P138 Inter-observer agreement and sensitivity of definitions for hypoventilation in children with neuromuscular disorders

Abstract Background hypoventilation occurs in neuromuscular disorders with respiratory muscle weakness. There are numerous definitions of hypoventilation. This study aimed to determine which definitions of hypoventilation had the highest inter-observer agreement and sensitivity in children with neuromuscular disorders. Methods twenty polysomnograms from children with neuromuscular disorders were examined by six paediatric sleep physicians to determine whether hypoventilation was present and which definitions of hypoventilation were used. Inter-observer agreement for the diagnosis of hypoventilation was calculated, with sensitivity determined in an exploratory analysis. Results mean inter-observer agreement was excellent at 82% (κ=0.533), higher for the absence (84%) than the presence (70%) of hypoventilation. Some physicians used more than one definition to diagnose hypoventilation for the same polysomnogram. Even when agreement was substantial, the definitions chosen varied between physicians.The objective defiitions used most frequently when inter-observer agreement was substantial that hypoventilation was present were; average increase in pCO2 ≥3mmHg from NREM to REM (n=15), increase of ≥10mmHg from awake to asleep (n=10) and pCO2 >50mmHg >25% total sleep time (n=8). The sensitivity for these definitions was 83%, 67% and 50%. Conclusions even when inter-observer agreement was high that hypoventilation was present, there was a lack of consistency between physicians when defining hypoventilation. An average increase in pCO2 ≥3mmHg from NREM to REM sleep had the highest inter-observer agreement and sensitivity. Alternative definitions in cases without a rise of pCO2 in REM are pCO2 >50mmHg for 25% total sleep time and/or an increase of ≥10mmHg from awake to asleep.