The partner’s insecure attachment, depression and psychological well-being as predictors of diurnal cortisol patterns for breast cancer survivors and their spouses

Abstract

The purpose of this study was to explore whether stress from individual’s and partner’s depression, anxiety, sleep disturbances, insecure attachment and meaning in life were predictors of diurnal cortisol patterns in breast cancer survivors and their spouses. Thirty-four couple dyads participated in this eight-month follow-up study. The breast cancer survivors and their spouses completed the Medical Outcomes Study Sleep scale, the Beck Depression Inventory-II, the State Trait Anxiety Inventory, the Experiences in Close Relationships – Revised scale and the Meaning in Life Questionnaire, and they collected salivary cortisol at home at the time of awakening, 30 and 45 min after waking and at 1200 h, 1700 h and 2100 h. Diurnal cortisol slopes of survivors and spouses are positively correlated. But the factors associated with diurnal cortisol patterns are different between survivors and spouses. For survivors, neither survivor individuals’ nor spouses’ psychosocial factors were the predictors of survivors’ diurnal cortisol patterns. For spouses, the survivors’ higher anxious attachment style was the main predictor of spouses’ flatter diurnal cortisol patterns. In conclusion, for spouses, psychophysiological stress responses are mainly influenced by breast cancer survivors’ insecure attachment. Future couple supportive care interventions can address survivors’ attachment styles in close relationships in order to improve neuroendocrine functions for both breast cancer survivors and their spouses.

• Anxiety
• attachment styles
• breast cancer survivors
• depression
• diurnal cortisol patterns
• meaning in life
• sleep disturbance
• spouses

Introduction

Increasing breast cancer survival rates have left greater numbers of survivors and their spouses to cope with the transition back to normal life. Such transitions often induce psychosocial stress relating to the fear of recurrence, the resumption of a sexual and intimate relationship and the uncertainty of life’s meaning (Alfano & Rowland, 2006; Ganz et al., 2002; Hodgkinson et al., 2007). Patients with breast cancer and their spouses appear to have higher psychological distress when both of them hold a negative cognitive appraisal of stress (Bigatti et al., 2012). In addition, there are interrelationships of emotional stress response (depression) and daily life adjustment (quality of life) between survivors and spouses (Awadalla et al., 2007; Mellon et al., 2006). The negative responses to cumulative stress reflected in flatter diurnal cortisol patterns have been considered a predictor of shorter survival rates in patients with breast cancer (Sephton et al., 2000). For spouses of patients with cancer, how they perceived cancer-specific stress was known to be correlated with increases in their physical symptoms and changes in immune function (Wells-Di Gregorio et al., 2012). It is unclear whether or not the stress from partner’s psychological distress is associated with negative cortisol stress responses in breast cancer survivors and their spouses.

Attachment styles describe the nature of emotional bonds in close interpersonal relationships (Fraley et al., 2000). It has been found that patients with metastatic cancer and their spouses’ anxious and avoidant attachment styles are positively correlated with their depression (Braun et al., 2007; Rodin et al., 2007). This finding suggests insecure attachment in close relationships might be one of the main sources of psychological stress for both patients and spouses. Studying the impact of partners’ attachment styles between breast cancer survivors and their spouses as a predictor of diurnal cortisol patterns might clarify whether stress from close relationships during survivorship could be related to changes in the neuroendocrine stress response.

Depression is a common emotional distress among breast cancer women and their spouses (Braun et al., 2007; Giese-Davis et al., 2006). Increases of cortisol levels likely occur in depressed patients with metastatic breast cancer (Giese-Davis et al., 2006). However, in cross-sectional studies (Abercrombie et al., 2004; Vedhara et al., 2006), depression and anxiety symptoms were not significantly related to diurnal cortisol patterns in patients with breast cancer. Spouse caregivers of patients with advanced cancer are regarded as a high risk group for depression (Braun et al., 2007) but it is unclear about the impacts of depression on cortisol responses in spouses. Moreover, a longitudinal study needs to clarify the impact of partners’ depression and anxiety on cortisol responses between patients with breast cancer and their spouses.

