Normal polysomnography parameters in healthy adults: a systematic review and meta-analysis

Abstract

Background: Existing normal polysomnography values are not truly normative as they are based on small sample sizes due to the fact that polysomnography is expensive and burdensome to obtain. There is a clear need for a large sample of truly normative data for clinical management and research. This study is a comprehensive meta-analysis of adult polysomnography parameters scored using recent criteria and establishes normative values adjusted for age and sex.

Methods: For this meta-analysis of adult polysomnography parameters, we searched Scopus for studies of any design published between Jan 1, 2007, and July 31, 2016, that reported polysomnographic parameters scored using recent American Academy of Sleep Medicine criteria (2007 or 2012) collected during an overnight level 1 in-laboratory sleep study in healthy controls. We excluded studies of patients with conditions or subjected to treatments that might affect sleep and studies not available in English. Study endpoints were the pooled estimates of 14 reported polysomnographic parameters. Estimates for each parameter were pooled using a random-effects meta-analysis. The influence of age and sex was ascertained using multivariate mixed-effects meta-regressions. This study is registered with PROSPERO, number CRD42017074319.

Findings: Of 3712 articles, 169 studies, comprising 5273 participants, were eligible for inclusion. We report normative data stratified by age and sex. For each decade of age, total sleep time decreased by 10·1 min (95% CI 7·5 to 12·8), sleep efficiency decreased by 2·1% (1·5 to 2·6), wake after sleep onset increased by 9·7 min (6·9 to 12·4), sleep onset latency increased by 1·1 min (0·3 to 1·9), arousal index increased by 2·1 events per h (1·5 to 2·6), percentage of N1 sleep increased by 0·5% (0·1 to 0·8), apnea-hypopnea index increased by 1·2 events per h (0·9 to 1·4), mean oxygen saturation decreased by 0·6% (0·5 to 0·7), minimum oxygen saturation decreased by 1·8% (1·3 to 2·3), and periodic limb movement index increased by 1·2 events per h (0·8 to 1·6). Changes with age in the percentage of N2 sleep (0·0%, 95% CI -0·1 to 0·1), N3 sleep (-0·1%, -0·1 to 0·0), and rapid eye movement (REM) sleep (0·0%, -0·1 to 0·0) were not significant. Every 10% increase in the percentage of male participants was associated with reduced REM latency (0·9 min decrease, 95% CI 0·1 to 1·6) and mean oxygen saturation (0·1% decrease, 0·0 to 0·1), and greater arousal index (0·3 events per h, 0·0 to 0·5) and apnea-hypopnea index (0·2 events per h, 0·1 to 0·3).

Interpretation: These normative values serve as a useful control reference for clinicians and for future research where it might be difficult to obtain polysomnographic controls. The resulting normative trends by age and sex might also be hypothesis-generating for a broad range of investigations.

Funding: None.