Introduction
The DSM-IV emphasizes both unexpected panic attacks and worry about future panic attacks and their consequences as essential features of panic disorder (APA, 2000). Panic attacks are understood as short-term defensive reactions that occur in response to imminent or perceived danger; the anxious response is characterized by a cascade of body symptoms, such as palpitations, shortness of breath, nausea or dizziness, as well as fearful cognitions including fear of dying, fainting or loss of control. Besides the occurrence of actual panic attacks, patients with panic disorder usually experience a continuous worry about future panic attacks. This anticipatory anxiety is typically associated with a pertinent hypervigilance in regard to sensations linked to panic attacks (e.g. Schmidt, Lerew and Trakowski, Reference Schmidt, Lerew and Jackson1997). In a more general sense, anticipatory anxiety has been characterized as a negative affect state that is accompanied by both somatic symptoms and appraisals of one's abilities to cope with potentially negative events (Barlow, Reference Barlow2000). Cognitive formulations of panic disorder emphasize anticipatory anxiety as a consequence of the apprehension of events (such as panic attacks) as both highly aversive and uncontrollable, leading to attempts to avoid those events or prevent their anticipated negative consequences (e.g. Salkovskis, Clark and Gelder, Reference Salkovskis, Clark and Gelder1996; White, Brown, Somers and Barlow, Reference White, Brown, Somers and Barlow2006).
Patients often experience anticipatory anxiety as general distress or worry that can remain incapacitating even when spontaneous attacks have largely remitted (Katon, Vitaliano, Anderson, Jones and Russo, Reference Katon, Vitaliano, Anderson, Jones and Russo1987; Lang, Reference Lang2008), and it may therefore account for an important portion of the impairment associated with panic disorder. Several studies have shown that high levels of anticipatory anxiety are more strongly associated with avoidance behaviour than the actual frequency of panic attacks (e.g. Craske, Rapee and Barlow, Reference Craske, Rapee and Barlow1988; Rachman and Lopatka, Reference Rachman and Lopatka1986); thus anticipatory anxiety seems to play an important role in determining severity and impairment in panic disorder.
Despite the obvious relevance for understanding panic disorder, anticipatory anxiety in panic disorder has rarely been addressed in research. However, there are several theoretical assumptions in regard to panic disorder that might inform about the natural occurrence of anticipatory anxiety. Cognitive conceptualizations, for instance, emphasize the mutual relationship between panic attacks and anticipatory anxiety: panic attacks experienced in the past contribute to the anticipation of future panic, leading to an increase in self-monitoring, and the oversensitive perception of critical body symptoms that in turn - at least when misinterpreted - provoke panic. Anticipatory anxiety is thus seen both as a consequence of experiencing panic attacks as well as one important antecedent of further panic attacks (Clark, Reference Clark1986). Recent bio-psycho-social models of panic disorder suggest that panic attacks and anticipatory anxiety represent distinct functional systems that overlap but that can also be differentiated at a functional and a phenomenological level (Bouton, Mineka and Barlow, Reference Bouton, Mineka and Barlow2001). It is assumed that anticipatory anxiety might, similar to panic attacks, be strongly influenced by conditioning processes, and that anticipatory anxiety is elicited by similar external and internal cues that are associated with previous panic episodes (Bouton et al., Reference Bouton, Mineka and Barlow2001). Thus, anticipatory anxiety might fluctuate in response to the presence of cues that are also assumed to trigger panic, such as body symptoms, or contextual variables.
Although there is laboratory and experimental evidence for these assumptions, there is a paucity of studies examining relevant processes in real-life settings. In addition, the few empirical studies addressing anticipatory anxiety most often solely focused on its relation to panic attacks. Cross-sectional studies surprisingly could not find any significant relations between panic attacks and anticipatory anxiety (Adler, Craske, Kirshenbaum and Barlow, Reference Adler, Craske, Kirshenbaum and Barlow1989; Basoglu et al., Reference Basoglu, Marks, Kilic, Swinson, Noshirvani and Kuch1994). In contrast, Craske, Glover and DeCola (Reference Craske, Glover and DeCola1995), using data from a 2-week self-monitoring period, found higher rates of anticipatory anxiety and worry after uncued panic attacks. Another study using ambulatory assessment found anticipatory anxiety to be the only predictor of panic attacks, supporting its importance in panic onset (Kenardy, Fried, Kraemer and Taylor, Reference Kenardy, Fried, Kraemer and Taylor1992). In line with that finding, the majority of patients in a retrospective study reported that panic attacks were preceded by heightened anxiety levels. Longer and more intense episodes of anxiety were associated with more severe panic attacks (Basoglu, Marks and Sengün, Reference Basoglu, Marks and Sengün1992). Thus, anticipatory anxiety was considered to augment panic reactions. Evidence for an independent role of anticipatory anxiety in maintenance of panic disorder stems from Rodebaugh, Curran and Chambless (Reference Rodebaugh, Curran and Chambless2002), who also used an ambulatory approach to monitor anticipatory anxiety over a 10-day period. They found daily anxiety ratings to be influenced by panic expectancy in the morning, and anxiety levels from the previous day. Anticipatory anxiety was best predicted by a trait-like expectancy factor but not by general distress the day before. The authors suggested anticipatory anxiety to represent a stable trait that maintains anxiety in panic patients irrespective of actual experiences of panic.