One study (Manne et al., 2004) found that breast cancer survivors and their spouses experience post-traumatic enlightenment after cancer, such as accepting the situation, and appreciating life (Kim & Carver, 2007). Moreover, spouses’ finding positive meaning in caregiving experiences was positively correlated with life satisfaction. These results suggest that spiritual growth from finding positive meaning in life might produce a protective factor for psychological adjustment. Questions arise, however, about whether this protective factor could predict (a) neuroendocrine stress responses for both breast cancer survivors and their spouses and (b) how partners’ positive view of cancer experience could influence cortisol responses of survivors and their spouses.

In summary, negative psychosocial distress (insecure attachment and depression) and positive psychological well-being (meaning in life) commonly occur in breast cancer survivors and their spouses. However, it is unknown whether stress from individual or partner’s psychosocial distress and psychological well-being are the main factors associated with cortisol stress responses for breast cancer survivors and spouses. This study aims to test hypothesis that individual’s and partner’s factors of depression, anxiety, sleep disturbances, insecure attachment in close relationships and meaning of life are the predictors of diurnal cortisol patterns in both breast cancer survivors and spouses.

Methods

This study is approved by Hospital Institutional Ethics Review Board for research on human subjects. Thirty-four dyads of breast cancer survivors and their spouses participated in this longitudinal study. Patients with breast cancer who are aged between 18 and 65 years and completed active treatments were recruited from the outpatient department of surgery in a breast cancer centre of a general hospital. Those diagnosed with both breast cancer and another type of cancer and those with adrenal function disorders (such as Cushing syndrome, Addison’s disease, adrenal tumor or pituitary tumor) were excluded. Inclusion criteria of spouses of survivors include living with the survivors, aged 18 years or older and absence of adrenal function disorders. The research assistant explained to breast cancer survivors and/or spouses in the waiting room located at the outpatient department of surgery in the breast cancer centre of a general hospital about the purposes, risks, benefits and ethical issues involved in the study, both verbally and in writing. The spouses who were not present with the survivors in the hospital were recruited through the survivors, who were asked for permission for the researchers to contact them. With the permission of the survivors and their spouses, the research assistant explained the details of this study and then collected informed consent. After the participants gave their written consent, they received the self-report questionnaires, the neutral cotton Salivette tubes (Sarstedt, Germany) and instructions for collecting saliva cortisol at home. A total of 189 breast cancer survivors were invited, but 155 declined to participate mainly due to their spouses not wanting to participate in this study. The questionnaires and Salivette tubes for the follow-up tests were sent to each participant. The 34 dyads agreed to complete baseline assessment (T0) and all follow-up tests: T1 (in the second month), T2 (in the fifth month) and T3 (in the eighth month). However, one spouse at T1 and one couple dropped out at T2, resulting in a 94% retention rate with 33 breast cancer survivors and 32 spouses completing all the follow-up data.

Over 60% of the survivors’ cancer stages were stage I (26.3%) or stage II (42.1%). Most of the survivors did not have cancer metastasis (64%), while 34.2% of the participants had had lymph nodes test positive. Regarding active treatment history, they all had received surgical treatment for breast cancer, most of them had received chemotherapy (75%) and 44.7% of survivors had received radiotherapy. At the post-active treatment stage, 22.4% of the survivors had received hormone therapy. The mean duration between completion of active cancer treatments and study entry was 3.31 years, ranging from 1.91 to 10.89 years. The mean duration between having undergone breast cancer surgery and study entry was 3.82 years, with a range of 1.94–10.89 years. Over 50% (53.9%) of the survivors had a family history of cancer. Table 1 indicates more spouses were currently employed than survivors. There were borderline significant differences in age between the two groups: the mean age of survivors was 49.6 years, ranging from 21 to 63 years; and the mean age of spouses was 53.7 years, ranging from 35 to 70 years. Despite these differences, there were similarities in education levels and with religious belief, as well as baseline scores of self-reported scales. Diurnal cortisol slopes of survivors and spouses are similar and they are positively correlated (r = 0.373, p = 0.03). However, the factors associated with diurnal cortisol patterns differed between survivors and spouses, which are described in the next section.

