Effects of Relationship Maintenance on Psychological Distress and Dyadic Adjustment Among Couples Coping with Lung Cancer

Abstract

OBJECTIVE:

Relationship maintenance strategies help to ensure the continuation of valued relationships by keeping them at a certain level of intimacy. We evaluated how lung cancer patients’ and spouses’ efforts to maintain their relationships affected their psychological and marital adjustment over time.

DESIGN:

Psychosocial questionnaires were administered within 1 month of lung cancer treatment initiation (baseline) and 3 and 6 months later to 158 lung cancer patients and their spouses.

MAIN OUTCOME MEASURES:

Study outcomes were global severity index scores on the Brief Symptom Inventory, and total scores on the Dyadic Adjustment Scale.

RESULTS:

Multilevel modeling analyses using the Actor-Partner Intedependence Model showed that, regardless of gender or social role (i.e., patient or spouse), individuals who engaged in the strategies of positivity, networks, and shared tasks reported less distress at baseline than other subjects. Over time, the effects of providing more assurances and experiencing a partner’s increased reliance on social networks differed: patient distress was exacerbated, and spouse distress was alleviated. Couples where both partners engaged in more frequent maintenance behaviors reported greater dyadic adjustment at baseline and over time.

CONCLUSION:

For couples coping with lung cancer, the initial treatment period may be an important time that sets the tone for future spousal interactions. Engaging in relationship maintenance during this stressful time may help mold more resilient relationships and facilitate adjustment as the disease progresses.

“A successful marriage is an edifice that must be rebuilt every day.”

Andre Maurois (1885–1967)

Social support from spouses is an important predictor of patients’ adaptation to cancer (Giese-Davis, Hermanson, Koopman, Weibel, & Spiegel, 2000; Manne, Pape, Taylor, & Dougherty, 1999). Positive marital interaction is associated with greater quality of life (QOL) among patients with cancer (Manne et al., 2004), who often refer to their spouses as their most valuable source of support (Neuling & Winefield, 1988). While most marriages are resilient in the face of cancer (Rieker, Fitzgerald, & Kalish, 1990; Schmidt et al., 1993), the risk for psychological (Hagedoorn, Buunk, Kuijer, Wobbes, & Sanderman, 2000) and marital distress (Northouse, Templin, Mood, & Oberst, 1998) increases, with some patients blaming their disease for relationship problems and even the dissolution of their marriages (Hodges, Humphris, & Macfarlane, 2005; Kornblith, Anderson, & Cella, 1990).

The marital literature has shown that stressors in domains external to the relationship can change partners’ relationship satisfaction and their interaction patterns (Bodenmann, 1997; Bolger, DeLongis, Kessler, & Wethington, 1989). This phenomenon is known as stress spillover. Chronic stressors (like cancer) may spill over and adversely affect marital well-being because they introduce opportunities for conflict and strain that the couple would not have otherwise experienced (Karney, Story, & Bradbury, 2005). For example, cancer can alter a couple’s finances, division of labor, and interaction patterns (Burman & Margolin, 1992). Indeed, many couples report problems communicating after cancer (Lichtman, Taylor, & Wood, 1987). This is concerning because communication problems are a primary cause of marital dissolution (Cleek & Pearson, 1985) and are linked to declines in marital quality (Caughlin, 2002) and a lack of social support (Reynolds & Perrin, 2004).

The interpersonal nature of the cancer experience necessitates a greater understanding of the ways in which couples maintain their relationships while coping with cancer (Ey, Compas, Epping-Jordan, & Worsham, 1998). Some researchers have argued that marital adjustment is an important component of the support process (Revenson, 2003), and, in turn, to psychological adjustment to cancer (Banthia et al., 2003). From this perspective, understanding the ways couples maintain caring, supportive relationships may aid the development of interventions to minimize the potentially negative effects of cancer on psychological and marital adjustment.

Relationship Maintenance

Relationship maintenance strategies are communication behaviors used to ensure the continuation of valued relationships by keeping them at a certain level of intimacy (Ayres, 1983). Canary and Stafford (1992) identified five such strategies: 1) positivity, or interacting with one’s partner in a cheerful and optimistic manner; 2) openness, which refers to discussing and disclosing information about the relationship with one’s partner; 3) assurances, which are messages that stress one’s desires to continue the relationship; 4) social networks, which entails relying on, or interacting with common relatives/friends; and, 5) shared tasks, which involves maintaining the relationship by partaking in everyday activities such as housework. Although positivity, openness, and assurances most clearly represent communication-based approaches for maintaining relationships, Stafford and colleagues (2000) have argued that sharing tasks and using common social networks are communicative in the sense that they take on a symbolic value and indicate commitment to the relationship and to one’s partner.

Maintenance strategies promote important relational characteristics (i.e., liking, commitment) that motivate people to engage in other pro-relationship behaviors over time (Canary, Stafford, & Semic, 2002), help repair relationships, and prevent them from decaying (Dindia & Baxter, 1987; Guerrero, Eloy, & Wabnik, 1993). Theory and research also suggest that they promote relational resilience. In a longitudinal study of healthy married couples, Canary et al. (2002) showed that marital resilience was related to how spouses maintained their relationships in the face of normative stressors; Luthar et al. (2000) conceptualized resilience as supportive behaviors that promote positive accommodation in the face of major adversity. Despite their differences, both studies portrayed positive adaptation to stressors as vital to maintaining relationship quality. Patterson (2002) has argued that understanding resilience depends on the identification of processes that moderate the relationship between a family’s exposure to risk and their ability to maintain competence and accomplish family functions. Because couples coping with cancer must deal with the crisis of diagnosis and the daily challenges of living with the sequelae of cancer and its treatment, understanding the strategies that allow couples to adapt and re-achieve relationship homeostasis is important. Although we are unaware of any studies before this one that have examined relationship maintenance in couples coping with cancer, focusing on the spousal relationship by engaging in these behaviors may enhance social support and fortify couples against the effects of cancer-related stress spillover on their relationships. It is also important to examine these associations over time; the effects of some maintenance strategies may not be immediately apparent. Likewise, maintenance strategies that are beneficial during acute stress (i.e., treatment initiation) may or may not have long term benefit.