Taken together, these findings do not provide a clear picture about the natural relationship between anticipatory anxiety and panic attacks. These mixed findings might result from neglecting potential interacting variables, such as cognitive appraisals of one's ability to cope, or individual levels of anxiety sensitivity, a more general disposition to fear anxious responses that is often regarded as a risk factor in the aetiology of panic attacks and panic disorder (Schmidt, Lerew and Jackson, Reference Schmidt, Lerew and Jackson1997, Reference Schmidt, Lerew and Jackson1999). Laboratory findings suggested that anxiety responses to biological challenges in clinical and non-clinical subjects are mediated by threat appraisals and perceived control (Sanderson, Rapee and Barlow, Reference Sanderson, Rapee and Barlow1989; Telch, Silverman and Schmidt, Reference Telch, Silverman and Schmidt1996; Telch et al., Reference Telch, Smits, Brown, Dement, Powers, Lee and Pai2010; Zvolensky, Eifert, Lejuez and McNeil, Reference Zvolensky, Eifert, Lejuez and McNeil1999). Perceived ability to cope might therefore reduce negative effects of panic. Although these findings are highly valuable for explaining variance in anxious responding, they are restricted to laboratory settings. Cognitive mediation of negative effects of panic attacks on anticipatory anxiety has not yet been addressed in a real life setting.
The present study aims to extend the existing knowledge in two directions: i) by providing a description of the natural fluctuations in anticipatory anxiety in everyday experience; and ii) by examining theoretically derived hypotheses about the correlates of anticipatory anxiety, and its relation to panic attacks. Based on assumptions of Bouton et al. (Reference Bouton, Mineka and Barlow2001) it is assumed that i) external and internal cues, typically assumed to evoke panic, such as being in agoraphobic situations, experiencing arousal due to bodily symptoms or negative affect, as well as consumption of psychotropic substances, are significantly associated with increased ratings of anticipatory anxiety. In line with the cognitive model of panic attacks, it is further assumed that ii) higher levels of anticipatory anxiety precede panic attacks, and that experiencing panic attacks in turn will increase fear of future panic.
In addition, two further research questions address potential interacting variables: anxiety sensitivity, and perceived ability to cope with panic. It is assumed that iii) higher levels of anxiety sensitivity will increase the relationship between panic attacks and anticipatory anxiety, whereas iv) higher levels of perceived ability to cope will reduce the impact of panic attacks on anticipatory anxiety (Telch et al., Reference Telch, Silverman and Schmidt1996, Reference Telch, Smits, Brown, Dement, Powers, Lee and Pai2010).
