A newborn hearing screen listens in two directions

Airman 1st Class Travis Beihl photographed his newborn daughter Mackenzie during a hearing test at Memorial Hospital in Gulfport, Mississippi, in 2016. The transducers and leads make physiological screening visible, but the source caption does not identify the test as OAE or AABR.[10]

The newborn in the cover photograph lies awake in her mother's arms, a knit cap pulled low and red and blue transducers held over her ears. The official caption calls the procedure a hearing test without naming its modality, so the photograph cannot tell us whether the equipment is listening for an acoustic echo or collecting an electrical response. That ambiguity makes it a useful entrance to this story.[10]

Another newborn hearing screen looks simpler. A soft probe sits in the ear canal, plays a sound and listens for energy made inside the ear to return. It needs no visible scalp electrodes. The two procedures are automated auditory brainstem response, or AABR, and otoacoustic emissions, or OAE. Both turn an infant who cannot follow instructions into a source of objective physiological evidence. They do not, however, ask the body the same question.[8]

That difference is the hinge of their comparative history. Newborn screening did not advance by replacing a primitive test with one universally superior machine. It learned to listen in two directions: outward from the active cochlea, and forward through synchronized neural transmission. OAE became a fast way to check the cochlear outer hair cells; AABR sampled more of the auditory pathway and could reveal a neural failure that a healthy cochlear echo might conceal. Each also acquired a blind spot. Modern screening protocols are the institutional memory of those limits.[6][8]

1971 and 1978: two signals become measurable

The brainstem signal entered the scientific record first. In 1971, physiologists Don Jewett and John Williston published Auditory-Evoked Far Fields Averaged from the Scalp of Humans. The title describes the essential move: a sound-locked electrical response, much smaller than the surrounding electrical activity, could be recovered by repeatedly presenting a stimulus and averaging what followed. Components arriving within the first several milliseconds could be recorded at the scalp and associated with activity below the cerebral cortex.[1]

That work supplied the foundation for auditory brainstem response testing. In a newborn screen, earphones or couplers deliver clicks or other brief sounds while surface electrodes collect voltage changes. Automation compares the accumulated waveform with the device's pass criteria. This is not a meter for whether a baby consciously “heard” or understood the sound. AABR measures sufficiently synchronized activity through the cochlea, auditory nerve and brainstem at a selected screening level; it does not test cortical perception, language or every possible hearing threshold.[8]

The second signal travelled the other way. In 1978, physicist David Kemp reported a delayed acoustic response recorded in the closed human ear canal after a near-threshold click. The slowly decaying component appeared after about 5 milliseconds, was present in the normal ears he tested and absent in ears with cochlear deafness. Kemp cautiously located its likely origin in a nonlinear cochlear mechanism.[2]

That caution still matters. An otoacoustic emission is not merely the test click bouncing off the eardrum. It is acoustic energy associated with the cochlea's active mechanics, particularly outer hair-cell function, that can be detected by a sensitive microphone in the probe. A reproducible emission is therefore evidence that this part of the auditory system is working at the tested frequencies and levels. It is not evidence that the auditory nerve carried the signal normally onward.[2][8]

Placed side by side, the discoveries offered something behavioral observation could not: useful evidence while an infant slept. But they opened different windows. OAE asked whether the cochlea produced an acoustic answer. ABR asked whether a synchronized electrical answer travelled into the brainstem. The eventual screening system would need both questions because neither was a synonym for hearing as a whole.

From high-risk lists to every nursery

For years, infant hearing programs concentrated on babies with recognized risk factors. That strategy was administratively manageable and biologically incomplete. A March 1993 National Institutes of Health consensus conference recommended universal newborn hearing screening. When an NIDCD working group translated that ambition into statewide protocol advice in 1997, it made the reason explicit: programs restricted to risk indicators identified only about 50% of infants with significant hearing impairment.[6]

The 1997 report accepted three routes—ABR, transient-evoked or distortion-product OAE, or a combination—and refused to declare a universal winner. It asked programs to judge the population, equipment, staff, test environment and follow-up capacity. It also set operational targets: after the initial screen and before outpatient rescreening, referrals should fall to no more than about 5% among newborns without risk indicators and 8% among those at risk. A screening method had to work not only in an acoustics paper but in a maternity ward, in each ear, before a family disappeared from the pathway.[6]

Two large programs showed that each technology could carry universal screening, although their results are not a head-to-head trial. From 1993 through 1996, Rhode Island used a two-stage transient-evoked OAE protocol across all eight maternity hospitals. Among 53,121 surviving newborns, the program reported permanent hearing loss at roughly two per 1,000. As the program matured, mean age at confirmation fell from 8.7 months to 3.5 months, and mean age at amplification from 13.3 months to 5.7 months.[3]

In Hawaii, James Mason and Kenneth Herrmann reported a five-year universal AABR program covering 10,372 infants. The nursery screened 96% of them. Its false-positive rate was 3.5% after the first screen and 0.2% after a two-stage procedure; bilateral loss requiring amplification was found in 1.4 infants per 1,000. The published experience demonstrated that electrodes, averaging and an automated pass rule could operate at nursery scale, not only in a specialist laboratory.[4]

