تعدیل هراس در واکنش به چالش CO2 35٪ با کنترل توجه

Abstract Attentional control (AC) is an individual difference variable indexing the ability to voluntarily focus attention and shift attention when desired. AC is thought to impact the experience of fear by facilitating the disengagement of attention from threat and promoting the deployment of attentional resources toward regulatory or coping strategies. Whereas previous research has focused on visual threat cues, in the current study we examined whether this model also applies to interoceptive threat by evaluating the extent to which individual differences in AC moderated the relationship between trait anxiety and self-reported fear in response to a single vital capacity inhalation of a 35% CO2, 65% balanced O2 gas mixture. The sample comprised a large nonclinical group of young adults (N = 128). Results indicated that AC moderated the relationship between trait anxiety and fearful responding to the challenge. Findings suggest that AC plays a significant and clinically important role in modulating self-reported fear.

Results Manipulation Check for the CO2 Challenge To ascertain whether the challenge produced a fear response, a repeated measures ANOVA was used to examine whether there was significant change in challenge anxiety outcome variables from baseline to postchallenge. Consistent with expectation and prior work, a single VC inhalation of 35% CO2 produced significant increases in SUDS, F(1, 119) = 93.44, p < .001, η2 = .65, and API total, F(1, 119) = 331.0, p < .001, η2 = .73. Zero-Order Correlations Among AC, Trait Anxiety, and CO2 Reactivity Table 1 displays zero-order correlations for the main predictor and outcome variables of interest. AC was significantly negatively correlated with STAI as predicted (r = −.47, p < .001), and negatively correlated with API at all time points (r range = −.20 to –.27; all p < .05) but not SUDS (r range = −.16 to –.18; all p ≤ .15). As would be expected, STAI was correlated with API (r range = .27 to .48; all p < .001) and SUDS at all time points (r range = .30 to .38; all p < .001). ACS Subscale 1 (attentional focus) was more strongly related to API and SUDS at all time points (all ps < .047) than Subscale 2 (attentional switching; all ps > . 07; see Table 1). Table 1. Means, Standard Deviations, and Bivariate Correlations for Predictors and Dependent Variables Mean SD σ2 1 1a 1b 2 3a 3b 3c 3d 4a 4b 4c 1. ACS 51.9 8.48 72.05 -- a. Subscale 1 22.2 5.00 24.63 .91** -- b. Subscale 2 29.7 4.47 20.06 .89** .62** -- 2. STAI (Trait) 41.1 11.17 124.88 –.47** –.42** –.43** -- 3. API a. Baseline 2.04 2.62 6.88 –.27** –.32* –.17 .48** -- b. Prechallenge 1.53 2.17 4.73 –.20** –.23* –.12 .40** .83** -- c. Challenge 11.77 6.73 45.4 –.21** –.19* –.09 .27** .41** .41** -- d. Recovery 2.01 2.83 8.02 –.20* –.21* –.15 .40** .63** .68** .57** -- 4. SUDS a. Baseline 12.33 16.64 277.04 –.18 –.20* –.11 .33** .60** .50** .32** .34** -- b. Prechallenge 10.49 15.25 232.63 –.16 –.18* –.10 .30** .51** .51** .29** .32** .96** -- c. Challenge 35.87 16.33 693.56 –.17 –.23* –.05 .31** .42** .36** .66** .46** .55** .51** -- d. Recovery 12.37 16.50 272.51 –.18 –.20* –.11 .38** .53** .53** .32** .49** .83** .85** .56** Note. σ2 = variance; ACS = Attentional Control Scale; STAI = State-Trait Anxiety Inventory (Trait Scale); API = Acute Panic Inventory, Total Symptom Scale; SUDS = Subjective Units of Distress. * Correlation is significant at the .05 level (2-tailed), ** Correlation is significant at the .01 level (2-tailed). Table options Hierarchical Multiple Regression Analyses The interactive relationship between trait anxiety and AC was evaluated in relation to the primary dependent variables of SUDS and API symptoms immediately postchallenge and at recovery (5 minutes subsequent to the challenge). To do this, we used a hierarchical multiple regression procedure (Cohen & Cohen, 1983). Separate models were constructed for predicting SUDS and API immediately postchallenge, and also at a recovery period, 5 minutes subsequent to the challenge. This resulted in a total of four equations, corresponding to the two dependent variables (SUDS and API) at both postchallenge and recovery. We followed the procedure outlined by Baron and Kenny (1986) for determining moderational effects. In all regression models reported below, we controlled for the autoregressor (baseline SUDS or API) at Step 1 in the model, in order to evaluate whether the predictor variables accounted for the dependent variables above and beyond the effects of baseline trait anxiety. At Step 2 in the model, the main effects for STAI total score and ACS were simultaneously entered, in order to determine the individual main effects of these variables. At Step 3, the interaction term between STAI and ACS was entered into the model. Note that the prechallenge measurement