Prevalence and Factors Associated with Marital Distress among Hematopoietic Cell Transplant Survivors: Results from a Large Cross-sectional Study

Abstract

Background and objective

Medical and psychological sequelae of hematopoietic cell transplantation (HCT) are well-established. Less is known regarding impacts on intimate relationships. We sought to describe the prevalence of relationship distress among married HCT survivors, and to identify factors associated with relationship distress.

Study design

The study was cross-sectional. HCT survivors treated at an NCI-designated cancer center completed a survey that included measures of psychological and relationship distress (Revised Dyadic Adjustment Scale, RDAS), demographic, and medical factors.

Results

1047 respondents reported being married and completed the RDAS. Sample characteristics were mean age 58 (range 22–83), mean years post-HCT 12.3 (range <1–43), 42% female, and 89% non-Hispanic white. RDAS total adjustment scores were comparable to published community sample values. 19% of respondents were classified as relationship-distressed. Three factors were associated with greater odds of relationship distress: cancer and treatment distress (OR = 1.59, CI = 1.09, 2.31); depression (OR = 1.44, CI = 1.17, 1.76), and time post-transplant (OR = 1.03, CI = 1.01, 1.04). These findings were qualified by gender-stratified analyses. Depression was associated with relationship distress among females; the other two variables were associated with relationship distress among males.

Conclusions

Results paint an overall picture of positive long-term marital functioning among HCT survivors, with a subset reporting relationship distress. Screening is warranted to identify survivors at risk for relationship distress and to inform targeted intervention.

INTRODUCTION

As a primary social relationship, marriage is generally thought to be health-protective [1]. For example, married versus unmarried patients with cancer are less likely to present with metastatic disease and to die of the disease [2]. But being married per se is not necessarily advantageous. The health benefits of marriage are stronger for males than females [2, 1]. Relationship quality also matters [1, 3]. A meta-analysis of 126 articles and over 72,000 individuals found that better marital quality was associated with better physical health, including lower risk of mortality [3]. Mean effect sizes were small but likened to the strength of associations between health behaviors such as dietary intake and health outcomes [3].

Marital adjustment is defined as “a process by which married couples attain mutual gratification and achieve common goals while maintaining an appropriate degree of individuality” [4]. The term encompasses not only marital satisfaction, but also cohesion (companionship) and consensus (shared values) [5]. We focus here on marital adjustment and facets thereof among survivors of hematopoietic cell transplantation (HCT). While medical and psychological sequelae of this rigorous cancer treatment are well-established [6–11], less is known regarding impacts on intimate relationships. The HCT process takes a major toll on families [12–14]. Caregiving demands, often shouldered by a spouse if patients are married, are tremendous and do not end once patients are discharged from the transplant center [12, 15, 16]. Late sequelae require extended monitoring and treatment. For example, graft-versus-host disease (GVHD), a multi-system disorder resulting from donated marrow or peripheral blood stem cells attacking the host’s body, necessitates long-term use of corticosteroids and other immunosuppressive treatments [17]. Patients may delay or be unable to return to work, causing financial burdens [7]. Roles often shift and can be difficult to resume even as health recovers [18]. Family dynamics may be disrupted [19]. It is not surprising, then, that intimate relationships could be adversely affected, but very little empirical work has been conducted to examine the prevalence of, and factors associated with, relationship distress in this population.

Interestingly, research from the broader (solid tumor) cancer literature paints a generally positive picture of relationship quality following treatment. Among studies employing the Dyadic Adjustment Scale (DAS) [20], a widely used measure of relationship adjustment and satisfaction, cancer survivors’ mean values fall at or above norms provided by the scale developer [21–25]. Just a small handful of studies has examined relationship adjustment and/or satisfaction in HCT samples. In a study of 10-year HCT survivors, there was no difference in relationship satisfaction between the survivors and case-matched controls [26]. Bishop and colleagues assessed relationship quality among 177 HCT survivors as well as their partners and 133 married acquaintances without a history of cancer or HCT [27]. Survivors did not differ from acquaintances with regard to relationship adjustment or satisfaction; partners, however, reported lower levels of relationship satisfaction relative to both survivors and the acquaintances [27]. In a prospective, longitudinal study of 121 HCT survivors, relationship satisfaction was stable over time for male and female patients and male spouses but not female spouses who reported reductions in satisfaction relative to pre-transplant baseline, up to 5 years post-transplant [28]. Rates of relationship distress based on DAS total adjustment scores paralleled this pattern [28]. While informative, these studies tell us little about relationship adjustment among long-term survivors. Due to attrition, only 52 patients completed the DAS at the 5-year timepoint in the aforementioned longitudinal study [28]. Patients in the study conducted by Bishop and colleagues were on average 6 years post-transplant (M±SD = 6.7±3.1, median = 6.1). Neither of these investigations included analyses designed to identify factors associated with relationship distress, aside from gender [28]. Doing so may point to readily identifiable risk factors of clinical utility when monitoring psychosocial functioning.

