Which heart rate features tell us the baby is in trouble?

For fetal heart rate monitoring to work the way that people want it to, people need to be able to identify heart rate patterns that are strong predictors of either a healthy baby or one that is at risk for some kind of poor outcome. This is true whether you are using CTG monitoring or listening intermittently to the heart rate. Given that people have been listening to the fetal heart rate since the 1820s, you might think that we have this stuff well worked out by now.

The truth is that we don’t. There’s not a lot of current, quality research in this area. Whenever a new study that contributes to this knowledge comes out, I get a bit excited that researchers are working on filling this gap.

When I saw this paper (Wiley et al., 2026) I was excited to be able to read and share it with you. Let me take you through their research and what they found.

What did they do?

The goal was to compare different features of CTG tracings between a group of healthy babies and a group that had one or more poor outcomes. The conditions included as poor outcomes were:

Apgar score of <7 at 5 minutes
Seizures
Hypoxic ischaemic encephalopathy
Sepsis
Bronchopulmonary dysplasia
Intraventricular haemorrhage
Necrotising enterocolitis
Birth injury
Meconium aspiration
The use of ventilation for more than 6 hours
Neonatal death

To be included, babies had to be from a singleton pregnancy, without a known fetal anomaly, at least 37 weeks of gestation, born after a period of labour, and have at least 20 minutes of CTG tracing. CTG recordings from the last 20 to 120 minutes of the labour were reassessed by a team of senior obstetricians who were given no information other than the trace itself. The recordings were assessed using the definitions set out in the American College of Obstetrics and Gynaecology (ACOG) guidelines – identifying both individual features of the tracing (like accelerations and variability) and the categorisation of the trace. ACOG uses a three category system. In essence, category I is considered normal, II is abnormal, and III is very abnormal.

What did they find?

During the study period (2020 – 2021), the research team identified 49 babies with a poor outcome who met the entry criteria, and 3,117 babies who were healthy – 3166 babies all up. Only 11.4% of babies had a category I trace all through the final period of monitoring. 88.5% had a category II trace at some point during that period and less than 1% had a category III trace. The most common poor outcomes were low Apgar scores (27 babies, 0.9% of the total) and the use of ventilation (14 babies, 0.4% of the total). There was one recorded death (0.03%).

So which bits of the CTG helped to predict the babies born with a health issue?

Changes in the baseline? No. Neither the incidence of tachycardia nor of bradycardia were significantly different between healthy babies and those with poor outcomes once confounding variables were corrected for. They didn’t look at rising baseline as a feature.
Variability? No. None of absent, decreased, or increased variability were predictive for poor outcome.
Accelerations were just as likely among babies born in poor condition (51%) as those that were healthy (59%).
Decelerations. They looked at early decelerations, late decelerations, variable decelerations, and prolonged decelerations, and within these at mild moderate and severe forms of each. The only type that were different between the two groups of babies were severe variable decelerations (aOR 4.67, 95% CI 1.01 – 21.68).
Category I throughout the final part of labour was seen in 12% of babies who were healthy and 8% of babies with a poor outcome – the difference was not significant. This means that of the 49 babies with a poor outcome, 8% (1 in every 12 babies) had a normal CTG for the last 20 – 120 minutes of labour.
Category II. There were no differences in the proportion of category II traces either for all or any part of the final part of labour, with 88.5% of healthy babies and 89.8% of the babies with a poor outcome having a category II trace at some time.
Category III. There were also no significant differences here either.

The only useful feature was the presence of severe variable decelerations (lasting more than 60 seconds and dropping below 80 beats per minute). These were uncommon in both groups though – 0.8% of healthy babies (that’s 25 of the these babies) and 4.1% babies with a poor outcome (that’s 2 of the 49 babies).

Let that sink in for a moment.

ALL the signs traditionally used in the USA (and many other places) to attempt to identify a baby that isn’t doing well were not helpful, apart from one. The only sign that was useful was helpful was seen in the CTGs of 27 babies (out of 3166 babies, under 1%). And only 2 of those babies were actually in trouble (7% of 27).

Their aha! moment

In the discussion of their paper the authors pointed out that the false positive rate of CTGs in this study was between 92 to 98%. This is the proportion of babies with an abnormal heart rate pattern who were actually fine. This fits with previous research.

They reminded their readers of the biologically plausible theory that placental insufficiency leads to late decelerations and cord compression to variable decelerations, and therefore persistence of these types of decelerations (in category II traces) presumably leads to damage from low oxygen levels. Yet their findings don’t match what you would expect to see if this were true. They then have the same big aha moment I had many years ago, saying:

Given the poor positive predictive value of an abnormal fetal heart rate both in our study and others, we posit that presumed reasons for adverse neonatal outcomes should be revisited, if not revised.
p. 943

In other words – maybe we have the ENTIRE story about CTGs wrong.

Another claim often made about CTG monitoring is that a normal trace is very reassuring as it shows a reduced chance that a poor outcome is possible. That was not true in this case.

If the heart rate patterns we have been taught to watch out for can’t help to reassure women when the patterns are normal and the only abnormal feature that predicts anything meaningful happens in less than 1% of women’s labours – then maybe it is time to admit that CTGs are nonsense?

The first point I make in my book Monitoring Your Baby in Labour is this: “A lot of nonsense has been – and continues to be – said and written about the use of CTG monitoring in pregnancy and during labour. I believe it is well past time for that to stop.“

If you are looking for a concise and easy to read summary about evidence for all the various fetal monitoring options – grab a copy from your preferred book purchasing options. It comes as an ebook (no – you don’t need to own a Kindle to read it), a paperback, and a hardcover (which is handy if you want to pass it around from person to person and have it last ages). Click the button below for a comprehensive list of all the various places you can find it.

Buy your copy today

Details about where to buy it are here!

References

Wiley, R. L., Roberts, A. W., Cagino, K. A., Zullo, F., Mendez-Figueroa, H., & Chauhan, S. P. (2026). Characteristics and categories of fetal heart rate tracings and adverse neonatal outcomes at term. American Journal of Perinatology, 43(7), 940-948. https://doi.org/10.1055/a-2708-4947