1 Introduction

Aboriginal and Torres Strait Islander cancer patients (respectively called here Aboriginal) face barriers to accessing cancer care, including historical trauma [1], lack of cultural understanding or respect shown by health care providers, lack of timely access to culturally safe services, remoteness to specialist health services and differing cultural beliefs about cancer [2, 3]. These factors may influence the timeliness and extent to which Aboriginal people present for diagnosis and treatment, contributing to poorer outcomes and higher mortality [4, 5]. In addition, Aboriginal people in rural/regional/remote areas face difficulties leaving their home, family and community to access city-based cancer services [1, 6-8]. Social and family obligations are competing priorities with cancer treatments, causing treatment interruptions or early cessation. Studies have identified that Aboriginal cancer patients often have more comorbidities at diagnosis than non-Aboriginal cancer patients [9] further complicating care. Similar poor outcomes have been identified in other indigenous populations including Inuit and Māori peoples [10, 11].

Previous studies in various Australian jurisdictions have identified that Aboriginal people had more advanced cancers at presentation [12] and were less likely than non-Aboriginal people to undergo surgery for lung [13-15], head and neck [16], breast [17], prostate [13] and cervical cancers [18]. However, most of these studies are of patients treated more than 15 years ago. Few studies have evaluated the utilisation of radiotherapy or surgery in Aboriginal and non-Aboriginal people [19, 20].

2 Materials and Methods

This was a retrospective analysis of a de-identified linked dataset. We have reported using similar methodology when examining the radiotherapy utilisation of all NSW cancer patients [21].

The study protocol was developed with multiple indigenous co-authors (KG, SA, DS) and was supported by the Tharawal Aboriginal Corporation and approved by the NSW Population and Health Services Research Ethics Committee (NSW Ethics 2019/ETH01657). The study protocol and the paper were approved by the Aboriginal Health and Medical Research Ethics Committee (NSW AH&MRC 1770/21). This study brought together researchers from health services, epidemiological and data linkage fields, along with Aboriginal health professionals and researchers to conduct one of the largest and most comprehensive population-based studies on surgery and radiotherapy utilisation of Australian Aboriginal cancer patients given that a third of the national Aboriginal population live in NSW [22].

Data were obtained on patients diagnosed with a notifiable cancer (includes all age groups, all invasive cancers except non-melanoma skin cancer) from NSW Cancer Registry for the period 2009–2018. These data were linked to Admitted Patient Data Collection (APDC), and the Outpatient Radiation Oncology Dataset (OROD). Staging data were based on Degree of Spread as this is the staging system used by the NSW Cancer Registry. This system describes the stage in 5 categories—local, regional to adjacent organs, regional to lymph nodes, metastatic and unknown. Chemotherapy data were unavailable and therefore not examined. All data were probabilistically linked by the NSW Centre for Health Record Linkage, providing a de-identified database. With respect to identifying those with Aboriginal status, the Enhanced Reporting of Aboriginality (ERA) was used [23]. ERA is a method that improves reporting on the health of Aboriginal people from administrative data collections using record linkage. Enhanced reporting relies on having linked records from the same person collected from independent sources, such as hospital admissions, emergency department presentations, births and deaths. Each record in the chain of linked records contributes to the weight of evidence as to whether a person is truly Aboriginal but may have been recorded as being non-Aboriginal or ‘unknown’ on some records. It is important to mention that this method doesn't simply recover ‘missing’ Aboriginality values, but it often uncovers Aboriginal people coded as non-Aboriginal in the reference dataset. To apply the ERA method, the Aboriginal status data for every individual patient in the NSW Cancer Registry dataset were linked to APDC, Emergency Department Data Collection, Cause of Death Unit Record File, Registry of Births, Deaths and Marriages records and OROD. It was decided to exclude patients who declined to report their Aboriginal status (n = 32) and patients missing Aboriginal status (n = 2974).

Socio-economic status was classified into five quintiles (Quintile-1 most disadvantaged to Quintile-5 least disadvantaged) using the Index of Relative Socioeconomic Disadvantage (IRSD) [24]. Road distance from patient residence to nearest facility (including interstate facilities) was calculated using buffered geocodes of patient residence [25] and Geographic Information System Software (ArcGIS Desktop: Release-10. Redlands, CA: Environmental Systems Research Institute). Non-cancer-related comorbidity index scores, based on the Quan-modification of the Charlson comorbidity index calculation (CCI), [26] were calculated using the ICD-10 diagnosis codes from the APDC dataset, basing index calculations on the prior 2 years of admissions for each patient. Cells with less than 5 counts in them are suppressed to protect privacy.

Oncologic surgery episode was arbitrarily defined as any record of an oncological operation occurring within one year of a cancer diagnosis. The rate of radiotherapy was defined as any record of receiving external beam radiotherapy within one year of diagnosis. NSW patients who lived nearer to the border with other states may have received treatment in other states (cross borders) without being recorded in the NSW dataset. In our previous study [25] we found that 10,008/108,064 (9.3%) of patients accessed treatment across a state border. Therefore, patients whose closest treatment facility was across the border (cross borders) were excluded from the surgical and radiotherapy analysis. In addition, for the surgical and radiotherapy utilisation analyses, we also excluded patients who had more than one cancer diagnosis during the study period as the surgery and radiotherapy code could not be attributed to one specific cancer. For the surgical utilisation calculations, we also removed patients who had cancers where surgery is predominantly in outpatient clinics where the APDC does not capture these surgeries (melanoma) and also cancers where surgery does not play a predominant role (haematological malignancy).

