2.4 Masking and audiogram configurations

A practical complication in pure-tone audiometry: when the two ears have very different thresholds, a tone presented to the worse ear at a high level can be heard by the better ear via cross-hearing — the test reads the wrong ear. Masking is the technique that solves this. It is also one of the procedural details that consumes the most clinical time, because it has to be done carefully (the cure can be worse than the problem) and clinical decision-making about when to mask, how much to mask, and when to suspect masking is failing is a skill that takes years to develop.

This lesson develops the cross-hearing problem, the plateau method that solves it, and closes the chapter with the standard taxonomy of audiogram configurations and their clinical interpretation.

Cross-hearing and interaural attenuation

When a sound is presented to one ear at a level $L$ , some fraction of it reaches the other ear by transmission across the head. The amount lost in transit is the interaural attenuation (IA):

Air-conducted tones with supra-aural earphones: typically 40 dB (range 35–65 dB depending on frequency and individual anatomy).
Air-conducted tones with insert earphones: typically 55–70 dB (the insert phone leaves more head between the source and the contralateral cochlea).
Bone-conducted tones: approximately 0 dB. The skull is small relative to bone-conduction wavelengths, so both cochleae are excited essentially equally.

If a tone presented to the test ear at level $L_\text{test}$ reaches the non-test ear at level $L_\text{test} - \text{IA}$ , and that level exceeds the non-test ear’s own threshold, the patient hears the tone via the non-test ear — cross-hearing. The audiometer records a response, but the response came from the wrong ear. The threshold appears artificially low.

Concrete example: the patient’s left ear has a 70 dB HL threshold; the right ear has a 10 dB HL threshold. We present a 65 dB HL tone to the left ear with supra-aural earphones. The non-test ear receives $65 - 40 = 25$ dB HL — well above its 10 dB threshold. The patient responds. We record the left-ear threshold as 65 dB HL, but it’s actually higher; we’re measuring the right ear via the left earphone.

The masking solution

Add narrow-band noise to the non-test ear at a level that elevates its threshold above the cross-hearing floor. The noise prevents the non-test ear from hearing the cross-heard tone, forcing any response to come from the test ear.

This sounds straightforward but the practical execution is delicate, because the masker can cross-hear too. If the masker is too loud, it can exceed its own IA and elevate the test ear’s threshold — overmasking — and we end up measuring a threshold that’s artificially high.

The fundamental masking dilemma: the masker must be loud enough to silence the non-test ear, but not so loud that it leaks back across to the test ear. The window between those two conditions is the plateau.

The plateau method

The standard clinical procedure (Hood’s plateau method):

With no masker, find the apparent threshold of the test ear (which may include cross-hearing).
Add a small amount of masker (say 10 dB) to the non-test ear. Re-measure the test-ear threshold.
Continue increasing the masker in 10-dB steps, re-measuring the test-ear threshold each time.
Plateau: at some masker level, the test-ear threshold stops increasing. This is the true threshold. Three consecutive 10-dB increases in the masker that produce no further change in test-ear threshold define the plateau.
If the masker is increased further and the test-ear threshold begins rising again, overmasking has set in. Back off.

test ear true threshold = 60 dB HL non-test ear threshold = 15 dB HL

masker level in non-test ear = 20 dB HL

The audiologist needs the *true* threshold of the test ear (green dashed line). At low masker levels, the patient hears the test tone via cross-hearing through the non-test ear, and the audiometer reads an artificially low threshold (the blue trace climbs from low values up to the green line). The masking plateau (shaded gold) is the range of masker levels where the apparent threshold equals the true threshold — the masker has silenced the non-test ear but is not yet leaking back across to the test ear. Push the masker too high and overmasking begins to elevate the apparent threshold above the truth. The plateau method finds the threshold by sweeping the masker through this region. Interaural attenuation is taken as 40 dB for air-conducted tones (a standard clinical assumption); the BC route used by the crossover masker is taken as 0 dB.

The interactive lets you set the test-ear true threshold, the non-test-ear threshold, and the masker level. The blue curve is the apparent threshold the audiometer reads as a function of masker level: it starts low (cross-hearing dominates), rises into the plateau (the truth), and rises further when overmasking kicks in. The plateau is the diagnostic sweet spot. The state readout indicates which regime you are in.

