Background
An estimated 30–50% of the general population in Norway will experience a mental disorder in their lifetimes, 10–20% will experience a substance use disorder (SUD), and about 30% will experience cancer [
1]. Thus, during their lifetimes, many people will experience an illness of a partner or other loved one across different illness domains. The illness not only will affect the patient but also will impact the partner, and several studies have found that the partner’s quality of life (QoL) is negatively affected and typically lower than that of the general population [
2,
3]. According to some studies, partner QoL can be even lower than that of the patient [
2,
4].
QoL can be affected for several reasons. A loved one’s illness may raise concerns and worries about the future [
2], which in turn can lead to stress, fatigue, and sleep deprivation [
2,
5]. Such factors can influence physical and mental health negatively, and anxiety and depression can be among the consequences [
6‐
8]. Physical and mental health form two integral components of QoL [
9,
10], and when they are affected, QoL will typically be perceived as impaired.
Another typical component of QoL is the social domain, or how people rate their social relations [
11]. A stressful event such as the illness of a loved one can be hard to deal with especially for partners who have no one with whom to share their problems [
12‐
14]. Conversely, higher perceived social support has consistently been associated with higher QoL [
2,
3,
14,
15]. Social life within the family is often affected as well; family cohesion may become disrupted, and the partner’s familial capacity can be weakened, as when the partner’s care is directed towards the needs of the patient at the expense of the needs of children in the household [
16,
17]. Several studies have reported correlations between QoL and impaired family life [
2,
5,
14].
Studies also have examined the association between socio-demographic variables and the QoL of those who have an ill partner [
5‐
7,
14,
18]. Although there is no obvious reason why gender, for example, would be associated with higher or lower QoL, some studies suggest that the QoL of female partners is more affected than that of male partners [
5,
8,
14]. Likewise, other socio-demographic variables and QoL might not be expected to interact automatically – i.e., people can be content with their lives despite difficult life circumstances [
9] – but adverse life conditions such as unemployment, higher financial burdens, or poverty would add to the strain when people experience stressful life events such as a loved one’s illness. Factors like these could reduce the ability to cope with the situation and in turn negatively affect QoL, as some studies also have suggested [
2,
5,
12]. Lack of engagement in work and/or school activity in themselves negatively affect QoL in a normative population [
19]. More specifically, partners of mentally ill patients or SUD patients often experience stigma attached to the illness or substance abuse [
20,
21], and some may even report overwhelming feelings of guilt and shame [
22,
23]. Such additional emotional burdens may cause partners, especially in these two illness domains [
24], to withdraw from social networks, further eroding QoL.
In this study, we compared QoL in partners of patients with illness across several illness domains, including somatic health, mental health, and SUD. To our knowledge, no studies have addressed this question by comparing these domains in this way. We hypothesized that:
1.
Partners of ill patients would have a lower QoL compared to the normative population.
2.
Partners of patients with mental illness or SUD would have a lower QoL compared to partners of patients with somatic illnesses.
The study aims were to (1) explore differences in socio-demographic, social/familial, and health variables and perceived QoL between partner groups and (2) to identify factors associated with QoL.
Results
Differences between partner groups
The total sample consisted of 213 partners: 116 in the somatic illness domain, 72 in the mental illness domain, and 25 in the SUD domain (Table
1). We found significant differences across groups. The proportion of women was higher in partners in the SUD group. Partners in this group also reported having significantly lower income, lower education level, and less work/school activity.
