Anxiety sensitivity and menstrual cycle reactivity: Psychophysiological and self-report differences

Joumd

of AnxietyLGorders,Vol. 10,No. 5, pp. 393-410.1996 copyright 0 19% ElsevierseicnccLtd Printed in the USA. All rights reserved 0887-6189% S15.00+ .OLl

PII s(#187-6185(%poo19-9

Anxiety Sensitivity and Menstrual Reactivity: Psychophysiological Self-Report Differences

Cycle and

SANDRAT. SIGMON, PH.D., CHRISTINE M. FINK, PH.D., KELLY J. ROHAN, B.A., AND LISA A. HOTOW, B.A. University

of Maine

Abstract - The role of anxiety sensitivity in the etiology and maituetumce of various anxiety disorders has received increased attention over the past decade. To date, no studies have empirically addressed the relationship between anxiety sensitivity, physiological reactivity, and self-reports of anxiety symptomatology across the menstrual cycle. Jn this study, high- and low-anxiety sensitivity women in either the premenstrual or intermenstmaJ phase of the menstrual cycle completed questionnaires and listened to anxiety and neutral scenes while psychophysiological data were collected. In addition, mood ratings were obtained at baseline and after scene presentations. High anxiety sensitivity participants scored higher on measures of anxiety, depression, and menstrual distress than low anxiety sensitivity females. Premensttually. high anxiety sensitivity females exhihited greater skin conductance response frequency and magnitude to the anxiety scenescompared to the other thtee groups. Similar results were ohtained when initial levels of state anxiety and panic history were controlled for statistically. Furthermore., high anxiety sensitivity females reported more anxiety and depressed mood following presentation of anxiety scenes. Implications of these results for the mediion of menstrual cycle timing and anxiety sensitivity hypotheses are discussed.

Anxiety sensitivity refers to a fear of anxiety symptoms stemming from the belief that the experience of such symptoms will lead to harmful consequences (e.g., heart attack, mental instability, loss of consciousness; Taylor, Koch, McNally, 8z Crockett, 1992). Individuals high in anxiety sensitivity are more likely to monitor their physiological symptoms and continue to focus their Portions of this manuscript were presented at the annual meeting of the Association for the Advancement of Behavior Therapy, November, 1994, Atlanta. The authors would like to thank Janice Zeman, Jeffrey Hecker, and Diana Dorhofer for their helpful comments on an earlier version of this manuscript. Requests for reprints should he sent to Sandra T. Sigmon, Ph.D., University of Maine, 5742 Little Hall, Orono, Maine, 04469-5742. 393

394

S. T. SIGMON

ET AL.

attention on internal stimuli (Peterson dc Reiss, 1987). Alternatively, individuals low in anxiety sensitivity typically do not focus on their physiological arousal and, therefore, are less likely to believe that their symptoms of anxiety will have negative outcomes. Research has indicated that the anxiety sensitivity construct has implications for a variety of anxiety-related phenomena and may represent a cognitive risk factor in the development of panic disorder (Maller & Reiss, 1992). The debate continues, however, as to whether anxiety sensitivity represents a separate construct from trait anxiety (e.g., Borden & Lister, 1994). In expanding the parameters of anxiety sensitivity, recent studies have investigated the relation between self-report of anxiety and physiological arousal in response to stressful tasks. In a sample of college students scoring in the high, middle, and low ranges of the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986), Shostak and Peterson (1990) assessed EMG and systolic blood pressure in participants who engaged in a mental arithmetic task. Results indicated that all three groups exhibited similar increases in EMG and similar blood pressure responses across pretask, posttask, and postrelaxation phases. Individuals in the high anxiety sensitivity group, however, reported experiencing more physiological symptoms and anxiety following the stressful task than the other two groups. Because perceptions of arousal, not actual physiological changes, differentiated the three anxiety sensitivity groups, the authors proposed that individual differences in “anxiety sensitivity beliefs are a primary determinant of self-reported anxiousness in reaction to physiological arousal” (Shostak & Peterson, 1990, p. 518). Along related lines, Borden and Lister (1994) examined physiological arousal and associated cog&ions in college students with varying levels of anxiety sensitivity and recent experiences of panic attacks. Consistent with the findings of Shostak and Peterson (1990). groups varying in anxiety sensitivity demonstrated similar physiological changes throughout baseline, relaxation, and stress phases. Participants with a history of panic, however, exhibited greater physiological arousal during baseline. In contrast to Shostak and Peterson (1990), individual differences in anxiety sensitivity did not differentiate physiological responses across the experimental phases when state anxiety was controlled for in subsequent analyses. Preliminary findings suggested that a recent history of panic attacks resulted in increased reports of subjective distress during baseline and that nonpanickers reported more distress during a mental arithmetic task. Borden and Lister (1994) concluded that conditioning experiences associated with a history of panic attacks may be more predictive of differential responding on physiological and self-report measures of distress than anxiety sensitivity. In the above two studies, the samples of college students contained both males and females. Most of the participants, however, were female (65%). Shostak and Peterson (1990) even used different norms to assign males and females to the high, middle, and low anxiety sensitivity groups. Although pre-

