Developing and testing a measure of dyadic efficacy for married women with rheumatoid arthritis and their spouses

Authors


Abstract

Objective

Interpersonal relationships and self-efficacy have each been independently studied in the context of coping with chronic illness. To examine a new type of interpersonal efficacy in couples coping with rheumatoid arthritis (RA), we developed and tested new measures of perceptions of arthritis dyadic efficacy. We assessed both partners' perceptions of confidence about working together as a team to manage women's illness-related challenges.

Methods

First, a 3-phase pilot study was conducted with interviews, expert review, and pretesting to develop items. Next, the psychometric properties of new measures were tested in 190 women with RA and their husbands. Exploratory analyses were conducted, Cronbach's alphas were calculated for each factor, and construct validity was examined with Pearson's correlations at baseline, 4-month, and 8-month followup.

Results

Factor analyses yielded 3 factors in the wife and husband versions, assessing dyadic behaviors concerning arthritis problem solving and emotions, arthritis symptom management, and arthritis-related couple outcomes. All items loaded >0.60, Cronbach's alphas for all subscales were >0.88, and initial evidence of construct validity was demonstrated. Finally, the initial factor structure was replicated with additional exploratory factor analyses in the same sample at 4-month and 8-month followup.

Conclusion

Three short, reliable subscales resulted for couples coping with RA for use in interpersonal health research. These instruments facilitate viewing illness adaptation processes in a dyadic manner.

INTRODUCTION

Teamwork within close relationships may be especially important in the context of a progressive, unpredictable illness such as rheumatoid arthritis (RA). RA is characterized by progressive joint destruction and disability (1). Its physical and psychological consequences affect, and are affected by, interpersonal relationships (2). Conceptualizing and operationalizing how close relationships enhance adjustment to RA may reveal focus points for interventions to enhance quality of life. Although social support is perhaps the most extensively studied interpersonal variable, its mechanisms are not fully understood (3). Therefore, alternative conceptual approaches to interpersonal processes may be beneficial. Self-efficacy is associated with positive change in health status in arthritis management interventions (4), and in the present study, we examined efficacy as an interpersonal process.

Self-efficacy is part of social cognitive theory (5), and collective efficacy, an extension of this concept (6), has been studied in communities but not in smaller interpersonal groupings such as dyads. Forming beliefs about RA management capabilities occurs in an interpersonal environment. In this study, we focused on marriage, one of the closest personal relationships (3). According to Bandura (6), “if people are to pool their resources and work together successfully, the members have to perform roles with a high sense of efficacy.”

Although we are not aware of any prior measures of efficacy for a dyad as an integral unit, individual self-efficacy has been assessed simultaneously in male myocardial infarction patients and their wives (7–9). Husband self-efficacy was positively associated with dependence on wives and marital quality, and was inversely associated with protective buffering, or hiding concerns from wives (7). In addition, wife self-efficacy (or confidence about assisting her husband's recovery) was positively associated with patient self-efficacy. These findings are consistent with other research showing that both partners' ratings of patient self-efficacy predicted men's future physical performance (8). Rohrbaugh and colleagues (9) further found that both patient and partner ratings of patient self-efficacy independently predicted survival in individuals with heart failure. These studies suggest that examining self-efficacy in both couple members is important in recovery from myocardial infarction.

To extend this research, we looked beyond self-efficacy in each partner, and instead examined each spouse's sense of the dyad's efficacy as a unit in the context of a progressive illness. We defined this new concept, arthritis dyadic efficacy, as an individual's perceptions of confidence about his or her shared ability with a partner to manage arthritis-related problems as a couple. Thus, dyadic efficacy is concerned with partner beliefs about the dyad's efficacy rather than an individual's contribution to the couple's efficacy.

Our pilot interviews revealed that couple members often reported trying to work together to manage arthritis and that working together was a satisfying experience. We therefore expected that dyadic efficacy would be positively associated with marital satisfaction, marriage quality, and perceptions of support. Furthermore, because self-efficacy influences affective processes, such as the emotions individuals associate with their actions (10, 11), and because self-efficacy is related to facing challenges with less distress (11), we expected that dyadic efficacy would be positively associated with psychological adjustment. Also, because having a sense of mastery for managing arthritis as a couple likely depends upon the severity of the illness challenges and the length of time one has had arthritis and been married, we expected that dyadic efficacy would vary by illness and marriage variables.

