robust. However,findings using physiologic measures of sleep and hotflashes are
less clear. Early studies using polysomnography (PSG) showed waking episodes
occurring at the time of a hotflash occurrence (Erlik et al. 1981 ). However, later
studies showed less consistent associations between hotflashes and waking epi-
sodes (Woodward and Freedman 1994 ; Freedman and Roehrs 2004 , 2006 , 2007 ).
Freedman and Roehrs ( 2004 ) examined the relationship of hotflashes and waking
episodes, variouslyfinding no relationship or significant associations between hot
flashes and waking episodes only in thefirst half of the night [proposed to be due to
the thermoregulatory alterations of rapid eye movement sleep suppressing hotfla-
shes in the second half of the night ( 2006 and 2007 )]. In a subcohort of SWAN
participants with hotflashes, sleep was assessed via both self-report and actigraphy
and hotflashes via both event marker and SCL. Participants’self-report of more
overnight hotflashes upon waking, but not SCL-assessed hotflashes during sleep,
was associated with significantly poorer actigraphy-assessed sleep (Thurston et al.
2012b). Thus, in this study, it appeared that women who woke from a poor night of
sleep tended to recall having more hotflashes (even in the absence of hotflashes
being detected). These studies have collectively called into question whether hot
flashes wake women up.
An exception to this broad pattern offindings is a study of 20 women in whom hot
flashes were induced by suppressing the reproductive axis via a
gonadotropin-releasing hormone agonist (Joffe et al. 2013 ). In this study, partici-
pants’sleep efficiency, the percentage of time in bed that participants slept, was
assessed via wrist actigraphy and hotflashes assessed via 24-h ambulatory SCL
monitoring at baseline and after administration of the medication (Joffe et al. 2013 ).
Women who developed hotflashes with the medication, as indexed by SCL, had
significant deficits in sleep efficiency (median 2.6%, range 0.9–10.2%). In contrast,
women who were not found to develop hotflashes had a median improvement
(median 4.2%, range 0.8–7.5%) in their sleep efficiency (Joffe et al. 2013 ). Similarly,
in another subcohort of SWAN participants, women with more reported hotflashes
(physiologic hotflashes were not assessed) had higher electroencephalogram beta
power during sleep, indicative of higher cortical arousal (Campbell et al. 2011 ).
It is notable that similar to hotflashes, self-reported and physiologically assessed
sleep (via PSG or actigraphy) are moderately correlated (e.g.,r= 0.31–0.59)
(McCall and McCall 2012 ). Physiologic measures of sleep, such as actigraphy and
PSG, examine the quantity, depth, or continuity of sleep, whereas sleep quality
(commonly assessed by the Pittsburgh Sleep Quality Index; Buysse et al. 1989 )
involves the subjective assessment of the quality of sleep and sleep disturbances
retrospectively over a longer time period (e.g., 1 month). This is relevant to the
measure of hotflashes as these distinct measures can lead to interesting differential
correlations between hotflashes and sleep measures. However, it is notable that
sleep researchers typically regard self-report measures and physiologic measures of
sleep as representing two distinct, yet equally informative, constructs.
In sum, there is clear evidence that the subjective experience of hotflashes is
strongly related to poorer self-reported sleep. However,findings are less clear from
the more limited number of studies using physiologic measures of both sleep and
242 W.I. Fisher and R.C. Thurston