The effect of acute respiratory events and respiratory stimulants on EEG-recorded brain activity in neonates: A systematic review

Highlights • Few studies investigated EEG changes during apnoea and in relation to respiratory stimulants in neonates.• EEG suppression is observed during some apnoeas but diverse definitions of apnoea and EEG measures limit inference.• Respiratory stimulants increase EEG continuity compared with before use.


Introduction
Neonates, especially those born prematurely, have irregular breathing patterns with frequent pauses in their breathing -a manifestation of the developmental immaturity of brain respiratory control mechanisms (Abu-Shaweesh and Martin, 2008;Picone et al., 2014;Williamson et al., 2021).Apnoea -often defined as a pause in breathing of at least 20 s, or shorter if accompanied by bradycardia or cyanosis (Committee on Fetus and Newborn.American Academy of Pediatrics, 2003;Elder et al., 2013) -is one of the commonest neonatal respiratory emergencies, and occurs frequently in very preterm infants (Eichenwald, 2016;Pergolizzi et al., 2022), with the risk extending beyond term gestation in some infants (Eichenwald et al., 1997).Moreover, respira-tory conditions are a common reason for admission to the neonatal unit, for both term and preterm infants (Gallacher et al., 2016).Indeed, the term infant, despite having a relatively mature central nervous system (CNS) may present with acute respiratory events including apnoea (Levin et al., 2017) secondary to CNS related complications, metabolic, infectious or obstructive causes (Patrinos and Martin, 2017).
As the neonatal brain is developing, it is vulnerable to hypoxic insults, and pauses in respiration can impair cerebral perfusion and oxygenation (Choi et al., 2021;Horne et al., 2018;Martin et al., 2011).Apnoea and hypoxia have been associated with long-term effects, including poor neurodevelopmental outcomes (Janvier et al., 2004;Pillekamp et al., 2007;Poets et al., 2015).In contrast, pharmacological therapies including caffeine, aminophylline and doxapram (routinely used in the Newborn Intensive Care Unit either prophylactically or for treatment of apnoea of prematurity), can reduce the incidence of apnoea (Henderson-Smart and De Paoli, 2010) and the risk of adverse neurocognitive out-comes (Schmidt et al., 2007).However, these studies do not necessarily show that apnoea has a causal association with neurodevelopmental outcomes (Williamson et al., 2021).
Key to understanding the potentially cyclical relationship between brain development and apnoea is to identify how changes in respiration impact brain function, and to compare changes in brain activity during apnoea, with changes in brain activity during periods of other acute respiratory events (such as shallow breathing) (Williamson et al., 2021).Moreover, investigating changes in brain function with the use of respiratory stimulants will enable a greater understanding of how such interventions potentially mitigate the effects of apnoea on the developing brain and their neuroprotective function.Brain activity is essential for development -in animal models, blocking or reducing activity during critical periods of development leads to abnormal, weak, thalamocortical connectivity and a lack of structure within the cortex (Ghosh and Shatz, 1992;Kanold et al., 2003;Tolner et al., 2012).Altering the pattern of activity in a computational model of the developing cortex disrupts synaptic connectivity formation (Hartley et al., 2020).In human neonates, reduced cortical activity early in life is associated with poor neurodevelopmental outcomes (Ranasinghe et al., 2015;Wikström et al., 2012).Understanding whether and how brain activity is altered during apnoea, how this compares with brain activity changes during other respiratory events, and how brain activity is altered by respiratory stimulants will help elucidate the mechanism by which apnoea impacts the developing brain.
We conducted a systematic review to determine the current understanding of how brain activity recorded using EEG changes during periods of acute respiratory events (e.g., apnoea, periodic breathing, shallow breathing, tachypnoea, bradypnea, etc.) and with the use of respiratory stimulants in human neonates between 28 and 42 weeks postmenstrual age (PMA).Specifically, we aimed to identify the current knowledge regarding (1) how respiratory and EEG signals co-vary in neonates across different PMA during normal breathing, (2) the differences in the EEG signals of neonates between periods of normal respiration and periods of acute respiratory events, (3) the relationship between the characteristics of acute respiratory events (e.g., duration, degree of severity) and brain activity changes, and (4) the effect, if any, of respiratory stimulants such as caffeine citrate, doxapram, or aminophylline on EEG features.

