Newborn auditory brainstem response and sudden infant death syndrome

Sudden infant death syndrome (SIDS)—the sudden and unexplained death of a seemingly healthy infant, <1 year old—may be associated with abnormalities in the brain regions that underlie breathing and arousal during sleep. While post‐mortem studies suggest abnormalities in SIDS infants' brainstems, there are no studies of these infants' brainstem function before death. One way to assess the function of the brainstem is with auditory brainstem response (ABR), a routine hearing‐screening method that noninvasively measures the brainstem's response to sound. We hypothesize that anomalies in newborns' ABR measures may predict SIDS. Indeed, previous studies identified abnormalities in ABR characteristics in small samples of near‐miss SIDS infants hospitalized for infant apnea syndrome. However, there is a need to examine the ABRs of infants who died of SIDS. Therefore, in the current study, we propose integrating two secondary datasets to examine newborns' ABRs (N = 156,972), including those who later died of SIDS (n = ~42; .27 out of every 1000 infants), using existing archived records of neonatal ABR results from a sample of newborns born in Florida. We hypothesize that infants who die from SIDS are more likely than non‐SIDS infants to have abnormal ABRs as newborns. Understanding the association between SIDS and ABR may facilitate more accurate identification of an infant's risk for SIDS at birth, enabling increased monitoring, which may facilitate interventions and improve survivorship.


| INTRODUCTION
Sudden infant death syndrome (SIDS) refers to the sudden death of a healthy infant under 1 year of age, usually during sleep, with no explanation as to why the death occurred after investigation, autopsy, and review of clinical history.In 2022, over 1400 infants died of SIDS in the U.S. (Ely & Driscoll, 2023), with rates rising disproportionately among Black infants since 2015 (Shapiro-Mendoza et al., 2023).Better identifying early risk factors for SIDS may aid in understanding the underlying causes and facilitate interventions.
The causes of SIDS are complex and not yet understood (for a review: Moon et al., 2022).The brainstem has been the focus of much research, as it may have a potential role in SIDS (for a review: Bright et al., 2018; also see: Kato et al., 2003;Khan et al., 1992;Poets, 2004;Schechtman et al., 1996;Sridhar et al., 2003).When respiratory failure occurs during sleep, the brainstem is supposed to trigger protective arousal and cardiorespiratory responses (Filiano & Kinney, 1994;Kinney et al., 2005;Kinney & Thach, 2009).However, in infants who died of SIDS, there may be incomplete arousal or failed response of the cardiorespiratory system, suggesting that there might be functional deficits in the brainstem (Kato et al., 2003;Khan et al., 1992;Poets, 2004;Schechtman et al., 1996;Sridhar et al., 2003).
In the current study, we propose to explore a potential early indicator of SIDS: the auditory brainstem response (ABR).ABR is a noninvasive, routine, universally used newborn hearing screening that measures brainstem function (Brama & Sohmer, 1977).If we can better understand the association between SIDS and ABR, we may more accurately identify which infants are at elevated risk of SIDS.This early identification could facilitate more targeted care, including better sleep monitoring, genetic screening, and parent education to reduce environmental risk factors (Glinge et al., 2023;Zhao et al., 2016).Early identification and intervention for those infants at higher risk of SIDS may increase survival rates and thereby save thousands of lives annually (Wigfield et al., 1994).

