Confidence Limits of Word Identification Scores Derived Using Nonlinear Quantile Regression

Area of Science:

• Audiology
• Speech Perception Research
• Hearing Science

Background:

• The relationship between sensorineural hearing loss severity and maximum speech identification scores (PBmax) is crucial for audiological diagnosis.
• PBmax scores exhibit significant inter-subject variability for a given hearing loss degree, necessitating the establishment of normative boundaries.
• Understanding the lower limit of expected PBmax is vital for identifying individuals with unexpectedly poor speech recognition.

Purpose of the Study:

• To determine the lower boundary of maximum speech identification scores (PBmax) in individuals with sensorineural hearing loss.
• To establish a reference point for identifying 'disproportionately' poor speech identification relative to the degree of hearing loss.
• To provide a tool for more accurate audiological diagnosis and rehabilitation planning.

Main Methods:

• PBmax scores were measured using the Dantale I word list in 1,961 young and older subjects with sensorineural hearing loss.
• Speech testing was conducted at 30 dB above the speech reception threshold or at the most comfortable level.
• A nonlinear quantile regression approach was used to estimate the lower limit of the one-tailed 95% confidence limit (CL) for PBmax scores across a range of pure-tone averages.

Main Results:

• The study successfully estimated the lower boundary (95% CL) for PBmax scores as a function of pure-tone average.
• A method was developed to identify PBmax scores that fall below this established lower boundary.
• This lower boundary provides a statistical reference for assessing speech identification performance.

Conclusions:

• A specific lower boundary for PBmax scores in sensorineural hearing loss was defined using quantile regression.
• Measured PBmax scores falling below this 95% confidence limit may indicate disproportionately poor speech identification.
• This finding aids in the clinical identification of atypical speech recognition performance in individuals with hearing loss.