Methods
This systematic review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al., 2021).The protocol for this systematic review was registered in PROSPERO (CRD42022339873) (Usman et al., 2022).

Exposure and comparison group
The exposure variables were defined as any abnormal, irregular, or dysfunctional acute respiratory event such as apnoea, periodic breathing, tachypnoea, bradypnea, sighing; and any breathing pattern such as shallow breathing, hyperpnea or respiratory-related muscle contractions, like hiccups.Geographical and institutional variations in the definition of operational terms exist and any article with a clear description of the exposure variable was included.
We considered within-subject comparison with a period of normal respiratory recording, (i.e., respiratory-event-free period), or a between-subject comparison with a neonatal cohort having normal respiratory rate, (i.e., the average rate between 40 and 60 breaths per minute and not experiencing any abnormal respiratory event) as a valid comparison for any exposure.

Review inclusion and exclusion criteria
All study designs reporting on acute respiratory events or the use of respiratory stimulants and EEG-recorded brain activity in human neonates between 28 and 42 weeks postmenstrual age (PMA) were included.We excluded the following: review articles, systematic reviews, commentaries, questionnaires, survey reports, and studies with wrong study population (animals, non-neonatal e.g., adults or older children).We also excluded studies involving neonates with cardiovascular malformations; neurological abnormalities (e.g., central nervous system malformations, seizures, hypoxic-ischemic encephalopathy), and intraventricular haemorrhage.Additionally, for the comparison group we excluded studies that involved neonates with pneumonia, bronchiolitis, congenital respiratory malformations or any abnormal respiratory event.Finally, we excluded studies that used EEG recordings for sleep staging only.

Search strategy
We used a combination of Medical Subject Headings (MeSH) terms and controlled keywords including three subject domains, namely EEG, breathing and infant categories; and an appropriate database-specific search strategy across eight databases (MEDLINE, Embase, Global Health, PsycINFO, Cumulative Index of Nursing and Allied Health Literature, Cochrane Library, Science Citation Index and Conference Proceedings Citation Index via Web of Science, and the grey literature on ProQuest).All search strategies are provided in full in Appendix A. Only studies reported in English were included with no publication year restriction up to 2022.The initial search was conducted in January 2022 and updated in August 2022 to capture recent publications.We also performed a backward citation search of all included articles.

Study selection
Article selection was performed using Rayyan software for systematic reviews (Ouzzani et al., 2016) based on the prespecified inclusion and exclusion criteria.Following de-duplication on Rayyan of the uploaded articles from the search results, initial title and abstract screening was done by two reviewers (F.U., S.M.).Before formal screening commenced, piloting of the selection process on a subset of 20 randomly selected articles against the set inclusion and exclusion criteria was performed by the two reviewers and the arbitrator to ensure the selection criteria were applied consistently.Each reviewer independently screened the articles for inclusion and was blinded to the decision by the other reviewer.At the end of the first selection process, the reviewers' decisions were compared, and discrepancies were resolved through discussion.Where differences still existed after discussion, a third reviewer (C.H.) acted as an arbitrator.Following title and abstract screening, full-text screening was conducted of the remaining articles independently by the same two reviewers.The arbitrator resolved any discrepancies.The reason for exclusion was documented for each ineligible article.

Data extraction
Review-specific data extraction fields were developed, piloted, and refined before final data extraction.Information was transcribed onto an excel sheet including the first author's name, journal title, year, country, and World Health Organisation region of study; study period, study design, sample size, age (gestational, postmenstrual or postnatal), infant category (term or preterm) and sub-population (e.g.infants with low-birth weight) of infants studied (where specified); type of exposure variable, including def-inition, comparison group (where stated), type of respiratory stimulant used (if any); EEG type (conventional or amplitudeintegrated), montage, number of channels, corresponding EEG findings; and cardiovascular monitoring where reported.No authors were contacted because the selected articles detailed all relevant information.

Outcome measures description
The primary outcomes were any EEG feature and changes thereof recorded using conventional and amplitude-integrated EEG and related to neonatal respiration, namely (but not limited to): 1. Band power (absolute and relative band powers of the full EEG signal in filtered bandpass range, and the delta, theta, alpha and beta bands); 2. Frequencies (average frequency of the full EEG signal; and in the delta, theta, alpha and beta band, 90 % spectral edge frequency); 3. Amplitudes (minimum, maximum and average absolute amplitude in the delta, theta, alpha and beta bands); 4. Continuity pattern (inter-burst interval duration, degree of continuity, occurrence of burst suppression patterns); 5. Entropy (multiscale entropy, approximate entropy, sample entropy, power spectral entropy).