This report describes the development and testing of new measures of dyadic efficacy for couples coping with RA. Items were developed in a pilot study and then tested in a larger survey study.

MATERIALS AND METHODS

Item development study.

Item generation for the dyadic efficacy instruments was guided by the arthritis self-efficacy measure developed by Lorig et al (12), which assesses individuals' confidence in their ability to manage pain, function physically, and control other arthritis symptoms. Several items from the existing instrument by Lorig et al were rewritten in a dyadic manner for the new dyadic efficacy measures. In addition, to further inform item development, in-person interviews were conducted with a convenience sample of 12 couples in which one partner had osteoarthritis in order to define ways in which couples work together to manage arthritis-related challenges. As shown in Table 1, recurring interview themes revealed important teamwork behaviors and challenges to teamwork.

Table 1. Teamwork behaviors and challenges in couples coping with arthritis: in-person interview findings
ThemeRepresentative quote
Concept I: teamwork behaviors 
 Provide understanding to one another“Feeling that he understands my pain is most important.”
 Communication“Little things help like when he calls to check on how I'm feeling.”
 Hands-on help“He rubs my back even when he's tired.”
 Physical assistance“He does all the heavy lifting.”
 Taking action together“We eat a healthy diet together and try to walk together.”
 Providing distraction“She tries to distract me from the pain with a focus on the good things in our life like our grandkids.”
 Providing motivation“He is a great supporter and encourages me to do things on my own.”
Concept II: challenges to teamwork 
 Severe pain“It's hard to concentrate on anything but not feeling well.”
 When both partners are having a bad day“That understanding is extra hard when you've both had a bad day.”
 When you feel helpless“It's so frustrating when you feel like there's nothing you can do.”
 The cyclic nature of arthritis“It's hard for me to understand the arthritis ups and downs so how can I expect him to understand?”

Based on comments from interviews, additional dyadic efficacy items were drafted and instructions for the new measures were written to emphasize that individuals should report their perceptions of the team's efficacy rather than perceptions of their own individual abilities. Furthermore, items were sorted into 2 sets, one covering patient outcomes and the other covering couple outcomes, because couples reported working together as a team on both areas. Dyadic efficacy patient outcomes items included themes such as pain, fatigue, treatment, emotions, and daily activities. Dyadic efficacy couple outcomes items included themes such as emotions, shared activities, and responsibilities in both partners, as well as depending on each other and taking action together.

Next, 7 researchers (independent of study investigators) with collective expertise in health behavior, couples research, rheumatology, pain management, interpersonal psychology, and self-efficacy provided written feedback on measure content, item clarity, and wording. Their recommendations led to a revision of the instructions and the revision of several items for better conceptual fit and clarity.

Finally, dyadic efficacy instruments were pretested among a convenience sample of married women with RA or osteoarthritis and their husbands. Of the 13 couples contacted, both partners in 9 couples (69%) completed questionnaires and provided written feedback. Sixteen individuals in these participating couples (89%) also provided followup information by telephone. Respondents had varied opinions about how important teamwork was with respect to different areas of arthritis management (e.g., “We don't work well as a team to manage my treatment, but I don't think he should be involved in that”). Therefore, a set of questions was designed to capture teamwork standards, or opinions about how involved partners should be on different aspects of arthritis management. Finally, vague items were deleted and the protocol was revised to promote confidential survey completion among partners. When item development was finalized, the psychometric properties of dyadic efficacy instruments were examined.

Sample and protocol.

The dyadic efficacy instrument was administered to 190 married women with RA and their husbands. Recruitment efforts included distributing project information through local Arthritis Foundation chapter publications, arthritis Web sites, and rheumatology offices. Volunteer married women with a confirmed diagnosis of RA for at least 1 year, who did not also have fibromyalgia or systemic lupus erythematosus, and their husbands were enrolled. Couple members completed separate mailed surveys at baseline, 4-month, and 8-month followup.

Measures.