Table 1
Characteristics of participants (N = 213), with data presented as N (%) or mean (SD) / median (Interquartile range, IQR) [italics]
Age, years |
43 (7)/43 (9)
|
39 (8)/38 (15)
|
36 (10)/35 (15)
|
41 (8)/41 (12)
| <0.001 | 0.004 | 0.001 | ns. |
Gender, women | 35 (30) | 13 (18) | 16 (64) | 100 (38) | <0.001 | ns. | 0.001 | <0.001 |
Work/school activityc
| 89 (26)/109 (94) | 85 (33)/63 (88) | 52 (46)/15 (60) | 83 (33)/187 (88) | <0.001 | ns. | <0.001 | 0.003 |
Educational level |
- Primary education | 13 (11) | 5 (7) | 6 (24) | 24 (11) | | | | |
- High school | 37 (32) | 36 (50) | 14 (56) | 87 (41) | 0.003 | 0.044 | 0.003 | 0.023 |
- College/university | 66 (57) | 31 (43) | 5 (20) | 102 (48) | | | | |
Incomed
|
986′ (439)/900′ (380)
|
770′ (271)/700′ (250)
|
481′ (224)/450′ (210)
|
849′ (403)/800′ (385)
| <0.001 | <0.001 | <0.001 | <0.001 |
Social support (ISEL) |
38.2 (6.2)/38.5 (10.8)
|
37.3 (7.3)/37.0 (10.8)
|
37.2 (6.9)/38.0 (13.0)
|
37.8 (6.7)/38.0 (10.0)
| 0.767 | | | |
Family cohesion (FACES-III) |
42.3 (5.4)/43.0 (8.0)
|
41.4 (5.9)/42.0 (9.8)
|
40.6 (8.0)/41.0 (9.5)
|
41.8 (6.0)/43.0 (9.0)
| 0.649 | | | |
Perceived family capacity influenced by patient’s illnesse
|
0.9 (0.8)/0.8 (1.5)
|
1.2 (1.0)/1.0 (1.8)
|
1.0 (0.9)/0.6 (1.5)
|
1.0 (0.9)/0.9 (1.5)
| 0.053 | | | |
Perceived concern for childrenf
|
0.9 (0.9)/1.0 (2.0)
|
1.0 (1.1)/0.0 (2.0)
|
0.4 (0.8)/0.0 (0.0)
|
0.8 (1.0)/0.0 (2.0)
| 0.036 | ns. | 0.013 | 0.017 |
Substance use (CAGE-AID), cut-off >2 | 2 (2) | 2 (3) | 3 (12) | 7 (3) | 0.031 | ns. | 0.012 | ns. |
Perceived psychological distress (SCL-10) |
1.42 (0.43)/1.30 (0.68)
|
1.49 (0.58)/1.30 (0.80)
|
1.40 (0.55)/1.20 (0.40)
|
1.44 (0.50)/1.30 (0.70)
| 0.877 | | | |
Quality of Life (QoL-5) |
0.72 (0.13)/0.73 (0.17)
|
0.68 (0.16)/0.68 (0.23)
|
0.73 (0.13)/0.73 (0.20)
|
0.71 (0.14)/0.70 (0.23)
| 0.122 | | | |
The mean score on the family cohesion scale (FACES-III) was above the cut-offs for lack of cohesion and on the positive side of the social support scale (ISEL). The partners’ perceived capacity in the family was affected only modestly by the illness of the patient, as evidenced by a mean score close to the term “slightly affected” on the scale. There were no significant differences across groups in these variables. In terms of concern for the child/children in the family; the participants had little worry for the child (a mean score of ≤1 on the scale), with the lowest score in the SUD group.
In terms of health variables, only 7 (3%) scored above the cut-off the for severe substance use problems (CAGE-AID), with a slightly higher proportion of problematic substance use in the SUD partner group. Regarding perceived psychological distress (SCL-10), the mean score (1.44, SD 0.50) was below the pathological cut-off for all three groups, with 39 (18%) participants scoring above the cut-off for psychological distress. No differences in perceived psychological distress (SCL-10) emerged among the three groups.
QoL scores were similar to those of the normative population for the sample as a whole (0.71, SD 0.14), with no significant differences among groups (Table
1). A small proportion of the sample (13%) reported a markedly low QoL (<0.55).
Variables associated with QoL
In bivariate analyses, age and substance use (CAGE-AID) had p-values above the recommended lax criterion (p > 0.2); thus, they were excluded from further analyses and from the following model.
