Study design and subjects

This is a cross-sectional study using internet-based surveys. The internet survey company we used is a Japanese company specializing in academic research. For the patient survey, in January 2010, we screened all registered members of the panel of internet-survey company A and selected those who developed cancer within the previous 5 years and experienced social problems. For the spouse survey, in the same manner in November 2016, we selected spouses of patients who developed cancer within the previous 5 years. Subjects were 259 patients and 259 patients’ spouses and all cancer types were selected.

Assessment indicators

Analysis methods

First, sex, age group, and presence or absence of recurrence were selected as covariates that should be adjusted based on clinical judgment, and propensity scores for these covariates were estimated using logistic regression models. Based on the calculated propensity scores, the patient and spouse groups were matched in a 1:1 ratio using nearest neighbor matching by sampling without replacement. Frequency distributions of the matched data were generated for background factors in each group to assess the balance among the covariates.

Second, differences in the degree of social difficulties experienced by the patient and spouse group were analyzed by independent two-sample t-tests. Furthermore, t-tests were also performed to analyze differences in the degree of the difficulties for the following combinations of subgroups: between the younger generation (≤39 years) and middle-aged and older generation (≥40 years), and between male patients/female spouses and female patients/male spouses.

The significance level was set at 5%. SPSS for Windows 23 was used to perform statistical analyses.

Subcategories of social difficulties experienced by cancer patients and spouses according to the KJ method (affinity diagram)

According to the KJ method, social difficulties were classified into 14 labels. Specifically, the 60 items were classified into the following subcategories: “difficulty in performing activities of daily living,” “difficulty in seeking expert advice on the disease state and treatment,” “complaints with health care providers,” “lack of information on treatment and disease state,” “lack of information on self-care,” “conflict over family relationships,” “concerns for family members,” “difficulty in planning life,” “conflict over relationships with non-family members,” “difficulty in adapting to changes in the social environment,” “lack of local support services,” “difficulty in solving work-related issues,” “difficulty in making financial arrangements,” and “lack of information on welfare services available during treatment.” The α coefficients for the internal consistency of the subcategories are shown in Table 1.

Matching for the characteristics and propensity scores of the subjects

When the patients and spouses were matched based on the propensity scores calculated from logistic regression models, 259 subjects were selected from each of the patient and spouse groups (Fig. 1). Table 2 shows the frequency distributions of the covariates and other background factors for each group. Regarding the covariates, the frequency distributions of sex, age group, and presence or absence of recurrence were completely matched between the patient and spouse groups, confirming that the groups were well balanced and matched. Regarding other background factors, cancer sites varied in both the patient and spouse groups, while most participating patients and patients of most participating spouses were being treated or followed up after the completion of treatment.

	Patients n (%)	Spouses n (%)
Patients’ sex
Men	128 (49.4)	128 (49.4)
Women	131 (50.6)	131 (50.6)
Patients’ age
20–29 years	11 (4.2)	11 (4.2)
30–39 years	44 (17.0)	44 (17.0)
40–49 years	67 (25.9)	67 (25.9)
50–59 years	85 (32.8)	85 (32.8)
60- years	52 (20.1)	52 (20.1)
The presence or absence of recurrence
Primary occurrence	62 (23.9)	62 (23.9)
Recurrence	197 (76.1)	197 (76.1)
Cancer site (including duplicates)
Lung cancer	34 (13.1)	29 (11.2)
Prostate cancer	25 (9.7)	13 (5.0)
Renal cancer	19 (7.3)	6 (2.3)
Bladder cancer	19 (7.3)	5 (1.9)
Testicular cancer	15 (5.8)	5 (1.9)
Gastric cancer	48 (18.5)	31 (12.0)
Esophageal cancer	20 (7.7)	5 (1.9)
Colorectal cancer	49 (18.9)	50 (19.3)
Liver cancer	17 (6.6)	6 (2.3)
Gallbladder cancer	13 (5.0)	2 (0.8)
Pancreatic cancer	12 (4.6)	3 (1.2)
Breast cancer	79 (30.5)	63 (24.3)
Thyroid cancer	20 (7.7)	6 (2.3)
Head and neck/oral cancer	11 (4.2)	7 (2.7)
Uterine cancer	27 (10.4)	12 (4.6)
Ovarian cancer	18 (6.9)	3 (1.2)
Leukemia	17 (6.6)	8 (3.1)
Malignant lymphoma	18 (6.9)	15 (5.8)
Malignant bone tumor	10 (3.9)	1 (0.4)
Brain tumor	17 (6.6)	1 (0.4)
Skin cancer	14 (5.4)	2 (0.8)
Cancer of unknown primary	5 (1.9)	0 (0.0)
Others	18 (6.9)	24 (9.3)
Treatments (including duplicates)
Surgery	217 (83.8)	208 (80.3)
Radiation therapy	92 (35.5)	95 (36.7)
Chemotherapy	147 (56.8)	101 (39.0)
Patients’ treatment status
Patients are currently under cancer treatment	138 (53.3)	118 (45.6)
Cancer was cured, and treatment was completed.	90 (34.7)	120 (46.3)
Cancer is not cured, but aggressive treatment has been completed.	12 (4.6)	12 (4.6)
Others	19 (7.3)	9 (3.5)
Academic background
Junior high-school graduates	8 (3.1)	10 (3.9)
High-school graduates	68 (26.3)	73 (28.2)
Vocational school/junior college graduates	53 (20.5)	65 (25.1)
Four-year college graduates	130 (50.2)	111 (42.9)
Work
Full-time employment	98 (37.8)	140 (54.1)
Part-time employment	28 (10.8)	35 (13.5)
Full-time housewife	37 (14.3)	55 (21.2)
Leave of absence because of the disease	15 (5.8)	1 (0.4)
Voluntary retirement because of the disease	21 (8.1)	6 (2.3)
Dismissal (closure of business) because of the disease	11 (4.2)	0 (0.0)
Mandatory retirement	23 (8.9)	12 (4.6)
Unemployed	11 (4.2)	6 (2.3)
Others	15 (5.8)	4 (1.5)