point (prior to the 10-minute adaptation period) was not used as the autoregressor in our models, in order to be consistent with previous studies utilizing CO2 challenge paradigms, in which the baseline period (after adaptation) is most commonly used as an autoregressor (Richey et al., 2010 and Schmidt and Zvolensky, 2007). Results of the four regression equations are presented in Table 2 and Table 3. Means and standard deviations for dependent variables are presented in Table 4, separated by high and low ACS and STAI on the basis of a median split. With respect to trait anxiety, there was no main effect for trait anxiety on SUDS immediately postchallenge, however, a main effect for trait anxiety on SUDS was observed at recovery (β = .18, p < .01, t = 3.03) but no main effect for trait anxiety on API was observed at either time point. With respect to ACS we did not observe a simple main effect for this predictor on either dependent variable at postchallenge or recovery. Regarding the predicted relationship between STAI and AC, there was a significant interaction at the third level of the model, for both API (ΔR2 = .06, t = 2.94, p = .004) and SUDS (ΔR2 = .05, t = 3.02, p = .003) at postchallenge, but not at recovery (API: ΔR2 = .01, t = 0.56, p = .572; SUDS: ΔR2 = .01, t = 1.27, p = .206) indicating that AC moderated the relationship between trait anxiety and distress immediately subsequent to the CO2 inhalation, but not 5 minutes after the challenge. 1 Table 2. Regression Results Predicting SUDS and API at Postchallenge Predicted Variable Predictors in Set ΔR2 t for Each Predictor β p Challenge SUDS Step 1 .26 Baseline SUDS 6.32 .51 <.001 Step 2 .02, .03a, .03b STAI-Trait 1.77 .17 .079 ACS Total 0.11 .01 .909 ACS Subscale 1a − 0.94 –.08 .351 ACS Subscale 2b 2.20 .11 .217 Step 3 .05, .07a, .03b ACS Total × STAI 3.02 .24 .003 ACS Subscale 1 × STAIa 3.06 .89. <.001 ACS Subscale 2 × STAIb 2.30 .86 <.05 Challenge API Step 1 .17 Baseline API 4.91 .41 <.001 Step 2 .02, .02a, .02b STAI-Trait 1.10 .12 .27 ACS Total − 0.39 –.04 .697 ACS Subscale 1a 1.59 .11 .494 ACS Subscale 2b 0.44 .01 .965 Step 3 .06, .04a, .04b ACS Total × STAI 2.94 .24 .004 ACS Subscale 1 × STAIa 2.29 .89 <.05 ACS Subscale 2 × STAIb 2.19 .87 <.05 a Subscale analyses for attentional focus (Subscale 1), b Subscale analyses for attentional switching (Subscale 2). To conserve space, subscale analyses are displayed in the context of regression analyses for ACS total scale score above; however, separate regression equations were constructed for each subscale analysis, utilizing the residualized scores at Step 2 and centered interaction terms at Step 3 in the model, after controlling for the relevant autoregressor at Step 1. Table options Table 3. Regression Results Predicting SUDS and API at Recovery Predicted Variable Predictors in Set ΔR2 t for Each Predictor β p Recovery SUDS Step 1 .72 Baseline SUDS 16.97 .85 <.001 Step 2 .02, .02a, .02b STAI-Trait 3.03 .18 .006 ACS Total 1.51 .08 .148 ACS Subscale 1a 0.95 .05 .342 ACS Subscale 2b 1.58 .09 .115 Step 3 .01, .01a, .01b ACS Total × STAI 1.27 .06 .206 ACS Subscale 1 × STAIa − 1.00 –.17 .164 ACS Subscale 2 × STAIb 1.22 .28 .226 Recovery API Step 1 .48 Baseline API 10.31 .69 <.001 Step 2 .02, .02a, .02b STAI-Trait 1.78 .15 .077 ACS Total − 0.19 –.15 .849 ACS Subscale 1a − 0.29 –.03 .953 ACS Subscale 2b 0.24 .03 .778 Step 3 .01, .00a, .01b ACS Total × STAI 0.56 .04 .572 ACS Subscale 1 × STAIa 0.67 .22 .355 ACS Subscale 2 X STAIb 0.12 .05 .915 a Subscale analyses for attentional focus (Subscale 1), b Subscale analyses for attentional switching (Subscale 2). To conserve space, subscale analyses are displayed in the context of regression analyses for ACS total scale score above; however, separate regression equations were constructed for each subscale analysis, utilizing the residualized scores at Step 2 and centered interaction terms at Step 3 in the model, after controlling for the relevant autoregressor at Step 1. Table options Table 4. Means and Standard Deviations for SUDS and API by Time and Group High ACS (N = 54; 62% female) Low ACS (N = 59; 56% female) Prechallenge Baseline Postchallenge Recovery Prechallenge Baseline Postchallenge Recovery SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API Mean 9.31 1.50 8.14 1.27 31.66 10.62 9.47 1.44 14.01 2.55 11.66 1.84 39.37 13.32 13.19 2.64 SD 2.93 1.95 13.25 2.07 23.18 5.70 14.69 2.01 18.79 3.13 16.03 2.34 23.19 7.39 7.07 3.39 SE 1.76 0.26 1.78 0.28 3.15 0.77 2.01 0.27 2.44 0.40 2.08 0.30 3.72 0.96 2.24 0.44 High STAI (N = 55; 52% female) Low STAI (N = 58; 48% female) Prechallenge Baseline Postchallenge Recovery Prechallenge Baseline Postchallenge Recovery SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API SUDS API Mean 17.22 3.01 14.62 2.25 43.86 13.11 17.6 2.96 7.31 1.11 6.01 0.83 27.9 10.10 6.81 1.08 SD 18.91 3.25 