A couple-based intervention for patients and caregivers facing end-stage cancer: outcomes of a randomized controlled trial

Authors

  • Linda M. McLean,

    Corresponding author
    1. Department of Psychosocial Oncology and Palliative Care, University Health Network, Princess Margaret Hospital, Toronto, Ontario, Canada
    • Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Tara Walton,

    1. University Health Network, Cancer Survivorship Program, Toronto, Ontario, Canada
    Search for more papers by this author
  • Gary Rodin,

    1. Department of Psychosocial Oncology and Palliative Care, University Health Network, Princess Margaret Hospital, Toronto, Ontario, Canada
    2. Ontario Cancer Institute, Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Mary Jane Esplen,

    1. Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada
    2. de Souza Institute, University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Jennifer M. Jones

    1. Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada
    2. Cancer Survivorship Program, Toronto, Ontario, Canada
    Search for more papers by this author

Suite 206 200 St. Clair Avenue West, Department of Psychiatry, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada, M4V 1R1. E-mail: lmmclean@rogers.com

Abstract

Background

We evaluated the effect of Emotionally Focused Therapy (EFT) [intervention (INT)], modified for the advanced cancer population versus standard care [control (CTL)], on marital functioning and psychosocial outcomes among distressed couples. EFT examines the ways couples process patterns of interaction, facilitating change.

Methods

Forty-two patients with metastatic cancer and their caregivers were assigned randomly to receive eight sessions of EFT delivered by one EFT-trained psychologist or standard care provided by multidisciplinary clinicians. Partners completed measures of marital functioning, depression, hopelessness, empathic caregiving (patient), and caregiver burden (caregiver) before randomization at baseline (T0), post-intervention (T1), and 3-month post-intervention (T2). Data were analyzed using three-way analyses of covariance; main effects were treatment, patient status, and sex. Multilevel modeling was also employed.

Results

At T1, INT group led to significant improvement in marital functioning (p ≤ 0.0001) and patients' experience of caregiver empathic care (p = 0.02) versus CTL group. Effect sizes for these differences were in the large range (d = 1.00). In both groups, patients showed a marginally higher mean score for marital functioning compared with caregivers [INT: M = 56.3, standard deviation (SD) = 4.6 vs M = 54.3, SD = 4.5; CTL group: M = 43.4, SD = 10.3 vs M = 42.4, SD = 6.8, respectively]. At T2, results were maintained.

Conclusions

The effectiveness of and adherence to the EFT couple intervention for patients facing advanced cancer were demonstrated in this study. Larger randomized control trials are needed to confirm these findings. Qualitative studies may help to further understand the mechanism by which such an intervention may act. Copyright © 2011 John Wiley & Sons, Ltd.

Introduction

The diagnosis of cancer has profound consequences for both patients and their family members [1-4], and its progression brings further challenges for couples that include imposed changes in responsibilities [5] and the threat of mortality [6, 7]. Distress, including depression and hopelessness, is very common in both partners when facing a diagnosis of metastatic cancer [8-14]. Social support, an important protective factor for patients with cancer [15], may be most important from a spouse [16, 17].

Although practical support provided by a spouse may facilitate the adjustment of patients with cancer, emotional support may be more closely linked to their psychological well-being [16]. Emotional support refers to the verbal and non-verbal communication of concern and care and includes the ability to listen, to empathize, and to offer comfort and reassurance [16]. The perception of empathic responses from their spouse is seen as salient to patients' experience of positive interactions in that relationship [16]. Communication between patients and spouse caregivers about respective cancer-related concerns can negatively affect the couple's relationship [17]. In that regard, negative, avoidant, and hostile marital interactions can compromise the relational experience of both partners at the end-of-life and can impact the bereaved caregiver following the patients' death [18-21]. In contrast, couples with strong and positive attachment bonds can draw on this stable foundation of mutual trust, affection, and respect during the challenges of terminal cancer [22-30].

The recent trend toward longer survival and home care of patients with cancer has tended to increase the involvement of spouses in the primary caregiving role [8]. For patients with advanced disease, the need for caregiving assistance increases as the disease progresses, with a varying emotional impact on spouses [31]. Spouses who are less satisfied with the marital relationship may experience more caregiving burden [8]. Caregiver burden refers to both objective caregiving, including care-related tasks and time involved in caregiving, as well as subjective caregiving, including the caregiver's experience and feelings about their role as caregiver [8].

