Psychosocial risk, symptom burden, and concerns in families affected by childhood cancer

The revised Psychosocial Assessment Tool (PATrev) is a common family-level risk-based screening tool for pediatric oncology that has gained support for its ability to predict, at diagnosis, the degree of psychosocial support a family may require throughout the treatment trajectory. However, ongoing screening for symptoms and concerns (e.g., feeling alone, understanding treatment) remains underutilized. Resource limitations necessitate triaging and intervention based on need and risk. Given the widespread use of the PATrev, we sought to explore the association between family psychosocial risk, symptom burden (as measured by the revised Edmonton Symptom Assessment System (ESAS-r)), and concerns (as measured by the Canadian Problem Checklist (CPC)). Families (n = 87) with children ≤ 18 years of age (M = 11.72, male: 62.1%) on or off treatment for cancer were recruited from the Alberta Children’s Hospital. One parent from each family completed the PATrev and the CPC. Participants 8–18 years of age completed the ESAS-r. Results. Risk category (universal/low risk = 67.8%, targeted/intermediate risk = 26.4%, clinical/high risk = 5.7%) predicted symptom burden (F[2, 63.07] = 4.57, p = .014) and concerns (F[2, 82.06] = 16.79, p < .001), such that universal risk was associated with significantly lower symptom burden and fewer concerns. Family psychosocial risk is associated with cross-sectionally identified concerns and symptom burden, suggesting that resources might be prioritized for families with the greatest predicted need. Future research should evaluate the predictive validity of the PATrev to identify longitudinal concerns and symptom burden throughout the cancer trajectory.


3
Background While a diagnosis of childhood cancer is often extremely distressing to those families affected, the majority of families adjust well. Yet, for some, this distress persists well into treatment and survivorship [1]. For this subset of families, psychosocial concerns and burdensome symptoms can dramatically decrease quality of life [2,3]. Left unaddressed, these challenges may negatively affect satisfaction with care, a family's ability to cope, longterm mental health, and potentially mortality of the child [3][4][5][6]. Even after treatment has ended, symptoms such as pain and fatigue may persist for many years into adulthood [7][8][9][10]. In response, standards for the psychosocial care of children with cancer and their families were published in 2015. These 15 evidence-based standards advocate for routine screening and assessment of psychosocial health needs within pediatric oncology [11,12]. While these standards represent an important step forward for psychosocial care, there remains little guidance to direct healthcare professionals about how psychosocial difficulties should be assessed.
In contrast, routine screening of psychosocial health is well established within adult oncology and implemented widely in countries such as Canada and Australia [13,14]. Here, psychosocial screening has primarily focused on ongoing assessment of practical and biopsychosocial concerns (e.g., difficulty making treatment decisions, feeling like a burden to others, via the Canadian Problem Checklist) as well as symptom burden (e.g., pain, fatigue, via the Edmonton Symptom Assessment System). For adults, effective screening using these two tools is associated with fewer emergency room visits, increased patient engagement in treatment, and potentially improved survival [15][16][17].
While distress screening is less well established within pediatric oncology, there has been advocacy for a familycentered approach, given the important social context of family dynamic for children and adolescents undergoing treatment [1,18]. Differences in distress and coping may exist between parents. For example, some research suggests that mothers may experience more distress at diagnosis than fathers, with diminishing differences over time [19]. Mothers are more likely to use emotion-focused coping strategies (e.g., social support, information seeking), while fathers are more likely to use problem-focused strategies (e.g., problem-solving, providing financial support) [19]. In addition, traditional gender roles are often amplified when a child is diagnosed with cancer, with mothers more likely to shoulder increased caregiving demands, meaning that fathers may be seen less than mothers by their child's healthcare team [19]. There may also be important sociocultural differences with regard to distress and coping [20]. Thus, how a family is coping may not be evident and might vary significantly between family members, all of whom may also have a significant impact on a child's functioning and well-being. In response to this, the Psychosocial Assessment Tool (PAT) was developed as a screening tool to be used at diagnosis, to identify families likely to require greater psychosocial support and intervention [21]. Families are differentiated by three tiers of psychosocial risk: universal (distressed but resilient); targeted (acute or elevated distress); and clinical (severe, escalating, or persistent distress). The three tiers of psychosocial risk are then used to facilitate referral to appropriate resources and treatment. Indeed, the PAT has shown feasibility and a strong ability to predictively differentiate families by the degree of psychosocial support required throughout treatment and into survivorship [22][23][24][25].
Symptom screening may be especially important within pediatric oncology settings, as children may not report symptoms until they become severely problematic [26,27]. Some children may believe debilitating symptoms are necessary to cure their cancer, be reluctant to worry their family, or worry about burdening their healthcare team [28]. When symptoms are known by families, greater symptom burden and unaddressed concerns may contribute to difficulty making treatment decisions and adhering to treatment recommendations [18,29]. In survivorship, unmet needs pertaining to symptom management are prevalent and associated with increased fear of cancer recurrence [10]. As such, it is exceedingly important that concerns and symptoms are identified systematically and early to ensure families can be given the appropriate resources and tools to address symptoms and concerns, and navigate treatment and survivorship. Given the use of familial level psychosocial risk screening in pediatric oncology, we sought to explore the association between psychosocial family risk, parent-identified concerns, and patient-reported symptom burden in a pediatric oncology sample. We hypothesized that higher psychosocial risk category would correspond to more parentidentified concerns and greater symptom burden as reported by patients.