The first step of the hierarchical regression (socio-demographic variables) (Table
2) showed that income and work/school activity were significantly associated with QoL. This model explained 6% (R
2) of the variance. In the second step of the hierarchical regression, we added social/familial variables; family cohesion (FACES-III), perceived social support (ISEL), and perceived worry/concern about the child/children were significantly associated with QoL (Table
2). This model explained 33% (R
2) of the variance in QoL. The final model included health variables, and only two variables were significantly associated with QoL: perceived family cohesion and psychological distress. Perceived family cohesion was positively associated with QoL while psychological distress (SCL-10) was a negative predictor (beta = −0.16; 95% CI = −0.20/−0.13,
p < 0.001; Table
2). The final model explained 56% (R
2) of the variance in QoL (Table
2).
Table 2
Factors associated with QoL (N = 266)
Groupc
| 0.02 (−0.01/0.05) | 0.141 | 0.02 (−0.01/0.04) | 0.242 | 0.00 (−0.02/0.03) | 0.715 |
Gender | −0.02 (−0.07/−0.02) | 0.279 | −0.03 (−0.07/0.01) | 0.141 | −0.01 (−0.04/0.03) | 0.758 |
Education | 0.01 (−0.02/0.04) | 0.431 | 0.01 (−0.02/0.04) | 0.498 | 0.00 (−0.02/0.03) | 0.760 |
Work/school activityd
| 0.00 (0.00/0.00) | 0.050 | 0.00 (0.00/0.00) | 0.241 | 0.00 (−0.00/0.00) | 0.669 |
Incomee
| 0.01 (0.00/0.01) | 0.030 | 0.00 (−0.00/0.01) | 0.180 | 0.00 (−0.00/0.01) | 0.437 |
Social / familial variables |
Family cohesion (FACES-III)f
| | | 0.05 (0.02/0.08) | 0.003 | 0.05 (0.02/0.07) | 0.001 |
Social support (ISEL)g
| | | 0.07 (0.03/0.10) | 0.001 | 0.03 (−0.00/0.06) | 0.091 |
Concern about child Family capacity | | | −0.03 (−0.05/−0.01) −0.01 (−0.04/0.01) | 0.002 0.186 | −0.01 (−0.03/0.01) −0.01 (−0.02/0.01) | 0.206 0.558 |
Health variables |
Psychological distress (SCL-10) | | | | | −0.16 (−0.20/−0.13) | <0.001 |
Discussion
Some socio-demographic variables differed significantly among the groups in this study; partners in the SUD group differed significantly in terms of gender (being female), lower work/school activity, lower educational level, and lower income. The QoL score for the total sample was similar to that of a normative population sample, with no significant differences in QoL among groups. In a regression model, perceived family cohesion was positively associated with QoL whereas psychological distress was negatively related to it. The model explained 56% of the variation in QoL.
The normality of the QoL scores in this population was unexpected in light of the known strain of having an ill partner [
8,
12,
14,
18]. Previous studies among partners to ill patients showed that if the patient received treatment, the impact on the partner’s QoL was positive [
8,
52,
53]. Our participants were recruited during a treatment period for the ill parent, which may in part explain the unexpectedly high QoL in our sample. However, in the long run, treatment does not necessarily lead to a better QoL in the partner if the patient does not have a remission [
6,
53,
54]. Nonetheless, 13% reported a markedly low QoL. This finding indicated that a relatively small proportion of the sample seemed to struggle more with their life situation.
The lack of significant differences in QoL between groups was also surprising and was contrary to our hypothesis. The partners in the SUD group were worse off in terms of some socio-demographic conditions; for example, they were less likely to being occupied with work or school, and had a poorer educational level and income than the other two groups, but these findings were not reflected in a poorer QoL at the group level. In general, poorer socio-demographic conditions seem to affect QoL negatively [
9]; however, the subjective experience of such conditions affects QoL more than the ‘objective’ differences [
55]. Thus, in line with other research [
56,
57], overall QoL is more than a measurement of objective demographic conditions; it reflects how the individual relates to these conditions. An alternative explanation may be that when the patient receives treatment, a partner in the SUD group experiences a relatively greater relief from worries and burdens and perhaps perceives a temporal relief from their worries [
54]. Thus, their QoL score may have been overestimated at this specific point in time.