Spousal Relationships and Lung Cancer

For couples, cancer-caused stressors may vary by cancer site, stage, and other disease characteristics; however, most descriptive and intervention research has focused on couples coping with cancer diagnosed at typically early stages such as breast cancer, than on cancers diagnosed at typically advanced stages such as lung cancer. Further, differences in relationship and psychological responses to cancer may be due to gender-related tendencies to solicit and provide emotional support. Disentangling the dual effects of gender (male/female) and social role (patient/spouse) is difficult because most relevant studies have been conducted in breast cancer (Figueiredo, Fries, & Ingram, 2004; Manne et al., 2006) rather than cancers affecting men and women more equally, like lung cancer (American Cancer Society, 2007). One exception was a Northouse et al. (2000) study of couples coping with colorectal cancer. That study found that partners reported more emotional distress than patients regardless of gender.

Maintaining a healthy marriage may be particularly important for lung cancer patients, as they have a poor prognosis (American Cancer Society, 2007) and report a variety of symptoms, including pain, dyspnea, fatigue, functional decline, and anorexia (Munro & Potter, 1996); 15% to 44% report some form of depression (Hopwood & Stephens, 2000). Spouses may experience equal or greater distress than cancer patients (Northouse, Mood, Templin, Mellon, & George, 2000), though this has not been examined in lung cancer. Spouse well-being is doubly critical to the lung cancer patient, whose QOL may be enhanced by a healthy relationship and by a spouse able to provide caregiving.

Although lung cancer patients have a life-threatening illness and face an uncertain future in which higher relationship satisfaction may be important for continued QOL, no studies have directly examined their spousal relationships. This is surprising since lung cancer patients view their family as a significant QOL area (Montazeri, Milroy, Gillis, & McEwen, 1996). Lung cancer may increase the risk for relationship distress more than other cancers because of its morbidity (Hopwood & Stephens, 2000), and stigma or attributions of blame regarding its cause (Chapple, Ziebland, & McPherson, 2004). Limited research also indicates the social constraints unique to lung cancer may make it more difficult for couples to engage in open, honest discussions (Badr & Carmack Taylor, 2006; Zhang & Siminoff, 2003). In fact, a cross-sectional study of disease-free survivors of lung, colon, and prostate cancer found lung cancer survivors had the most problems communicating and interacting with their partners (Schag, Ganz, Wing, Sim, & Lee, 1994). Thus, lung cancer may introduce psychological and relationship challenges that may persist long after the initial diagnosis and treatment period. By engaging in relationship maintenance, patients and spouses may draw greater strength and support from each other and ultimately experience less distress and greater marital adjustment.

The purpose of this study was 1) to examine the associations of participants’ self-reported relationship maintenance behaviors within 1 month of treatment initiation (baseline) and their own/partner’s psychological distress and dyadic adjustment; 2) to examine these associations over time (3 and 6 months later); and, 3) to explore whether these associations are affected by one’s gender and/or social role. Our expectation was that one’s own use of maintenance strategies and one’s partner’s use of maintenance strategies at baseline would be positively associated with dyadic adjustment and negatively associated with psychological distress at baseline and over time. Although the marital literature suggests gender differences in relationship maintenance (Stafford & Canary, 1991), the effects of gender have not been simultaneously examined with the effects of social role in the context of cancer. Thus, we examined gender and social role in our analyses but did not formulate any formal hypotheses about their effects.

Methods

Procedure

This 6-month longitudinal study of lung cancer patients who were within 1 month of treatment initiation (baseline) and their spouses was approved by The University of Texas M. D. Anderson Cancer Center (MDACC) institutional review board. Potentially eligible patients were identified based on a review of their medical charts; during their clinic visits, research staff approached all those who met the inclusion criteria for participation. Eligibility was confirmed at the time of recruitment. Participants were included if they provided informed consent, spoke and understood English, and had a spouse or significant other with whom they had lived for at least 1 year. Patients also had to have a physician-rated Eastern Cooperative Oncology Group (ECOG) performance status score ≤ 2 (i.e., ambulatory and capable of all self-care but unable to carry out any work activities). To have a representative sample of patients, we used a stratified sampling procedure such that participants fell into one of three therapy/disease stage categories: patients undergoing curative surgery; non-metastatic inoperable patients receiving combined modality therapy; and patients with metastatic disease. Couples who agreed to participate were asked to complete their own surveys separately and to return them in individually sealed postage-paid envelopes. Follow-up surveys were mailed out 3 and 6 months later to couples who had returned the baseline survey. Participants received compensation worth $10 upon the return of each completed survey (up to $60 total per couple).

Measures

Psychological distress.

The 53-item Brief Symptom Inventory (BSI; Derogatis, 1993) yields 9 subscale scores and a Global Severity Index (GSI). T-scores ≥ 63 on the GSI or on two or more of the BSI’s 9 primary dimensions indicate distress. The BSI has been validated in a range of medical populations including cancer (Zabora, Brintzenhofeszoc, Curbow, Hooker, & Piantadosi, 2001).

Dyadic adjustment

Dyadic adjustment was assessed with the 32-item Dyadic Adjustment Scale (DAS; Spanier, 1976) which assesses 4 components of marital adjustment: consensus, satisfaction, cohesion, and affectional expression. The total score cutoff for marital distress is 97 (Jacobson, Schmaling, & Holtzworth-Munroe, 1987). In this study, internal consistency reliability for the DAS was high (α = .91 for patients; α = .90 for spouses).