3 Statistical Analysis

Chi-square tests were used to analyse the univariate relationship between radiotherapy and surgery utilisation with factors Aboriginality status, age group, degree of spread, area of remoteness, IRSD, year of diagnosis, Quan comorbidity index, distance from radiotherapy treatment centre and sex. Analysis was performed separately for each tumour site. A multivariable logistic regression was then used to analyse these factors with radiotherapy and surgery utilisation. As effect modification was expected, the following interactions were considered as part of the multivariable model; ARIA by degree of spread; Aboriginal status by ARIA; Aboriginal status by comorbidity; Aboriginal status by degree of spread; Aboriginal status by IRSD; and Aboriginal status by year group of diagnosis. Interactions amongst ARIA, IRSD, degree of spread, comorbidity and sex were also examined. Interactions that were significant remained in the final multivariable model. Hosmer–Lemeshow goodness-of-fit test [27] was used to check model fit. Probit models were considered for models with poor model fit. The statistical analysis was conducted in SAS Enterprise Guide version 8.2 and SPSS version 27. Within Aboriginal and non-Aboriginal patients, univariate comparisons were made between patients diagnosed by year groups (2009–2011, 2012–2015, 2016–2018) for differences in degree of spread, radiotherapy and surgical utilisation rates.

4 Results

For the 10-year study period, 389,992 people were diagnosed with 414,980 cancers in NSW (Table 1). Of these cancer patients, 8970 (2.3%) identified as Aboriginal. For the radiotherapy utilisation rate calculations, there were 53,410 (13.7%) patients excluded from this part of the analysis as they either lived closer to an interstate cancer facility (26,712, 6.8%) or had more than one cancer diagnosis during the study period (21,809, 5.6%) or diagnosed with death certificate only (4889, 1.3%), leaving 336,582 (86.3%) patients being analysed for radiotherapy utilisation. In addition, for surgery utilisation, patients who were diagnosed with melanoma, haematopoietic cancers, unknown and other cancers (80,895, 20.7%) and patients who had no record in the Admitted Patient dataset (11,608, 3.0%) were excluded, leaving 244,079 (62.6%) patients being analysed for surgery utilisation.

5 Discussion

The univariate analysis in this study has identified that cancer in Aboriginal patients occurred at a younger age and were more advanced at presentation. Aboriginal patients were also likely to have significantly more comorbidities and live a greater distance from a cancer treatment centre. Over the duration of this study, the proportions of patients with localised spread have increased, and the proportions of distant metastasis has decreased for both Aboriginal and non-Aboriginal patients, with some tumours showing a halving of the more advanced stages of disease. The change in degree of cancer spread was more pronounced for the Aboriginal population. This suggests efforts to eliminate disadvantage and improve access to screening and other medical services are proving effective. This also supports other recent findings of a significant improvement in life expectancy amongst Aboriginal people in the Northern Territory [28]. However, these strategies have not closed the gap completely and so further efforts at improving screening rates and patient education remain important in further closing the gap. Further investigation into any modifiable causes for earlier onset of cancer in the Aboriginal population is warranted.

Based on crude percentages, Aboriginal patients had higher radiotherapy utilisation rates than non-Aboriginal patients. The reason for the higher radiotherapy utilisation rates may be partly related to the fact that they have a higher proportion of some cancers where radiotherapy plays a more significant role such as lung (12.9% vs. 8.9%), cervix (1.6% v 0.6%) and head and neck (5.7% vs. 3.5%). In addition, patients with more advanced disease at presentation are also more likely to be recommended for postoperative radiotherapy due to higher risk of locoregional recurrence or more likely to be recommended for curative radiotherapy as the extent of disease may preclude surgery, as has been shown when modelling radiotherapy needs in low- and low- to middle-income countries where advanced cancers prevail [29]. Patients with metastatic disease will also more likely receive palliative radiotherapy rather than surgery.

Our data show that Aboriginal and non-Aboriginal patients had higher radiotherapy and surgery utilisation when compared with other published series. Gibberd et al. [14] reported that 30.8% of Aboriginal people and 39.5% of non-Aboriginal in NSW (2001–2007) received surgery for nonmetastatic non–small-cell lung cancer (NSCLC), compared with 36.8% and 46.5%, respectively in this study. Hall et al. [13] found that 9.5% of all Aboriginal and 12.9% for non-Aboriginal lung cancer patients in Western Australia (1982–2001) had surgery compared to 14% and 18.8%, respectively in this study. Moore et al. [16] reported that 43% of Aboriginal people and 50% of non-Aboriginal patients diagnosed with head and neck cancer in Queensland (1998–2004) received radiotherapy compared to 57% and 51%, respectively in this study. Valery et al. [19] in a matched cohort in Queensland (1997–2002), found that Aboriginal patients were less likely to undergo surgery than non-Aboriginal (48% versus 58%) and were less likely to receive radiotherapy. This study shows that 51% of Aboriginal patients underwent surgery compared to 57% for non-Aboriginal patients and 30% of Aboriginal patients received radiotherapy compared to 26% for non-Aboriginal patients. A West Australian matched study [20] showed that Aboriginal people were less likely to receive radiotherapy (OR 0.70, 95% CI 0.56–0.89, p = 0.41) or surgery (OR 0.57, 95% CI 0.45–0.73, p = 0.53), than non-Aboriginal people, which is opposite to what we have found.

