Design: retrospective multicentre cohort study across five tertiary academic centres, with multivariable analysis of predictors
Population: 68 adults implanted after more than ten years of unilateral severe-to-profound hearing loss; median duration 17 years (IQR 12–35)
Primary outcome: postoperative word recognition score and average daily device use
Effect: mean word recognition gain 28.4% (SD 31.6, P < 0.001); ≥20-point gain in 54.5% and ≥50-point gain in 32.7%; daily use 8.4 h in year one and 7.1 h thereafter; duration of deafness not predictive

Duration of auditory deprivation has long been treated as a reason not to implant: the working assumption is that a cochlear nerve unused for a decade will not usefully carry speech. A retrospective cohort across five tertiary academic centres examined 68 adults implanted after more than ten years of unilateral severe-to-profound hearing loss, with a median duration of 17 years and an interquartile range reaching to 35.

Most benefited. Mean word recognition improved by 28.4 percentage points (standard deviation 31.6, P < 0.001) among those tested with the same instrument before and after. Just over half (54.5%, 30 patients) gained at least 20 points and a third (32.7%, 18 patients) gained at least 50. Device use was substantial and sustained: a mean of 8.4 hours daily in the first year and 7.1 hours beyond it, which matters because an implant a patient does not wear is the failure mode this population was expected to have. On multivariable analysis, duration of deafness did not predict postoperative word recognition.

The standard deviation of 31.6 against a mean gain of 28.4 says the other half of the story: outcomes varied enormously, and some patients gained little. That argues for individualised candidacy assessment rather than for implanting everyone with an old dead ear. But it does remove the blanket exclusion — a patient deaf in one ear for fifteen years should be assessed on their own merits and not turned away on the calendar.

Do not exclude a patient from cochlear implant assessment on duration of deafness alone.
Counsel honestly on the spread of outcomes, not the mean — some patients gain little.
Discuss expected daily wear time; use, not implantation, is what determines benefit.
Assess the contralateral ear's function, since the decision in single-sided deafness is different from a bilateral one.
Refer for formal candidacy assessment rather than making the decision in a general clinic.

Why it matters

A criterion used to refuse referral turned out not to predict the outcome it was being used to predict.

The statistics, in plain English

A mean gain of 28.4 percentage points with a standard deviation of 31.6 means the variation between patients was larger than the average effect — this is a group in which some people improved dramatically and some did not improve at all, and quoting the mean alone to a patient would be misleading. That duration of deafness 'was not predictive' on multivariable analysis is a negative finding in 68 patients, which is a small number for a regression with several covariates; it argues against duration being a strong predictor rather than establishing that it has no effect. This is retrospective tertiary-centre data, so the patients selected for implantation were already thought likely to benefit.