Secondary Logo

Journal Logo

Original Article

We think we can: development of the Dyadic Efficacy Scale for Cancer

Brosseau, Danielle C.a,∗; Braeken, Johanb; Carmack, Cindy L.c; Rosberger, Zeevd,e,f; Körner, Annetta,d,e,f,g

Author Information
Journal of Psychosocial Oncology Research and Practice: October-December 2021 - Volume 3 - Issue 4 - p e066
doi: 10.1097/OR9.0000000000000066

Abstract

Measurement advances are needed for research to keep pace with the growing conceptual acceptance that patients’ and partners’ responses to a cancer diagnosis occur within a dynamic interdependent system.[1,2] As conceptualized in Berg and Upchurch's developmental-contextual model,[3] this approach involves the adoption of a dyadic lens when considering the appraisal, coping and adjustment of patients and their partners following a cancer diagnosis. In keeping with the need for dyadic conceptualizations, Sterba et al[4] expanded on existing individually focused models of self-efficacy by developing a model of dyadic efficacy among couples (N = 190) managing rheumatoid arthritis. In the context of coping with a chronic illness, dyadic efficacy was defined as “an individual's perceptions of confidence about his or her shared ability with a partner to manage (illness)-related problems.”[4] Sterba et al discovered that higher dyadic efficacy was associated with fewer depressive symptoms for both women with rheumatoid arthritis and their husbands. In this same study, dyadic efficacy was also positively associated with relationship satisfaction and quality for both patients and partners. Dyadic efficacy rests on systemic principles that highlight the relational embeddedness of confidence appraisals. For individuals embedded in a patient-partner dyad, understandings of efficacy need to not only acknowledge confidence in the individual's ability to cope with cancer and its treatment but also assess efficacy for the abilities of the couple as a unit.

To our knowledge, dyadic efficacy has not yet been investigated among cancer patients or their partners. Researchers have, however, examined self-efficacy generally and with respect to several processes of adaptation to cancer including communication, symptom management, emotional functioning, and coping.[5,6] The term self-efficacy was first coined by Bandura[7] and refers to individuals’ “judgments of their capabilities to organize and execute courses of actions required to attain designated types of performances.”[8] Higher self-efficacy expectations have predominantly been associated with lower levels of psychological distress and higher ratings of quality of life among patients and their partners following a cancer diagnosis.[9,10] Researchers have also begun to demonstrate that psychosocial interventions can be used to improve cancer patients’ and their partners’ self-efficacy.[11–13] More specifically, researchers have demonstrated that educational or skills-based interventions could be used to improve spousal caregivers’ confidence to support their partner, care for themselves[11] and to manage stress and other changes associated with cancer[12]. Northouse et al[13] increased the self-efficacy of both patients with advanced cancer and their family caregiver using a dyadic intervention which provided information and support.

Applying Sterba et al's work[4,14] and the efficacy component of social cognitive theory[8] to the cancer context, cancer-related dyadic efficacy represents an individual's judgment of his or her joint capability with a partner to manage challenges posed by cancer and its treatment. An empirically based measure of dyadic efficacy will enable examinations of the role dyadic efficacy plays in couples’ relational and psychological outcomes following a cancer diagnosis.

Objectives and hypotheses

The objective of this study was to evaluate the Dyadic Efficacy Scale for Cancer (DESC). This objective was accomplished by: conducting item-level analyses, confirming the factor structure, and presenting evidence of the scale's reliability and validity. It was anticipated that cancer-related dyadic efficacy would be negatively correlated with psychological distress, and positively correlated, but not synonymous with, relationship functioning, and self-efficacy. All hypotheses relate to both patient and partner data.

Methods

Scale development

The DESC was developed using an exploratory sequential mixed methods design. Gehlbach and Brinkworth's[15] and DeVellis’[16] guidelines informed the item and scale development procedures which included 3 phases: construct and content development, the development of an item pool that was evaluated through expert review and pilot testing, and administration of the scale to a development sample for psychometric testing. Phase 3, the psychometric evaluation of the DESC, is the focus of the present paper.

Phases 1 and 2, encompassing construct and item pool development were completed through a multi-step process of consultation with lay and content experts. A detailed account of the conceptual grounding for this scale and procedures related to item development and evaluation have been provided by Brosseau et al[17] elsewhere. To sum, the exploratory phase of the study involved focus group (N = 5) consultations with lay experts—individuals diagnosed with cancer and their partners (N = 17)—to establish a description of dyadic efficacy as it applied to the process of coping with cancer and its treatment and to identify content domains for the assessment of cancer-related dyadic efficacy. Reflexive thematic analysis[18] was used to identify 3 main conceptual themes: cancer-related dyadic efficacy is multidimensional, consistent with relational functioning and distinct from self-efficacy. Three main content themes were also identified. Items were developed to assess cancer-related dyadic efficacy across the following 3 content domains: dyadic efficacy for managing illness intrusions, the emotional response of the patient and the partner, and communication and care for children. The pool of scale items was then evaluated for relevance, clarity, and content breadth by experts in psychosocial oncology (N = 5) and by a second sample of individuals diagnosed with cancer and their partners (N = 41). Ethical approval for this multi-site study (MP-05-2015-254, 14-078) was obtained from the research ethics committee of le center intégré universitaire de santé et de services sociaux du Centre-Ouest-de-l’Île-de-Montréal, Québec, Canada.