Table 1. Demographic and baseline characteristics of the breast cancer survivors and spouses.

	Survivors (n = 34)	Spouses (n = 34)	t/χ^2	p
Age (mean ± SD), years	49.6 ± 9.2	53.7 ± 8.5	1.92	0.059
Education			0.95	0.32
High school graduate/nongraduates, %	50.0/50.0	61.8/38.2
Working status			14.78	<0.001^a
Employed/unemployed, %	44.1/55.9	88.2/11.8
Religion			0.11	0.74
With/without, %	85.3/14.7	82.4/17.6
Anti-depressant medication			1.02	0.31
Yes/No, %	2.9/97.1	0/100
Sleeping pill			1.88	0.17
Yes/No, %	20.6/79.4	8.8/91.2
BDI-II (mean + SD)	8.7 ± 9.6	7.4 ± 9.4	−0.60	0.55
Anxiety state of STAI	40.7 ± 12.1	39.3 ± 10.4	−0.51	0.61
MLQ presence	26.1 ± 6.1	26.5 ± 5.4	0.27	0.78
MLQ search	25.2 ± 7.3	26.2 ± 5.0	0.64	0.52
ECR-R anxiety attachment style	43.9 ± 20.6	51.4 ± 20.0	1.52	0.13
ECR-R avoidance attachment style	37.2 ± 15.9	38.2 ± 15.5	0.26	0.79
Sleep problem index	28.1 ± 15.6	24.7 ± 15.8	−0.89	0.37
Diurnal cortisol slope
(log nmol/l per hour)	−1.09 ± 0.36	−1.10 ± 0.41	−0.158	0.87

^ap < 0.001.

Medical Outcomes Study Sleep scale (Spritzer & Hays, 2003) contains six subscales: sleep disturbance, snoring, shortness of breath or headache during sleep, sleep adequacy, sleep somnolence and sleep quantity. Five of the six scales (except sleep quantity) were combined into a nine-item sleep problem index. Higher scores on the sleep problem index indicated higher levels of sleep problems. An adequate internal consistency for the sleep problem index was reported in this study (alpha = 0.831 for breast cancer survivors and 0.867 for their spouses). Beck Depression Inventory-II (BDI-II) scale (Beck et al., 1961) has total scores ranging from 0 to 63. The total scores are classified within four levels of depression: 0–13, normal; 14–19, mild; 20–28, moderate; and 29–63, severe. State Trait Anxiety Inventory (STAI) (Spielberger et al., 1970), a measure of anxiety, produces a range of scores from 20 to 80, with higher scores indicating higher levels of anxiety state. In this study, the BDI-II scale and STAI both had good internal consistency (BDI-II alpha = 0.935 for breast cancer survivors and 0.936 for their spouses; STAI alpha = 0.962 for breast cancer survivors and 0.949 for their spouses). Experiences in Close Relationships – Revised (ECR-R) scale (Fraley et al., 2000) measures two types of attachment styles: an 18-item attachment-related anxiety subscale assesses fear of rejection and abandonment by others and an 18-item attachment-related avoidance subscale assesses discomfort with closeness and dependency or a reluctance to be intimate with others. The participants were asked to respond on a 7-point Likert-type scale ranging from 1 (disagree strongly) to 7 (agree strongly), which generated a score for each subscale ranging from 18 to 126. Higher scores indicate more insecure attachment styles. Good internal consistency for the attachment-related anxiety subscale and attachment-related avoidance subscale have been reported in this study (anxious attachment style alpha = 0.913 for breast cancer survivors and 0.856 for their spouses; avoidant attachment style alpha = 0.922 for breast cancer survivors and 0.876 for their spouses). Meaning in Life Questionnaire (MLQ) (Steger et al., 2006) measures psychological well-being using two scales: a 5-item MLQ-Presence scale, measuring the subjective sense that one’s life is meaningful, and a 5-item MLQ-Search scale, measuring one’s drive and orientation toward finding meaning in life. MLQ is measured on a 7-point Likert-type scale ranging from 1 (absolutely untrue) to 7 (absolutely true). Total scores range from 5 to 35 for each subscale. Higher scores indicate higher levels of psychological well-being. Adequate internal consistency was reported in this study (MLQ-Presence scale alpha = 0.86 for breast cancer survivors and 0.82 for their spouses; MLQ-Search scale alpha = 0.94 for breast cancer survivors and 0.85 for their spouses).