Risk of bias (Quality) assessment
Two reviewers (F.U., C.H.) independently evaluated the risk of bias for the included studies using the Joanna Briggs Institute (JBI) critical appraisal tool for systematic reviews (Aromataris and Munn, 2020).This tool was chosen as it offers a range of checklists for different study designs.Each study-specific appraisal checklist had the options of ''yes", ''no", ''unclear", or ''not applicable" -discrepancies between reviewers were resolved through discussion.Due to the low number of studies identified for inclusion in the review, we used the JBI tools to assess the quality of conduct and reporting and not to make exclusion decisions for the included articles.Checklist results are given for all included studies in Appendix B.

Data synthesis
A PRISMA flowchart was used to depict the outcome of the database searches and the selection process.A narrative approach described by the Centre for Reviews and Dissemination for systematic reviews (Centre for Reviews and Dissemination, 2009;Popay et al., 2006) was used to summarise all empirical evidence detailing the relationship between respiratory changes and EEG features.Details for all included articles were summarised in tables.A thematic description of the study findings was discussed while exploring variations in baseline characteristics of the study population, methodology, and differences in exposure and outcome measures.During narrative synthesis, three distinct themes were identified in the articles identified -those investigating EEG changes during apnoea, those investigating EEG changes during other respiratory events, and those investigating EEG changes to respiratory stimulants; these themes are described separately in the results.Finally, the limitations and strengths of the review were explored using the quality and data completeness of the included studies.

Results
The electronic databases search identified 234 relevant articles after removing 33 duplicates.Most (n = 176, 75.2 %) were excluded following the title and abstract review, and a further 46 (19.6 %) were excluded following the full-text screen.Overall, 14 articles comprising a total of 537 infants were included in the review synthesis (Fig. 1).Half of the studies (n = 7/14, 50 %) involved preterm infants only -with one study focusing on low-birthweight infants, while one (7.1 %) study included term infants only.Most of the publications were from Europe (n = 9/14, 64.3 %), almost a quarter from the region of the Americas (n = 3/14, 21.4 %) and one each from the Eastern Mediterranean and Western Pacific regions (n = 1/14, 6.3 %, each).No studies were reported from Africa.Only one study (7.1 %) was an RCT while the others were descriptive in design (cross-sectional, n = 8/14, 57.1 %; cohort studies, n = 4/14, 28.6 %; or case reports, n = 1/14, 7.1 %).Two studies were reported as conference abstracts only.
Most studies (n = 9/14, 64.3 %) focused primarily on the relationship between apnoea and EEG activity, one (7.1 %) reported other respiratory changes (hiccups) while just over a quarter (n = 4/14, 28.6 %) investigated the effect of respiratory stimulants on EEG features.We did not identify any study that examined how normal respiration and EEG signals co-vary in neonates.The EEG measures evaluated were diverse (Tables 1-3).Conventional EEG (cEEG) was used in 9 studies (64.3 %), amplitude-integrated EEG (aEEG) in 3 studies (21.4 %), one (7.1 %) study used both aEEG and cEEG, and in one study (7.1 %) the EEG method was not specified (this was a conference abstract).A bipolar EEG montage was used during most (n = 10/14, 71.4 %) recordings.Of note, aEEG was used in all 4 studies investigating the effect of respiratory stimulants (Table 3).