Participants
Ethics approval was obtained from the Health Research Ethics Board of Alberta: Cancer Committee (HREBA. CC-16-0274). Participants were recruited from the Alberta Children's Hospital over a 24-month period, as part of a larger study [2]. Families were eligible to participate if patients were 18 years of age or younger; on active treatment for cancer, or had completed cancer therapy; were fluent in English; and had at least one parent who was fluent in English and was willing to participate. Families who were deemed eligible by their primary healthcare team were approached by a research assistant during a standard clinic appointment. Informed written consent was obtained from all participating parents and patients who were 18 years of age, and verbal assent was obtained from participating children and adolescents who were 8-18 years of age. Parents completed a demographics questionnaire, the PAT-revised edition (adapted for families in Canada), and the CPC. For children 7 years of age or younger, only parent-reported data was collected. Children and adolescents between 8 and 18 years of age completed the ESAS-revised edition. All questionnaires were completed on paper and returned to the research assistant via a sealed envelope, either directly or via a primary healthcare provider.

Psychosocial Assessment Tool (PATrev) [21, 30] The
PATrev is a comprehensive family-level psychosocial risk assessment tool that was developed for use in pediatric oncology settings and has been adapted for a Canadian setting. The PATrev contains 57 items across seven domains: family structure and resources; social support; child problems; sibling problems; caregiver stress reactions; family problems; and family beliefs. Total scores range from 0 to 7, with higher scores indicating higher psychosocial family risk. Families are classified as belonging to one of three categories: universal (lowest risk, scores 0.00-0.99), targeted (elevated risk, scores 1.00-1.99), or clinical (highest risk, scores 2.00-7.00). The PATrev has previously shown strong construct validity, discriminative validity, inter-rater reliability (0.77), internal consistency reliability (0.85), and test-retest reliability (0.75) [30].

Edmonton Symptom Assessment System (ESAS-r) [31]
The ESAS-r is a 10-item self-report measure of the most common symptoms experienced by cancer patients including Pain, Tiredness, Nausea, Depression, Anxiety, Drowsiness, Appetite, Wellbeing, Shortness of Breath, and Other. Each item is rated on a 10-point scale of severity and can be summed to provide a total score for each patient (scale: 0-100). Patient self-report data were collected for participants aged 8-18. The ESAS-r is widely used in both clinical practice and research worldwide, and it has been translated into over 20 languages (Hui & Bruera, 2017). While the ESAS-r was developed for use with adults, available evidence suggests children ≥ 8 years of age can provide valid responses on self-report symptom measures [32]. Previous research has found the ESAS-r to produce good internal consistency reliability (0.79), test-retest reliability (0.86), and convergent validity (0.56-0.85) [33].

