ARTICLE IN PRESS

Behaviour Research and Therapy 45 (2007) 2628–2639 www.elsevier.com/locate/brat

Inhibitory control and symptom severity in late life generalized anxiety disorder$ Rebecca B. Price!, Jan Mohlman Department of Psychology, Rutgers, the State University of New Jersey, 152 Frelinghuysen Road, Piscataway, NJ 08854, USA Received 20 February 2007; received in revised form 23 May 2007; accepted 18 June 2007

Abstract Contemporary models of generalized anxiety disorder (GAD) posit that worry functions as an avoidance strategy. During worry, individuals inhibit threat-related imagery in order to minimize autonomic reactivity to phobic topics. This conceptualization of worry suggests a role for the executive system in exerting top-down inhibitory control (IC) over threat processing. We tested the hypothesis that better performance on an IC task would be associated with greater severity of worry and concomitant anxious mood. Forty-three older adults (age 60–77) with GAD completed the Stroop color word task and a battery of self-report symptom measures. Fifteen of the GAD patients were paired with age- and sex-matched non-anxious controls. In the full GAD sample, age-normed t-scores of Stroop performance were positively correlated with measures of worry and trait anxiety, but not anxious arousal or depression. Positive relationships between IC and symptom severity were upheld in the smaller subsample of GAD patients, while in the matched control group, no relationships between Stroop scores and clinical measures were observed. Patients and controls did not differ in Stroop performance. In the context of a disorder-specific tendency to make maladaptive use of executive functions, better IC may be associated with more severe symptomatology. r 2007 Elsevier Ltd. All rights reserved. Keywords: Generalized anxiety disorder; Worry; Executive functioning; Inhibitory control; Anxiety and aging

Introduction Worry as a cognitive avoidance strategy Generalized anxiety disorder (GAD) is a prevalent and disabling disorder characterized by pervasive, excessive, and uncontrollable worry (American Psychiatric Association, 2001). Contemporary models of GAD emphasize two central features of the condition: the persistence of pervasive and uncontrollable worry, and the notable absence of many of the somatic features (e.g. faintness, heart palpitations, and excessive sweating) that reliably characterize other anxiety disorders (Marten et al., 1994). For example, Borkovec’s avoidance theory $

This research was supported in part by a National Alliance for Research on Schizophrenia and Depression (NARSAD) Young Investigator Grant (YI2002) awarded to the second author. !Corresponding author. Tel.: +1 347 200 4234; fax: +1 732 445 2263. E-mail address: [email protected] (R.B. Price). 0005-7967/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2007.06.007

ARTICLE IN PRESS R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

2629

of worry (Borkovec, 1994; Borkovec, Alcaine, & Behar, 2004; Borkovec, Ray, & Stober, 1998) posits that worry is a cognitive avoidance strategy used to decrease discomfort level in the face of anxiety-provoking stimuli. Laboratory studies of voluntary worry have shown that worry is comprised of thoughts, as opposed to images (Borkovec & Inz, 1990), and, in contrast to imaginal processing, is associated with a reduction in autonomic nervous system response to fear material both during (Vrana, Cuthbert, & Lang, 1986) and after (Borkovec & Hu, 1990; Peasley-Miklus & Vrana, 2000) worry. Because worry is a successful strategy for the short-term suppression of anxious reactivity, worry is negatively reinforced as a coping tactic (e.g., Mowrer, 1947). However, worry is ultimately maladaptive because its strictly verbal nature precludes the comprehensive emotional processing (e.g., mental imagery, intense feelings of arousal, and imagining the ‘worst outcome’) believed to be necessary for fear extinction (Foa & Kozak, 1986). In the avoidance conceptualization of worry, avoidance of unwanted emotional/imaginal content is achieved through a selective processing strategy in which attention is directed toward verbal information and away from mental images. In a sense, the behavioral avoidance that characterizes other anxiety disorders is turned inward, necessitating a degree of cognitive dexterity in order to direct attention narrowly and avoid emotional contact with anxiety-provoking stimuli. This selective processing style may require proficiency in the effortful control of attention. During worry, an individual devotes attentional resources to the renumerative linguistic processing of threatening stimuli while holding at bay other demands on processing capacity. Because imaginal representations of the same aversive stimulus are likely to be primed repeatedly during the course of the worry episode (Borkovec et al., 2004), this task may require considerable inhibitory control (IC) of attention in order to maintain top-down attenuation of the fear response. IC and the neurocognitive substrates of worry The executive functions governed by the prefrontal cortex (PFC; Fuster, 1997) may play a crucial role in modulating the body’s response to threat during worry. When multiple processing routines are available, it is the role of the executive system to inhibit task-irrelevant processing while directing attentional resources toward task-relevant processing (e.g., Smith & Jonides, 1999). This skill becomes most crucial when the context promotes the retrieval of response tendencies that stand in conflict to the current task. IC or the executive ability to inhibit task-irrelevant processing, may be particularly relevant to the maintenance of GAD symptoms. During an episode of worry, the executive system may impose a strict processing routine that keeps verbal representations of the worrisome topic quickly retrievable while simultaneously inhibiting the activation of threat-related images, even though the conditions of worry (i.e., elaborative processing of a threatening subject) are likely to prime imaginal representations and emotional responding. Although a role for IC during worry has intuitive appeal, as Borkovec, Alcaine, and Behar (2004) note, the mechanism by which worry accomplishes suppression of somatic reactions has yet to be defined. To our knowledge no work has been done to directly test the role of executive functions in worry. Executive skills, including IC, are known to be implemented primarily by the PFC (e.g., Miller & Cohen, 2001; Smith & Jonides, 1999). Although executive processes and PFC activation are not necessarily synonymous, evidence of PFC recruitment during worry would be expected if IC and worry are indeed closely linked. To date, neuroimaging studies of worry and GAD have been scant; nevertheless, several increases in prefrontal indices have been documented, including increased PFC neuronal integrity (Mathew et al., 2004) and increased PFC activity in GAD patients (Hoehn-Saric, Schlund, & Wong, 2004; Wu et al., 1991) as well as in non-anxious participants who are prompted to worry (Hoehn-Saric, Lee, McLeod, & Wong, 2005). In contrast to the notion that worry relies on IC, the contradictory assertion that inhibitory deficits contribute to GAD pathology has also been made. A substantial body of research documents involuntary attentional biases toward the encoding of threat-related information in anxious individuals, including individuals with GAD (for a review, see MacLeod & Rutherford, 2004). There is presently no consensus regarding the cognitive mechanism that best accounts for these biases (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van Ijzendoorn, 2007). While some theoretical models imply that a primary deficit in controlled, top-down processing abilities such as IC leaves GAD patients with little ability to adaptively modulate threat processing (MacLeod & Rutherford, 1992, 2004), others emphasize the role of early evaluative processes that operate prior to the engagement of higher-order executive functions like IC. For

