حافظه شرح حال دچار اختلال شده در سوگ و داغداری پیچیده

Abstract This study investigated the relationship between autobiographical memory and goals in complicated grief (CG). Twenty-four individuals with CG and 21 bereaved individuals without CG completed an autobiographical memory task and a personal goals task. CG participants were more likely to recall memories related to their loss, particularly in response to negative cues. There was a positive relationship between the proportion of loss-related memories recalled and the proportion of grief-related goals held by individuals after controlling for symptom level. Individuals with CG also showed impaired retrieval of specific autobiographical memories in response to both positive and negative cues. These results suggest that CG is characterised by impaired and biased retrieval of specific autobiographical memories. These patterns are consistent with propositions stemming from the self-memory models of autobiographical memory.

Results Participant characteristics Table 1 indicates that participants in the two groups did not differ in terms of age, time since the death, or relationship to the deceased. As expected, participants in the CG group had significantly higher scores than participants in the No-CG group on the CGA [t(28.62) = −24.17, p < .001], BDI-II [t(27.10) = −10.38, p < .001], and CAPS [t(23.67) = −13.06, p < .001]. In the CG sample, 9 (38%) met criteria for PTSD on the CAPS-2 and 19 (79%) met criteria for major depression on the SCID. The groups performed equally well on the measures of working memory and verbal intelligence. Table 1. Participant characteristics. Complicated grief n = 24 No-complicated grief n = 21 Mean age 50.67 (13.49) [range 22–79] 50.76 (15.97) [range 23–80] CGA interview 33.25 (4.39) 10.29 (1.45) Relationship of the deceased Partner 42% 43% Child 21% 14% Parent 29% 33% Sibling 8% 10% Years since death 3.83 (2.88) 3.32 (1.96) Sudden deatha 61% 38% BDI-II 31.25 (11.56) 5.0 (3.57) CAPS-2 54.92 (19.45) 2.67 (2.19) LNS 10.42 (1.50) 10.91 (2.66) NART 33.61 (7.45) 34.42 (5.48) Grief goals (proportion) .2 (.16) .05 (.10) Note: Standard deviations appear in parentheses. CGA = Complicated Grief Assessment; BDI-II = Beck Depression Inventory, Second Edition; CAPS-2 = Clinical Administered PTSD Scale, Version 2; LNS = Letter Number Sequence Task; NART = National Adult Reading Test. a Sudden death refers to a death that occurred with less than one week's notice. Table options Memory content Table 2 presents the mean proportion of memories coded into each of the three content categories (Loss, Person, Other). A 2 (Participant Group) × 3 (Content Category) × 2 (Cue Valence) mixed model analysis of variance (ANOVA) showed a significant main effect for Content (F(2, 86) = 147.05, p < .001), significant 2-way interactions between Group and Content (F(2, 86) = 8.31, p < .002), and Valence and Content (F(1, 43) = 7.20, p < .002), and a significant 3-way interaction between Group, Content and Valence (F(2, 86) = 6.65, p < .001). To clarify these effects, separate 2 (Participant Group) × 3 (Content Category) ANOVAs were conducted for positive and negative cues. These analyses indicated that for positive cues there was a significant main effect for Content [F(1,43) = 94.57, p < .001], with participants recalling more Other memories than Loss or Person memories. For negative cues there was a significant main effect for Content [F(1,43) = 49.06, p < .001], and a significant Content × Group interaction [F(1,43) = 18.71, p < .001]. Participants recalled more Other memories in response to negative cue words, however the CG group were more likely to recall a Loss memory in response to a negative cue than the No-CG group [t(43) = −4.46, p < .001]. Table 2. Mean proportion of loss, person and other memories. Loss Person Other Positive Negative Positive Negative Positive Negative Complicated grief .13 (.15) .39 (.27) .15 (.21) .05 (.10) .72 (.23) .56 (.24) No-complicated grief .07 (.10) .10 (.13) .18 (.21) .07 (.13) .75 (.22) .83 (.22) Note: Standard deviations appear in parentheses. Table options Relationship between memory content and goals Table 1 displays the mean proportion of grief-related goals provided by participants. CG participants provided a greater proportion of grief-related goals than the No-CG group, t(42) = −3.66, p < .01. To explore the relationship between memory content and goals, correlation coefficients were calculated between grief goals, CGA scores, and the total proportion of Loss memories recalled by participants (with an adjusted alpha of p = .016). This analysis indicated that there was a significant correlation between the proportion of Loss memories and the proportion of grief-related goals (r = .48, p < .002). To examine the extent to which this relationship existed independent of symptomatology, a hierarchical multiple regression was conducted. CGA, BDI-II and PTSD scores were entered on Steps 1, 2 and 3 respectively, and the proportion of grief-related goals was entered on Step 4. The results of the final step of this analysis are presented in Table 3. Overall, this equation was significant, F(5,44) = 8.58, p < .001. CGA scores accounted for 32% of variance in the proportion of Loss memories. PTSD and depression did not explain any additional variance in this data. On Step 4 grief-related goals accounted for an additional 6% of the variance in the proportion of Loss memories (p < .04). 