Relationship maintenance

Relationship maintenance was assessed using the 29-item Relationship Maintenance Strategies Measure (RMSM), comprising 5 subscales: positivity, openness, assurances, social networks, and sharing tasks (Stafford & Canary, 1991). Participants were asked to describe how they maintained their relationships with their partners over the last two weeks (1 = strongly disagree to 7 = strongly agree). Internal consistency reliabilities for the RMSM subscales ranged from α = .85 to .93 for patients and α = .76 to .92 for spouses.

Sample

During recruitment, research staff approached 460 patients and their partners. Of these, 116 patients (25%) were ineligible (55 were beyond the treatment requirement; 35 were single and did not have a significant other with whom they currently lived; 14 did not have a confirmed lung cancer diagnosis; 9 did not speak English; and 3 could not provide informed consent). Of the 344 eligible patients remaining, 74 (22%) refused to participate (31 were too distressed; 34 were not interested; 5 felt too old to participate, 3 were too busy, and 1 had been approached by too many research studies). Those who consented but did not return surveys within 2 weeks were reminded to do so with phone calls, letters, and a second set of surveys. Still, 92 of the 270 couples did not return the baseline survey. In 20 of the 178 remaining couples, only one partner returned the survey (11 patients and 9 spouses), resulting in 158 couples with complete baseline data.

Comparisons were made between participants and non-participants (those who refused and those who consented but did not return the surveys) based on available data for age, type of cancer (NSCLC vs. SCLC), therapy/disease stage category, ECOG performance status, and race/ethnicity). The only significant differences were for ECOG χ² (2, 314)=11.85, p=.003, and ethnicity χ² (1, 342)=3.77, p=.05; 94% of non-participants and 87% of participants had an ECOG of 0 or 1, and 20% of non-participants and 14% of participants were minorities.

Two patients who completed the baseline survey died before the 3-month assessment, so only 156 3-month surveys were mailed. Of the 130 couples who returned the 3-month surveys, complete data (from both partners) was obtained from 119 couples (75%). Before the 6-month survey was mailed, 4 additional patients died, 3 couples dropped out, and 9 couples were lost to follow-up resulting in a third mail-out of 140 surveys. Fourteen couples did not return the 6-month survey because the patient had died. Of the 108 couples (68% of the original 158) who returned the 6-month surveys, 97 couples (69% of the 140 surveys mailed) had complete data. Comparisons were made between study completers and non-completers based on age, type of cancer, therapy/disease stage category, performance status, race/ethnicity, dyadic adjustment, and psychological distress. However, no significant between-group differences were found.

Analysis Plan

Descriptive statistics (means, standard deviations, and pooled within-couple correlations) were calculated for each of the major study variables at each assessment time point. A series of one-way analyses of variance (ANOVAs) were conducted to examine whether there were any differences in the variables of interest by disease stage. Over-time differences based on social role and gender were also tested using a series of repeated measures ANOVAs; time (baseline, 3-month, and 6-month assessments) was treated as a within-subjects factor, and gender (man vs. woman) and social role (patient vs. spouse) were treated as between-subjects factors. Interactions between time, gender, and social role were also examined.

Baseline analyses.

Because data from married couples tend to be related, analyses must adjust for this non-independence so that statistical significance tests are not biased, and model the interdependence or mutual influence process itself. The Actor-Partner Interdependence Model (APIM) accomplishes both goals by utilizing a multilevel modeling approach in which data from two dyad members are treated as nested scores within the same group (i.e., marital dyad) (Kenny, Kashy, & Cook, 2006). The APIM suggests that a person’s independent variable score affects his or her own dependent variable score (known as the actor effect), and his or her partner’s dependent variable score (known as the partner, or cross-spouse effect). We can also determine whether these effects differ depending on whether the actor is a patient or spouse and/or a man or a woman. The APIM can be specified as a mixed model with random effects representing individuals and couples. One of its strengths is that it can handle missing data and maximize the use of existing data, which is particularly useful in longitudinal studies.

A series of APIM analyses using SAS Proc Mixed were conducted to examine the baseline actor and partner associations for each relationship maintenance behavior with the outcomes of psychological distress (BSI GSI scores) and dyadic adjustment (DAS total scores), controlling for participant age. We also tested whether the association between a specific relationship maintenance behavior and the outcome of interest differed depending on gender (1 = men and −1 = women) and/or social role (1 = patients and −1 = spouses). The continuous predictor variables were standardized, and the heterogeneous compound symmetry option in SAS was used to allow the error terms to differ for the two dyad members. The effect size r, associated with each t was calculated using the formula $\sqrt{t^{\text{2}}/t^2+df}$ (Snijders & Bosker, 1999).

Longitudinal analyses.

In analyzing longitudinal dyadic data, three factors (time, person, and dyad) are involved. Researchers often err by considering these data to be a three-level nested model in which time points are nested within persons and persons are nested within dyads. However, the level of time is the same for both members of the dyad. If the three-level nested model is assumed, the correlation between the two partner’s intercepts is constrained to be positive and the correlation between the two members’ errors at each time is assumed to be zero. An alternative approach that does not force these constraints is an expansion of the two-intercept model recommended by Kenny, Kashy, & Cook (2006). With this approach, the effect of time is averaged and the effects of past on present are estimated for both partners.

We examined psychological distress (BSI GSI scores) and marital adjustment (DAS scores) across the 3 and 6 month follow-ups as a function of baseline RMSM scores controlling for Time 1 (baseline) scores on the outcome variables. The longitudinal dataset was structured with one record for each time point for each dyad member. Lagged values of the dependent variables were grand-mean centered (across couples and time) to allow for straightforward interpretation of the intercepts. We also allowed the errors to be auto-correlated as suggested by Bolger and Shrout (2007) by defining the error structure as a lag 1 autoregressive structure. In SAS, this is accomplished by the TYPE=UN@AR(1) option in the repeated statement.