ARTICLE IN PRESS 2630

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

example, Mogg and Bradley (1998) propose that anxious biases stem primarily from a low threshold for threat appraisal. In this conceptualization, once a stimulus has been preattentively evaluated as a threat, both highand low-anxious individuals will experience similar interruptions of ongoing activity, suggesting a minimal role for top-down inhibitory deficits in producing biased response patterns to emotional stimuli. A direct test of the relationship between GAD symptoms and IC using non-emotional stimuli may shed light on this issue, as emotionally neutral stimuli will be unlikely to receive a threat appraisal from either anxious or non-anxious individuals. If inhibitory deficits do contribute fundamentally to GAD pathology, then we might expect poorer IC, even on a non-emotional task, to correspond to increased severity of GAD symptoms. Conversely, if IC is intact in GAD patients and is, in fact, recruited during worry in order to achieve inhibition of fearful imagery, we would predict just the opposite—better scores on a general measure of IC should be seen in conjunction with more consistent reinforcement for worry as an avoidance tactic, leading subsequently to more worry and a more severe experience of related anxiety variables. Present study These disparate predictions regarding the relationship between IC, worry, and GAD symptomatology suggest the need for direct empirical investigation. We collected data from older adults with GAD because in geriatric populations, executive dysfunction is relatively common even in the absence of disease, trauma, or neurotoxic events (Albert, 1994; Craik, Anderson, Kerr, & Li, 1995; West, 1996). The cognitive developmental literature highlights age-related performance declines in tasks requiring a high degree of cognitive control (including IC) occurring as a function of normal aging (Braver & Barch, 2002; Hedden & Gabrieli, 2004). If IC plays a key role in GAD symptoms such as worry, then individual differences in IC brought on by age may impact the frequency with which worry is employed as an arousal avoidance strategy. This population therefore provides both an opportunity and an impetus to examine the relationship between IC and GAD symptoms. Considerable heterogeneity in IC allows us to test the hypothesis that GAD symptoms are exacerbated through application of top-down executive skills. We tested the hypothesis that individuals with GAD would possess age-normative, rather than impaired, capacities for IC as indexed by the Stroop (1935) color word (Stroop C-W) task, a measure of IC that requires participants to identify the color of ink in which words are printed while inhibiting the prepotent response to read the (incongruent) color name. Furthermore, we predicted that an individual with greater IC would be capable of applying their maladaptive worry habit to greater (detrimental) effect. Therefore, we predicted that better Stroop C-W scores would be seen in conjunction with greater severity of worry and concomitant increases in trait anxiety and anxious mood. Conversely, we did not expect to find a positive relationship between IC and a measure of anxious arousal, since the avoidance conceptualization of worry suggests that successful application of IC during worry should result in decreased somatic anxiety, at least in the short-term. Similarly, we expected no positive association between IC and measures of depression in our sample, based on previous studies demonstrating an inverse relationship between executive measures and depression (e.g. Derryberry & Rothbart, 1988; Van den Berg, Oldehinkel, Brilman, Bouhuys, & Ormel, 2000). To test the specificity of these relationships to a clinical population, we included a non-anxious control group matched on age, sex, and handedness for comparison. We did not expect to find any correlation between Stroop C-W performance and GAD measures in the non-anxious controls, because we assumed that IC would be deleterious only in the context of a GAD-specific avoidance strategy that engages it. Methods Participants Participants were recruited through media advertisements from the New York, NY and Syracuse, NY communities as part of two clinical treatment trials for late life GAD. The total patient sample consisted of 43 adults aged 60 and over (mean ¼ 66.1, SD ¼ 5.57, range ¼ 60–77) meeting DSM-IV criteria for a primary diagnosis of GAD. Descriptive statistics for the full sample can be found in Table 1. The participants had comorbid diagnoses of major depressive disorder (23%), dysthymia (19%), specific phobia (12%), social

ARTICLE IN PRESS 2631

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639 Table 1 Descriptive statistics for GAD full sample, GAD subsample, and controls GAD full sample (n ¼ 43)

GAD subsample (n ¼ 15)