ANXIETY

SENSITIVITY

395

vious research has suggested gender differences in anxiety sensitivity (Peterson & Reiss, 1987), no studies have examined gender-related variables that might influence or mediate the relationship between anxiety sensitivity, physiological arousal, and self-reports of anxiety symptoms. In particular, it might be important to investigate the relationship between menstrual cycle timing and anxiety sensitivity. Increased anxiety represents one of several physiological and psychological symptoms that has been associated anecdotally and empirically with the premenstrual phase of the menstrual cycle. Studies focusing on the relationships between physiological responses, self-reports of anxiety symptoms, and menstrual cycle timing have produced equivocal results depending upon whether retrospective or prospective methodologies were used. Similar to studies pertaining to the general female population (e.g., Golub, 1976; Veith et al., 1984), retrospective self-reports of increased anxiety during the premenstrual phase have been found for women with agoraphobia (e.g., Breier, Chamey, Heninger, 1986) as well as panic disorder (e.g., Cameron, Kuttesch, McPhee, & Curtis, 1988). Studies employing prospective methodology to investigate the relationship between self-reports of anxiety and menstrual cycle timing have, however, also produced conflicting results. Results of several studies did not reveal increases in reports of anxiety symptoms or panic attacks during the premenstrual phase (Cameron et al., 1988; Cook et al., 1990), whereas other researchers have found increases in self-reports of both anxiety and panic attacks (e.g., Kaspi, Otto, Pollack, Eppinger, & Rosenbaum, 1994). Different definitions and assessment of menstrual cycle phase, as well as differences in methodology and other measurements, may have contributed to these disparate findings. Although one study (Roth et al., 1986) has examined nonspecific physiological arousal in agoraphobic women with panic attacks, no studies to date have investigated changes in physiological arousal across the menstrual cycle in anxiety-disordered individuals. Attempts to account for the conflicting results in the anxiety and menstrual cycle literature regarding retrospective and prospective accounts of menstrual symptoms have focused on body sensations and cultural expectations hypotheses. According to the body sensations hypothesis, hormonal fluctuations are responsible for the production of bodily sensations or noticeable changes in arousal levels (Klebanov & Jemmott, 1992). In other words, a woman’s report of menstrual symptoms represents a veridical account of physiological changes that are occurring in her body. Reports of physical symptoms across the menstrual cycle are thought to be less affected by suggestion than reports of psychological symptoms (Metcalf & Hudson, 1985). According to expectation accounts, cultural expectations or negative cultural stereotypes influence women’s perceptions regarding their menstrual cycle symptoms (Klebanov & Jemmott, 1992). Therefore, a woman may report more menstrual symptoms premenstrually based on knowledge of cycle timing and a tendency to attribute any physiological changes in a negative manner, regardless of the experience

396

S. T. SIGhiON

ET AL.

of actual symptoms. As such, an expectations hypothesis would predict that females will be more likely to attend to normal body sensations and interpret these sensations as related to the premenstrual phase of the menstrual cycle. Body sensations and expectations hypotheses may both have relevance for how women differing in anxiety sensitivity report menstrual symptoms. Theoretically, throughout the menstrual cycle, females high in anxiety sensitivity should be more attentive to changes in their physiological state and more likely to focus on subsequent changes in their arousal levels. Consistent with the body sensations hypothesis, females high in anxiety sensitivity may be more sensitive to menstrual cycle physiological changes due to their heightened awareness of physical sensations. In accord with the expectations hypothesis, females high in anxiety sensitivity may expect to experience more intense symptoms premenstrually, leading to self-report differences. In addition, both hypotheses may have implications for psychophysiological responding. Reiss (1987) proposed a vicious cycle hypothesis in which anxiety sensitivity operates as a risk factor for panic attacks, and panic attacks further increase anxiety sensitivity. If a high anxiety sensitivity female is more aware of and focused on physiological changes occurring during the premenstrual phase, this may lead to greater arousal and the reporting of more intense premenstrual symptoms. Because of the potential for menstrual cycle timing to influence psychophysiological responses and self-reports of anxiety, the present study investigated this variable in females with different levels of anxiety sensitivity. Although previous anxiety sensitivity studies have utilized mental stress tasks to minimize the potential for demand characteristics, these types of tasks may also create performance anxiety for all individuals, regardless of anxiety sensitivity level. In an effort to increase external validity, the present study improved upon previous research by using stimuli that were more relevant to everyday situations that might elicit anxiety. Due to conflicting results concerning retrospective and prospective accounts of menstrual cycle symptoms, measures of both types of information were obtained in the present study. Although previous studies have not consistently found differences in self-report measures of mood, this study included mood measures of anxiety and depression. It was predicted that high anxiety sensitivity females in the premenstrual phase would exhibit greater arousal to anxiety scenes and report more anxiety after listening to the anxiety scenes in comparison to the other three groups. METHOD Participants

College females (N = 337) were initially screened in large group sessions with the Anxiety Sensitivity Index (ASI; Reiss et al., 1986). In addition, participants completed questionnaires assessing state and trait anxiety, depres-

ANXIEI’Y

SENSlllVlTY

397

sion, panic, and menstrual distress. Participants were asked to indicate their phone number if they would like to be contacted for the experimental phase of the study. Females with normal menstrual cycles (26-35 days in length) who scored in the lower 25% (a score of 15 or lower) or in the upper 25% (a score of 25 or greater) on the AS1 and who indicated interest in participating (ZV= 123) were eligible for the second part of the study. Of the 123 females who were eligible, only 86 were able to be contacted for participation in the experimental part of the study. On the ASI, the low anxiety sensitivity (AS) participants scored an average of 9.76 (SD = 3.50) and the high AS females scored an average of 36.34 (SD = 5.96). Based upon random assignment, high- and low-AS participants were scheduled to return for the experimental session either during the intermenstrual (Days 8-22) or premenstrual phase (Days 24-28) of their next menstrual cycle. Necessary adjustments were made in scheduling for women with slightly longer or shorter menstrual cycles. MEASURES

Anxiety Sensitivity Index. The AS1 (Reiss & McNally,

1985; Reiss et al., 1986) is a 16-item self-report measure that assesses an individual’s fear of anxiety and anxiety symptoms. Scores can range from 0 to 64 with a mean of 18.4 (SD = 8.5) for nonclinical participants (Peterson & Reiss, 1987). The AS1 has been demonstrated to have acceptable reliability and validity in previous research (Reiss et al., 1986, Peterson & Heilbronner, 1985). In a subsequent study (Peterson & HeilbroMer, 1987), the AS1 had an alpha coefficient of .88.