Family cohesion were retained as significant factors associated with QoL in the final regression model. This outcome has been seen in previous studies among partners of patients with illness and accentuates the importance of perceived proximity and cohesion in close relations to retain the QoL [
2,
14,
54]. The experience of instability and insecurity that partners of ill patients report may affect perceived family cohesion and also underlie the negative influence on QoL [
5,
14].
Psychological distress (SCL-10) was the strongest variable explaining variations in QoL. A one-point gain (higher psychological distress) resulted in a 0.16 lower QoL-5 score in the final adjusted model, suggesting a substantial influence when applying the clinical interpretation of the scale [
46]. The fit of the model was also strengthened considerably, and the explained variance in QoL increased from 33% to 56% with inclusion of this clinical variable. Feelings of hopelessness, worry, stress, and depression have been observed in partners of somatic or mentally ill patients [
12,
18], as has anxiety in relatives of SUD patients [
4,
6]. Such negative emotions may underlie the psychological distress reported here, which in turn strongly predicted worse QoL. High psychological distress would likely make an individual less able to cope well with a difficult situation arising when a close relative suffers from an illness. Other studies also report strong correlations between psychological distress and poor QoL [
2,
6‐
8,
52], affirming the findings of our final model. However, with the present design, we cannot discern whether the reported psychological distress existed before the illness or was a reaction to having an illness in the family.
Methodological considerations
The strengths of the study include an acceptable sample size and inclusion of groups of respondents who have not been compared before; previous studies tend to focus on separate domains. However, some limitations must be kept in mind. The sample size per group may not have been large enough for detecting statistical significant differences between them. Furthermore, the participants were recruited while the ill parent was in treatment, which might limit the representativeness of the findings. The participants in most benchmark studies in this field have an average age at least 10 years greater than in our study [
7,
8,
12‐
14,
53]. The sample is therefore mainly representative of middle-aged partners and time periods when the ill parent is enrolled in treatment. Although socio-demographic variables differed among the groups, the findings indicate that we did not recruit respondents with extreme economic or social disturbances in their lives. One possible question is if those who did not participate experienced more disturbances compared to those who did [
12]. The attrition analysis showed that there was a lower inclusion rate in the mental health and SUD illness domain, indicating that our results may be positively biased in these two illness domains. Further attrition analysis was not possible because administrative data on non-inclusion were insufficiently registered. The limited sample size per group also prevented us from examining whether there were different associations between independent variables and QoL across groups, i.e. with separate regression analyses for each group. In spite of the limitations, the findings provide important information about obstacles and facilitators of QoL in partners, which may be informative for further research and interventions.
Implications
Although the findings indicate that the sample as a whole reported a QoL score in line with the general population, some respondents still reported a markedly low QoL. We suggest that such brief QoL tools can be used to capture those who are struggling most with their life situation.
Acknowledgments
The authors are grateful to the respondents who shared their experiences. We would like to acknowledge and thank all of the partners who participated in this study and all of the institutions that made it possible for us to recruit them. These institutions include Sørlandet Hospital Trust, Akershus University Hospital Trust, Vestre Viken Hospital Trust, Helse Stavanger University Hospital Trust, Rogaland A-senter, and Nordland Hospital Trust. Furthermore, we would like to express our gratitude to the coordinators, research assistants, and PhD students who collected the data. We would also like to thank the Regional Center for Child and Adolescent Mental Health, Eastern and Southern Norway, for technical support in collection of data, and the network BarnsBeste (Children’s Best Interest) for contributing with user perspective to this study.