After being informed and instructed about the study, all participants collected their salivary cortisol responses using neutral cotton Salivette tubes in their homes at six time points: upon waking, 30 and 45 min after waking and at 1200 h, 1700 h and 2100 h. The participants were instructed about collection and storage procedures. They collected saliva samples only once on a weekday at each measurement point. The participants were told not to brush their teeth before completing the first saliva sampling of the day and not to eat before the first three collections had been completed. For the remaining three samples, participants were asked not to eat during the 30 min immediately before they collected the samples. Except for the above restrictions, participants followed their normal daily routines on the sample collection day. The participants received the saliva kits for the follow-up tests through mail and repeated the same collection procedures.

To determine the diurnal cortisol patterns, β values for diurnal cortisol slopes (log nmol/l per hour) were calculated based on a regression of the log-transformed mean (standard deviation) cortisol levels (nmol ± l) at each of the six measuring times. Salivary cortisol levels (nmol ± l) were skewed positively, so we used the natural logarithm to transform the raw cortisol levels to obtain an unskewed distribution for further analysis. Steeper slopes (log nmol/l per hour), which are characterized by a normal descending profile from high morning cortisol levels to lower evening cortisol levels, are indicated by smaller β values for the slope of the regression. Flatter slopes that do not show descending cortisol levels as the day progresses are indicated by larger β values, which show slower declines (Sephton et al., 2000).

Repeated measures analysis of variance was used to examine whether significant change occurred in the means of the diurnal cortisol slopes and psychosocial measures from the baseline levels (T0) across the second (T1), fifth (T2) to eighth (T3) months of follow-up among survivors and spouses. The F value generated by the statistic test of within-subject effects showed the significant values for the change in means by time for the main variables of diurnal cortisol slopes, BDI-II depressive levels, STAI anxiety levels, ECR-R attachment styles, MLQ meaning in life and Sleep Problems index.

The generalized estimating equations (GEE) model was used to test what were the predictors for the changes in diurnal cortisol patterns across the eight-month follow-up period. GEE was applied for analyzing the longitudinal data of this study for a number of reasons: all cases (even drop-out cases) could be included by using available pairs to estimate working correlation parameters for the total sample, correlated errors for bias estimations could be managed, and this model had the ability to analyze the within-subject correlations in responses to the dependent variable (diurnal cortisol patterns) at the four measurement time points (Liang & Zeger, 1986). Because sociodemographic data (age, educational levels, with/without being employed and with/without religion) and body mass index were not correlated with diurnal cortisol slopes among survivors and spouses (p > 0.05), these data were not used for further GEE analysis. In the GEE model, psychosocial variables (BDI depressive symptoms, STAI anxiety symptoms, presence and search for meaning in life, anxiety and avoidance types of attachment styles in close relationships and sleep problem index) were placed as independent variables to explore their roles in predicting diurnal cortisol patterns. Moreover, to examine the effects of mutual psychological distress and psychophysiological stress responses on diurnal cortisol patterns between breast cancer survivors and their spouses, the partners’ psychosocial variables and partner’s diurnal cortisol slopes were also included in the GEE model in addition to their individual psychosocial variables. Coefficients of independent variables were used to demonstrate the power of independent variables predicting the changes in diurnal cortisol patterns over eight months. The same procedures were conducted separately for breast cancer survivors and their spouses. To consider correction for multiple comparisons in two models of GEE analysis, the significance p value was designated 0.025 using Bonferroni correction.