The present study is unique in that it is large (over 1,000 survivors) and includes persons transplanted at different lifespan timepoints (from childhood to older adulthood) and at different stages of survivorship (from within one year to over 40 years post-transplantation). The purpose of the study is three-fold: (1) to describe the prevalence and characteristics of relationship distress in this sample; (2) to compare HCT survivors to published general population values [29] of relationship adjustment; and (3) to identify factors associated with relationship distress. Regarding the latter, factors of interest were demographic characteristics such as gender, age, and education; medical factors thought to impede or challenge recovery such as treatment intensity, time since transplantation, relapse, and severity of chronic GVHD; and psychological distress (both general and cancer-specific). Selection of these factors is supported in part by a study of survivors of childhood cancer where older age at diagnosis, greater treatment intensity, and higher anxiety were associated with lower relationship satisfaction [30]. Rationale for examination of the role of gender is supported by research from the broader (non-medical) literature on marital satisfaction, suggesting lower levels of satisfaction among women versus men [31, 32].

MATERIALS AND METHODS

Design and participants

The study design was cross-sectional. HCT survivors transplanted at an NCI-designated comprehensive cancer care center with a large transplant program were mailed an annual survey. The survey contained a core set of items and a domain-specific module. The version administered between July 1, 2014 and June 30, 2015 included a module focused on sexual functioning and relationship quality (see Syrjala et al., 2020 [33] for a presentation of the sexual functioning findings). All procedures were approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center in Seattle, WA. In addition to the patient-reported outcomes from the survey, medical records were extracted to gather key clinical variables: diagnosis for which respondents were transplanted, type of transplant, number of transplants, preparative regimen intensity (myeloablative versus non-myeloablative), and cytogenetic or morphologic relapse (following the most recent transplant if more than one). Patient-reported measures are described below.

Relationship quality

The Revised Dyadic Adjustment Scale (RDAS) was used to measure relationship quality [34]. This 14-item scale, an abbreviated version of the 32-item Dyadic Adjustment Scale [20], assesses relationship satisfaction, cohesion, and consensus in the context of a marriage or cohabiting partnership. Response formats vary across items. For consensus, 6 items such as “career decisions,” “demonstrations of affection,” and “conventionality (correct or proper behavior” are rated on a 6-point agreement scale, indicating the extent to which the respondent and their partner agree on the given matter [items 3, 4, 6, 7, 12 and 15 from the full DAS]. For satisfaction, 4 items such as, “How often do you and your partner quarrel?” and “Do you ever regret that you are married?” (reverse-scored) are rated on a 6-point frequency scale [items 16, 20, 21 and 22 from the full DAS]. For cohesion, 3 items such as “have a stimulating exchange of ideas” and “calmly discuss something” are rated on a 6-point frequency scale [items 25, 27 and 28 from the full DAS]. A fourth cohesion item [item 24 from the full DAS] assesses the frequency with which respondents and their partner “engage in outside interests together.” Items are summed to create subscale scores and a total adjustment score. Consensus subscale scores can range from 0–30, satisfaction subscale scores from 0–20, cohesion subscale scores from 0–19, and total adjustment scores from 0–69. Higher values indicate greater relationship quality. Validity of the RDAS has been demonstrated by adequate positive association with the Marital Adjustment Test and ability to discriminate between clinical and non-clinical samples [34]. Internal consistency values based on the present sample ranged from 0.80 for satisfaction and cohesion to 0.87 for total adjustment. Per Crane et al. [35], continuous scores were dichotomized to determine prevalence using the following cut-offs, with values below the cutoff coded as relationship-distressed: 22 for consensus, 14 for satisfaction, 11 for cohesion, and 48 for total adjustment.

