Risk factors of unmet needs among women with breast cancer in the post‐treatment phase

Abstract Objective Unmet health care needs require additional care resources to achieve optimal patient well‐being. In this nationwide study we examined associations between a number of risk factors and unmet needs after treatment among women with breast cancer, while taking into account their health care practices. We expected that more care use would be associated with lower levels of unmet needs. Methods A multicenter, prospective, observational design was employed. Women with primary breast cancer completed questionnaires 6 and 15 months post‐diagnosis. Medical data were retrieved from medical records. Direct and indirect associations between sociodemographic and clinical risk factors, distress, care use, and unmet needs were investigated with structural equation modeling. Results Seven hundred forty‐six participants completed both questionnaires (response rate 73.7%). The care services received were not negatively associated with the reported levels of unmet needs after treatment. Comorbidity was associated with higher physical and daily living needs. Higher age was associated with higher health system‐related and informational needs. Having had chemotherapy and a mastectomy were associated with higher sexuality needs and breast cancer‐specific issues, respectively. A higher level of distress was associated with higher levels of unmet need in all domains. Conclusions Clinicians may use these results to timely identify which women are at risk of developing specific unmet needs after treatment. Evidence‐based, cost‐effective (online) interventions that target distress, the most influential risk factor, should be further implemented and disseminated among patients and clinicians.

respectively. A higher level of distress was associated with higher levels of unmet need in all domains.
Conclusions: Clinicians may use these results to timely identify which women are at risk of developing specific unmet needs after treatment. Evidence-based, costeffective (online) interventions that target distress, the most influential risk factor, should be further implemented and disseminated among patients and clinicians.
cancer, distress, needs assessment, oncology, women with breast cancer 1 | BACKGROUND Breast cancer is the most frequently diagnosed cancer among women worldwide. 1 Thanks to the introduction of early detection programs and increasingly successful treatments, the number of breast cancer survivors keeps rising. Yet, this positive development entails an increasing number of survivors with disease-and treatment-related problems. Many report fatigue, psychological distress, poor physical fitness, motion restriction, lymphedema, sleep problems, cognitive problems, and menopausal symptoms. [2][3][4][5] Many women with breast cancer receive medical, paramedical, psychosocial, or complementary care to cope with these problems, up to more than 10 years after diagnosis. 6 For most women, the received resources are sufficient. 7 However, a considerable proportion of women with breast cancer express an unmet need for support, indicating that they would like additional help. The prevalence of specific needs may reach up to 70%. The highest needs are generally found in the psychological and health system-related and informational domain, with fear of the cancer spreading or recurring being the most prevalent. 8 To ensure that these needs are adequately addressed, clinicians might benefit from knowing which women with breast cancer are most at risk of developing unmet needs. A considerable number of studies addressed this topic. A systematic review found that, for instance, younger women, with advanced stage breast cancer, treated with chemotherapy, or those who experience a higher level of distress, report increased unmet needs. Risk factors may differ between domains. For example, women with a higher level of education report greater unmet need in the sexuality, and/or health system-related and informational domains, while women with a lower level of education generally report greater unmet need in the psychological and/or patient care and support domains. 8 These insights show promise in identifying the women with breast cancer most in need for additional support. However, more research is warranted. First, most studies employed cross-sectional designs to address risk factors of unmet need domains during cancer treatment, when the levels of unmet needs are highest. Far less studies examined risk factors of need domains shortly after treatment, when a greater number of patients experience unmet needs. 9 Patients' need for support may rise as they may miss regular contact with health care professionals and/or experience treatment-induced side effects. 10 Therefore, more prospective studies regarding risk factors of specific unmet need domains in the post-treatment phase are required. 11 Furthermore, health care needs, by definition, refer to problems that require an action or additional care resources to achieve optimal well-being. 8,9,12 This implies that received care resources and one's level of unmet need may be associated. 13,14 Yet, to the best of our knowledge, the extent to which different types of received health care, such as medical, paramedical, or psychosocial care, influence the relation between risk factors and specific unmet need domains, has not been investigated.
The current study aims to extend existing insights by examining associations between risk factors and unmet needs of women with breast cancer post-treatment, while taking into account varying types of care use. We hypothesized that higher levels of health care use are associated with lower levels of remaining unmet needs, thus that care use helps fulfill existing needs.
Based on the literature, we included age, educational level, cancer stage, types of treatment, and distress as sociodemographic, clinical, and psychosocial risk factors. 8 We additionally included type of care insurance and comorbidity as possible risk factors that deserve more research attention. 14,15 Patients may refrain from physical or psychological treatment if their insurance does not fully cover the costs. Consequently, they may experience higher levels of unmet need in the physical and daily living and psychological domains. Having one or more comorbid disorders may especially influence unmet need in the physical and daily living domain.
Finally, we included previous psychosocial treatment as a potential risk factor. Breast cancer patients with a history of mental illness are at higher risk of developing cancer-related distress. 16 If not adequately addressed, they may experience higher levels of psychological need over time.
Most of the included sociodemographic and clinical factors are also known risk factors for a higher level of distress. 17 As such, distress might be a more proximal risk factor, and possibly a mediating factor between the other risk factors and patients' levels of unmet needs. Therefore, we further hypothesized that the included sociodemographic and clinical factors influence unmet needs directly and indirectly through distress.