| The brainstem's potential role in SIDS
Numerous studies indicate that brainstem abnormalities may be implicated in SIDS (for reviews: Bright et al., 2018;Kinney, 2009;Machaalani & Waters, 2014;Paine et al., 2014).Infants who die of SIDS may have an incomplete arousal response-a process regulated by the brainstem.For example, one sleep clinic study in 7-to 19-week-olds found that infants who died of SIDS, compared to infants who did not die of SIDS, had fewer complete arousal responses during sleep (Kato et al., 2003).Three other studies reported similar observations in infants' cardiorespiratory records from home monitors.The first study recorded 2.5-to 10-month-olds' heart rate and respiration rates in waveforms which revealed bradycardia (slowing heart rate) in all infants before death, indicating a failure in the cardiorespiratory response mechanism of the brainstem (Meny et al., 1994).The second study recorded 1-to 6-month-olds' time and cause of the first alarm, respiratory rate, heart rate, respiratory patterns (regular or non-regular, or gasping), number and duration of gasps, time to first gasp, and time to a heart rate below 15 beats per minute, which revealed that most monitors were alarmed because of bradycardia in infants just before they died of SIDS (Poets et al., 1999).These infants did not auto-resuscitate (i.e., regulate breathing and heart rate without assistance) through the gasp autoresuscitation mechanism (which they defined as slower breath frequency, longer breathing duration, and higher gasping amplitudes).
The third study examined unusual gasping patterns, longer gasp duration, and a lack of heart rate increases after gasps.Infants (3 weeks to 21 months old) who died from SIDS had a weaker reaction in the gasp auto-resuscitation mechanism compared to healthy infants and those who died from other identified causes (Sridhar et al., 2003).
In addition to these physiological indicators shortly before death, post-mortem studies observed that infants who died of SIDS had brainstem abnormalities (Bright et al., 2018;Machaalani & Waters, 2014;Paine et al., 2014).For example, a series of studies reported a lower rate of serotonin (5-HT) receptor binding in the brainstem compared to infants who died of other causes which may prevent an infant from responding to low oxygen or high carbon dioxide levels in the blood (Kinney, 2009;Kinney et al., 2005;Kinney & Haynes, 2019;Ozawa & Okado, 2002).These abnormalities are theorized to be linked to the aforementioned essential processescardiorespiratory control, sleep regulation, and arousal-potentially due to disrupted development or maturational delays in the brainstem (Bright et al., 2018;Duncan & Byard, 2018).
Together, these studies point to disruptions in brainstem function as possible indicators of SIDS.However, no study has examined potential early identifiable markers of brainstem abnormalities well before life-threatening cardiorespiratory episodes or death.

| Auditory brainstem response
To prevent SIDS, we need to develop better protocols to identify infants' risk of SIDS before any documented symptoms.One way