Panic Attack Questionnaire. The Panic Attack Questionnaire (FAQ; Norton, Dorward, & Cox, 1986) assessesthe nature and frequency of panic symptoms according to DSM-ZZZ(American Psychiatric Association, 1980) criteria. The PAQ has been used in the panic literature to gather information concerning panic symptoms, severity of symptoms, panic history, and family history of panic in both clinical and nonclinical participants. In this study, we were interested in the number of panic attacks that participants had experienced, the severity of the attacks, and family history of panic attacks. Borden and Lister (1994) used a categorical variable of panic/no panic to assessfor the different effects of panic history. In addition to the presence or absence of panic, the number of attacks and associated distress of the attacks may be distinguishing factors in physiological sensitivity.

State-Trait Anxiety Inventory.

The State-Trait Anxiety Inventory Spielberger, Gorsuch, & Lushene, 1970) measures both transitory states and more stable trait anxiety. Adequate levels of reliability and have been established with the STAI through extensive use with both and nonclinical populations (Spielberger, 1988). Alpha coefficients

(STAI; anxiety validity clinical for the

398

S.T. SIGMON

ETAL.

revised STAI for both the state and trait scales average .90 for the normative samples (Spielberger, 1988). Inventory The Beck Depression Inventory (BDI; Beck & Steer, 1984) assessessymptoms of depressed mood and has been used widely in clinical and nonclinical samples. The BDI has been found to have acceptable levels of reliability and validity (Beck, Steer, & Garbin, 1988). The average coefficient alpha for the BDI with normal populations is .87 (Steer & Beck, 1988). Beck Depression

Distress Questionnaire. The Menstrual Distress Questionnaire (MDQ; Moos, 1969, 1985) is a 46-&m inventory that assesses a variety of symptoms that can be experienced during the menstrual cycle. Factor analyses have revealed eight factors: pain (e.g., headache, cramps), concentration (e.g., forgetfulness, distractible), behavioral change (e.g., stay at home, avoid social activities), autonomic reactions (e.g., dizziness, faintness, cold sweats), water retention (e.g., weight gain, swelling), negative affect (e.g., crying, anxiety), arousal (e.g., orderliness, feelings of well-being), and control (e.g., chest pains, heart pounding). It should be noted that the arousal factor pertains to positive symptoms relating to arousal and the control factor relates to symptoms usually associated with panic attacks. Individuals can complete the MDQ with regard to their last menstrual cycle (i.e., retrospective) or with regard to their experience of various symptoms over the past few days (i.e., prospective). The MDQ has demonstrated acceptable reliability and validity in subsequent research with average coefficient alphas for the subscales ranging from 64 to .88 (Moos, 1985). Menstrual

Profile of Moods State. The Profile of Mood States (POMS; McNair, Lorr, 8z Droppleman, 1971) was designed to assess transitory mood states. Research has indicated that this measure has acceptable reliability and validity with coefficient alphas for the subscales averaging 90 or above (McNair et al., 1971). Only the Tension-Anxiety and Depression-Dejection subscales were utilized in this study. The anxiety subscale consists of adjectives that describe somatic tension and the depression subscale consists of adjectives that describe feelings of personal inadequacy and sadness. Psychophysiological

Apparatus and Measurement

Skin conductance measurement was obtained utilizing the PSU unit of Contact Precision Instruments (London, U.K.) hardware in combination with PSYLAB software (Dow, 1991). Two Ag-AgCl electrodes were filled with a standard conductance paste in accordance with the recommendations of Fowles et al. (1981). The electrodes were attached via adhesive collars to the medial phalanxes of the two middle fingers on the participant’s nondominant hand. A constant voltage (0.6V) was applied through the skin conductance transducer.

ANXIETY

SENSlTWlTY

399

Baseline skin conductance (SCL) recordings were obtained for the first 5 min, and the average SCL was obtained at the end of the baseline period. After presentation of neutral and anxiety scenes, significant skin conductance responses (SCR) .05 microsiemens above the individual’s baseline SCL were recorded. Only the SCR that occurred during the 10-s period after the offset of each scene were utilized in subsequent analyses. To examine SCR magnitude differences, averages of the magnitudes were calculated for each block of scenes. Logarithmic transformations of SCL and SCR magnitude were conducted in order to minimize the effects of outliers on the data and to normalize the distribution of scores (Venables & Christie, 1980). Scenes

In order to provide participants with scenes of neutral and anxiety-provoking content, a pilot study was conducted. Scenes (N = 56) varying in affective content were rated by 100 undergraduates on a Likert scale (e.g., 1 = not very anxiety provoking, 4 = neutral, 7 = very anxiety provoking). Scenes that received the highest respective means for neutrality or anxiety (and the lowest standard deviations) were selected for the study. Four neutral content scenes (e.g., “Your telephone rings and a telephone company representative asks if you are interested in switching phone companies. You tell them no and hang up.“) and four anxiety content scenes (e.g., “You are sitting down to relax after a hectic day. Suddenly, you can’t breathe, your chest feels tight, and you think, this is it, I am going to die.“) were chosen. These prerecorded audiotaped scenes were presented via headphones, and participants were asked to vividly imagine themselves in each situation and to concentrate on their thoughts and feelings. Each scene lasted approximately 7 to 15 s, with a 20-s pause between each scene. Procedure

