The variability of the apnoea±hypopnoea index LIA RITA AZEREDO BITTENCOURT 1,2 , DEBORAH SUCHECKI 2 , SERGIO TUFIK 2 , CLO Â VIS PERES 3 , SO Ã NIA MARIA TOGEIRO 1,2 , MAURI Â CIO DA CUNHA BAGNATO 1,2 and LUIZ EDUARDO NERY 1 1 Respiratory Division, Universidade Federal de SaÄo Paulo/Escola Paulista de Medicina, Brazil; 2 Department of Psychobiology, Universidade Federal de SaÄo Paulo/Escola Paulista de Medicina, Brazil, and 3 Biostatistic Division, Universidade Federal de SaÄo Paulo/Escola Paulista de Medicina, Brazil Accepted in revised form 9 April 2001; received 11 September 2000 INTRODUCTION The apnoea/hypopnoea index (AHI) is the most frequently used measurement to characterize a sleep respiratory disorder. From the clinical point of view, the AHI is used for the classi®cation of severity of obstructive sleep apnoea/hypop- noea syndrome (OSAHS) (American Academy of Sleep Medicine 1999) and as a mortality marker (He et al. 1998). In research, this index is utilized to describe population distribution of this sleep disorder (Young et al. 1993). Despite its popularity as a primary measurement of the disease, there are limiting factors to its application. Until the publication of American Academy of Sleep Medicine (1999), there was a lack of standardized criteria to de®ne hypopnoea, one of the key-components of the AHI. A second limitation for the use of AHI is the lack of a universally accepted cut- point to de®ne OSAHS. Some groups employ values of ®ve (Young et al. 1993, 1996, 1997a, b), 10 (Flemmons et al. 1994), 15 (Engleman et al. 1997), 20 (He et al. 1998) or even 30 events h )1 (Redline et al. 1998). Another aspect of vital importance and the motivation for the present study is the question of how variable AHI is from one night to the next. There are few reports on this matter and the contribution given by studies that employ test±retest are still recommended (American Academy of Sleep Medicine 1999). The few studies that explore AHI variability are contradictory, insofar as some describe index variability whereas others do not. The diverging points may be related to methodological Correspondence: Dr Lia Rita A. Bittencourt, Universidade Federal de SaÄ o Paulo, Rua NapoleaÄ o de Barros, 925, Vila Clementino ± SP 04024- 002, SaÄo Paulo, Brazil. Tel.: 55 11 539-0155; fax: 55 11 572 5092; e-mail: lia@psicobio.epm.br J. Sleep Res. (2001) 10, 245±251 SUMMARY This study was designed to evaluate the variability of the apnoea±hypopnoea index (AHI) in 20 patients with obstructive sleep apnoea±hypopnoea syndrome (OSAHS) and to determine possible relationships of this variability with other polysomnographic parameters. The subjects were recorded on four consecutive nights. The mean AHI values were not signi®cantly altered throughout the four recording nights (P  0.67). The intraclass correlation coecient of the AHI on the four nights was 0.92. However, the Bland and Altman plot showed that, individually, the AHI presented an important variability, which was not related to its initial value. In regard to the OSAHS severity, 50% of the patients changed the classi®cation from the ®rst to the subsequent nights. Thirteen of the 20 patients (65%) presented a variation in the AHI value equal or higher than 10 events h ±1 . When we evaluated the AHI mean values for a speci®c body position and sleep stage, no dierence was observed among the nights. In both supine and lateral±ventral decubitus, higher AHI was observed during Stages 1 and 2 than the other stages. Additionally, the AHI during Stages 1 and 2 and REM sleep was higher on the supine than on the lateral±ventral decubitus. The AHI in OSAHS patients presented a good correlation among the four recording nights; however, a signi®cant individual variability should be considered, especially when AHI is applied in OSAHS classi®cation or as a criterion of therapeutic success. KEYWORDS apnoea/hypopnoea index, obstructive sleep apnoea/hypopnoea syn- drome, variability Ó 2001 European Sleep Research Society 245