The Utility of Multi-dimensional Symptom Assessment Scales for Children Treated in a Tertiary Paediatric Oncology Unit: a Prospective Survey-based Study


 Background: Children undergoing cancer-directed treatment experience distressing symptoms. Multi-dimensional patient-reported symptom assessment scales have been validated in children with cancer, but are not routinely used in clinical practice.Aim: To describe the symptom prevalence and burden for children receiving treatment in a paediatric oncology unit, as described by both children where possible, and their parents.Methods: Prospective survey-based study during which the Memorial Symptom Assessment Scale was administered to children and parents. Participants were offered the opportunity to complete the survey on multiple occasions. Demographic and clinical data were obtained from electronic medical records. Institutional ethics approval was granted (HREC37022A).Results: Forty-one children were recruited, aged 8 months to 18 years and 54% were female. In total, 54 parent surveys and 29 child surveys were completed. The vast majority of surveys (81%) were completed in the inpatient setting, and more than half within 10 days of chemotherapy. Haematological malignancies predominated. There was a median of 4.8 months between diagnosis and recruitment. Eleven children died after the study closed; no patients died during the study period. Children aged 10 – 18 reported an average of 15 symptoms. Symptom prevalence did not always correlate with distress.Conclusions: Exploration of the impact of a symptom, and not just its presence, is vital for patient-centred care and can be achieved using multi-dimensional symptom scales. Both the child and caregiver’s voices should be obtained where possible. Further studies are needed to explore how these scales can be used to identify distress and guide supportive care delivery.

care and can be achieved using multi-dimensional symptom scales. Both the child and caregiver's voices should be obtained where possible. Further studies are needed to explore how these scales can be used to identify distress and guide supportive care delivery.

Background
Although symptom self-report is generally considered gold standard, children's experiences have historically been assessed by proxy through parents, carers or clinicians, and only rarely through patientreported outcomes 1 . Our understanding that children undergoing oncology disease-directed therapy, particularly those approaching end-of-life, suffer signi cant symptom burden and poor health-related quality of life predominantly comes from bereaved parents and/or clinicians in retrospective studies [2][3][4] .
With a shift toward patient-reported outcomes, tools such as the Memorial Symptom Assessment Scale (MSAS) and Memorial Symptom Assessment Scale -Short Form (MSAS-SF) have been developed and validated 5,6 . These have now been adapted and used in children with cancer, including an 8-item version for children aged 7 to 12 years 7 and a 30 or 31-item version for children aged 10 to 18 years 8,9 . Patient-reported MSAS data support that children with cancer suffer from physical and psychological symptoms, including pain, lack of energy, nausea and feeling sad or nervous 1,7,8 . Symptoms may be related to underlying disease, interventions and/or treatments 10 . Multiple symptoms are usually experienced 1,7−9 . A strength of the multi-dimensional MSAS is that in addition to enquiring about the presence of a symptom, it assesses its impact on and importance to the patient by clarifying its frequency, severity and associated distress 5 . This enables clinicians to tailor their attention and treatment to the patient's most troubling symptoms.
The PediQUEST study conducted across three centres in the USA is the largest paediatric study to use the MSAS. The MSAS was administered 549 times in teenagers and 249 times in children aged 7 to 12 years.
In addition, a proxy version was completed 212 times by parents of children aged 2 to 6 years 1 . Pain (48%), fatigue (46%), drowsiness (39%) and irritability (37%) were the most prevalent symptoms, causing a high degree of distress. The study highlights that symptom burden is present and distressing both during cancer-directed treatment and at end-of-life 1 .
We sought to use the MSAS tools to explore the prevalence and impact of physical and psychological symptoms, for a cohort of children treated by the oncology unit in a tertiary Australasian hospital. We were interested in obtaining and comparing both parent and patient reports where possible.