Results

Descriptive Analyses

Patients’ average age was 62.86 years (SD = 10.14). The majority were male (62.7%); not Hispanic or Latino (95.3%); white (88.2%); educated with some college level credits, a 2-year degree or higher (61.5%); retired (50.6%); and married (97.6%). Disease stage at study entry was 16.3% Stage 1, 14.5% Stage 2, 32.5% Stage 3, and 36.7% Stage 4. Of the non-metastatic patients, 29% were undergoing curative surgery and 36% were undergoing combined modality therapy. Spouses’ average age was 60.39 (SD=11.08). The majority were female (67.1%); not Hispanic or Latino (96.4%); white (91.5%); educated with some college level credits, a 2-year degree, or higher (58.1%); and employed full-time (41.8%).

Analysis of baseline DAS scores show that the sample was maritally satisfied (M = 118.82, SD = 18.46). Twenty-three spouses (14%) and 18 patients (11%) scored below the DAS cut-off for marital distress. Fifty-two patients (29%) and 50 spouses (33%) met the BSI criteria for distress at baseline. Table 1 shows the Means and SD’s on the major study variables by gender and social role for each assessment period. Table 2 shows the pooled correlations between patients and spouses across assessments. With the exception of shared tasks, low to moderate associations between patients’ and spouses’ scores on the major study variables were found (r=.19 to .50). The association between patient and spouse reports of dyadic adjustment, positivity, openness, and networks grew significantly weaker over time, whereas reports of assurances grew stronger over time. No significant differences based on disease stage were found for any of the relationship maintenance strategies (p’s = .47 to .99), or for the study outcomes of psychological distress (p=.82) and dyadic adjustment (p=.56).

Table 1.

Means ± Standard Deviations for Major Study Variables by Role and Sex

Variable	Patients		Spouses
	Men	Women	Men	Women
RMSMa
Positivity
Baseline	5.55 ± 1.04	5.41 ± 1.10	5.82 ± 1.09	5.84 ± .94
3 months	5.44 ± 1.16	5.66 ± .96	5.74 ± 1.03	5.71 ± 1.01
6 months	5.47 ± 1.08	5.54 ± 1.08	6.07 ± .91	5.77 ± .90
Openness
Baseline	4.98 ± 1.31	5.04 ± 1.29	4.98 ± 1.15	5.29 ± 1.18
3 months	4.90 ± 1.30	4.97 ± 1.24	4.89 ± 1.15	5.09 ± 1.12
6 months§	4.92 ± 1.36	4.93 ± 1.32	5.26 ± 1.08	5.00 ± 1.09
Assurances
Baseline	5.92 ± 1.08	5.91 ± 1.15	6.13 ± 1.15	6.22 ± .97
3 months	5.83 ± 1.22	6.02 ± 1.09	6.06 ± .85	6.16 ± .93
6 months	5.98 ± .93	5.99 ± 1.12	6.37 ± .88	6.08 ± .77
Network
Baseline	5.22 ± 1.28	5.34 ± 1.39	5.53 ± 1.12	5.59 ± 1.20
3 months	5.26 ± 1.21	5.51 ± 1.21	5.26 ± 1.13	5.40 ± 1.34
6 months	5.23 ± 1.20	5.43 ± 1.21	5.63 ± .98	5.33 ± 1.24
Tasks
Baseline†§	5.59 ± 1.23	5.77 ± 1.29	5.95 ± 1.20	6.34 ± .80
3 months‡§	5.38 ± 1.39	5.72 ± 1.29	6.12 ± 1.01	6.22 ± .79
6 months‡§	5.50 ± 1.43	5.60 ± 1.32	6.36 ± .79	6.20 ± .70
BSI GSIb
Baseline	.38 ± .35	.47 ± .41	.38 ± .39	.48 ± .42
3 months†‡	.40 ± .40	.44 ± .27	.24 ± .26	.45 ± .38
6 months†	.38 ± .38	.48 ± .36	.20 ± .31	.50 ± .40
DASc
Baseline	122.17 ± 17.14	115.52 ± 20.63	119.59 ± 17.67	117.07 ± 18.48
3 months	117.11 ± 20.17	119.25 ± 17.23	120.29 ± 17.94	113.98 ± 22.40
6 months†	121.71 ± 13.95	118.89 ± 16.35	126.61 ± 16.19	114.92 ± 19.36

^aRelationship Maintenance Strategies Measure scores

^bBrief Symptom Inventory Global Severity Index raw scores

^cDyadic Adjustment Scale total score

^†Mean scores for male and female spouses are significantly different at the p<.05 level

^‡Mean scores for male patients and male spouses are significantly different at the p<.05 level

^§Mean scores for female patients and female spouses are significantly different at the p<.05 level

Table 2.

Pooled Within-Couple Correlations at Each Assessment

Variable	Effect size (r)	95% Confidence interval	P
RMSM†
Positivity
Baseline	.21	.10 – .32	<.001
3 months	.23a	.14 – .38	<.001
6 months	.15b	.001 – .29	<.05
Openness
Baseline	.24	.13 – .34	<.001
3 months	.26a	.14 – .38	<.001
6 months	.19b	.04 – .32	<.01
Assurances
Baseline	.20a	.09 – .31	<.001
3 months	.29b	.17 – .40	<.001
6 months	.30b	.17 – .43	<.001
Networks
Baseline	.43a	.33 – .51	<.001
3 months	.34b	.22 – .46	<.001
6 months	.34b	.20 – .46	<.001
Shared tasks
Baseline	−.04	−.16 – .07	NS
3 months	.04	−.09 – .17	NS
6 months	−.21	−.34 – −.06	NS
BSI GSI‡
Baseline	.19	.07 – .30	<.001
3 months	.22	.09 – .34	<.001
6 months	.19	.04 – .34	<.001
DAS§
Baseline	.50a	.41 – .58	<.001
3 months	.35b	.23 – .46	<.001
6 months	.28c	.14 – .41	<.001

Note. For variables where significant within-couple associations were found, differences across time were examined. For each of these variables, effect sizes with superscripts of different letters significantly differ at the p≤.05 level.