Effect of apnoea on EEG brain activity
Nine of 14 (64.3 %) articles described the effect of apnoea on brain activity.All were observational studies with a combined sample size of 209 infants; over half (n = 5/9, 55.6 %) of the studies assessed both term and preterm infants (Table 1).All studies had a within-subject control, comparing EEG epochs during periods of normal breathing with apnoeic episodes.The definitions used for the exposure variable (apnoea), where stated, were diverse.Apnoea was defined in some articles to be as short as 3 s in duration, whereas other articles defined such periods as respiratory pauses, and apnoea as a pause of at least 20 s (Table 1).
Of the 9 studies, simultaneous apnoea and EEG suppression were reported in three articles (one of which is a conference abstract) (Deuel, 1973;Low et al., 2012;Wulbrand and Bentele, 1994), a generalised reduction in EEG amplitude was reported in two studies (Bridgers et al., 1985;Fenichel et al., 1980), and a reduction in the mean and standard deviation of absolute EEG power in one study (Schramm et al., 2000).Holthausen et al. (1999) reported a decrease in amplitude only in the delta and theta bands.One study showed a significant post-apnoea frequency change, with some infants having an increase -while others a decrease -in frequency compared with the control period one minute before the apnoea (Curzi-Dascalova et al., 2000).Significant increases in the magnitude squared coherence, and nonlinear generalized synchronization index during apnoea relative to control periods were also reported (as a conference abstract) (Manas et al., 2014).
In contrast, no change in EEG brain activity was observed during most episodes of apnoea studied by Fenichel et al. (1980) (n = 28/35 apnoeas, 80 %) and Low et al. (2012) (n = 6/8 apnoeas, 75 %).Similarly, no change in amplitude in some apnoeas (proportion not provided by authors) was reported by Bridgers et al. (1985) and in 62.7 % of apnoeas observed by Curzi-Dascalova et al. (2000).Only one study (Holthausen et al., 1999) compared preterm infants based on age and reported a significant reduction in theta and delta amplitude during apnoea in term infants !41 weeks; this change was not significant in younger infants.

Effect of other respiratory changes on EEG
One of the 14 studies (7.1 %) reported respiratory changes other than apnoea, describing EEG-recorded evoked responses to hiccups (Whitehead et al., 2019) (Table 2).This was a cross-sectional study involving 13 term and preterm infants (10 of whom met the inclusion criteria for this review).The authors identified three distinct hiccup-related potentials with peaks occurring at 16, 125 and 310 ms after contraction of the diaphragm, predominantly in central regions.

Relationship between characteristics of acute respiratory events and brain activity
Three studies (Deuel, 1973;Fenichel et al., 1980;Low et al., 2012) evaluated EEG changes relative to apnoea severity; all reported EEG amplitude decreases and burst suppression during some, but not all, long apnoeas, and, in some cases, in shorter apnoeas.The description of apnoea duration in the studies was non-uniform.Fenichel et al. (1980) defined long (!20 s) and short ( 19 s) apnoeas, and reported 6 of 19 short apnoeas had mild amplitude suppression, and amplitude suppression only during 1 out of 16 long apnoeas.Deuel (1973) reported burst suppression at the start of respiratory pauses and apnoeas (Table 1).Low et al. (2012) found in one infant (case report) that changes in oxygen saturation below 20 % during two episodes of prolonged apnoea requiring resuscitation were associated with complete   Outcome measures: Absolute power, relative power, median frequency, and peak frequency for sub delta (0.4 ± 1.5 Hz), delta (1.5 ± 3.5 Hz), theta (3.5 ± 7.5 Hz), alpha (7.5 ± 12.5 Hz), beta 1 (12.5 ± 19.5 Hz), and beta 2 (19.5 ± 25.0 Hz) band using spectral analysis.Sleep states were automatically annotated using a polysomnographic device.
A significant reduction in the mean and standard deviation of absolute EEG power during apnoeas in the sub delta, delta, theta and alpha frequency bands compared to before and after periods of apnoea and the apnoeafree phases of active sleep was observed.In the beta 1 band, significantly lower absolute power during the apnoea compared with before apnoea and the apnoeafree periods occurred (not different to after the apnoea).No significant power differences in the beta 2 band were observed.The highest relative reduction of 45 % was in the theta band.There were no differences in the means and standard deviations of peak frequencies in all the frequency bands.
No  EEG burst suppression; this was not observed in shorter episodes of apnoea in the same infant.Most (n = 7/9, 77.8 %) of the studies simultaneously recorded heart rate and/or oxygen saturation with EEG monitoring (Table 1), however, two of these studies (Schramm et al., 2000;Bridgers et al., 1985) did not describe the vital signs changes in relation to brain activity changes.Fenichel et al. (1980) did not observe a consistent relationship with heart rate changes and EEG suppression and Curzi-Dascalova et al. (2000) reported no correlation between changes in EEG frequency during apnoea and changes in heart rate or oxygen saturation (Table 1).Deuel (1973) reported that EEG suppression occurred simultaneously with apnoea, whereas bradycardia occurred approximately 15 s later in the single infant that met this review's inclusion criteria (Table 1).
Curzi-Dascalova et al. (2000) found that changes in EEG frequency were dependent on apnoea type (with greater EEG changes following obstructive apnoea compared with central apnoea), amplitude modification and baseline EEG frequency in their study of five infants, but frequency changes were not related to apnoea duration or sleep state.

