DISCUSSION
Our study provides evidence pointing to inequity in outcome for Aboriginal children and for children from remote and very remote localities with acute leukaemia in Australia. There was no association found between MRD-based response and either ethnicity or remoteness of residence, suggesting the lower survival among Aboriginal children and those from remote areas is independent of this measure.
It is widely known that cancer is more likely to be fatal in Indigenous adults compared to non-Indigenous counterparts3. There have been few studies performed in Indigenous children, particularly within a specific cancer type. A study by Valery et al. revealed a reduced 5-year survival of 75% for Indigenous children with any type of cancer in Australia compared to 82.3% for non-Indigenous children3. Furthermore, Indigenous children had 1.36 times greater risk of death within 5 years of diagnosis after adjustments for rurality of residence, SES, cancer diagnosis and year of diagnosis3,13.
Previous studies worldwide have shown higher overall cancer incidence and mortality in minority ethnic groups, such as Maori and Pacific Islander people in New Zealand (NZ)4,5. Black children with acute leukaemia and those from other ethnic or racial minority groups in the United States of America have inferior survival compared to Caucasian children6-8. Measures have been taken to address the inequalities in cancer outcome and incidence between Maori and non-Maori people, such as introducing specific services for Maori people4. Similar measures are required to address the gap in survival rates for Aboriginal children with leukaemia in Australia.
It has been estimated that 76% of Indigenous children under 15 years of age in NT reside in remote and very remote parts of Australia compared to 13% in SA and 36% in WA14. However, data on childhood cancer incidence by remoteness of residence and SES are scarce15. Consistent with previous studies2, our study has shown Aboriginal children were less likely to live in major cities. Applying the ARIA remoteness index, this showed that Aboriginal children were more likely to live in communities further away from medical centres (Table 1). Furthermore, no Aboriginal children in our study lived in the least disadvantaged areas. Whilst these numbers were small, this trend warrants further research.
A study by Youlden et al. revealed that children living in more urban areas in Australia had higher cancer rates overall, particularly for leukaemia and lymphoma15. It was postulated that the reduced incidence of childhood cancers in remote and very remote areas could potentially be related to reduced incidence of cancer in Indigenous residents or missed diagnosis due to non-presentation. The incidence rates of childhood cancer were slightly higher among children in the least socioeconomically disadvantaged areas in Australia compared to those in the most disadvantaged areas15. Whilst this study did not yield population data for SA, NT and WA, analysing rates of leukaemia between areas of remoteness and different SES should be considered for future studies.
Worldwide, there has been inconclusive evidence that geographic residence impacts on overall survival9 although previous Australian studies have indicated poorer rates of survival for children with cancer who live in more isolated parts9. One would expect that with increasing remoteness comes reduced access to healthcare and increased challenges associated with diagnosis, treatment and long-term follow-up15. It is not known whether the reduced survival in remote areas is due to the larger relative percentage of Indigenous people9.
Generally, an inferior outcome has been observed in children with cancer from lower or middle-income countries; however, results from low socioeconomic demographics in high-income countries are less consistent16. A study from Queensland in the early 1980s reported a significant difference in incidence and survival among children with ALL across different SES, despite similar treatments being received9,17. This trend was also observed in Maori and Pacific people in NZ5. Further research is clearly warranted to evaluate causes of poorer outcomes in children from remote areas, lower socioeconomic advantage and from various ethnicities.
An overrepresentation of AML patients in Aboriginal children diagnosed and treated in SA (WCH) between 1978 and 1998 was first reported in an audit by Bartle et al. in 19991. These patients were less likely to speak English and more likely to come from traditional Aboriginal communities1. This was confirmed 12 years later by Rotte et al. , who again found relatively more cases of AML diagnosed among Aboriginal children at WCH between 1997 and 20112. The latest cohort of patients from the same geographical area (SA/NT) reported here has also shown a higher proportion of AML cases and again without obvious morphological phenotype or cytogenetic abnormality1. The over-representation of AML in Aboriginal children in SA/NT observed in three successive cohorts since 19781-2 warrants further investigation and interestingly was not observed here in Aboriginal children from WA. Our study also seems to indicate lower overall survival of children with AML compared to current Australian Childhood Cancer Statistics (69.8% vs. 77.6%)18, although this may be because the latter only includes children diagnosed before the age of 15 rather than 18.
Despite small patient numbers, there was a significant difference observed in clinical trial enrolment and follow-up among Aboriginal children. Previous studies have reported lower participation rates in clinical trials2, which is likely to contribute to a less optimal outcome given less rigorous adherence to diagnostic and therapeutic guidelines2. Cultural and language barriers can also make discussions regarding randomised trial and informed consent challenging2.
Consistent with previous studies2, we found a lower rate of follow-up among Aboriginal children, with a statistically significant increase in those lost to follow-up. As discussed previously, this may reflect inequality in healthcare access for those in rural and remote communities. Improvements could be made by expediting use of tele-conferencing to support a shared care model and utilisation of rural outreach oncology nurses to enhance communication with local healthcare workers2. Furthermore, simplification of care, community involvement, peer support and decentralisation of care may also improve service delivery for Aboriginal and remote people19.
Strengths from this study are the high level of data quality obtained from the contributing centres. Limitations include the relatively small cohort size and subsequently small number of Aboriginal children included, leading to reduced statistical power. Furthermore, given only patients were included from SA, NT and WA, there is not complete population coverage of data. This could potentially introduce bias, however this risk is small given most children diagnosed with leukaemia in these states/territories are referred to the major paediatric oncology treating hospitals included in the study. Indigenous status was defined by self-assessment at the involved centres and the accuracy of these data is unknown. Nevertheless, any misclassification of Indigenous status is likely to be random with respect to the factors examined here, thus is unlikely to have biased our results.