Repatriation is one of the key milestones in neonatal care. A baby who once needed the resources of a tertiary unit is now stable enough to continue care closer to home, and the transfer is recorded and often celebrated as progress.

Yet many parents describe the days around a move as more frightening than the acute events they had already survived. They become hypervigilant again, struggle to sleep, question decisions they had previously accepted, and find reassurance ineffective.

It is tempting to read this as anxiety, poor coping, or a lack of trust in the receiving team. The more accurate reading is neurobiological.

I am going to start with lived experience, because after all, that is why I am here.

What a move felt like from the inside

I never moved units in my NICU season, but I did during my PICU season, and it was probably one of the most distressing experiences I had.

We were closer to home, but we were not in our usual local hospital where I knew the routines, I knew the staff, I knew where to get a cup of tea, I knew when ward round was, I knew how to set up my camp bed. We were in the thick of winter and within an hour of extubation we were being moved to a PICU closer to home.

The change of pace, where alarms were tended to by parents. The change of expectations, where no one wore gloves or aprons in the local PICU. No one, or perhaps two people, at the foot of my son's bed.

I felt like everything was on me, in a hospital setting, following a ten day stint in PICU. And it was in that moment I realised that as scary as PICU was, I had found safety within it through the regimented and highly clinical care. After all, my son was back, once again on a ventilator, needing intensive care to save his life.

I remember becoming so distressed that I begged a consultant to send us back to our local hospital. There was no medical reason to keep him where he was, but I was struggling to cope.

This is something I have heard hundreds of times caring for families at Leo's. That repatriation to a closer neonatal unit can go unrecognised as a distressing event, in the guise of it meaning one step closer to home, which is the aim after all.

A transfer is not only a change of location. It is a large, simultaneous change in almost every sensory, relational and procedural cue that a parent's nervous system had been using to infer that their baby was safe.

What follows sets out the mechanisms involved, drawn from predictive processing, autonomic theory, interoception and the stress and allostasis literature, and translates them into practical implications for how teams can manage the transition.

The brain is a predictive organ

Contemporary neuroscience frames the brain not as a passive receiver of sensation but as an active inference system. Rather than waiting for events and reacting, the brain continuously generates predictions about the external world and the internal state of the body, then compares those predictions against incoming signals. The key here is that this is based on our understanding of the world, and our lived experience.

Where the two diverge, a prediction error is generated, flagging that the internal model needs updating (Friston, 2010; Friston, Thornton and Clark, 2012).

Several elements of this framework matter clinically.

Priors are the learned and inherited expectations the brain brings to a situation. We can think of this as the parents' learned experiences of neonatal care in a tertiary setting.

Prediction error is the mismatch between expectation and evidence.

Precision weighting is the confidence the brain assigns to a given signal relative to its priors, effectively the gain on incoming information (Friston, 2010; Bottemanne, 2025; Feldman et al., 2024).

When this happens under sustained uncertainty and threat, the precision tends to shift toward threat consistent hypotheses, so ambiguous signals are interpreted conservatively, as danger rather than as noise.

This logic applies to the body as much as to the outside world. Interoceptive experience is not a simple readout of visceral signals. It is the product of descending predictions about internal state interacting with ascending bodily input (Critchley and Garfinkel, 2017; Barrett and Simmons, 2015).

The implication is that what a parent feels in their body during a transfer is shaped by what their brain has learned to expect, not only by what is physiologically occurring.

What the tertiary unit had encoded

Over days or months in the tertiary unit, a parent's brain builds a detailed generative model of where and how their baby is kept safe.

One to one or close nursing, senior medical presence, particular alarms that were answered by the person stood at the bottom of the incubator or cot, a familiar physical layout, a hospital they have learned to navigate and a predictable rhythm of care become bound together with high emotional salience.

The amygdala tags these cues for significance and the hippocampus binds them into contextual memory, so the environment itself becomes encoded as a template for safety (LaBar and Cabeza, 2006; Sumadevi, 2024).

This binding is strengthened by the conditions under which it occurs. Emotionally arousing events recruit noradrenergic and glucocorticoid signalling that enhances consolidation of the central, threat relevant features of an experience (McGaugh, 2013; Henckens et al., 2009).

