咸宁学院本科毕业论文：
relationship between attachment security and depressive symptoms, regression analyses were performed for each attachment security domain (anxious and avoidant attachment),controlling for each social support domain (emotional/informational, tangible, affectionate,positive social interaction, and global social support ,tested separately) as specified by Baron andKenny [59]. Separate analyses were performed for attachment avoidance and for attachment
anxiety .Sobel’s [60] z-test was used to determine whether ornot the decrease in the main effect of attachment security on depressive symptoms was statistically significant, after controlling for the social support domain (each domain was tested separately).
Results
Sample characteristics
A total of 1014 patients with Stage IV (metastatic) GI or Stage IIIA, IIIB, or IV (recurrent or metastatic) lung cancer who attended the ambulatory outpatient clinics at PMH between
November 1, 2002, and January 31, 2006, were considered for participation in this study. Two hundred and eighty-three of these patients were excluded due to language barriers en ? 262T or cognitive impairment en ? 21T: Of the 731 eligible patients (483 GI and 248 lung), 434 (59.4%) consented to participate in the study (283 GI (58.9%) and 151 lung (60.9%)). Of the 434 consenting patients, 17 subsequently died, 80 withdrew from the study prior to returning their questionnaires, and 8 had not returned their questionnaires at the time of the following analyses. Three hundred and twenty-nine participants returned their questionnaires, and 326 participants provided enough data to be included in the analyses (participation rate, 44.6%; completion rate, 75.1%).Patients who refused study participation en ? 297T did not differ from the 326 participants in terms of gender or primary cancer site (GI vs lung) (the only data available for those who refused). The majority of patients who refused, declined to be
approached about the study, and therefore a reason for refusal was not obtained; of those that did provide a reason, most indicated they were not interested, and/or felt too ill or too tired to participate. A limited amount of demographic information was documented at recruitment for the consenting patients who either did not return their questionnaires en ? 8T; returned largely incomplete questionnaires en ? 3T; or subsequently withdrew en ? 80T or died en ? 17T before returning their questionnaires (i.e. 108 non-completers).Comparison of the
non-completers with the 326 participants, using independent samples t-tests and Pearson chi-square tests (controlling for Bonferroni correction adjusting type-1 error rate), indicated that participants had somewhat better functional status as rated on the KPS (81.3_9.1 vs77.3_12.1; t(410)?3.64, p50.001), compared tonon-completers. Participants and
non-completers did not differ in terms of age, gender, marital status (married/common law vs other), education (4high school vs less), living arrangements (alone vs other), country of birth (Canada vs other), primary language (English vs other), primary disease site (GI vs lung), or duration of illness. Characteristics of the sample, including demographic and medical data, attachment security, social support, disease burden, and depressive symptoms are summarized in Table 1. The 326 participants ranged in age from 24 to 88 years and 207 (63.5%) were less than 65 years of age. One hundred and sixty-nine (51.8%) participants were within 12 months of diagnosis at the time of recruitment into the study. Participants reported a median of 11 close friends and 8 close relatives, and the majority reported adequate social support, with only 69 (21.2%) reporting less than adequate global support (i.e. ratings of ‘none’ to ‘some of the time,’ or 570 using transformed MOS-SSS scores, across the four domains of support). Mean differences with respect to depression, attachment security (avoidant or anxious), and social support domains were not detected across demographic indicators, after controlling for type-1 error with a Bonferroni correction of 0.02. The median number of physical symptoms reported on this measure was 6.0 (range 0 – 25), and the median symptom severity score was 2.0 (corresponding to‘moderate’). Commonly reported physical symptoms included: lack of energy (72.4%), feeling drowsy (51.5%), and pain (50.9%). Based on a BDI-II score of 515; a
22
咸宁学院本科毕业论文：
cut-off score shown to have optimal sensitivity and specificity for the diagnosisof major depression in cancer populations [49], 76 (23.3%) participants had significant symptoms of depression. Table 2 presents the univariable association of depressive symptoms with the demographic, medical,attachment, social support, and disease burden variables. The severity of depressive symptoms,as defined by the BDI-II total score, was inversely related to age and all dimensions of social support, and was positively correlated with both attachment anxiety and avoidance. Disease burden indices were significantly positively associated with
depressive symptoms. The degree of attachment anxiety, attachment avoidance, and perceived social support (emotional/informational and global social support) individually contributed to the protection from depressive symptoms. The correlations between depression, social support and attachment security (avoidance and anxiety) areincluded in Table 2.
