Cancer Council SA commissioned the production of An Atlas of Cancer in South Australia by the Public Health Information Development Unit of the University of Adelaide, to identify areas where primary and secondary preventive programs might be better directed to improve cancer outcomes in rural South Australia. The Atlas illustrated the benefit of using data from multiple sources together to highlight inequalities in cancer risk in regional and remote compared with metropolitan areas. Differences in survival were also presented, including important ones requiring immediate attention, but in most instances the differences were small and suggestive of reasonably equitable access to critical services. Based on Atlas data, we have made recommendations regarding cancer-control initiatives needed to reduce inequalities in cancer risk and outcomes in South Australia, particularly in high risk populations such as Aboriginal and Torres Strait Islander people. Acquisition of data for the Atlas proved to be a slow and difficult process. There was good support from many data custodians, but also major barriers, including some that proved insurmountable within the two-year period of the project. Major differences existed in data access and approval processes, and in resource availability to extract data. There is a pressing need to improve data governance arrangements to increase access to existing Australian data to guide cancer-control initiatives.
A quarter of the South Australian population resides outside of metropolitan Adelaide and the local Adelaide hills.1 South Australia has a highly centralised distribution of cancer treatment facilities, with all radiotherapy services and most chemotherapy services provided in metropolitan Adelaide. A key aspect of the Cancer Council SA Strategic Plan for 2012-2015 is to “identify areas in secondary prevention … where investment by Cancer Council SA would be productive to improve cancer outcomes in rural South Australia.” To advance this strategy, and to guide primary prevention and therapeutic initiatives, Cancer Council SA commissioned an atlas from the Public Health Information Development Unit of the University of Adelaide to show inequalities between metropolitan and country residents of South Australia in cancer risk factors, cancer incidence and outcomes.
The final report, An Atlas of Cancer in South Australia (referred to in this paper as ‘the Atlas’), provides an overview of patterns of cancer and cancer risk factors, and includes a focus on rural and remote communities, residents of areas of socioeconomic disadvantage and Aboriginal and Torres Strait Islander people.2
The purpose of this paper is: (1) to outline key observations from the Atlas and demonstrate the benefit of using combination data from multiple sources, such as health surveys, screening registers, vaccination registries, cancer registries and treatment databases, for population assessment; and (2) to report on the data retrieval process for the Atlas. Data from different sources are frequently reported independently in Australia, but they have complementary qualities and when viewed together, they can provide a better overview of service needs than when considered independently.
The methodology included a review of scientific literature, both journal publications and ‘grey literature’ reports that were known to Cancer Council SA and Public Health Information Development Unit. The purpose was to describe what already was known about geographic differences in cancer incidence and mortality, risk-factor prevalence, screening uptake and case survivals in rural South Australia, low socio-economic areas and Aboriginal people. Methodological details are provided in the Atlas.2 Risk data and cancer rates were age-standardised, using conventional direct standardisation methods and the Australian 2001 reference population, to facilitate comparisons.
The Atlas used data from population surveys, the cancer registry, screening registers and administrative databases to describe geographic differences in risk factor prevalence, screening participation and outcomes, cancer incidence and mortality, descriptors of cancer stage at diagnosis, five-year relative survivals by statistical local area, remoteness of residence from Adelaide, and extent of socioeconomic disadvantage of residential area.2
Cancer is responsible for about 28% of South Australian deaths and is one of the largest contributors to disease-related morbidity and mortality.4 About one in three South Australians diagnosed with cancer resides in a country region rather than metropolitan Adelaide.5