At baseline, sociodemographic variables assessed in both partners' surveys included date of birth, race, years of education, and years married. Physician ratings included a confirmation of women's RA diagnoses and functional status ratings (13) obtained by fax from treating rheumatologists at baseline. Functional status ranged from class I (completely able to perform usual activities of daily living) to class IV (limited ability to perform usual self-care, vocational, and avocational activities). Additional measures described below were included in baseline and followup surveys; Cronbach's alphas for the current sample are reported.

Dyadic efficacy was examined with 25 items assessing teamwork affecting patient outcomes and 27 items assessing teamwork affecting couple outcomes in both partners' surveys. Each item contained the following introduction: “How confident are you that you and your spouse can work together as a team to …” followed by a specific behavior. Respondents circled one number on a scale from 0 (not at all confident) to 10 (extremely confident) that best represented their confidence in current abilities to perform each task with their spouse. Higher scores reflected higher dyadic efficacy.

Four items developed in a previous arthritis couples study assessed perceptions of support over the past month. This unidimensional scale operationalized support beliefs as the extent to which one's spouse was sensitive to one's needs and interested in helping solve problems, along with satisfaction concerning a spouse's sensitivity to needs and interest in helping solve problems (α = 0.93 for wives and α = 0.92 for husbands). Marital satisfaction was measured using the Kansas Marital Scale (14) (α = 0.96 for wives and α = 0.97 for husbands) to assess respondents' satisfaction with marriage overall, with their spouse as a partner, and with their relationship with spouse. Finally, marriage quality was examined with the Quality Marriage Index (15) (α = 0.94 for wives and α = 0.94 for husbands) to assess respondents' ratings of their marriage (e.g., whether the relationship was stable, strong, made one feel like a part of a team).

Teamwork standards assessed the extent to which wives and husbands believed arthritis should be handled as a team (versus individually). Four questions examined standards about day-to-day management, long-term planning, arthritis treatment, and emotions related to arthritis. Respondents circled one number on a scale from 1 (should be handled by me/her as an individual) to 10 (should be handled by us together as a team) for each question. The 4 items all loaded onto 1 factor in factor analysis; therefore a teamwork standards average score was calculated (α = 0.80 for wives and α = 0.86 for husbands).

Arthritis understanding was examined by a wife's ratings of the extent to which she believed her husband understood 10 different aspects of her arthritis (e.g., consequences, treatment, unpredictability, emotions; α = 0.95). Each husband rated the extent to which he believed he understood the same aspects of his wife's arthritis (α = 0.87).

Measures of positive and negative affect (16) (α = 0.90 for wife positive affect, α = 0.89 for wife negative affect, α = 0.91 for husband positive affect, and α = 0.86 for husband negative affect), depression (17) (α = 0.91 for wives and α = 0.86 for husbands), satisfaction with life (18) (α = 0.89 for wives and α = 0.88 for husbands), and optimism (19) (α = 0.82 for wife optimism, α = 0.87 for wife pessimism, α = 0.87 for husband optimism, and α = 0.88 for husband pessimism) were included. In a higher order factor analysis, all of the standardized average scores for these measures loaded highly onto 1 factor. A psychological adjustment score was calculated by summing each of the standardized average scores (α = 0.80 for wives and α = 0.82 for husbands); higher scores reflected better adjustment.

General health was assessed for both partners using 3 questions adapted from the Health Survey Short Form (20). These questions examined each partner's current health on a 5-point scale ranging from poor to excellent along with health limitations and worry about health over the past month on 6-point scales ranging from none of the time to all of the time.

Arthritis functioning was assessed in wives using 7 subscales (mobility, hand and finger function, walking and bending, arm function, self-care, household tasks, and pain) of the Arthritis Impact Measurement Scale (21). A factor analysis showed that all 7 standardized average scores loaded highly onto 1 factor; therefore an arthritis disability score was calculated by summing standardized average scores for each of the subscales (α = 0.79). Higher scores reflected higher disability.

Statistical analysis.