In order to test these hypotheses in a naturalistic setting, a computerized Ecological Momentary Assessment (EMA) approach was used. EMA has become increasingly popular in research on psychopathology as it overcomes some of the limitations of traditional assessment methods. By repeated within-day assessments in the participants’ natural environment, it allows for modelling within-person processes as well as interpersonal differences over time, and in their relation to covariates. Real-time monitoring is thought to reduce retrospective recall biases, and thus to provide ecologically valid data (Piasecki, Hufford, Solhan and Trull, Reference Piasecki, Hufford, Solhan and Trull2007; Stone and Shiffman, Reference Stone and Shiffman1994). Most studies with clinical populations demonstrate good to excellent compliance to EMA procedures, and there is little evidence for reactivity effects (e.g. Johnson et al., Reference Johnson, Grondin, Barrault, Faytout, Helbig and Husky2009). Studies comparing EMA with retrospective questionnaires suggest better accuracy of EMA data, especially with regard to estimates of symptom frequency or duration (e.g. Samo, Tucker and Vuchinich, Reference Samo, Tucker and Vuchinich1989; Shiffman et al., Reference Shiffman, Engberg, Paty, Perz, Gnys, Kassel and Hickcox1997; Stone, Broderick, Shiffman and Schwartz, Reference Stone, Broderick, Shiffman and Schwartz2004). However, EMA still relies on self-report, and it is therefore not free of response biases. To this date, EMA was most often applied in severe psychiatric disordered populations, but its feasibility in anxiety samples has also been demonstrated (Helbig, Lang, Swendsen, Hoyer and Wittchen, Reference Helbig, Lang, Swendsen, Hoyer and Wittchen2009). Previous ambulatory assessment studies on panic attacks have yielded relevant findings in regard to occurrence and phenomenology of panic attacks (see Alpers, Reference Alpers2009). In extension to prior studies, the current study assessed anticipatory anxiety, panic attacks, and perceived ability to cope with panic simultaneously.
Material and method
Participants and study procedures
Data were collected from a clinical trial on the effects of therapy preparation during a CBT wait list that was conducted at an outpatient psychotherapy treatment centre (Helbig et al., Reference Helbig, Lang, Swendsen, Hoyer and Wittchen2009). Consecutive patients with panic disorder and agoraphobia were asked to participate in this study. Diagnoses were established using a computerized standardized clinical interview based on the Composite International Diagnostic Interview (DIA-X/CIDI; Wittchen and Pfister, Reference Wittchen and Pfister1997). All interviews were conducted by licensed assessors, who had received a 2-day training workshop. Diagnoses obtained by the DIA-X were validated by an experienced clinician in an additional diagnostic session (Hoyer, Ruhl, Scholz and Wittchen, Reference Hoyer, Ruhl., Scholz and Wittchen2006). Given their written informed consent, patients were asked to complete a one-week EMA period at therapy start. Design and procedures of the study are described elsewhere in more detail (Helbig et al., Reference Helbig, Lang, Swendsen, Hoyer and Wittchen2009).
Of 30 patients who were asked to participate, 25 gave their informed consent. Reasons for non-agreement included employment status (unable to react to random signals, n = 3) and unwillingness to monitor symptoms (n = 2). After the wait list period, 3 patients did not commence treatment. One patient withdrew the informed consent, yielding a total sample of 21 participants.
All participants were Caucasian. Average age was 33.8 (SD = 12.2), with a range of 18–60 years; 76% were females. The majority of participants (66%) were currently not in employment due to academic studies, maternal leave, or unemployment. Almost half of patients (43%) had at least one comorbid diagnosis; 33% (n = 7) were diagnosed with a depressive disorder and three patients each reported additional anxiety disorders or somatoform disorders. About one-third of participants (n = 8) were treated with antidepressant medication. Ten patients (48%) reported an intermittent use of prescribed anxiolytics.
EMA procedures and measures
At their first therapy appointment, participants received a PSION Revo handheld to capture everyday experiences, in addition to a 30-minute instruction and training for using the handheld computer. Participants were advised to respond to every EMA signal. Assessment comprised a one-week EMA period with five random signals per day. Signals were scheduled between 8 a.m. and 10 p.m., with a minimum interval of 2 hours between the signals. Starting day and time varied between the participants. Data sets from the days when the device was given out or returned were excluded from analyses, yielding an equal number of 30 data entries per participant. Of 630 possible EMA records, a total of 549 question sets were completed (total response rate: 87.1%). Individual response rates ranged from 36.6% to 100%, whereas the majority of participants (71.4%) answered at least 90% of all possible question sets, indicating good compliance with the assessment protocol. After the assessment period, all participants except one evaluated self-monitoring procedures as neutral or positive in regard to their general well-being.
The EMA questionnaire comprised 45 fixed questions and – depending on patients’ answers – 20 supplementary questions. Items were mostly taken from previous EMA studies (see Johnson et al., Reference Johnson, Grondin, Barrault, Faytout, Helbig and Husky2009) and comprised ratings, multiple choice questions, and free-response options. Similar EMA procedures have demonstrated not only their clinical feasibility, including little indication of reactivity, but also validity in assessing psychiatric patients (e.g. Johnson, Barrault, Nadeau and Swendsen, Reference Johnson, Barrault, Nadeau and Swendsen2009). Variables that were used in the present analyses are described below.
