Background

Cancer and the family

The National Cancer Institute assumes that 24 % of adult cancer patients are parents of children under legal age [1]. A number of findings suggest that a parental cancer diagnosis and treatment affect the whole family unit [2]. Cancer presents a major stressor and often creates multiple problems for families with significant mental and somatic implications for family members [2–4]. When a parent gets a cancer diagnosis, the burdens of the disease and cancer treatments are added to the normal burdens of family life. Parents with cancer are concerned about their children’s well-being and seem to be torn between their own and their children’s needs, and usual parental roles could change [5]. Cancer patients with children under 18 years (compared to those without) had a five times higher risk of developing a stress disorder and a higher risk for the development of anxiety disorders in the course of the disease [6].

Regarding the impact of parental cancer on children, study results are inconsistent [4, 7, 8]. There is preliminary evidence that cancer in a parent may be a significant stressor for school-aged children and adolescents [4, 9]. Further, previous findings indicated that there are significant associations between parents’ psychosocial distress, family functioning, family communication, and child functioning [4, 10–13].

There is a considerable body of evidence showing that the levels of psychological distress among partners of cancer patients are equal to or sometimes even greater than those of the patients [14]. Longitudinal studies have shown increased psychological distress (particularly for anxiety) in both female and male partners [15]. Partners of cancer patients also have a significantly higher psychological distress and lower quality of life compared to the general population—particularly in patients with advanced disease [14, 16].

Dyadic relationships

Cancer in a parent can cause changes within the couple relationship [17]. Several studies documented that partnership satisfaction had a positive effect on the psychological distress of patients and partners as well as the functioning of the children [8, 18–21]. Couples commonly reported improvements in marital relationship and growing closer after cancer [17, 22]. Patients and partners who reported negative dyadic changes showed lower quality of life and higher levels of anxiety and depression [17]. It is therefore advisable not to consider the situation of the patient and the partner independently, but to analyze the dyadic relationships between the two, e.g., with the structural equation model described by Kenny and colleagues [23]. They underline that “families have been understood as complex organizations in which socially and biologically prescribed role requirements and processes of positive and negative feedback have dramatic effects on family members’ behavior.”

Social network and employment

A cancer disease in middle adulthood has consequences not only for the parenthood and family life but also for the social network and working life [1, 8]. The importance of social support and a stable social network is widely validated [24]. Several studies have reported a positive association between helpful social support from family and friends and cancer patients’ psychological distress and quality of life [25, 26]. Furthermore, the quality of the support network of the parents was a significant moderator of child functioning [4].

For parents aged 30 to 50, employment and work plays an important role. Previous research findings have shown the positive impact of employment on the patients’ and partners’ psychological well-being [27, 28]. Cancer and cancer treatments can lead to limitations in work ability, employment, and satisfaction with work among cancer patients and their partners [29].

The impact of parenthood on the psychosocial situation of cancer patients and partners has received little research attention. To our knowledge, there have been no prospective studies addressing this topic so far and previous findings on parental cancer have shown different results [30]. In the review of the literature on the experience of parents with cancer who are caring for children, Semple and McCance found a predominance of studies using solely qualitative methodology and the majority of studies focused on the experience of mothers [1]. Furthermore, previous studies mostly assessed the situation of the patient and their partner independently; the dyadic relationships were not recognized.

Therefore, the first aim of our study was to examine psychological distress (anxiety and depression) of cancer patients with children under 18 years and their partners in a longitudinal design. The second aim was to analyze dyadic relationships with a structural equation model to estimate the actor-partner interdependence model.

Our work centered on the following questions:

  1. 1.

    What levels of anxiety and depression do cancer patients with children under 18 years and their partners experience over a 1-year period post therapy?

  2. 2.

    What is the perceived social support and partnership satisfaction of cancer patients with children under 18 years and their partners over a 1-year period post therapy?

  3. 3.

    What is the dyadic relationship between patients’ and partners’ psychological distress and the predictor variables social support, partnership satisfaction, and employment?

Methods

Study design

Our study was part of the multi-site research project “Psychosocial Services for Children of Parents with Cancer”, conducted in five German cities from 2009 to 2012 and supported by the German Cancer Aid (Deutsche Krebshilfe, grant no. 108303). We evaluated a longitudinal study of cancer patients and their partners; analyses were based on data from three assessment points (t1 = post therapy, t2 = 6 months after t1, and t3 = 12 months after t1).

Participants

At the University Hospital of Leipzig and collaborating hospitals in the Leipzig region, cancer patients with children younger than 18 years and no palliative treatment situation were recruited. Of the 218 patients approached, 26.1 % (n = 57) rejected participation, mainly based on rational reasons (e.g., no interest). One hundred sixty-one patients (73.9 %) met the inclusion criteria (77 % females). At t2 and t3, 124 and 121 patients could be surveyed, respectively. In the longitudinal analysis, 106 patients with complete data were included.

