Are subjective assessments of snoring sounds reliable?

Abstract

The aim of this study is to investigate whether snoring sounds can be reliably, i.e. reproducibly and comparably, assessed by the listener. This is an absolute prerequisite if an objective method is to provide a valid representation of subjective assessments of snoring sounds. Fifty-three subjects, both men and women, from different age categories assessed snoring sequences in terms of their annoyance level. This was done using a paired comparison model with ten sequences, once at original volume and once at adjusted peak level. In addition, assessments of 50 sequences were made using a visual analog scale (VAS). In the first paired comparison the assessments displayed very high consistency (K = 0.94) and high concordance (W = 0.83). The results showed very good agreement between the sexes (r s = 0.98), between different levels of snoring-related experience (r s = 0.99) and between different ages (r s = 0.98). At adjusted peak level there was also very high consistency (K = 0.91) although concordance was clearly lower (W = 0.50). The rankings in the two paired comparisons showed a high degree of correlation (r s = 0.81). The assessments on the VAS revealed good inter-rater reliability (α = 0.73) and a high degree of correlation with the result of the first paired comparison (r s = 0.93). Snoring sounds can be assessed in a reproducible and comparable manner in terms of their annoyance level. There are no gender-specific differences, and age-specific and experience-specific differences also appear to be negligible. While volume plays a dominant role, other characteristics of the sounds are also sufficient to permit an assessment.