General distress

The Patient Health Questionnaire is a widely used, valid and reliable measure of depressive and anxious symptomatology. The PHQ-4 provides a factor analytically-derived ultra-brief version [36]. Respondents are asked to rate the extent to which they have been bothered by each of 4 problems on a 0–3 scale (not at all to nearly every day). Two items are summed to assess anxiety (feeling nervous, anxious, or on edge; not being able to stop or control worrying) and two are summed to assess depression (feeling down, depressed, or hopeless; little interest or pleasure in doing things). Internal consistency values based on the present sample were 0.85 for both subscales.

Cancer and treatment distress

The Cancer Treatment and Distress scale (CTXD) is a valid and reliable measure of distress or worry related to cancer treatment, developed with HCT recipients [37]. Items such as “long term effects of treatment,” “being a burden to other people,” and “changes in my appearance” are rated with respect to how much distress or worry they caused in the past week (none to severe, coded 0–3). Six subscales assess uncertainty, health burden, family strain, identity, finances, and medical demands. We used the standard scoring of a total mean across all items. Internal consistency (Cronbach’s alpha) in the present sample was 0.95. Validity of the CTXD in HCT patients has been demonstrated by substantial positive correlations with other widely-used measures of generic depression and distress [37].

Chronic graft-versus-host disease

Survivors of allogeneic transplantation were asked, “Do you currently have chronic GVHD?” Those responding affirmatively were asked to characterize the disease “now” as mild, moderate, or severe. Responses were dichotomized across these two items as none or mild versus moderate or severe.

Functional status

Following Lee et al. [38], respondents were asked to select from 9 phrases the one that best described their level of activity during the past two weeks, mapping on to standard Karnofsky Performance Status Scale definitions [39]. Options ranged from “I am in the hospital most of the time” (scored 10%) to “I can do my normal activity, and I have no complaints about my health” (scored 100%). Based on the distribution of this variable, we dichotomized scores as 90–100% versus 80% or lower.

Statistical analyses

Analyses were conducted using Statistical Package for the Social Sciences 28.0. Descriptive statistics afforded characterization of the sample with respect to demographic, clinical, and psychosocial characteristics. Single sample t-tests were conducted to compare our HCT RDAS total adjustment score values to norms reported by Anderson and colleagues [29]. Specifically, Anderson et al. presented RDAS total adjustment score estimates based on two sets of samples: 16 community samples (N = 4967) and 6 clinical samples (couples receiving therapy for relationship distress, N = 1228). These values were 52.95 and 40.58, respectively [29].

Binary logistic regression was used to identify factors associated with relationship distress (dyadic total adjustment scores < 48). We analyzed the following factors in separate univariate models: age at first transplant, age at survey, time post-transplant, gender (male, female), race/ethnicity (non-Hispanic white, other), educational status (< 4-year degree, 4-year degree or higher), work status (employed full-time or part-time, other), type of transplant (autologous, allogeneic related, allogeneic unrelated), number of transplants (one, two or three), treatment intensity (myeloablative, non-myeloablative), functional status (80% or less, 90–100%), patient-reported chronic graft-versus-host disease severity (none or mild, moderate or severe), disease relapse (yes, no), PHQ depression, PHQ anxiety, and cancer and treatment distress. Variables associated with relationship distress in univariate models were entered into a multivariate model. Multicollinearity was not observed as evidenced by tolerance values > .1 and variance inflation factors < 3.

RESULTS

The annual long-term follow-up survey was sent to 4214 HCT survivors. Among the 4214, 1706 completed a question about marital status embedded in the module on sexual functioning and relationship quality. Of note, those who completed the marital status item were younger and closer in time to transplantation as compared to those who did not, M (SD) = 57.8 (12.5) versus 60.7 (13.2) for age, p = .002; and 12.2 (9.5) versus 13.1 (9.9) for years post-transplant, p < .001. Among the 1706 who responded to the question about marital status, 1269 reported being married, 87 living with a partner, and 350 not partnered (145 single, 130 divorced, 22 separated, and 53 widowed). Please see Supplemental Table 1 for descriptive characteristics of the married, living with a partner, and non-partnered subgroups with regard to age, sex, pre-transplant clinical characteristics and, for those completing the RDAS, dyadic adjustment. In brief, the married group was older than the other two subgroups at the time of both transplantation and survey completion. The subgroups also differed by gender. The married group was comprised of more males than females (57% versus 43%) while the non-partnered group was comprised of more females than males (63% versus 37%). The living with partner group, in contrast, was evenly distributed with regard to gender. Importantly, the married and living with partner groups did not differ on the dependent variable of interest, dyadic adjustment.