Couple-based interventions for patients with cancer have the potential to reduce marital and psychological distress and caregiver burden and to promote relational growth, when one partner is facing end-of-life [3, 32-36]. Moreover, the bereavement process may be eased and grief therapy facilitated [21]. However, despite advocacy for couples' interventions focused on the specific needs of patients with terminal cancer and their spouses [12, 18, 20, 26, 34, 37-39], the suffering of such couples remains largely under-recognized and undertreated [33].

The majority of studies that have evaluated couples' interventions for patients with cancer have been limited by small sample sizes, varied study designs, interventions, and measures with only a handful focusing on terminal cancer [2, 13, 17, 32, 40-43]. There have been three randomized control trials (RCTs) evaluating couple-based interventions for couples facing advanced cancer [17, 42, 43]. Porter and colleagues [17] examined the relational benefit of partner-assisted emotional disclosure in 130 patients with gastrointestinal cancer and their partners. Findings showed that when partners avoided discussion of cancer-related concerns, the partner-assisted emotional disclosure condition resulted in improvements in relationship quality and intimacy compared with the education/support condition.

Northouse et al. [42] conducted a prospective longitudinal RCT with 134 patient–family caregiver dyads to assess the ‘Family involvement, Optimistic attitude, Coping effectiveness, Uncertainty reduction, Symptom management (FOCUS) program’, a family intervention based on the model of stress appraisal of Lazarus and Folkman [44, 45]. Patients in the ‘FOCUS program’ reported significantly less hopelessness and less negative appraisal of illness than the control group; this was not maintained at 3-month and 6-month follow-up. Caregivers in the FOCUS program also reported less negative appraisal of caregiving, but this was not sustained at post-intervention follow-up. Finally, Kuijer et al. [43] conducted a smaller RCT to evaluate a brief cognitive intervention based on equity theory. Couples in the intervention showed significant moderate effects with respect to their perceptions of the give-and-take balance (d = 0.50) and higher levels of relationship quality, the latter decreasing at 3-month follow-up. Patients also reported lower levels of psychological distress following the intervention, whereas in contrast, partners did not.

In an attempt to address some of the limitations in the existing literature, we developed a theoretically based couple intervention using Emotionally Focused Therapy (EFT), adapted for use in couples facing non-curative metastatic cancer (i.e., advanced disease) [46-48]. EFT is a well-established treatment for couples in non-cancer populations including those with depression and trauma [46-50]. EFT arises from the synthesis of experiential and systemic approaches. The focus is on intrapsychic processes, interpersonal processes, and a theory of adult love, which is viewed as an attachment process [47]. Attachment behaviors are germane in the face of metastatic cancer due to separation and loss, activating separation distress [35, 46]. Distressed relationships are thought of as insecure bonds in which healthy attachment needs are unable to be met due to rigid patterns of interacting, inhibiting emotional engagement. In emphasizing the quality of the attachment and emotional involvement between spouses, EFT can address both patients and partners' experience of chronic and terminal illness as well as their relationship dynamics, to increase more engagement, flexible response patterns and to strengthen the attachment bond, mitigating grief and loss [47]. A meta-analysis of RCTs of EFT in the general population showed a 70–73% recovery rate from marital distress (Dyadic Adjustment Scale) compared with wait-list control groups [48, 51]. The weighted mean effect size in four of the most rigorous studies was d = 1.31 [48, 52].

Building on the positive results of a pilot evaluation [3], we conducted an RCT to further evaluate this intervention. To our knowledge, this study represents the first RCT examining the impact of a modified EFT protocol for advanced cancer care. We hypothesized that following EFT, the intervention (INT) group would demonstrate a greater increase in marital functioning post-intervention when compared with the control (CTL) group. We also expected that the INT group would have a greater decrease in depression, hopelessness, and spouse caregiver burden scores and a greater increase in scores of patients' perception of spouse caregiver empathic behavior.

Methods

Recruitment and study participants

Participants were recruited from Princess Margaret Hospital (PMH), Canada's largest comprehensive cancer center and part of the University Health Network (UHN) in Toronto, Ontario. The study was approved by the UHN Research Ethics Board. Potential participants were identified if they had requested a psychosocial referral for couple distress or if there was perceived clinical need noted during routine attendance in PMH clinical outpatient clinics by nurses and physicians that led to subsequent referral to the Psychosocial Oncology and Palliative Care Program (POPC). They were given a letter of invitation explaining the study and referred to the clinical research assistant who then provided further information. Those who expressed interest in participating were screened by the clinical research assistant to confirm eligibility, and then voluntary written informed consent was obtained.