Buffering the effect of disease burden on depressive symptoms
Regression analyses examining possible buffering effects revealed that although greater perceived social support was associated with decreased depressive symptoms (see Table 2), no significant interaction was found between each social support indicator and each disease burden score (all standardized-beta coefficients 50:01; all p-values 40:05). That is, greater perceived social support (across the four domains of support) was associated with less depressive symptoms regardless of the level of disease burden. Similarly, the degree of attachment avoidance was positively associated with the level of depressive symptoms (see Table 3),but did not significantly interact with any of the five disease burden indicators (all p-values 40:60).This suggests that lower attachment avoidance is associated with less
depressive symptoms regardless of the level of disease burden. Regression analyses revealed that attachment anxiety was positively related to depressive symptoms and significantly interacted with four of the five disease burden indicators (see Table 4). The significant
interactions between attachment anxiety and physical symptom frequency, physical symptom severity, physical symptom distress, and GDI (tested separately; see Table 4) indicated that high attachment anxiety was associated with a steeper increase in depressive symptoms from low to high levels of disease burden (for each of the disease burden indicators) than the increase observed in participants with low attachment anxiety. Figure 1 graphically depicts the observed buffering effect for‘high’ vs ‘low’ attachment anxiety at ‘high’ and‘low’ levels of disease burden (i.e. physical symptom frequency, severity, distress, and GDI) depicted at 1 standard deviation above and below the sample mean [61].
Mediating the relationship of depressive symptoms to attachment security
Attachment avoidance and attachment anxiety were both found to be significantly (all p-values 50:001) associated with depressive symptoms such that depressive symptoms increased an average of 3 points for every one point increase in attachment anxiety and an average of 1.4 points for every one point increase in attachment avoidance (both attachment indices range from 1 to 7). Mediational analysis was conducted to test whether perceived social support (all domains were tested separately) could account for the relationship between attachment security (attachment avoidance and attachment anxiety, tested separately) and depressive symptoms. The following conditions, previously described by Baron and Kenny [59], were examined:(a) a significant association between the independent variable and the mediator; (b) a significant association between the mediator and dependent variable; and (c) the reduction of the association between the independent variable andthe dependent variable after statistically controlling for the mediator. The first condition tests the effect of attachment security (attachment avoidance and attachment anxiety, tested separately) on perceived social support (all domains tested separately).Regression analysis indicated that attachment
avoidance and attachment anxiety were significantly (all p-values 50:001) associated with all domains of social support (Path A in Figure 2),indicating increased attachment insecurity (either avoidant or anxious) results in decreased perceived social support. The second
23
咸宁学院本科毕业论文：
condition tests the effect of perceived social support on depressive symptoms.All domains of social support were found to be significantly (all p-values 50:001) associated with depressive symptoms (Path B in Figure 2),indicating that decreased perceived social support was associated with increased depressive symptoms.The third condition was tested with 10 separate stepwise multiple regression analyses. In each regression model, the depressive symptoms variable was regressed onto attachment security (attachment avoidance or attachment anxiety,tested separately) in the first step, then in the second step a specific domain of social support (emotional/informational, tangible, affectionate,positive social
interaction, and global, all tested separately) was added to the model and the change in model parameters was examined. Table 5 presents the calculated standardized beta coefficients from the mediation analysis of socialsupport, attachment security, and depressive symptoms. Generally, perceived social support more strongly mediated the relationship between attachment avoidance and depressive symptoms relative to its effect on the relationship between attachment anxiety and depressive symptoms. Attachment avoidance was fully mediated by emotional/informationalsupport (z ? 4:54; p50:001), resulting in an 81% effect reduction of attachment avoidance to non-significant level. Attachment anxiety was partially mediated by emotional/informational support (z ? 3:28; p ? 0:001), with a 17% reduction in its effect. Tangible support (z ? 3:13; p ? 0:002) fully mediated the prediction of depressive symptoms by attachment avoidance, reducing the effect by 37% to a non-significant level. Tangible support partially mediated (z ? 2:35; p ? 0:019) the effect of attachment anxiety on depressive symptoms, reducing its effect by10%.Affectionate support fully mediated (z ? 3:44; p50:001) the prediction of depressive symptoms by attachment avoidance, with an effect reduction of 48% to a non-significant level. Affectionate support partially mediated (z ? 2:04; p ? 0:042) the prediction of depressive symptoms by attachment anxiety, reducing its effect by 10%. Positive social interaction support fully mediated (z ? 3:74; p50:001) the relationship between depressive symptoms and attachment avoidance, an effect reduction of 56% to a non-significant level, and partially mediated (z ? 2:67; p ? 0:008) the relationship between depressive symptoms and attachment anxiety, an effect reduction of 12%. Global social support fully mediated (z ? 4:40; p50:001) the effect of attachment avoidance on depressive symptoms, reducing its effect by 75% to a nonsignificant level, and partially mediated (z ? 3:14; p ? 0:002) the effect of attachment anxiety, reducing its effect by 17%. Overall, each domain of social support fully mediated the relationship between depression and attachment avoidance, however, partial mediation was observed between depression and attachment anxiety.