Even though remoteness from treatment facilities is a more frequent problem proportionally for Aboriginal patients, the radiotherapy utilisation for Aboriginal patients is higher than that of non-Aboriginal patients, suggesting that many remote patients can access radiotherapy. Receipt of radiotherapy may also be influenced by socioeconomic factors and the out-of-pocket cost and side effects of other treatment such as surgery [30].

After stratifying by tumour type, and accounting for patient factors, significant increases in the odds of radiotherapy for Aboriginal people compared with non-Aboriginal people were observed in prostate, lung and rectum (with no comorbidities) cancer, whilst decreased odds were observed in oesophageal (with 1 comorbidity) cancer. When analysing surgery utilisation, there were decreases in the odds of surgery for Aboriginal people in prostate (for patients diagnosed from 2016 to 2018), breast (in outer regional areas), lung, head and neck (in patients with 2 or more comorbidities), stomach and liver cancers. These associations are highly complex. We have not performed multiple comparisons as this study is exploratory and will require further examination to confirm any observed associations. This would be of particular interest in the findings specific to examples such as Aboriginal patients being more likely to receive radiotherapy as an alternative to surgery for oesophageal cancer when comorbidities exist and Aboriginal head and neck cancer patients being less likely to receive surgery as comorbidities rise. However, other observations did not appear to have any valid reason for the association (e.g., lesser chance of Aboriginal patients having surgery in outer regional centres but not in very remote centres). The fact that the goodness of fit was poor for a number of tumours suggest that other variables not within the current dataset may be partly responsible for variations in treatment utilisation.

This study has strengths and limitations. The strengths include the size of the NSW dataset and access to linked surgery and radiotherapy data. One possible limitation is that patients closer to the borders may have missing treatment data as they cross the border for treatment. We have managed this by excluding those patients from the specific analyses related to surgical and radiotherapy utilisation. Another limitation is that it is likely that identification of Aboriginal status remains incompletely captured in administrative databases, in addition to problems associated with incompleteness of data such as Degree of Spread and the lack of availability of systemic therapy data to complete the picture. Interestingly, in the NSW dataset used by Supramaniam et al. [17], only 1% of breast cancer patients identified as being Aboriginal compared with 2.3% in our study, suggesting better capture of Aboriginal status in recent administrative datasets. The difference may also be partly attributed to using ERA method to enhance the reporting of cancer outcomes of Aboriginal people in NSW [23]. It is possible that some of these data deficiencies create some bias as it would be more likely that stage and Aboriginal status are more accurately captured for those undergoing treatment compared to those diagnosed in the community and not referred. Another limitation includes that this study did not analyse overall or cancer-specific survival against whether surgery, radiotherapy or both were administered as this would need to be done separately for each tumour site and degree of spread and is therefore beyond the scope of this paper. Future studies on individual cancer types could address these issues. The recording of comorbidity status relies on data from the Admitted Patient Data Collection, therefore relying on inpatient episodes of care; this will underreport the comorbidity status of all patient groups as comorbidity presence will not be scored for patients who have not had an inpatient episode during that time (although those having inpatient surgery for their cancer will have comorbidity recorded).

6 Conclusion

This study provides a comprehensive assessment of Aboriginal tumour type and degree of spread, surgical and radiotherapy treatment utilisation in NSW and provides a baseline for further improvements in cancer care. Cancer degree of spread and the presence of comorbidities remains a greater issue for Aboriginal people along with earlier onset of cancer but is improving. Encouragingly, the degree of spread improved significantly throughout the study period suggesting strategies to improve screening, and earlier presentation has been of benefit. Access to radiotherapy improved significantly for Aboriginal patients in NSW during the past 10 years. However, differences in surgical and radiotherapy utilisation exist, partially explained by the greater degree of spread and presence of comorbidity in Aboriginal patients. Future work will assess survival in this cohort of patients.

Ethics Statement

The study protocol was approved by the NSW Population and Health Services Research Ethics Committee (NSW Ethics 2019/ETH01657) and the Aboriginal Health and Medical Research Ethics Committee (NSW AH&MRC 1770/21). The Aboriginal Health and Medical Research Ethics Committee also approved the final version of this paper prior to journal submission. There were no live participants in this study and therefore informed consent was not required.

Conflicts of Interest

The authors certify that they have no conflicts of interest with this manuscript. Professor Shalini Vinod is on the Editorial Board of JMIRO.