Methods
The study was conducted in a tertiary paediatric hospital where the oncology unit provides care to children with haematological malignancies, as well as brain and solid tumours. The unit also performs bone marrow transplants for both malignant and non-malignant conditions. A prospective study was conducted between 31/05/2017 and 05/11/17, following institutional ethics approval (HREC37022A).
Parents and children were asked to complete symptom surveys, and supplementary demographic and clinical data were collected from electronic medical records. Although the MSAS 7-12 form has only been validated in children aged 7 to 12 years, we included children as young as 5 years of age providing they were able to understand and answer the questions and complete the survey with parental assistance. The MSAS 10-18 was administered to children aged 10-18 years of age. A proxy version of the 31-item MSAS 10-18 was administered to parents, regardless of the age of their child. This form was selected as it enquires about more symptoms than the MSAS 7-12 form. All parents were invited to participate, regardless of whether their child was eligible or chose to participate.
Participants were identi ed using electronic medical record patient lists. Eligibility criteria were deliberately broad, in order not to exclude patients who may have symptom burden. Patients were eligible if they were being treated by the oncology department as an inpatient or outpatient, with a carer available to provide consent. Written and spoken English language were required. Participants were offered the option to complete the forms electronically or on paper. Once recruited, children and parents were offered the opportunity to complete the survey at a minimum of weekly intervals until the close of the study.
The MSAS survey forms were entered into REDCap©, a secure web-based database housed in the institution server 11 . For those who elected to complete surveys electronically, surveys were distributed electronically using a REDCap© link, with responses stored in the REDCap© database. Data from paper surveys were transcribed into the REDCap© database. Additional patient demographic and clinical data collected from the electronic medical records were entered and stored in REDCap©.
The study was not intended to in uence standard care. However, given the potential for symptoms and distress to be unmasked, symptoms of concern were escalated to the oncology team who enacted referral to the pain, palliative care or psycho-oncology teams as needed.
Surveys were included if at least 25% of the survey was completed. Data were analysed in REDCap© and Excel, using descriptive statistics. Parent and child surveys were not matched, and data were summarised and presented separately for both groups.

Patient demographic and clinical characteristics
Forty-one children of 172 receiving therapy over the period described were recruited; parents of all children and 25 children completed surveys (12 children aged 5 to 9 years, and 13 children aged 10 to 18 years).
One child aged 5 years, and one child aged 6 years completed the MSAS 7-12, with the remainder aged between 7 and 9 years. Most children and parents completed the survey just once, but eight parents completed it between two and four times, and four children completed it between two and three times. In total, 54 parent surveys, 14 MSAS 7-12 surveys and 15 MSAS 10-18 surveys were completed.
Most children had a malignant condition, while three were undergoing bone marrow transplant for an immunological condition. Most surveys were completed in the inpatient setting in the context of receiving disease-targeted and/or supportive treatments. More than half the surveys were completed within 10 days of chemotherapy administration. Haematological malignancies (leukemia or lymphoma) were represented more frequently than solid organ and brain tumours. The duration from diagnosis to recruitment ranged from several days to 6.5 years, with a median of 4.8 months. Eleven (27%) children died between 4 months and 3.3 years of being recruited to the study. No child died during the study period. Table 1 summarises the demographic, clinical and treatment details for patients. Tables 2 and 3 summarise child self-reports and parent proxy-reports of symptom prevalence, frequency, severity and distress. Figure 1 illustrates the degree of distress from each symptom, with responses of 'quite a bit' or 'very much' to the question 'How much did it bother or distress them/you?' considered high distress.