Results

Change in means for diurnal cortisol slopes and psychosocial measures during eight months follow-up among survivors and spouses

In Table 2, there were no significant changes from baseline scores of diurnal cortisol slopes, anxiety and depressive levels, anxious and avoidant styles of attachment in close relationships, view of presence of and search for meaning in life and sleep problem index (p > 0.025) for either survivors or spouses. These results indicate that diurnal cortisol slopes and all psychosocial status maintained stable patterns for both survivors and spouses during the eight-month follow-up period.

Table 2. The mean and standard deviations (SD) of psychosocial measures and diurnal cortisol slopes at four measurement times during eight months follow-up among survivors and spouses.

	T0 Mean (SD)	T1 Mean (SD)	T2 Mean (SD)	T3 Mean (SD)	Statistical analysis F, p
Survivor individuals
BDI-II	9.09 (9.76)	8.16 (7.66)	7.91 (6.44)	6.50 (5.09)	F = 1.978 p = 0.149
Anxiety state of STAI	40.97 (12.35)	39.78 (11.80)	39.38 (11.61)	38.28 (8.99)	F = 1.099 p = 0.354
MLQ presence	26.00 (6.25)	26.21 (6.63)	26.62 (6.39)	28.00 (5.74)	F = 2.648 p = 0.054
MLQ search	25.25 (7.43)	24.90 (8.03)	25.03 (7.93)	25.81 (8.60)	F = 0.359 p = 0.783
ECR-R anxiety attachment style	45.67 (20.77)	45.09 (20.45)	43.87 (17.22)	41.61 (16.02)	F = 0.822 p = 0.485
ECR-R avoidance attachment style	37.7 (16.15)	39.65 (18.48)	39.90 (21.70)	38.40 (14.95)	F = 0.367 p = 0.735
Sleep problem index	28.62 (15.91)	28.47 (13.98)	28.45 (14.22)	24.34 (13.23)	F = 1.699 p = 0.173
Diurnal cortisol slope (log nmol/l per hour)	−1.06 (0.35)	−1.23 (0.44)	−1.25 (0.61)	−1.25 (0.67)	F = 1.099 p = 0.354
Spouse individuals
BDI-II	7.81 (9.68)	6.19 (6.88)	7.32 (8.73)	5.65 (6.24)	F = 1.836 p = 0.146
Anxiety state of STAI	39.23 (10.92)	38.94 (11.57)	37.77 (12.91)	36.45 (11.77)	F = 1.343 p = 0.266
MLQ presence	26.90 (5.43)	28.22 (5.52)	27.93 (5.63)	29.06 (5.62)	F = 2.558 p = 0.060
MLQ search	26.35 (5.12)	26.70 (6.22)	25.70 (6.10)	27.12 (6.22)	F = 0.712 p = 0.547
ECR-R anxiety attachment style	51.16 (19.40)	51.66 (19.62)	51.56 (20.07)	46.90 (21.26)	F = 1.166 p = 0.328
ECR-R avoidance attachment style	38.09 (16.07)	37.93 (15.27)	39.45 (17.14)	37.06 (14.99)	F = 0.385 p = 0.713
Sleep problem index	24.42 (16.01)	24.21 (13.75)	24.64 (12.62)	23.70 (13.16)	F = 0.099 p = 0.960
Diurnal cortisol slope (log nmol/l per hour)	−1.11 (0.42)	−1.07 (0.55)	−1.10 (0.43)	−1.15 (0.47)	F = 0.292 p = 0.831