Controls (n ¼ 14)a

Age % Female % Retired % Caucasian % African American % Hispanic

66.1 (5.57) 51 58 84 12 5

67.7 (5.75) 60 60 87 13 0

68.0 (5.14) 60 80 94 6 0

Clinical measures BAI STAI-T PSWQ HRSA BDI HRSD

17.02 55.16 57.16 – 19.53 –

18.80 52.60 54.40 19.80 16.47 12.33

4.93 (6.18)** 28.50 (6.47)** 40.43 (11.86)** 2.64 (1.69)** 5.93 (4.21)* 1.21 (1.25)**

Neuropsych AMNART Digits Stroop Color Stroop C-Word

116.65 (7.2) 51.05 (10.0) 53.19 (3.6) 45.63 (11.6)

(10.2) (12.2) (10.7) (10.0)

(10.23) (15.71) (8.48) (8.74) (10.73) (5.61)

116.81 (6.79) 50.53 (6.29) 52.27 (3.24) 45.87 (9.36)

121.65 (7.27) 50.07 (5.61) 53.00 (2.35) 45.57 (9.25)

Note: Group comparisons of GAD subsample and controls made by independent measures t-tests; AMNART scores given are a calculated estimate of Verbal IQ, all other neuropsychological measures are reported as age-normed t-scores; BAI ¼ Beck Anxiety Inventory; STAI-T ¼ State Trait Anxiety Inventory, Trait subscale; PSWQ ¼ Penn State Worry Questionnaire; HRSA ¼ Hamilton Rating Scale for Anxiety; BDI ¼ Beck Depression Inventory; HRSD ¼ Hamilton Rating Scale for Depression; AMNART ¼ American Nelson Adult Reading Test; Digits ¼ Digit Span subtask of the Wechsler Adult Intelligence Scale, 3rd edition; Stroop C ¼ 2-min Stroop Color task; and Stroop C-W ¼ 2-min Stroop Color-Word task. *po.05; **po.01; all p’s adjusted for multiple comparisons. a Only 14 control participants completed all clinical and neuropsychological measures.

anxiety disorder (16%), and panic disorder (14%). Participants were required to have intact basic cognitive skills (Mini-Mental State Exam score X23), to read and write in English, to have no history of suicidality in the previous 6 months, to have never experienced psychotic symptoms, and to be free of anxiolytic and antidepressant medications. Because it was thought that the accuracy of self-report responses, as well as Stroop performance, could be influenced by third factors such as cognitive ability and motivation, we also explored whether IC was related to clinicians’ symptom ratings. A subset of the GAD patients (n ¼ 15) volunteered to complete additional clinician-rated assessments and were paired with age- and sex-matched control participants (n ¼ 15) also recruited from the Syracuse, NY community. Control participants were subject to the same exclusion criteria as above, with the additional criterion that they be free of all current and past DSM-IV Axis I psychopathology. Control participants completed all clinical and neuropsychological assessments. One control participant withdrew prior to completion of all assessments; his data were excluded from these analyses. Neuropsychological assessments IC was assessed by the color-word trial of the 2-minute Stroop Test (Trenerry, Crosson, DeBoe, & Leber, 1989). The Stroop C-W task is the paradigmatic test of executive inhibition ability (Stroop, 1935). Participants are presented with a list of 112 written color names, printed in an incongruent color of ink. The task requires that participants identify the color of ink that each word is written in while inhibiting the prepotent response to read the written name. Raw scores were computed by subtracting the number of errors from the total number of items completed within a 2 min period. To eliminate the possibility of ceiling effects, participants

ARTICLE IN PRESS 2632

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

were instructed to return to the start of the list and continue the task for the full 2 min if they should complete all items before the time limit was reached. Raw scores were converted to age-normed t-scores based on normative data published in the Stroop Test manual (Trenerry et al., 1989). Simple attention and reaction time were assessed by the Stroop color (Stroop C) trial immediately prior to Stroop C-W completion. Participants were presented with a list of 112 incongruent color names printed in colored ink, quite similar to the list presented in the Stroop C-W task, and required to read the written color names aloud as quickly as possible. This task is expected to tap a similar skill set to the Stroop C-W task, with the important exception that successful performance does not require executive inhibition of a prepotent response. Vocabulary and reading ability were assessed by the American Nelson Adult Reading Test (AMNART; Grober & Sliwinski, 1991) and were combined with years of education to compute a standardized estimate of verbal IQ (Grober & Sliwinski, 1991). Verbal working memory capacity was assessed by the Digit Span subscale of the WAIS-III (Wechsler, 1997). All measures are standard neuropsychological testing instruments with well-established normative data for older adults. Raw scores on the Stroop C and Digit Span tasks were converted to t-scores on the basis of test manual guidelines. Clinical measures Diagnoses were established by the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID; First, Gibbon, Spitzer, & Williams, 1995). A random sample of 12 audiotaped SCIDs rated by a blind assessor yielded a k coefficient of .92 for the diagnosis of GAD; however, prior to their visit all participants passed a preliminary phone screen designed to ensure a high likelihood of meeting GAD criteria, which may have led to an inflated k value for SCID diagnoses. Participants completed a packet of self-report questionnaires to assess anxiety and depression symptom severity. Primary anxiety measures included the Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990) and the State-Trait Anxiety Inventory— Trait subscale (STAI-T; Spielberger, Gorsuch, & Lushene, 1983). Anxious arousal was assessed by the Beck Anxiety Inventory (BAI; Beck & Steer, 1990). Depressive symptomatology was assessed using the Beck Depression Inventory (BDI; Beck & Steer, 1987). Internal consistency coefficients for self-report measures ranged from .77 to .92. To obtain clinician-rated measurements of anxiety and depression, the Hamilton Rating Scales for anxiety (HRSA; Hamilton, 1959) and depression (HRSD; Hamilton, 1960) were administered by trained graduate students. These measures were administered to the smaller subset of GAD patients and matched control participants only. Inter-rater reliability on the Hamilton scales was adequate (k ¼ .67 for HRSA; k ¼ .72 for HRSD). All clinical measures have sound psychometric properties for use with older populations (e.g., Beck & Stanley, 2001; Beck, Stanley, & Zebb, 1999). Results Full GAD sample analyses In our sample of 43 GAD patients, Stroop C-W t-scores were distributed slightly below the standardized norms for this age group (mean t-score ¼ 45.63, SD ¼ 11.6, range ¼ 28–69). Mean scores on all anxiety measures were reflective of clinically significant anxiety (Antony, Orsillo, & Roemer, 2001). Descriptive statistics for all clinical and neuropsychological measures are presented in Table 1. Bivariate correlations were performed to test the relationship between Stroop C-W scores and the four clinical measures obtained from the full sample. In the patient group, Stroop C-W scores were positively correlated with scores on the PSWQ (r ¼ .44, po.01) and the STAI-T (r ¼ .35, po.05). Stroop C-W scores and scores on the BAI (r ¼ ".04, n.s.) and BDI (r ¼ .04, n.s.) were unrelated. To separate the role of IC from more basic cognitive skills (e.g. simple attention, verbal IQ, working memory), we performed partial correlations controlling for three basic cognitive measures: Stroop C task, AMNART, and Digit Span. Because age was marginally correlated with Stroop C-W performance (r ¼ ".27, po.10), we also controlled for age. The inclusion of these covariates did not alter significance for any analysis. Pearson’s r values for the partial correlations between the four clinical variables and Stroop C-W score are presented in Table 2.