Exploratory principal axis factor analyses were conducted on the items intended to assess dyadic efficacy using promax rotation in SAS, version 9.1 (SAS Institute, Cary, NC). Factors were retained when scree plots showed a substantial drop in the amount of information provided when an additional factor was included and when eigenvalues were ∼1.0 (22). Items with standardized regression coefficients (referred to as factor loadings hereafter) >0.5 were retained when they also did not cross-load (or load on any other factor >0.3) (23). Identified item groupings were then examined for internal consistency by computing Cronbach's alphas (24). Items with the lowest item-to-total correlations in both wife and husband versions were dropped in sequence to optimize brevity and reliability. Additional exploratory factor analyses were conducted on the same full set of items at 4-month and 8-month followup to assess the stability of the factor structure. Test–retest reliability was assessed with Pearson's correlations of dyadic efficacy item sets at baseline and 4-month followup. Finally, correlation matrices were examined to explore convergent validity by examining relationships between each of the wife and husband versions of the dyadic efficacy subscales and wife and husband sociodemographic, marital, psychological, and disability variables. Due to the number of Pearson's correlation tests, the P value was set at 0.003 (25) (0.05 divided by 15 tests per comparison).

RESULTS

Participant characteristics.

Of 258 eligible couples, 7 couples (3%) declined initially and 61 couples (24%) declined later in the study. Reasons for declining included health problems, being too busy, or one partner not wanting to participate. Baseline surveys were completed by 190 couples; of these, 173 (91%) and 165 (87%) couples completed 4-month and 8-month surveys, respectively. As shown in Table 2, participants varied widely in sociodemographic characteristics and functional status.

Table 2. Participant characteristics*
VariableWivesHusbands
  • *

    Values are the percentage unless otherwise indicated. RA = rheumatoid arthritis; NA = not applicable.

Age, mean ± SD years49 ± 12.951 ± 13.6
Race  
 White9393
 African American44
 American Indian11.5
 Hispanic1.51
 Other0.50.5
Education  
 7–12 years (high school or below)1926
 13–16 years (some college)5546
 Graduate or professional school2628
Years married, mean ± SD22 ± 14.922 ± 14.9
Years diagnosed with RA, mean ± SD14 ± 10.9NA
Physician-rated functional status NA
 Class I25 
 Class II37 
 Class III27 
 Class IV5 
Usual pain over last month NA
 Mean ± SD3.0 ± 1.0 
 None9.5 
 Very mild45.7 
 Mild29.5 
 Moderate13.7 
 Severe1.6 

Factor analyses.

To confirm the separation of dyadic efficacy items into 2 categories (patient and couple outcomes), an initial factor analysis including all 52 items was conducted. Results confirmed the separation of items into 2 factors as interview respondents had asserted.

The next set of analyses examined patient outcomes and couple outcomes item sets separately to define the factor structure further. For the 25 patient outcomes dyadic efficacy items for wives, the factor analysis revealed 2 factors (second eigenvalue = 1.85) explaining 67.1% of the total variance (communalities were fixed at 1.0 for all analyses) with 3 cross-loaded items. Those 3 items were dropped, resulting in a 2-factor solution explaining 67.9% of the total variance (α = 0.96 for factor 1 and α = 0.92 for factor 2). Factor loadings ranged from 0.54 to 0.92 for items comprising factor 1 and from 0.54 to 0.85 for items comprising factor 2.

For the 25 patient outcomes dyadic efficacy items for husbands, factor analyses revealed 2 factors (second eigenvalue = 1.30) explaining 69.1% of the total variance with 4 cross-loaded items. Those 4 items were dropped, resulting in a 2-factor solution explaining 69.3% of the total variance (α = 0.96 for factor 1 and α = 0.93 for factor 2). Factor loadings ranged from 0.55 to 0.94 for items comprising factor 1 and from 0.51 to 0.87 for items comprising factor 2.

Next, in seeking parallel versions for the 2 patient outcomes dyadic efficacy subscales for wives and husbands, items in both versions with the lowest item-to-total correlations were dropped in sequence to optimize scale brevity and reliability. Items loading strongly on the same factor for both wives and husbands were preferentially retained. Two 5-item subscales for wives (interfactor correlation = 0.61) and 2 identical 5-item subscales for husbands (interfactor correlation = 0.64) resulted. The first subscale addressed dyadic efficacy about arthritis problem solving and emotions (mean 6.8, range 0.4–10, α = 0.93 for wives; mean 7.3, range 0–10, α = 0.92 for husbands). The second subscale addressed dyadic efficacy about arthritis symptom management (mean 6.2, range 0–10, α = 0.89 for wives; mean 6.3, range 0–10, α = 0.89 for husbands). Factor loadings and items for factors 1 and 2 are displayed in Table 3.