Our analysis sample is limited to the subgroup of respondents who both reported being married and completed the full RDAS (N = 1047). Table 1 presents demographic, clinical, and psychosocial characteristics of this sample. Age at first transplant varied considerably, from 1 year to 75 years (mean = 45), as did age at the time of survey completion, from 22 to 83 years (mean = 58). Fifty-eight percent of respondents were male, 42% female. The majority (89%) was non-Hispanic and white. Sixty-one percent had earned a college degree or higher, and 55% were either in school or employed full- or part-time. Most respondents had received one transplant (89%). Type of transplant was fairly evenly distributed: 31% autologous, 39% allogeneic with the donor related to the survivor, and 30% allogeneic with an unrelated donor. The majority of transplants, moreover, were myeloablative (92%) meaning that the procedure was preceded by high intensity conditioning (chemotherapy and/or total body irradiation). Functional status at the time of survey completion was high for the majority (77%).

Table 1.

Demographic and clinical characteristics of the sample

N	1047
Age at first transplant in years, M (SD); range	45.92 (14.65); 1–75
Age at first transplant in years, n (%)
<18	42 (4.0)
18–39	311 (29.7)
40–59	507 (48.4)
60+	187 (17.9)
Age at time of survey in years, M (SD); range	58.24 (10.98); 22–83
Time post-transplant in years, M (SD); range	12.29 (9.53); 0.45–43
Time post-transplant, n (%)
<10 years	498 (47.6)
10–19 years	300 (28.7)
20+ years	249 (23.8)
Gender, n (%)
Male	608 (58.1)
Female	439 (41.9)
Race/ethnicity, n (%)
Non-Hispanic and white	933 (89.1)
Hispanic and/or non-white	105 (10.0)
Unknown	9 (0.9)
Educational status, n (%)
< 4-year degree	405 (38.7)
≥ 4-year degree	639 (61.0)
Unknown	3 (0.3)
Work status, n (%)
Employed and/or school full- or part-time	571 (54.5)
Homemaker/ retired/ disabled/ unemployed	471 (45.0)
Unknown	5 (0.5)
Type of transplant, n (%)
Autologous	328 (31.3)
Allogeneic related	408 (39.0)
Allogeneic unrelated	311 (29.7)
Conditioning regimen, n (%)
Myeloablative	964 (92.1)
Non-myeloablative	83 (7.9)
Number of transplants, n (%)
One	936 (89.4)
Two or three	111 (10.6)
Diagnosis, n (%)
Acute leukemia	244 (23.3)
Chronic leukemia	217 (20.7)
Lymphoma	235 (22.4)
Myeloma	158 (15.1)
Other malignant	140 (13.4)
Non-malignant	53 (5.1)
Current chronic GVHD, n (%) moderate or severe	69 (6.6)
Cytogenetic or morphologic relapse, n (%)	99 (9.5%)
Karnofsky performance status, n (% )
90–100%	803 (76.7)
80% or lower	237 (22.6)
Unknown	7 (0.7)
Patient Health Questionnaire depression, M (SD)	0.53 (1.03)
Patient Health Questionnaire depression, n (%) elevated	48 (4.6)
Patient Health Questionnaire anxiety, M (SD)	0.75 (1.27)
Patient Health Questionnaire anxiety, n (%) elevated	82 (7.8)
Cancer and Treatment Distress, M (SD)	0.58 (0.54)
Cancer and Treatment Distress, n (%) elevated	256 (24.5)
RDAS total relationship adjustment, M (SD)	53.38 (7.74)
RDAS consensus, M (SD)	24.64 (4.04)
RDAS satisfaction, M (SD)	16.20 (2.22)
RDAS cohesion, M (SD)	12.54 (3.37)
RDAS total relationship adjustment, n (%) relationship-distressed	199 (19.0)
RDAS consensus, n (%) relationship-distressed	180 (17.2)
RDAS satisfaction, n (%) relationship-distressed	114 (10.9)
RDAS cohesion, n (%) relationship-distressed	264 (25.2)

Turning to RDAS values, total adjustment scores ranged from 14 to 69, with a mean of 53. Nineteen percent of respondents scored in the relationship-distressed range based on these scores (below the cut-off of 48). Using subscale cut-off scores, relationship distress ranged from 11% for satisfaction to 17% for consensus and 25% for cohesion (bottom of Table 1).