Anticipatory anxiety. Anticipatory anxiety was assessed with one item. Participants were asked to rate their current anxiety in regard to a future panic attack on a 6-point Likert scale from “0 – not at all” to “5 – very much”. This formulation was chosen in order to ensure panic specificity of anxiety ratings.
Perceived ability to cope. Panic-related self-efficacy was operationalized as confidence in being able to cope with future panic attacks. The item had to be evaluated using a rating on a 6-point Likert scale from “0 – not at all” to “5 – absolutely convinced”.
Body symptoms and panic attacks. The questionnaire included nine body symptoms taken from DSM-IV symptom list for panic disorder (e.g. palpitations, shortness of breath, dizziness) that should be rated in regard to their current intensity (Likert scale: “0 – not at all” to “5 – extremely intense”). For analyses, intensity ratings for all body symptoms were summed, yielding an index of overall symptom severity at one assessment point. Reliability of the derived scale was good (Cronbach alpha: .81). There was one additional question enquiring about panic attacks in the previous assessment interval (yes/no).
Panic-related cognitions. Panic related cognitions were assessed with two questions. The first question asked for idiosyncratic panic beliefs in regard to a potential aversive outcome of panic attacks. Five typical panic-related beliefs (fainting, having a heart attack, going insane, losing control, dying), and an open ended answering option were provided. The second question asked for a momentary estimation of the perceived likelihood that the aforementioned outcome would happen in case of panic. This variable was labelled as “catastrophic cognitions”.
Situational variables. Situational variables included questions regarding concurrent situation (“Where are you right now?”), and company (“Who is with you?”). The first question was assessed by a categorical response option, including an open ended category. In regard to the second question, several suggestions were made that had to be answered dichotomously (yes/no). Records were post hoc collapsed into situational categories “being in a public place”, “being at home or at a family member's home”, “being at work or at school” as well as “being alone”.
Negative affect. Affect was measured with a series of mood adjectives derived from a circumplex affect model (Tellegen, Reference Tellegen, Tuma and Master1985). Negative affect was calculated as a sum score of five associated items: nervous, bored, sad, angry and fearful. Cronbach alpha for this score was satisfactory (α = .60).
Substance consumption. Substance consumption (tobacco, alcohol, anxiolytics) during the previous assessment interval was assessed by dichotomous questions (consumption yes/no).
Anxiety sensitivity. Anxiety sensitivity was assessed with the revised Anxiety Sensitivity Index (ASI-R; Taylor and Cox, Reference Taylor and Cox1998). The ASI-R comprises 36 items assessing fearful responses when experiencing anxiety symptoms. The questionnaire has demonstrated excellent reliability (Cronbach α > .90; Deacon, Abramowitz, Woods and Tolin, Reference Deacon, Abramowitz, Woods and Tolin2003), and validity. Mean score in the current sample was 57.8 (SD = 21.0).
Statistical analyses
Data clearing, descriptive analyses and correlational analyses were carried out with the statistical package PASW Statistics 18.0.2. Two data files were prepared: A level-1 file containing available EMA data for each participant, and a level-2 file including baseline information about the participants, as well as questionnaire data (ASI-R). Covariation of anticipatory anxiety with categorical variables was analyzed with a two-level hierarchical model using HLM 6.04 for Windows (Raudenbush, Bryk and Congdon, Reference Raudenbush, Bryk and Congdon2000). Five hundred and forty-nine observations were made for the 21 participants. Weekday and daytime of assessment turned out to be related to anticipatory anxiety (see Results section), and were thus - besides ratings of anticipatory anxiety - considered as coefficients at level 1. All coefficients at level 1 were entered grand-mean-centred in the model. Further variables of interest at level 1, such as situational features or the occurrence of panic attacks, were treated as between-subjects factor. At level 2, sex and age were considered as coefficients; however, preliminary analyses indicated no significant influence of these variables on the associations between anticipatory anxiety and other variables. Thus, results are reported neglecting these level-2 coefficients. Examining the impact of anxiety sensitivity (hypothesis 3), all analyses were repeated with the baseline ASI-R score as a coefficient at level 2. Detailed description of analyses for specific research questions are described in the Results section. Single missing values were excluded while computing the models.
Effect sizes for contrast scores were calculated based on correlations between dependent and independent variables (r). Effect sizes r > .10 are considered as small effect; r > .25 is considered as medium, and r > .45 is considered as large effect (Rosenthal, Rosnow and Rubin, Reference Rosenthal, Rosnow and Rubin2000).