The associations of diurnal cortisol slopes with psychosocial measures among survivors and spouse at the baseline condition

For survivors, initial flatter diurnal cortisol slopes were only significantly associated with higher baseline scores for view of presence of meaning in life (r = 0.423, p = 0.014), while the rest of psychosocial measures were not related to diurnal cortisol slope. Survivors’ view of presence of meaning in life was negatively associated with anxiety (r = −0.379, p = 0.027), depression (r = −0.564, p = 0.001) and sleep problem index (r = −0.393, p = 0.022). Anxiety levels were positively correlated with depressive levels (r = 0.62, p < 0.001). Anxiety and depression were both positively correlated with anxious attachment style (r = 0.422, p = 0.013; r = 0.436, p = 0.010) and sleep problem index (r = 0.644, p < 0.001; r = 0.576, p < 0.001). But survivors’ individual diurnal cortisol slopes at baseline were not associated with their spouses’ psychosocial status (p > 0.05).

For spouses, initial flatter diurnal cortisol slopes were only correlated with higher baseline scores for avoidant attachment style (r = 0.413, p = 0.015). Spouses’ avoidant attachment style was positively related to sleep problem index (r = 0.367, p = 0.033). There were no significant associations of spouses’ view of presence of meaning in life with depressive levels (r = −0.322, p = 0.063). Anxiety levels were positively correlated with depressive levels (r = 0.732, p < 0.001). Anxiety and depression were both positively correlated sleep problem index (r = 0.579, p < 0.001; r = 0.441, p = 0.009). However, spouses’ individual diurnal cortisol slopes at the baseline condition were not associated with survivors’ psychosocial status (p > 0.05).

Survivors’ initial diurnal cortisol slopes were positively correlated with spouses’ initial diurnal cortisol slopes (r = 0.373, p = 0.032). For the correlations of psychosocial status between survivors and spouses, sleep problem index was positively associated with anxiety and depressive levels between survivors and spouses (r = 0.370, p = 0.031; r = 0.394, p = 0.021). Anxiety levels, depressive levels, attachment styles and meaning in life were not related between survivors and spouses (p > 0.05).

The predictors of the changes in diurnal cortisol slopes over eight months among breast cancer survivors

For breast cancer survivors, in Table 3, the findings indicated that all survivor individuals and spouse variables in GEE model were not correlated with changes of breast cancer survivor’s diurnal cortisol slopes during the eight-month follow-up period.

Table 3. Breast cancer survivor individuals’ and spouses’ psychosocial factors as predictors of diurnal cortisol slope over a period of eight months in breast cancer survivors.

Predictors	Beta	SE	95% CI lower	95% CI upper	Wald χ^2	p
Survivor individuals
BDI-II	−0.005	0.0068	−0.018	0.008	0.590	0.442
Anxiety state of STAI	0.003	0.0059	−0.008	0.015	0.350	0.554
MLQ presence	0.009	0.0075	−0.005	0.024	1.530	0.216
MLQ search	−0.007	0.0059	−0.019	0.004	1.538	0.215
ECR-R anxiety attachment style	0.000	0.0022	−0.004	0.005	0.012	0.911
ECR-R avoidance attachment style	0.001	0.0032	−0.005	0.007	0.107	0.744
Sleep problem index	−0.004	0.0030	−0.01	0.002	1.939	0.164
Spouses
BDI	−0.011	0.0091	−0.029	0.006	1.553	0.213
Anxiety state of STAI	0.000	0.0067	−0.013	0.013	0.002	0.963
MLQ presence	−0.021	0.0106	−0.042	0.000	3.771	0.052
MLQ search	0.010	0.0105	−0.010	0.031	0.956	0.328
ECR-R anxiety attachment style	0.001	0.0022	−0.003	0.006	0.448	0.503
ECR-R avoidance attachment style	−0.003	0.0044	−0.011	0.006	0.411	0.521
Sleep problem index	0.000	0.0036	−0.008	0.007	0.020	0.889
Diurnal cortisol slope (log nmol/l per hour)	0.135	0.1131	−0.087	0.356	1.416	0.234

Working correlation matrix structure: AR(1).

