«ATTENTIONAL BIAS IN PATIENTS WITH IMPLANTABLE CARDIOVERTER DEFIBRILLATORS: EXAMINING MECHANISMS OF HYPERVIGILENCE AND ANXIETY By NEHA K. DIXIT A ...»
Cue Word Valence and Arousal Ratings To determine if the words selected for the emotional manipulation in the dot probe task resulted in differential valence and arousal ratings within and between groups, post-task SAM assessment valence and arousal ratings were analyzed using separate ANOVAs; with group as a between-subjects factor and word category (Pleasant, Neutral, Unpleasant, Clinical) as a withinsubject factor.
Results indicate that the words were, indeed, effective in producing the desired effects: (1) Both arousal and valence ratings for the Pleasant, Neutral, and Unpleasant words were consistent with expectations, (2) the two groups did not differ in ratings of these standard words, (3) but VF patients rated the clinically-relevant words as significantly more unpleasant than AF controls.
Cue Word Valence Ratings Analyses of valence ratings demonstrated a main effect of valence, F(3,110) = 734.58, p0.001, p.001, η2 =.99. This main effect was qualified by a significant Group x Valence interaction F(3,110) = 11.32, p0.001, η2 =.92. Follow-up independent samples t-test revealed that ICD patients rated the clinically-relevant words as significantly more unpleasant than AF patients, t(76) = 3.771, p0.001; yet they did not differ in their rating of the other word categories (Figure 3-3).
Cue Word Arousal Ratings Analyses of arousal ratings revealed a significant main effect of word category, F(3, 110) = 274.29, p0.001, η2 =.87. In general both groups rated pleasant, neutral and clinical words as more arousing than neutral words. No group by valence interaction was seen for arousal F(3, 110) = 3.07, p.08, η2 =.12 suggesting that ICD and AF patients found the words, equally arousing despite a priori predictions (Figure 3-4).
Shock and Reaction Time To examine the unique effects of shock on task performance, ICD recipients were grouped into “shock” and “no shock” categories. Fourteen ICD recipients received one or more shocks and 21 recipients had no history of shock. There were no significant effects of presence or absence of shock with respect to RTs, F(1,36) = 2.15, p.53, η2 =.08. Additionally, no significant effect of valence F(3,102) =.774, p.49, η2 =.13 or Shock x Valence interaction was found F(3,102) = 2.15, p.09, η2 =.23. Finally, no Shock x Congruency x Valence interaction was found as hypothesized for RTs, F(3,102) = 1.288, p0.33, η2 =.07 (Figure 3-5).
Shock and Error Rates There were no significant effects of presence or absence of shock on error rates, F(1,36)=.57, p.45, η2 =.08. Additionally, no significant effect of valence F(3,102) =.1.037, p.35, η2 =.02 or Shock x Valence interaction was found F(3,102) =.90, p.96, η2 =.05.
Finally, no Shock x Congruency x Valence interaction was found for errors, F(3,102) = 1.04, p0.37, η2 =.02 (Figure 3-6).
To further examine the impact of ICD shock on psychosocial measures of state, trait, and shock related anxiety, additional ANOVAs were run using only ICD shock as the group variable.
No significant differences between groups were found for state anxiety, F(1,36)=.87, p.35, η2 =.12, trait anxiety, F(1,36)=.28, p.59, η2 =.15, and shock anxiety (FSAS), F(1,36)=.68, p.41, η2 =.20.
General attentional bias scores were calculated to qualitatively examine trends in the data by subtracting incongruent RT trials from congruent RT trials (Figure 3-7). While no significant differences emerged between groups (as discussed above); there was a pattern of decreased bias to unpleasant and clinically-relevant stimuli compared to pleasant and neutral stimuli.
To examine the hypothesis that magnitude of bias was positively correlated with both ICD and AF patients’ anxiety levels, Pearson correlations were calculated between bias scores and psychosocial measures of state, and trait anxiety. No significant relationships emerged amongst bias scores and anxiety all ps.05 Psychosocial Data Psychosocial measures described in the methods were used to characterize the sample.