At t1, 115 partners of the surveyed patients agreed to participate in the study. At t2 and t3, 80 and 81, respectively, filled out the questionnaire. In the structural equation model to estimate the actor-partner interdependence model, we included 81 patient-partner dyads. The recruitment process and non-responder analyses have been described in more detail elsewhere [28]. Those rejecting participation did not differ significantly from participants in gender or age (p > 0.05).

Comparison group

Data obtained on the study sample’s anxiety and depression (Hospital Anxiety and Depression Scale, HADS) were compared to German normative data from a representative survey (N = 1337)—matched by age (40.1 years), partnership (71 %), and parenthood (at least one child under 18 years lives in the household, 75 %) [31].

Instruments

To assess psychological distress, we used the HADS which measures levels of anxiety and depression in physically ill adults [32]. For each scale, a cumulative score (0 to 21) can be created, reflecting the severity of symptomatology: 0–7 (low), 8–10 (moderate), and ≥11 (high). The total HAD Scale, used as a measure for psychological distress, has a theoretical score range between 0 and 42 with a cutoff ≥15 indicating high psychological distress. The authors report Cronbach’s alpha of 0.80 (anxiety)/0.81 (depression) in their sample.

To measure social support, we used the Oslo 3-item Social Support Scale (OSS-3) [33]. The three items cover different fields of social support, and the OSS sum score ranges from 3 to 14 and is calculated by summarizing the raw scores of the items. Scores 3–8, 9–11, and 12–14 indicate “poor,” “moderate,” and “strong” social support. Its structure and reliability (Cronbach’s alpha = 0.60) have not been well-documented despite their widespread use in several European countries.

Perceived quality of partnership was determined with the short version of the Abbreviated Dyadic Adjustment Scale (ADAS, German translation of Köppe et al.’s FBZ-K) [34, 35]. The ADAS contains seven items. Three areas of partnership and the couple’s concordance on them are assessed (0 = always disagree to 5 = always agree). Three other items assess the occurrence of positive behaviors (0 = never to 5 = more than once a day). One item assesses how happy the respondent is with their relationship (0 = very unhappy to 5 = very happy). Answers are added up (reliability α = 0.82), and a sum score of less than 17.1 indicates a low satisfaction with the partnership; values above 22.5 indicate medium to high levels of satisfaction. The short version of the used scale has a good reliability (Cronbach’s α = 0.82) [35].

Data analyses

All statistical analyses were performed using SPSS 20. For bivariate comparisons, we used Pearson’s chi-square tests (Fisher’s exact test, respectively) or Student’s t tests. We analyzed the dyadic relationship of three variables (social support: OSS sum score; partnership satisfaction: ADAS sum score; and employment: 0 = no, 1 = yes) with the psychological distress of the patients and partners (HADS total score).

We used a structural equation model to estimate the actor-partner interdependence model (distinguishable dyads) [23]. The model that contains predictor variables that are between dyads, within dyads, and mixed is called the actor-partner interdependence model or APIM. APIM analyses combine two regression models which are interrelated through a covariance between endogenous variables and the residuals. The interdependence between dyadic data is explicitly considered within this approach. It should be noted that “partner” has two different meanings here: generally, the partner is the spouse of the patients, but the partner effect (as opposite to the actor effect) in the model means the impact of the other, the non-actor, which can also be the patient. The computed models were saturated (df = 0), and no measures of fit could be obtained. All metric variables were z-standardized using the overall means and standard deviations (e.g., 1 mean and 1 standard deviation for patients and partners together), and binary variables were used as they were (0/1-dummy-coded). We report the unstandardized coefficients of the standardized variables, which can be interpreted as the standardized coefficients. Therefore, it is possible to compare the coefficients of different scales without losing metric equivalence (see [36], p. 361, footnote 2).

Sample

Sample characteristics are shown in Table 1. Eight patients and partners were interviewed who did not live in the same household as their children, as they mainly lived with the other parent due to a divorce.

Table 1 Sample characteristics of patients (N = 161) and partners (N = 115)
Full size table

Results

Psychological distress over a 1-year period post therapy

Levels of anxiety and depression of the patients and partners over a 1-year period post therapy are shown in Table 2. We found no significant changes in anxiety and depression of the patient (anxiety: p = 0.955 (t1 and t2), p = 0.444 (t1–t3); depression: p = 0.762 (t1 and t2), p = 0.636 (t1–t3)). Levels of anxiety and depression of the partners had decreased significantly from t1 to t3 (anxiety: p = 0.057 (t1 and t2), p = 0.002 (t1–t3); depression: p = 0.427 (t1 and t2), p = 0.018 (t1–t3)).