NS = not significant

^†Relationship Maintenance Strategies Measure scores

^‡Brief Symptom Inventory Global Severity Index raw scores

^§Dyadic Adjustment Scale total score

No significant changes in dyadic adjustment, positivity, openness, and assurances were found. A significant time × gender effect was found (F(2, 160) = 5.48, p=.005) indicating that women became more distressed and men became less distressed over time, regardless of role. A significant time × health effect was found for social networks (F(2,160) = 3.25, p=.04), indicating that spouses relied more on social networks over time than did patients, regardless of gender. Finally, a significant time × health effect was found for shared tasks (F(2,160)=4.65, p=.01), indicating that patients performed fewer tasks and spouses performed more tasks over time.

Baseline Analyses

Psychological distress.

Although we found no significant actor or partner interactions between any of the relationship maintenance behaviors and either gender or social role, significant main effects were found for the actor effects of positivity (t(140) = −4.46, p = .001; r = .34), common social networks (t(140) = −3.63, p = .003; r = .28), and shared tasks (t(140) = −2.62, p = .01; r = .21). In each case, more relationship maintenance was associated with less distress. Overall, the cross-spouse (partner) effects for relationship maintenance were not significant. For illustrative purposes, Figure 1 depicts the mixed models coefficients for the actor and partner associations of positivity, social networks, and shared tasks with psychological distress for patients and spouses separately. Also depicted is the correlation, r, between patient and spouse relationship maintenance scores and the correlation between the two error terms, e, which represents the residual non-independence in the outcome scores. By examining the individual coefficients for the actor and partner effects for patients and spouses separately, we discovered that even though the overall partner effects for relationship maintenance had no significant association with patient distress, patient engagement in the maintenance strategies of positivity, networks, and shared tasks was significantly negatively associated with spouse distress.

Figure 1.

Results of APIM baseline analysis regressing BSI GSI scores on patient and spouse self-reports of positivity, social networks, and shared tasks.

Note: **p≤.01, *p≤.05

Unless otherwise specified (i.e., r=correlation, e=error term), model coefficients for the actor and partner effects of engaging in different relationship maintenance strategies on psychological distress for both patients and spouses are presented.

Dyadic adjustment.

Similar to the findings for psychological distress, no significant actor or partner interactions between any of the relationship maintenance behaviors and either gender or social role were found. Results of the main effects analyses showed that both the actor (t(140) = 11.74, p = .001; r = .69) and partner (t(140) = 7.47, p = .001; r = .52) effects for positivity were significant, as were the actor and partner effects for openness (t_actor (140) = 8.55, p = .001; r = .57; t_partner (140) = 5.31, p = .001; r = .40), assurances (t_actor (140) = 9.47, p = .001; r = .61; t_partner (140) = 6.14, p = .001; r = .45), social networks (t_actor (140) = 8.67, p = .001; r = .58; t_partner (140) = 3.59, p = .001; r = .28), and shared tasks (t_actor (140) = 6.31, p = .001; r = .46; t_partner (140) = 5.08, p = .001; r = .38). Collectively, these findings suggest that individuals benefit in terms of greater dyadic adjustment when they engage in more relationship maintenance and when their partners engage in more relationship maintenance. Although the effect size estimates (r) suggest that actor effects were stronger than partner effects, no significant differences between actor and partner effects were found. To further clarify the associations, Figure 2 depicts the mixed models coefficients for these analyses for patients and spouses separately. Also depicted is the correlation, r, between patient and spouse relationship maintenance scores and the correlation between the two error terms, e.

Figure 2.

Results of APIM baseline analysis regressing DAS scores on patient and spouse self-reports of relational maintenance.

Note: **p≤.01, *p≤.05

Unless otherwise specified (i.e., r=correlation, e=error term), figure 2 presents the model coefficients for the actor and partner effects of engaging in different relationship maintenance strategies on dyadic adjustment for both patients and spouses.

To determine whether engaging in relationship maintenance behaviors at diagnosis/during treatment initiation had any effects across the 3 and 6 month assessments for patients and spouses, a series of longitudinal analyses were conducted.

Longitudinal Analyses

Psychological distress.

No significant associations with gender were found. Given this, we re-ran the analyses with only social role and the actor and partner effects of each of the five relationship maintenance behaviors included in the model statement. Interactions between social role and the actor and partner effects were also included. No significant actor, partner, or interaction effects were found for positivity, openness, or shared tasks.

Table 3 details the results for assurances and social networks. Although the effects of one’s partner engaging in these relationship maintenance strategies (partner effect) and the interaction between partner’s maintenance and actor’s social role (whether the actor was a patient or spouse) were not significant, the effects of one engaging in assurances and social networks (actor effects) and the interactions between the actor engaging in these relationship maintenance strategies and the actor’s social role were significant. Analysis of the coefficients for the actor’s maintenance × actor’s social role interaction showed that patients who provided more assurances to their partners at baseline reported slightly more distress over time (b = .05), whereas spouses who reported providing more assurances at baseline reported slightly less distress over time (b = −.05). A slightly different pattern emerged for common social networks. Namely, the actor effect of social networks and the actor social networks × actor’s social role interaction were not significant, but the partner effect of social networks and the partner social networks × actor’s social role interaction were significant. Analysis of the mixed model coefficients for the interaction showed that patients reported more distress when their spouses relied more on common social networks (b = .05) and spouses reported less distress when patients relied more on common social networks (b = - .05).

Table 3.

