Emotional Antecedents of Hot Flashes During Daily Life
REBECCA C. THURSTON, PHD, JAMES A. BLUMENTHAL, PHD, MICHAEL A. BABYAK, PHD, AND ANDREW SHERWOOD, PHD
Objective: Hot flashes are among the most frequently reported menopausal symptoms. However, little is known about factors
associated with their occurrence. Moreover, despite the wide use of self-report hot flash measures, little is known about their
concordance with physiological flashes. This study evaluated emotional and behavioral antecedents of subjectively and objectively
measured hot flashes during daily life. It also examined individual differences predicting concordance between objective and
subjective hot flashes. Methods: Forty-two perimenopausal or postmenopausal women (mean age ? 50.5 ? 4.8 years) reporting
daily hot flashes completed 2 days of ambulatory sternal skin conductance monitoring, behavioral diaries 3 times an hour, and
psychometric questionnaires. Hot flashes meeting objective physiological criteria and subjectively reported flashes not meeting
physiological criteria were assessed. Likelihood of hot flashes following emotions and activities were examined in a case-crossover
analysis. Results: Relative to nonflash control times, objective hot flashes were more likely after increased happiness, relaxation,
and feelings of control, and less likely after increased frustration, sadness, and stress. Conversely, subjective hot flashes not meeting
physiological criteria were more likely after increased frustration and decreased feelings of control. Questionnaires revealed
increased negative mood and negative attitudes were associated with fewer objective flashes and higher false-positive reporting
rates. Conclusion: Increased positive and decreased negative emotions were associated with objective hot flashes, whereas
increased negative and decreased positive emotions were associated with subjective flashes not meeting physiological criteria. The
anecdotal association between negative emotions and hot flashes may be the result of self-reported flashes lacking physiological
corroboration. Key words: menopause, hot flashes, hot flushes, vasomotor symptoms, emotions, stress.
HRT ? hormone replacement therapy; BDI-II ? Beck Depression
Inventory, Second Revision; STAI ? State Trait Anxiety Inventory;
DSI ? Daily Stress Inventory; SCL-90-R ? Symptom Checklist–90,
Revised; RR ? rate ratio.
naturally menopausal American women reporting hot flashes.
Among surgically menopausal women, prevalence estimates
are even higher (3,4). Reported hot flashes have been associ-
ated with negative mood (5,6) and subjective sleep distur-
bance (7) and are a leading cause of treatment-seeking among
menopausal women (8). Despite their prevalence and apparent
impact on women’s lives, little is known about factors asso-
ciated with their occurrence, including individual differences
rendering women more vulnerable to hot flashes and behav-
ioral factors directly preceding their occurrence.
One commonly reported antecedent of hot flashes is neg-
ative emotional arousal. In 1 study of 506 women, the major-
ity of participants (59%) cited negative emotions as the lead-
ing trigger of hot flashes (9). Other largely anecdotal reports
have linked negative emotions to hot flashes (10,11), and
indirect evidence from relaxation intervention studies has sug-
gested a relationship between negative emotional arousal and
hot flashes (12–16). One laboratory study (17) designed to
provoke hot flashes with behavioral stressors suggested an
impact of stressors on hot flashes. However, most existing
studies have been very small in size (12,16–18), uncontrolled
(12,16), or assessed self-reported flashes only (12,15,16,18).
No studies have examined emotions as precursors of physio-
ot flashes are considered the “classic” symptom of meno-
pause, with estimates ranging from 57% (1) to 74% (2) of
logical hot flashes in an ambulatory context. Thus, the role of
negative emotions in hot flash occurrence remains unclear.
Most information regarding hot flashes has been derived
from studies of self-reported hot flashes. Despite the wide use
of these measures, little is known about their relationship to
objectively defined physiological hot flashes. Importantly,
physical symptom reporting may be subject to recall and
reporting biases (19,20). One such influence is mood, with
evidence linking negative affect with somatic magnification or
amplification (21,22). However, little is known about the
concordance between objective and subjective hot flashes
during daily life, and particularly factors influencing their
Recent technologic advances have enabled physiological
measurement of hot flashes. Laboratory studies have estab-
lished sternal skin conductance as the most sensitive and
specific physiological measure of hot flashes (23–25), and
unlike palmar and plantar skin conductance, is relatively un-
responsive psychologic stimuli (24,26). Sternal skin conduc-
tance has been adapted and validated (24,27) for use in an
ambulatory context with high levels of sensitivity (72–100%)
(24,27) and specificity (70–86%) (24,27). Physiological mea-
sures are particularly important in studies of mood and emo-
tions, given the influence of psychologic factors on symptom
perception and reporting (21,22).
This study examined the role of emotional arousal in the
occurrence of hot flashes during daily life. This study also
examined the influence of moods and emotions on hot flash
reporting. Hot flashes were physiologically measured through
ambulatory sternal skin conductance. The case-crossover an-
alytic technique, previously used to examine triggers of myo-
cardial events (28,29), was used to examine precursors of hot
flashes. We hypothesized that hot flashes would be more
likely to occur after elevated negative emotion and would be
less likely to occur after increased positive emotion compared
with control times. We hypothesized that women with more
frequent hot flashes would have increased negative mood.
Moreover, we hypothesized that women with more negative
From Duke University Medical Center, Department of Psychiatry and
Behavioral Sciences, Durham, North Carolina.
Rebecca Thurston is currently a Robert Wood Johnson Health and Society
Scholar at Harvard University.
Address correspondence and reprint requests to Rebecca C. Thurston, PhD,
Harvard School of Public Health, 677 Huntington Avenue, 7th Floor, Boston,
MA 02115-6096. E-mail: email@example.com.
Received for publication March 23, 2004; revision received July 26, 2004.