Table 3. Items and factor loadings for dyadic efficacy concerning patient outcomes items at baseline (T1), 4-month followup (T2), and 8-month followup (T3)*
 WifeHusband
T1T2T3T1T2T3
  • *

    Values are the factor loading unless otherwise indicated. Each item begins with the statement “How confident are you that you and your spouse can work together as a team to… .”

Factor 1: dyadic efficacy concerning problem solving and emotions (Cronbach's alpha)0.930.940.930.920.930.91
 Item      
  Deal with your/her arthritis frustrations?0.900.920.790.780.680.96
  Help you/her keep a positive attitude about living with arthritis?0.910.910.910.840.570.79
  Rearrange your/her activities when you are having a bad day?0.910.860.900.850.970.72
  Decide how best to tackle your/her course of treatment?0.710.690.850.800.640.82
  Keep up your/her activity level?0.780.600.790.920.840.84
Factor 2: dyadic efficacy concerning arthritis symptom management (Cronbach's alpha)0.890.930.900.890.920.89
 Item      
  Help you/her avoid pain?0.880.880.880.620.770.75
  Keep your/her arthritis from interfering with your/her sleep?0.790.710.790.960.610.63
  Decrease your/her pain?0.840.860.850.610.800.58
  Manage your/her arthritis symptoms?0.740.710.800.570.540.79
  Control your/her pain with methods other than taking medication?0.700.830.730.710.990.96

Using identical methods, the factor analysis of the 27-item couple outcomes dyadic efficacy item set revealed a 1-factor solution for both the wife version (eigenvalue = 18.7) and the husband version (eigenvalue = 20.4). A 1-factor solution for the wife version resulted in high loadings for all 27 items (from 0.65 to 0.90), explaining 69.2% of the total variance (α = 0.98). A 1-factor solution for the husband version also resulted in high loadings for all 27 items (from 0.73 to 0.93), explaining 75.7% of the total variance (α = 0.99). Dropping items with the lowest item-to-total correlations in both the wife and husband versions yielded a 6-item scale for wives and husbands assessing dyadic efficacy about arthritis-related couple outcomes (mean 7.8, range 0.3–10, α = 0.96 for wives; mean 8.1, range 0–10, α = 0.97 for husbands). Factor loadings and items for factor 3 are displayed in Table 4.

Table 4. Items and factor loadings for dyadic efficacy concerning couple outcomes items at baseline (T1), 4-month followup (T2), and 8-month followup (T3)*
 WifeHusband
T1T2T3T1T2T3
  • *

    Values are the factor loading unless otherwise indicated. Each item begins with the statement “How confident are you that you and your spouse can work together as a team to… .”

Dyadic efficacy concerning arthritis-related couple outcomes (Cronbach's alpha)0.960.970.970.970.950.97
 Item      
  Maintain positive attitudes?0.920.940.940.930.920.93
  Encourage each other?0.930.930.940.930.900.94
  Deal together with the unpredictable nature of arthritis?0.900.920.920.920.870.93
  Work around the difficulties of arthritis?0.890.920.940.920.880.90
  Keep each other's spirits high?0.910.950.940.940.920.94
  Focus together on the good things in your life?0.910.950.940.930.920.93

Identical exploratory factor analysis procedures assessing the stability of the factor structure of the 52 dyadic efficacy items at 4-month followup (n = 173 couples) and 8-month followup (n = 165 couples) replicated the original factor structure. Factor loadings and Cronbach's alphas at followup are shown in Tables 3 and 4. Also, 4-month test–retest correlations ranged from 0.62 to 0.69 for wives and from 0.68 to 0.73 for husbands (all P < 0.0001).

Construct validity.

Selected wife and husband Pearson's correlations are shown in Tables 5 and 6, respectively. The 3 wife dyadic efficacy subscales were highly positively correlated with one another (r = 0.54–0.77, P < 0.0001) and with each of the 3 husband dyadic efficacy subscales (r = 0.32–0.55, P < 0.0001).

