Night-to-night variation in snoring sound severity: one night studies are not reliable

To quantify the night-to-night variation in snoring severity; to compare this with inter-subject variation in snoring intensity: to compare multinight mean snoring scores with self-reported subjective scores.

Prospective observational study.

Subjects were recorded during sleep at their own homes.

Twenty patients with socially disruptive snoring awaiting surgery.

Over four consecutive nights using a solid-state sound recording device, the mean, standard deviation and intra-class correlation coefficient were calculated for (a) the loudest 1% of sound, (b) snore frequency and (c) total snore duration. Results were correlated with Snoring Symptom Inventory scores assessed immediately prior to these recordings.

Overall mean and intrasubject standard deviation for the loudest 1% of sound was 65.0 (+/-4.1) dB, for snore frequency was 245 (+/-104) per hour and for total snore duration was 4.3% (+/-2.1). Intraclass correlation coefficients were 0.78, 0.74 and 0.67, respectively, suggesting only moderate reliability of these outcome measures. No significant correlation was found between objective and subjective scores for either endpoint.

Natural night-to-night variation in snoring severity represents a significant proportion of overall snoring variance, thus one night studies of snoring are not reliable. The random error associated with one-night studies exceeds the expected effect size of snoring interventions and so multi-night studies of at least four nights are recommended to reduce the error. However, even multi-night objective measurements correlate poorly with subjective scores of snoring.