137Psychosomatic Medicine 67:137–146 (2005)
Copyright © 2005 by the American Psychosomatic Society
mood would have an increased false-positive reporting rate.
Finally, we explored the relationship between “false-positive”
hot flashes (reported subjectively but lacking physiological
corroboration) and emotions.
Forty-two perimenopausal or postmenopausal women were recruited from
the community through flyers, public service announcements, and newspaper
advertisements. The following were required for study entry: 1) female; 2) age
40 to 60; 3) reporting daily hot flashes; 4) postmenopausal (amenorrhea ?12
months), late perimenopausal (amenorrhea 3–12 months), or early perimeno-
pausal (irregularity in menstrual cycle frequency/length) status; and 5) pro-
vision of informed consent and ability to follow study procedures.
The following were exclusionary criteria: 1 taking the following medica-
tions: hormone replacement therapy (HRT), oral contraceptives, clonidine,
tamoxifen, raloxifene, pellagral, tibolone; 2) hysterectomy without bilateral
oophorectomy (as a result of inability to establish menopausal status); 3)
history of menstrual irregularities in peak reproductive years; and 4) history
of medical or psychiatric conditions associated with hot flash sensations
(panic disorder, pheochromocytoma, leukemia, pancreatic tumor). All partic-
ipants were compensated $50.
Participants underwent telephone and in-person screening. Eligible par-
ticipants were equipped with an ambulatory skin conductance monitor be-
tween 7:30 and 10:00 AM on a typical weekday, wore the monitor during the
day and overnight at home, removed it on morning waking, and returned to
the laboratory. They repeated this protocol on a second workday within 2
weeks. Two days of monitoring were estimated to yield adequate power to
evaluate study hypotheses, to establish reliable hot flash estimates, and to
provide time and participant matched control periods while minimizing par-
During waking monitoring hours, participants completed a structured
behavioral diary, which they were prompted to complete on a fixed schedule
3 times an hour through a portable signaling device worn around the wrist or
on the belt (WatchMinder Training and Reminder System; Advanced Multi-
Media Designs, Great Neck, NY). They were also instructed to complete a
diary entry during a subjective hot flash. Participants were instructed to report
subjectively experienced hot flashes by pushing the monitor’s event buttons
and completing a diary page. Participants completed a battery of psychometric
questionnaires on the first monitoring day.
Ambulatory Skin Conductance Monitoring
Hot flashes were objectively measured with a Biolog ambulatory skin
conductance monitor (UFI, model 3991/1-SCL; Morro Bay, CA), a light-
weight, portable, single-channel device allowing continuous measurement of
sternal skin conductance during daily life. It contains 2 MB of memory, is
powered by 1 9-V battery, and samples at 1 Hz (once/second) with a 0.5
constant voltage circuit (30). Two Medi-trace silver/silver chloride electrodes
(Graphic Controls, Buffalo, NY), 1.5 cm in diameter filled with 0.05 M
potassium chloride Unibase/glycol paste (31), were affixed to the sternum. Its
2 event mark buttons, pressed simultaneously, provides a date- and time-
stamped subjective event report. Participants were instructed to avoid engag-
ing in rigorous activities or showering while wearing the monitor.
Data were downloaded into a personal computer after each day and
scanned for hot flashes visually and using DPS Software Support Package
(UFI). Events meeting hot flash criterion of ?2 micro-mho (?mho) increase
in a 30-second period were electronically flagged and visually inspected by a
trained analyst in 5-minute, 1-minute, or 30-second windows to distinguish
flash from artifact. Additionally, all data were visually inspected in 10-minute
intervals to ensure all hot flash events were coded. The minimum interflash
interval, during which no flashes were scored, was 20 minutes after flash
onset (24,27). Independent coding of 10% of files by expert J. Carpenter
(27,32), who was blinded to participants’ diary recordings, revealed ? ? 0.74,
indicating adequate interrater reliability.
Subjective hot flash reports by diary or event marker were compared with
physiologically recorded or objective hot flashes to establish reporting rates.
A true positive flash was a subjective flash report accompanied by the
physiological criterion within 5 minutes. A false-negative flash was an ob-
jective hot flash with no subjective report within 5 minutes, and a false-
positive flash was a subjective flash report not meeting the physiological
criterion within 5 minutes (23,27). A true negative was a 20-minute interval,
corresponding to the interflash scoring interval, without a reported or objec-
tive flash. True-positive (true positive/total waking objective flashes), false-
positive (false-positive/total waking subjective flashes), false-negative (false-
negative/total waking objective flashes), and true-negative (true-negative/total
waking 20-minute intervals without objective flash) reporting rates were
Participants completed a pocket-sized, structured paper diary during wak-
ing monitoring hours. The validated behavioral diary of Hedges and col-
leagues (33), predictive of ischemic events (28,34), was adapted for hot
flashes (9,18). Each page assessed time of entry, dichotomous ratings of hot
flash and physical exertion occurrence at time of entry, indication of caffeine,
tobacco, and alcohol consumption since last entry, and 5-point ratings (“not at
all” to “very much”) of the following emotions: frustrated, sad, tired, stress,
relaxed, happy, and in control, and mental and physical effort.
Participants completed a battery of questionnaires selected for established
reliability and validity, use in prior menopause studies, and assessment of key
domains relevant to hot flashes.
1. The Attitudes Toward Menopause scale is 7 items assessing stereotyp-
ical attitudes about menopause (35,36) shown to be predictive of meno-
pausal symptom reporting (37).
2. The Beck Depression Inventory–II (BDI-II) (37) assesses 21 symptoms
experienced within the previous 2 weeks. It is a reliable and valid measure
(38) and a preferred scale for depressive symptom assessment among
symptomatic menopausal women (39).
3. The Daily Stress Inventory (DSI) (40) assesses the occurrence and
perceived stress associated with 58 events occurring over 24 hours.