When participants returned for the experimental part of the study, they completed the state version of the STAI, the BDI, and a prospective version of the MDQ asking them to rate their experience of menstrual symptoms over the past few days. In addition, participants’ menstrual cycle phase was verified through self-report. Participants were then taken to a windowless and soundproof labo ratory room and seated in a comfortable chair While the experimental procedure was described to the individual, skin conductance electrodes were attached. Participants then sat quietly for 5 min during the baseline phase. To assessmood following baseline, participants completed the first POMS. Participants were told that they would be hearing different types of scenes over the headphones and were asked to try to imagine themselves in each situation. After the presentation of a block of four neutral scenes, participants completed the second PQMS and rated the scenes they had just heard on a l-7 Likert scale (1 = very pleasant, 7 = very unpleasant). After listening to a block

400

S. T. SIGMON

J5T AL.

of four anxiety scenes, participants completed the third POMS and rated the scenes on the pleasantness scale. Participants were debriefed regarding the experimental hypotheses and given experimental credit for their participation. In addition, all participants were given a referral list for counseling in accordance with university human subjects guidelines. RESULTS Means and standard deviations for the questionnaires completed at the initial screening session and at the experimental session are presented in Tables 1 and 2. Questionnaire Measures Screening session. Analyses of the questionnaires were conducted using analyses of variance (ANOVA) and multivariate analyses of variance (MANOVA) with post hoc comparisons employing ‘Iwrey’s HSD tests of significance. In order to control for Type I errors in post hoc comparisons, a conservative alpha of .OOl was used. For questionnaires completed at the screening session, comparisons were made between low- and high-AS females. Although an ANOVA on the PAQ revealed that high-AS females reported experiencing more panic attacks in the last year, F(1.26) = 4.44, p < .05, this effect was not significant at the .OOl level. High- and low-AS females who had experienced panic attacks in the past year did not significantly dit%r in their reported distress, F(1, 24) = 1.70, m. Although high-AS females cited more family members who had experienced a panic attack compared to low-AS females, F(1, 46) = 7.69, p < .Ol, this effect was not significant at the .OOl level. In addition, high-AS females reported more trait anxiety, F(1, 82) = 34.98, p < .OOOl; more state anxiety, F(1, 82) = 27.11, p < .OOOl; and more symptoms of depression, F(1,82) = 43.19, p < .OOOl than low-AS females. Results of a MANOVA on the retrospective MDQ, Wilks’s lambda, F(8,76) = 5.24, p < .OOOl, and subsequent tivariate analyses’ indicated that high-AS females scored significantly higher than low-AS females on all subscales (all ps < .OOOl), with the exception of the arousal scale. Experimental session. Similar analyses were conducted on the questionnaires administered at the experimental session, except that menstrual cycle phase (intermenstrual and premenstrual) was added as an additional factor to the ‘Univariate analyses of variances F values for each sukale of the retrospective MDQ are as follows: Pain, F(1, 83) = 32.04, p < .ooOl; Concentration. F’(l. 83) = 18.42, p < .ooOl; Behavioral Change, F(l, 83) = 28.20. p < .OOOl; Autonomic Reactions, F(1, 83) = 19.11, p < .OOOl; Water Retention, F(1, 83) = 16.21, p c .OOOl: Negative Affect, F(1, 83) = 31.75, p < .OW; Arousal, F(1,83)=3.17,n.r,andControl,F(1.83)=25.14.p~.0001.

ANXlEl’Y

-

401

SENSlllVlTV

TABLE 1 INITIAL scREENlNcMmsultls High AS

Anxiety Sensitivity Index No. Panic attacks in last year Distress of the attack No. relatives with panic STAI - trait anxiety STAI - state anxiety BDI Menstrual Distress Questionnaire Subscales Pain Concentration Behavioral Change Autonomic Reactions Water Retention Negative Affect Arousal Control

Low AS

M

SD

36.34, 5.95, 2.44,

3.%

O.%, 46.64, 44.61,

14.52,

20.37, 20.63, 17.05, 9.98, 13.77, 34.84 13.58, 11.84

5.96 1.09 0.99 12.44 14.36

M

SD

9.79 3.00,

3.50 1.87 0.76 0.63 13.42 11.33 4.77

1.71, 0.34 29.95, 28.38,

9.14

4.11

7.61

12.12, 12.17,

9.69 8.45 5.56 6.86 12.85 6.35 6.98

8.8% 5.61b 8.55, 20.4%

11.24, 6.3%

5.82 7.01 5.32 3.51 4.34 10.47 5.04 1.32

Nore. Row means having the same subscript are not significantly different at p c .OOl.

analyses. A MANOVA on the prospective MDQ, Willrs’s lambda, F(8.70) = 3.84, p < .OOl, and subsequent univariate analyses2 revealed that high-AS females scored significantly higher on all subscales than low-AS females (all p’s < .OOOl) with the exception of the arousal subscale. There was no significant effect for menstrual cycle phase on the prospective MDQ. An ANOVA on state anxiety scores revealed a group main effect, F( 1,80) = 14.08, p < .OOl, with high-AS females scoring higher than low-AS females. An ANOVA on the depression measure (BDI) yielded similar results in that high-AS females reported more symptoms than low AS females, F(1, 80) = 22.78, p < .OOOl. Menstrual cycle phase had no effect on measures of state anxiety and depression. Psychophysiological Measurement