Goodness of fit: Quasi Likelihood under Independence Model Criterion (QIC) = 65.12.

Corrected Quasi Likelihood under Independence Model Criterion (QICC) = 67.404.

The predictors of the changes in diurnal cortisol slopes over eight months among spouses

Table 4 indicated that survivors’ anxious attachment styles predicted the changes of spouses’ diurnal cortisol slopes. There were the significant associations of survivors’ higher scores for anxious attachment style (β = 0.005, χ² = 6.946, p = 0.008) with spouses’ flatter diurnal cortisol slopes. The rest of the spouse individuals and survivor variables in GEE model were not correlated with changes of spouse’s diurnal cortisol slopes over the eight-month follow-up period.

Table 4. Spouse individuals’ and survivors’ psychosocial factors as predictors of diurnal cortisol slope over a period of eight months in spouses.

Predictors	Beta	SE	95% CI lower	95% CI upper	Wald χ^2	p
Spouse individuals
BDI-II	0.000	0.0081	−0.017	0.015	0.011	0.915
Anxiety state of STAI	0.005	0.0066	−0.008	0.018	0.609	0.435
MLQ presence	−0.008	0.0124	−0.032	0.016	0.420	0.517
MLQ search	0.003	0.0103	−0.017	0.023	0.074	0.786
ECR-R anxiety attachment style	7.412E-5	0.0021	−0.004	0.004	0.001	0.971
ECR-R avoidance attachment style	0.006	0.0042	−0.003	0.014	1.814	0.178
Sleep problem index	0.000	0.0035	−0.008	0.006	0.035	0.851
Survivors
BDI	0.012	0.0069	−0.002	0.025	2.895	0.089
Anxiety state of STAI	−0.007	0.0064	−0.019	0.006	1.089	0.297
MLQ presence	0.021	0.0100	0.001	0.041	4.429	0.035
MLQ search	−0.005	0.0079	−0.020	0.011	0.346	0.556
ECR-R anxiety attachment style	0.005	0.0018	0.001	0.008	6.946	0.008^a
ECR-R avoidance attachment style	0.000	0.0026	−0.006	0.004	0.142	0.706
Sleep problem index	0.003	0.0038	−0.005	0.010	0.469	0.493
Diurnal cortisol slope (log nmol/l per hour)	0.062	0.0736	−0.082	0.206	0.708	0.400

^ap < 0.025.

Working correlation matrix structure: AR(1).

Goodness of fit: Quasi Likelihood under Independence Model Criterion (QIC) = 70.980.

Corrected Quasi Likelihood under Independence Model Criterion (QICC) = 59.517.

Discussion

This study indicated that diurnal cortisol slopes of survivors and spouses are positively correlated. However, the factors associated with diurnal cortisol patterns are different between survivors and spouses. For breast cancer survivors, neither survivor individuals nor spouse variables in the GEE model were correlated with changes of breast cancer survivor’s diurnal cortisol slopes during the eight-month follow-up period. For spouses, cortisol stress responses are associated with survivors’ factors rather than their own psychosocial factors. These findings indicate that there were significant associations of survivors’ higher scores for anxious attachment style with the spouses’ flatter diurnal cortisol patterns. In summary, the findings of this study demonstrate that psychophysiological stress responses of survivors and their spouses are interrelated. Survivors’ anxious attachment style in close relationship was the main predictor of their spouses’ negative cortisol stress responses during the eight-month follow ups.