Participants

Two hundred eighty-seven dyads were represented in the sample (Table 1). Although efforts were made to recruit complete dyads, eligible patients and partners were able to participate alone. Data were received from 191 complete dyads leaving 96 dyads represented by a patient (n = 70) or a partner (n = 26) alone. In sum, data from 261 patients and 217 partners were available for analysis. Four percent of the participating couples represented homosexual partnerships (n = 4 male/male, n = 3 female/female). The sexual orientation of patients and partners who participated alone was unknown. Eligible patient participants were diagnosed with cancer within the past 2 years, were presently receiving treatment (or within 6 months of completion) and were involved in a committed relationship of at least one year (ie, dating, common-law, married). Partners meeting the relationship criterion were eligible provided their spouse met medical criteria. All eligible participants were 18 years of age or older (patient age, M = 60 years, range = 18–93 years; partner age, M = 59 years, range = 19–85 years), able to read and comprehend English, and able to provide informed consent.

Table 1 - Demographic and medical characteristics of the sample presented by role.
Patients (n = 261) Partners (n = 217)


Characteristic n % n %
Sex
 Female 149 57 98 45
 Male 109 42 119 55
Marital status
 Married 202 77 171 79
 Common-law 36 14 32 15
 Cohabiting 12 5 6 3
 Dating 8 3 7 3
Highest level of education
 Primary school (grades 1–6) 2 1 4 2
 Secondary school (grades 7–11) 53 20 50 23
 Vocational or technical training 78 30 59 27
 University degree 123 47 102 47
Employment status
 Full-time work 50 19 76 36
 Part-time work 21 8 18 8
 Leave from work 35 13 8 4
 Retired/not employed outside home 105 40 101 47
 Unemployed 9 3 7 3
 Disability assistance 35 13 3 1
 Social assistance 1 .5
Annual household income
 <$40,000 51 19 39 20
 >$40,000–< $60,000 36 14 30 16
 >$60,000–< $80,000 32 12 29 15
 >$80,000–< $100,000 35 13 30 16
 >$100,000 75 29 63 33
Ethnicity
 White 214 82 187 87
 Chinese 2 1 1 .5
 Asian (South/Southeast/West) 5 2 6 3
 Black 9 3 6 3
 Indigenous, First Nations, Inuit & Métis 2 1 1 .5
 Filipino 7 3 2 1
 Latin American 6 2 2 1
 Arab 2 1 3 1
 Korean 1 1 1 .5
 Other 12 5 7 3
Cancer type
 Breast 84 32 (64) (30)
 Gastrointestinal 44 17 (34) (16)
 Blood 26 10 (26) (12)
 Lung 20 8 (19) (9)
 Prostate 18 7 (17) (8)
 Gynecological 16 6 (14) (7)
 Head and neck 11 4 (8) (4)
 Melanoma 11 4 (8) (4)
 Kidney 5 2 (4) (2)
 Bladder 3 1 (2) (1)
 Brain 2 1 (2) (1)
 Other 11 4 (8) (4)
Stage at diagnosis
 0 10 4 (5) (2)
 I 23 9 (24) (12)
 II 49 19 (30) (15)
 III 43 17 (36) (18)
 IV 51 20 (51) (25)
 Unknown/alternate staging 75 29 (59) (29)
Treatment Received
 Chemotherapy 196 75 (159) (74)
 Radiation therapy 127 49 (100) (47)
 Surgery 93 36 (75) (35)
 Transplant 3 1 (5) (2)
 Hormonal therapy 47 18 (34) (16)
 Other 39 15 (31) (14)
Category totals may not equal 100% due to missing values and/or rounding. Partner data displayed in parentheses indicate medical information about the partner's spouse.

Procedures

A multimodal recruitment strategy was used to facilitate paper-and-pencil or online survey completion. The most effective recruitment strategy occurred face-to-face in clinic and treatment waiting rooms at 2 urban cancer centers in Québec, Canada. Of the 612 eligible dyads (or patient/partner representative of a dyad) recruited in person, 45% completed participation in the study (Fig. 1). This response rate was well within the range (25%–90%) of response rates from similar populations reported in a recent systematic review.[19] It was not possible to compute a comprehensive response rate because the number of eligible participants who received an email invitation, viewed a paper advertisement, or were informed about the study by a health care professional was unknown. Data were not collected regarding the reasons why 55% of those who were eligible and accepted a participation package did not go on to complete the study. Some of these potential participants were never reached upon follow-up, others expressed an ongoing intent to complete participation and others indicated that they were no longer interested.

F1
FIGURE 1:
Face-to-face recruitment rate. Recruitment was tracked based on dyads. Data totals indicate dyads represented by both members or a solo patient or a solo partner participant. aAn additional 13 dyads consented online and 10 of these completed the study. These dyads were not included in the totals displayed above to avoid inflating the response rate for face to face recruitment.