Studies reporting on the effect of respiratory stimulants on EEG
Four of 14 (28.6 %) of the included studies described the effect of respiratory stimulants (used for apnoea treatment or given prophylactically) on the EEG in a total of 315 infants (Table 3).Caffeine in varying doses was used in all four studies (Table 3), and aminophylline use was reported in one study (Yang et al., 2019).All four studies were of preterm infants, one was an RCT (Yang et al., 2019) and three were cohort studies (Dix et al., 2018;Hassanein et al., 2015;Supcun et al., 2010).
The studies summarised EEG changes before, during, and after the use of respiratory stimulants, and all used amplitudeintegrated EEG.None of the included studies found a significant difference in the apnoea rate or respiratory rate after stimulant introduction compared with before, however, in many cases the stimulant was given prophylactically.The aEEG features assessed included changes in continuity pattern, arousal from sleep (Hassanein et al., 2015;Yang et al., 2019), spontaneous activity transient (SAT) rate and interval, (Dix et al., 2018), aEEG amplitude (Dix et al., 2018;Supcun et al., 2010;Yang et al., 2019), voltage and bandwidth (Yang et al., 2019).Importantly, no respiratory stimulant studies investigated how changes in EEG during apnoea were altered with stimulant use -they only investigated EEG changes in relation to the start of drug therapy.They also did not explore simultaneous interactions between EEG and changes in heart rate, oxygen saturation or respiration.

Quality assessment of the included articles
Comments on quality assessment for each of the included studies are summarised in Tables 1-3 and the results of the JBI checklists are given in Appendix B. Three papers (Curzi-Dascalova et al., 2000;Low et al., 2012;Whitehead et al., 2019) were very well described and clear in terms of subjects, exposure and outcome measures.Other studies lacked detail in one or more key areas, such as subject inclusion criteria, methodology related to respiratory measurement, EEG measurement or analysis.Two papers (Manas et al., 2014;Wulbrand and Bentele, 1994) were conference abstracts and so details were very limited; we chose to include these studies as details were sufficient for inclusion, however, it should be noted that these were not full articles.Several studies (Bridgers et al., 1985;Deuel, 1973;Fenichel et al., 1980) used visual EEG assessment (in part as they were from the 1980s or before) -this is the clinical gold standard, nevertheless, it is subjective and may lack reproducibility.Statistical analysis was lacking in several studies, and where statistical analysis was used, most papers did not appear to correct for multiple comparisons.