A 2 (Groups: High and Low Anxiety Sensitivity) x 2 (Menstrual Cycle Phase: Premenstrual, Intermenstrual) ANOVA on baseline SCL revealed a 2Univariate analyses of variances F values for each subscale of the Imqective MDQ are as follows: Pain, F(1, 77) = 9.34, p < .Ol; Concentration, F(1. 77) = 18.95. p < .OOOl; Behavioral Change, F(1, 77) = 25.99, p < .OOOl; Autonomic Reactions, F(1. 77) = 5.43, p c .Ol; Water Retention, F(l.77) = 6.14,~ < .Ol; Negative Affect, F(l. 77) = 19.92.~ < .ooOl; Arousal, F(1 ,77) = 0.78, N, and Control, F(l.77) = 9.48. p < .Ol.

402

S. T. SIGMON

I3 AL.

TABLE 2 EWERIMWTAL SESSIONMEASURES High AS

Low AS

M

SD

M

SD

State anxiety BDI

43.27, 11.59,

12.88 8.29

31.61t, 3.9%

15.37 5.68

Menstrual Distress Questionnaire Subscales Pain Concentration Behavioral Change Autonomic Reactions Water Retention Negative Affect Arousal Control

15.95, 19.93, 15.90, 6.61, 9.32, 27.85, 17.24, 10.68,

6.88 9.55 7.95 3.47 5.16 13.07 7.12 5.55

11.61b 12.69b 8.74 5.02t, 6.82, 17.41, 16.07, 7.73,

6.54 6.96 4.08 2.27 3.41 10.01 5.14 2.64

Nore. Row means having the same subscript are not significantly different atp < .OOl.

significant main effect for AS group only, F(1, 77) = 5.23, p < .03. High-AS females (M = SO) exhibited higher baseline SCL compared to low AS participants (it4 = .13). There was no significant main effect for menstrual cycle phase or interaction of group and phase. This analysis was repeated controlling for initial state anxiety level and panic history. In this study, panic history was operationally defined as the number of panic attacks experienced in the past year. No significant differences emerged for baseline SCL when the effects of state anxiety level and panic history were removed. In order to compare psychophysiological responses across neutral and anxiety scenes, a 2 (Groups: High and Low Anxiety Sensitivity) x 2 (Menstrual Cycle Phase: Premenstrual, Intermenstrual) x 2 (Occasions: Neutral Scenes, Anxiety Scenes) repeated-measures ANOVA on SCR frequency was conducted using Greenhouse Geisser corrections for repeated measures. A significant three-way interaction revealed that high-AS females in the premenstrual phase exhibited more SCR to the anxiety scenes when compared to the females in the other three conditions, F(1.75) = 5.20,~ c .03. Figure 1 depicts this interaction. To determine whether SCR magnitude varied, a 2 (Groups) x 2 (Menstrual Cycle Phase) x 2 (Occasions) repeated-measures ANOVA was performed. Similar to the results for SCR frequency, there was a significant three-way interaction indicating that high-AS females in the premenstrual phase exhibited skin conductance responses greater in magnitude for the anxiety scenes compared to females in the other groups, F(1, 77) = 6.56, p < .Ol. These results are presented in Figure 2. In addition, high-AS females in the intermenstrual phase had significantly greater responses than low AS females in either the intermenstrual or premenstrual phases.

ANXIETY

403

SENStTIVlTY

4 3.5 3 2.5 2 1.5 1 0.5 0

Reme.nsttual Neutral

Intemenstmal Neutral

Premensmlal Anxiety

Intermenstrual Anxiety

Menstrual Cycle Phasex SceneType FIG. 1. SCR PREQ~~NCY.

n HighAS 11[111 LowAS I

Intermenstrual Premenstrual Intermenstrual Premenstrual Neutral Anxiety Anxiety Neutral Menstrual Cycle Phasex SceneType FIG. 2. LOGARITHMIC

TRANSFORMAllONS

OF

sc!lt

MAGNITUDE.

S. T. SIGMON

404

IR AL.

As mentioned previously, higher levels of state anxiety may influence psychophysiological responding. In addition, previous research demonstrated that panic history needs to be addressed in psychophysiological research with anxiety sensitivity participants (Borden & Lister, 1994). To examine the contribution of these factors to the current results, psychophysiological data were reanalyzed controlling for initial levels of state anxiety and panic history. The results for SCR frequency and SCR magnitude did not differ when initial levels of state anxiety and panic history were added as covariates in the subsequent analyses. Mood Measures

An ANOVA revealed that high-AS females reported more anxiety and depression than low-AS females on all three measurement occasions (after baseline, after the neutral scenes, after the anxiety scenes), F(1, 78) = 25.57, p < BOO1 and F(l, 78) = 10.48, p < .Ol, respectively. Menstrual cycle phase did not differentially affect the results. Means and standard deviations are presented in Table 3. Additional analyses were performed to examine if mood changed within groups across the three occasions and to control for initial levels of state anxiety. Results indicated that in both the high- and low-AS groups, females reported more anxiety, F(2, 162) = 4.53, p < .Ol, and depression, F(2, 162) = 6.53, p < .OOl, after listening to the anxiety scenes compared to baseline and neutral scenes. Mean anxiety and depression mood scores did not differ across baseline and neutral scene occasions. To assess changes in mood from the presentation of neutral to anxiety scenes, an analysis of covariance with state anxiety levels and neutral mood scores as covariates was conducted on the mood ratings after the presentation of the anxiety scenes. The results indicated that high-AS females reported TABLE MOOD

3

MEGWREY

Hi@ AS Occasion

Low AS

M

SD

M

SD

FOMS-AtlXiety Baseline After neutral scenes After anxiety scenes

8.75, 8.77, 14.Mt,

7.40 6.88 8.60

3.26b -b S.87t,

3.84 4.15 5.87

POMS-Depression Baseline Atier neutral scenes A!kr anxiety scenes

8.34 6.70, 12.27,

9.81 8.98 12.23

2.69t, 2.46b 3.33,

6.53 6.66 7.31

Note. Row means having the same subscript

are not significantly

different

at p c .oOl .

