The effect of experience on the sensitivity and specificity of the whispered voice test: a diagnostic accuracy study

To determine the sensitivity and specificity of the whispered voice test (WVT) in detecting hearing loss when administered by practitioners with different levels of experience.

Diagnostic accuracy study of WVT, through acoustic analysis of whispers of experienced and inexperienced practitioners (experiment 1) and behavioural validation of these recordings (experiment 2).

Research institute with a pool of patients sourced from local clinics in the Greater Glasgow area.

22 people had their whispers recorded and analysed in experiment 1; 4 older experienced (OE), 4 older inexperienced (OI) and 14 younger inexperienced (YI). In experiment 2, 73 people (112 individual ears) took part in a digit recognition task using 2 OE and 2 YI whisperers from experiment 1.

Average level (dB sound pressure level) across frequency, average level across all utterances (dB A) and within/across-digit deviation (dB A) for experiment 1. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of WVT for experiment 2.

In experiment 1, OE whisperers were 8-10 dB more intense than inexperienced whisperers across all whispered utterances. Variability was low and comparable regardless of age or experience. In experiment 2, at an optimum threshold of 40 dB HL, sensitivity and specificity were 63% (95% CI of 58% to 68%) and 93% (92% to 94%), respectively, for OE whisperers. PPV was 56% (51% to 61%), NPV was 95% (94% to 96%). For YI whisperers at an optimum threshold of 29 dB HL, sensitivity and specificity were 80% (78% to 82%) and 52% (50% to 55%), respectively. PPV was 65% (63% to 67%) and NPV was 70% (67% to 72%).

WVT is an effective screening test, providing the level of the whisperer is considered when setting the test's hearing-loss criterion. Possible implications are voice measurement while training for inexperienced whisperers.