Discussion
This study of 326 patients with metastatic GI or lung cancer demonstrates a strong relationship of depressive symptoms to physical distress, perceived social support and
insecure attachment, both anxious and avoidant. Depressive symptoms were inversely related to age, as has been found in studies of depression in the community [62], but were not related to gender, marital status, living arrangement, socioeconomic status, country of birth, primary language, education, or duration of illness. The degree of physical suffering was considerable in these patients, who reported a mean of seven different physical symptoms, with more than two-thirds reporting fatigue and approximately half reporting pain and diminished alertness. Significant depressive symptoms were also relatively common with more than one-fifth of the sample reporting depressive symptoms in the range of severity that is often associated with clinical depression. The findings of this study are compatible with the view that adjustment to the multiple stresses and losses associated with advanced disease depends on individual, social, and disease-related factors. Previous studies have demonstrated the role of social
support in protecting the emotional well-being of cancer patients [63]. However, the experience of support depends not only upon the presence of significant others but also on the capacity to
24
咸宁学院本科毕业论文：
elicit and accept support and the ability to experience others as supportive. The latter is associated with what has been called a secure attachment style, which also tends to be
associated with the capacity to communicate and share emotional experience [21].This study sample reported a relatively high degree of social support, with mean scores over 80 on all dimensions of support, which corresponds to adequate support available ‘most of the time’ to ‘all of the time.’ Attachment security, as measured on the subscale of anxious (but not
avoidant) attachment, was found to buffer the impact of physical distress on the presence of depressive symptoms, to a degree that was both statistically significant and clinically meaningful. This would be compatible with the view that the expectation of, and flexible capacity to accept support from others is associated with increased resilience in the face of increasing disease burden. A similar buffering effect was not demonstrated for perceived social support, although such a relationship has been demonstrated in other medical populations [11,12].The protective effect of attachment security in the present study was
partially mediated through the perception of social support. This is consistent with previous research demonstrating that individuals who are insecurely attached are less likely to elicit social support and less likely to perceive social linteractions and relationships as supportive [26]
.These data suggest that the individual characteristic of attachment security protects from depressive symptoms both in its own right and also through its impact on perceived social support. This effect was greater for avoidant than for anxious attachment, perhaps because of the stronger association of anxious attachment with emotional distress and because avoidant attachment styles may be more likely to negatively affect the elicitation of social support.The finding that attachment security buffers the mood lowering effects of disease burden is
compatible with other research suggesting that attachment security has a protective effect on the emergence of distress under stress-related circumstances that activate the attachment system. In particular, this has been demonstrated in the management of death anxiety [64,65]. Why a buffering effect was found for attachment security but not for social support in this study deserves some consideration. It may be that the relatively high mean ratings of social support in this sample are related to the tendency for advanced disease to elicit support and/or that there may be social desirability attached to the perception of social support in this context. These factors may contribute to a greater overall tendency of patients with metastatic cancer to report more perceived support. In addition, since interaction effects are generally tested with much less power than are main effects, the failure to find a significant interaction effect for social support in this sample cannot rule out the possibility of its existence.The buffering effect of attachment security on depressive symptoms in this sample and its partial mediation through social support suggest that the interaction of individual, social, and disease-related factors may need to be considered in order to understand the emergence of depressive
symptoms in patients with metastatic cancer. Attachment security appears to be an important protective factor in this context, but these effects are partially mediated through its
relationship to the elicitation and/or perception of social support. Supportive interventions for cancer patients may be optimized,as we have noted in the palliative care setting [31], by taking into account individual differences in attachment needs, preferences and expectations. Whereas more avoidant individuals may benefit from support which is not perceived to
undermine their own sense of autonomy and self-reliance, those more anxiously attached may benefit most from support in which predictability and availability are clearly delineated. The tendency of avoidant individuals to initially minimize distress and their need for support may interfere with subsequent adaptation in the face of advanced disease, in which reliance on others is inevitably required. In that regard, avoidant attachment styles have been shown to be associated with stress generation in individuals who are mildly depressed[66]. Although both social support and attachment style are associated with depression, the interaction between these two factors may also be relevant in the generation of, or protection from, distress.