1 That is, goals remained a significant predictor of memory content after controlling for symptoms. We now turn to examine the relationship between CG and memory specificity. Table 3. Summary of hierarchical regression models for memory content. Variable B SE B β Loss memories CGA −.00 .00 −.21 BDI .01 .00 .55 PTSD .111 .06 .28 Grief goals .34 .16 .33 Note: CGA = Complicated Grief Assessment, BDI = Beck Depression inventory – Version 2, PTSD = PSTD diagnosis on the CAPS-2. Step 1 R2 = .34, ΔR2 = .34; Step 2 R2 = .35, ΔR2 = .01; Step 3 R2 = .40, ΔR2 = .05; Step 4 R2 = .46, ΔR2 = .06. Table options Memory recall specificity Table 4 presents the mean number of specific, categoric, and extended memories recalled by participants. A 2 (Participant Group) × 2 (Cue Valence) ANOVA on specific memories indicated a main effect for Participant Group (F(1, 43) = 38.75, p < .001); CG participants retrieved fewer specific memories than no-CG participants. A 2 (Participant Group) × 2 (Cue Valence) ANOVA of categoric memories indicated a significant main effect for Participant Group (F(1,43) = 37.72, p < .001); CG participants retrieved more categoric memories than No-CG participants. A 2 (Participant Group) × 2 (Cue Valence) ANOVA of extended memories indicated no significant effects. Table 4. Mean number of specific and general memories recalled. Specific Categoric Extended Positive Negative Positive Negative Positive negative Complicated grief 2.33 (1.46) 2.46 (1.44) 1.67 (1.34) 1.58 (1.21) .29 (.50) .42 (.55) No-complicated grief 4.78 (.75) 4.19 (.93 .14 (.35) .29 (.64) .10 (.30) .14 (.36) Note: Standard deviations appear in parentheses. Table options To explore the relative contributions of depression, PTSD and CG symptoms to this pattern of results, two hierarchical multiple regression analyses were conducted. The first analysis predicted number of specific memories recalled, and the second analysis predicted the number of categoric memories recalled. In both analyses, depression, PTSD, and other variables predicted by the CARFAX model (Williams, 2006) to contribute to overgeneral memory, were entered into the analysis on separate steps, prior to the inclusion of CG symptom scores. Specifically, BDI-II scores were entered at Step 1, PTSD diagnostic status was entered on Step 2. Measures of executive functioning (the WAIS Letter Number Sequencing Task score) and rumination (RRS Brooding subscale) were entered on Steps 3 and 4 respectively, and at Step 5 CGA scores were entered. Due to the absence of any differences in recall specificity according to cue valence, responses for positive and negative cue words were combined. Initial analyses were conducted using dimensional CAPS-2 scores rather than PTSD diagnostic status, however due to the strong correlations between CAPS-2 and CGA scores, CAPS-2 scores were replaced with PTSD status (see Field, 2005 and Tabacknick and Fidell, 2001). The change in predictors reduced indicators of multicollinearity but did not impact the overall outcome of the analysis with respect to CG. Table 5 presents the final model for both equations. BDI-II scores accounted for 15% of the variance in specific recall. PTSD, LNS and Rumination (brooding) did not significantly predict the variance. When CGA interview scores were added at Step 5, an additional 48% of the variance in specific recall was explained, F(5,39) = 17.65, p < .001. For categoric recall, BDI-II scores accounted for 24% of the variance. PTSD, LNS and Rumination (brooding) were not significant predictors. The inclusion of CGA scores increased the predicted variance in categoric recall by 39%, F(5,39) = 14.95, p < .001. Table 5. Summary of hierarchical regression models for memory specificity. Specific recall Categoric recall Variable B SE B β Variable B SE B β Step 4 Step 4 BDI .13 .04 .73 BDI −.07 .03 –.52 LNS .05 .13 .03 LNS .06 .10 .06 RRSb .10 .11 .16 RRSb −.05 .08 −.12 CGA −.34 .05 −1.43 CGA .22 .04 1.28 Note: BDI-II = Beck Depression inventory – Version 2, LNS = Letter Number Sequencing Task, RRSb = Ruminative Response Style (brooding subscale), CGA = Complicated Grief Assessment. Results for Specificity: Step 1 R2 = .15, ΔR2 = .15; Step 2 R2 = .15, ΔR2 = .00; Step 3 R2 = .16, ΔR2 = .01; Step 4 1 R2 = .64, ΔR2 = .48. Results for Categoric Recall: Step 1 R2 = .24, ΔR2 = .24; Step 2 R2 = .24, ΔR2 = .00; Step 3 R2 = .25, ΔR2 = .01; Step 4 R2 = .63, ΔR2 = .38. Table options Recall specificity and memory content Finally, we examine the relationship between memory content and memory specificity. The mean proportion of memories coded as specific in each content category are presented in Table 6. Due to the low frequency of responses in several categories, positive and negative cues were combined for analysis. Planned between group t-test comparisons (adjusted alpha p < .016) indicated that the proportion of specific memories recalled by the CG group was significantly less than the No-CG group in both the Loss and Other categories [Loss: t(30.33) = 5.35, p < .001, Other: t(31.39) = 5.26, p < .001]. The groups did not differ in proportion of specific Person memories [t(18.63) = −2.16, p < .045]. Table 6. Mean proportion of specific memories according to content category. CG No-CG Positive Negative Positive Negative Loss specific .29 (.45) .41 (.40) 1.0 (.00) .93 (.18) Person specific .41 (.49) .20 (.44) .66 (.50) .91 (.20) Other specific .62 (.37) .60 (.41) .98 (.27) .91 (.17) Note: Standard deviations appear in parentheses. Table options Paired t-tests examining within group differences indicated that in the No-CG group, the proportion of specific memories was significantly greater than the proportion of general memories for all categories [Loss, t(11) = 8.0, p < .001, Person t(11) = 3.43, p < .007, Other t(20) = 17.22, p < .001; adjusted alpha p < .016]. In the CG group, the proportions of specific and general memories did not differ significantly across categories.