ARTICLE IN PRESS 2633

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

Table 2 Partial correlations (controlling for age and basic cognitive skills) of Stroop color-word performance with anxiety and depression measures in the GAD full sample, GAD subsample, and controls Stroop Color-Word T-score Symptom measure

GAD full sample (n ¼ 43)

GAD subsample (n ¼ 15)

Controls (n ¼ 14)

Anxiety measures PSWQ STAI-T BAI HRSA

.40* .37* ".02 –

.75** .59 .72* .77**

.03 .22 ".03 ".08

Depression measures BDI HRSD

".03 –

.25 .26

.29 ".20

Note: PSWQ ¼ Penn State Worry Questionnaire; STAI-T ¼ State Trait Anxiety Inventory, Trait subscale; BAI ¼ Beck Anxiety Inventory; HRSA ¼ Hamilton Rating Scale for Anxiety; BDI ¼ Beck Depression Inventory; and HRSD ¼ Hamilton Rating Scale for Depression. *po.05; **po.01.

Small group analyses In the subgroup of GAD patients (n ¼ 15) and matched control participants, independent samples t-tests revealed that the GAD group exhibited more severe symptomatology than the controls on all clinical measures (PSWQ, BAI, STAI-T, BDI, HRSA, and HRSD), adjusting for multiple comparisons (t(28)X3.43, po.008). The groups did not differ significantly on Stroop C-W performance (t(28) ¼ ".09, n.s.) or on any basic cognitive measure. Descriptive statistics for the two small groups are summarized in Table 1. Bivariate correlations were performed separately for the patient and control groups to assess the relationship between IC and clinical measures in the presence or absence of GAD psychopathology. In the patient subsample, Stroop C-W scores were positively correlated with scores on the PSWQ (r ¼ .63, po.013), the HRSA (r ¼ .59, po.05), and, contrary to predictions, the BAI (r ¼ .68, po.013). Stroop performance and STAI-T scores showed a positive relationship that did not reach significance (r ¼ .49, p ¼ .067). As expected, Stroop performance was not related to either measure of depression (BDI, HRSD). In the control group, Stroop C-W performance was not significantly correlated with any clinical measure. As in the full GAD sample analysis, we performed partial correlations controlling for age and three basic cognitive measures: Stroop C task, AMNART, and Digit Span. The addition of these covariates did not alter significance for any analysis. Pearson’s r values for all partial correlations are presented in Table 2. The correlational analyses described above revealed significant relationships between Stroop performance and three clinical measures (HRSA, BAI, and PSWQ) that were present in the GAD group but not the nonanxious controls. To model the observed interaction between GAD status (patient vs. control) and IC statistically, a stepwise regression was performed on each of these three self-report anxiety variables. In step 1, the clinical measure of interest was regressed on the four control variables (age, Stroop C, Digits, and Amnart). In step 2, GAD status (GAD patients coded as 0, non-anxious controls coded as 1) and meandeviated Stroop C-W scores were entered. In step 3, the GAD status # Stroop C-W score cross-product term was entered. The final regression model was significant for all three outcome measures (F(7,21) ¼ 2.71, 6.01, 19.17 for the PSWQ, BAI, and HRSA, respectively; all p’so.05). Applying a Bonferroni adjustment for multiple comparisons resulted in a loss of significance for the PSWQ only. Change in R2 at step 1 was not significant for any measure, indicating that the control variables did not significantly improve the prediction of any clinical variable. However, in the final model for the HRSA, coefficient terms for two control variables, age (t(28) ¼ 2.05, p ¼ .053) and Digits (t(27) ¼ 2.05, p ¼ .053), were marginally significant, a pattern not present in the models for the other two clinical measures. The increase in R2 when the interaction factor was entered (between steps 2 and 3) was significant for the HRSA (F(1,21) ¼ 12.07, po.01) and the BAI (F(1,21) ¼ 7.58, po.05), indicating that, as expected, the interaction between GAD status and Stroop C-W