25
咸宁学院本科毕业论文：
However, caution must be exercised in the interpretation of these findings because of the crosssectional nature of this study. Longitudinal studies are needed to confirm the protective value of attachment security and social support in patients with advanced disease.
Acknowledgements
We thank our research staff and colleagues whocontributed to this project,including: Leanne Ferreira, Anne Rydall, Rinat Nissim, our many student and volunteer research
assistants, the PMH GI and lung clinic staff and volunteers who facilitated recruitment for our study, Larry Stitt who provided statistical expertise, and especially our study participants who gave their time and effort to help us better understand the experience of living with cancer. This study was supported by grants from the Canadian Institutes of Health Research (CIHR#MOP-62861; GR) and York University (LG). The authors gratefully acknowledge this support.
Role of the funding source
The sponsors of this study had no role in study design, data collection, data analysis, interpretation of findings, or writing of this report.
Conflict of interest statement
None declared.
References
1. Rodin GM, Nolan RP, Katz MR. Depression (in theMedically Ill). In APPI Textbook of Psychosomatic Medicine, Levenson J (ed.). American Psychiatric Press,Inc.: Washington, DC, 2004;113–127.
2. Peveler R, Carson A, Rodin G. Depression in medical patients. Br Med J 2002;325:149–152.
3. Cobb S. Social support as a moderator of life stress. Psychosom Med 1976;38:300–314. 4. Uchino BN, Cacioppo JT, Kiecolt-Glaser JK. The relationship between social support and physiological processes: a review with emphasis on underlying mechanisms and implications for health. Psychol Bull1996;119:488–531.
5. Gurevich M, Devins GM, Wilson C, McCready D, Marmar CR, Rodin GM. Stress
response syndromes in women undergoing mammography: a comparison of women with and without a history of breast cancer.Psychosom Med 2004;66:104–112.
6. Murberg TA, Bru E, Aarsland T, Svebak S. Social support, social disability and their role as predictors of depression among patients with congestive heart failure.Scand J Soc Med 1998;26:87–95.
7. Bisschop MI, Kriegsman DM, Beekman AT, Deeg DJ.Chronic diseases and depression: the modifying role of psychosocial resources. Soc Sci Med 2004;59:721–733.
8. Kornblith AB, Herndon JE, Zucherman E et al. Social support as a buffer to the psychological impact of stressful life events in women with breast cancer. Cancer 2001;91:443–453.
9. Cohen S, Syme SL. Issues in the study and application of social support. In Social Support and Health, Cohen S, Syme SL (eds). Academic Press: San Diego, 1985;3–22.
10.Olstad R, Sexton H, Sogaard AJ. A prospective population study of the social support buffer hypothesis, specific stressors and mental distress. Soc Psychiatry Psychiatr Epidemiol 2001;36:582–589.
11.Katz MR, Irish JC, Devins GM, Rodin GM, Gullane PJ. Psychosocial adjustment in head and neck cancer: the impact of disfigurement, gender and social support.Head Neck 2003;25:103–112.
12. Koopman C, Hermanson K, Diamond S, Angell K, Spiegel D. Social support, life stress, pain and emotional adjustment to advanced breast cancer. Psycho-Oncology 1998;7:101–111.
26