Cochlear Implant Atlas
CI Atlas · Candidacy & Evaluation · Module 04

4The audiometric criteria

Candidacy ultimately comes down to numbers, and the numbers reward careful interpretation. The degree of loss sets the scene, but the decisive figure is the best-aided sentence score, which must fall below a criterion that has crept upward over the years toward fifty or sixty per cent. A subtlety underpins it all: the score depends heavily on how loud the test speech is. Played loudly, the same patient scores far better than they ever do in real conversation; played at conversational or soft levels, the score drops to reflect the everyday difficulty that actually warrants an implant. Candidacy testing therefore deliberately uses softer presentation levels, and the exact criterion is shaped further by device labelling, programme rules and which ear is being considered. This module is about reading those numbers honestly.

TTwo numbers

Two figures anchor the audiometric criteria. The degree of loss (pure-tone average) describes severity, and most candidates sit in the severe-to-profound range — though, as the criteria expanded, the decisive figure shifted to the second: the best-aided sentence score, which must fall below a criterion for the patient to qualify.

Test soft, not loud — a quiet presentation level reveals the real deficit

050100aided score (%)candidacy criterion506070presentation level (dB SPL)

The same patient scores very differently depending on how loud the test speech is. At 70 dB SPL — louder than typical conversation — scores are flattered and reproducible but unrealistic; drop to 60 or 50 dB SPL and the score can fall sharply, exposing the difficulty the patient meets in real life. Candidacy testing therefore uses conversational and soft levels, not loud ones, so a borderline patient is not wrongly excluded by an easy test. Schematic, after Donaldson & Allen and Firszt.

CWhy the level matters

The same patient can pass or fail depending on how loud the test speech is. Presented at 70 dB SPL — louder than typical conversation — scores are higher and more reproducible but unrealistic. Dropped to 60 or 50 dB SPL, scores fall, exposing the difficulty the patient meets in life. Candidacy therefore tests at conversational and soft levels, which better reflect real listening, so a struggling patient is not excluded by an artificially easy test.[2004]

CWhere the line sits

The exact criterion is not universal. It is shaped by regulatory labelling (FDA-approved indications for a given device), by funding/programme rules (which can differ from device labels), and by the accumulating evidence that has pushed the allowable aided-speech ceiling toward 50–60%. The principle beneath the moving number is constant: implant when the best aided performance is poor enough that the implant is likely to do better.[2004]

CWhich ear is tested

Criteria also specify which ear the scores apply to. Historically the better ear had to be poor enough to qualify; expanded indications now allow implantation of a much poorer ear alongside an aidable better ear (asymmetric loss and single-sided deafness, Module 9), and the ear-specific aided scores feed directly into the ear-selection decision (Module 11). Reading the criteria means knowing not just the cut-off but which ear it is being applied to, and at what level.

TCDevice-specific cutoffs

The exact numbers are device- and condition-specific. A typical adult FDA criterion asks for sentence recognition around ≤50% in the ear to be implanted and ≤60% in the contralateral ear, with ≤50% best-aided for bilateral implantation; CMS is tighter at ≤40% best-aided. The electric-acoustic (EAS) window is different again — good low-frequency hearing (normal through ~1500 Hz) sloping to ≥80 dB in the highs, with generous word-score limits. The reason candidacy tests at conversational ~60 dB SPL rather than a loud 70 dB SPL is methodological: ordinary conversation sits near 60 dB SPL, and while post-operative scores are similar at the two levels, pre-operative scores are poorer at 60 — so testing loud artificially inflates the pre-op score and can wrongly disqualify a true candidate.

The cutoffs are device- and condition-specific — and the ear matters

FDA adult (typical)≤50%Bilateral CICMS / MedicareEAS / hybrid (Iowa)10–60% wordsOlder child (words)

There is no one number. Most FDA adult labels ask for a sentence score around ≤50% in the ear to be implanted and ≤60% in the contralateral ear, with ≤50% best-aided for bilateral implantation; CMS is tighter at ≤40% best-aided. The EAS/hybrid window is different again — good low-frequency hearing (normal through ~1500 Hz) sloping to ≥80 dB in the highs, with generous word-score limits. And the condition tested matters: a binaural best-aided score can mask a poorer ear that genuinely qualifies, so the ear-to-be-implanted is tested on its own. Illustrative; verify against current device instructions for use. Schematic.

Case 11.4 · It depends how loud you test
A borderline patient scores 70% on aided sentences at 70 dB SPL but only 40% at 50 dB SPL. The clinic debates which level should inform candidacy.

Which level is more appropriate, and why?

Self-assessment — Module 42 questions
Question 1 · Trainee

Why does candidacy testing use conversational/soft presentation levels rather than loud ones?

Question 2 · Clinician

What shapes exactly where the candidacy speech-score line sits?

Tracked locally in your browser — see /progress for the dashboard.