The result is that the cues a parent associates with their baby's survival are not neutral preferences. They are deeply learned safety signals, encoded as such because of the levels of threat that were at play.

The transfer as a model violation event

When the environment changes in a single move, reality stops matching the model the brain has built, and a large prediction error is generated.

From a predictive processing perspective the system must rapidly update, and under uncertainty it commonly raises precision around threat cues in order to avoid missing danger (Friston, 2010; Feldman et al., 2024).

Subjectively this does not present as a measured reappraisal. We may see a parent with tension, restlessness, a need to control and regain agency, and a pull to monitor everything, because the brain is signalling that something consequential has changed before any narrative explanation is available.

Crucially, the direction of the clinical change does not determine the size of the prediction error. A move that is unambiguously good news can still constitute a substantial violation of the model, because the model was built around the specific conditions of the previous unit, not around an abstract idea of recovery.

The autonomic response: threat detection and the defence cascade

Beneath conscious appraisal, the brain continuously evaluates cues of safety, danger and life threat through fast, subcortical threat-detection circuits centred on the amygdala (Kozlowska et al., 2015). In polyvagal terms, this rapid and precognitive evaluation is called neuroception (Porges, 2011).

Repeated exposure to uncertainty and threat can lower the threshold of this threat-detection system, biasing it toward danger, a predictable adaptation in an environment of alarms, procedures and ambiguous prognostic information.

When the system detects threat, the body does not select from a menu of states. It recruits a patterned, ordered survival sequence known as the defence cascade, organised across the amygdala, hypothalamus, periaqueductal gray and autonomic nuclei (Kozlowska et al., 2015).

The point that matters most clinically is that arousal is always the first response. The cascade opens with hypothalamic and sympathetic activation, raising heart rate, respiratory rate and muscle tone, and this mobilisation is recruited from the bottom up, before cognition has interpreted the situation (Kozlowska et al., 2015).

Only from this aroused state can the system branch onward, into the active defence states we are more familiar with, such as fight or flight, into freeze where mobilisation is inhibited but sympathetic tone remains, or, under overwhelming and inescapable threat, into collapsed immobility.

A key point here is that shutdown is not the opposite of mobilisation. It is a later stage that the body reaches only after mobilisation has already occurred.

This is why a transfer reliably produces activation rather than relief. At the moment of the move, the cues that had signalled safety are removed or altered at once, the threat-detection system registers the mismatch as threat, and the cascade begins where it always begins, with sympathetic arousal and mobilisation.

The receiving environment may be entirely appropriate for the baby to progress toward home, but threat detection is not an objective audit. It is a rapid, subcortical comparison against learned templates, and a parent's first physiological answer to that mismatch is to mobilise, not to settle.

Decoding the specific threat cues of repatriation

The generic statement that change is stressful is insufficient for neonatal clinical practice. It helps to be specific about which features of a transfer the nervous system is likely to read as threat, and why.

A lower nurse to baby ratio. In the tertiary unit, intensive surveillance had become part of the parent's felt sense of safety. A reduced ratio reflects an appropriately lower acuity, but the threat system can encode it as fewer eyes on the baby, that is, as the withdrawal of a safety signal the brain had come to rely on. Parents frequently compensate by increasing their own monitoring, attempting to fill the gap their model expects another person to fill.

Less frequent consultant presence. Senior medical presence is often paired, through repeated experience, with the capacity to rescue. When consultants are no longer on the ward as often, that association is disturbed, even when the baby is now well enough not to require it. The absence is filed as elevated risk. This is learned survival logic rather than a considered judgement about the competence of the receiving team.

New staff, new routines, new sensory environment. Every unfamiliar face, handover, new parking system to work out, and procedure introduces uncertainty, and uncertainty is the variable that drives the system to increase the weighting of threat consistent priors while reducing confidence in reassurance (Critchley and Garfinkel, 2017; Feldman et al., 2024).

This is one reason the first days after a move can feel more destabilising than the original crisis.

Interoception when the monitors are removed

In a higher acuity setting, numerical monitoring offloads some of the work of inferring the baby's state, and gives parents a window into their baby.