Discussion
This systematic review aimed to investigate the effect of both acute respiratory events and respiratory stimulants on brain activity recorded using EEG in neonates.We identified 14 studies conducted up until August 2022, involving a relatively small number of infants.Nine studies investigated EEG changes in relation to episodes of apnoea.Neonatal apnoea-related EEG changes were inconsistent, with EEG suppression, amplitude and post-apnoea frequency reduction observed during some, but not all, episodes of apnoea.The factors that drive these differences are yet to be elucidated, with differences across studies in definition of apnoea, and other changes in vital signs (e.g., bradycardia and desaturation), making comparison across studies difficult.One study investigated other respiratory events -characterising evoked cortical responses to hiccups, observed in both preterm and term infants.Four studies investigated EEG changes in relation to respiratory stimulants, demonstrating increased EEG continuity, arousability, lower edge amplitude and boundary voltage values and decreased SAT interval, bandwidth and narrow-band upper boundary voltage.
EEG suppression, amplitude and frequency reduction during some apnoeic episodes suggests neuronal desynchrony with diffuse brain inactivity, most likely from cerebral anoxia (Shanker et al., 2021).Brain activity is critical in brain development (Ghosh and Shatz, 1992;Kanold et al., 2003;Ranasinghe et al., 2015;Tolner et al., 2012;Wikström et al., 2012) and apnoea has been associated with poor neurodevelopmental outcomes (Greene et al., 2014;Janvier et al., 2004;Pillekamp et al., 2007;Poets et al., 2015).Nevertheless, the extent to which apnoea contributes to (and is not just correlated with) poorer outcomes remains unclear (Erickson et al., 2021;Williamson et al., 2021).EEG suppression during apnoea may be neuroprotective to reduce cell energy consumption (Low et al., 2012), however, very long periods of oxygen deprivation, such as those in hypoxicischaemic brain injury, are associated with prolonged EEG suppression, which in turn is an indicator of poor outcome (Dereymaeker et al., 2019;Douglass et al., 2002).It is not known how long an apnoea can last in an infant without brain damage occurring (Low et al., 2012) or whether periods of EEG suppression during apnoea are related to neurodevelopmental outcomes.While Low et al. (2012) reported in a single infant, complete EEG suppression with extremely low oxygen level during a prolonged apnoea, Fenichel et al. (1980) did not observe a significant relationship between EEG changes and apnoea duration.The latter study also had a small sample size and the relationship with other factors such as the degree of oxygen desaturation and bradycardia (Pichler et al., 2003) which likely play an important synergistic role in brain function, were not clear.Curzi-Dascalova et al. (2000) found no correlation between EEG frequency changes and changes in heart rate or oxygen saturation.However, there is likely a complex interplay between factors such as apnoea duration, bradycardia, and oxygen desaturation with EEG changes.Further work in larger samples of infants is needed to understand these relationships.
EEG activity changes greatly with gestational age from a relatively discontinuous pattern in the extremely preterm infant to a more continuous pattern by term (André et al., 2010).Only one study considered EEG changes during apnoea in relation to the age of the infant, observing a significant amplitude reduction in infants > 41 weeks (Holthausen et al., 1999).In younger infants, similar trends were observed; the authors may have been underpowered to observe a significant effect in these age groups.To date, studies have not investigated if there is a relationship between the number of apnoeas an infant experiences and EEG changes, and it is unclear if the frequency of apnoea impacts brain function.
We only identified one study that investigated other acute respiratory events - Whitehead et al. (2019) identified event-related potentials in response to hiccups in both term and preterm infants.This demonstrates that contraction of respiratory muscles provides input to the brain, even in young preterm infants, which may drive the establishment of interoceptive circuits (Whitehead et al., 2019).Further studies in this area and those examining respiratory interactions during periods of normal breathing are key to understanding the formation of brain-respiratory networks in preterm infants and factors that disrupt their development.Moreover, we did not identify any studies which examined other respiratory events, such as shallow breathing or periodic breathing.Further studies in this area will enable a comparison with findings on the effect of apnoea on brain function.
We included four studies that investigated the effects of respiratory stimulants on the EEG.We excluded three other papers investigating the effect of respiratory stimulants on the EEGtwo which examined the effect of respiratory stimulant use on sleep states only (Curzi-Dascalova et al., 2002;Seppä-Moilanen et al., 2022) and one study which involved neonates with clinical seizures, cardiovascular, and CNS abnormalities (Czaba-Hnizdo et al., 2014).Of the included studies, one study -an RCT (Yang et al., 2019) -compared the effect of caffeine and aminophylline, while the other three (Dix et al., 2018;Hassanein et al., 2015;Supcun et al., 2010) studied the effect of caffeine only.The four studies did not specifically look at EEG changes during apnoea or in relation to respiration, rather, they evaluated the effect of methylxanthines on EEG activity.Although previous studies have reported decreased apnoea rate with the use of methylxanthines (Henderson-Smart and De Paoli, 2010), none of the included studies found a significant difference in the apnoea or respiratory rate after respiratory stimulant use compared with before, though this may be because in many cases stimulants were given prophylactically soon after birth.All four studies showed changes in the EEG with the use of respiratory stimulants, but further work is needed to ascertain whether and how these changes are related to other factors, such as age, dose, and comorbidities, and to understand how these changes in EEG might affect brain development.Understanding the effect of respiratory stimulants on brain activity may help drive individualised drug dosing in infants (Hartley, 2021;Hartley et al., 2021;Kumar et al., 2020).
This review was limited by the small number of studies, many of which had very small sample sizes.Two of the papers were conference abstracts with limited information (Manas et al., 2014;Wulbrand and Bentele, 1994), one was a case report (Low et al., 2012) which have known study design limitations (Nissen and Wynn, 2014), and three (Bridgers et al., 1985;Deuel, 1973;Fenichel et al., 1980) were published 40 to 50 years ago, and so findings may be obsolete with improvements in medical technology.There were diverse definitions for the exposure variable i.e., the definition of apnoea -with wide variation in minimum duration, different EEG outcome measures, methods of analysis (visual versus quantitative assessment) and methods of apnoea detection (which may also be unreliable in infants (Adjei et al., 2021)) making cross-study comparison challenging.The heterogeneity of studies also prevented a meta-analysis and power calculations with pooled data for future studies.Some studies used few EEG channels which can result in important loss of spatial information (Cherian et al., 2008;Ebersole et al., 1983).Further, the studies investigating respiratory stimulants used aEEG, which is less sensitive than cEEG (Falsaperla et al., 2022;Kadivar et al., 2019).To our knowledge, this is the first review summarising existing data on this topic.Our findings highlight the need for future large prospective research investigating the relationship between EEG and respiratory activity in infants.