Results
Natural course and diurnal patterns of anticipatory anxiety
Ratings of anticipatory anxiety substantially varied within and between persons. Average ratings ranged from 0 up to M = 3.3, with a total mean score of 1.3 (SD = 1.5) over all subjects and measurements. About one-third of participants (61.9%) reported substantial variation of anticipatory anxiety with at least four different anxiety levels over time. There were no gender or age differences in means or within-subject variance, and no differences in regard to antidepressant medication (yes/no).
Multi-level modelling was used to examine diurnal changes in anticipatory anxiety. There were significant differences in average ratings between weekdays and weekends, with higher rates during the week (weekdays: M = 1.4, SD = 1.5 vs. weekends: M = 1.1, SD = 1.4; t = −2.17; p = .030). There were also significant differences in regard to time of day: participants reported higher levels of anxiety in the mornings (M = 1.4; SD = 1.4) compared to afternoon and evening ratings (M = 1.2, resp. M = 1.2; t = 2.13; p = .033).
Anticipatory anxiety and its relation to external and internal cues
In order to identify potential cues linked to anticipatory anxiety, we examined anticipatory anxiety levels in relation to internal and external variables that are known to be associated with panic experiences. For dichotomous variables, such as being in agoraphobic situations, and consumption of psychotropic substances, the primary method of analysis was contrasting the mean of anticipatory anxiety in assessment occasions that featured the interesting variable compared to occasions in which this was not the case. As day- and week-time were found to be associated with anticipatory anxiety, these variables were included as coefficients. Table 1 summarizes descriptive information and results of statistical comparisons.
Table 1. Anticipatory anxiety in relation to situational and internal variables
Notes: 1Assessment occasions featuring a given variable value; 2Effect size r *p<.01; **p<.001
Anticipatory anxiety was found to vary significantly in relation to certain situations. Being in a public place was associated with higher mean ratings of anxiety, whereas being alone (vs. accompanied) was not. Having experienced a panic attack in the previous assessment interval doubled ratings of anticipatory anxiety, indicating a strong relationship between panic attacks and subsequent increases in anxiety. Although we found lower rates of anticipatory anxiety after consumption of any psychotropic substance (especially alcohol and anxiolytics); these differences lacked statistical significance, possibly due to the rather low consumption rates.
Anticipatory anxiety was also associated with idiosyncratic panic-related cognitions. Fear of going insane due to panic was significantly linked to higher anticipatory anxiety ratings, compared to any other panic-related belief (see Table 1). Lowest ratings of anticipatory anxiety were found with regard to fear of losing control (M = 0.6; SD = 1.3). We also found a clear relationship between anticipatory anxiety and the perceived likelihood that the feared outcome would occur in case of panic, indicated by a significant rank coefficient (τ = .602; p < .001), We calculated the Kendall rank coefficient (tau) as data were not normally distributed. Significant, despite lower associations, were also found between anticipatory anxiety and negative affect (τ = .440; p < .001) as well as number of bodily symptoms experienced (τ = .399; p < .001). These associations remained stable when assessment time was partialled out.
Relationship between anticipatory anxiety and actual panic attacks
In the next step, we examined relations between anticipatory anxiety and panic attacks. During the sampling period, 13 participants (61.9%) reported a total of 64 panic attacks. Number of panic attacks experienced ranged from 1 to 9. In contrast, eight patients reported no panic attack during the sampling period. We found no sex differences with regards to number of panic attacks experienced, and no association to age. Patients with antidepressant medication tended to report fewer panic attacks than patients without medication; however, this difference failed statistical significance (M = 3.5 vs. 2.3; t(19) = 0.95; p = .352). Inspecting the temporal distribution of panic attacks, they were found to occur less frequently in the afternoons compared to morning and evening assessments (OR = 0.41; CI [0.18; 0.95]. Panic frequency did not differ between morning and evening occasions, and also did not vary with weekday. For subsequent analyses, only patients reporting panic attacks were taken into account. Two hypotheses were tested: 1) Heightened levels of anticipatory anxiety precede panic attacks; and 2) experiencing panic attacks increases subsequent ratings of anticipatory anxiety.