There are inequalities in cancer risk, cancer rates and cancer survival for regional and remote populations compared with metropolitan Adelaide residents, although the Atlas showed the differences in survival to be generally quite small and suggestive of reasonable state-wide access to critical services.2 Cancer risk factors more common in regional and remote than metropolitan areas included excess sun exposure, tobacco smoking, high-risk alcohol consumption and excess body weight.2 Areas of socioeconomic disadvantage and long travelling distances to screening and specialist cancer services were also evident in country South Australia.2 Treatment for cancer is complex and multidisciplinary in nature, with a need for specialist centres, which can complicate access for many regional and remote residents where population numbers are insufficient to support specialist centres.2
The incidence of cancer types with lower prospects of survival, such as cancers of the lung, stomach, pharynx/oesophagus, liver, pancreas and unknown organ site, is elevated in Aboriginal and Torres Strait Islander people, which would contribute to geographic differences seen in South Australia. However, the recording of Aboriginal and Torres Strait Islander status in South Australian and other cancer registries has been unreliable, such that differences by ethnicity remain poorly defined.2, 6
Compared with the metropolitan Adelaide population, residents of remote populations have elevated rates of tobacco smoking, alcohol consumption, obesity, low physical activity and excess sun exposure.2 Data from the National Health Survey (2007-2008) were used to identify patterns in tobacco smoking, alcohol consumption, body weight and physical activity. SA Health Omnibus Study data were used to provide information on excess sun exposure.2
Tobacco smoking and alcohol consumption
The prevalence of smoking was higher in metropolitan areas of high socioeconomic disadvantage and non-metropolitan areas than in other metropolitan areas, with statistically significant differences (p<0.05) in age-standardised rates of greater than 10% reported.2 Of males aged 18 years and over living in non-metropolitan areas, survey data indicated that 26.5% were current smokers, compared with 23.6% in metropolitan Adelaide (Rate Ratio (RR) = 1.12).2 The corresponding figures for females were 20.8% and 16.5% respectively (RR = 1.26).2 The percentages of adults reporting alcohol consumption at dangerous levels were relatively low, averaging 4.7% overall, but the percentage for regional and remote areas of 5.9% was 1.4 times higher than the corresponding 4.3% for metropolitan Adelaide.2
Body weight and physical activity
The geographical distribution of overweight and obesity differed between males and females.2 The proportions of males found to be overweight or obese were similar in Adelaide and non-metropolitan areas, with 37.6% classified as overweight and 16.8% as obese. Atlas maps for males for metropolitan areas showed distinct patterns, with overweight concentrating in areas of higher socioeconomic (SES) status and obesity concentrating in lower SES areas.
The estimated proportions of females who were overweight or obese were also similar in metropolitan and non-metropolitan areas. The obesity rate in females (17.4%) was comparable to the male rate, whereas the proportion of women who were overweight (26.5%) was about a third lower than for males. Atlas maps showed a concentration of overweight and obesity in females in lower SES areas. The distribution of overweight and obesity in females was more variable than in males, demonstrating a weaker association with socioeconomic disadvantage than seen for males.2
Consistent with the distribution of excess body weight, the proportion of residents aged 15 years and over who were physically inactive was higher in country areas than in metropolitan Adelaide.2 High levels of physical inactivity were also strongly associated with socioeconomic disadvantage.2
Excess sun exposure
The annual South Australian Health Omnibus Survey was used to describe excess sun exposure. Survey participants were asked if they had been sunburnt in the previous summer and whether they were regularly compliant with the five sun protection behaviours (i.e. wearing a hat, SPF30+ sunscreen, clothes that covered all of the arms and legs, and sunglasses, and using shade).
There was little variation reported between metropolitan and regional areas regarding sunburn experience and engagement in sun protection behaviours, but there was a difference in sun exposure for populations in very remote areas.2 Residents of very remote areas were reported to be three times more compliant with recommended sun protection behaviours, but nonetheless reported a sunburn rate more than 50% higher than for other South Australians.2 It is possible that very remote residents, while taking additional measures to mitigate increased risk of excess sun exposure in country living, may need to intensify these measures to nullify their increased risk.2
There are three national population-based cancer screening programs in Australia, namely, BreastScreen, the National Cervical Screening Program, and the National Bowel Cancer Screening Program.