Table 2 Anxiety and depression of patients (N = 106) and partners (N = 65) over a 1-year period post therapy, HADS cutoff (0–7 = low, 8–10 = moderate, ≥11 = high), and HADS mean scores
Full size table

In bivariate analyses, none of the examined sociodemographic variables (gender, age, and age of the children) revealed a significant relationship to symptoms of anxiety and depression (p > 0.05; exception: female partners were more anxious at t1 (p = 0.033), and older partners were more depressed at t3 (p = 0.009)).

As illustrated in Fig. 1, at all points of assessment, the level of anxiety in patients and partners was significantly higher than that in the representative survey of the general population (patients at t1–t3: p < 0.001, Cohen’s d = 0.485–0.577; partner at t1–t3: p < 0.01, Cohen’s d = 0.434–0.836). Regarding levels of depression, there were no differences between patients, partners, and the comparison group (patients at t1–t3: p > 0.05, Cohen’s d = 0.044–0.127; partner at t1–t3: p > 0.05, Cohen’s d = 0.042–0.146).

Fig. 1
figure 1

Clinically relevant levels of anxiety and depression of the patients (N = 106) and partners (N = 65) over a 1-year period post therapy in comparison to the general population (N = 1337; matched by age, partnership, and children), HADS cutoff (≥11 = high)

Full size image

Social support and partnership satisfaction over a 1-year period post therapy

The perceived social support of the patients decreased over the 1-year period post therapy (p = 0.027). Nevertheless, at all times of measurement, most of the patients reported feeling well or very well socially supported (Table 3). We found no significant changes in social support of the partners over time (t test for paired samples, p = 0.949).

Table 3 Social support and partnership satisfaction of patients and partners over a 1-year period post therapy, OSS-3 cutoff (3–8 = low, 9–11 = middle, and 12–14 = high) and OSS-3 sum score (range 3–14), and ADAS cutoff (<17.1 = low, 17.1–22.5 = middle, >22.5 = high) and ADAS mean scores
Full size table

Patients and partners mostly reported a high partnership satisfaction (Table 3); there were no significant changes over time (patient: p = 0.485; partner: p = 0.856).

Dyadic relationships

We did not find a significant difference between partners and patients in terms of anxiety and depression on pair level at any time of measurement (t test for paired samples, t1: N = 113 couples/t2 and t3: N = 74 couples; p = 0.275–0.908). Furthermore, patients and partners did not differ significantly regarding social support (t test for paired samples, p = 0.131–0.804) and partnership satisfaction (t test for paired samples, t1: p = 0.284/t2: p = 0.291/t3: p = 0.377).

To analyze the dyadic relationships, we used the following structural equation model and parameters to estimate the APIM (Fig. 2).

Fig. 2
figure 2

Parameters of the structural equation model to estimate the APIM; X = t1 (post therapy), Y = t2 (6 months after t1), a = patient, b = partner, e = error probability

Full size image

Table 4 shows the partner and actor effects on psychological distress (HADS total score) for the predictors social support (OSS sum score), partnership satisfaction (ADAS sum score), and employment (yes/no).

Table 4 Predictors of psychological distress in patients and partners (criterion: HADS total score (structural equation model), N = 81 couples)
Full size table

Observing social support, both actor effects were negative, showing a decrease of distress when social support increases. Increases of social support by one unit (1 overall standard deviation above the overall mean) led to a decrease of psychological distress for patients (beta = 0.206, p < 0.2) and partners (beta = 0.448, p < 0.001).

The partnership satisfaction showed no significant effects for patients, while for partners, the distress decreased by 0.448 (p < 0.001) deviations when satisfaction increases by 1 unit.

The psychological distress of the patient (in case of unemployment of partner and patient) was 0.216 overall standard deviations higher than the overall mean of distress (p ≥ 0.2). For partners, the level of psychological distress was 0.419 deviations higher (p < 0.05), when both were not employed. The significant actor effect of employment—which decreases the level of distress—was stronger for partners (patient, −0.361, p < 0.2; partner, −0.509, p < 0.01).

The patients’ social support, employment, and partnership satisfaction showed no significant effects on psychological distress of the partner, and vice versa (no partner effects, all error probabilities were greater than 0.2).

Conclusion

The study examined the level and the course of psychological distress (anxiety and depression) in cancer patients with children under 18 years and their partners based on three measurement times, starting after or up to 12 months after cancer therapy. Data were collected within a Germany-wide research project.