Canadian Problem Checklist (CPC) [15]
The CPC is an evidence-based [34,35] 21-item self-report checklist designed to query common concerns that cancer patients experience and is used across Canada as a part of a Screening for Distress toolkit. It includes six broad domains: Emotional, Spiritual, Practical, Social/Family, Informational, and Physical. This tool is intended to be used in conjunction with the ESAS-r and indicates concerns over the week preceding completion [14,15]. Clinically, the CPC is used as a qualitative tool to communicate possible concerns with the healthcare team (i.e., via items that are endorsed). For the present study, a summary score was calculated by counting the number of items endorsed on this checklist by each participant to produce a "total score" from 0 to 21, as has been previously described [36]. Parent-reported data were collected. While psychometric properties of the CPC have yet to be established, psychometric evaluation may be of limited value given the checklist format [37].

Data analysis
All analyses were performed using SPSS 24.0. Descriptive analyses were performed to describe participant characteristics and distributions of items endorsed on the ESAS-r and CPC. For our primary analyses, multilevel modelling was used to account for the nested structure of participant data within phase of cancer treatment (on treatment vs off treatment). Multilevel modelling assumes that micro-level data (i.e., participant data) are dependent within macrolevel groups (i.e., treatment phase), and thus accounts for nonindependence of data by simultaneously estimating variability at each level [38]. Two multilevel models were conducted to evaluate our hypotheses. For model 1, ESASr and PATrev data (level-1) were nested within treatment phase (on treatment vs off treatment; level-2). The PATrev was entered as a predictor of the ESAS-r, and patient's sex and age were included as covariates given their associations with pediatric cancer diagnoses, treatment, and outcomes [39][40][41]. For model 2, CPC and PATrev data (level-1) were nested within treatment phase (on treatment vs off treatment; level-2). The PATrev was entered as a predictor of the CPC and patient's sex and age were included as covariates. Fixed effects (i.e., coefficients for PATrev) were interpreted for adjusted models that produced a significant omnibus test with a = 0.05.

Participants
A sample of 87 families were included in the final analyses (on treatment n = 40; off treatment n = 47). Of these, 14 families included patients ≤ 7 years of age at the time of data collection and as such, only parent-reported data (demographics form, PATrev, CPC) were collected from these families. Overall, patients (62.1% male) were on average 11.72 years of age at the time of data collection. The most common diagnoses were leukemia and lymphoma (45.9%), followed by solid tumors (28.7%) and malignancies of the central nervous system (16.1%). Parent respondents included mothers (91.9%), fathers (7.0%), and one grandparent (1.2%). Of 186 families identified as eligible and invited to participate, 99 declined or did not return complete packages. Additional information about study recruitment can be found in Fig. 1. There were no significant differences between participants and those who declined or did not complete packages with respect to sex, age, diagnosis, age at diagnosis, treatment status (on vs off treatment), or type of treatment received (surgery, chemotherapy, radiation, transplant).
For families undergoing active treatment, 27 (67.5%) were classified into the universal risk category, 12 (30.0%) were classified into the targeted risk category, and one family (2.5%) was classified into the clinical risk category. For those families who had completed therapy, 32 (68.1%) were classified into the universal risk category, 11 (23.4%) were classified into the targeted risk category, and four (8.5%) were classified into the clinical risk category (see Fig. 2).
Additional patient and family demographics can be found in Table 1.

Edmonton Symptom Assessment System
Of 73 patient participants between 8 and 18 years of age, 65 completed the ESAS-r and were included in analyses. Internal consistency reliability for the current sample was good (α = 0.90). Out of a possible score of 100, children and adolescents in the universal risk category reported an average symptom summary score of 11.02. The most highly rated symptom was tiredness (M = 2.26, SD = 2.  Table 2. Results of the multilevel model indicated that PATrev risk category was significantly associated with patient's self-reported ESAS-r summary score F (2, 63.07) = 4.57,  Table 3.