ANxImY SENSlllvlTY

405

more anxious and depressed mood following the anxiety scenes even when controlling for initial anxiety and mood levels, F( 1, 77) = 3.66, p < .05, F(1,77) = 11.62,~ < .OOl, respectively. Ratings of the Scenes

A 2 (Group) x 2 (Menstrual Cycle Phase) x 2 (Occasions: Neutral and Anxiety Scenes) ANOVA on the pleasantness ratings revealed a main effect for occasions only, F( 1,77) = 333.82, p < .OOOl. All participants rated the neutral and anxiety scenes similarly regardless of anxiety sensitivity grouping or menstrual cycle phase. The mean ratings for the neutral (M = 2.89) and anxiety scenes (M = 6.10) fell in the expected directions. Relationship Between Self-Repot?Measuresand Psychophysiological Measures

Because of tbe mixed results that have been obtained in previous research examining menstrual cycle reactivity, this study examined the relationship between self-report measures of anxiety and psychophysiological responding to the anxiety scenes. Frequency of SCR to the anxiety scenes was positively correlated with AS1 scores (r = .40, p < -001) and with levels of state anxiety (r = .34, p < .Ol). Interestingly, SCR magnitude was also positively correlated with the AS1 (r = .42, p < .OOOl) and with state anxiety (r = .42, p c .ooOl). In this study, both AS1 and state anxiety levels appeared to be comparably ptedictive of physiological responses to anxiety scenes. Similar to previous research (e.g., Shostak & Peterson, 1990), the AS1 was also positively correlated with trait anxiety (r = .61, p < .OOOl) and with state anxiety levels (r = .44, p < .OOOl). DISCUSSION In this study, the general pattern of results suggests that anxiety sensitivity and menstrual cycle timing are infhiential factors in observed psychophysiological response differences. Initially, baseline differences in SCL were found between low and high anxiety sensitivity females. Because of the positive correlation between initial levels of state anxiety and anxiety sensitivity, it is important to untangle the effects of state anxiety from the effects of anxiety sensitivity. After controlling for effects of initial levels of state anxiety and panic history, however, differences between the anxiety sensitivity groups at baseline were not significant. These results correspond to those of a previous study (Borden & Lister, 1994). which found that individuals high and low in anxiety sensitivity did not differ on baseline epidermal levels. Direct comparisons with this study, however, are difficult due to different psychophysiological recording systems. Although Shostak and Peterson (1990) did not find any baseline psychophysiological differences between anxiety sensitivity groups, it should be noted that they used EMG and blood pressure measures.

406

S.T. SIGMON

El-AL.

Arguably, the most important finding was that premenstrual females high in anxiety sensitivity exhibited greater psychophysiological reactions to anxiety stimuli as measured by SCR frequency and magnitude. These findings remained consistent even after controlling for the effects of initial state anxiety and for panic history. Previous research has not demonstrated psychophysiological differences in individuals varying in anxiety sensitivity levels (Borden & Lister, 1994; Shostak & Peterson, 1990). Neither of the two published studies examining anxiety sensitivity and physiological responding in males and females found significant effects for anxiety sensitivity. Several reasons could account for the discrepancies between past results and the findings in the present study. Both of the previous studies used a mentally challenging task to avoid any obvious demands for greater psychophysiological responding. It could be argued that the anxiety scenes utilized in the present study represented more of an obvious demand for individuals predisposed to respond in an anxious manner to daily anxiety-provoking events. Although the obviousness of the stimuli or tasks used may be debated, both types were chosen to elicit arousal. It may even be that mental tasks are intended to provoke even more arousal, perhaps due to performance anxieties or because of a social demand to please the experimenter. This study utilized anxiety stimuli to increase both the external and ecological validity of the study by examining responses to stimuli that bear resemblance to situations encountered in everyday experiences. Another possibility that accounts for the discrepancy between our results and those of previous studies pertains to the type of psychophysiological measurement used and how the data were managed. Skin conductance continues to be one of the most sensitive physiological measurements for assessing emotional arousal and has been widely used for this purpose. In the Borden and Lister (1994) study measuring epidermal activity, no mention was made of the type of electrolyte medium used nor of any of the standard data transformations on skin conductance recordings. Thus, methodological differences between the two studies may well account for the inconsistent findings. In skin conductance recordings, extreme responses can skew the data in either direction, either minimizing differences or maximizing them. In this study, standardized recording procedures and recommended statistical approaches to analyzing the data were followed (Venables & Christie, 1980). The finding that menstrual cycle phase also affects psychophysiological responding in anxiety sensitivity females has important implications. Previous research on anxiety sensitivity has not examined menstrual phase as a possible mediating variable in obtained results. This may be critical given that most studies investigating anxiety sensitivity use college populations with the majority of the sample being female. The question remains as to why high anxiety sensitivity females in the premenstrual phase had stronger reactions to the anxiety stimuli than females in the other three groups. A more complete explanation may reside in a blending of the bodily sensations and expectations hypotheses with anxiety sensitivity theory.