17.38 2.68 26.67 6.90 18.83 3.46 12.84 1.41 11.72 1.26 24.54 6.05 1.97 1.62 SE 2.48 0.42 2.26 0.34 3.47 0.89 2.49 0.45 1.67 0.18 1.52 0.16 3.21 0.79 1.57 0.21 Table options Significant interactions were examined graphically (Cohen & Cohen, 1983) and analytically (Holmbeck, 2002) to determine their form and significance (see Figure 1 and Figure 2). To analytically probe the interactions, we examined the simple slopes at various levels of ACS. Specifically, we evaluated the relationship between STAI and the two dependent variables (SUDS, API) as a function of ACS score (above and below the median). Consistent with prediction, we found a positive relationship at postchallenge between STAI and SUDS for individuals low in AC (N = 59, 56% female; β = .42, p < .001, t = 3.37), indicating that at lower levels of AC, there is a significant positive relationship between trait anxiety and fear in response to the challenge. For individuals high in AC (N = 54, 62% female), a robust inverse relationship was observed (β = −.43, p < .001, t = 3.25), indicating that individuals high in AC reported significantly less fear, even when also endorsing high levels of trait anxiety. For API, we found a similar pattern of results, although when using STAI to predict API the simple slopes for low (β = .04, p = 79, t = .26) and high levels of ACS (β = −.15, p = .56, t = .41) were individually nonsignificant (see Figure 2). We did not probe interactions at 5 minutes subsequent to the challenge, because there was no significant moderational effect of ACS for either dependent measure (API, SUDS). We also assessed whether these subgroups differed in terms of IV. To do this, we divided the residual volume (in liters) and divided by the participant's total VC and multiplied this figure by 100 to determine the percentage of total VC inhaled. We then compared the groups (high vs. low ACS) in a between-groups ANOVA, and found no difference in the deviation from total VC inhaled, F(1, 111) = .93, p = .42. Interaction of ACS and STAI predicting SUDS during CO2 challenge. Note. Residual ... Figure 1. Interaction of ACS and STAI predicting SUDS during CO2 challenge. Note. Residual scores for both SUDS and STAI were utilized in hierarchical regression analyses and are therefore depicted as such on X and Y axes. Figure options Interaction of ACS and STAI predicting API symptoms during CO2 challenge. Note. ... Figure 2. Interaction of ACS and STAI predicting API symptoms during CO2 challenge. Note. Residual scores for both API and STAI were utilized in hierarchical regression analyses and are therefore depicted as such on X and Y axes. Figure options ACS Subscale Analysis The ACS can be decomposed into two distinct subscales reflecting attentional focus (Subscale 1) and attentional switching (Subscale 2). We analyzed both subscales in a hierarchical multiple regression framework as outlined above, substituting in either the first or second subscale in place of the total scale score at Step 2, while controlling for the autoregressor at Step 1. Centered interaction terms (STAI × Subscale) were created for each subscale score for use at Step 3 in the model. We found that Subscale 1 interacted significantly with trait anxiety at Step 3 in the model to predict SUDS (ΔR2 = .07, β = .89, p < .001, t = 3.06), as did Subscale 2 (ΔR2 = .03, β = .86, p < .05, t = 2.30). This was also true when evaluating API as the primary dependent variable using the subscales at Step 3 (Subscale 1: ΔR2 = .04, β = .89, p < .05, t = 2.29; Subscale 2: ΔR2 = .04, β = .87, p < .05, t = 2.19). As expected, neither subscale interacted with STAI to predict SUDS or API at recovery (see Table 2 and Table 3). Exploration of Differential Findings Due to Sex Because previous research has identified that males and females differ in their autonomic and evaluative responses to panicogenic inhalations of CO2-enriched gas (Kelly et al., 2006 and Monkul et al., 2010), we also explored whether our results differed as a function of sex. We first tested whether males and females differed in baseline indices of AC and trait anxiety. A between-groups ANOVA (male vs. female) revealed no differences between males and females in either ACS total score, F(1, 119) = 1.9, p = .20, or STAI total score, F(1, 119) = .84.1, p = .36. Similarly, in a series of between-groups ANOVAs, males and females did not differ in terms of API total symptoms or SUDS at any time point (all ps > .28). Adding sex as an additional covariate in the multiple regression models revealed that the difference in the responses of males and females to the CO2 challenge was not a result of differences between the sexes (i.e., the significance of the observed relationships did not change in any level of any hierarchical regression model probing ACS full-scale and subscale analyses when using sex as a covariate).