Table 5. Wife dyadic efficacy Pearson's correlations with wife variables
Wife variablesWife dyadic efficacy subscales
Arthritis problem solving and emotionsArthritis symptom managementArthritis-related couple outcomes
  • *

    P < 0.0001.

  • P < 0.003.

Marital satisfaction0.55*0.39*0.67*
Perceptions of support0.64*0.49*0.73*
Marriage quality0.50*0.33*0.66*
Beliefs about how well husband understands arthritis0.47*0.37*0.51*
Psychological adjustment0.41*0.37*0.47*
Teamwork standards0.48*0.37*0.37*
Depression−0.32*−0.27−0.34*
Health worry−0.20−0.16−0.17
Age0.080.220.03
Race−0.05−0.09−0.07
Years married0.070.190.05
Years since diagnosis0.060.210.03
Pain−0.14−0.16−0.04
Self-reported illness disability−0.10−0.06−0.02
Physician-reported functional status−0.07−0.016−0.07
Table 6. Husband dyadic efficacy Pearson's correlations with husband variables and wife illness variables
Husband variablesHusband dyadic efficacy subscales
Arthritis problem solving and emotionsArthritis symptom managementArthritis-related couple outcomes
  • *

    P < 0.0001.

  • P < 0.003.

Marital satisfaction0.54*0.44*0.75*
Perceptions of support0.50*0.38*0.67*
Marriage quality0.49*0.37*0.70*
Beliefs about how well he understands wife's arthritis0.260.240.30*
Psychological adjustment0.40*0.30*0.53*
Teamwork standards0.53*0.45*0.48*
Depression−0.32*−0.20−0.45*
Wife health worry−0.14−0.15−0.10
Age−0.02−0.010.01
Race−0.06−0.16−0.06
Years married−0.06−0.08−0.03
Years since diagnosis−0.06−0.05−0.05
Wife pain−0.14−0.10−0.08
Wife self-reported illness disability−0.07−0.050.002
Wife physician-reported functional status−0.12−0.05−0.12

All 3 wife dyadic efficacy subscales were significantly positively associated with wife marital satisfaction, perceptions of support, marriage quality, beliefs about husband arthritis understanding, psychological adjustment, and teamwork standards. Also, all wife dyadic efficacy subscales were significantly inversely associated with wife depression. Wife dyadic efficacy concerning arthritis symptom management was also significantly positively associated with age and number of years living with RA, and was marginally associated with marriage length (P = 0.009). Neither wife's self-reported disability nor physician's report of patient's current functional status was related to dyadic efficacy.

All 3 husband dyadic efficacy subscales were significantly positively associated with husband marital satisfaction, perceptions of support, marriage quality, beliefs about his own arthritis understanding, psychological adjustment, and teamwork standards. Also, all 3 husband dyadic efficacy subscales were significantly inversely associated with husband depression. Husband dyadic efficacy was not associated with sociodemographic or wife illness variables.

Further analyses examined the associations of wife dyadic efficacy with husband variables and husband dyadic efficacy with wife variables. A similar pattern of findings resulted; for example, wife dyadic efficacy subscales were significantly positively associated with her own and her husband's marital satisfaction, perceptions of support, marriage quality, and psychological adjustment. The relationships between wife dyadic efficacy and wife variables were higher in magnitude than those between wife dyadic efficacy and husband variables.

DISCUSSION

We developed short, reliable, and easy to administer instruments assessing perceptions of arthritis dyadic efficacy. The instruments include items assessing dyadic behaviors about arthritis problem solving and emotions, arthritis symptom management, and arthritis-related couple outcomes in parallel versions for partners in couples coping with RA. High internal consistency was demonstrated with items of the 3 subscales each loading highly onto only 1 factor, being highly intercorrelated, and explaining a high proportion of total variance in the item set. Furthermore, the factor structure was the same at 3 time points and in parallel separate versions for wives and husbands. Finally, initial evidence of construct validity was demonstrated.