Eligibility required patients with metastatic cancer, who were English speaking, ≥18 years old, in a romantic partnership of ≥1 year, endorsing marital distress (Revised Dyadic Adjustment Scale (RDAS) ≤ 47) [53] in minimally one partner, not currently in couple therapy, and a patient Karnofsky Performance Status score [54] of ≥60. Patients or caregivers were excluded if one had any significant cognitive deficits (Short Orientation-Memory-Concentration Test; cut-off score of <20 is equivalent to >10 errors) [55], if the patient was too ill to participate or if either partner had major psychiatric illness (e.g., psychosis, schizophrenia).

Enrolment and randomization procedure

The study was a two-group RCT. All participating couples were first mailed the baseline questionnaire package (T0). Once completed and returned, the couples were randomly assigned to EFT (INT) or standard care (CTL) in a 1:1 ratio by the University of Toronto's statistician for this study (BT). Intervention assignment was stratified by patient sex. Recruitment and randomization continued until 22 couples were assigned to each treatment group. Because of the obvious variation in the treatment groups, participants were not blinded to their assignments. However, the study personnel who entered and checked the data were blinded to condition assignment.

Description of treatment groups

Couple participants in the INT group (n = 22) received an eight-session EFT intervention [47, 56] adapted for use with couples where one partner has advanced metastatic cancer [3]. We modified and manualized EFT to address the particular issues that challenge such couples [35]. Our primary goals for this intervention were to facilitate marital relationships by changing habitual and distressing patterns of interaction, to increase mutual understanding and emotional engagement, and to strengthen the marital bond. Themes that were interwoven throughout the modified EFT have been described in case studies [32, 35, 36] and influenced by others' works [e.g., [3, 57, 58]] and included some of the following [e.g., [16, 32, 57]]: impact of a terminal diagnosis and effective communication; control of physical symptoms and communication with their medical team; assessment of grief versus clinical depression; the couple's need for decision making; fears, the couple's changing perceptions of time, and how to spend it in a meaningful way; physical changes and decline in the patient that result in further role changes; importance of being attuned to oneself and partner, the process of reviewing one's life; and existential issues that include meaning, value, and spirituality.

One-hour weekly couple sessions (M = 7.7, SD = 0.94, median = 8, mode = 8) were delivered by one EFT-trained psychologist (LM) and occurred over a 2–3-month period. Sessions took place at PMH clinical offices or at alternative locations in four of the INT group couples, including home (n = 2) and/or inpatient hospital room (n = 2), to accommodate needs and to maximize adherence.

Couple participants in the CTL group (n = 20) received standard care provided by the POPC department. Social work consultations account for two-thirds of the psychosocial care, and usually involve provision of practical or instrumental care or supportive interactions drawing on varying theoretical approaches such as cognitive behavioral therapy, psychoeducational, and eclectic. Systems theory is a core foundation underpinning social work thinking. The goal is to alleviate emotional distress in patients, partners, or family members. Of the remainder referred to psychiatry or psychology, theoretical orientation depends upon the individual clinician and may be cognitive behavioral therapy, interpersonal therapy, supportive-expressive, or psychodynamic. Psychiatrists render pharmacotherapeutic intervention where needed. Patients and their partners facing metastatic cancer may be followed weekly, bi-weekly, or monthly until end-of-life, determined by patient/clinician assessment of need. Following an initial consultation of all couple participants by social workers (n = 10), psychiatrists (n = 5), and psychologists (n = 3), participants were provided with therapeutic care. Individual therapy was then provided in 20% (n = 5) of the couples (M = 6, median = 7, mode = 7), with 80% receiving couples' therapy. The individual therapists utilized varied approaches. The number of sessions in the CTL group (n = 20) ranged from two to eight (M = 5.22, SD = 2.29, median = 6, mode = 6), and they were provided in clinical offices or, in the case of three couples (16.6%), their inpatient rooms.

Assessment of primary and secondary outcomes

Couple participants in both study groups completed all study measures at baseline (T0) (before random assignment), immediately post-intervention (T1), and at 3-month post-intervention follow-up (T2). The primary time point of interest was post-intervention (T1), and the primary outcome was marital functioning.

Primary outcome

Marital functioning

The Revised Dyadic Adjustment Scale [53, 59] is a standardized and validated 14-item self-report that is widely used to evaluate both individual and dyadic adjustments in distressed relationships. Total scores range from 0 to 69 with a reliable cut-off score of 48 to classify individuals and/or couples as maritally distressed (≤47) or non-distressed (≥48) [53]. Lower scores on this measure reflect higher marital distress. The RDAS had a Cronbach's α = 0.90 in the current study. The validity of the RDAS in cancer populations has been established [37].