The BreastScreen SA program was introduced in 1991 as a joint Commonwealth/SA Government initiative.7 Approximately two thirds of 50-69 year-old SA women were attending for biennial screening around 2000, but there was a reduction of about 12% in screening in absolute terms in both non-metropolitan and metropolitan areas from 2002 to 2009.2 This reflected a growth in size of the screening cohort that outstripped service capacity.7
The Atlas showed an association between lower screening participation and lower SES area.2 Also, the decline in screening participation was largest for women in the lowest SES areas, which widened the gap in participation between the lowest and highest SES areas from 9% to 13% in absolute terms.2 A weak positive correlation between screening participation and breast cancer incidence was also evident, which may reflect increases in the detection of breast cancer cases from screening, and associated lead-time and related effects, rather than true elevations in incidence.2,10
The National Cervical Screening program was also introduced in 1991.2 Screening participation rates were mapped in the Atlas. Participation by women aged 20-69 years declined by about 6% in absolute terms across South Australia between 2002 and 2009 to 60.7%. Rates were similar for Adelaide and non-metropolitan areas in general, but appreciably lower for women in very remote regions. As with breast cancer screening, the decline in participation was most marked among residents of the most disadvantaged areas.2
The National Bowel Cancer Screening Program was introduced in 2006, whereby invitations for Faecal Occult Blood Test (FOBT) screening were sent to Australians turning 55 and 65 years. The program subsequently was extended to include those turning 50 years, and then 60 years, and further extensions are planned. In 2010, participation rates were similar for Adelaide and non-metropolitan areas, but lower in very remote areas. Females had a higher participation rate than males. There was a pronounced, positive correlation of participation with higher SES area both for males (r=0.64) and females (r=0.71).2
South Australian cancer incidence (all cancer types combined) is high by world standards, but comparable to that for Australia overall and other developed countries, including the United States.2 The age-standardised incidence rate for South Australia in 2008 was 601 invasive cancers (all types) per 100,000 for males and 404 per 100,000 for females.2,4Incidence rates increased between 2003 and 2008 for both men and women. Incidence rates increased for those cancers targeted in population-based screening and allied early detection activity, including breast, bowel and prostate cancers, and for those likely to be more detectable through advances in ultrasound imaging, MRI, PET and other diagnostic technology, such as thyroid and kidney cancers.4
For all cancers combined, a 1.1% annual increase in incidence occurred for males from 2004-2008, largely attributable to increases in prostate cancer detection, while a 0.6% annual increase occurred for females, largely due to small increases in bowel and lung cancer diagnoses.2,4 The Atlas showed that increases in cancer incidence were evenly distributed across socioeconomic quintiles, both in metropolitan and non-metropolitan areas.2
The Atlas showed rural residents to have similar age-standardised incidence rates to metropolitan residents for all types of cancer except lip cancer, where incidence rates in more remote areas were nearly twice those in metropolitan areas (figure 1).2 This observation is consistent with data published by the SA Cancer Registry since 1980.4,5Lip cancers mostly occur on the outer vermilion border of the lower lip and their higher incidence in non-metropolitan areas is attributed to excess sun exposure.5 The incidence of lip cancer is often high in populations with a high incidence of non-melanoma skin cancers (NMSC), including basal and squamous cell carcinomas, but NMSC is not recorded in the SA Cancer Registry (or most other Australian registries).4,8-11 Because of the high incidence of lip cancer in rural areas, it is expected that the rates of NMSC would also be high in those areas. While NMSC and lip cancers rarely cause death, they place a large burden on the health system.4,10
Conversely, the Atlas showed that the incidence of invasive female breast cancer was about 8% lower in more remote than metropolitan areas.2 This difference is thought to be due in part to differences in reproductive history, with younger age at first pregnancy and higher parity occurring more frequently in more remote areas.2 Differences in use of hormone replacement therapy may also contribute, but the reasons for these differences require further investigation.
In metropolitan Adelaide, breast cancer incidence is highest for women in the highest SES quintile (least disadvantaged) and lowest for women in the lowest SES quintile (most disadvantaged). The incidence of female breast cancer increased by approximately 40% in metropolitan and non-metropolitan areas between 1986 and 2008.2 This coincided with the introduction of the BreastScreen program, but the extent to which the increase reflected lead time effects of screening, a potential contribution from over-diagnosis, changes in diagnostic practices, and real increases in incidence due to changes in reproductive and related lifestyle factors is unclear.7 It is evident that mammography screening of 50-69 year-olds has reduced breast cancer mortality in Australia and South Australia specifically, by around 40%.2, 7
In contrast to the increase in the incidence for all invasive cancers collectively between 1986 and 2008, rates of lung cancer in men aged 20 years and over declined by nearly 10% in metropolitan Adelaide.2 This reduction was not seen in non-metropolitan areas where the incidence increased marginally, and with particularly high incidence rates occurring in very remote areas. The incidence of lung cancer among women increased substantially, by about 48% over the same period in metropolitan and non-metropolitan areas, with the largest increase seen in the lowest SES quintile.2
Aboriginal and Torres Strait Islander people comprise approximately 2% of the South Australian population, with over 50% residing in rural and remote areas. Overall, Aboriginal and Torres Strait Islander people experience higher incidence rates than other South Australians of more life-threatening cancer types, including cancers of the lung, stomach, pharynx/oesophagus, liver, pancreas and unknown organ sites.6 Cervical cancer incidence is also elevated in Aboriginal and Torres Strait Islander people.2 This cancer pattern points to high levels of modifiable risk factors in the Aboriginal population, including smoking, excess alcohol consumption and poor diet, as well as higher levels of HPV, hepatitis B and helicobacter pylori infections, which could be targeted in cancer control interventions.2,10 It is important though that the broader social determinants of these cancers are addressed, since these would underpin most of the risk-factor elevations.