ARTICLE IN PRESS 2634

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

Table 3 Stepwise regression statistics for variables regressed on HRSA, BAI, and PSWQ b

Step 1 Age Stroop C Digits Amnart R2

HRSA

BAI

PSWQ

.19y .14 .20y .03 .21

".02 .23 ".06 .07 .13

.14 .07 ".17 .15 .05

Step 2 GAD status Stroop C-W R2 DR2

".83*** .58*** .79*** .58***

".70*** .67** .55** .42***

".64*** .54* .39 .34**

Step 3 GAD # Stroop R2 DR2

".43** .87*** .08**

".53* .67*** .12*

".44y .47* .08y

Note: All standardized coefficient (b) values are those obtained in the final model; HRSA ¼ Hamilton Rating Scale for Anxiety; BAI ¼ Beck Anxiety Inventory; and PSWQ ¼ Penn State Worry Questionnaire. y po.10; *po.05; **po.01; *** po.001.

Table 4 Stroop C-W coefficient statistics for multivariate regressions on HRSA, BAI, and PSWQ with group set to 0 (GAD) or 1 (controls) GAD (n ¼ 15)

Control (n ¼ 14)

HRSA b b t(28) p

.68 .58 4.73 o.001

".05 ".04 ".31 n.s.

BAI b b t(28) p

.80 .67 3.51 .002

".12 ".10 ".46 n.s.

PSWQ b b t(28) p

.73 .54 2.25 .036

".14 ".10 ".38 n.s.

Note: HRSA ¼ Hamilton Rating Scale for Anxiety; BAI ¼ Beck Anxiety Inventory; and PSWQ ¼ Penn State Worry Questionnaire.

performance significantly improved prediction of self-reported anxiety beyond either factor treated independently. For the PSWQ, the increase in R2 with the addition of the cross-product term was present as a non-significant trend (F(1,21) ¼ 3.34, p ¼ .08). Regression statistics for all analyses are presented in Table 3. Simple slope analyses revealed that for GAD patients, IC significantly predicted HRSA (t(28) ¼ 4.73, po.001), BAI (t(28) ¼ 3.51, po.01), and PSWQ (t(28) ¼ 2.25, po.05) in the final model, whereas for individuals in the control group, IC was not a significant predictor of any clinical measure (all p’s4.65).

ARTICLE IN PRESS R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

2635

Again, applying a Bonferroni adjustment for multiple comparisons resulted in a loss of significance for the PSWQ only. Table 4 summarizes Stroop C-W regression coefficient statistics for GAD and control participant subgroups. Discussion General discussion In our full sample of older adults with GAD, IC was positively associated with self-report measures of worry (PSWQ) and trait anxiety (STAI-T), but not measures of anxious arousal (BAI) or depression (BDI). These correlations held even when we controlled for age and basic cognitive skills. In a smaller subset of these individuals who completed additional assessments, the positive correlation between IC and worry remained significant. Furthermore, additional measures of anxiety symptomatology were positively related to IC, including clinician-rated anxiety (HRSA), and, unexpectedly, anxious arousal (BAI), while there was no evidence of a relationship between IC and any clinical measure in a sample of age- and sex-matched nonanxious controls. In regression analyses of combined data from these 15 GAD patients and 14 age- and sexmatched controls, a model that included the interaction factor between GAD status and Stroop C-W performance explained significantly more variance in HRSA, BAI, and (marginally) PSWQ scores than a model including these two predictors and control variables alone, indicating that the predictive power of IC is contingent upon the presence of the GAD syndrome. In other words, only within the context of GAD is increased IC associated with a more severe anxiety profile. For the HRSA only, regression coefficients for two control variables (age and digit span) approached significance in the final model (p ¼ .053), suggesting that demographic and basic cognitive variables may factor into clinicians’ ratings of a patient’s anxiety, while not affecting the patients’ own self-reports. The fact that the Hamilton scales (HRSA and HRSD) yielded similar findings to the self-report measures suggests that the observed relationships between IC and symptom severity cannot be easily explained by individual differences in accuracy, motivation level, or attention to detail during questionnaire completion. However, our Hamilton scale findings must be interpreted with caution given that estimates of inter-rater reliability were somewhat low. One major implication of these findings is that the behaviors that typify GAD seem not to stem from an underlying deficit in global IC, but from a consistent failure to make use of inhibitory executive skills in adaptive ways. In light of previous findings from emotional Stroop (eStroop) task paradigms, individuals with GAD may exhibit an encoding bias for threat-relevant stimuli (Macleod & Rutherford, 2004) which is sometimes interpreted as a failure of IC. However, our results suggest that performance differences on emotional tasks like the eStroop do not generalize to non-emotional tasks, as there were no group differences in performance when GAD patients were compared to age- and sex-matched non-anxious controls. Though the discrepancy between Stroop C-W and eStroop results might be explained by stimulus properties alone (i.e., the presence of an emotional stimulus might be necessary in order to invoke an anxious processing style), our findings have potential implications for theoretical models that seek to explain attentional bias in GAD patients, suggesting that a deficiency in top-down IC may be less fundamental than other theorized mechanisms of bias, such as a tendency to assign high threat value to a wide range of stimuli (Mogg & Bradley, 1998). In fact, rather than demonstrating IC deficits, individuals with GAD appear to suffer more severely from symptoms of worry and anxiety when their IC exceeds normative averages, suggesting a potential role for inhibitory skills in the processes that uphold or exacerbate GAD symptoms. Because worry is thought to involve inhibition of threat-related image processing (Borkovec et al., 2004), executive IC may facilitate engagement in worry as a maladaptive avoidance strategy. Our results indicate that IC is associated not only with an increased propensity to worry, but also with increased frequency of anxious mood as indexed by the STAI-T and the HRSA. These findings are consistent with models of GAD that propose an instrumental role for worry in maintaining and promoting habitual anxious mood states. One inconsistency in our findings concerns the extent to which habitually high levels of anxious arousal, as indexed by the BAI, are also associated with increased IC. In our full GAD sample these measures were uncorrelated, while in a subset of the same GAD patients, a significant positive relationship was observed. It is unclear what sample characteristics may account for the disparate findings other than differential sample size.

