5.2 Transient-evoked OAEs and newborn screening
The transient-evoked OAE is the OAE in its simplest form: a brief impulsive stimulus excites the entire cochlea at once, and the cochlea returns a frequency-dispersed echo over the next 15–20 milliseconds. The stimulus is typically an 80 µs rectangular click, presented at 80–85 dB peak-equivalent SPL through a probe earphone sealed into the ear canal. The same probe contains a sensitive microphone (electret, low noise floor around 5–10 µV in the ear-canal volume) that records both the click itself and everything that follows. Averaging across hundreds of click repetitions extracts the response from the broadband ambient noise.
A brief click excites the entire cochlea simultaneously. Each cochlear place generates its own re-emission at its characteristic frequency, but with a place-dependent travel time — high frequencies (basal places, near the oval window) return first (~5 ms), low frequencies (apical places) return later (~16 ms). The recorded probe pressure is the sum of these latency-dispersed echoes. Damage to the outer hair cells at a particular cochlear region selectively suppresses the corresponding latency component. The newborn-screening pass criterion is typically 3 of 4 bands present at SNR ≥ 3 dB. Above ~30–40 dB HL of sensorineural loss in any band, the OAE in that band disappears entirely — making the TEOAE an extremely sensitive screen for mild cochlear pathology, but unable to characterise the *degree* of loss.
Why the response is latency-dispersed
The click is broadband — its spectrum is essentially flat from DC to several kHz. It enters the cochlea and sets up a traveling wave that moves from the basal (high-frequency) end toward the apical (low-frequency) end. Each cochlear place receives the click and re-emits at its characteristic frequency, but each place also takes a different amount of time for the traveling wave to reach it. From Hearing 4.4 — The place map refresher →, the traveling wave’s group delay roughly follows
where is a normalisation constant set by cochlear geometry; the higher the characteristic frequency, the earlier its echo arrives. For human cochleae:
| CF | Group delay |
|---|---|
| 4000 Hz | ~5 ms |
| 2000 Hz | ~8 ms |
| 1000 Hz | ~12 ms |
| 750 Hz | ~16 ms |
| 500 Hz | ~20 ms |
The recorded TEOAE waveform is the sum of these latency-dispersed band echoes. Frequency-domain analysis of the waveform (a windowed FFT, or band-pass filtering) decomposes the response into per-band amplitudes and noise floors. A typical clinical screener reports SNR (signal-to-noise ratio) in four to six octave or half-octave bands centred at 1000, 1500, 2000, 3000, and 4000 Hz; the pass criterion is generally 3 of 4 bands at SNR ≥ 3 dB, or stricter thresholds in some protocols.
The clinical procedure
The TEOAE test takes about 30 seconds per ear when conditions are favourable. The probe is sealed in the canal. The instrument:
- Calibrates the canal SPL by playing a calibration tone, computes the canal volume, and adjusts the click amplitude to deliver the target peak-equivalent SPL.
- Plays a sequence of about 260 clicks at 50 ms intervals.
- Averages the responses in two interleaved buffers (Buffer A: odd-numbered clicks; Buffer B: even-numbered clicks). The coherent signal (the OAE) shows in both buffers; incoherent noise (movement, swallowing, EMG) cancels between them.
- Reports per-band SNR by comparing the cross-buffer signal (A+B) to the cross-buffer noise (A−B).
The instrument refuses to declare a pass while the noise level is excessive; it pauses and resumes when the infant is quiet again. Total test time, including pauses, is typically 30 seconds to 2 minutes per ear.
Newborn hearing screening
The United States adopted universal newborn hearing screening (UNHS) in stages from 1999 to 2007; by 2026 essentially every birthing hospital in the developed world screens every newborn before discharge. The standard protocol is the 1-3-6 rule from the Joint Committee on Infant Hearing:
- Screen by 1 month of age.
- Diagnose by 3 months if the screen identifies a concern.
- Intervene by 6 months with hearing aids, cochlear implants, or sign-language exposure if confirmed.
The screen itself is a two-stage protocol. The first stage in healthy newborns is OAE (TEOAE or DPOAE). Infants who refer on the OAE stage repeat the OAE test; those who refer again receive a second-stage automated auditory brainstem response (AABR) test. The AABR has its own clinical chapter (Ch 6) but is the same principle: a non-behavioural, scalp-electrode test that gives a pass/refer decision. Infants who refer on AABR are referred to audiology for a full diagnostic evaluation.
The two-stage screen catches three populations:
- Conductive losses present at birth (effusion, atresia) — fail the OAE because middle-ear fluid blocks the signal; usually pass the AABR after a recheck once the fluid clears.
- Sensorineural losses above ~30 dB HL — fail both stages and are referred for diagnostic workup.
- Auditory neuropathy spectrum disorder (ANSD) — pass the OAE (outer hair cells healthy) but fail the AABR (auditory-nerve dyssynchrony); discovered only because the two-stage protocol catches the OAE-AABR dissociation. UNHS has approximately tripled the diagnosed incidence of ANSD by surfacing cases that would otherwise have been missed.
The screening yield is large. Roughly 1.4 of every 1000 healthy newborns has permanent congenital hearing loss; this number triples in NICU populations. Without UNHS, the average age at diagnosis was 24 months in the early 1990s; with UNHS, by 2010 the median age at diagnosis had fallen below 3 months and intervention had moved correspondingly earlier. Multiple longitudinal studies (Yoshinaga-Itano et al. 1998 and follow-ups) have shown that early-intervened deaf children have language outcomes equivalent to hearing peers, while late-intervened children show measurable delays — a number that has made UNHS one of the highest-leverage public-health interventions in pediatric medicine.
Limitations and pitfalls
TEOAE is sensitive but blunt. It cannot:
- Quantify the degree of loss when present. OAEs are present or absent in each band; if absent, the loss is “at least 30–40 dB HL” with no further resolution.
- Distinguish sensorineural from conductive loss without companion tympanometry and AABR.
- Detect cochlear loss confined to the apex (< 500 Hz), because the TEOAE has poor low-frequency response (the noise floor rises rapidly toward DC).
- Detect retrocochlear or central pathology at all.
For diagnostic depth, the DPOAE (next lesson) takes over: by sweeping primary frequencies, it produces a frequency-resolved analogue of the audiogram and quantifies the OAE amplitude rather than just present/absent.