Actor-Partner Interdependence Model (APIM) Analysis of the Over-Time Actor and Partner Effects of Assurances and Networks on Psychological Distress in Couples Coping with Lung Cancer

	Effects of assurances				Effects of networks
	B	SE	t	Effect size (r)	B	SE	t	Effect size (r)
Patient intercept	.45	.04	-	-	.44	.04	-	-
Spouse intercept	.41	.05	-	-	.41	.12
Actor’s Social Role	.04	.06	.54	.03	.03	.08	.33	.02
Actor’s Maintenance	−.04	.02	−2.10*	.11	−.03	.02	−1.89#	.10
Partner’s Maintenance	−.02	.02	−.83	.04	−.05	.02	−2.39*	.13
Actor’s Maintenance × Actor’s Social Role			2.51**	.13			.87	.05
Effects of patient’s maintenance on patient’s distress	.05	.02			.02	.02
Effects of spouse’s maintenance on spouse’s distress	−.05	.02			−.05	.02
Partner’s Maintenance × Actor’s Social Role			1.41	.08			3.26***	.17
Effects of patient’s maintenance on spouse’s distress	−.03	.02			−.05	.02
Effects of spouse’s maintenance on patient’s distress	.03	.02			.05	.02

Note. B = raw coefficient, SE = standard error; effect size r associated with each t: r = $\sqrt{t^{\text{2}}/t^2+df}$,

^#p<.10,

^*p<.05,

^**p<.01,

^***p<.005

Dyadic adjustment.

Similar to the findings for psychological distress, no significant associations with gender were found. Thus, the analyses were re-run with social role and the actor and partner effects of the each of five relationship maintenance behaviors included in the model statement. Interactions between social role and the actor and partner effects were also included. Table 4 presents the results of the APIM mixed models analyses.

Table 4.

Actor-Partner Interdependence Model (APIM) Analysis of the Over-Time Actor and Partner Effects of Relationship Maintenance Strategies on Dyadic Adjustment in Couples Coping with Lung Cancer

	Effects of positivity				Effects of openness
	B	SE	t	Effect size (r)	B	SE	t	Effect size (r)
Patient intercept	119.93	1.57	-	-	119.65	1.62	-	-
Spouse intercept	117.21	1.70	-	-	117.78	1.78
Actor’s Social Role	2.53	1.68	1.50	.08	2.07	3.21	.64	.03
Actor’s Maintenance	6.70	1.30	5.16***	.27	3.33	1.27	2.63**	.10
Partner’s Maintenance	5.56	1.33	4.17**	.22	3.36	1.14	2.95***	.16
Actor’s Maintenance × Actor’s Social Role			−.32	.02			.83	.04
Effects of patient’s maintenance on patient’s DAS	4.85	1.17			3.50	.91
Effects of spouse’s maintenance on spouse’s DAS	6.59	1.28			3.76	1.17
Partner’s Maintenance × Actor’s Social Role			−1.76#	.10			−2.23*	.12
Effects of patient’s maintenance on spouse’s DAS	1.91	1.20			.13	1.07
Effects of spouse’s maintenance on patient’s DAS	3.89	1.37			2.67	1.09
	Effects of assurances				Effects of networks
	B	SE	t	Effect size (r)	B	SE	t	Effect size (r)
Patient intercept	119.85	2.38	-	-	119.47	1.58	-	-
Spouse intercept	117.37	1.99			117.52	2.00
Actor’s Social Role	2.39	1.71	1.40	.07	2.12	1.77	1.21	.06
Actor’s Maintenance	5.89	1.39	4.23***	.22	3.45	1.21	2.79***	.15
Partner’s Maintenance	4.65	1.26	3.69***	.19	3.59	1.66	2.99***	.16
Actor’s Maintenance × Actor’s Social role			.08	.004			.84	.04
Effects of patient’s maintenance on patient’s DAS	4.79	1.08			4.28	.94
Effects of spouse’s maintenance on spouse’s DAS	6.06	1.35			4.57	1.09
Partner’s Maintenance × Actor’s Social Role			−1.85#	.10			−1.73#	.11
Effects of patient’s maintenance on spouse’s DAS	3.89	1.23			1.72	1.00
Effects of spouse’s maintenance on patient’s DAS	.82	1.27			3.46	1.14
	Effects of shared tasks
	B	SE	t	Effect size (r)
Patient intercept	119.47	1.58	-	-
Spouse intercept	117.52	1.99
Actor’s Social Role	1.90	1.95	.97	.05
Actor’s Maintenance	2.85	1.50	1.89#	.10
Partner’s Maintenance	3.35	1.16	2.89*	.15
Actor’s Maintenance × Actor’s Social role			−.17	.01
Effects of patient’s maintenance on patient’s DAS	1.31	.99
Effects of spouse’s maintenance on spouse’s DAS	2.23	1.44
Partner’s Maintenance × Actor’s Social Role			−.88	.05
Effects of patient’s maintenance on spouse’s DAS	1.30	1.26
Effects of spouse’s maintenance on patient’s DAS	2.64	1.12

Note. B = raw coefficient, SE = standard error, DAS = Dyadic Adjustment Scale total score; effect size r associated with each t: r = $\sqrt{t^{\text{2}}/t^2+df}$,

^#p<.10,

^*p<.05,

^**p<.01,

^***p<.005

Significant actor effects were found for positivity, indicating that one’s own positivity enhanced his/her dyadic adjustment over time. Significant partner effects were also found, indicating that patients and spouses had better dyadic adjustment over time when their partners were more positive at baseline. Although the actor’s positivity × actor’s social role interaction was not significant, the partner’s positivity × actor’s social role interaction approached significance (p = .08), suggesting a trend for spouses to benefit more from patient positivity (b = 3.89) than patients benefited from spouse positivity (b = 1.91) in terms of greater dyadic adjustment over time.