ICD patients and AF controls were similar in their levels of psychological distress, such that they endorsed low levels of depression; similar levels of state anxiety, cardiac specific quality of life and general mental health (Table 2-1). Notably, ICD patients demonstrated significantly more trait anxiety than AF controls t(78) = 1.29, p.04, as well as worse physical health than AF controls t(78) =. -2.321, p.03. Additionally, when examining ICD patients alone, female recipients endorsed a greater level of shock related anxiety than male patients.
Pearson correlations were calculated amongst psychosocial measures of anxiety. As expected, state anxiety was positively correlated with trait anxiety r(78) =.709, p.01, suggesting, that AF and ICD patients who experience greater levels of moment to moment anxiety are more likely to be generally anxious individuals. Additionally, because this study highlighted participants’ experience of living with an ICD, state anxiety was positively correlated with shock related anxiety r(37) =.412, p.01.
-29 Figure 3-7. Average attentional bias scores (incongruent-congruent). Negative scores indicate decreased bias.
Our study is the first to examine the relationships between disease-related anxiety and attentional processing in patients living with ICDs. While many studies have examined ICD and AF patient functioning from a biomedical perspective, evaluating symptoms, quality of life outcomes, and medical outcomes (Sears, 2004; Sears & Conti 2005; Godemann 2004.), this is the first examination of potential attentional bias in this patient population. Four main findings emerged from the research: First, both ICD and AF patients showed a significant attentional bias towards congruent information, irrespective of emotional valence of the task word cue. Second and unexpectedly, ICD patients committed more errors overall than AF patients. Third, ICD patients demonstrated worse general physical functioning and greater trait anxiety as a group compared to AF patients. Contrary to predictions, however, the presence or absence of ICD shock did not distinguish this difference. Finally, ICD patients found the clinically-relevant cue words more unpleasant than did AF patients.
A primary aim of this study was to examine the ICD patients’ ability to disengage attention from shock-related (cardiac specific) information using a modified dot-probe paradigm;
that is, to determine if they exhibit evidence of a specific attentional bias towards clinicallyrelevant stimuli. The hypothesis was that ICD patients would be slower to respond to incongruent clinical trials compared to AF patients. Results demonstrated that regardless of group (ICD or AF control) or emotional valence of the words presented in the task, both groups were slower to respond to a probe presented at a congruent than incongruent location, suggesting a “reverse” congruency effect. This finding is novel and generally inconsistent with the literature on dot probe and Stroop tasks which consistently demonstrate that subjects have longer reaction times under incongruent than congruent conditions. The literature also demonstrates that emotional valence of the cue (e.g., word, picture, face) differentially affects a person’s ability to respond to corresponding probes such that incongruent trials with affectively arousing words, give rise to disproportionately slower responses.
Although the finding that participants were slower to respond to a probe presented directly after an emotional word is surprising, past research has previously suggested similar inconsistencies, particularly in older populations. Fox and colleagues (2005) demonstrated that older adults with high levels of state anxiety did not demonstrate a classical interference effect on an emotional Stroop task. In fact, their findings demonstrated that older adults have a tendency to disregard the threat content of task relevant information and perform similarly on both congruent and incongruent trials. The findings from this study and another by Mather and Carstensen (2003) also suggest that older adults with low levels of anxiety tend to avoid attending to negative emotional material. One explanation of the findings in this study is that age may have factored into the response style of participants. Our participants did not endorse clinically significant levels of state anxiety and were older than most cohorts in the classical dotprobe literature. As such, they may have disregarded the emotional content of the words during the task, focusing more on the directions asking them to respond to the probe.
The “reverse” congruency effect in this task may be better understood through paradigms in the literature examining attentional engagement. Posner and colleagues (e.g. Posner, Cohen & Rafel, 1982) demonstrated the concept of attentional engagement vs. disengagement to spatial location using a cued target paradigm. In this task, participants were instructed to focus on a fixation point between two rectangles, wait for a cue (the brightening of one rectangle) and respond to congruent or incongruent probes presented in one of the two rectangles. Participants in Posner’s experiment were faster to respond to congruent than incongruent trials, suggesting an effect of cue dependency.