4. The Symptom Checklist-90-R (SCL-90-R) is a 90-item inventory
assessing 9 symptom dimensions (41), including somatization, considered
in these analyses. It is reliable, valid, and has strong internal consistency
5. The Spielberger State-Trait Anxiety Inventory (STAI) (42), a 40-item
scale with strong psychometric properties (42), assesses anxiety present at
testing (state anxiety) and the propensity toward anxiety (trait anxiety).
The primary analysis involved identifying antecedents of 1) “objective”
hot flashes (hot flashes meeting physiological criteria); and 2) “false-positive”
hot flashes (reported hot flashes failing to meet physiological criterion). In the
case-crossover analytic technique (28,29), exposure before an event is com-
pared with exposure before 1 or more nonevent control periods (43). Each
participant passes through exposure and nonexposure periods, serving as her
own control and eliminating between-person confounds.
Diary entries before an objective or false-positive flash were matched to
entries during control periods. Waking flashes only were included as a result
of absence of reporting during sleep. A 30-minute case period preceding
flashes was the minimum interval with maximum diary entry and associated
flash retention. When case periods contained more than 1 entry, the entry
closest to the flash was used. Two controls were used: a 30-minute interval
not preceding a flash on the opposite monitoring day matched on time and
participant and all 30-minute intervals not preceding flashes matched on
participant. Minimum induction periods and carryover effects were zero.
Odds ratios and confidence intervals for flash antecedents were estimated
with generalized estimating equations (GEE), a class of generalized linear
R. C. THURSTON et al.
138 Psychosomatic Medicine 67:137–146 (2005)
models for nonindependent error structures as with repeated observations on
individuals (44), controlling for time of day. A model with a binomial
outcome distribution and a logit link was estimated. An exchangeable struc-
ture was applied to the covariance matrix after examination of actual and
estimated covariance matrices. Time of day was divided into 3 periods and
dummy-coded with evening as reference. Predictors were diary-reported
emotions, physical effort and exertion, and tobacco, caffeine, or alcohol use.
The effect period was zero for analyses of emotion during flashes.
Relations between physiological or false-positive flash frequency and
individual characteristics were estimated within a generalized linear model
with a Poisson distribution and a log link, offset by monitoring duration. A
dispersion parameter estimated based on Pearson’s chi-square was applied to
the variance function to correct for overdispersion. Predictors were depres-
sion, anxiety, somatization, daily stress, and attitudes toward menopause scale
sums. A sum score for continuous Attitudes Toward Menopause scale items
was derived by reverse scoring negative items, yielding a possible score range
of 6 to 24, higher scores indicating more positive attitudes (45). Daily stress
scores were averaged over 2 days.
Relations between true-positive, true-negative, false-positive, and false-
negative reporting ratios, and psychologic variables were estimated within a
generalized linear model with a logit link. A dispersion parameter estimated
based on Pearson’s chi-square was applied to the variance function. All
analyses were conducted with and without antidepressants in the model.
Unadjusted models are presented given the lack of impact of this covariate on
results. Goodness of model fit was determined from examination of deviance
and log likelihood values. Analyses were conducted using SAS V8.0 (SAS
Institute, Cary, NC).
Forty-two perimenopausal and postmenopausal women
were recruited from the Research Triangle community (Ra-
leigh, Durham, and Chapel Hill, NC). They completed 2
monitoring days, yielding 84 monitoring days.
Demographic and Medical Characteristics
Participants’ average age was 50.5 years (standard devia-
tion [SD], 4.8; range, 40–60) and over half (50.5%, n ? 21)
belonged to a minority ethnic group. Although no participants
had medical or psychiatric conditions associated with hot flash
sensations, participants reported other conditions, most com-
monly hypertension (28.6%, n ? 12), arthritis (21.4%, n ? 9),
and hyperlipidemia (14.3% n ? 6). All participants reported
daily hot flashes. Among early perimenopausal women, the
average duration of menstrual irregularities was 15.4 months
(SD, 9.0), all had irregularities for ?6 months, and none had
a history of irregularities during peak reproductive years. Most
(88.1%, n ? 37) women retained their uterus and both ovaries,
although 2 had undergone hysterectomy with bilateral oopho-
rectomy and 3 women had had unilateral oophorectomy with-
out hysterectomy. No women had taken exogenous estrogen
or progesterone (eg, HRT, oral contraceptives) or other med-
ications known to affect hot flashes within a month of study
participation. However, 2 (4.8%) women were taking soy
supplements, 5 (11.9%) were taking vitamin E, and 7 (16.7%)
were taking antidepressants. Participant characteristics are
presented in Table 1.
The mean BDI-II score was 10.5 (SD, 10.6; range, 0–50),
indicating minimal depressive symptoms (38). However, ap-
proximately 21.4% (n ? 9) of participants had mild to mod-
erate (14–28) and 7.1% (n ? 3) had severe depressive symp-
toms (?28). Participants exhibited typical state (mean, 35.0;
SD, 11.6; range, 20–60) and trait anxiety (mean, 37.9; SD,
10.4; range, 20–67) levels (42). Attitudes Toward Menopause
scores revealed women’s attitudes to be relatively neutral
(mean, 16.2; SD, 2.8; range, 11–24). Somatization (mean,
0.67; SD, 0.66; range, 0–28) was slightly elevated (58th
percentile) relative to nonpsychiatric females (41). Although
several scale items assessed hot flash-like symptoms (eg, hot
and cold spells), this scale was not a proxy for hot flashes,
with somatization unrelated to flash frequency (rate ratio
[RR], 0.79; 95% confidence interval [CI], 0.58–1.09; p ? not
significant). Finally, participants indicated relatively low daily
stress (mean, 39.0; SD, 29.9; range, 4–142) compared with
normative female data (40).