Secondary outcomes

Psychological symptoms

The Beck Depression Inventory-II (BDI-II) [60] is a 21-item self-report measure assessing intensity of depressive symptoms (Cronbach's α = 0.92) and has been used extensively in cancer populations [61-63]. Total scores range from 0 to 63, with higher scores reflecting higher levels of depression. A cut-off score of ≥15 has been shown to have high sensitivity and specificity for symptoms of major depression in cancer populations (>90%) [63]. The Beck Hopelessness Scale (BHS) [64] is a 20-item true/false scale developed to quantify hopelessness and negative expectancies. Total scores range from 0 to 20, with higher scores reflecting increased hopelessness (Cronbach's α = 0.88). The BHS has been used in terminally ill cancer populations [65, 66], with a cut-off score of ≥8 recommended for this population [65, 67].

Caregiver burden

Two subscales (Demand/Difficulty) of the Caregiver Burden Scale were used to access objective and subjective caregiving burden (caregivers only) [68]. The Demand subscale measures how much time a caregiver devoted to 14 caregiving tasks (Cronbach's α = 0.92). Ratings are made on a five-point scale ranging from 1 (little/no time) to 5 (a great deal of time). The Difficulty subscale assesses subjective caregiving burden and measures how difficult caregivers appraised the undertaking of the 14 different tasks. Ratings are made on a five-point scale ranging from 1 (little/no difficulty) to 5 (a great deal of difficulty). Scores on each subscale range from 0 to 70 with higher scores reflective of greater objective and/or subjective caregiver burden (Cronbach's α = 0.93).

Patient's perspective of caregiver empathic behavior

The 10-item Relationship-Focused Coping Scale (RFCS) [69] was used to measure the cognitive-affective and behavioral aspects of empathic responses of the caregiver (patient only). Ratings are made on a five-point scale ranging from 0 (does not describe my caregiver very well) to 4 (describes my caregiver very well). Scores range from 0 to 40 with higher scores indicative of higher use of empathic responding (Cronbach's α = 0.89).

Assessment of demographic and medical data, adherence, and adverse events

Clinical and demographic characteristics were obtained at baseline through a brief demographic questionnaire. Patient medical data were then confirmed through medical chart review. The history of marital distress pre-dating the cancer diagnosis was obtained during the initial couple assessment. Efforts were made to retain participants (i.e., telephone follow-up, appointment time, and location flexibility). Recruitment continued until an adequate sample size was reached to achieve statistical power.

Clinicians involved in both study groups monitored adherence, as well as the occurrence of adverse events for the purposes of tracking and addressing any that occurred. Twenty-five per cent of the INT groups' sessions were randomly selected for audio taping (determined by University of Toronto statistician). These recordings were reviewed by an approved EFT-trained psychologist (SJ), founder of the Ottawa Family and Couple Institute's Externship program and head of the International Center for Excellence in EFT to ensure treatment fidelity.

Data analyses

On the basis of the RDAS scores from our pilot study, we calculated that 36 couples (18 couples in INT and 18 couples in CTL) were required to detect an intervention effect of seven points (SD = 8.6) on the RDAS scale with 0.90 power [3].

Descriptive and inferential statistics were completed using sas V 9.2 (SAS Institute, Cary, NC, USA). All statistical tests were two-sided with α = 0.05. We calculated means, standard deviations, minima, and maxima for primary and secondary outcomes at all three periods and conducted comparative analyses of the INT and CTL groups at baseline using two-tailed t-tests. In cases where the variances of the two groups were not significantly different at baseline (T0), we used the pooled t-test method. When group variances were statistically different, we used the Satterthwaite method.

To test the effectiveness of the INT on marital distress (primary outcome), psychological distress, empathic caregiving, and caregiver burden (secondary outcomes), we performed three-way analyses of covariance (ANCOVAs) (PROC GLM in SAS) with T0 scores as covariates and T1 outcome measures as dependent variables. The main effects tested were treatment group (INT, CTL), patient status (patient, caregiver), and sex (female, male). All possible two-way interactions with treatment were tested. To determine if effects were maintained, the same statistical tests were used to examine 3-month follow-up data (outcomes at T2 using T1 values as covariates). We also used multilevel modeling to allow for the possibility of a negative correlation between members of a couple.