In order to analyze these relations, we categorized assessment times according to reports of panic attacks into: a) occasions at which a preceding panic attack was reported (occasion after panic attack); b) occasions prior to a panic report (occasion before panic attack); and c) occasions unrelated to panic reports. Using this categorical variable, we first analyzed differences in anticipatory anxiety comparing occasions before and after panic attacks to occasions unrelated to panic attacks. Results are depicted in Figure 1 (columns for “total sample”).
Figure 1. Levels of anticipatory anxiety in relation to panic experiences, and perceived ability to cope with panic (only patients reporting panic attacks; n = 317 assessment occasions)
As predicted, anticipatory anxiety significantly increased after experiencing panic attacks (t(317) = 4.99; p < .001; r = .30); however, we could not find elevated levels of anticipatory anxiety preceding panic attacks (t(317) = 0.31; p = .759, r = .03).
Impact of dispositional levels of anxiety sensitivity
Anxiety sensitivity is likely to increase the vulnerability for panic attacks, and anxious responding to bodily symptoms. We thus assumed that higher levels of anxiety sensitivity would increase the impact of panic attacks on anticipatory anxiety. We therefore re-examined differences in anticipatory anxiety after panic attacks compared to occasions unrelated to panic attacks, using the ASI-R scores as coefficient at level 2.
Within the HLM model, baseline anxiety sensitivity failed to significantly contribute to anticipatory anxiety ratings, when comparing occasions unrelated to panic attacks, and occasions preceding panic attacks. The difference between occasions after experiencing panic attacks and occasions unrelated to panic attacks remained stable, when controlling for anxiety sensitivity (t(194) = 2.49; p = .015). There was also no significant relationship between subjects’ mean rating of anticipatory anxiety, and baseline level of anxiety sensitivity (τ = −.289; p = .098).
Impact of perceived ability to cope
Self-efficacy is often thought to moderate emotional and behavioural consequences of adverse experiences. We therefore tried to elucidate the impact of perceived ability to cope with panic on the association between anticipatory anxiety and panic attacks. Mean ratings of the perceived ability to cope with panic did not differ in regard to gender or medication status, and were not related to age. There was a rather small but significant negative association between overall ratings of anticipatory anxiety and perceived ability to cope with panic attacks (τ = −.227; p < .001). When inspecting only patients reporting panic attacks, this association was somewhat stronger. Experience of panic attacks did not influence subsequent ratings of perceived ability to cope; thus, panic-related self-efficacy was not generally affected by panic attacks. However, it clearly showed that high levels of panic-related self-efficacy were associated with lower ratings of anticipatory anxiety to any assessment occasion (see Figure 1).
In order to analyze whether panic-related self-efficacy attenuated the impact of panic attacks on anticipatory anxiety, we compared ratings of anticipatory anxiety after the experience of panic attacks for high vs. low perceived ability to cope (median split). Low perceived ability to cope was significantly related to higher ratings of anticipatory anxiety after panic attacks when compared to high levels of panic-related coping abilities (t(62) = −2.06, p = .044). Reports of low perceived ability to cope were also associated with higher ratings of negative affect (t(60) = −6.04, p < .001), and panic-related cognitions (t(61) = −2.12, p = .038) after panic attacks. There was no such impact of panic-related self-efficacy on reported number of bodily symptoms experienced after panic. Frequency of alcohol or anxiolytics consumption after panic attacks did not differ in regard to self-efficacy.
Discussion
The present investigation aimed at exploring the natural course and correlates of anticipatory anxiety in patients with panic disorder. Further, the impact of anxiety sensitivity and perceived ability to cope with panic attacks on the relationship between anticipatory anxiety and panic attacks was scrutinized. Based on theoretical conceptions, it was assumed that cues, typically linked to panic attacks, would also be associated with anticipatory anxiety. It was further hypothesized that panic attacks and anticipatory anxiety would mutually interact, and that this relationship would systematically depend on baseline levels of anxiety sensitivity, and the perceived control over panic. In the following paragraphs, we first review descriptive findings on the natural course of anticipatory anxiety, before addressing results related to our hypotheses.