In single sample t-tests comparing our RDAS total adjustment scores to pooled sample means reported by Anderson et al. [29], values for the HCT sample (M = 53.38, SD = 7.74) did not differ significantly from the pooled mean of 52.95 based on the community samples, t(1046) = 1.81, p = .071. In contrast, values for the HCT sample were higher (meaning better adjusted) on average as compared to the pooled mean of 40.58 for the couples therapy samples, t(1046) = 53.52, p < .001.

Table 2 displays results of univariate logistic regression analyses for factors associated with relationship distress (RDAS total adjustment under the cut-off value of 48). Four variables were significantly associated with greater odds of relationship distress in univariate models: time post-transplant in years, PHQ anxiety, PHQ depression, and cancer and treatment distress (CTXD). These four variables were entered into a multivariate model (lower portion of Table 2). Three emerged as significant in the final model: time post-transplant, PHQ depression, and cancer and treatment distress. The odds of relationship distress were greater for those with greater depression, those with greater cancer and treatment distress, and those further in time since transplantation.

Table 2.

Logistic regression analyses of relationship distress

UNIVARIATE MODELS	OR (95% CI)	p
Age at first transplant (continuous)	1.00 (0.99, 1.01)	.457
Age at time of survey (continuous)	1.01 (0.99, 1.02)	.330
Time post-transplant in years (continuous)	1.02 (1.002, 1.04)	.024
Gender		.235
Male	1.00
Female	0.83 (.60, 1.13)
Ethnicity and race		.249
Hispanic and/or non-white	1.00
Non-Hispanic and white	0.75 (0.46, 1.22)
Educational status		.897
Less than four-year degree	1.00
Four-year degree or higher	0.98 (0.71, 1.34)
Employment status		.692
Homemaker/ retired/ disabled/ unemployed	1.00
Employed or in school full- or part-time	0.94 (0.69, 1.28)
Type of transplant		.483
Autologous	1.00
Allogeneic related	0.98 (0.65, 1.47)	.915
Allogeneic unrelated	1.20 (0.82, 1.74)	.345
Transplant number		.592
One	1.00
Two or three	0.87 (0.52, 1.46)
Conditioning regimen		.721
Myeloablative	1.00
Non-myeloablative	1.11 (0.63, 1.93)
Karnofsky performance status		.181
80% or lower	1.00
90–100%	0.78 (0.55, 1.12)
Current cGVHD		.550
None or mild	1.00
Moderate or severe	1.20 (0.66, 2.17)
Relapse		.769
No	1.00
Yes	1.08 (0.64, 1.81)
Patient Health Questionnaire anxiety (continuous)	1.19 (1.06, 1.33)	.002
Patient Health Questionnaire depression (continuous)	1.46 (1.28, 1.67)	<.001
Cancer and treatment distress (continuous)	1.83 (1.40, 2.39)	<.001
MULTIVARIATE MODELS	OR (95% CI)	p
Years post-transplant	1.03 (1.01, 1.04)	.004
Patient Health Questionnaire anxiety	0.88 (0.74, 1.04)	.136
Patient Health Questionnaire depression	1.44 (1.17, 1.76)	<.001
Cancer and treatment distress	1.59 (1.09, 2.31)	.015

Post-hoc analyses

While gender was not associated with relationship distress in the aforementioned models (Table 2), per suggestion of an anonymous reviewer and given well-established gender differences in associations between relationship quality and health [1], we re-ran the logistic regression analyses stratifying by gender, i.e., separately for males and females. Results are presented in Supplementary Table 2. For males, four variables were associated with greater odds of relationship distress in univariate models: time post-transplant (p = .04), PHQ anxiety (p = .008), PHQ depression (p < .001), and cancer-related distress (p < .001). Two of these emerged as predictive in the final multivariate model, time post-transplant and cancer-related distress. For males, odds of relationship distress were greater with longer time since transplant (OR = 1.03, p = .008) and greater cancer-related distress (OR = 2.03, p = .005). For females, three variables were associated with greater odds of relationship distress in univariate models: PHQ anxiety (p = .08), PHQ depression (p < .001), and cancer-related distress (p = .029). Just one of these emerged as predictive in the final multivariate model, depression. For females, odds of relationship distress were greater with greater depression (OR = 1.87, p < .001).