The SA cancer registry does not routinely collect stage of progression of cancer at diagnosis, although tumour size and thickness are collected for breast cancers and melanomas respectively.
The Atlas showed that the percentage of invasive breast cancers classified as large (i.e. greater than 30mm in diameter) was higher in more remote areas than in Adelaide for females aged 40-49 years and 70 years and over, but not for females in the BreastScreen SA target screening group aged 50-69 years.2 This may reflect effects of BreastScreen SA in reducing socio-demographic inequality in its target population. Collation and analysis of quality data and more intensive research is required to determine the effectiveness of screening in various age groups outside the recommended target. There is a need to focus on the promotion of early breast cancer detection in the more remote areas, especially for Aboriginal and Torres Strait Islander females, who are diagnosed at a more advanced stage and experience lower survival from breast cancer than other South Australian females.2,10
Similarly, the Atlas showed that the percentage of invasive melanomas classified as thick (i.e. thickness greater than 1.5mm) at time of diagnosis was higher in non-metropolitan regions. Thickness was found to be larger in more remote areas in all age groups, with a statistically significant difference applying in the 50-59 and 60-69 year old age groups.2 These findings, along with increased lip cancer rates, highlight the need for priority to be given to rural and remote areas in early skin cancer detection programs and alongside the promotion of sun protection.
Advances in screening, diagnostic technologies, treatment and service delivery are leading to improved cancer survival in Australia and other developed countries.2Significant all-cancer survival gains have been realised over the past two decades, although an unacceptably high case fatality persists in the Aboriginal and Torres Strait Islander populations.6
Survival from cancer in South Australia is high by world standards, with a five-year survival for all cancers collectively of 61%.2 This is comparable to the leading United States figure of 63%, and is much higher than the 48% survival in Europe.2
The Atlas showed that case survivals for all cancers collectively tended to be lower for residents of more remote areas than Adelaide residents. The difference, although reaching statistical significance (p<0.05), was small (i.e. a five-year survival of 62% compared with 64%).2 It could reflect more advanced stages at diagnosis for some cancers, as well as differences in treatment management and timely access to services. Further research is needed to explore survival differences across South Australia and reasons for differences. Although the survival differences reported in the Atlas were generally small, attention should be given to strengthening regional service access and referral patterns so that these differences do not increase. The numbers of health care practitioners per capita decrease with degree of remoteness, limiting opportunities for country residents to gain professional health advice and referrals for specialised oncology care.2
There are inadequate data to accurately assess cancer outcomes in Aboriginal and Torres Strait Islander people due to incomplete documentation of ethnicity status.6 Nonetheless, available data suggest that Aboriginal and Torres Strait Islander residents have lower cancer survival than other South Australians for all cancers combined, but possibly not lung cancer, for which there may be a small survival advantage (figure 2).6 Survival from cancer is particularly low in Aboriginal communities in very remote areas (e.g. the Far North and Far West).6 Aboriginal and Torres Strait Islander people experience more lethal types of cancers.6 They are also diagnosed at more advanced stages, have higher rates of accompanying diabetes and other co-morbidities, and often experience poorer access to specialist treatment services.6
Australia has comprehensive cancer-related data, but we were aware at the beginning of the Atlas project of access barriers and were uncertain whether these barriers would prevent sufficient data access in the two-year project time frame.3 Many data custodians were involved across multiple jurisdictions and data custodian organisations. Examples include the Australian Bureau of Statistics, Commonwealth Department of Health, Commonwealth Department of Human Services, Australian Institute of Health and Welfare, and South Australian Department of Health and Ageing. The data sources covered cancer and cancer screening registries administered by government departments and contracting agencies, plus university-based and other data holdings. Because we wished to test the utility of using multiple data sources together for Cancer Council service planning and evaluation, we wished to see whether retrieving these data was straightforward.