Descriptive analyses of temporal patterns of panic attacks and anticipatory anxiety over the course of a week revealed that panic attacks were similarly probable on work days and weekends. In contrast, anticipatory anxiety was significantly higher on weekdays (as opposed to weekends). This finding may indicate that both symptom manifestations may be differently related to workload, stress, daily hassles, and leisure time. In regard to time of day, panic attacks were less likely to occur in the afternoon as opposed to other daytimes; there were no differences between morning and evening occasions in panic frequency. Prior studies have already suggested daytime-related differences in panic frequency. Kenardy et al. (Reference Kenardy, Fried, Kraemer and Taylor1992) found panic attacks to be less frequent in mornings and evenings, whereas Margraf, Taylor, Ehlers, Roth and Agras (Reference Margraf, Taylor, Ehlers, Roth and Agras1987) found spontaneous attacks to be evenly distributed throughout daytime whereas situational attacks were less frequent in evenings.
In contrast, ratings of anticipatory anxiety were highest in the morning and significantly decreased over the day. This result of morning peaks of anxiety corresponds to data indicating that persons with persistent anxiety problems have higher morning cortisol levels and a higher cortisol awakening response compared to controls (Greaves-Lord et al., Reference Greaves-Lord, Ferdinand, Oldehinkel, Sondeijker, Ormel and Verhulst2007). Both cortisol parameters might indicate a stronger physiological response to the process of awakening that, in turn, might be perceived as anxious apprehension. Thus, anticipatory anxiety might partly reflect biological changes in HPA-axis reactivity or base rate.
The different temporal patterns of anticipatory anxiety and panic attacks might also reflect the proposed differences between fear and anxiety reactions. Fear and panic are thought to represent an immediate threat reaction, whereas anxiety is rather future-orientated and inhibiting (e.g. Barlow, Reference Barlow and Barlow2004; Lang, Bradley and Cuthbert, Reference Lang, Bradley and Cuthbert1998). Previous findings from Rodebaugh et al. (Reference Rodebaugh, Curran and Chambless2002) already supported this distinction by suggesting that panic expectancy might be influenced by a trait-like component, which is rather independent from actual experiences of panic attacks.
However, modern learning theories propose that both panic attacks and anticipatory anxiety might emerge from similar conditioning processes (Bouton et al., Reference Bouton, Mineka and Barlow2001). It was assumed that anticipatory anxiety should be related to internal and external stimuli that are known to provoke panic. Actually, our findings largely support this assumption. Ratings of anticipatory anxiety increased in situations that might be regarded as agoraphobic, such as being in public places as opposed to being in private. Interestingly, we found no association between anxiety levels and being alone vs. being accompanied; however, this might reflect differential tendencies of panic patients to feel reassured or distressed by company. In correspondence with theoretical formulations by Barlow (Reference Barlow2000), higher ratings of anticipatory anxiety were also largely associated with negative affect, catastrophic cognitions, and, to a somewhat smaller degree, to perceptions of body symptoms. Interestingly, individual panic-related beliefs also related differently to anticipatory anxiety, with fear of going insane being associated with highest, fear of losing control with lowest anxiety ratings. This confirms the importance of a cognitive appraisal-component in anticipatory anxiety. There was some evidence that preceding consumption of psychotropic substances might lower anticipatory anxiety, although it remains unclear whether this finding reflects attributional or biological effects of substance use. As anticipatory anxiety also persisted in absence of panic attacks, it has to be assumed that anticipatory anxiety substantially contributes to impairment, and thus needs to be targeted and reduced in treatment in order to achieve stable remission in patients with panic disorder.
In regard to relationships between panic attacks and anticipatory anxiety, we assumed a mutual relationship, although previous studies have yielded rather heterogeneous results. Findings disconfirm our hypothesis of a bi-directional relationship as they show drastically increased rates of anticipatory anxiety after panic attacks, but no heightened levels of anticipatory anxiety preceding panic attacks (see also Craske et al., Reference Craske, Glover and DeCola1995). Obviously, anticipatory anxiety results from - or is at least strengthened by - experiencing panic attacks (and not vice versa). This association is meaningful from a learning theory perspective as panic attacks represent a potent internal cue for the activation of memory and emotional networks associated with anxiety; it also is meaningful from a cognitive perspective, as panic attacks constitute an event that often elicits catastrophic misinterpretation in patients with panic disorder.
But why does higher anticipatory anxiety not prompt further panic attacks? Anticipatory anxiety might make catastrophic explanations of internal cues more accessible, but the present results suggest that anticipatory anxiety may also have other functions. It might, for instance, lead to avoidance or safety-seeking behaviours, which - at least in the short run -could render panic attacks less probable. This assumption of an anticipatory anxiety – behaviour link should be addressed in future studies.