Although waiting room recruitment was the most effective strategy employed, 23% of individuals approached in the waiting room declined to be screened. It is possible that this result reflects the broad recruitment strategy used. Research assistants approached individuals waiting for a clinic or treatment appointment, provided information about the research project and asked if the individual was interested in learning more about the study. Reasons for declining screening were not systematically recorded but included factors such as obvious ineligibility (ie, the individual approached was not a patient or a partner, the patient was not partnered) or a lack of interest in research on couples coping with cancer. Potential participants who expressed interest in the study were screened for eligibility and provided with a participant package (consent forms, surveys, follow-up survey invitation and a pre-addressed stamped return envelope). Up to 3 follow-up calls/emails were attempted (1 week apart) if questionnaires were not received after 1 week. Participants invited via email were provided with a URL link to the online survey and instructed to complete eligibility screening, consent and the study questionnaires using the web-based Qualtrics software.[20]

Measures

All participants first completed a demographic and medical questionnaire followed by the dyadic efficacy scale. To limit the influence of carry over effects, a balanced Latin Square design was used to establish an incomplete counterbalanced order (5 options) for the remaining measures.

Demographics and medical history

The demographic and medical questionnaire included questions about participants’ basic demographics (ie, age, sex, ethnicity), relational and family factors (ie, relationship status, relationship length, parental status), education and employment (ie, household income, level of education, employment status), and physical and psychiatric history (eg, cancer diagnosis, cancer treatments, psychiatric diagnosis).

Dyadic efficacy

Patients and partners were presented with 50 parallel items. Each item began with the following item stem, I am confident that we can work together as a team to. .. . The items were presented in 4 sections: dyadic efficacy for managing illness intrusions (related to the patient's physical experience, the medical system, social life, couple life and ongoing responsibilities), patient affect, partner affect, and childcare. It was anticipated that the scale would elicit a general factor due to the theoretical proposal that cancer-related dyadic efficacy rests on the couples’ relational functioning and multidimensionality reflective of the 4 domains queried. Participants responded using a 9-point numeric scale with descriptive labels identifying each end point, 1 (not at all confident), 9 (completely confident) and the midpoint, 5 (somewhat confident). Higher scores reflected higher dyadic efficacy.

Self-efficacy

Patients’ self-efficacy was assessed using the Cancer Behavior Inventory—Brief (CBI-B).[21,22] The CBI-B is a unidimensional scale containing 14 items that assess 4 domains of cancer-related self-efficacy: maintaining independence and positive attitude, participating in medical care, coping and stress management, and managing affect. Cronbach α = .91. Partners’ self-efficacy was assessed using the Caregiver Inventory (CGI).[23] The CGI consists of 21 items related to self-efficacy for managing medical information, caring for the patient, caring for oneself, and managing difficult interactions and emotions. Cronbach α = .93. Higher scores indicated higher self-efficacy for both measures.

Relationship functioning

The Revised Dyadic Adjustment Scale (RDAS) was used to assess relationship functioning.[24,25] The RDAS includes 14 items assessing dyadic consensus, dyadic satisfaction, and dyadic cohesion. Higher scores reflected higher functioning. Patients’ Cronbach α = 87; partners’ Cronbach α = 89.

Psychological distress

Participants’ psychological distress was measured using the Hospital Anxiety and Depression Scale (HADS).[26] The HADS is a 14-item scale divided into two 7-item subscales (anxiety and depression). Higher scores indicated higher symptom levels. The anxiety symptoms subscale Cronbach's αpatients = .83; αpartners = 84. The depressive symptoms subscale Cronbach αpatients = .84; αpartners = 82.

Data analyses

Preliminary analyses

Item-level analyses were conducted to select the best candidate items for inclusion in model testing. Statistical data, conceptual considerations (ie, content redundancy, clarity) and readability informed the exclusion of items. Intercorrelations among items were examined using a correlation matrix. Corrected item-total correlations and item distribution data (ie, measures of central tendency, SD) were computed and reviewed. Statistical reasons for item exclusion focused on items with poor inter-item correlations, low item-scale correlations, a noncentral mean, or poor variability. Based on DeVellis’[16] recommendations, item correlations were prioritized over item reduction based on means and variances. The statistical performance of item pairs with highly similar conceptual content was compared. Items with better performing distributions or when equivalent, items evaluated as clearer by consensus of the research team were retained. Items demanding higher reading comprehension (> grade 9) based on Lexile scores of readability were excluded. To ensure parallel patient/partner measures, item decisions were applied identically to each item set.

Factor analysis

Using the dyad as the unit of analysis, a confirmatory factor analysis (CFA) with a mirrored bifactor model structure among both patients and partners was conducted (Fig. 2).[27] The model was tested using 26 retained items from the initial pool that reflected the 3 components of the scale design (ie, illness intrusions, patient-affect, and partner-affect). In the bifactor structure, all 26 items were constrained to load on a general primary factor and also on one of the 3 correlated secondary factors, orthogonal to the primary factor. Items presented in the same questionnaire section loaded on the same secondary factor. This bifactor structure was mirrored between patients and partners. To account for the dyadic study design, correlations among residual terms of the patient and partner side pertaining to the same item were included and the common factors from both sides were allowed to intercorrelate. A variance decomposition based on the bifactor model was conducted to investigate the dimensionality and reliability of the scale. A CFA, rather than an exploratory factor analysis, was warranted due to the substantial conceptual research done to identify a theoretical scale structure.[17]

F2
FIGURE 2:
Standardized path diagram of the mirrored bifactor model for the Dyadic Efficacy Scale for Cancer (DESC). There are 26 items in total in the questionnaire for both patients (A) as well as partners (B): DE1 to DE26 (see bold block containers). FA1 is a primary factor on which all items load on that were responded to by the patient group (A); FA2.1 to FA2.3 are correlated secondary factors (see RAA and RBB) orthogonal to the primary factor, for which only items with the same stem / from the same questionnaire component load on (see sub-blocks inside the bold block). This bifactor structure is mirrored in similar fashion for the partner group (B). To account for the dyadic study design (ie, patient-partner couples), correlations (r ε(A)ε(B)) among residual terms pertaining to the same item (εDE(i)A & εDE(i)B) are allowed, as well as correlations among the common factors from both sides (RAB).