When that external data is reduced, parental attention often turns inward, onto the baby's body and onto their own. Subtle cues, a pause in breathing, a change in colour, a small movement, acquire heightened salience, integrated through insular networks that map bodily state and contribute to subjective feeling (Namkung et al., 2017; Critchley and Garfinkel, 2017).

Under sustained stress, the precision assigned to these interoceptive signals can shift toward threat, so the same bodily sensation that once read as ordinary may now read as urgent (Schulz and Vögele, 2015; Harricharan et al., 2021).

When a parent says that something feels off, this is clinically meaningful detection under uncertainty, not irrationality.

A dismissive response is not neutral. It adds social and sensory information that can increase uncertainty, heighten threat inference and erode trust both in the team and in the parent's own perception and self-belief over time (Harricharan et al., 2021).

Allostasis and the cost of anticipatory regulation

The stress involved in all of this is not only psychological.

Allostasis describes the maintenance of stability through anticipatory change, the brain adjusting physiological set points in advance of predicted demand rather than only correcting deviations after the fact (McEwen, 1998; Bobba-Alves et al., 2022). This is efficient in the short term, but when prediction driven mobilisation becomes chronic it accrues a cumulative cost known as allostatic load (McEwen, 1998; Juster et al., 2010).

Uncertainty, unpredictability and a perceived loss of control are particularly potent drivers of anticipatory allostasis. A transfer concentrates all three, which is why the physiological burden can rise at precisely the moment the clinical picture is improving.

Parents may be simultaneously sleep deprived, anticipating worst case outcomes, sensorily overloaded by a new environment and acutely mobilised by unfamiliar alarms, and that layering is what shifts stress from adaptive deviation toward sustained load.

Why reassurance underperforms

If the destabilisation is generated by a violated predictive model, then verbal reassurance alone is mismatched to the problem. The model updates through repeated, lived experience that the new environment is safe, not through being told that it is. This is where felt safety becomes an integral part of care.

There is a lag between the clinical change and the nervous system catching up, and during that lag threat weighted priors remain dominant (Friston, 2010; McEwen, Nasca and Gray, 2016). Reassurance offered without reducing uncertainty may fail simply because the inferential model has not yet been given the evidence it needs to update.

Safety has to be experienced repeatedly before it is encoded as real. This is slow biology, and it is also where the opportunity lies.

Implications for clinical practice

Understanding the mechanism points directly to what helps. The aim at transfer is not to eliminate vigilance but to give the nervous system the conditions it needs to recalibrate, by reducing uncertainty and allowing safety to repeat.

• Treat the handover to the family through a nervous system lens. A supportive, compassionate, structured and paced orientation to the new unit, its routines, its staff and how concerns will be escalated directly reduces the uncertainty that is driving threat weighting. Explicitly explain the changes that will come, verbally and in writing. Ask parents if they need time to reflect on the changes coming.
• Maximise predictability. Consistent team language, clear timelines and explicit escalation planning lower prediction error and support autonomic regulation more effectively than general reassurance.
• Normalise the response explicitly through validation. Naming that the brain learned to feel safe somewhere else, and that bracing is expected, reduces shame, and lower shame reduces the salience amplification that maintains the alarm.
• Receive interoceptive reports collaboratively. Responding to "something feels off" with "let us look at this together" preserves dignity, supports regulated inference and maintains trust, without committing to the concern being correct.
• Allow graded, repeated experiences of safety. Each calm, uneventful day on the new unit is data the system uses to update its model. Those micro moments of safety are crucial in building capacity.

Reframing the behaviour

When a parent appears hypervigilant, mistrustful or dysregulated after a transfer, what you are most likely witnessing is a nervous system that learned, under high stakes, that vigilance kept a baby alive, and that has not yet been given enough lived evidence to update.

This is adaptation, not pathology.

Recognising it as such changes the clinical response from managing a difficult parent to supporting a recalibrating system, and that shift is where safer, more compassionate transitions begin.

This is nervous system informed neonatal care.

Lottie

This article is drawn from the NICU Neuroscience master course, a training programme that brings neuroscience and lived experience together to protect the mental health and wellbeing of neonatal families and the teams who care for them. 

It is mapped to WHO, NICE, BAPM, GFCNI, the Royal College of Midwives, the Royal College of Nursing, the Royal College of Paediatrics and Child Health, and the Family Integrated Care framework.