As the present study examined hypotheses derived from laboratory studies on cognitive mediation of anxious responding, there is further evidence for the importance of cognitive appraisals in explaining individual reactions to panic. It was assumed that perceived ability to cope with panic as an indicator of self-efficacy would impact the relationship between anticipatory anxiety and panic attacks. Our data suggest that perceived ability to cope with panic is largely unaffected by the experience of panic attacks, but it is clearly related to anticipatory anxiety, suggesting panic-related self-efficacy to be a rather stable variable that might partially overlap with anxious apprehension. As expected, lower rates of self-efficacy were associated with intensified negative emotional and cognitive consequences of panic attacks. Thus, patients with lower perceived coping abilities might process experiences of panic attacks to a larger degree as catastrophic. It might be assumed that these patients are more likely to develop secondary symptoms such as agoraphobic avoidance (see also Telch, Brouilard, Telch, Agras and Taylor, Reference Telch, Brouilard, Telch, Agras and Taylor1989), or depression. In terms of clinical implications, low self-efficacy might indicate the need for cognitive interventions before exposing patients to potentially panic-triggering cues. Similarly, recent analyses of extinction learning research stress the importance of the experience of control in the face of anxiety (Craske et al., Reference Craske, Kircanski, Zelikowsky, Mystkowski, Chowdhury and Baker2008). Diagnostic assessment should therefore include measures on perceived control or self-efficacy, such as the anxiety control questionnaire, that was specifically developed for anxiety disorders (Rapee, Craske, Brown and Barlow, Reference Rapee, Craske, Brown and Barlow1996).
Surprisingly, we found no impact of dispositional anxiety sensitivity, neither on anticipatory anxiety as a panic consequence, nor on anticipatory anxiety levels in general. It might be concluded that anxiety sensitivity rather explains the onset than the course of panic disorder, further strengthening the importance of the cognitive mediation hypothesis of panic- and anxiety reactions in the maintenance of panic disorder.
Our results should be interpreted cautiously due to methodological study limitations. First, sample size was small, indicating insufficient power to calculated robust estimates within the multilevel model (Hox and Maas, Reference Hox and Maas2001), a fact that was in part compensated for by the multitude of assessment points per individual. Nevertheless, variance in some situational variables seemed rather low, perhaps reflecting high levels of agoraphobic avoidance. This sample bias might have prevented stronger effects for these variables. Anticipatory anxiety was assessed with one global item only; further studies should also include items regarding panic expectancy (likelihood of future panic attacks) as a second component of panic-related anticipatory anxiety. To our knowledge, there is no study assessing both aspects, and thus the extent to which the different aspects represent two distinct facets within the individual experience of anticipatory anxiety. There are also some limitations in regard to using signal contingent EMA for examining anticipatory anxiety and panic attacks. We cannot exclude potential effects of self-monitoring on perceptions of anxiety levels, although in most cases reactivity effects in EMA are found to be rather low (Johnson et al., Reference Johnson, Grondin, Barrault, Faytout, Helbig and Husky2009). Using an event-contingent sampling strategy might have further advanced valid assessment of panic attack consequences; however, previous studies have suggested that patients do not always comply with event-sampling strategies in panic disorder (Margraf et al., Reference Margraf, Taylor, Ehlers, Roth and Agras1987). In addition, there might be further variables of interest that moderate the examined relationships, such as trait anxiety, depressive symptoms or general tendencies to worry. Those variables should be included in further analyses.
Taken together, findings support assumptions of anticipatory anxiety and panic attacks as inter-correlated but also distinct features of panic disorder. Internal and external cues known to be associated with panic attacks are also linked to anticipatory anxiety. Results also show that consequences of panic attacks in the natural environment were related to cognitive appraisals, indicating perceived ability to cope with panic as a potentially relevant variable in the aetiology and maintenance of panic disorder. This line of research seems promising for explaining intra- and inter-individual differences in responses to anxiety provoking events, such as the development of panic disorder after experiencing panic attacks. In this regard, it might be promising to further scrutinize which variables determine high or low levels of perceived ability to cope with panic attacks.
Despite the given limitations, the study demonstrates the methodological potential and additional contribution of an online, signal-based, within-person approach to study clinical symptoms (Helbig et al., Reference Helbig, Lang, Swendsen, Hoyer and Wittchen2009). In further studies it might be promising to combine both psychophysiological and psychological measures of everyday experiences in order to further disentangle underlying mechanisms in the maintenance of panic disorder.
Comments
No Comments have been published for this article.