TABLE 1. Demographic Characteristics of Study Participants
No. (%) of Participants (n ? 42)
Age (mean yrs)
Household income (median)*
Marital status (%)
No. of children (%)
50.5 (SD ? 4.8)
*Three participants declined to answer.
†Hysterectomy with bilateral oophorectomy.
‡One participant is represented in more than 1 category.
SD ? standard deviation; SSRI ? selective serotonin reuptake inhibitor.
EMOTIONAL ANTECEDENTS OF HOT FLASHES
139Psychosomatic Medicine 67:137–146 (2005)
The participants completed 3416 diary entries, with a mean
(SD) of 81.3 (15.7) and a range of 32 to 108 entries per person.
Diary negative emotion ratings were low. The average
(SD) frustration, sadness, stressed, and tired ratings were 0.41
(0.81), 0.21 (0.56), 0.69 (0.95), and 0.97 (1.15), respectively.
High (?2) negative emotion levels were infrequent, with
4.0%, 1.5%, and 6.4% of entries associated with high frustra-
tion, sadness, and stress, respectively. Approximately 14.4%
entries indicated high tired ratings. High negative emotion
levels were evenly distributed, with 57.1%, 33.3%, 66.7%,
and 78.6% of women indicating high frustration, sadness,
stress, and tiredness, respectively, in at least 1 entry.
Positive emotions were high, with average (SD) ratings of
happiness, relaxation, and control of 2.69 (1.18), 2.30 (1.26),
and 3.01 (1.09), respectively. High (?2) positive emotion
levels were common, with 66.2%, 49.2%, and 72.9% of en-
tries indicating high happiness, relaxation, and control, respec-
tively. High positive emotions were evenly distributed, with
95.2%, 97.6%, and 100% of participants reporting high hap-
piness, control, and relaxation, respectively, in at least 1 entry.
Physical exertion was rare, indicated in 116 (3.4%) of
entries, which was expected given instructions to refrain from
rigorous physical activities during monitoring. Mean (SD)
physical effort ratings were also low, at 0.92 (0.87), and only
5.2% of entries indicated high (?2) physical effort. Tobacco,
caffeine, and alcohol consumption was rare, reported in 2.5%,
4.1%, and 1.1% of entries, respectively.
Ambulatory Hot Flash Monitoring
We distinguished two types of hot flashes: objective hot
flashes, or flashes meeting physiological criterion (?2 ?mho
increase in 30-second period) irrespective of self-report, and
subjective hot flashes, or subjectively-reported hot flashes
lacking a corroborating physiological flash. We considered
these latter flashes “false-positive” flashes.
Objective Hot Flashes
Each participant underwent an average (SD) of 27.5 (2.7)
waking and 11.3 (4.4) sleeping monitoring hours for a total of
1153.5 waking and 475.7 sleeping monitoring hours across
participants. They experienced 923 objective flashes, includ-
ing 742 waking and 181 sleeping flashes. Each woman expe-
rienced an average (SD) of 8.8 (5.6) waking (median, 8; range,
1–25) and 3.7 (2.2) sleeping (median, 2; range, 0–8) flashes
per day. For the sample as a whole, participants experienced a
flash every 1.5 hours while awake and every 2.6 hours while
asleep. Compared with evening flashes (6 to 10 PM), flashes
were more likely in late morning/early afternoon (10 AM to
1:59 PM; odds ratio [OR], 1.39; 95% CI, 1.13–1.71; p ? .002)
and late afternoon/early evening (2 to 5:59 PM; OR, 1.50; 95%
CI, 1.23–1.84; p ? .0001). Odds of hot flashes in late morn-
ing/early afternoon relative to late afternoon/early evening
were similar. See Figure 1 for an example of a sternal skin
conductance-recorded hot flash.
Adjusted for waking monitoring duration, the number of
waking objective flashes was not significantly related to
menopausal status, surgical status, oophorectomy, past HRT
use, supplement use, age, race, income, employment, marital
status, smoking, or alcohol consumption. However, women
with more children (RR, 1.32; 95% CI, 1.08–1.61; p ? .006)
had a higher hot flash rate. Women with a high school edu-
cation had twice the flash frequency (RR, 2.21; 95% CI,
Figure 1. Objectively recorded hot flash.
R. C. THURSTON et al.
140 Psychosomatic Medicine 67:137–146 (2005)
1.15–4.11; p ? .01) relative to women with a graduate edu-
cation. Moreover, regular aerobic exercise was associated
with a marginally lower (RR, 0.73; 95% CI, 0.51–1.04; p ?
.08) and antidepressant use with a significantly lower hot flash
rate (RR, 0.56; 95% CI, 0.22–0.98; p ? .04).
More negative psychologic states were associated with fewer
flashes, controlling for monitoring duration. For every 1-unit in-
for trait anxiety. Women with more positive attitudes toward meno-
pause had more frequent hot flashes (RR, 1.07; 95% CI, 1.01–1.15;
p ? .02). The frequency of objective hot flashes was not signifi-
cantly related to depression, somatization, or daily stress (see Table
2). Results were largely unchanged restricting analyses to women
not taking antidepressants.
Examination of hot flash reporting indicated 44.0% of
objective hot flashes were accompanied by a subjective report
(true-positive reporting rate), and 56.0% of objective hot
flashes were not accompanied by subjective report (false-
negative reporting rate; see Table 3). Women with higher state
anxiety (OR, 1.03; 95% CI, 1.00–1.06; p ? .06) and daily
stress (OR, 1.01; 95% CI, 1.00–1.01; p ? .05) had a margin-
ally increased false-negative reporting rate. True-positive re-
sults were symmetric to false-negative results.