For the strategy of openness, the actor’s openness × actor’s social role interaction was not significant; patients and spouses did not significantly differ with respect to the effects of their own openness on their own dyadic adjustment. However, the partner’s openness × actor’s social role interaction was significant; spouses’ dyadic adjustment benefited more from patient openness (b = 2.67) than patients’ dyadic adjustment benefited from spouse openness (b = .13).

For assurances, although the actor’s assurances × actor’s social role interaction was not significant, the partner’s assurances × actor’s social role interaction approached significance (p = .07), suggesting a trend for spouses to benefit over time in terms of greater dyadic adjustment from patient assurances (b = 3.89) than for patients to benefit from spouse assurances (b = .82). For social networks, the actor’s social networks × actor’s social role interaction was not significant; however, the partner’s social networks × actor’s social role interaction approached significance (p = .08), showing a trend for spouses to benefit more from patient reliance on common social networks (b = 3.46) than for patients to benefit from spouse reliance on common social networks (b = 1.72). Finally, for tasks, no significant interactions with social role were found. The actor main effect for tasks approached significance (p < .10), and the partner main effect was significant (p<.05), suggesting that patients and spouses reported greater dyadic adjustment when their partners performed more shared tasks.

Discussion

This study showed how engaging in relationship maintenance behaviors within 1 month of treatment initiation for lung cancer was associated with the psychological and marital adjustment of both partners and how efforts to maintain or enhance the relationship during this period affected psychological and marital adjustment over time. Individuals who engaged in pro-relationship behaviors adapted to living with lung cancer better than those who did not: specifically, patients and spouses who reported greater use of the maintenance strategies of positivity and networks, and patients who reported engaging in more shared tasks reported less distress at baseline. When individuals communicate positivity and hopefulness to their partners at diagnosis and take a more active role in the relationship by sharing tasks and seeking support from common social networks, they may feel they are doing something constructive to alleviate their own and their partners’ stress. In the process, they may experience less distress at diagnosis, regardless of gender or social role.

While our findings are consistent with research emanating from the social support literature showing that providing support is beneficial to the provider as well as to the recipient (Brown, Neese, Vinokur, & Smith, 2003), they also extend this literature by suggesting that the benefits of engaging in relationship maintenance may depend not only on the outcome being examined (e.g., psychological distress or dyadic adjustment), but also may vary over time for patients and spouses. Specifically, we found that own and partner engagement in each of the five maintenance strategies helped to enhance patient and spouse dyadic adjustment over time. In contrast, providing assurances and partners’ reliance on social networks were significantly associated with psychological distress over time and their effects varied depending on social role. These findings are important because they suggest that psychological and marital adjustment to cancer may be tied together by relationship processes. They also highlight the utility of taking a “relationship-perspective” when studying the role of marriage in couples’ psychosocial adjustment to cancer.

Taking a relationship-perspective means that it is not only important to view the marriage as a resource for both partners to draw on but also to consider what each partner invests (or puts) into the marriage to maintain equilibrium and/or enhance relationship quality. From this perspective, one would expect that focusing greater attention onto the relationship and investing in it by engaging in relationship maintenance (during a stressful period like a lung cancer diagnosis and treatment) would help strengthen the marital bond over time, resulting in both partners reporting a more satisfactory relationship. At the same time, individual differences such as social role (patient or spouse) may play a more prominent part when it comes to the “withdrawals” that are made from the marriage and their impact on psychological adjustment. Because of their different roles in the marriage, patients and spouses may differ with respect to what they expect or need from their partner and their relationship. This in turn may affect the impact of own or partner engagement in specific maintenance behaviors on psychological adjustment. This idea is supported by our baseline analyses in which patients’ positivity, reliance on common social networks, and engagement in shared tasks were significantly associated with spouse distress whereas the same behaviors enacted by spouses were not significantly associated with patient distress. One reason for this may be that patients’ who perform more tasks and seek outside support are doing physically better, and those who are more positive, may have less distress. This in turn may relate to their spouses being less distressed. At the same time, in a study of couples coping with chronic illness, Michela (1987) reported a related finding and postulated that, “[The patient’s] experience is filtered through concerns about surviving and recovering, while [the spouse’s] experience is filtered through the meaning of the marital relationship – what the marriage has provided, and hence, what is threatened by the [patient’s] potential death or what is lost by his disability” (p. 272). Thus, one avenue for future research is to examine whether factors such as spouses’ anticipatory grief surrounding the impending loss of not only their partner, but also their relationship attenuate the positive impact of patients’ maintenance on spouses’ adjustment.

Consistent with studies demonstrating gender differences in cancer-related distress (Hagedoorn, Buunk, Kuijer, Wobbes, & Sanderman, 2000), we found that women’s distress increased and men’s decreased over time, regardless of social role. However, no significant gender differences were found regarding the effects of relationship maintenance on either distress or dyadic adjustment. Gendered tendencies to engage in maintenance behaviors may become less salient when a life-threatening illness like lung cancer demands that partners adapt quickly to their new patient and caregiver roles. More research on mediators and moderators of these associations is needed to clarify whether the lack of baseline differences between patients and spouses is due to the equally devastating impact of a lung cancer diagnosis and whether changes in needs or expectations – as partners become entrenched in their roles as caregivers or patients – cause differences to develop over time.

Although few differences between patients and partners were evident at baseline, one unanticipated finding was that even though no significant over-time changes for patients and spouses on mean positivity, openness, assurances, and dyadic adjustment scores were found, the within couple correlations on these variables remained significant but grew significantly weaker - particularly from the 3 to 6 month assessments. Thus, from a couple-level perspective, social role differences in expectations regarding one’s relationship and the support received from one’s partner may intensify over time, which may suggest greater need to direct more attention onto the relationship and for both partners to engage in relationship maintenance behaviors as cancer progresses.