Significance
Our study will compare the brainstem activity of newborns who later died of sudden infant death syndrome (SIDS) to healthy newborns.No prior research has identified markers of newborn brainstem abnormalities before life-threatening cardiorespiratory episodes or death.This study will be the first to assess functional brainstorm abnormalities in SIDS using newborns' auditory brainstem responses (ABRs) to predict an increased likelihood of dying from SIDS in a large sample of newborns (~150,000).A neural marker of SIDS, like ABR, could prevent numerous infant deaths with little burden on parents or clinicians.
to prospectively examine the health of the brainstem is with ABRalso known as brainstem auditory evoked potential (BAEP) and brainstem auditory evoked response (BAER; Brinsmead et al., 2010;Kileny et al., 1982;Orlowski et al., 1982).ABR is a routine, noninvasive procedure that measures the inner ear and brainstem response to sound for newborn hearing screening (Brama & Sohmer, 1977).
This procedure presents acoustic stimuli to each ear, and the infants' brainstem response is recorded with a scalp electrode.This test is performed at birth, does not harm the baby, and can be done while the child is awake or asleep.ABR is an established screening test for hearing used by approximately 40% of countries worldwide (Neumann et al., 2020).
ABR provides objective information about whether and where there is dysfunction in the neural circuitry connecting the inner ear and brainstem based on the characteristics of each specific waveform.See Table 1 for details.
Past research suggests that there is information in infants' ABR measures that may be associated with SIDS.Specifically, one report of case studies of near-miss SIDS infants (n = 10; 2-to 8-months old)-identified after hospital referrals as having prolonged episodes of sleep apnea with accompanying cyanosis (changes in body tissue color due to low blood oxygen)-revealed abnormal ABR recordings when compared to healthy infants (Orlowski et al., 1979).
Abnormalities were observed in near-miss SIDS infants' peak latency, wave amplitude, wave shape, peak presence, response stability, intra-ear peak latency, and inter-ear peak latency.No other irregularities in additional clinical measurements were reported-including routine laboratory studies, continuous monitoring in the intensive care unit, computerized tomography (CT) scan of the head, and electroencephalogram (EEG)-suggesting ABR may be unique in providing this information.Only ABR appeared sensitive enough to detect abnormalities.
Another study compared ABR hospital records of 36 infants (42 weeks to 12 months old) at risk of SIDS (identified as infants with infant apnea syndrome) to an age-matched control group of 25 infants (Orlowski et al., 1982).Infant apnea syndrome was identified in infants who previously experienced a life-threatening event characterized by a prolonged sleep apnea episode and who required resuscitation.Results showed bilateral (n = 15) and unilateral (left side n = 17) abnormalities in the ABRs from the at-risk infants.Specifically, infants at-risk of SIDS displayed greater latencies in the peaks of waves I, III, and V, and shorter amplitudes in the wave III peak.
Only two studies directly compared SIDS infants' ABRs to non-SIDS infants' ABRs.Stockard (1982) included 1 SIDS infant and found normal-for-age interpeak latencies for waves I-V, but wave I was prolonged-for-intensity, consistent with mild conductive hearing loss, which they interpret as a common transient finding in premature infants.Brinsmead et al. (2010) compared 18 SIDS infants to a large sample of non-SIDS infants, but only explored whether there was a pass or fail on the ABR test and did not examine the qualities of the ABR signal itself.They detected no association between failing the hearing test and dying of SIDS in their sample.
In sum, existing studies of at-risk SIDS cases suggest there may be structural and functional abnormalities in the brainstem, which may be evident in ABR hearing screening measures taken at birth.Yet, no clear association has been documented between ABR data records and actual SIDS cases as this association has yet to be studied (Dahl et al., 2021).

| Current study
We will explore the association between the brainstem and SIDS by comparing the newborn ABR waveforms (i.e., absolute latency to wave V's peak) of infants who later died of SIDS to infants who did not die of SIDS.To accomplish this, we propose integrating two secondary datasets: vital statistics (birth and death) records and ABR screening data.We hypothesize that newborns who die from SIDS, compared to non-SIDS newborns, will have abnormal ABRs.

Structures measured Brainstem parts and functions
Wave

| Secondary dataset linkage
We will obtain vital statistics (birth and death) record data from the Florida Department of Health (FDOH) and ABR data Pediatrix Medical Group.The ABR records include 156,972 newborns born in Florida from 2009 to 2015.Birth and death record data will be linked with ABR data.We will integrate these datasets to examine newborn ABRs, including those of infants who later died of SIDS.These two datasets will be linked using infants' names, dates of birth, birth locations (counties), and sex, using the Python Record Linkage Toolkit (De Bruin, 2019).We will hand-review cases to verify and maximize the number of record matches.Records will be excluded if the child was over 28 days old during the ABR test.After data linkage, the integrated dataset will be deidentified to maintain confidentiality (i.e., name, date of birth, and location information will be deleted).

| Newborn FDOH data
A request to the state of Florida was submitted for access to the infants' demographic data, including birth and death information.We will identify infants based on Florida's definition of SIDS: "Sudden Infant Death Syndrome, also known as SIDS, occurs when a healthy baby under the age of one year dies while sleeping, and no specific cause of the death is found" (Florida Department of Health, 2023).
We will exclude cases in which other causes of death were noted.
From these data, infants will be identified in the ABR data as SIDS infants if their death occurred under 1 year and the only cause of