When masking is needed

Not every audiogram needs masking. The decision rule:

AC masking is needed when $L_\text{AC, test} \geq L_\text{BC, non-test} + \text{IA}$ . I.e., when the AC level in the test ear is high enough to cross over and exceed the non-test ear’s BC threshold. (We use the non-test BC not AC, because cross-hearing happens via the BC route once the signal has crossed the head.)
BC masking is needed when there is any air-bone gap in either ear, because IA for bone conduction is 0 — the BC stimulus reaches both cochleae nearly equally regardless of where the oscillator is placed.

In practice, masking is required surprisingly often. A patient with a 30-dB unilateral conductive loss already requires AC masking to verify the loss is unilateral; BC measurements with any ABG require masking essentially always.

Audiogram configurations: the standard taxonomy

The audiogram is summarised by three descriptors: degree (severity, by PTA), type (sensorineural / conductive / mixed, by ABG), and configuration (shape).

The canonical configurations:

Flat

Thresholds within roughly 20 dB across the test frequencies. Common in:

Otosclerosis at advanced stages
Profound congenital sensorineural loss
Some autoimmune inner-ear disease

Sloping (high-frequency loss)

Thresholds worsen as frequency increases. The most common configuration in adult-onset hearing loss. Etiologies:

Presbycusis — gradual, symmetric, age-related; typically mild to moderate at high frequencies in the 50s, progressing through the 60s and 70s.
Noise-induced hearing loss — bilateral if exposure was symmetric. Usually combines a baseline sloping pattern with a 4-kHz notch (see below).

Rising (low-frequency loss)

Thresholds worsen as frequency decreases. Less common in adults. Etiologies:

Early Ménière’s disease — fluctuating low-frequency sensorineural loss with associated vertigo and aural fullness.
Some hereditary syndromes — Pendred syndrome and others.

Notched

Thresholds dramatically worse at one frequency than at adjacent ones. The most common pattern:

Noise notch at 4 kHz — the audiometric signature of noise-induced hearing loss. The 4-kHz vulnerability has two physical contributors: the ear canal’s resonance amplifies sound at 2–3 kHz, so by the time energy reaches the basilar membrane it is concentrated around 3–4 kHz, and the resulting peak pressures damage the corresponding cochlear region first. The notch may be at 3, 4, or 6 kHz depending on the individual, but 4 kHz is the canonical placement.

Mid-frequency thresholds depressed, with low and high frequencies relatively spared. Etiologies:

Often genetic / hereditary (multiple identified gene variants associated)
Some forms of congenital conductive loss
Otosclerosis at certain stages can present this way

The opposite — low and high frequencies depressed, mid spared. Very rare; usually genetic.

Asymmetric

A loss substantially different between the two ears (typically defined as a ≥ 15-dB difference at any one frequency or a ≥ 10-dB difference in PTA). Asymmetric sensorineural loss is always worked up for retrocochlear pathology — typically MRI to rule out an acoustic neuroma (vestibular schwannoma).

Reading an audiogram in practice

The clinical short script when looking at an unfamiliar audiogram:

Degree — compute the PTA on each ear; assign a degree category.
Type — examine the ABG at each frequency in each ear; conclude sensorineural / conductive / mixed.
Configuration — describe the shape (flat / sloping / rising / notched / cookie-bite).
Symmetry — compare the two ears; flag any asymmetry.
Anchor to history — does the audiometric picture match the patient’s history? Adult with bilateral symmetric high-frequency sloping sensorineural loss and a history of construction work: classic NIHL. Sudden unilateral sensorineural loss in a 50-year-old: emergency workup.

The audiogram is read in seconds by an experienced audiologist. The categorisations above are the vocabulary that experience runs on.

Closing the chapter

That closes Chapter 2 — the audiogram. Across the four lessons we developed pure-tone audiometry as a behavioural test (2.1), the four decibel scales an audiologist juggles (2.2), air-conduction and bone-conduction thresholds and the air-bone gap that distinguishes conductive from sensorineural loss (2.3), and the cross-hearing problem with the masking plateau that solves it (this lesson).

The audiogram is the substrate for almost every subsequent decision the audiologist makes. Subsequent chapters in this book — speech audiometry, tympanometry, OAEs, hearing-aid fitting — all assume the reader can read an audiogram and use it to anchor interpretation. Done well, the audiogram takes ten minutes to administer and informs everything that follows. Done sloppily — without masking, with inattentive presentation, with the wrong calibration — and the rest of the workup is built on sand. It is the foundation tool of the field, and the chapter’s length here reflects the weight it carries.