsis, as there are more babies with central lines for a longer time.

In order to have valid comparisons, and to decide how to improve our outcomes, we need to know, not just the cause of death, but also the mode of death.

This data from Edouard Verhagen and his collaborators show how these approaches can differ between different NICUs. (Verhagen AAE, et al. Categorizing Neonatal Deaths: A Cross-Cultural Study in the United States, Canada, and The Netherlands. The Journal of pediatrics. 2010;156(1):33-7.)

Intervention Physiology Groningen (n = 68) Montreal (n = 43) Wisconsin (n = 43) Chicago (n = 29) Total
Died in DR
Category A, died while receiving CPR Unstable 0 0 3 (7%) 0 3
Category B, withholding CPR, comfort care Unstable 16 (24%) 9 (21%) 4 (9%) 0 29
Died in the NICU
Category A, died while receiving CPR Unstable 3 (4%) 5 (12%) 4 (9%) 9 (31%) 21
Category B, withholding CPR, died on the respirator Unstable 1 (1%) 1 (2%) 0 5 (17%) 7
Category C, extubation to let the moribund child die in parents’ arms Unstable 35 (52%) 13 (30%) 17 (40%) 15 (52%) 80
Category D, elective extubation for quality-of-life reasons Stable 13 (19%) 15 (35%) 15 (35%) 0 43


The approach is also interestingly variable between the NICU and the PICU, even when they are separated by a single floor in the same establishement, here again are data from our hospital (Fontana MS, et al. Modes of Death in Pediatrics: Differences in the Ethical Approach in Neonatal and Pediatric Patients. The Journal of pediatrics. 2013.)

Intervention Physiology PICU (n = 68), n (%) NICU (n = 77), n (%) P value
Died with CPR (no WH or WD) Unstable 4 (6) 5 (7) NS
Died on ventilator (WH CPR, no WD) Unstable 35 (51) 4 (5) <.05
WH/WD (moribund children) Unstable 18 (27) 27 (35) NS
WH/WD (for QOL reasons) Stable 11 (16) 41 (53) <.05

The conclusion of this part of the talk was that for epidemiologic studies :

  • Reports should include all babies who eventually delivered at very preterm gestation (either alive or stillborn); and specify:
  • How many were alive when the mother was admitted to obstetrics.
  • How many were terminations of pregnancy for anomalies
  • How many were terminations of pregnancy/inductions of labour for threatened extreme prematurity
  • How many resuscitations not attempted for extreme prematurity
  • How many failed resuscitations
  • How many admissions to the NICU
  • How many deaths in the NICU
  • Causes and modes of death in the NICU
  • Only by reporting how the numbers change between admission of the mother, delivery of the baby, admission to the NICU, and discharge can we have valid comparisons that will allow us to move forward.

About Keith Barrington

I am a neonatologist and clinical researcher at Sainte Justine University Health Center in Montréal
This entry was posted in Neonatal Research and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.