DISCUSSION

To our knowledge, this is the largest sample to date in which relationship adjustment following HCT has been investigated and that provides results on a large proportion of married survivors more than 10 years post-transplant (50%). Generally speaking, results paint a picture of positive marital functioning in these survivors. Mean total dyadic adjustment scores for our sample were on par with published values based on pooled community samples and superior to values based on pooled couple therapy samples. While evidence is convincing that HCT can alter family roles [40], burden caregiving partners [12], disrupt employment [7], and strain finances [41], marital functioning from the survivors’ perspectives does not seem to be affected, on average, in the long run.

Despite this overall picture of relational well-being, a subset of respondents was classified as relationship-distressed: 19% based on total adjustment scores, with the largest deficits seen for cohesion scores (25%). The latter finding suggests that interventions designed to enhance marital functioning may benefit from couple-based approaches that address and discuss strategies to boost cohesion, for example, having couples find or newly explore intersecting points of interest and ways to pursue them together.

Results from logistic regression analyses combining males and females point to three factors associated with relationship distress following HCT that could be used to help identify those who may benefit from couple intervention or for whom marriages may be improved by addressing relationship needs: longer-term survivorship and those with elevated depression or cancer and treatment distress. Because the study is cross-sectional, we do not know what might drive the association between years post-transplant and relationship distress. It could be that complications of the disease and treatment simply take their toll over time and that this adversely affects survivors’ intimate relationships. However, this sample was relatively high in functional status. The association could also be driven by length of the relationship, but this cannot be ascertained because we did not assess marriage duration or marital disruptions. Similarly, we do not know if the relationship being rated in the survey pre-dated the transplant, and it has been shown that relationships formed after completion of treatment are more vulnerable [42].

Other factors associated with relationship distress were psychological in nature. The odds of relationship distress were greater with greater depressive symptoms as measured by the PHQ. This fits with findings from the general marital literature citing robust associations between depression and relationship distress [1, 43]. Results of analyses stratified by gender, however, qualify this finding. The association between depression and relationship distress was statistically significant for female but not male survivors. This makes sense in part based on findings from the broader psychological literature on gender differences in self-construal. For women, self-construals are typically more inter-dependent and communal whereas those for men are more self-contained [44]. Thus women may be more adversely affected by relational problems. Other work suggests that females’ perception of the quality of their marriage may depend more on the perceived quality of interactions with their spouse, and that females approach such interactions with greater sensitivity to the behavior of their partner [45]. Indeed, observational studies of marital interactions indicate that sustainment of emotional expression (measured by fundamental frequency, a vocal indicator of emotional arousal) over the course of a conversation is important for females, affording processing of the problem at hand. Interruption of this sustainment by a male partner is maladaptive, relating to greater psychological symptoms [46]. Future research assessing marital communication in the context of cancer is needed to gain a more nuanced understanding of mechanisms by which communication dynamics might affect important psychological and even clinical outcomes for both male and female survivors [47].

Interestingly, general anxiety as measured by the PHQ was not associated with relationship distress in a multivariate model but treatment-related distress was. This suggests that worry about transplant-related issues, even years following the event, may adversely impact relationship quality. The Cancer and Treatment Distress measure instructs respondents to rate how much distress or worry a given item such as medical problems caused them in the past week, even if the event has not happened. We know in fact that treatment-related distress can linger for quite some time [10]. Our result linking the construct to relationship distress extends our understanding of the impacts of HCT to the interpersonal realm. Here too, findings are qualified by gender-stratified analyses. Cancer-treatment distress was positively associated with relationship distress among males but not females. This suggests that for the male survivor, relationship quality may be more closely tied not to their general psychological state but instead to cancer- and treatment-specific concerns. Indeed, results from a large national cross-sectional study of cancer survivors identifying psychosocial and demographic predictors of divorce or separation after diagnosis found that for males, only fear of cancer recurrence was associated with greater odds of divorce or separation [48]. Interestingly, in line with our results linking depression to relationship distress for women, greater odds of divorce or separation were predicted by greater emotional distress for females but not males [48]. These disparate patterns bear consideration for future research investigations and may indicate that screening for relationship distress should be targeted, with greater watch and attention to cancer-specific distress for males and to elevated depressive symptoms for females.