ARTICLE IN PRESS 2636

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

Others have remarked that older adults with GAD are more likely than younger adults to express anxiety symptoms in terms of somatic complaints (Beck & Averill, 2004), a distinction which could reflect either a difference in symptom experience or a cohort effect stemming from a relative unwillingness to discuss emotional experiences in overt terms (Lipton & Schaffer, 1988). Thus, it may be that in the older adult population, GAD symptomatology fosters diverse response tendencies on measures of anxious arousal such as the BAI, making the measure less consistent in its relationship to other variables across multiple samples. Given the theoretical relevance of somatic anxiety symptoms in the avoidance conceptualization of worry, the relationship between anxious arousal, worry, and IC warrants further exploration, ideally with direct measurement of somatic symptoms (e.g., heart rate variability and skin conductance) supplementing selfreport indices. Study limitations The generalizability of our results may be qualified by our relatively small sample size and by the age range and specific demographics of our sample. The association between executive performance and GAD symptoms we observed may be an age-specific effect that would not generalize to a younger sample, given that older adults differ from younger adults on IC tasks (e.g., Hasher, Stoltzfus, Zacks, & Rypma, 1991). Although the k coefficient for GAD diagnoses reflected strong inter-rater reliability in our sample, diagnoses were based on a single interview rather than the repeated independent interviews recommended by some authors (Borkovec, Newman, Pincus, & Lytle, 2002) due to the relatively poor diagnostic reliability of GAD (Brown, DiNardo, Lehman, & Campbell, 2001). The generalizability of our findings to GAD patients at large may be limited given that our group means for the PSWQ (in both the GAD full and subsamples) fell below clinical cut-points (62, 65) established through receiver operating characteristic (ROC) analyses in younger samples (Behar, Alcaine, Zuellig, & Borkovec, 2003; Fresco, Mennin, Heimberg, & Turk, 2003). However, our values were only slightly lower than previously reported means for older GAD patients (e.g., Stanley, Beck et al., 2003; Stanley, Diefenbach et al., 2003) and higher than the older adult diagnostic cut-point of 50 determined by ROC analysis of geriatric primary care patients (Stanley, Diefenbach et al., 2003). Our results are further limited by our reliance on correlational analyses. This design precludes a more detailed exploration of the causal relationships between IC, worry, and other symptoms of anxiety, and leaves our findings open to alternative explanations. We cannot rule out the possibility that the observed relationships resulted from a causal influence of worry on IC—i.e., that frequent worry leads to more finetuned inhibitory skill—or from a third, uncontrolled factor. However, if frequent worry, or any other factor closely linked to clinical anxiety, can causally account for the observed pattern of Stroop performance, we would expect a performance benefit for GAD participants over age- and sex-matched non-anxious controls. This group difference in Stroop C-W performance was not observed, despite significantly elevated levels of worry and anxiety in the clinical group. Furthermore, no associations between anxiety scores and Stroop C-W performance were found in the non-anxious control group, suggesting that the link between clinical and inhibitory measures hinges on a disorder-specific pattern of behavior, not on more broadly applicable factors that contribute to Stroop C-W performance across a range of clinical and non-clinical populations. Therefore, we suggest that our data evidence an integral role for IC in worry (the maladaptive behavior that best typifies GAD), exacerbating the syndrome by facilitating cognitive avoidance. Implications of findings Though qualified by our study’s limitations, our results indicate that GAD is not characterized by an underlying deficit in IC at the broad level of the executive system. Our findings caution against a simplistic characterization of emotional psychopathology at large, or anxiety disorders in particular, as invariably denoting inadequate top-down IC. Our findings are apparently in contrast to recent neurocognitive findings from other psychiatric populations. For example, in depression, behaviors such as depressive rumination and negative memory biases have been posited to stem from a deficit in top-down inhibition of the amygdala by the PFC, resulting in involuntary sustained processing of negative information (Davidson, 2000; Drevets, 1999; Siegle, Steinhauer, Thase, Stenger, & Carter, 2002; Siegle, Thompson, Carter, Steinhauer, & Thase,