Emotions and Behaviors Associated With Objective Hot
To characterize factors associated with the occurrence of
objectively monitored hot flashes, the likelihood of hot flashes
after diary-rated emotions or behaviors were examined rela-
tive to control periods.
The case-crossover analysis revealed that objective flashes
were less likely after increased negative emotion. Compared
with diary entries not preceding flashes (Case-crossover re-
sults are using the control of all nonflash diary entries matched
on participant. Analyses using the control of 1 nonflash diary
entry on the alternate day of monitoring matched on time and
participant yielded no significant results.), the odds of hot
flash occurrence were 0.85 (95% CI, 0.74–0.98; p ? .004)
after increased frustration, 0.70 (95% CI, 0.54–0.91; p ?
.005) after increased sadness, and 0.82 (95% CI, 0.73–0.93;
p ? .001) after increased stress, controlling for time of day.
Findings for tired were not significant.
Objective hot flashes were more likely to occur after in-
creased positive emotion. Compared with diary entries not
preceding flashes, odds of hot flash occurrence was 1.17 (95%
CI, 1.03–1.30; p ? .004) after increased happiness, 1.12 (95%
CI, 1.03–1.23; p ? .008) after increased relaxation, and 1.16
(95% CI, 1.01–1.35; p ? .006) after increased control, con-
trolling for time of day (see Figure 2). Notably, odds ratios are
those associated with 1-unit increases in emotion ratings.
Two diary measures of physical activity were considered: a
dichotomous physical exertion report and a continuous physical
effort rating. Results indicated a marginally increased likelihood
of hot flashes after physical exertion (OR, 1.49; 95% CI, 0.99–
2.25; p ? .05), controlling for time of day. Hot flashes were
significantly more likely after high (?2; OR, 1.51; 95% CI,
1.18–1.95; p ? .001) versus low (?2) physical effort. No sig-
nificant pattern emerged for tobacco, caffeine, and alcohol use.
Emotions and Behaviors Occurring During Objective
No significant pattern of emotions or physical activity was
observed at the time of flash onset, both for all objective hot
flashes and only those that were reported. However, objective
flashes were more likely during caffeine use (OR, 2.24; 95%
CI, 1.31–3.81; p ? .003) relative to control times. Given diary
instructions to indicate consumption since the last entry, this
finding indicates an increased likelihood of hot flashes during
or in 20 minutes after caffeine use.
Together, results indicate that objective hot flashes were
less likely to occur after increased negative emotion, more
likely after increased positive emotion, and more likely after
high physical effort. Hot flashes were also more likely during
TABLE 2. Individual Characteristics and Frequency of Waking
Objective Hot Flashes
Parity (for each additional
Regular aerobic exercise
TABLE 3. Concordance Between Objective and Subjective Waking Hot Flashes
Objective Hot Flash (n)
Yes NoTotal Subjective
555 Subjective hot flash (n)Yes
EMOTIONAL ANTECEDENTS OF HOT FLASHES
141Psychosomatic Medicine 67:137–146 (2005)
or shortly after caffeine use. Emotional patterns before hot
flashes were not markedly altered controlling for physical
exertion, physical effort, or caffeine use. They were largely
unchanged simultaneously controlling for antidepressant use,
parity, education, anxiety, and attitudes toward menopause.
Finally, restricting analyses to only objective hot flashes that
were subjectively reported did not markedly alter results.
Thirty-seven participants experienced 208 false-positive
flashes (subjectively reported but lacking physiological cor-
roboration). A mean (SD) of 2.8 (4.2) false-positive flashes
(median, 2; range, 1–28) were reported per woman per day.
False-positive flashes showed no significant diurnal rhythm.
Controlling for monitoring duration, women with increased
somatization (RR, 1.57; 95% CI, 1.05–2.33; p ? .03) had more
frequent false-positive flashes and African American women
(RR, 1.90; 95% CI, 1.03–3.50; p ? .04) had more frequent
false-positive flashes relative to women of other ethnic back-
The proportion of subjectively reported flashes lacking a corre-
sponding objective flash (false-positive reporting rate) was 43.2%
(see Table 3). The percentage of 20-minute monitoring intervals
lacking an objective and a subjective hot flash (true-negative report-
Figure 2.Odds ratios of objective hot flash occurrence in 30 minutes after emotional arousal.
R. C. THURSTON et al.
142Psychosomatic Medicine 67:137–146 (2005)
ing rate) was 92.3%. Increased depression (OR, 1.04; 95% CI,
1.001–1.08; p ? .04), state anxiety (OR, 1.05; 95% CI, 1.01–1.10;
p ? .01), trait anxiety (OR, 1.05; 95% CI, 1.01–1.10; p ? .02), and
somatization (OR, 2.76; 95% CI, 1.36–5.57; p ? .005), were asso-
ciated with an increased false-positive reporting rate. More positive
attitudes toward menopause was associated with a lower (OR, 0.90;
95% CI, 0.7–0.99; p ? .04) false-positive reporting rate. Women
anxiety, somatization, and more negative attitudes were associated
with increased reports of hot flashes lacking a corresponding objec-
tive hot flash.
Emotions and Behaviors Associated With False-Positive
To characterize factors associated with occurrence of false-
positive hot flashes, the likelihood of false-positive hot flashes
after diary-rated emotions or behaviors were examined rela-
tive to control periods.
Results indicated that the likelihood of a false-positive
flash was significantly increased after increased frustration
(OR, 1.19; 95% CI, 1.01–1.40; p ? .04) and significantly
decreased after increased feelings of control (OR, 0.82; 95%
Figure 3.Odds ratios of a false-positive (FP) flash report in 30 minutes after emotional arousal.
EMOTIONAL ANTECEDENTS OF HOT FLASHES
143 Psychosomatic Medicine 67:137–146 (2005)
CI, 0.71–0.94; p ? .005) (see Figure 3). Odds ratios are those
associated with every 1-unit increase in emotion.