| Newborn ABR data
We requested and received approval in June 2023 for ABR data from around 150,000 infants born in Florida from the Pediatrix Medical Group Soundata database.These screenings were performed using the Smart Screener-Plus2 (Intelligent Hearing Systems Corp) typically within 2 days after birth.We will use ABR recordings from the first test the infant passed with both ears simultaneously.If unavailable, we will use recordings from the last screening pass of each ear individually.It is common for infants to have more than one screening because of their state (e.g., crying, moving, and fighting), noise in the recordings, fluids in the middle ear, or vernix present in the ear canal (Chung et al., 2019).The ABRs were evoked with 100 μs clicks at 35 dB in a normal hearing level (nHL).Left and right ear responses were acquired simultaneously using different stimulation rates for each ear.The left ear was stimulated at a rate of 77 Hz while the right ear was stimulated at a rate of 79 Hz.In addition to raw ABR records, the screening results include system-automated response detection information and acquisition parameters (Delgado et al., 2021;Delgado & Lim, 2010).
An ABR waveform typically consists of three major peaks (I, III, and V) (Hall, 2006).We will analyze the left and right ears separately to control for any effects of the stimulation rates.We will primarily focus on peak V because it is the main component at the lower intensities used for screening (Worthington & Peters, 1980).An additional advantage of using peak V is that it is the easiest to label using an automated routine to find the Vn (negative component) automatically in large samples (e.g., Delgado et al., 2021).
These ABR data come from various hospitals, including private, semi-private, and public, metropolitan and rural, and serve socioeconomically diverse populations, from both large cities and smaller towns, across 17 counties in Florida.In these hospitals, ABRs are administered to all newborns.The Pediatrix ABR data include quality measures such as residual noise and artifact rejection sweep count that have already objectively assessed data quality.

| Data analysis
Data analyses will be conducted using R (version 4.3.2) and RStudio (version 2023.09.1).We will explore whether newborns' ABR responses predict later SIDS.We will conduct a logistic regression analysis with SIDS outcome as the dependent variable and the absolute peak latency to peak V from each ear as two primary predictors, which is appropriate at this screening intensity (Bachmann & Hall, 1998;Schwartz & Berry, 1985).If available in the SIDS infants, the wave morphology of the individual SIDS infants will be qualitatively contrasted with the grand average for the normative waveform.
Specific risk factors that put infants at elevated risk for SIDS were previously identified (Athanasakis et al., 2011).We will descriptively report the available risk factors in the current study (Table 2).When risk factor information is available for both groups (SIDS and control infants), we will include them as control variables in our logistic regression analysis (i.e., infant sex, gestational age, and nursery-status).Unfortunately, we will not have all risk factors of interest for both groups so we cannot control for all of these variables (e.g., maternal smoking and maternal alcohol consumption).
We conducted an a priori power analysis using G*Power version 3.1.9.6 (Faul et al., 2007) to determine the minimum sample size required to test the study hypothesis.Results indicated the required sample size to achieve 80% power for detecting a large odds ratio (4.14; i.e., quadrupling the odds of SIDS with every unit increase in the predictor), at a significance criterion of α = .05,was N = 84 (n = 42 for SIDS cases; n = 42 for controls) for a logistic regression analysis with SIDS (yes/no) as an outcome, including three infant control variables-infant sex, gestational age, nursery-status-that are expected to moderately correlate with newborn ABR absolute peak latency to peak V. Thus, the estimated population sample size of N = 156,972-with SIDS affecting .27 in every 1000 infants according to the Florida Department of Health (Florida Health Charts, 2023)-is adequate to test the study hypothesis.

| Missing data
Every infant discharged from the hospitals (studied here) was screened for hearing impairments using ABRs, so we do not anticipate missing ABR data to be a large issue.Only in rare circumstances, due to a technical issue (e.g., equipment failure) or an unsuccessful ABR attempt (e.g., an extremely fussy newborn), would we have missing ABR data.We will report how much ABR data are missing and for what reasons separately for SIDS and non-SIDS infants.In all other cases, ABR waveform peaks will be coded as present or absent.An absent waveform reflects the absence of the expected (typical) waveform (which is different from "missing data").
If we have more than 20% of missingness in the ABR data, we will use a random forest model as an alternative analysis.The random forest model is a widely used machine learning approach for data classification based on a set of predictors (Breiman, 2001).
This approach is effective in handling missing data and imbalanced categories (i.e., in the event we have too few SIDS cases to provide stable estimates; Hapfelmeier et al., 2014;Hapfelmeier & Ulm, 2014).

| Limitations and future directions
The current study will focus only on specific counties in the state of Florida in which ABR data were available.As our power analysis revealed, we may be underpowered to detect small effect size group differences, so null results should be interpreted with caution.However, given that there are no previous studies directly TA B L E 2 SIDS risk factors to be reported descriptively.

