Cochlear Implant Atlas
CI Atlas · Candidacy & Evaluation · Module 08

8The actuarial odds model

A single audiometric cut-off is a blunt instrument: it treats every ear at the threshold as identical when their prospects are not. The most rigorous answer to candidacy reframes it as a probability — the statistical odds that a given ear will understand more with an implant than with its hearing aid. The UK Cochlear Implant Study Group built exactly this, combining an ear's pre-operative speech score with its duration of deafness into an actuarial equation and recommending implantation only when the odds of improvement reach about four in five. The Melbourne group reached a parallel, more generous rule from their own outcome data. Because these models are computed per ear, two ears with identical hearing can carry very different candidacy — and the maths makes explicit why implanting the poorer or shorter-deaf ear is often right. This module is about candidacy as odds.

TFrom a cut-off to a probability

A fixed cut-off asks “is the score below X?” The actuarial model asks a better question: “what are the odds this ear will do better with an implant than with its aid?” That reframes candidacy as a per-ear probability, computed from more than one variable, and lets borderline decisions rest on evidence rather than a single line.

Candidacy as per-ear odds — duration of deafness moves the needle

70%Left ear44%Right ear
Higher odds of improvementImplant the LEFT ear (ticks mark the ~75% Dowell and ~80% UKCISG thresholds).

Instead of one fixed audiometric cut-off, the actuarial model asks a probabilistic, per-ear question: what are the odds this ear will score better with an implant than with its aid? The UK group combined pre-op speech score and duration of deafness into an odds equation and recommended implanting when the odds are at least about 4-in-5 (80%); the Dowell/Melbourne rule sets a similar ~3-in-4 bar with more generous score limits. Because odds fall steeply with duration — roughly 4:1 at a few years deaf toward 1:1 at 30 years — two ears with identical hearing can differ, and the model directly justifies implanting the poorer or shorter-deaf ear. Schematic.

CThe UKCISG odds rule

The UK Cochlear Implant Study Group defined candidacy by the odds that an ear would score better after implantation than with its hearing aid, recommending implantation when those odds were at least about 4-in-5 (80%). They built this across three companion papers — one establishing equivalent effectiveness, one a cost-effectiveness analysis, and one the prospective actuarial criterion.[2004][2004] An MRC actuarial equation combined the pre-operative speech score and the duration of profound deafnessto compute each ear's odds.

COdds fall with duration

The model makes the dominance of duration of deafness quantitative: the odds of exceeding a given sentence score fall steeply from roughly 4:1 at a few years deaf toward about 1:1 at thirty years. Two ears with identical current hearing but different histories therefore have different candidacy — which is exactly why the same audiogram does not give the same answer in every patient, and why the model directly justifies implanting the poorer or shorter-deaf ear. The economic companion paper tied this to cost-effectiveness, the bridge to the access and funding view of selection.[2004]

CThe Dowell/Melbourne redefinition

The Melbourne group reached a parallel rule from their own users' outcomes: a candidate if best-aided sentence scores are up to about 70%, and up to about 40% in the ear to be implanted — giving roughly 3-in-4 odds of improvement, and retrospectively almost all such patients improved.[2004] Like the UK model, it is probabilistic and residual-hearing-tolerant: it lets in patients with substantial aided scores whom an old fixed cut-off would have excluded. Together these actuarial rules replaced a single number with a per-ear estimate of benefit — the intellectual core of modern candidacy.

Odds of improvement fall steeply with duration of deafness

1:12.5:14:1odds of improvement0102030duration of deafness (years)

This is the curve underneath the per-ear gauge. The actuarial models express candidacy as the odds that an implanted ear beats its aid, and those odds decay steeply with the duration of deafness — from roughly 4:1 at a few years toward about 1:1 by thirty years. It is the quantitative face of “duration of deafness is the dominant predictor,” and the reason a recently-deafened ear is a far stronger candidate than a long-silent one with the same audiogram. Schematic.

Case 11.8 · Same audiogram, different odds
Two ears in one patient have identical aided sentence scores, but the right has been deaf 5 years and the left 28 years. The team uses an actuarial model to decide.

What does the model conclude?

Self-assessment — Module 82 questions
Question 1 · Trainee

How does the actuarial model define candidacy?

Question 2 · Clinician

Why can two ears with identical aided scores have different candidacy under this model?

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