Physical Activity and Health Behaviors
False-positive flashes were significantly more likely after
physical exertion (OR, 2.71; 95% CI, 1.61–4.62; p ? .0002)
compared with control times. No significant patterns for to-
bacco, alcohol, or caffeine use were observed.
Emotions and Behaviors Occurring During False-
Analyses of emotions coinciding with false-positive flashes
revealed that false-positive flashes were marginally more
likely during increased frustration (OR, 1.26; 95% CI, 0.99–
1.59; p ? .05) and physical exertion (OR, 2.39; 95% CI,
0.88–6.55; p ? .09) and marginally less likely during in-
creased happiness (OR, 0.87; 95% CI, 0.74–1.01; p ? .07)
compared with control times. False-positive flashes were sig-
nificantly more likely during tobacco use (OR, 2.92; 95% CI,
1.32–6.55; p ? .008), which, given diary instructions, indi-
cates that false-positive flashes were more likely during or in
20 minutes after tobacco use compared with control times.
In contrast to objective hot flashes, false-positive flashes
were more likely after increased frustration and decreased
feelings of control. They were also more likely after physical
exertion and during or after tobacco use. Moreover, the expe-
rience of a false-positive flash was characterized by slightly
increased negative emotion.
Contrary to our expectations, objectively measured hot
flashes were significantly less likely after increased sadness,
stress, and frustration and were significantly more likely after
increased happiness, relaxation, and feeling in control. More-
over, increased anxiety and negative attitudes toward meno-
pause were associated with fewer hot flashes. Thus, the in-
verse relationship between negative psychologic states and hot
flashes was consistent across between-individual psychologic
factors and within-individual emotions.
False-positive flashes, or flash reports lacking correspond-
ing objective physiological evidence, were common, repre-
senting over 43% of reported flashes. In contrast to objective
physiological flashes, post hoc analyses revealed that false-
positive flashes were more likely after elevated frustration and
decreased feelings of control. Moreover, increased depres-
sion, state anxiety, trait anxiety, somatization, and more neg-
ative attitudes toward menopause were associated with an
increased proportion of reported flashes lacking a correspond-
ing objective flash.
Taken as a whole, results suggest that positive emotions
were associated with objective hot flashes and negative emo-
tions associated with false-positive flashes. This pattern oc-
curred across diary-rated emotions and questionnaire-assessed
mood and attitudes. Thus, commonly held beliefs that “stress”
or negative emotion can trigger hot flashes may be the result
of the experience of false-positive flashes with negative emo-
Emotions and behaviors affect symptom perception and
reporting. For example, psychologic processes are central to
theories of pain perception (46). In laboratory studies, in-
creased reporting of aches and pain (22) and decreased toler-
ance for experimental pain (47) is evident after negative mood
induction. In contrast, increased pain tolerance is evident with
positive mood induction (47). Moreover, in a study of patients
experiencing ischemic heart disease (48), 66% of anginal pain
reports occurred in the absence of ischemia, and physical
exertion, physical effort, and negative emotion increased the
likelihood of angina reports. Thus, emotions and behaviors are
associated with a range of physical symptom reporting.
Emotions were associated with the occurrence of hot
flashes. Mechanisms behind these relationships are not well-
characterized, in part as a result of limited understanding of
hot flash physiology. Current research indicates a narrowed
thermoneutral zone among menopausal women (49), with hot
flashes potentially representing a heat-dissipating mechanism.
Research also suggests a role of central norepinephrine acting
on the hypothalamic thermoregulatory center in hot flash
occurrence (50,51). However, other systems, including sero-
tonergic (52) and opiate (53) systems have been implicated.
Moreover, other yet unexamined systems may be involved.
Oxytocin, a neurohormone associated with relaxation, certain
positive emotions, and affiliative behaviors (54,55), when
centrally infused, increases core body temperature (54). How-
ever, given the limited understanding of hot flash physiology,
these links remain speculative.
Hot flashes are believed to be aversive or distressing (9).
However, in this study, most hot flashes were not reported and
there was no significant pattern of emotions during hot
flashes, both for all objective flashes and objective hot flashes
that were reported. Notably, false-positive flashes were asso-
ciated with marginally increased frustration and decreased
happiness at the time of the false-positive flash. These results
challenge the notion that physiological hot flashes are neces-
sarily accompanied by emotional distress. Moreover, it sug-
gests potential clinical significance of false-positive flashes.
Women with lower education and more children had more
frequent hot flashes. Epidemiologic evidence comparing
women with and without reported hot flashes have shown
similar relationships for education (1,3,5) and to a lesser
extent for parity (1). Mechanisms behind these associations
are not known but may include differences in body composi-
tion, health behaviors, ovarian function, and HRT use. This
study is notable for demonstrating these relationships with
physiological flash frequency as opposed to single-item self-
report measures of hot flashes.
Hot flashes were related to health behaviors. They were
more likely after high physical effort, consistent with in-
creased core body temperature acting as a trigger (49). How-
ever, regular aerobic exercisers had somewhat fewer waking
hot flashes, consistent with epidemiologic studies suggesting
hot flashes are less likely (1,7) or less severe (56) among
R. C. THURSTON et al.
144Psychosomatic Medicine 67:137–146 (2005)
regular exercisers, although previous results have been incon-
sistent and limited by brief self-report measures. Caffeine may
be a trigger, with objective flashes over twice as likely during
or after caffeine use. Finally, findings suggested that sensa-
tions associated with physical exertion and tobacco use may
have been perceived as hot flashes in light of their association
with false-positive flashes. Thus, preliminary results suggest a
role of health behaviors in the occurrence of, propensity
toward, and perception of hot flashes.
