Snoring: Is It in the Ear of the Beholder?

Summary: Complaint of snoring, which is usually voiced by a patient's bedpartner, frequently leads to investigations in the sleep laboratory that are designed to assess snoring objectively and determine whether it is a symptom of sleep apnea. How well this subjective complaint of the listener is confirmed by the objective measurement of snoring is not known. Consequently, we designed a study i) to test the validity of self-perception of snoring and ii) to compare subjective perception of snoring by the sleep technologist with objective measurement of its frequency and loudness. We studied 613 unselected patients referred to our sleep clinic because of snoring and suspicion of sleep apnea. They all had nocturnal polysomnography that included measurements of snoring, expressed as the number of snores per hour of sleep [snoring index (SI)] and mean (dBmean) and maximum (dBmax) nocturnal sound intensity. Following the sleep study, the technologist (and patient) independently rated a patient's snoring as none, mild, moderate or severe. Kruskall-Wallis test, Spearman rank correlations and Cohen's kappa statistics were used to compare the groups, examine the correlations between subjective and objective measurements, and check the agreement between them. We found that i) most patients who snore (471 of 613) were unaware of this behavior; ii) snoring index increased progressively from the “none” group (median SI = 26) to the “severe” group (median SI = 755), with a significant but moderate correlation between the technologist's impression and SI (r = 0.68, p < 0.0005) and poor correlation between the technologist's impression and dBmean (r = 0.34, p < 0.0005); and iii) most of those patients who claim to be able to judge the severity of their snoring have a different perception of it than the listener, with poor agreement between patient's and technologist's subjective impression (Cohen's kappa = 0.22, 95% confidence limits: 0.14−0.33). We conclude that to a large extent snoring is “in the ear of the beholder”, showing only moderate correlation with objective measurement.