ARTICLE IN PRESS R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

2637

2007). Similarly, Kent and Rauch (2004) have proposed a neuroanatomical model of PTSD in which amygdalar hyper-responsivity is coupled with inadequate top-down control from the anterior cingulate cortex (a region involved in IC and habituation to feared stimuli) as well as the hippocampus, leading to a dysregulated conditioned fear response. If GAD differs from other emotional disorders in the directionality of the relationship between executive functions and symptom severity, this could imply that a subset of the higher-order neurocognitive processes involved in the etiology and/or maintenance of GAD are quite distinct from these other disorders, despite considerable overlap in the subjective experience of symptoms such as anxious mood, worry, and negative affect. Future studies should seek to explore the relationship between GAD symptoms and additional executive functions not yet examined (e.g., task switching, sequence planning, and abstract reasoning). Increasing our understanding of these distinctions on the neurocognitive level may eventually inform the development of treatments to target specific disorders, or patients with specific cognitive-affective profiles, most effectively. Acknowledgments This research was supported in part by a National Alliance for Research on Schizophrenia and Depression (NARSAD) Young Investigator Grant (YI2002) awarded to the second author. We thank G. Terence Wilson, Ph.D. and Brian Chu, Ph.D. for their comments on an earlier draft of this manuscript. References Albert, M. S. (1994). Age related changes in cognitive function. In M. L. Albert, & J. E. Knoefel (Eds.), Clinical neurology of aging (pp. 314–328). New York: Oxford University Press. American Psychiatric Association. (2001). Diagnostic and statistical manual of mental disorders—Text revision (4th ed.). Washington, DC: APA. Antony, M. M., Orsillo, S. M., & Roemer, L. (2001). Practitioner’s guide to empirically based measures of anxiety. New York: Kluwer Academic/Plenum Publishers. Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A Meta-analytic study. Psychological Bulletin, 133, 1–24. Beck, J. G., & Averill, P. M. (2004). Older adults. In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder (pp. 409–433). New York: The Guilford Press. Beck, J. G., & Stanley, M. A. (2001). Assessment of anxiety disorders in older adults: Current concerns, future prospects. In M. M. Antony, S. M. Orsillo, & L. Roemer (Eds.), Practitioner’s guide to empirically based measures of anxiety (pp. 43–47). New York: Kluwer Academic/Plenum Publishers. Beck, J. G., Stanley, M., & Zebb, B. (1999). Effectiveness of the Hamilton Anxiety Rating Scale with older generalized anxiety disorder patients. Journal of Clinical Geropsychology, 5, 281–292. Beck, A. T., & Steer, R. A. (1987). Beck depression inventory: Manual. San Antonio, TX: The Psychiatric Corporation. Beck, A. T., & Steer, R. A. (1990). Manual for the beck anxiety inventory. San Antonio, TX: Psychological Corporation. Behar, E., Alcaine, O., Zuellig, A. R., & Borkovec, T. D. (2003). Screening for generalized anxiety disorder using the Penn State Worry Questionnaire: A receiver operating characteristics analysis. Journal of Behavior Therapy and Experimental Psychiatry, 34, 25–43. Borkovec, T. D. (1994). The nature, functions, and origins of worry. In G. C. L. Davey, & F. Tallis (Eds.), Worrying: Perspectives on theory, assessment and treatment (pp. 5–33). Oxford: Wiley. Borkovec, T. D., Alcaine, O. M., & Behar, E. (2004). Avoidance theory of worry and generalized anxiety disorder. In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder (pp. 77–108). New York: The Guilford Press. Borkovec, T. D., & Hu, S. (1990). The effect of worry on cardiovascular response to phobic imagery. Behaviour Research and Therapy, 28, 69–73. Borkovec, T. D., & Inz, J. (1990). The effect of worry on cardiovascular response to phobic imagery. Behaviour Research and Therapy, 28, 69–73. Borkovec, T. D., Newman, M. G., Pincus, A. L., & Lytle, R. (2002). A component analysis of cognitive-behavioral therapy for generalized anxiety disorder and the role of interpersonal problems. Journal of Consulting and Clinical Psychology, 70, 288–298. Borkovec, T. D., Ray, W. J., & Stober, J. (1998). Worry: A cognitive phenomenon intimately linked to affective, physiological, and interpersonal behavioral processes. Cognitive Therapy and Research, 22, 561–576. Braver, T. S., & Barch, D. M. (2002). A theory of cognitive control, aging cognition, and neuromodulation. Neuroscience and Biobehavioral Reviews, 26, 809–817. Brown, T. A., DiNardo, P. A., Lehman, C. L., & Campbell, L. A. (2001). Reliability of DSM-IV anxiety and mood disorders: Implications for the classification of emotional disorders. Journal of Abnormal Psychology, 110, 49–58.