This study has several important strengths. To our knowl-
edge, it is currently the largest and most ethnically diverse
sample of women studied to date in relation to physiological
hot flashes. Moreover, it is the only study evaluating emo-
tional antecedents of hot flashes in an ambulatory setting
using the case-crossover analytic technique, prospectively as-
sessed emotion, and physiological hot flash recording. How-
ever, despite the relatively large sample of women and over
1629 person-hours of monitoring, the study may have lacked
adequate power for exploratory and between-person analyses.
Although the ethnic diversity and low flash frequency eligi-
bility criteria increase the study’s generalizability, given the
size and the volunteer nature of the sample, participants may
not be representative of menopausal women. Moreover, al-
though sternal skin conductance is the most sensitive and
specific measure of hot flashes (23–25,27), further validation
of this measure under varying conditions is merited. Further-
more, false-positive flashes were few relative to physiological
flashes and were examined in a post hoc fashion. These results
must be regarded as tentative. Finally, given the low fre-
quency of physical exertion and alcohol, caffeine, and tobacco
use, these results should be regarded as suggestive of these
In summary, study results suggest an inverse relation be-
tween negative emotions and physiological flashes a positive
relation between negative emotions and false-positive flashes.
Given these findings, the anecdotal association between neg-
ative emotion or stress and hot flashes may actually be based
on reported flashes lacking physiological corroboration. This
finding is important given the current reliance on self-report
measures of hot flashes, indicating flash reports may diverge
from physiological flashes markedly among certain individu-
als or in the context of certain emotions. These findings
underscore the importance of both objective and subjective
measures of hot flashes. Both objective and subjective mea-
sures may be particularly important in studies evaluating be-
havioral or pharmacologic interventions designed to reduce
hot flashes that may simultaneously affect psychological func-
This research was conducted in partial fulfillment of requirements
for a doctoral degree in clinical health psychology at Duke Univer-
sity. The authors thank Ginger Henshall for help with data prepa-
ration/programming and Janet Carpenter for help with reliability
coding and sternal skin conductance monitoring.
1. Gold EB, Sternfeld B, Kelsey JL, Brown C, Mouton C, Reame N,
Salamone L, Stellato R. Relation of demographic and lifestyle factors to
symptoms in a multi-racial/ethnic population of women 40–55 years of
age. Am J Epidemiol 2000;152:463–73.
2. von Muhlen DG, Kritz-Silverstein D, Barrett-Connor E. A community-
based study of menopause symptoms and estrogen replacement in older
women. Maturitas 1995;22:71–8.
3. Avis NE, Stellato R, Crawford S, Bromberger J, Ganz P, Cain V,
Kagawa-Singer M. Is there a menopausal syndrome? Menopausal status
and symptoms across racial/ethnic groups. Soc Sci Med 2001;52:345–56.
4. Bachmann GA. Vasomotor flushes in menopausal women. Am J Obstet
5. Collins A, Landgren BM. Reproductive health, use of estrogen and
experience of symptoms in perimenopausal women: a population-based
study. Maturitas 1994;20:101–11.
6. Bromberger JT, Assmann SF, Avis NE, Schocken M, Kravitz HM,
Cordal A. Persistent mood symptoms in a multiethnic community cohort
of pre- and perimenopausal women. Am J Epidemiol 2003;158:347–56.
7. Dennerstein L. Well-being, symptoms and the menopausal transition.
8. Guthrie JR, Dennerstein L, Taffe JR, Donnelly V. Health care-seeking for
menopausal problems. Climacteric 2003;6:112–7.
9. Kronenberg F. Hot flashes: epidemiology and physiology. Ann N Y Acad
Sci 1990;592:52–86; discussion 123–33.
10. Rogers J. The menopause. N Engl J Med 1956;254:697–704.
11. Molnar GW. Body temperatures during menopausal hot flashes. J Appl
12. Stevenson DW, Delprato DJ. Multiple component self-control program
for menopausal hot flashes. J Behav Ther Exp Psychiatry 1983;14:
13. Germaine LM, Freedman RR. Behavioral treatment of menopausal hot
flashes: evaluation by objective methods. J Consult Clin Psychol 1984;
14. Freedman RR, Woodward S. Behavioral treatment of menopausal hot
flushes: evaluation by ambulatory monitoring. Am J Obstet Gynecol
15. Irvin JH, Domar AD, Clark C, Zuttermeister PC, Friedman R. The effects
of relaxation response training on menopausal symptoms. J Psychosom
Obstet Gynaecol 1996;17:202–7.
16. Wijma K, Melin A, Nedstrand E, Hammar M. Treatment of menopausal
symptoms with applied relaxation: a pilot study. J Behav Ther Exp
17. Swartzman LC, Edelberg R, Kemmann E. Impact of stress on objectively
recorded menopausal hot flushes and on flush report bias. Health Psychol
18. Gannon L, Hansel S, Goodwin J. Correlates of menopausal hot flashes.
J Behav Med 1987;10:277–85.
19. Bradburn NM, Rips LJ, Shevell SK. Answering autobiographical
questions: the impact of memory and inference on surveys. Science
20. Stone AA, Shiffman S. Capturing momentary, self-report data: a proposal
for reporting guidelines. Ann Behav Med 2002;24:236–43.
21. Barsky AJ, Goodson JD, Lane RS, Cleary PD. The amplification of
somatic symptoms. Psychosom Med 1988;50:510–9.
22. Salovey P, Birnbaum D. Influence of mood on health-relevant cognitions.
J Pers Soc Psychol 1989;57:539–51.
23. de Bakker IP, Everaerd W. Measurement of menopausal hot flushes:
validation and cross-validation. Maturitas 1996;25:87–98.
24. Freedman RR. Laboratory and ambulatory monitoring of menopausal hot
flashes. Psychophysiology 1989;26:573–9.
