Australasian Sleep Technologists Association

Articles & Links

Articles

Reliability of manual and automatic scoring of paediatric sleep studies

INTER & INTRA-RATER RELIABILITY OF MANUAL AND AUTOMATIC SLEEP, RESPIRATORY AND AROUSAL SCORING IN A PAEDIATRIC POPULATION.

Printer friendly version

Rebecca Smith1, Rosemary Horne2, Nicole Verginis1, Margot Davey1
1 Melbourne Children's Sleep Unit, Dept of Respiratory and Sleep Medicine, Monash Medical Centre, and 2 Ritchie Centre for Baby Health Research, Monash University, Victoria, 3168

Introduction: In paediatric polysomnography (PSG) there is currently no automatic analysis software available, hence studies are routinely manually scored. It is therefore necessary to ensure good inter and intra-rater reliability. Previous investigations in adult laboratories have observed sleep staging agreement between 65-90%. There is less inter-rater agreement for respiratory disturbance (RDI) and arousal indices (ArI).

Aim: To evaluate the inter- and intra-rater reliability in sleep staging, respiratory disturbance and arousal indices of scorers within the same paediatric sleep laboratory.

Methods: Five overnight PSG's (age range; 4-8 years), recorded using Compumedics, were scored manually by 5 sleep technologists (experience 9 months - 6 years) and automatically using Compumedics software. To determine intra-rater agreement the same technologist re-scored one of the PSG recordings with an average of one month between scorings. Agreement values for sleep staging were calculated on an epoch-by-epoch basis. RDI and ArI were also compared.

Results: The average inter-rater agreement was 87% (range 80-92%; kappa 0.64 - 0.90) and the average intra-rater agreement was 89% (range 87-91%; kappa 0.78 -0.88). There was no statistically significant difference between or within raters for RDI or ArI. Average agreement between manually and computer scored sleep staging was 23% (range 26-31%; kappa 0.004 - 0.16). There was a statistically significant difference between manually and automatically scored ArI (p < 0.01), but not for RDI, however automatic analysis often scored inappropriate respiratory events.

Conclusions: This study has provided quality assurance for our sleep laboratory, which should routinely be conducted in all paediatric sleep laboratories. Future studies should investigate agreement between paediatric sleep laboratories.

Key words: reliability, sleep staging, paediatric laboratory, agreement

Links

Sleep and related Associations

Sleep Research