ANXIETY

SENSITIVITY

407

According to anxiety sensitivity hypotheses, high anxiety sensitivity females may be more likely to monitor or focus on internal stimuli. These individuals may constantly scan and assessfor any symptoms that may have potential negative consequences. According to the bodily sensations hypothesis, hormonal changes are thought to be responsible for many of the reported symptoms that occur premenstrually. Nonspecific arousal may be slightly higher at this time due to expectations regarding premenstrual symptoms (i.e., expectations hypothesis). Given these assumptions, differences in reactions to certain events (i.e., anxiety-provoking) as a function of anxiety sensitivity may be more likely. Consistent with this rationale, high anxiety sensitivity females had the greatest reactions premenstrually to anxiety scenes. High anxiety sensitivity females in the intermenstrual phase, however, had greater SCR magnitude to anxiety scenes when compared to low anxiety sensitivity females. Although the current results suggest that it is important to assess menstrual cycle phase, an assessment of anxiety sensitivity continues to be important in determining differential psychophysiological responses. Our results also indicate that self-reports of mood (i.e., anxiety and depression) changed significantly after the presentation of the anxiety scenes. Specifically, high anxiety sensitivity females reported more anxiety and depression following the presentation of anxiety scenes even after controlling for the effects of initial state anxiety and mood. Although it might be expected that the anxiety scenes would differentially impact mood depending on an individual’s level of anxiety sensitivity, it is important to note that all participants’ moods changed in accordance with the respective scenes. In addition, all participants rated the pleasantness of the scenes in the expected directions. These results contrast sharply with those of Borden and Lister (1994), who did not find that participants’ level of anxiety sensitivity distinguished reports of subjective distress during different phases of the experiment. Although these researchers controlled for initial levels of state anxiety as well, they used a general measure of subjective distress. Shostak and Peterson (1990) did not find pretask anxiety differences but did find baseline to posttask differences for high and low anxiety sensitivity participants. They did not, however, control for prior levels of state anxiety. Contrary to Borden and Lister’s findings (1994), panic history and prior anxiety levels did not affect self-reports of mood changes in this study. According to anxiety sensitivity hypotheses, self-reports of distress should occur as arousal increases because high anxiety sensitivity individuals focus more on internal stimuli. The present results are more in accord with tenets of the anxiety sensitivity construct. Reports of mood changes in the present study appeared to be more influenced by anxiety sensitivity rather than by menstrual cycle phase. Differences in retrospective (screening session) and prospective reports (experimental session) of menstrual symptoms varied only by level of anxiety sensitivity. Although previous research has found that actual menstrual cycle

408

S. T. SIGMON

FX AL.

phase is important when females complete menstrual symptom questionnaires (e.g., AuBuchon & Calhoun, 1985), anxiety sensitivity proved to be a more important variable than cycle phase in this study. For both retrospective and prospective accounts, high anxiety sensitivity females reported a more intense experience of symptoms, regardless of menstrual cycle phase. According to the anxiety sensitivity construct, anxiety sensitivity individuals are more aware of physiological changes and are more likely to monitor them. This preoccupation and attention to physiological sensations may account for the present fmdings. Future research needs to address the question of whether high anxiety sensitivity females are more likely to expect worse symptoms based on cultural stereotypes. According to the expectation hypothesis, one would have predicted an effect for menstrual cycle phase for the prospective reports of menstrual symp torns. Prospective reports of menstrual symptoms, however, are thought to be less influenced by expectations (Metcalf & Hudson, 1985). Although the expectations hypothesis may account for the high anxiety sensitivity females’ reports of menstrual symptoms retrospectively, they cannot completely account for the prospective differences obtained. According to tbe body sensations hypothesis, hormonal fluctuations are responsible for increased reports of premenstrual symptoms. It has also been proposed that physical symptoms are less susceptible to expectation or suggestion accounts than are affective symptoms (Metcalf & Hudson, 1985). Thus, high anxiety sensitivity females’ reports of more severe symptoms across the menstrual cycle may reflect an increased general self-focus. It should be noted that participants in the present study obviously knew which phase of the menstrual cycle they were in when they attended the experimental session. Although knowledge of the purposes of retrospective accounts of menstrual symptoms has been a problem, it has been less of one for prospective self-report accounts (Klebanov & Jemmott, 1992). Independent verifications of the cycle phase either through hormonal assays or daily temperature readings would be preferable. The present findings suggest that menstrual cycle timing may be an important variable to consider when conducting physiological research with anxiety sensitivity populations. Given that a large number of the individuals assessed in anxiety sensitivity studies are female, cycle timing may differentially affect the results. Another important implication of this research pertains to the nonclinical nature of our sample of students. The finding that females divided into high and low categories on a measure of anxiety sensitivity exhibited different physiological responding to anxiety stimuli may have implications for the understanding of the development of anxiety disorders in females. In addition, differences in psychophysiological responses to anxiety stimuli according to menstrual cycle timing may represent a plausible explanation for clinical accounts of increased anxiety at certain phases of the menstrual cycle. Although menstrual cycle timing had less of an impact on self-report measures