Full information maximum likelihood was used to handle missing data under the missing at random assumption and made use of all available information for each individual. Robust (Huber-White) standard errors for all estimated parameters and a scaled goodness-of-fit χ2 for statistical inference were applied. Model fit was evaluated based on commonly recommended goodness-of-fit indices,[28] including the χ2 test of exact model fit, the root mean square error of approximation (RMSEA: ≤.08 = acceptable, ≤.05 = good) to assess close fit, the comparative fit index (CFI: ≥.90 = acceptable, ≥.95 = good) contrasting to a null independence model, and the standardized root mean square residual (SRMR: ≤.08 = acceptable, ≤.05 = good). CFA was conducted using Mplus (Version 8).[29]

Reliability and validity

Given the strengths of model-based reliability indices over measures of internal consistency,[30] McDonald's omega (≥.70 = adequate) was used to assess the reliability of the unit-weighted total and subscale scores. Convergent validity of the DESC was evaluated by examining Pearson product moment correlations of the hypothesized relationships between dyadic efficacy and established measures of related constructs.[16] Reliability and validity analyses were conducted using IBM SPSS software (Version 24).

Results

Preliminary analyses

Data screening were conducted to identify entry errors and assess missing values and normality assumptions. Due to a small number of participants with dependent children living at home (n = 49 patients, n = 33 partners), 5 items specific to dyadic efficacy for communicating with and caring for children were excluded from this analysis. Descriptive statistics and item-level analyses were conducted on the remaining 45 equivalent patient and partner items. Age, sex, relationship length, relationship quality, and distress were compared between complete dyads and those represented by a patient or partner only. The only group difference found was for partner sex, χ2(1, N = 217) = 4.89, P = .03, ϕ = .15. The percentage of male partners was higher (58%) among complete dyads than among the sample of partners participating alone (35%).

Corrected item-total correlations and item means were of limited use for item reduction purposes due to highly similar results across items. Five items with limited response ranges were candidates for exclusion but were retained for the current analysis given this stage of scale evaluation and due to their conceptual relevance as previously evaluated by lay and expert reviewers.[17] Nineteen grouped or paired items with high inter-item correlations and conceptual similarity were reduced to 8 items (eg, “discuss our concerns with my doctor(s)" was retained over “talk openly with my medical team"). When possible, the item with greater variance was retained; when item variance was identical or negligibly different, the item that authors DB and AK judged to be more direct was retained. An additional 4 items that had content similarity with retained items but were more broadly worded were excluded by authors DB and AK. For example, an item quering patients’ and partners’ confidence in their conjoint ability to “make sense of their feelings about cancer" was more specifically addressed by items querying patients’ and partners’ confidence to manage fears, worries or low mood. The 4 items with the highest Flesch-Kincaid Grade Level[31] (FKGL) scores were also removed. The FKGL scores of these items ranged from 9.8 to 13.4, suggesting that a reading comprehension level typically achieved at the end of grade 9 or later was required for accurate understanding. The above-described item-level analyses led to the exclusion of 19 items, leaving 26 patient and 26 partner items to be carried forward for model testing (see Appendix A for DESC–Patient version, https://links.lww.com/OR9/A30). Descriptive statistics, including item-level missing values were reviewed before model testing (see supplement digital content SDC Table 1, https://links.lww.com/OR9/A29).

Factor analysis

The overall fit of the CFA reflected adequate to strong goodness-of-fit indices: χ2(1170) = 2090, P < .001; RMSEA = .05, P = .14, 90% CI .05 to .06]; SRMR = .05; CFI = .90. All 26 patient items loaded significantly on the patient primary factor (FA1) with factor loadings ranging from .52 to .86 (Table 2). Apart from 3 items (DE4B, DE16B, and DE19B), all partner items loaded significantly on the partner primary factor (FB1) with factor loadings ranging from .56 to .79. The primary factors explained 54% and 43% of the variance in patients’ and partners’ dyadic efficacy respectively (see SDC Table 2, https://links.lww.com/OR9/A29 for variance decomposition).[32] Item responses were more strongly correlated for patients than for partners across the full scale and within each secondary factor item set. The 3 secondary factors examined were supported in the patient but not in the partner model. For patients, item sets reflecting dyadic efficacy for managing illness intrusions (FA2.1), one's own affect (FA2.2) and a partner's affect (FA2.3) demonstrated residual common variance within each item set, above that which was explained by the general factor (known as local item dependencies). This multidimensionality accounted for an additional 18% of the variance in patients’ dyadic efficacy. In the partner sample, there was evidence for secondary factors with respect to partners’ dyadic efficacy for managing the patient's affect (FB2.2) and their own affect (FB2.3), but not for illness intrusions (FB2.1). Secondary factors accounted for 21% of the variance in partners’ dyadic efficacy. To sum, the model was able to account for a total of 72% and 64% of the variance in patients’ and partners’ cancer-related dyadic efficacy.