Risk factor Association with SIDS
Infant specific Sex a Males are at higher risk than females (Mitchell & Stewart, 1997) Gestational age a Premature, <37 weeks of gestation, are at higher risk than full-term (Malloy, 2007) Birth weight Low birth weight-<2500 g-are at higher risk than healthy birth weight (Malloy & Hoffman, 1995) Plurality Twins are at higher risk than singletons (Asif et al., 2023; although see Luke & Brown, 2006 who reported lower risk among multiples) Nursery-status a Neonatal intensive care unit (NICU) infants are at higher risk compared to non-NICU/well-baby infants (Friedmann et al., 2017) Parity Increasing parity increases risk (Lisonkova et al., 2012;Thompson & Mitchell, 2006) Parent specific

Maternal age
Higher risk if it is the first pregnancy and the mother is <20 years old compared to if the mother >20 years old (Willinger, 1995) Maternal race/ethnicity Higher risk among non-Hispanic Black mothers compared to mothers of other racial/ethnic backgrounds (Getahun et al., 2004;Leu et al., 2020) Maternal smoking (during pregnancy) Higher risk if the mother engaged in smoking during pregnancy compared to non-smoking mothers (Dybing & Sanner, 1999) Maternal alcohol consumption (during pregnancy) Higher risk with alcohol consumption during pregnancy compared to without consumption (O' Leary et al., 2013) Prenatal care Prenatal visits (number of visits) Higher risk with fewer prenatal visits or a lack of prenatal care (Getahun et al., 2004;Schlaud et al., 1996;Willinger, 1995) Note: We will report descriptive statistics for these factors to the extent that the available data allow us to do so.a Indicates variables that will be included as control variables in the logistic regression analysis.
comparing ABR measures of SIDS infants to non-SIDS infants, the current study reflects a first step.Nonetheless, replications across larger and other populations (e.g., outside the U.S.) will be necessary.While we cannot conduct a population-wide study because ABR data are not available for the entire population of Florida (or the U.S.) at this time, future research may identify ways to obtain a larger sample.However, because not all current hearing screening devices store ABR recordings, this will require further collaboration from other equipment manufacturers.
In the current study, we are unable to control for all variables related to SIDS.Ideally, we would include additional control variables associated with SIDS risk in our analysis, like quality of prenatal care (Getahun et al., 2004;Schlaud et al., 1996;Willinger, 1995).
However, we will descriptively report all available variables related to SIDS that are not accessible for both SIDS and control groups.
This will allow us to better generalize our population in comparison to other populations across the U.S.
Of note, SIDS is a catch-all category when no other cause of death can be identified, and therefore is heterogeneous, including various individual and environmental factors (Byard, 2018;Wojcik et al., 2023).While beyond the scope of the current study, research is needed to clarify the extent to which ABR is a useful predictor of each specific SIDS phenotype, as medical knowledge advances and SIDS subtypes are identified.

| Expected results and potential implications
Expected findings from the current proposed study may aid in pre- vention and intervention efforts by identifying infants' risk for SIDS soon after birth.If our results suggest that infants who die from SIDS are more likely to have abnormal ABRs as newborns, then ABR could be used as an early marker to facilitate identifying an infant's risk for SIDS at birth.With this identification, clinicians and parents could increase monitoring of infants during sleep and facilitate interventions to improve survivorship.D ECL A R ATI O N O F TR A N S PA R EN C YThe authors, reviewers, and editors affirm that in accordance with the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.