25. Tataryn IV, Lomax P, Meldrum DR, Bajorek JG, Chesarek W, Judd HL.
Objective techniques for the assessment of postmenopausal hot flashes.
Obstet Gynecol 1981;57:340–4.
26. Kuno Y. The Physiology of Human Perspiration; 1934.
27. Carpenter JS, Andrykowski MA, Freedman RR, Munn R. Feasibility and
psychometrics of an ambulatory hot flash monitoring device. Menopause
28. Gullette EC, Blumenthal JA, Babyak M, Jiang W, Waugh RA, Frid DJ,
O’Connor CM, Morris JJ, Krantz DS. Effects of mental stress on myo-
cardial ischemia during daily life. JAMA 1997;277:1521–6.
29. Mittleman MA, Maclure M, Tofler GH, Sherwood JB, Goldberg RJ,
Muller JE. Triggering of acute myocardial infarction by heavy physical
EMOTIONAL ANTECEDENTS OF HOT FLASHES
145 Psychosomatic Medicine 67:137–146 (2005)
exertion. Protection against triggering by regular exertion. Determinants Download full-text
of Myocardial Infarction Onset Study Investigators. N Engl J Med
30. Lykken DT, Venables PH. Direct measurement of skin conductance: a
proposal for standardization. Psychophysiology 1971;8:656–72.
31. Schneider RE, Fowles DC. A convenient, non-hydrating electrolyte me-
dium for the measurement of electrodermal activity. Psychophysiology
32. Carpenter JS, Gautam S, Freedman RR, Andrykowski M. Circadian
rhythm of objectively recorded hot flashes in postmenopausal breast
cancer survivors. Menopause 2001;8:181–8.
33. Hedges SM, Krantz DS, Contrada RJ, Rozanski AR. Development of a
diary for use with ambulatory monitoring of mood, activities, and phys-
iological function. Journal of Psychopathology & Behavioral Assessment
34. Gabbay FH, Krantz DS, Kop WJ, Hedges SM, Klein J, Gottdiener JS,
Rozanski A. Triggers of myocardial ischemia during daily life in patients
with coronary artery disease: physical and mental activities, anger and
smoking. J Am Coll Cardiol 1996;27:585–92.
35. Kaufert PA. Women and their health in the middle years: a Manitoba
project. Soc Sci Med 1984;18:279–81.
36. Lock M. Ambiguities of aging: Japanese experience and perceptions of
menopause. Cult Med Psychiatry 1986;10:23–46.
37. Avis NE, McKinlay SM. A longitudinal analysis of women’s attitudes
toward the menopause: results from the Massachusetts Women’s Health
Study. Maturitas 1991;13:65–79.
38. Beck AT, Steer RA, Brown GK. Beck Depression Inventory Manual, 2nd
ed. Orlando: The Psychological Corp; 1996.
39. Gath D. The assessment of depression in peri-menopausal women. Ma-
40. Brantley P, Waggoner C, Jones G, Rappaport N. A daily stress inventory:
development, reliability, and validity. J Behav Med 1987;10:61–74.
41. Derogatis LR. SCL-90-R, Administration, Scoring, and Procedures Man-
ual for the Revised Version, 2nd ed. Towson, MD: Clinical Psychometric
42. Spielberger CD. Manual for the State-Trait Anxiety Inventory. Palo Alto:
Consulting Psychologists Press; 1983.
43. Maclure M. The case-crossover design: a method for studying transient
effects on the risk of acute events. Am J Epidemiol 1991;133:144–53.
44. Liang KY, Zeger SL. Regression analysis for correlated data. Annu Rev
Public Health 1993;14:43–68.
45. Sommer B, Avis N, Meyer P, Ory M, Madden T, Kagawa-Singer M,
Mouton C, Rasor NO, Adler S. Attitudes toward menopause and aging
across ethnic/racial groups. Psychosom Med 1999;61:868–75.
46. Melzack R, Wall PD. Pain mechanisms: a new theory. Science 1965;150:
47. Zelman DC, Howland EW, Nichols SN, Cleeland CS. The effects of
induced mood on laboratory pain. Pain 1991;46:105–11.
48. Krantz DS, Hedges SM, Gabbay FH, Klein J, Falconer JJ, Merz CN,
Gottdiener JS, Lutz H, Rozanski A. Triggers of angina and ST-segment
depression in ambulatory patients with coronary artery disease: evidence
for an uncoupling of angina and ischemia. Am Heart J 1994;128:703–12.
49. Freedman RR, Krell W. Reduced thermoregulatory null zone in post-
menopausal women with hot flashes. Am J Obstet Gynecol 1999;181:
50. Freedman RR. Biochemical, metabolic, and vascular mechanisms in
menopausal hot flashes. Fertil Steril 1998;70:332–7.
51. Freedman RR, Woodward S, Sabharwal SC. Alpha 2-adrenergic mech-
anism in menopausal hot flushes. Obstet Gynecol 1990;76:573–8.
52. Berendsen HH. The role of serotonin in hot flushes. Maturitas 2000;36:
53. Genazzani AR, Petraglia F, Facchinetti F, Facchini V, Volpe A, Ales-
sandrini G. Increase of proopiomelanocortin-related peptides during sub-
jective menopausal flushes. Am J Obstet Gynecol 1984;149:775–9.
54. Argiolas A, Gessa GL. Central functions of oxytocin. Neurosci Biobehav
55. Uvnas-Moberg K. Oxytocin linked antistress effects—the relaxation and
growth response. Acta Physiol Scand Suppl 1997;640:38–42.
56. Hammar M, Berg G, Lindgren R. Does physical exercise influence the
frequency of postmenopausal hot flushes? Acta Obstet Gynecol Scand
R. C. THURSTON et al.
146 Psychosomatic Medicine 67:137–146 (2005)