In this study, higher survivors’ anxious attachment style correlated with their spouses’ flatter diurnal cortisol patterns suggests that breast cancer survivors’ feelings of insecurity in close relationships contribute to negative physiological stress responses in their spouses. This study found that survivors’ anxious attachment style was positively associated with their anxiety and depression. The results suggest that survivors’ fear of rejection and abandonment by spouses in close relationship likely produces negative emotions, which might influence their spouses to have negative cortisol responses. The previous study found that high anxious attachment styles of patients with cancer were associated with their partners’ burden and psychological distress in the caregiving process (Braun et al., 2011). The authors suggest that this was because patients with cancer with insecure attachment likely appeared to be too dependent and demanding to allow love and closeness, too self-doubting about their ability to be loved and accepted and were often dissatisfied with their partners’ support (Braun et al., 2011). The finding demonstrates that breast cancer survivors’ insecurity in close relationships might influence their spouses’ emotional distress and marital adjustment, which likely produces abnormal cortisol responses to stress.

As for unexpected findings, individuals’ and partners’ respective depressive or anxiety symptoms did not predict diurnal cortisol patterns for breast cancer survivors or their spouses, although these symptoms are related to cortisol responses in the studies on patients with major depressive disorders (Hsiao et al., 2010). The reasons for this finding might be related to the low mean averages of BDI-II depression and STAI anxiety scores at the baseline condition in breast cancer survivors and their spouses in this study; these measurements were classified within the normal range for both depression and state of anxiety.

The main limitation of this study is a poor response rate (34/189 = 18%) due to most spouses being unwilling to participate, which limits the generalization of the findings. This limitation might reflect the findings on men’s help-seeking behaviors. Traditional masculine norms and negative attitudes toward help-seeking tend to discourage men from seeking help from others (Sánchez et al., 2013). Future studies are recommended to explore spouses’ needs and the barriers to obtaining help from professionals during the survivorship period. Moreover, in this study, the low mean averages of BDI-II depression and STAI anxiety scores and low percentages of receiving antidepressant medication and sleeping pill in survivors and their spouses suggest that our participants are likely to have high levels of psychological functioning. Moreover, most of the breast cancer survivors in this sample were not in metastasis, cancer stages mainly were I or II, and the minimum duration between completion of active cancer treatments and study entry was 3.31 years. These limitations influence the generalization of the results to survivors with metastatic breast cancer, advanced cancer stages, those just completing active cancer treatments and couples (survivors and spouses) with severe psychological distress. Future studies need to include a more representative sample that shows greater variation in cancer characteristics and psychological distress in order to demonstrate the main predictors of diurnal cortisol patterns among breast cancer survivors and their spouses. Despite the limitations, this study with a longitudinal design brings a new finding of the effects of survivors’ psychological distress on spouses’ neuroendocrine stress responses. Future interventions for couples can address survivors’ attachment styles in close relationships in order to improve neuroendocrine functions for both breast cancer survivors and their spouses while they are coping with stress during the survivorship period.

Declaration of interest

All authors of this study declare that we all have no conflicts of interest including any financial, personal or other relationships with other people or organizations within three years of beginning the work submitted that could inappropriately influence, or be perceived to influence, our work.

This study was supported by a grant from the Frontier and Innovative Research of National Taiwan University (98R0304).

Acknowledgements

The authors are grateful to the staff from Foundation of Breast Cancer Prevention and Treatment, Miss Yun-Ting Lee and Yu-Han Tseng for their assistance in data collection and Miss Ming-Ru Wang for her help in cortisol analysis. The authors acknowledge statistical assistance provided by the Taiwan Clinical Trial Bioinformatics and Statistical Center, Training Center and Pharmacogenomics Laboratory (which is founded by National Research Program for Biopharmaceuticals (NRPB) at the National Science Council of Taiwan; NSC 102-2325-B-002-088).