Discussion
Our ndings are consistent with existing literature that children undergoing oncological treatment experience a high degree of suffering, related to several symptoms at any given time 1,7,8 . These symptoms are both physical and psychological, and can cause a high degree of distress.
We included children regardless of the timepoint in their diagnosis and treatment trajectory. Our inclusion criteria varied from other studies as we recruited children undergoing bone marrow transplantation for non-malignant conditions and children younger than 7 years if they were deemed able to understand and respond to the MSAS 7-12 form. The rationale for this was our hypothesis that neither of these groups are immune from suffering and both can express this suffering. We acknowledge that the MSAS 7-12 has not been validated in children younger than 7, however the younger children in our study appeared capable and keen to participate. While our numbers are too small to compare subgroups, we would support including both these groups in future studies.
Prevalence of some symptoms and mean number of symptoms experienced were higher in our study than that reported by others who have also explored symptoms for children with cancer who are not just at end of life 1,7,8 . For example, pain was reported by 79% of children and 89% of parents in our study, compared to 48% in the PediQUEST study 1 , 32% in the MSAS 7-12 validation study 7 , and 49% in the MSAS 10-18 study 8 . In our study, an average of 5 and 18 symptoms was reported for children aged 5 to 9 and 10-18 respectively, compared with 1.9 and 12.7 in the MSAS 7-12 and MSAS 10-18 validation studies. Parents reported a median of 14 symptoms, compared with 5 in the PediQUEST study (parent and children surveys).
The higher prevalence of symptoms in our study is likely related to the higher proportion of inpatient survey completion (81% in our study versus 11% in the PediQUEST study 1 ). In our study, 61% of children had received chemotherapy in the preceding 10 days, a factor associated with a higher symptom burden in the PediQUEST study 1 . In addition, more intensive therapies due to the inclusion of children early in their cancer treatment course and a higher proportion of children with haematological malignancies, are likely to have contributed to higher symptom burden and report Parent and child reports of symptom presence and distress were usually consistent in our study. For example, sadness was reported in 52% of child surveys and 59% of adult surveys. When present, sadness caused high distress 33% and 28% of the time according to child and parent reports respectively. Similarly, pain was reported in 79% and 89% of child and parent surveys respectively, with resulting high distress 43% and 48% of the time when present. Parent and child reports of distress did not always correlate however, for example, children seemed more distressed by vomiting and dizziness, and less distressed by headaches than parents may have recognised. This speaks to the importance of obtaining and hearing the child's voice, and not solely relying on proxy reports.
We feel that our most pertinent study nding is the importance of multi-dimensional symptom assessment. Unlike the PediQUEST study ndings that when a symptom was present it usually caused high distress 1 , in our study the presence of a symptom didn't necessarily correlate with distress. Symptoms may be present with high frequency but cause little distress and therefore require little clinical attention; conversely, infrequent symptoms such as dysphagia may cause disproportionately high distress and warrant aggressive proactive symptom management.
We propose that it is crucial to re ect on how we assess, consider and use the four dimensions of the MSAS for a given child. It is possible that untreated frequent or severe symptoms may over time lead to more distress on subsequent assessment, causing disproportionate negative impact on quality of life in these vulnerable patients that could be ameliorated with a more active expectant approach. Future studies could focus on how multi-dimensional symptom assessment tools might identify predictors of distress, which could in turn guide intensity of symptom control or promote protocols for timely referral to supportive care teams.
Ultimately, patient-centred care requires exploration of the patient's voice and agenda. Knowledge that a symptom is present is less useful than understanding its impact, and this is the very reason for the development of multi-dimensional symptom assessment scales such as the MSAS versions.
Understanding the degree and source of distress in a patient's own words can guide symptom management so that time and resources are invested in a way that optimally addresses patients' priorities.

Strengths and Limitations
Page 13/15 The numbers in our study are low, and don't allow comparison across subgroups like disease type and episode of care setting. Unfortunately, we did not collect data about the incidence of distress being escalated to the oncology team, but this would have been useful and should be included in future studies. Ultimately, the prospective design, and inclusion of both the children and parent voices are strengths of the study, and our ndings contribute to emerging literature that includes the child's voice. In addition, the high proportion of inpatients in our study complements and adds to existing literature which has predominantly been based in the outpatient setting.

Conclusion
Our study supports the value and importance of multi-dimensional symptom assessment. It is not simply enough to know whether a symptom is present, but rather its impact. This cannot be presumed by clinicians. Tools such as the MSAS can be used to obtain self-reports from children, and these can be complemented by parent reports. Future studies are needed to explore how child-friendly multidimensional symptom assessment tools may enhance child-centred care, identify distress and guide care delivery.

Declarations
Funding: The study was supported by the Victorian Palliative Medicine Training Program through a Palliative Medicine Fellowship granted to Naomi T Katz.

Figure 1
Degree of distress. For each symptom, the top bar represents self-report by children, and the bottom bar represents proxy report by parents.