Obstructive sleep apnoea (OSA) is thought to be worse during rapid eye move
ment (REM) sleep. REM rebound in the late postoperative period can follow t
he REM suppression shown to occur after some types of surgery. This is thou
ght to worsen nocturnal episodic hypoxaemia, leading to greater cardio-resp
iratory risk. We set out to determine if OSA was a REM-predominant phenomen
on. We reviewed the sleep clinic records of 64 consecutive patients with a
diagnosis of OSA on full overnight polysomnography and sufficient data to d
etermine the presence of a sleep stage predominance. OSA was diagnosed if t
he number of apnoeas/hypopnoeas per hour of sleep, the respiratory disturba
nce index (RDI), was greater than 10. The variables recorded for the purpos
es of this study were the RDI and the minimum blood oxygen saturation using
pulse oximetry (Sp(O2min)) for both REM and non-rapid eye movement (NREM)
sleep. All values are presented as mean (SD). The Wilcoxon signed rank test
was used for statistical analysis. The means for NREM and REM RDI were, re
spectively, 36 (26) and 38 (27) per hour (P=0.96). In 32 of the 64 patients
(50%) the RDI in NREM was greater than in REM. Thirty-one (48%) had a larg
er number during REM. One patient had identical RDIs for both REM and NREM.
Sixty-two patients had satisfactory pulse oximetry recordings for both NRE
M and REM, and the mean Sp(O2min) values were, respectively, 84 (7) and 82
(13)% (P=0.15). Twenty-nine patients (47%) had a lower Sp(O2min) in REM (se
ven by more than 10% and two by more than 40%), while 24 (39%) were lower i
n NREM (two by more than 10%). Nine patients (14%) had identical values in
REM and NREM. In contrast to suggestions that OSA is a REM-predominant phen
omenon, this study suggests that respiratory disturbance is not greatly aff
ected by sleep stage, in most patients. While a small number clearly desatu
rate much more during REM, the majority do not. Thus, postoperative REM reb
ound may worsen OSA in some patients, but in many it may do otherwise. The
implications of postoperative sleep disturbance are therefore likely to be
more complex than previously suggested.