Table 2 - Mirrored bifactor model for the Dyadic Efficacy Scale for Cancer.
Patients Partners


Item FA1 FA2.1 FA2.2 FA2.3 r ε(A)ε(B) FB1 FB2.1 FB2.2 FB2.3
1 .70 .20 .03 .70 −.21
2 .74 .19 .14 .68 −.18
3 .74 .23∗∗∗ .31 .67 −.06
4 .52 .48∗∗ .32∗∗ .47 −.30
5 .58 .28∗∗ .71 .59 −.15
6 .66 .39 .33 .66 −.16
7 .81 .06 .38 .79 −.01
8 .74 .43 .16 .67 −.22
9 .79 .27∗∗ .13 .79 −.19
10 .60 .67 .35 .69 −.17
11 .72 .45∗∗ .34 .69 −.32
12 .75 .35∗∗ .18 .77 −.15
13 .78 .08 .43 .75 −.14
14 .73 .24∗∗∗ .09 .72 −.24
15 .80 .24 −.11 .56 −.51
16 .74 .45∗∗ −.11 .58 −.62
17 .86 .30∗∗∗ .03 .65∗∗∗ .61∗∗
18 .77 .47 .04 .59∗∗∗ .56
19 .76 .57 −.01 .54 .77∗∗
20 .75 .55 .15 .61∗∗∗ .68∗∗
21 .83 .35 .16 .63∗∗ .55∗∗∗
22 .79 .47 .01 .66 .62
23 .75 .55 .04 .63 .68
24 .70 .67 −.06 .64 .70
25 .69 .66 −.08 .59 .72
26 .67 .57 −.07 .61 .61
Model fit indices: χ2(1170) = 2090, P < .001; RMSEA = .05, P = .143, 90% CI .05–.06; SRMR = .05; CFI = .90. F1 is a primary factor, F2.1 to F2.3 are secondary factors, orthogonal to the primary factor, on which only items from the same questionnaire component load on. The bifactor structure is mirrored between patients and partners, with additional residual correlations (rε(A)ε(B)) among all item residuals and inter-factor correlations to account for the dyadic design (ie, patient-partner couples). Residual correlations where allowed for item pair 1 and 2 (rε(A)ε(A) = .47 (.08), P < .001; rε(A)ε(A) = .24 (.09), P = .008) and item pair 21 and 26 (rε(A)ε(A) = .33 (.08), P < .001; rε(A)ε(A) = .32 (.12), P = .006).
P < .001.
∗∗P < .01.
∗∗∗P < .05.

Within-dyad correlations

The primary factors were strongly correlated between patients and partners (r = .67, P < .001; see SDC Table 3, https://links.lww.com/OR9/A29). Within-dyad correlations among the secondary affect management factors were approximately ≥.3, with only the correlations between a patient's dyadic efficacy for managing partner affect (FA2.3) and the 2 affect management factors among partners (FB2.2 and FB2.3) significantly different from 0 (r = .39, P < .01; r = .37, P < .01). Significant local dyadic dependencies were demonstrated for 9 items reflecting dyadic efficacy for managing multiple illness intrusions (Table 2). Responses to these items were correlated more strongly within dyads than was anticipated based on inter-factor correlations. In other words, the residual variance associated with these 9 patient and partner items may be explained, at least in part, by a dyadic-level factor not accounted for by the model. Items related to affect management did not demonstrate such residual within-dyad dependencies.

Reliability and validity

The reliability of the unit-weighted total score was estimated to be ΩA = .72 and ΩB = .64 for patients and partners, respectively. Reliability for the unit-weighted subscale scores among patients was estimated to be ΩA2.1 = .63 for illness intrusions, ΩA2.2 = .84 for patient-affect and ΩA2.3 = .87 for partner-affect. Reliability among partners was estimated to be ΩA2.2 = .71 for patient-affect and ΩA2.3 = .84 for partner-affect. The reliability results presented here support multiple unit-weighted scoring options for the DESC.[33] Among patients, reliability indices support the use of a total DESC score and the computation of patient affect and partner affect subscale scores. Support for scoring the illness intrusions subscale among patients was near adequate. Among partners, support for a total DESC score was near adequate. Computations of subscale scores for patient and partner affect among partners were well-supported. Dyadic efficacy for managing illness intrusions was not an identified secondary factor among partners. Convergent validity of the DESC was examined by considering hypothesized relationships between cancer-related dyadic efficacy and psychological distress, self-efficacy, and relationship quality. The hypothesized relationships and the direction of association were confirmed for each of the correlations examined (Table 3).