By 2000, the Joint Committee on Infant Hearing had converted detection into a timetable: physiological screening by 1 month, audiologic diagnosis by 3 months and intervention by 6 months—the 1-3-6 goals. The statement noted that until shortly before then, average identification in the United States had been around 30 months. The decisive comparison was no longer OAE versus AABR in isolation. It was either objective screen versus waiting for behavior to make hearing loss unmistakable.[7]

Why the cheaper sequence developed a hidden cost

OAE has practical attractions in a well-baby nursery. The probe is quick to place, disposables can cost less, and no skin electrodes are required. Its sensitivity to the cochlea can also be a nuisance in the first days after birth: vernix in the ear canal, retained middle-ear fluid, poor probe fit, movement or environmental noise can produce a “not pass” that disappears on a technically clean repeat. A not-pass result is a referral signal, not a diagnosis.[8]

AABR samples a longer biological route. That gives it the crucial ability to flag auditory neuropathy, in which outer hair-cell activity may remain measurable while neural firing is absent or poorly synchronized. An infant with that pattern can pass OAE and not pass ABR. Yet AABR also requires a quiet infant, clean electrode contact and control of electrical and muscle noise. Common screening levels are slightly higher—about 40–45 dB HL in the 2019 JCIH review, compared with about 30–35 dB HL for OAE—so AABR is not simply the more sensitive test for every loss.[8]

Those tradeoffs encouraged a tempting two-stage design: screen everyone with OAE, then give AABR only to babies who do not pass. The sequence reduces the number sent out of the hospital for diagnostic work, because some babies who fail the more disturbance-sensitive OAE will pass AABR. It also creates a category that deserves attention: OAE not pass, AABR pass.

A seven-center study published in 2005 followed that category after 86,634 infants underwent a two-stage OAE/AABR protocol. Of 1,524 eligible infants enrolled for follow-up, 973 received diagnostic testing at a mean age of 9.7 months. Twenty-one infants, involving 30 ears, had permanent hearing loss; 77% of the affected ears had mild loss. On the study's assumptions, the authors estimated that about 23% of infants with permanent hearing loss at nine months could have passed this two-stage newborn protocol.[5]

The finding does not prove that every OAE-fail/AABR-pass baby has hearing loss; most do not. Nor was follow-up complete. It shows a narrower and more important point: when two screens use different physiology and thresholds, “passed the second test” cannot be read as “the first test was wrong.” The second stage can successfully reduce false referrals and still permit some real, predominantly mild losses to leave the screening pathway.[5][8]

The nursery split is a division of risk

The 2019 JCIH position statement preserves both tools but assigns them differently. In the well-baby nursery, either OAE or AABR, or both, can be used. OAE followed by AABR remains a common two-stage option. The committee describes auditory neuropathy in well babies as uncommon—an estimated 6 to 30 per 100,000 births, with limited and variable evidence—so programs must weigh that rare miss against equipment access and the larger danger that delayed rescreening becomes no rescreening.[8]

The neonatal intensive care unit changes the balance. NICU infants have more elevated hearing thresholds and greater risk of auditory neuropathy, including risk associated with severe hyperbilirubinemia and some intensive-care exposures. JCIH therefore recommends AABR for infants who receive NICU care. This is not a claim that AABR is superior in the abstract. It is a decision to sample the neural pathway where neural disorder is more plausible.[8]

Even there, one technology cannot erase the other's information. OAE can reveal outer hair-cell function and may flag mild cochlear loss below a particular AABR screening level. AABR can reveal disordered synchrony that an OAE-only screen misses. A full pediatric diagnostic evaluation may use ABR, OAE, middle-ear measures and later behavioral testing together. The automated nursery versions are sorting gates; they are not shortened diagnostic examinations.[8]

This is the central historical comparison. OAE and AABR survived together because their disagreement is sometimes information. A cochlea can answer while the nerve fails to synchronize. A brainstem response can meet a device's pass rule while a milder cochlear loss sits below that rule. The responsible protocol does not pretend those cases are impossible. It chooses which miss is least acceptable for a given population, controls referrals, and builds a route for the children who need another look.

A pass is an event; screening is a system

Near-universal coverage can hide a weak handoff. In the CDC's 2022 summary from 55 U.S. jurisdictions, 3,547,774 of 3,616,776 births, or 98.1%, had documented hearing screening; 95.6% were screened before 1 month. The final or most recent screen was not passed by 62,508 infants.[9]

Downstream, the record thinned. Among infants who did not pass or who went directly to diagnostic evaluation, 55.2% had a documented diagnosis, while 34.0% were classified as lost to follow-up or lost to documentation. These categories cannot tell us how many children received care that a registry failed to capture. They do show that a hospital's successful pass/not-pass event and a public-health program's successful identification pathway are different achievements.[9]

Neither OAE nor AABR can detect every mild, frequency-specific, progressive or delayed-onset loss. Neither can arrange a pediatric audiology appointment, explain a result without panic, reconnect a family after a move, or begin language access and other chosen interventions. That is why the 1-3-6 sequence joins three verbs—screen, diagnose, intervene—and why continuing surveillance matters even after a pass.[7][8][9]

The sleeping infant makes the first step look almost effortless. Historically, it was anything but. One line of research learned to extract a neural response from scalp noise; another discovered that a healthy cochlea sends sound back into the ear canal. Public health then had to decide when each answer was enough, when disagreement mattered and how quickly a child could move from risk to understanding. The machines made hearing screenable before behavior could reveal it. Their limits made follow-up non-negotiable.

cronfeed.work