The content validity of the dyadic efficacy instrument comes from its development methods. First, a valid measure of arthritis self-efficacy (12) was modeled to cover areas of importance to individuals with arthritis, namely, controlling pain and disability. Second, interviews with couples coping with arthritis guided item generation so that items reflected behaviors and challenges couples faced. Third, experts reviewed items, providing unique perspectives based on research and practice experience. Finally, the instruments were formally pretested to examine usability.

As expected, individuals who believed that they worked better as a team with their partner to manage arthritis-related problems also reported better marriages, believed their spouse understood their arthritis more, had better psychological adjustment, and believed arthritis should be handled as a team. We speculated that the new scales might be sensitive to illness and marriage variables, but found no relationships between dyadic efficacy and illness disability, pain, or marriage length. However, we did find that older women who had RA longer also had higher dyadic efficacy concerning arthritis symptom management; this subscale was also marginally associated with length of marriage. Therefore, the concept that this subscale assessed potentially required the most effort, or years of experience with the illness to master, and future studies should further examine its relationship to marriage length. The other 2 dyadic efficacy subscales appeared to be less skills-based and possibly more a function of the quality of the marriage. Husband dyadic efficacy was not affected by age or illness duration and differences between husband and wife versions require more study.

The current study did not assess each partner's self-efficacy. Bandura (26) acknowledged that collective efficacy “is not simply the sum of the efficacy beliefs of individual [group] members” but “an emergent group-level property.” It is important as a next step to demonstrate that dyadic efficacy is a distinct concept, and is more than a simple combination of 2 partners' self-efficacy perceptions. The relationship between self-efficacy and dyadic efficacy is currently being examined in another study (27).

Future studies should also examine dyadic efficacy in more diverse samples; we included volunteers who were mainly white and had high levels of marital and illness functioning. In addition, all women in this study had RA. Although we suspect that instruments will have similar properties when the male is the patient, this hypothesis requires empirical validation.

Based on our interviews with couples and research showing beneficial effects of social support, we assumed that working together as a team in couples was beneficial for arthritis management. That is, social support has been associated with improved psychological adjustment (28) and less pain (29) in arthritis. This literature supports the idea that working together as a team is beneficial to both parties in a couple. However, a body of literature also shows that if support promotes dependence, coping efforts may be compromised. Because support can involve both beneficial and harmful consequences (30), it is important to examine how dyadic efficacy relates to support and coping behaviors. In addition, due to the strong positive associations of dyadic efficacy with marital variables, future studies should examine the discriminant and predictive validities of these new measures and whether dyadic efficacy is a component of support (e.g., does dyadic efficacy predict arthritis-specific support or adaptive coping behaviors above and beyond other measures of marital functioning?).

Furthermore, future studies should examine the potentially unique contributions of wife and husband perceptions of dyadic efficacy to outcomes. For example, wife dyadic efficacy may relate to different outcomes than husband dyadic efficacy. Also, a discrepancy between dyadic efficacy beliefs in partners may be related to poor outcomes. With more study, new ways to compute couple-level data from wife and husband scores may prove informative.

Longer-term studies in other types of dyads may provide a deeper understanding of how dyadic efficacy functions. In addition, future work should examine the feasibility of interventions to enhance dyadic efficacy in couples in clinical and community settings. In arthritis self-management programs (4), self-efficacy has been successfully enhanced, leading to positive physical and behavioral changes. After gaining a better understanding of how dyadic efficacy functions and the modifiable behaviors it represents, it may be possible to add to such programs to promote dyadic efficacy in couples.

This research offers support for a dyadic approach to understanding how couples cope with a long-term illness. Partners' enhanced sense of confidence concerning the ability to work together may be the first step involved in constructive dyadic behaviors. These new instruments can be used by researchers interested in interpersonal relationships and health.

AUTHOR CONTRIBUTIONS

Dr. Sterba had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Sterba, R. DeVellis, Lewis.

Acquisition of data. Sterba.

Analysis and interpretation of data. Sterba, R. DeVellis, Lewis, Baucom, Jordan, B. DeVellis.

Manuscript preparation. Sterba, R. DeVellis, Lewis, Baucom, Jordan, B. DeVellis.

Statistical analysis. Sterba, R. DeVellis.

Ancillary