Table 3 - Correlations between the DESC total, DESC subscales, and measures of psychological distress, relationship quality and self-efficacy by patient (N = 261) and partner (N = 217) role.
Measure 1 2 3 4 5 6 7 8 M Pt SD Pt
1. DESC total −.42 −.42 .47 .62 7.36 1.42
2. Illness intrusions DE .83 .77 −.38 −.41 .43 .59 7.49 1.49
3. Patient affect DE .82 −.43 −.39 .45 .60 7.24 1.72
4. Partner affect DE .82 −.41 −.35 .46 .57 7.07 1.70
5. Anxiety symptoms −.41 −.36 −.47 .59 -.26 -.53 6.25 3.86
6. Depressive symptom −.50 −.45 −.52 .59 -.22 -.54 4.63 3.72
7. Relationship quality .52 .50 .45 −.17∗∗ −.29 .35 50.83 9.96
8. Self-efficacy .77 .73 .67 −.40 −.50 .51 7.21 1.29
M Sp 7.46 7.21 6.98 7.17 4.52 51.35 7.26
SD Sp 1.62 1.47 1.71 3.74 3.28 8.72 1.07
DE = dyadic efficacy, Pt = patient, Sp = spouse/partner. Correlations for the patient sample are presented above the diagonal, and correlations for the partner sample are presented below the diagonal. Correlations for disease management in the partner sample were not calculated as there was no evidence to support this secondary factor. Correlations between DESC total scores and secondary DESC factors were not presented due to the overlap of items underlying their score.
P < .001.
∗∗P < .05.

Discussion

The DESC was developed using a rigorous multiphase design with careful attention to foundational construct and item development procedures. The bifactor model identified is consistent with conceptual work that highlighted both the stability of cancer-related dyadic efficacy as a function of couples’ preexisting relational functioning and an anticipated multidimensionality of efficacy based on the challenge being managed.[17] The strength of the general factor, particularly among patients, may suggest that patients’ perceptions of cancer-related dyadic efficacy were more strongly influenced by an overarching confidence in the dyad's capability than by differences in the specific type of challenge encountered. However, the additional variance captured by the secondary factors suggests that subtle differences in dyadic efficacy across domains were present. The suggestion that cancer-related dyadic efficacy reflects both a general dyadic efficacy and a domain-specific dyadic efficacy echoes an unresolved discussion in the literature surrounding the presence of a stable sense of self-efficacy in one's capability versus domain-specific appraisals.[7,34] Results based on the DESC indicate that a general perception of the dyad's efficacy may be a stronger underlying contributor to patients’ perceptions of confidence than to partners.

Dyadic efficacy for affect-management was well accounted for by the respective patient and partner factors. Items reflecting the management of illness intrusions resulted in higher item residuals and the presence of residual interdependence within dyads. Confidence to conjointly manage illness intrusions may be more strongly influenced by idiosyncratic differences specific to the functioning of a couple or by disease-related factors that were not accounted for by this model. For example, differences in the extent of change to a patient's physical appearance are strongly influenced by cancer type and treatment. Patients diagnosed with cancers of the head and neck and younger breast cancer patients commonly report higher levels of body image dissatisfaction.[35,36] These initial results suggest that the influence of shared demographic, relational or disease factors on couples’ dyadic efficacy may be more substantial for illness intrusions than for items reflecting efficacy for affect management.

Adequate to strong evidence was provided for the reliability and validity of computing a DESC total score and the 3 patient subscales and 2 partner subscales. Following this initial evaluation, well-supported DESC scoring options include: unidimensional total score for patients and affect management subscale scores for patients and partners. Additional scale development efforts should reevaluate and consider adjustments to improve the reliability of a DESC total score among partners and the illness intrusion items among both groups. Consistent with previous research on dyadic or collective efficacy, the DESC performed as expected in relation to related constructs.[4,14,37] Sterba et al[4] reported that dyadic efficacy for rheumatoid arthritis management was highly correlated with relationship quality, noting that further examination of the constructs’ distinctiveness should be reviewed. Greater distinctiveness was achieved in our sample with more moderate associations found between cancer-related dyadic efficacy and relationship quality. Dyadic efficacy and self-efficacy were highly correlated in our samples, particularly among partner participants. It will be essential to further examine this association to ensure adequate differentiation between a partner's dyadic efficacy and a partner's self-efficacy for caregiving has been achieved.

Clinical research and applications

The study of couple-focused interventions is young when compared to other areas of clinical focus in psychosocial oncology. The consideration of cancer-related dyadic efficacy adds an additional dimension to the commonly examined individual and social resources studied among couples coping with cancer (eg, self-efficacy, relationship functioning).[6,38] With theoretical and empirical support for the association of efficacy and psychological outcomes following illness,[4,39] the DESC provides a promising new systemic target for psychosocial researchers to pursue. The development of the DESC opens up possibilities for the design and evaluation of dyadic efficacy enhancing interventions in cancer care and enables the assessment of a novel outcome when evaluating existing couples’ interventions. Adherence to the notion of, “we think we can" may have dynamic influences on couples’ adjustment to cancer. According to Badr and Acitelli,[40] dyadic efficacy may also help to explain the influence of dyadic coping behaviors on individual and relational outcomes among couples coping with illness. Further experimental research would be useful here as cancer-related dyadic efficacy could be conceptualized as an appraisal within the developmental-contextual model of couples’ coping,[3] thereby influencing the couples’ enactment of individual or dyadic coping strategies.

The 26-item DESC can be completed by most respondents in 10 minutes. The DESC could be a valuable assessment tool for clinicians providing support for couples coping with cancer. Knowledge of patients’ and partners’ baseline appraisals of confidence to manage the effects of cancer and more specifically, their own and their partners affect, would be a valuable component of the early clinical picture of a couples’ functioning. The structure of the DESC also responds to the need to consider partners’ responses to a loved one's disease in their own right, rather than holding a more restricted view of partners as unidimensional care providers.[3] This was accomplished as equivalent patient/partner versions of the DESC were examined and perhaps more uniquely, through the assessment of the couples’ confidence to manage the partner's emotional response to cancer as well.