After cancer therapy, 15 % of the patients showed high levels of anxiety and 8 % had relevant levels of depression. Recent studies on this topic found similar levels of psychological distress [37–40]. Not only patients but also their partners showed high levels of anxiety (26 %). High levels of psychological distress have repeatedly been found in patients’ partners [41–43]. In a longitudinal study with couples dealing with colorectal cancer, patients and partners showed increased levels of psychological distress prior to surgery in comparison with a reference group [44]. We also found higher levels of anxiety in patients and partners compared to the general population. Similar mean scores for anxiety and depression were reported in a study with patients living with myeloma and their partners [20].

One year after cancer therapy, there were no significant changes in anxiety and depression of the patient. Above all, the anxiety symptoms remained temporarily stable at a high level and were higher than those in the general population. Levels of anxiety of the partners had decreased significantly over time, but they were, at all points of measurement, significantly higher than those in the general population. The findings add to the body of evidence showing considerable psychological long-term consequences of cancer and its treatment for both patients and partners. The results also suggest that the illness experience is a huge burden for both patients and partners, highlighting the high incidence of anxiety. In our study, high social support leads to a decrease of psychological distress for patients and partners. This result is consistent with other study findings and underlines the importance of continuing social support for both cancer patients and their partners [45, 46].

The findings of this study suggest that unemployment was associated with an increased likelihood of psychological distress in patients and partners. The positive impact of employment for psychological well-being of cancer patients and their partners has repeatedly been demonstrated [24, 47–49]. Through the integration into the daily working routine, patients and partners gain normalcy, structure, a sense of meaning, and support in the context of disease and treatment. Also, financial worries were lower when the patient’s partner was employed.

We found no significant differences on pair level in terms of anxiety, depression, social support, and partnership satisfaction. At least one other study was similarly unable to find significant differences of perceived relationship satisfaction within couples [50]. This could be explained by the finding that couples coping with cancer tend to react as an emotional system rather than as individuals [3].

For partners, the psychological distress decreased when partnership satisfaction increased. Similar results were reported from a study by Stafford and Judd, in which poor relationship satisfaction predicted greater anxiety [50], and by Lewis et al. who found a significant association between high depression in partners of breast cancer patients and low perceived quality of the partnership [49].

Through dyadic analyses, we could show that this burden can influence both partners within the couple. The complexity of a couple’s relationship and the entanglements of a couple’s joint distress should therefore receive more attention in clinical routines. Patients’ distress is often determined by their relationship and should thus be treated in a way that incorporates their relationship. In a clinical setting, this means that patients’ partners should be included in a systematic fashion, e.g., in diagnostic conversations. This could help patients adjust better to the disease and strengthen their partnership, both of which can ultimately also have a positive impact on the disease trajectory itself, e.g., through improved compliance, constructive communication about stress, and increased mutual health awareness.

Strengths and limitations

The strength of our work is that the study is one of the first investigations to use a longitudinal design and validated measures with parents with cancer and their partners. Our work focused on the dyadic adjustment between patient and partner over time in the context of a cancer disease in the family, using a structural equation model for dyadic data analyses. This methodological approach reflects the complexity of a couple’s relationship and the entanglements of a couple’s joint distress. Anxiety and depression of the patients and partners were compared to normative data from a large representative population survey matched by age, partnership, and parenthood.

A limitation of the work is that the aspect of parenthood played a subordinate role. Cancer affects the care of children, and it would be important to know more about the changes in parenting of cancer patients and their spouses when young children are involved. For parents who are responsible for the well-being of their children, it is a challenge, to cope well with the cancer and the treatment and, at the same time, to be good parents [5]. Study results which included the situation of the children were published elsewhere [28]. When evaluating the study results, one should consider that patients in a palliative situation or families where the patient had died in the course of the survey period were not included in or excluded from the study. It is possible that non-participants are characterized by more severe psychosocial complaints. The Hospital Anxiety and Depression Scale is a screening instrument. However, further diagnostic interviews are required for the evaluation of clinically confirmed mental disorders. Furthermore, diagnosis-related analyses would provide more precise and accurate findings; there is a risk of limited results due to the heterogeneity of the sample.

Implications and future research

With increasing age at pregnancy, the likelihood of cancer affecting a family with children under 18 years is rising and the issue of parents with cancer is gaining importance. So far, longitudinal studies have been lacking, which assess long-term consequences of parental cancer also for family members. Our results emphasize the need for long-term psychological and social support going beyond the acute treatment for both cancer patients and their partners. Counselors working with the family system can help families identify effective ways to cope with needed role changes, particularly during the treatment phase. In future studies on this issue, a dyadic perspective that focuses on the interdependence of distress should also be taken.