Half-night polysomnography: how is it compared to full-night polysomnography?

I read with interest the paper by FANFULLA et al. [1] on half-night polysomnography (HN-PSG). The authors found that there was a significant difference between apnoeahypopnoea indices (AHIs) recorded in the first and second portion of a standardized full-night polysomnography (FN-PSG). The statistical difference was more significant in patients who did not have rapid eye movement (REM) sleep during the first half of the night. The authors concluded that HN-PSG was not appropriate for evaluating patients with sleep-disordered breathing. In considering whether HN-PSG may be used to diagnose and assess the severity of sleep apnoea, the level of statistical difference between indices measured using HN-PSG and FN-PSG is less relevant than the magnitude of the difference. BLAND and ALTMAN [2] described a graphical technique to compare the difference between readings by two methods to the mean of those two readings. The mean difference indicates the bias and the standard deviation of the differences indicates the error between the two methods. I used the Bland-Altman plot to examine the agreement between HN-PSG and FN-PSG in 37 consecutive patients who were diagnosed with obstructive sleep apnoea syndrome (OSAS) according to the International Classification of Sleep Disorders [3] and had a AHI >10 events·h-1. The 37 subjects were predominately male (male:female 35:2) and 42.1±10 yrs of age (mean±SD). The mean body mass index was 28.8±6.3. During HN-PSG and FN-PSG, the mean total sleep times were 179±43 and 412±54 min, respectively. Ten subjects (27%) had no REM sleep in the first half of the night. During HN-PSG and FN-PSG, the AHI (mean±SD) were 42.3±30.6 and 44.4±27.1, respectively, the number of Š4% oxygen desaturation per hour were 16.4±20.4 and 17.6±18.3, respectively, and the arousal index (mean±SD) scored according to the American Sleep Disorders Association (ASDA) criteria [4] were 49± 26.3 and 49.2±24.2, respectively. There was no significant difference between the parameters measured during HNPSG and FN-PSG. The Bland-Altman plot of the AHIs recorded during HN-PSG and FN-PSG were shown in figure 1. The mean and SD of the differences in AHI using HN-PSG and FN-PSG were 2 and 7.5 events·h-1, respectively. The limits of agreement (mean±2SD) and 95% confidence interval for the mean were -13 to +17 and -0.5 to +4 events·h-1, respectively. The difference between AHIs measured during HN-PSG and FN-PSG was <10 events·h-1 for most patients with mean AHI >30 events·h-1. The magnitude of the difference was less clinically significant than that in subjects with mean AHI <30 events·h-1. I repeated the analysis in patients with mean AHI between 10 and 30 events·h-1 (n=17). The results were: mean = 4.8; SD = 7.1; limits of agreement = -9.4 –19; and 95% confidence interval = 1.1–8.5 events·h-1. HN-PSG considerably underestimated the severity of sleep apnoea and the AHI recorded during HN-PSG notably differed from that by a standardized full-night study. The results demonstrated that although there was no significant difference between the indices recorded during HN-PSG and FN-PSG, the actual magnitude of the difference could only be revealed by the Bland-Altman plot I agree with FANFULLA et al. [1] that HN-PSG has a limited value for evaluating patients with sleep-disordered breathing. The assessment of the severity of sleep apnoea using HN-PSG in patients with mild OSAS is far from accurate.