Differences in the degree of social difficulties experienced by the patient and spouse groups

Comparison of difficulty between patients and spouses

The degree of difficulty was higher in the patient than the spouse group for all subcategories except “difficulty in seeking expert advice on the disease state and treatment (lack of opportunities to consult for patient transfer arrangement, hospital selection, second opinion, psychological counseling, etc.).” Patients and partners are interdependent in that cancer impacts on their shared life, both emotionally and practically. However, no conclusion has been reached on whose distress is more severe because there are conflicting reports. While some reports indicate that distress severity is comparable between patients and their spouses or partners, other reports indicate that patients’ distress is more severe than that of their spouses or partners, and there are even reports indicating that distress of spouses or partners is more severe than that of patients [17]. Hodges et al. [18] investigated mental distress of spouses or partners of patients over the course of disease and confirmed that the distress of spouses or partners significantly correlated with that of patients. Then, they indicated that the mental distress of spouses or partners gradually increases after diagnosis and becomes more strongly correlated with that of patients. Because currently treated patients and spouses of such patients accounted for 50% of the subjects in the present study, the degree of difficulty perceived by the spouses might not have been as high as that of difficulty perceived by the patients. However, we assume that the degree of “difficulty in seeking expert advice on the disease state and treatment” in the spouses was as high as that in the patients because spouses were greatly involved in treatment of patients soon after diagnosis.

Difficulty perceived by the young spouses

In the young subgroup, the degree of difficulty perceived by the spouses was as high as that of difficulty perceived by the patients for all subcategories except “difficulty in adapting to changes in the social environment (I feel that my appearance has changed or that I am treated differently).” People aged 39 years or younger, who were classified as the younger generation in the present study, are called “adolescents and young adults (AYA).” Cancer patients in the AYA generation experience, after diagnosis and treatment, not only difficulties associated with social relationships, work, academic background, property, etc., but also many physical and psychosocial problems, such as interruptions to romantic and/or intimate relationships, reconsideration of family planning, infertility, and body image dissatisfaction [19, 20]. For this reason, young caregivers in this generation often seem to feel burdened [21, 22] and to perceive a high degree of difficulty.

Difficulty perceived by female spouses

For the combination of a male patient and a female spouse, the degree of difficulty perceived by the spouses was as high as that of difficulty perceived by the patients for “difficulty in seeking expert advice on the disease state and treatment (lack of opportunities to consult for patient transfer arrangement, hospital selection, second opinion, psychological counseling, etc.),” “complaint with health care providers,” “lack of information on treatment and disease state,” “lack of information on self-care (lack of knowledge on nutritional needs of patients or how to deal with anxiety),” “concerns for family members,” “lack of local support services,” “difficulty in making financial arrangements,” and “lack of information on welfare services available during treatment (lack of knowledge on the nursing-care insurance system or nursing-care facilities and equipment).” Generally, compared to men who take care of their wives with cancer, women who take care of their husbands with cancer have higher mental morbidity (high levels of distress, depression, and anxiety, and a low level of mental health), physical morbidity (low physical health score, decreased physical function, and loss of physical fitness), and social morbidity (low satisfaction in marriage and limited social support) [12, 23, 24]. Ussher et al. [25] attributed this to the fact that women caregivers are positioned as all-encompassing expert careers, expected to be competent at decision-making, a range of physical caring tasks, and provision of emotional support for the person with cancer. The consequences of this positioning are over-responsibility and self-sacrifice, physical costs and overwhelming emotions. Men caregivers positioned caring as a competency task which they had mastered, and which provided them with satisfaction.

Limitations

The present study has several limitations. First, the results may have been affected by measurement bias because of the use of data from internet-based surveys. Data reliability is limited by the facts that the participants in these surveys determined whether they met the eligibility criteria and that data on diagnosis, treatment regimens, etc., were self-reported. Thus, in future studies, more accurate medical data need to be collected by conducting questionnaire or interview surveys at medical institutions in combination with review of medical records. However, a merit of an internet-based survey is that it guarantees anonymity and allows participants to respond without worrying about health care providers. Second, because of arrangements for this study and funding issues, several years passed between the patient and spouse surveys, and we were unable to collect data from patient-and-spouse pairs. These facts may also have contributed to measurement bias. The degree of social difficulties differed between the patients and spouses in the present study because both groups might have had different underlying problems. Thus, in future studies, patient-and-spouse pairs need to be targeted and surveyed around the same time. However, in this study, comparison was made while the differences in background problems were minimized as much as possible by matching propensity scores for data on sex, age group, and presence or absence of recurrence. Third, an analysis of non-responder or cancer patients who did not experience social difficulties is key to gaining information about a possible sample bias that might impact study results. This was not done in this study. Fourth, an unstandardized scale was used in this study. That is why we prepared the scale using the KJ method and confirmed the reliability with an α coefficient.