ARTICLE IN PRESS 2638

R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

Craik, F. I. M., Anderson, N. D., Kerr, S. A., & Li, K. Z. (1995). Memory changes in normal aging. In A. D. Baddeley, B. A. Wilson, & F. N. Watts (Eds.), Handbook of memory disorders (pp. 211–242). Chichester, England: Wiley. Davidson, R. J. (2000). Affective style, psychopathology, and resilience: Brain mechanisms and plasticity. American Psychologist, 55, 1196–1214. Derryberry, D., & Rothbart, M. K. (1988). Affect, arousal, and attention as components of temperament. Journal of Personality and Social Psychology, 55, 958–966. Drevets, W. C. (1999). Prefrontal cortical-amygdalar metabolism in major depression. Annals of the New York Academy of Sciences, 877, 614–637. First, M. B., Gibbon, M., Spitzer, R. L., & Williams, J. B. W. (1995). Structured clinical interview for DSM-IV Axis I disorders. SCID-I, version 2.0, October 1995, Final version. Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological Bulletin, 99, 20–35. Fresco, D. M., Mennin, D. S., Heimberg, R. G., & Turk, C. L. (2003). Using the Penn State Worry Questionnaire to identify individuals with generalized anxiety disorder: A receiver operating characteristic analysis. Journal of Behavior Therapy and Experimental Psychiatry, 34, 283–291. Fuster, J. M. (1997). The prefrontal cortex (3rd ed.). Philadelphia, PA: Lippincott-Raven. Grober, E., & Sliwinski, M. (1991). Development and validation of a model for estimating premorbid verbal intelligence in the elderly. Journal of Clinical and Experimental Neuropsychology, 13, 933–949. Hamilton, M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50–55. Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery and Psychiatry, 23, 56–62. Hasher, L., Stoltzfus, E. R., Zacks, R. T., & Rypma, B. (1991). Age and inhibition. Journal of Experimental Psychology: Learning, Memory, & Cognition, 17, 163–169. Hedden, T., & Gabrieli, J. D. E. (2004). Insights into the ageing mind: A view from cognitive neuroscience. Nature Reviews Neuroscience., 5, 87–96. Hoehn-Saric, R., Lee, J. S., McLeod, D. R., & Wong, D. F. (2005). Effect of worry on regional cerebral blood flow in nonanxious subjects. Psychiatry Research: Neuroimaging, 140, 259–269. Hoehn-Saric, R., Schlund, M. W., & Wong, S. H. (2004). Effects of citalopram on worry and brain activation in patients with generalized anxiety disorder. Psychiatry Research, 131, 11–21. Kent, J. M., & Rauch, S. L. (2004). Neuroimaging studies of anxiety disorders. In D. S. Charney, & E. J. Nestler (Eds.), Neurobiology of mental illness (2nd ed., pp. 639–660). New York: Oxford University Press, Inc. Lipton, M., & Schaffer, W. (1988). Physical symptoms related to post-traumatic stress disorder (PTSD) in an aging population. Military Medicine, 153, 316–318. MacLeod, C., & Rutherford, E. M. (1992). Anxiety and the selective processing of emotional information: Mediating roles of awareness, trait and state variables, and personal relevance of stimulus materials. Behaviour Research and Therapy, 30, 479–491. MacLeod, C., & Rutherford, E. (2004). Information-processing approaches: Assessing the selective functioning of attention, interpretation, and retrieval. In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder (pp. 109–142). New York: The Guilford Press. Marten, P. A., Brown, T. A., Barlow, D. H., Borkovec, T. D., Shear, M. K., & Lydiard, R. B. (1994). Evaluation of the ratings comprising the associated symptom criterion of DSM-III-R generalized anxiety disorder. Journal of Nervous and Mental Disease, 181, 676–682. Mathew, S. J., Mao, X., Coplan, J. D., Smith, E. L. P., Sackeim, H. A., Gorman, J. M., et al. (2004). Dorsolateral prefrontal cortical pathology in generalized anxiety disorder: A proton magnetic resonance spectroscopic imaging study. American Journal of Psychiatry, 161, 1119–1121. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validity of the Penn State Worry Scale. Behaviour Research and Therapy, 28, 487–495. Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167–202. Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational analysis of anxiety. Behaviour Research and Therapy, 36, 809–848. Mowrer, O. H. (1947). On the dual nature of learning: A re-interpretation of ‘‘conditioning’’ and ‘‘problem-solving’’. Harvard Educational Review, 17, 102–148. Peasley-Miklus, C., & Vrana, S. R. (2000). Effect of worrisome and relaxing thinking on fearful emotional processing. Behaviour Research and Therapy, 38, 129–144. Siegle, G. J., Steinhauer, S. R., Thase, M. E., Stenger, V. A., & Carter, C. S. (2002). Can’t shake that feeling: fMRI assessment of sustained amygdala activity in response to emotional information in depressed individuals. Biological Psychiatry, 51, 693–707. Siegle, G. J., Thompson, W., Carter, C. S., Steinhauer, S. R., & Thase, M. E. (2007). Increased amygdala and decreased dorsolateral prefrontal BOLD responses in unipolar depression: Related and independent features. Biological Psychiatry, 61, 198–209. Smith, E. E., & Jonides, J. (1999). Storage and executive processes in the frontal lobes. Science, 283, 1657–1661. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. (1983). State-trait anxiety inventory test manual for form Y. Palo Alto, CA: Consulting Psychological Press. Stanley, M. A., Beck, J. G., Novy, D. M., Averill, P. M., Swann, A. C., Diefenbach, G. J., et al. (2003). Cognitive behavioral treatment of late-life generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 71, 309–319. Stanley, M. A., Diefenbach, G. J., Hopko, D. R., Novy, D., Kunik, M. E., Wilson, N., et al. (2003). The nature of generalized anxiety in older primary care patients: Preliminary findings. Journal of Psychopathology and Behavioral Assessment, 25, 273–280. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–662.

ARTICLE IN PRESS R.B. Price, J. Mohlman / Behaviour Research and Therapy 45 (2007) 2628–2639

2639

Trenerry, M. R., Crosson, B., DeBoe, J., & Leber, W. R. (1989). Stroop neuropsychological screening test manual. Odessa, FL: Psychological Assessment Resources, Inc. Van den Berg, M. D., Oldehinkel, A. J., Brilman, E. I., Bouhuys, A. L., & Ormel, J. (2000). Correlates of symptomatic, minor and major depression in the elderly. Journal of Affective Disorders, 60, 87–95. Vrana, S. R., Cuthbert, B. N., & Lang, P. J. (1986). Fear imagery and text processing. Psychophysiology, 23, 247–253. Wechsler, D. (1997). Wechsler adult intelligence scale and Wechsler memory scale technical manual (3rd ed.). San Antonio, TX: The Psychological Corporation. West, R. L. (1996). An application of prefrontal cortex function theory to cognitive aging. Psychological Bulletin, 120, 272–292. Wu, J. C., Buchsbaum, M. S., Hershey, T. G., Hazlett, E., Sicotte, N., & Johnson, J. C. (1991). PET in generalized anxiety disorder. Biological Psychiatry, 29, 1181–1199.