Strengths of our study include the large sample, the diversity in types of transplants and diagnoses, inclusion of a large proportion of male respondents, and particularly the wide range of time since transplant, with a majority of survivors who were more than 10 years post-HCT and who are not represented in most studies of marital relationships in survivors. A further strength is the use of standardized measures. Nonetheless, limitations bear consideration. The sample was comprised mostly of persons identifying as non-Hispanic white (89%). This is commensurate with disparities in access to HCT [49] but limits generalizability. The heterogeneity of the sample with regard to medical characteristics, while a strength in one regard, is also a limitation. Autologous and allogeneic transplants, for example, pose different medical challenges such as chronic GVHD; distress is also known to persist longer for allogeneic transplant recipients [26]. Nonethless, transplant type was not associated with relationship distress, nor were other medical or functional variables of interest.

The cross-sectional design of the study restricts our ability to determine whether variables of interest influence relationship distress or vice versa, or if unidentified variables play a mechanistic role in the association. Importantly, the study was not dyadic in that we only assessed survivors, not partners. We do not know how partners perceive the relationship and therefore whether or not their perceptions are concordant with the survivors. Our previous work (based on a longitudinal cohort) indicated concordance in relationship satisfaction among patients and caregivers prior to transplant but discordance in particular within male survivor-female caregiver pairs over time, with female caregivers reporting reductions in satisfaction up to 5 years post-transplant [28]. Because we did not assess partners or gather patient-reported information on partners, we do not know the sex composition of the relationship our respondents were in at the time of the survey (same-sex or opposite-sex). Individuals who identify as other than heterosexual have been historically underrepresented in biomedical research [50]. Also, because we excluded partnered but not married survivors, we are unable to comment on other types of less officialized relationships, though we do know that the percentage of relationship-distressed persons is similar in both groups (Supplemental Table 1).

Future investigations are needed that address these limitations and examine both the intra- and inter-personal correlates of HCT on both survivors and partners. Longitudinal studies are essential to examine the trajectory of relationship quality and stability among long-term survivors. Couple-based interventions may be warranted for a subset of survivor dyads with relationship distress. How couples communicate plays a major role in their adaptation to the cancer experience [51]. Holding back from disclosure and communication patterns characterized by demand on the part of one dyad member followed by withdrawal on the part of the other have been associated with lower relationship satisfaction [52–54]. Constructive communication, in contrast, has been associated with greater relationship satisfaction [53], as has talking about the relationship [22]. Couple-based interventions in HCT are rare and most of this work has been pilot in nature, to determine feasibility and acceptability [14]. Considering the associations of depression and distress with relationship distress seen in our cohort, a couple-based intervention to assist couples coping with psychosocial issues after HCT could be beneficial. A pilot study with advanced gastrointestinal cancer survivors and their partners found preliminary efficacy of a communication intervention on relationship satisfaction [55]. A similar intervention may be of value in the HCT setting as it was conducted over video conference and may address travel barriers since three-quarters of survivors live at least 30 miles from their transplant center [56].

Beyond screening after transplant to determine referral needs, we suggest that information regarding marital quality be imparted to patients and partners prior to transplantation. Based on our findings, providers could convey a general message of positivity, that relationships overall seem similar to couples in the general public that have not faced transplant together, tempered with education on communication styles that facilitate positive marital adjustment.

Supplementary Material

1

Supplemental Table 1. Demographic, pre-transplant clinical, and relationship characteristics as a function of marital status

2

Supplemental Table 2. Logistic regression analyses of relationship distress stratified x gender

HIGHLIGHTS.

• On average, survivors’ ratings of the quality of their marriage were comparable to published community sample values and superior to values from clinical samples (those in couple therapy).

• 19% of survivors were classified as relationship-distressed.

• Odds of relationship distress were associated with greater depression for females and greater cancer-related distress for males.