Canadian Problem Checklist
Of 87 parents who participated, 84 completed the CPC and were included in analyses. Internal consistency reliability for the current sample was good (α = 0.87). Of 21 possible concerns, parents in the universal risk category endorsed, on average, 2.63 (SD = 3.18). The most common concerns in this group were fears (n = 22, 37.3%) and sleep (n = 15, 25.4%). Parents in the targeted risk category endorsed 6.43 items on average. Their most common concerns were fears (n = 17, 73.9%), followed by sleep (n = 15, 65.2%), sadness (n = 15, 65.2%), and work/school (n = 15, 65.2%). Finally, parents in the clinical risk group endorsed 8.50 concerns on average. Their most common concerns were fears (n = 4, 80.0%) and sleep (n = 4, 80.0%). Additional information about concerns endorsed can be found in Table 2.
Results of the multilevel model indicated that PATrev risk category was significantly associated with parent-reported CPC summary score F (2, 82.06) = 16.79, p < 0.001. Exploring further, there was a significant difference in CPC summary scores between those in the universal and targeted risk categories (b = 3.72, p < 0.001, SE = 0.80, 95% CI = 2.14, 5.30), as well as between the universal and clinical risk categories (b = 6.73, p < 0.001, SE = 1.68, 95% CI = 3.40, 10.07), such that those in the universal risk category endorsed significantly fewer concerns than those in the targeted or clinical risk categories. The difference between summary scores for those in the targeted and clinical risk categories did not reach significance (b = 3.01, p = 0.090, SE = 1.75, 95% CI = − 0.48, 6.50). Additional information about this model can be found in Table 3.