In addition, following the recommendation of Kendall and Grove [70] (i.e., clinical significance of treatment effects on the primary outcome), we calculated the percentage of patients and caregivers in both groups who were above and below clinically significant cut-offs on the RDAS scale [53, 59]. The proportion of patients and caregivers in the INT and CTL groups who improved, remained stable (stable non-distressed and stable distressed), or declined from T0–T1 and T0–T2 were then examined using chi-squared analyses. Effect sizes (Cohen's d) were calculated for significant findings at the primary endpoint (T1) using [d = (math formula)/s] where math formula is the mean for the EFT group at T1, math formula is the mean of the CTL group at T1, and s is the pooled standard deviation for the two groups.

Results

Recruitment occurred between January 2008 and June 2009. Of 45 potentially eligible couples, three (6.6%) couples refused to participate, two due to lack of interest, and one due to unexpected disease progression (see Figure 1). In total, 42 patients were enrolled into the study, along with their caregivers, defined as the primary caregiver of the patient and confirmed by both couple members. Twenty-two couple participants were randomly assigned to the INT group and 20 to the CTL group. Two couple participants (10%) in the CTL group dropped out after baseline data collection; one patient unexpectedly died of cancer, and one patient had progressive disease and was too ill to continue (both couples were followed by psychiatrists in POPC). The remaining 18 couples (90%) in the CTL group completed questionnaire packages at all three data collection points. Four patients (18.2%) in the INT group died prior to collection of post-intervention follow-up (T2). The INT group therefore had an 82% completion rate for the questionnaire packages at all three data collection points. Groups were balanced on baseline outcome measures except for the measure for depression (BDI-II), with the INT group having a significantly higher mean than the CTL group (M = 17.05, SD = 7.7, range 3–33 vs M = 10.90, SD = 6.87, range 1–28, respectively; t(78) = 3.72, p = 0.0004).

Figure 1.

Consort diagram of the inclusion, random assignment and the retention of the intervention group (EFT) and control group study participants. POPC, Psychosocial Oncology and Palliative Care; SOMC, Short Orientation-Memory-Concentration Test; RDAS, Revised Dyadic Adjustment Scale; KPS, Karnofsky Performance Status; EFT, Emotionally Focused Therapy; T0, baseline; T1, corresponds to post-intervention (eight sessions) of EFT (INT); T2, corresponds to 3-month post-intervention follow-up

Sample characteristics

Demographic and clinical characteristics are listed in Table 1. Couple participants who reported a history of marital conflict or difficulty (35.7%) were distributed similarly among INT (53.3%) and CTL (46.6%) study groups. Most participants (92.5%) had an income of ≥$60,000 per year. The most common types of cancer among patients are listed in Table 1. As expected, because of differences in the EFT protocol and standard care, a significant difference was found in the mean number of sessions attended by the INT group versus the CTL group (M = 7.73, SD = 0.94 vs M = 5.22, SD = 2.29, respectively, p ≤ 0.001). These variations were adjusted for in the statistical analyses.

Table 1. Baseline demographics and clinical characteristics of patients and caregivers by group assignment
CharacteristicPatientCaregiver
Intervention (EFT: n = 22)Control (SC: n = 20)p for differenceIntervention (EFT: n = 22)Control (SC: n = 20)p for difference
  1. EFT, Emotionally Focused Therapy; SC, standard care; M, mean; SD, standard deviation; SOMC, Short Orientation-Memory-Concentration Test; KPS, Karnofsky Performance Status; CNS, central nervous system; GI, gastrointestinal; Gyne, gynecological; GU, genito-urinary.