Study limitations

The model confirmed in this administration of the DESC should be further corroborated in a distinct sample. On average, respondents in the current sample had higher socioeconomic status and reported high levels of cancer-related dyadic efficacy and relationship quality. This may be due to a self-selection bias whereby individuals in higher functioning relationships are more likely to participate in a study examining couple-level outcomes. Efforts were made to limit this phenomenon by allowing patients and partners to participate alone but it is unknown if this strategy improved the sample heterogeneity. It may be advantageous to administer the DESC in a study examining either patients or partners, thereby minimizing the emphasis on couples when advertising and actively recruiting participants. The high levels of dyadic efficacy reported here may also reflect ceiling effects in the items themselves. The sample used for this study included 18% of patients and 13% of partners who were Indigenous, Black, persons of color. Furthermore, only 4% of complete dyads represented male/male or female/female partnerships. Additional inquiry is needed to consider whether the DESC is an accurate measure of cancer-related dyadic efficacy among more diverse patient and partner populations. Participation in the study was limited to those who could read and write in English. This eligibility criterion further influences the generalizability of the reported findings.

To ensure the DESC covered content domains most commonly experienced by individuals with cancer regardless of cancer diagnosis, necessary generality was used for items concerning physical symptoms and side-effects. The design and testing of supplementary subscales specific to a particular type of cancer would enable a more thorough assessment of dyadic efficacy for the management of physical symptoms. The use of subscales specific to cancer type has been used successfully for other scale sets (ie, Functional Assessment of Cancer Therapy scales).[41] An assessment of the scale's test–retest reliability and sensitivity to change would also need to be investigated before this tool could be used to evaluate interventions or be employed in clinical settings.

Conclusions

In his writings on self-efficacy, Bandura[8] asserted that enhancing efficacy not only reduces distress but also has the potential to increase one's capacity to persevere in the face of obstacles. In a setting where uncertainty ranks among the most common psychological stressors of cancer,[42] this assertion deserves further empirical inquiry. The DESC is a newly developed scale with initial evidence of its reliability and validity for assessing couples’ confidence to cope with the effects of a cancer diagnosis and its treatment. The DESC is thoroughly systemic in structure and design and will enable investigations of dyadic-efficacy interventions to better meet the interpersonal needs of cancer patients and their family members.

Conflicts of interest

The authors report no conflicts of interest.