Discussion
The aim of the current study was to assess the association between psychosocial family risk, parent-identified concerns, and patient-reported symptom burden in a pediatric oncology sample. Our hypothesis that greater psychosocial family risk would be associated with more parent-reported concerns and greater patient-reported symptom burden using cross-sectional data was partly supported by the data. Results indicated that those in the targeted and clinical levels of familial risk on the PATrev reported more concerns and higher symptom burden than those in the universal risk category during both active treatment and after treatment completion. No statistical differences in symptom burden between families in the targeted and clinical risk categories were present in the current sample. Our findings that the majority of families were identified as belonging to the universal risk category, followed by targeted and clinical risk categories, are consistent with the broader literature [18,21,42]. Though comprehensive research on child and teen symptoms in pediatric oncology is scant, the most commonly evaluated and reported symptoms   (6.45) in the literature include fatigue, sleep, pain, and psychosocial distress [7,28,43]. While our results reflect some of these symptoms (e.g., fatigue, pain, and anxiety) as the most intensely experienced, it is unclear which symptoms patients find most distressing, which may differ. Certainly, there is evidence for the association between the symptom frequency and worry among children and adolescents with cancer, but this does not always align [44]. Bottom-up research is also needed to identify which symptoms are most relevant for children and teens, as these symptoms may differ from the most frequently reported adult symptoms captured by the ESAS-r. Finally, future research should also work to better characterize the multidimensional aspects of symptoms experienced by this population (e.g., frequency, duration, distress, interference) [28].
Our results expand previous findings that familial psychosocial risk is related to psychosocial distress [12,45] by identifying relationships with symptom burden as well as other practical and psychosocial concerns. Previous research has explored the relationship between family factors (e.g., family cohesion, parent distress) and symptom burden as well as health-related quality of life, and found that family and parent factors are related to health outcomes [46]. The relationship between family risk and child health has been well-established in the general population [47,48]. Results of this work have called for a need to identify interventions focused on family function and parent distress as potential moderators of child health, although interventions focused on parent or family functioning within pediatric oncology are relatively rare [49][50][51].
The results of this study have important clinical implications. Regular screening for symptoms and concerns is needed to systematically identify and intervene as these challenges arise. The need for screening is further emphasized by the fact that symptoms and concerns were present during both active treatment and after treatment completion, and especially given that nearly a quarter of families in our study on active treatment indicated that they did not know about available resources. The PATrev can easily be administered to families of children currently undergoing treatment as well as to families of survivors to assess risk of psychosocial distress. Evidence for the feasibility of administration of the PATrev has previously been established [2]; however, more research is needed to demonstrate the ability to implement screening as part of regular follow-up care. Routine screening as part of regular follow-up care has been well established within adult oncology in countries such as Canada and Australia, with patient-reported outcome measures administered either electronically or via paper survey at time of entry, all regular follow-up appointments, and all critical time points along the cancer trajectory [13,14]. Furthermore, as screening is only a first step, it would be important to identify available resources and referral options that correspond to screening outcomes. Nevertheless, knowing that high familial risk is associated with larger numbers of total symptoms and higher levels of burden due to these symptoms, physicians can follow up with families who score in higher risk categories with an assessment of symptom burden.
The study was not without limitations. First, the crosssectional nature of the study impedes the ability to draw causal conclusions based on the findings. Without longitudinal data, it is unknown if the reported symptoms were due to cancer and its treatment or if they were present before diagnosis. Moreover, it is unclear whether familial risk leads to greater symptom burden or vice versa. Although the PAT has been designed to determine premorbid familial functioning, it may be that the experience of greater symptom burden leads to the reporting of greater psychosocial distress. Likely, these relationships are bidirectional. Second, 53% of invited participants either declined to participate or did not complete the questionnaire. While non-responders did not differ from the study sample on demographic or disease characteristics recorded, it is possible that other variables such as symptom burden could have differed between groups. For example, individuals with lower symptom burden and lower family-level stressors may have been more likely to participate in the study. Third, the ESAS-r and the CPC do not yet have evidence of validity in younger populations. Since this study was conducted, pediatric symptom screening tools have been established for use in pediatric oncology (e.g., the SSPedi, O'Sullivan, Dupuis [52]). As such, future studies should utilize larger sample sizes with longitudinal data to increase the confidence that the PATrev administered at diagnosis can predict higher symptom burden at a later time during treatment and survivorship using screening tools for a pediatric setting. It may be beneficial to design studies that follow families through treatment and into survivorship to assess symptom burden over time. In addition, given that the ages of pediatric patients who completed their own survey spanned from 8 to 18, it is possible that some of the young children in our study may have completed the ESAS-r with the help of their parents, which may have influenced reporting. Unfortunately, we do not have documentation of if and when this occurred. Our sample was also limited by the number of families in the targeted and clinical PAT categories, and as such, our analyses may have been underpowered.
Finally, it is important to highlight that all participants in our study were fluent in English, the majority of our sample identified as white (79%), nearly all respondents were mothers (92%), and mean income of fathers was quite high. Taken together, these characteristics reflect a sample largely composed of families of high socioeconomic status. As such, our results may not be generalizable to other, more diverse populations. Related to this and of particular importance, there is a significant need for psychosocial screening tools that are appropriate for a pediatric health setting and can be completed by non-English-speaking patients and families. These families may include new immigrants, those who are unfamiliar with our healthcare system, or may have added challenges identifying and engaging in needed resources due to language barriers. As such, it is especially critical that we work to identify and support these families. Last, while inter-rater reliability on the PATrev is strong, reflecting a high level of correlation among scores between parents, some evidence suggests that more subjective scales on the PAT (e.g., Stress Reactions, Family Beliefs) may be less well aligned and therefore benefit from both parents' input [30,53]. Given that our sample was predominately mothers, and that mothers may experience greater distress at diagnosis and be more information-seeking than fathers, a more balanced study sample may have produced lower average scores on the PATrev and CPC, particularly for those families early in the cancer trajectory. However, we reiterate that mothers are more often the primary caregivers in hospital and therefore we believe our results reflect the true hospital setting.
In sum, we found that familial psychosocial risk, as measured by the PAT, was significantly associated with the number of symptoms reported and overall burden on the patient and their family, as measured by the ESAS-r and CPC, respectively. Institutions already using the PATrev to assess familial psychosocial risk should follow up with families who fall within the targeted and clinical risk categories to assess the patient's symptom burden and ensure appropriate interventions are provided.

Conflict of interest
The authors declare no competing interests.
Consent to participate Informed written consent was obtained from all participating parents and patients who were 18 years of age. Verbal assent was obtained from participating children and adolescents who were 8-18 years of age.
Consent for publication Not applicable.