| Participants and procedure
Women with primary breast cancer diagnosed up to 6 months earlier in one of nine hospitals in the Netherlands (ie, six academic hospitals, two community hospitals, and one comprehensive cancer center) were eligible for the study, regardless of type of treatment. Patients were excluded if they were not literate in Dutch, younger than 18 years, and/or had a prognosis of 3 months or less.
Eligible patients were identified by their oncologist, cancer nurse, or nurse practitioner during a hospital visit. The clinician informed the patient about the study and asked whether she would consider participation. Subsequently, the investigator invited interested patients to participate by telephone or e-mail. Participating centers could exclude patients who were already participating in other studies. A more detailed description of the study procedure has previously been published. 7,18 Our aim was to include at least 900 participants, a sufficient number for testing multiple correlations, given the number of predictors. This number would suffice even when domain scores would be included separately, and taking into account a drop-out rate of 20%. 19,20

| Design
The study had a multicenter, prospective, observational design. Participants completed a self-report questionnaire at 6 months (time win-

| Sociodemographic and clinical factors
Age at diagnosis, educational level (8 response categories, including the option "other," see the legend of Table 2), type of health insurance (5 response categories: no health insurance, basic package, basic package and additional package without dental insurance, basic package and dental insurance, basic package with additional and dental insurance), number of comorbid conditions (17 response categories, including the options "none" and "other"), 21 and previous use of psychosocial services (yes/no) were assessed through self-report.
Cancer stage (via pTNM-classification) 22 and types of treatment were retrieved from medical records.

| Distress
Psychosocial distress was assessed with the validated Dutch version of the single item Distress Thermometer. 23, 24 The Thermometer is a visual analog scale, that measures the level of distress in the past week (score 0 "no distress at all" to 10 "extreme distress").

| Health care use
Health care use was assessed with a 24-item questionnaire. 7,18,21 The questionnaire measures how often in the past 3 months patients used specific types of medical (eg, visits to a surgeon), psychosocial (eg, visits to a psychologist), paramedical (eg, visits to a physical therapist), and supplementary care services (eg, use of paid child care) (response categories: 0, 1, 2, 3, 4, 5, more than 5 times). The legend of Table 2 provides an overview of the included services. We calculated a sum score for each type of care.

| Health care needs
Health care needs were assessed with the 34-item Supportive Care Needs Survey (SCNS-SF34), 25 and the 8-item SCNS breast cancerspecific module. 26 The SCNS measures patients' perceived needs over the past month in the psychological, health system-related and informational, physical and daily living, patient care and support, and sexuality domain (five response categories: no need, met need, some need, moderate need, high need). The additional module addresses breast cancer-specific needs, for example, related to experiencing lymphedema.
We calculated a Likert sum score for each domain, excluding the category "met need," such that a higher domain score indicated a higher level of unmet need for help (range 0 "no need" to 3 "high need"). This was an adaptation of the standard scoring procedure.
Cronbach's alpha coefficient in our study ranged from 0.86 to 0.96 for the SCNS-SF34 subscales and was 0.81 for the breast cancer-specific module. Finally, we transformed the scores to a scale ranging from 0 to 100 in order to facilitate comparisons across scores.

| Data analyses
Associations between the included risk factors, care use, and unmet needs over time were examined with structural equation modeling (SEM). 27 SEM has the ability to include multiple independent and dependent variables in one model, thus enabling simultaneous analysis of all hypothesized associations. Missing values ranged from 0% for age and the treatment-related variables to 9.5% for the unmet need score in the patient care and support domain. The full information maximum likelihood estimation method 28 was used to take missing data into account. 29

| Sample
Of 1353 women with breast cancer assessed, 1263 women were eligible, and 1012 agreed to participate. The analyses for this study included the 746 women who completed both questionnaires (73.7% of participants) (Figure 1). Most women were diagnosed with stage 1 or 2 invasive breast cancer, and were treated with lumpectomy and radiotherapy. Over 60% had one or more comorbid conditions (Table 1).
Participants did not differ significantly in age (groups based on median split, P > .10), cancer stage (chi-square, P > .10), or distress score 6 months post-diagnosis (P > .10) from nonrespondents, that is, from women who were approached by the investigator, but could not F I G U R E 1 Flowchart be reached or declined to participate. Participants who only completed the first questionnaire at 6 months post-diagnosis (n = 111) did not differ significantly in age (chi-square, P > .10), cancer stage (chisquare, P > .10), or distress score (t tests, P > .10) from those who completed both questionnaires (n = 746).

