Symptoms and suffering at the end of life in children with cancer: an Australian perspective

Our study was conducted at the Royal Children’s Hospital, Melbourne. Parents of children who had died of cancer over the period 1996–2004 were interviewed between February 2004 and August 2006. Parents were eligible for our study if they were English-speaking, had no known major psychiatric illness, and had had a child who died at least 1 year before study enrolment. Permission to contact the parents was also required from the child’s primary oncologist and the social worker responsible for the family’s care.

Parents who met eligibility criteria received a letter inviting them to participate in our study, along with a reply-paid “opt-in” or “opt-out” postcard. If parents returned an opt-in postcard, they were further contacted by telephone to arrange an interview.

Once parents had provided written informed consent, trained clinical psychologists conducted structured, in-person interviews and provided self-report questionnaires. The content of the interview and questionnaires was developed by one of us (J W) and has previously been reported.9 Some measures and language were adjusted for use in the Australian population. A summary of items in the structured interview and self-report questionnaire is given in Box 1. In most cases, one parent was interviewed, but occasionally, both parents asked to be present during the interview. The self-report questionnaire was completed once for each participating family.

In cases in which questions were left unanswered or information was unavailable, the elements in question were excluded from our analysis. Variables were graded with the use of five-point Likert scales. Statistical analysis was performed using Fisher’s exact probability test, the χ2 test for independence and the Mann–Whitney U test for independent observation.

Of 193 families who were identified as eligible for our study, 144 could be located and 100 agreed to participate. Parents from 96 families (response rate, 50%) were ultimately interviewed. Compared with all patients treated at the Royal Children’s Hospital between 1996 and 2004, members of our patient cohort did not differ significantly in age at diagnosis and type of diagnosis (data not shown). Interviews took place a mean of 4.5 years (SD, 2.1 years) after the child’s death, and lasted for an average of 144 minutes (range, 45–420 minutes). Of the 96 parents who were interviewed, 89 subsequently returned self-report questionnaires.

Recent guidelines for paediatric palliative care in Australia emphasise that high-quality medical care should be readily available and should focus on meeting the needs of not only patients but also their families and communities.13 To date, it has not been established whether the care of children dying of cancer in Australia achieves these goals. Our study cohort, recruited from the largest of the eight specialist paediatric oncology units in Australia, appears to be representative in age and tumour types of the childhood cancer population. Although data on parental factors such as age, ethnicity, educational level and religion were not available for all families whose children were cared for between 1996 and 2004, we believe that, allowing for the exclusion of non-English-speaking parents, the sample was representative.

It was pleasing to find that most families were offered out-of-hospital palliative care services. Because of the relative rarity of death from medical causes in childhood and the large geographical areas covered, this step often necessitates collaboration with adult-based palliative care services and local and regional health services. Without this cooperation, adequate services could not be provided. Overall, families appeared to be satisfied with this component of their child’s palliative care.

A key finding of our study was the lower level of aggressive, cancer-directed treatment during the end-of-life period compared with the level in a similar cohort in the US (Box 5). Whether this reflects a more enlightened approach to palliative care or a reduced capacity to offer novel and experimental treatments is unclear. Certainly, with regard to children in our study who received cancer-directed therapy in the end-of-life period, the primary reasons for most of the treatment were to relieve suffering and extend life. It is noteworthy that these treatments appeared to be ineffective, even for symptom control. It was not possible in our study to determine whether patients who received cancer-directed treatments had medical problems directly related to their tumour burden.

As in other international studies, pain, fatigue and poor appetite were the three most commonly reported symptoms during the end-of-life period. The rates of these symptoms in our Australian population appeared to be very similar to those reported previously in US, European and Japanese populations (Box 5). It was pleasing that, in our study, a substantial proportion of children who experienced significant suffering from symptoms received treatment. However, it was disappointing that current treatments appear inadequate, with success rates of under 50% for all symptoms and under 20% for symptoms such as fatigue, poor appetite, constipation and diarrhoea. Given the positive impact that specialist paediatric palliative care services have had overseas,14 it is hoped that their recent introduction in the state of Victoria will improve symptom control and lessen suffering for children dying of cancer.

A striking difference between our study and previously published international studies was the low proportion of patients who died in hospital and, more specifically, in intensive care units (Box 5). The one study that reported a higher proportion of children dying at home8 was skewed by the elimination of children who died from treatment-related complications and those who were treated for less than 6 months. As in other international studies, nearly all parents in our study indicated a preference for their child to die at home, so the high rate of deaths at home was a positive finding. Furthermore, our finding that most symptoms were just as likely to be treated at home as in hospital, and that treatment was no less likely to be successful, supports the feasibility of providing adequate care at home.15,16 In addition, in view of previously reported evidence of better psychological outcomes in parents whose child dies at home,15,17,18 these findings are reassuring. The universal provision of bereavement services may also be a vital component of recovery from such a traumatic event.

There are a number of shortcomings of our study that restrict interpretation of the results. Our findings, based on a single hospital, may not accurately reflect the experiences in other areas of Australia, particularly the more remote areas of central, northern and western Australia. In addition, the possibility of selection bias cannot be discounted. In particular, the exclusion of non-English-speaking parents and those with psychiatric conditions (which constituted about 10% of parents) may have underestimated important problems relating to communication and access to services, as well as issues of cultural diversity. Recall bias is another factor that cannot be excluded. The information collected was retrospective, relying on accurate recollection by parents some years after the death of their child. Although the significance of events surrounding the death of a child means that parents are not likely to forget them easily, it is possible that their interpretation may have been affected by their own subsequent bereavement experiences. Furthermore, parental impressions of their child’s symptoms and suffering may not accurately reflect the child’s actual experience. Data from adult studies suggest that proxies’ perceptions of patients’ pain alters (downwards) with time19 and is influenced by the caregiver burden20 — factors that were not accounted for in our study.

Relatively high rates of death at home and low rates of heroic medical interventions suggest a realistic approach to care of children with cancer at the end of life. However, many Australian children who die of cancer suffer from unresolved symptoms. Greater attention should be paid to palliative care for these children.

3 Proportion of children who experienced specific symptoms compared with the proportion who, according to parental reports, suffered “a lot” or “a great deal” as a result of these symptoms in the last month of life

4 Proportion of children treated for various symptoms and reported success rates of treatment in the last month of life