Conclusion
Evaluating the relationship between brain activity and respiration in newborn infants is key to understanding the impact of apnoea on neurodevelopment and the potentially cyclical relationship between brain development and respiratory dynamics.Current studies in this area are limited by small sample sizes, and inconsistent definitions of apnoea and EEG methods make study comparisons challenging.Further research is needed to explore the impact of a multitude of factors, such as age, duration of apnoea and oxygen saturation on changes in brain activity.Current studies investigating the effect of respiratory stimulants on brain activity in infants are limited and have investigated the effect of start of treatment on brain activity, with no studies exploring brain activity changes during apnoea.Further research is needed to investigate how brain activity changes are related to factors such as drug dose and to gain a mechanistic understanding of how changes in brain activity with stimulant use impact on brain development.Understanding the effects of apnoea and optimising treatment options is essential to improving the long-term outcomes of premature infants.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Were patient's demographic characteristics clearly described?Yes (excluding race) Was the patient's history clearly described and presented as a timeline?Yes Was the current clinical condition of the patient on presentation clearly described?Yes Were diagnostic tests or assessment methods and the results clearly described?Yes Was the intervention(s) or treatment procedure(s) clearly described?
Not applicable Was the post-intervention clinical condition clearly described?
Not applicable Were adverse events (harms) or unanticipated events identified and described?
Not applicable Does the case report provide takeaway lessons?Yes Unclear: aEEG methods clearly described, however, EEG assessment was visual.
Was the follow up time reported and sufficient to be long enough for outcomes to occur?
Unclear: Follow-up was for two hours after caffeine administration.Half-life of caffeine is much longer.
Unclear: Follow-up was for six hours after caffeine administration.Half-life of caffeine much longer.
Unclear: Follow-up for two hours after caffeine administration.Half-life of caffeine much longer. Yes.
Was follow up complete, and if not, were the reasons of loss to follow up described and explored?
Yes: Loss to follow up of caffeine group at 36 weeks not clearly reported, but this was not relevant for our review.

Yes.
Yes. Yes.Was appropriate statistical analysis used?
Yes, but no adjustment for multiple comparisons.
Yes, but no adjustment for multiple comparisons.
Yes, but no adjustment for multiple comparisons. Yes.

Fig. 1 .
Fig. 1.Prisma flow chart showing search outcome and article selection process for the review.

Table 1
Summary of included studies reporting the effect of apnoea on neonatal EEG features.

Table 2
Summary of included studies reporting on the effect of other respiratory changes on neonatal EEG.

Table 3
Summary of included studies reporting the effect of respiratory stimulants used for apnoea on neonatal EEG features.

Table B .
3 exp Respiration/ or (breath* or respirat* or apno* or apne* or breath* cess* or stop breath* or breath* hold* or period* breath* or inter breath interval* or interbreath interval*).mp. 4 1 and 2 and 3. 5 limit 4 to english language.1: JBI Critical Appraisal Checklist for the Included Case Report Studies

Table B .
2: JBI Critical Appraisal Checklist for the Included Cohort Studies F.Usman, S. Marchant, L. Baxter et al.