References

[1]. Jacobs JM, Shaffer KM, Nipp RD, Fishbein JN, MacDonald J, El-Jawahri A, et al. Distress is interdependent in patients and caregivers with newly diagnosed incurable cancers. Ann Behav Med 2017;51:1–13.
[2]. Revenson TA. Suls J, Wallston KA. Scenes from a marriage: examining support, coping, and gender within the context of chronic illness. Social Psychological Foundations of Health and Illness Malden, MA: Blackwell; 2003;530–559.
[3]. Berg CA, Upchurch R. A developmental-contextual model of couples coping with chronic illness across the adult life span. Psychol Bull 2007;133:920–954.
[4]. Sterba KR, DeVellis RF, Lewis MA, Baucom DH, Jordan JM, DeVellis B. Developing and testing a measure of dyadic efficacy for married women with rheumatoid arthritis and their spouses. Arthritis Rheum (Arthrit Care Res) 2007;57:294–302.
[5]. Weber BA, Roberts BL, Resnick M, Deimling G, Jaclene A, Zauszniewski CM, et al. The effect of dyadic intervention on self-efficacy, social support, and depression for men with prostate cancer. Psychooncology 2004;13:47–60.
[6]. Merluzzi TV, Philip EJ, Heitzmann Ruhf CA, Liu H, Yang M, Conley CC. Self-efficacy for coping with cancer: revision of the Cancer Behavior Inventory (Version 3.0). Psychol Assess 2018;30:486–499.
[7]. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev 1977;84:191–215.
[8]. Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs (NJ): Prentice-Hall; 1986.
[9]. Kjærside Nielsen B, Mehlsen M, Bonde Jensen A, Zachariae R. Cancer-related self-efficacy following a consultation with an oncologist. Psychooncology 2013;22:2095–2101.
[10]. Manne SL, Ostroff JS, Norton TR, Fox K, Grana G, Goldstein L. Cancer-specific self-efficacy and psychosocial and functional adaptation to early stage breast cancer. Ann Beh Med 2006;31:145–154.
[11]. Lewis FM, Griffith KA, Alzawad Z, Dawson PL, Zahlis EH, Shands ME. Helping her heal: randomized clinical trial to enhance dyadic outcomes in couples. Psychooncology 2019;28:430–438.
[12]. Northouse LL, Mood DW, Schafenacker A, Montie JE, Sandler HM, Forman JD, et al. Randomized clinical trial of a family intervention for prostate cancer patients and their spouses. Cancer 2007;110:2809–2818.
[13]. Northouse LL, Mood DW, Schafenacker A, Kalemberian G, Zalupski M, LoRusso P, et al. Randomized clinical trial of a brief and extensive dyadic intervention for advanced cancer patients and their family caregivers. Psychooncology 2013;22:555–563.
[14]. Sterba KR, Rabius V, Carpenter MJ, Villars P, Wiatrek D, McAlister A. Dyadic efficacy for smoking cessation: preliminary assessment of a new instrument. Nicotine Tob Res 2011;13:194–201.
[15]. Gehlbach H, Brinkworth ME. Measure twice, cut down error: a process for enhancing the validity of survey scales. Rev Gen Psychol 2011;15:380–387.
[16]. DeVellis RF. Scale Development. 3rd ed. Thousand Oaks (CA): Sage; 2012.
[17]. Brosseau DC, Palaez S, Annang B, Körner A. Contextualizing measurement: Establishing a construct and content foundation for the assessment of cancer-related dyadic effficacy (manuscript under review). 2021.
[18]. Braun V, Clarke V. Reflecting on reflexive thematic analysis. Qual Res Sport Exerc Health 2019;11:589–597.
[19]. Dagan M, Hagedoorn M. Response rates in studies of couples coping with cancer: a systematic review. Health Psychol 2014;33:845–852.
[20]. Qualtrics [computer program]. 2018.
[21]. Heitzmann CA, Merluzzi TV, Jean-Pierre P, Roscoe JA, Kirsh KL, Passik SD. Assessing self-efficacy for coping with cancer: development and psychometric analysis of the brief version of the Cancer Behavior Inventory (CBI-B). Psychooncology 2011;20:302–312.
[22]. Merluzzi TV, Sanchez MAM. Assessment of self-efficacy and coping with cancer: development and validation of the Cancer Behavior Inventory. Health Psychol 1997;16:163–170.
[23]. Merluzzi TV, Philip EJ, Vachon DO, Heitzmann CA. Assessment of self-efficacy for caregiving: the critical role of self-care in caregiver stress and burden. Palliat Support Care 2011;9:15–24.
[24]. Busby DM, Christensen C, Crane RD, Larson JH. A revision of the dyadic adjustment scale for use with distress and nondistressed couples: Construct hierarchy and multidimensional scales. J Marital Fam Ther 1995;21:289–308.
[25]. Spanier GB. Measuring dyadic adjustment: new scales for assessing the quality of marriage and similar dyads. J Marriage Fam 1976;38 (1):15–28.
[26]. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–370.
[27]. Reise SP. The rediscovery of bifactor measurement models. Multivar Behav Res 2012;47:667–696.
[28]. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional crtieria versus new alternatives. Struct Equ Modeling 1999;6:1–55.
[29]. Muthén LK, Muthén BO. Mplus Statistical Analysis With Latent Variables User's Guide. 8th ed Los Angeles, CA: 1998–2012.
[30]. Peters GY. The alpha and the omega of scale reliability and validity: Why and how to abandon Cronbach's alpha and the route towards more comprehensive assessment of scale quality. Eur Health Psychol 2014;16:56–69.
[31]. Kincaid JP, Fishburne RPJ, Rogers RL, Chissom BS. Derivation of new readability formulas (Automated Readability Index, Fog Count and Flesch Reading Ease Formula) for navy enlisted personnel. https://apps.dtic.mil/sti/citations/ADA006655. 1975. Accessed October 18, 2021.
[32]. Rodriguez A, Reise SP, Haviland MG. Evaluating bifactor models: calculating and interpreting statistical indices. Psychol Methods 2016;21:137–150.
[33]. Reise SP, Bonifay WE, Haviland MG. Scoring and modeling psychological measures in the presence of multidimensionality. J Pers Assess 2013;95:129–140.
[34]. Schwarzer R, Born A. Optimistic self-beliefs: assessment of general perceived self-efficacy in thirteen cultures. World Psychol 1997;3:177–190.
[35]. Fingeret MC, Teo I, Goettsch K. Body image: a critical psychosocial issue for patients with head and neck cancer. Curr Oncol Rep 2015;17:422–427.
[36]. Paterson CL, Lengacher CA, Donovan KA, Kip KE, Tofhiagen CS. Body image in younger breast cancer survivors: a systematic review. Cancer Nurs 2016;39:E39–E58.
[37]. Hou WK. Intrapersonal and interpersonal dimensions of cancer perception: a confirmatory factor analysis of the cancer experience and efficacy scale. Support Care Cancer 2010;18:561–571.
[38]. Traa MJ, De Vries J, Bodenmann G, Den Oudsten BL. Dyadic coping and relationship functioning in couples coping with cancer: a systematic review. Br J Health Psychol 2014;20:85–114.
[39]. Bandura A. Self-efficacy: The exercise of control. New York (NY): WH Freeman and Company; 1997.
[40]. Badr HJ, Acitelli LK. Re-thinking dyadic coping in the context of chronic illness. Curr Opin Psychol 2017;13:44–48.
[41]. Cella D, Tulsky DS, Gray G, Sarafian B, Linn E, Bonomi A, et al. The Functional Assessment of Cancer Therapy scale: Development and validation of the general measure. J Clin Oncol 1993;11:570–579.
[42]. Lebel S, Rosberger Z, Edgar L, Devins GM. Comparison of four common stressors across the breast cancer trajectory. J Psychosom Res 2007;63:225–232.
Keywords:

couples; dyadic efficacy; measurement; neoplasm; self-efficacy

Supplemental Digital Content

Copyright © 2021 The Authors. Published by Wolters Kluwer Health Inc., on behalf of the International Psycho-Oncology Society.