In the paradigm used in this study, it is possible that cue word reading prior to a congruent probe response resulted in facilitated attention to that spatial location such that participants attended to the word and responded to the probe in an “automatic” manner without paying conscious attention to the probe itself. Additionally, directing gaze towards a probe in the opposite location of the cue word (incongruent probe) may have required heightened disengagement of attentional resources (to make a correct probe response) such that participants could not dwell on the emotional content of the word presented. Finally, the magnitude of attentional bias in this study was in the opposite direction demonstrating decreased bias towards clinically-relevant and unpleasant words compared to neutral and pleasant words. This suggests that participants spent less time attending to the clinical words compared to the non-clinical words. While the result was not statistically significant, it may provide further evidence for the “reverse” congruency effect.
Contrary to expectations there was no effect of valence on task performance. A potential explanation for this may be that participants were experiencing heightened arousal during the entire task and may have allocated all possible resources to performing that task correctly at the expense of attending to the valence of the words. Participants may have been over aroused throughout the task such that valence specific effects were washed out. Direct measurements of autonomic arousal (i.e., skin conductance) (Lang et al. 2000) may have been useful in gaining information about physiological arousal in participants.
Another notable finding in this study was that ICD patients exhibited greater levels of trait anxiety and worse self-reported physical functioning compared to AF controls. It is plausible that this increased generalized anxiety led ICD patients to make more errors in the dotprobe task compared to AF controls. Further support for this hypothesis is provided by the fact that there was no speed-accuracy trade off in either group. ICD patients’ greater trait anxiety may also help to explain why no main effects of valence were found. Mogg and Bradley (1998) proposed that anxiety is the locus of the individual differences in a person’s threat appraisal mechanism. These researchers posit that high trait-anxious individuals appraise all levels of threat as subjectively greater than low trait-anxious individuals. As such, ICD patients in this sample may have reached a high level of “threat vigilance” by viewing personally-threatening (clinical) words and subsequently appraised other valences as more unpleasant across the trials, suggesting the presence of a “carryover effect.”
Another consideration given the general results of this study is the methodology employed to elicit evidence of emotional attentional bias. The traditional dot-probe task developed by McLeod and colleagues (1997) presented word pairs such that spatial attention during the “cue” phase of the task was captured across the screen with a different set of stimuli each time (e.g. a neutral word always accompanied an emotional arousing word). On the traditional task, trials can be examined to delineate preferences towards and away from threat related and neutral word pairs. The modified dot-probe task in this study was developed based on findings by Amir and colleagues (2004), who did not employ the use of word pairs when examining attentional bias in social phobics. Thus, it is possible that the task employed in this study may not have been robust enough to elicit an attentional bias at the magnitude of other dotprobe tasks. Another methodological difference that potentially influenced the results in this study is the lack of a probability manipulation (Amir et al., 2004, Fox et al., 2000). Other researchers have used a disproportionately larger set of congruent trials verses incongruent trials, priming participants’ response patterns and increasing the magnitude of bias observed.
Finally, participants’ ratings of the valence and arousal of the cue words, demonstrate a positive manipulation check. The ratings showed a disproportionate affective response to clinically-relevant stimuli in ICD patients, who rated the clinically-relevant (cardiac) words as more unpleasant compared to AF controls. One explanation of this finding is that the personal relevance of the clinical words was greater for the ICD patients than the AF patients. An example of a clinical word used in the task is “shock.” This word has clinical meaning for both groups. For ICD patients shock may refer to their device going off, and be a reminder to them that they have a life threatening arrhythmia. For AF patients shock may refer to an external defibrillation treatment that helps to control their symptoms, potentially alleviating them. While the word is relevant to both groups, it has the potential to connote greater negativity for ICD patients than AF patients.