AgeM = 51.83M = 49.450.503M = 48.82M = 50.890.570
SD = 9.10SD = 12.42 SD = 13.38SD = 9.27
Duration of partnershipM = 22.44M = 19.140.404M = 19.18M = 22.440.413
SD = 12.00SD = 12.61 SD = 12.69SD = 12.01
Number of childrenM = 1.83M = 1.410.140M = 1.45M = 1.60.481
SD = 0.78SD = 0.96 SD = 0.91SD = 0.97
Gender
Male10 (24.0%)9 (21.0%)0.94912 (29.0%)11 (26.0%)0.949
Female12 (29.0%)11 (26.0%) 10 (24.0%)9 (21.0%)
Language
English22 (52.0%)16 (38.0%)0.20022 (52.0%)17 (40.0%)0.346
French01 (2.5%) 01 (2.5%) 
Spanish01 (2.5%) 01 (2.5%) 
Other0(2.5%)  1 (2.5%)
Education
Graduate4 (9.5%)4 (9.5%)0.5345 (11.9%)5 (11.9%)0.754
Undergraduate10 (23.8%)7 (16.6%) 4 (9.5%)3 (7.1%) 
College3 (7.2%)4 (9.5%) 5 (11.9%)4 (9.5%) 
Trade school1 (2.4%)2 (4.8%) 3 (7.1%)3 (7.1%) 
High school3 (7.1%)1 (2.4%) 3 (7.1%)2 (4.8%) 
Grade school1 (2.4%)0 2 (4.8%)1 (2.5%) 
Professional school02 (4.8%) 02 (4.8%) 
SOMCM = 25.89M = 26.050.816M = 26.55M = 26.780.686
SD = 2.42SD = 1.81 SD = 2.15SD = 1.22 
KPSM = 71.11M = 69.090.547   
SD = 11.82SD = 9.21    
Diagnosis
Blood4 (9.5%)3 (7.1%)0.113   
Breast2 (4.8%)8 (19.0%)    
CNS1 (2.4%)4 (9.5%)    
GI1 (2.4%)1 (2.4%)    
Gyne4 (9.5%)2 (4.8%)    
Lung2 (4.8%)0    
GU2 (4.8%)2 (4.8%)    
Head and neck4 (9.5%)0    
Skin1 (2.4%)0    
Neuroendocrine1 (2.4%)0    
Years since diagnosisM = 3.94M = 1.970.054   
SD = 3.83SD = 2.37   
Treatment active
Yes7 (16.6%)5 (11.9%)0.271   
No15 (35.7%)15 (35.7%)   

Changes in primary and secondary outcomes at post-intervention

Table 2 presents the means of the primary and secondary outcome measures at each of the time points.

Table 2. Means and standard deviations at three points of measure for six outcome measures
OutcomeT0T1T2
INTCTLINTCTLINTCTL
M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)
  1. T0, baseline; T1, post-intervention; T2, 3-month post-intervention follow-up; INT, intervention group (received EFT); CTL, control group (received standard care); M, mean; SD, standard deviation; PT, patient; CG, caregiver; RDAS, Revised Dyadic Adjustment Scale; BDI-II, Beck Depression Inventory-II; BHS, Beck Hopelessness Scale; RFCS, Relationship-Focused Coping Scale; CBS1, Caregiver Burden Scale, Time subscale; CBS2, Caregiver Burden Scale, Difficulty subscale; EFT, Emotionally Focused Therapy.

  2. Sample sizes for T0 and T1 are 44 for INT (EFT) and 36 for CTL (standard care). Sample sizes for T2 are 36 INT (EFT) and 36 for CTL (standard care). Possible ranges of scales are shown next to the outcome in the left column.

RDAS, 0–69
PT43.14 (4.8)44.22 (7.6)56.27 (4.6)43.44 (10.3)55.39 (6.3)44.17 (10.2)
CG46.41 (7.0)42.94 (7.2)54.32 (4.6)42.39 (6.9)54.72 (5.8)44.56 (10.3)
BDI-II, 0–63
PT17.86 (7.7)13.11 (6.9)15.00 (6.4)12.94 (7.6)15.89 (11.7)14.33 (10.9)
CG16.23 (7.9)8.69 (6.3)13.82 (8.8)10.33 (7.7)13.33 (8.1)9.67 (7.3)
BHS, 0–20
PT6.69 (5.7)4.83 (4.3)7.78 (3.9)5.62 (5.4)6.95 (5.8)5.78 (6.0)
CG6.32 (4.7)4.56 (3.6)6.64 (5.1)4.72 (3.9)6.94 (5.4)5.39 (3.9)
RFCS, 0–40
PT30.05 (8.1)30.78 (8.8)33.73 (5.4)30.94 (9.6)32.67 (10.0)29.00 (10.1)
CBS1, 1–70
CG2.96 (0.9)2.52 (1.2)2.85 (1.0)2.60 (1.0)2.58 (1.2)2.70 (1.3)
CBS2, 1–70
CG1.97 (0.7)1.81 (0.8)1.90 (0.8)2.05 (1.0)1.84 (0.8)1.98 (1.1)

Marital functioning (primary outcome)

At T1 there was a significant difference on the RDAS (p < 0.0001), with the INT having higher mean scores (better marital functioning) than the CTL group. The actual calculated effect size for this difference was Cohen's d = 1.00, which is in the large range. We found that patients had marginally higher mean scores (i.e., experienced more marital distress) compared with caregivers in both the INT and the CTL groups (see Table 3). T0 scores (baseline) were highly correlated with T1 scores. There was no significant effect for sex, and there were no significant interactions for primary or secondary outcomes. The multilevel modeling analysis, with couple as one of the levels, provides the same conclusions, except that the effect of patient/caregiver status is even stronger than that reported in Table 3 (F(1,38.5) = 6.05, p = 0.02).