| Risk factors of distress
Patients reported on average a distress level of 3.93 (SD = 2.67) at 6 months post-diagnosis. Younger age (ß = −.18), having had chemotherapy (ß = .43), comorbidity (ß = .17), and psychosocial treatment before the breast cancer diagnosis (ß = .29) were found to be associated with a higher level of distress (P < .05). The predictors together explained 12.3% of variance, reflecting a small effect size (Table 2).

| DISCUSSION
This prospective, nationwide study identified risk factors of unmet needs among women with breast cancer after treatment, while taking into account their global reports of received care practices. Of the included sociodemographic, clinical, and psychosocial risk factors, higher age, having one or more comorbid disorders, having had chemotherapy or a mastectomy, and patients' level of distress were found to be significant direct risk factors.
The data, generally, did not support our hypothesis that women's reports of received care services are negatively related to their unmet needs. There were even some small positive associations, for example, With regard to the sociodemographic factors, there is strong evidence that younger age is associated with a higher number of unmet needs after cancer treatment. 11,33 Additionally, two studies among early cancer survivors reported associations between younger age and specific unmet need domains, namely the patient care and support, and the relationship/sexuality domain. 11 In contrast, our study showed a small but significant direct association between higher age and higher levels of unmet need in the health system-related and informational domain. While an overall effect of age on unmet needs has been established, we conclude that this might mask varying effects on separate underlying unmet needs domains. One explanation for our finding is that older patients are more likely to experience more comorbid or physical problems in addition to their cancer diagnosis, which requires additional support and information. Also, older cancer patients may have more difficulty with processing and recalling provided information due to decreasing cognitive and sensory functions. 34 In addition to a direct effect, we would like to highlight that

| Study limitations
The number of visits to care providers was assessed by self-report.
These results may be influenced by recall bias, leading to possible under-or overestimation of actual care practices. However, selfassessment allowed us to also assess the use of nonmedical types of care use, which are not standardly registered in medical files. Another study limitation is that many participants were recruited from radiotherapy departments. Therefore, women with breast cancer who do not receive radiotherapy, a minority, were underrepresented in our sample. A previous study based on data from a Dutch populationbased, regional cancer registry indicated that 17% of the women with breast cancer received systemic therapy without radiotherapy. 38 Furthermore, we were not able to gather information about women who declined to be approached for this study. Therefore, we could not determine our sample's representativeness in that regard. There was, however, no indication of a sample bias resulting from loss to follow-up.
Strengths of our study include its multicenter, prospective, design, its nationwide character, and large sample size. We addressed a key period in the disease trajectory that warrants further investigation of risk factors in relation to survivors' unmet needs. Indeed, thanks to the large sample size, we were able to simultaneously examine multiple risk factors of separate domains of unmet needs in concurrence with varying types of health care use of women with breast cancer.
To our knowledge, this is the first study to have done so.

| Future research suggestions
To ensure appropriate and cost-effective utilization of care services, the influence of health care use on varying domains of unmet needs post-treatment and in the survival phase deserves further attention.
We recommend employing standardized care use 39 and needs measurements 8 in order to allow comparisons across studies. Given the increasing number of women with breast cancer, we especially suggest further investigation of the association between psychosocial care use and unmet needs after a breast cancer diagnosis. Patients with informational or emotional needs may seek support from psychosocial providers, while their needs might be satisfied by participating in support programs, receiving written information, or using low-cost self-management resources. All future studies on this topic should include distress as a key risk factor, and preferably also as a mediating factor. Based on our results, we also recommend examination of types of comorbid disorders in relation to breast cancer patients' unmet needs after treatment.

| Clinical implications
Clinicians can use these results to identify in a timely manner those women with breast cancer who are at risk of developing higher levels of unmet needs after treatment. Distress was found to be the most relevant risk factor across need domains. Additionally, older women with breast cancer, with one or more comorbid disorders, who have had chemotherapy, or who had a mastectomy may have more specific needs. Higher age was associated with higher health system-related and informational needs. Comorbidity was associated with higher physical and daily living needs. Having had chemotherapy and a mastectomy were associated with higher sexuality needs and breast cancer-specific issues, respectively.
Taking into consideration the growing number of breast cancer patients and staff shortages, this raises the question how patients' needs can be adequately met, and perhaps even prevented. Based on our results and the literature, distress should be targeted as one of the most significant and modifiable risk factors of unmet needs after treatment. One possible cost-effective approach would be to address distress by use of a stepped care program, that is, watchful waiting, followed by guided self-help, problem-solving therapy, and psychotherapy or medication, if needed. Research shows that only a minority of patients will need the more resource-intensive practices. 40 Furthermore, the number of online self-management tools that target specific needs or distress is steadily growing. However, most of the promising, evidence-based, tools are not structurally updated and implemented in practice because of lack of funding after development. Also, potential users do not know which tools are available, and how to evaluate these tools in terms of quality, reliability, and privacy.
Therefore, based on our study results, we also strongly encourage endeavors to successfully implement and disseminate such low resource, technology-based aids among patients and clinicians.