It is also important to consider the meaning patients and partners may attribute to maintenance behaviors. For example, we found that providing more assurances to a partner and an increasing partner reliance on social networks exacerbated patient distress but alleviated spouse distress. One explanation may be that patients who expect their partners to be supportive become distressed if they instead find themselves having to provide all the support and reassurance. They may also become distressed when their partners rely on other social network members for support because they may feel they are no longer able to fulfill their role of primary confidant. In contrast, spouses may experience less distress when reassuring the patient of their commitment to the relationship because it reinforces their beliefs about what a supportive spouse should do. They may also experience relief when patients rely on other social network members because it decreases their own feelings of caregiver burden. Clearly, more research is needed to clarify the nature of these associations as alternate explanations are possible (e.g., spouses of distressed patients may be more likely to seek outside support, whereas patients who seek outside support may be less distressed, which may positively affect spouses’ adjustment).

Our findings are consistent with research showing that talking with supportive partners facilitates the cognitive and emotional processing of cancer (Lepore, 2001; Manne, 1999; Manne et al., 2004). Unlike studies focusing on cancer-related discussions, our focus was on the communication strategies couples use to maintain their relationships while coping with cancer. Shifting focus to the relationship may reduce constraints and buffer couples from the negative effects of cancer on their well-being. Although we did not explicitly examine constraints or cancer-related discussions, future research should consider how marital processes are affected by a cancer diagnosis and treatment. As cancer progresses, changes in maintenance behaviors and everyday patterns of relating might impair couples’ willingness or ability to engage in cancer-related discussions. Those who focus attention onto the relationship may find it easier to relate and provide social support which may ultimately help them to initiate supportive cancer-related discussions and facilitate positive adaptation to cancer.

This study has benefited from several methodological strengths. First, it is one of very few studies in cancer to have examined marital processes from both partners’ perspectives, and it is the first study we are aware of to have simultaneously examined lung cancer patients and their spouses. A second strength was the APIM multilevel modeling approach, which allowed us to control for the non-independence of partners’ responses and to examine whether gender and social role affected the association between relational maintenance strategies and individual/marital outcomes. Third, the longitudinal design permitted examination of changes in psychological and marital adjustment. Fourth, our data was collected from a relatively homogeneous sample which provided a conservative test of our hypotheses. Because all patients were within 1 month of treatment initiation, the likelihood that the effects resulted from uncontrolled differences in the length of illness was reduced.

Despite its strengths, this study had some limitations. Because the RMSM is a self-report measure it may not reflect actual spousal behaviors. The RMSM was also developed to assess everyday maintenance behaviors in healthy couples, so its validity with couples coping with cancer has not been evaluated. We also had a considerable number of overt (22%) and passive refusals (i.e., couples who consented but did not complete the baseline questionnaire; 26%). Although our participation rate is comparable to other published studies of couples in the oncology setting (de Groot et al., 2005; Manne et al., 2006), it is important to note that our sample comprised patients initiating treatment for a cancer with a relatively poor prognosis. Thus, some couples may have had too much to deal with and completing a long survey was not a priority. Refusal may have also been due to marital or psychological distress; however, we did not collect this data at recruitment and cannot determine whether those who refused were more or less distressed than study participants. Future studies may benefit from administering a brief, anonymous assessment of psychosocial distress at recruitment to allow for such comparisons. A related issue is the drop-out rate. We found no significant differences between completers and non-completers in terms of baseline psychological and marital distress; however, those who dropped out may have had sharper declines in these outcomes over time than those who completed the study.

Our sample was relatively homogeneous in terms of race/ethnicity and marital satisfaction. Because most participants were white, we had insufficient power to examine socio-cultural differences. Moreover, satisfied couples may find it more rewarding and important to maintain their relationships than unsatisfied couples. Future studies might benefit from more heterogeneous samples and make special efforts to recruit non-white patients. Comparing the maintenance strategies employed by maritally satisfied and dissatisfied couples might help to clarify existing research findings that suggest that the purpose for engaging in relationship maintenance (e.g., to sustain, enhance, or repair a relationship) can affect the selection and efficacy of these strategies (Dindia & Baxter, 1987). Because we did not assess relationship maintenance prior to diagnosis, we also do not know if having cancer made couples more aware of their relationships and/or the need to maintain them, or whether their maintenance patterns were the same before diagnosis. Finally, even though more relationship maintenance behaviors were associated with dyadic adjustment as opposed to psychological distress for both patient and partner over time, we cannot definitively say whether relationship maintenance affected one outcome more strongly than the other due to the fact that separate analyses were conducted for each outcome. We hope to address this issue in future research as dyadic data analytic methods become increasingly more flexible and sophisticated.

Despite its limitations, this study highlighted the effects of relational maintenance on the psychosocial adjustment of lung cancer patients and their partners. We found that patients and spouses who reported more relationship maintenance at baseline reported greater dyadic adjustment at baseline and over time. Although the effect sizes for our over-time analyses were small, the emphasis of this study was on the effects of engaging in maintenance strategies for both the actor and his or her partner and the effects shown have clear clinical implications. One interesting finding was that patients who provided more assurances reported more distress over time and spouses who received more assurances (from patients) reported greater dyadic adjustment over time. Even though providing assurances of one’s commitment to the relationship may be taxing for the patient, it may reinforce the notion in the mind of the spouse that the patient intends to fight the disease and not leave the spouse (by dying). This in turn may help to reinforce the spouse’s commitment to the relationship and thus positively affect his/her own perceptions of dyadic adjustment. When designing couples-focused interventions, researchers may thus need to adjust for the changing and sometimes conflicting needs of each partner. Our results lend support for the view that, in the context of lung cancer, the initial diagnosis and treatment period is a critical time which may set the tone for future marital interaction as well as the course of psychosocial adjustment. From a clinical perspective, helping lung cancer couples become more aware of the need to focus on and maintain their relationships during the initial treatment period may help to mold a more resilient relationship as the cancer progresses and the patient moves toward the end of life.