Table 3. Results of three-way analyses of covariance for the primary and secondary outcomes at T0 and T1
Measure at T1TreatmentMeasure at T0Patient statusSexLeast square means
F-valuep-valueF-valuep-valueF-valuep-value
  1. T0, baseline; T1, post-intervention; RDAS, Revised Dyadic Adjustment Scale; BDI-II, Beck Depression Inventory-II; BHS, Beck Hopelessness Scale; RFCS, Relationship-Focused Coping Scale; CBS1, Caregiver Burden Scale, Time subscale; CBS2, Caregiver Burden Scale, Difficulty subscale; ns, not significant.

RDAS total score80.68<0.000135.38<0.00014.260.04nsINT = 54.91; CTL = 43.60; PT = 50.50; CG = 48.01
BDI prorated total score0.550.4630.43<0.0001nsnsnsns
BHS prorated total score1.380.2438.87<0.0001nsnsnsns
RFCS total score5.840.0271.67<0.0001nsnsnsINT = 34.13; CTL = 30.74
CBS10.020.8827.78<0.0001nsnsnsns
CBS22.880.0949.69<0.0001nsnsnsns

Psychological symptoms

After controlling for baseline differences (p = 0.46), BDI-II did not remain significant in the INT group compared with the CTL group (p = 0.03). There was no difference in BHS between treatment groups.

Patient-perceived empathic behavior and caregiver burden

Patients in the INT group had significantly higher mean scores at T1, indicating higher patient-perceived caregiver empathic behavior, compared with patients in the CTL group (p = 0.02). The actual calculated effect size for this difference was Cohen's d = 1.00, which is in the large range. There was also no significant difference (p = 0.09) between groups in caregivers subjective difficulty in caregiving for their ill spouses.

Clinical significance of treatment effects on marital distress

We examined the proportion of patients and caregivers in each group who met criteria for a clinically significant improvement in marital distress across time points (see Figure 2). On the basis of individual RDAS scores, the proportion of INT patients versus CTL patients who were clinically improved was 3.25:1 [percentage: 91% of INT patients vs 28% of CTL patients (χ2 = 16.8, p < 0.0001)] and the proportion of INT versus CTL caregivers who were clinically improved was 4.54:1 [percentage: 50% of INT caregivers vs 11% of CTL caregivers (χ2 = 6.825, p = 0.01)] and moved from marital distress (RDAS ≤ 47) to non-marital distress (RDAS ≥ 48) from T0 to T1.

Figure 2.

Bar graphs showing the percentage of participants who had a change in score from baseline (T0) to 3-month post-intervention (T2) for marital distress. EFT, Emotionally Focused Therapy; SC, standard care; PT, patient; CG, caregiver

Sustainability of the intervention effects on marital distress, psychological distress, patient-perceived caregiver empathic responses, and caregiver burden

Using three-way ANCOVAs to control for baseline scores, we examined whether the INT and CTL groups differed on any of the outcome variables at T2. As at T1, the INT group had significantly higher RDAS scores compared with the CTL group (least square means: 55.04 vs 44.40, respectively, p ≤ 0.001). In addition, patients in the INT group had higher scores for RFCS compared with patients in the CTL group (least square means: 33.84 vs 28.25, p = 0.028). There were no significant differences for Caregiver Burden Scale, BDI-II or BHS scores.

We found that there were no significant changes from T1 to T2 in overall mean scores, suggesting stability of scores between post-intervention and 3-month post-intervention follow-up (see Table 4). In addition, a total of 89% of patients in the INT group versus 50% of the patients in the CTL group remained above the cut-off (non-distressed) on the RDAS scale at T2 (p = 0.01); 89% of INT caregivers versus 39% of CTL caregivers remained above the cut-off (p = 0.001).

Table 4. Results of three-way analyses of covariance for the primary and secondary outcomes at T2
Measure at T2TreatmentMeasure at T1Patient statusSex
F-valuep-valueF-valuep-value
  1. T1, post-intervention; T2, 3-month post-intervention follow-up; RDAS, Revised Dyadic Adjustment Scale; BDI-II, Beck Depression Inventory-II; BHS, Beck Hopelessness Scale; RFCS, Relationship-Focused Coping Scale; CBS1, Caregiver Burden Scale, Time subscale; CBS2, Caregiver Burden Scale, Difficulty subscale; ns, not significant.