ANXETY

SENSITMTY

409

of mood and menstrual symptoms, this variable needs to be addressed in future research with anxiety sensitivity populations as well as with anxiety-disordered samples. REFERENCES American Psychiatric Association. (1980). Diagnostic and statistical manualofmentaldisonlers (3rd ed.). Washington, DC: Author. Aubuchon, P. G.. & Calhoun, K. S. (1985). Menstrual cycle symptomatology: The role of social expectancy and experimental demand characteristics. Psychosomatic Mea’icine. 47.35-45. Beck, A. T., & Steer, R.A. (1984). Internal consistencies of the original and revised Beck Depression Inventories. Journal of Clinical Psychology, 40.1365-l 367. Beck, A. T, Steer, R A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8,77-100. Borden, J. W.. & Lister, S. C. (1994). The anxiety sensitivity construct: Cognitive reactions to physiological change. Journal ofAnriety Disoniers, 8.31 l-321. Bmier. A., Charney, D. L., & Heninger, G. R (1986). Agoraphobia with panic attacks: Development, diagnostic stability, and course of illness. Archives of Gene& Psychiany, 43.1029-1036. Cameron, 0. G., Kuttesch, D., McPhee, K, & Curtis, G. C. (1988). Menstrual fluctuation in the symptoms of panic anxiety. Joumul of Anxiety Disorders, 15,169-174. Cook, B. L., Noyes, R., Garvey, M. J., Beach, V., Sobotka, J., & Chaudhry, D. (1990). Anxiety and the menstrual cycle in panic disorder. Journal of Afictive Disoniers, 19.221-226. Dow, R. (1991). Psylab (version 2.0) [Computer software]. London, England Contact Precision Instruments. Fowles, D. C.. Christie, M. J.. Edelberg, R., Grings, W. W.. Lykken, D. T., & Venables, P. H. (1981). Publication recommendations for electrodermal measurements. Psychophysiology, 18, 232-239.

Golub, S. (1976). The magnitude of premenstrual anxiety and depression. Psychosomatic Medicine, 38,4-12. Kaspi, S. P., Otto, M. W., Pollack, M. H., Eppinger. S., & Rosenbaum, J. E (1994). Premenstrual exacerbation of symptoms in women with panic disorder. Journal of Anxiety Disorders, 8, 131-138. Klebanov, I? K., & Jemmott, J. B. (1992). Effects of expectations and bodily sensations on selfreports of premenstrual symptoms. Psychology of Women Quurterly, 16.289-310. Mailer, R. G., & Reiss, S. (1992). Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety

Disotirs,

6.241-247.

McNair, D. M., Lorr, M., & Droppleman. L. F. (1971). El7’S hfanual for the Profir of Mood Stares. San Diego, CA: Educational and Industrial Testing Service. McNally, R. J., & Lorenx. M. (1987). Anxiety sensitivity in agoraphobics. Journal of Behavior Therapy

and Experimental

Psychiatry,

18.3-l

1.

Metcalf, M. G., & Hudson, S. M. (1985). The premenstrual syndrome: Selection of women for pretreatment trials. Journal of Psychosomatic Research, 29,63 l-638. Moos, R. H. (1969). The development of a Menstrual Distress Questionnaire. Psychosomatic Medicine,

30.853-867. Perimenstrual Distress Questionnaitz.

Moos, R. H. (1985).

symptoms:

A manual

and overview

of research

with the

Stanford, CA: Author. Norton, G. R.. Donvard. J., & Cox, B. J. (1986). Factors associated with panic attacks in noncliiical subjects. Behavior Therapy, 17.239-252. Peterson, R. A., L Heilbronner, R. L. (1987). The Anxiety Sensitivity Index: Construct validity and factor analytic structure. Jownal ofAnxiety Dbomkrs, 1, 117-121. Peterson, R. A., & Reiss, S. (1987). Test nwuuzl for the Anxiety Sensitiviv Index. Palos Heights, IL: International Diagnostic Systems. Menstrual

410

S. T. SIGMON

E.T AL.

Reiss, S. (1987). Theoretical perspectives on the fear of anxiety. Clinical

Psydwlogy

Review,

7,

585-S%.

Reiss, S., & McNally, R. J. (1985). The expectancy model of fear. In S. Reiss & R R. Bootzin (Bds.), Theoretical issues in behavior therapy. New York: Academic Press. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behuviour Research and Therapy, 24, l-8. Roth, W. T., Tetch, M. J.. Taylor, C. B.. Sachitano, J. A., Gallen, C. G., Kopell. M. L., McClenahan, K. L., Agras, W. S., & Pfefferbaum, A. (1986). Autonomic characteristics of agoraphobia with panic attacks. Biological Psychiatry, 21.1133-1154. Shostak, B. B., & Peterson, R. A. (1990). Effects of anxiety sensitivity on emotional response.to a stress task. Behaviour

Research

and Therapy,

28.513-521.

Spielbetger, C. D. (1988). State-Trait Anxiety Inventory. In M. Hetsen & A. S. Bellack (Eds.), Dictbnury of Behavioral Techniques (pp. 448-450). New York: Pergamon Ptess. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manuul for rhe State-Trait Andy Inventory. Palo Alto, CA: Consulting Psychologists Press. Steer, R. A., & Beck, A. T. (1988). Beck Depression Inventory. In M. Hersen & A. S. Bellack @Is.), Dictionary of behavioral techniques (pp. 44-46). New York Pergamon Press. Taylor, S., Koch, W. J., McNally, R. J., & Crockett, D. J. (1992). Conceptualizations of anxiety sensitivity. Psychological Assessment, 4,24%250. Veith, J. L., Anderson, J., Slade, S. A., Thompson, P.,Laugel, G. R., & Getzlab, S. (1984). Plasma Bendorphin, pain thresholds and anxiety levels across the human menstrual cycle. Physiological Behavior,

32.3 l-34.

Venables, P. H., & Christie, M. J. (1980). Electrodermal activity. In I. Martin and P. H. Venables (Eds.), Techniques in psychophysiology (pp. 31-67). New York: John Wiley and Sons.