RDAS total score0.110.7491.08<0.0001nsns
BDI prorated total score0.170.6833.62<0.0001nsns
BHS prorated total score0.110.7578.35<0.0001nsns
RFCS total score0.090.7665.41<0.0001nsns
CBS11.480.2348.55<0.0001nsns
CBS20.020.8712.000.0015nsns

The least square means from the ANCOVA (i.e., treatment means adjusted for differences in sex and patient status) were examined and showed the difference between the two groups. Significant differences between EFT and standard care were found on the RDAS (+11.31 with a 95% CI, p ≤ 0.0001) and the RFCS (+3.44 with a 95% CI, p = 0.02).

Discussion

In this trial, we found that patients with metastatic cancer and their spouses who received a modified EFT demonstrated statistically and clinically significant improvements in marital functioning compared with those couples who received standard care. The change in treatment mean scores on the RDAS (+11.34), which were maintained at 3-month follow-up exceeded the minimum change score difference of seven points that we found in our pilot work [3]. None of the INT group declined, whereas one-third of the CTL group deteriorated in marital functioning at post-treatment.

Encouragingly, when comparing the proportion of patients with improved marital functioning in the INT versus CTL group, the proportion was 3.25:1 (i.e., 91% of INT patients vs 28% of CTL patients, respectively), whereas the proportion of caregivers in INT versus CTL was 4.54:1 (i.e., 50% of INT caregivers vs 11% of CTL caregivers, respectively). This degree of improvement is similar to other reported empirical EFT studies of couples in the non-cancer literature, which show an improvement in marital functioning (range 70–73%) [48]. The significant improvement may in part be due to our eligibility requirement of marital distress. Furthermore, this study is unique, compared with other couple interventions reported in the cancer literature [13, 17, 32, 40, 42, 43], in that a validated measure of marital functioning was employed as a primary outcome.

Additionally, our results showed that patients in the INT group reported significant improvement in their assessment of perceived caregivers' empathic behaviors when compared with the CTL group. It is possible that EFT increased the patients' perception of feeling heard, understood, accepted, and cared about. As expected, these findings suggest that the beneficial effects of EFT on both RDAS and RFCS scores occurred primarily during and at the end of the intervention (T1), being maintained at 3-month post-intervention as well. Receiving EFT may have helped the caregiver in trying to experience the patients' feelings and learning how to best help and comfort the patient as well as improving overall responsiveness.

The high participation rate, retention of couples in the study, and their compliance in regard to completion of questionnaires support the effectiveness of a couple-based intervention tailored for the advanced cancer population.

The study was limited by the relatively small sample size. Furthermore, the results are limited to couples who were referred by their clinical team and met the RDAS cut-off for marital distress. Moreover, the CTL group intervention varied on the basis of the approach considered most suitable by the clinician. Although this study has demonstrated that EFT can provide significant benefits to couples compared with standard care alone, future studies with more equitable control groups of standardized evidence-based couple therapies are now needed. It is unknown if further research with larger samples will demonstrate significant effects on secondary outcomes, such as depression [71]. Our findings in this study are in contrast to our pilot work where we found a significant reduction in mean BDI-II scores from T0 to T3 using EFT for a similar group of participants [3]. The lack of an effect of EFT on hopelessness in both the pilot study and this study is similar. It is possible that such an effect may occur when participants are selected for treatment on the basis of elevations in these symptoms (e.g., EFT studies, see [48]). Larger RCTs are now needed to confirm our findings. Additionally, qualitative studies may help to further understand the mechanism by which such an intervention may act as well as what may assist in the reduction of symptoms of depression and hopelessness.

In conclusion, our study provides evidence that this newly developed EFT intervention is beneficial in the terminal cancer population, improving both the quality of marital functioning and the patients' perceived experience of being empathically understood by their caregivers. The findings also suggest that couple-based interventions have the potential to result in relational growth during a very challenging and difficult time. The impact of terminal cancer on couples' relationship requires attention in future studies in palliative care programs [3, 8, 35, 36].

Conflict of interest

The authors have declared that there is no conflict of interest.

Author contributions

Conception and design: Linda Mclean, Gary Rodin, Tara Walton, Mary Jane Esplen, Jennifer Jones

Administrative support: Linda McLean, Tara Walton

Collection and assembly of data: Linda McLean, Tara Walton, Barbara Thomson

Data analysis and interpretation: Linda McLean, Tara Walton, Barbara Thomson, Gary Rodin, Jennifer Jones

Manuscript writing: Linda McLean, Tara Walton, Gary Rodin, Mary Jane Esplen, Jennifer Jones

Final approval of manuscript: Linda McLean, Tara Walton, Mary Jane Esplen, Gary Rodin, Jennifer Jones

Ancillary