Data retrieval proved to be a slow and difficult process involving many mail and telephone requests to data custodian organisations across jurisdictions. It was clear that many custodians did their best and were successful in providing the requested data expeditiously, but uncertainties and delays arose with others due to privacy concerns or because of difficult and variable approval processes for data release, some of which were cumbersome with multijurisdictional approval points. Also, in-house resources were often limited for data extraction. Generally, there was a lack of policy on charging and other aspects that generated uncertainty. As a consequence, data extraction took around two years overall, and at the end of the project, some key data were still not available and had to be excluded from the Atlas. Data held by Commonwealth and other interstate authorities were generally more difficult to retrieve for this South Australian project than data held locally in South Australia.
Ideally, the research group would like to have commissioned construction of a de-identified linked data set that incorporated cancer registry data, hospital inpatient co-morbidity and treatment data, public and private radiotherapy data, Medicare Benefits Schedule/Pharmaceutical Benefits Scheme data, cancer screening and Human Papilloma Virus vaccination registry data, and data from population-based cohorts on relevant environmental and behavioural exposures.3 Australia is well-practised in applying privacy-protecting data linkage methodologies to do this type of work, having gained extensive experience with the WA Data Linkage System and more recently with similar linkage systems in Queensland, NSW and SA/NT.3 Examples of many successful projects using registries, administrative data and data linkage were presented at a Menzies Foundation Data Linkage Workshop.3 However, these projects generally did not cover all cancers and broad aspects as risk factors, incidence, mortality and case survival. The data linkage option was not a realistic proposition for the Atlas due to the need to gain approvals from an insurmountably large number of research ethics committees and data-custodian organisations, given the time constraints. Barriers to data linkage have been described in a recent report of the Menzies Foundation Data Linkage Workshop and they were known to us at the start of this project.3 All these barriers appear to be resolvable, but a significant national commitment would be needed across all jurisdictions to gain agreement and implement more manageable data governance arrangements.”
Timely and routine access to multiple population-level data sources, such as those listed which include national as well as South Australian sources, would provide valuable evidence to Cancer Councils and other cancer-control agencies, and would facilitate better targeting of cancer control. Identification of inequalities across the cancer spectrum, from risk factor exposure through to mortality, and data showing how these patterns were changing over time in light of health promotion activities, population screening, changes to health care services delivery and policy changes, would assist Cancer Councils and other agencies to ensure that their resources were best utilised for cancer control.
The Atlas has provided an overview of differences in cancer risk, cancer rates and cancer outcomes across the South Australia population and shown the benefits of using combined data sources for this purpose. The Atlas was successful in bringing together data for residents of rural and remote areas, those living in areas of high socioeconomic disadvantage, and Aboriginal and Torres Strait Islander populations.
Importantly, the Atlas has provided much needed evidence on inequalities in risk and cancer outcomes for these populations. The information available was much broader than generally provided in routine reports from single data sources. The Atlas offers a breadth of data that enable more informed decision making and identification of targets for intervention (e.g. for health promotion, screening, treatment and health professional liaison). The data point to areas needing further research to identify high-risk groups for future programs. Baseline data have also been produced for assessing the effectiveness of such programs. The data can inform advocacy strategies at a local, regional and state level.
The Atlas demonstrates the value of using combination data sources to better understand cancer risk-factor prevalence, cancer rates and outcomes across the population. Data of much greater power could be assembled however, through linkage of data along the cancer trajectory for planning and evaluating preventive, screening, diagnostic, treatment and end-of-life care, and community support services. Opportunities to use existing Australian data to better inform the work of Cancer Councils and other cancer-control agencies clearly exist, but data governance arrangements need to be greatly simplified if there are to be timely data outputs. In particular, better access to nationally held data is required.
In the meantime, based on the project experience, the authors make the following recommendations in relation to South Australia.
Rural and remote communities and areas of most social disadvantage
Promote and advance the prevention, early detection and treatment of cancer in rural and remote areas:
Aboriginal and Torres Strait Islander populations
Work with Aboriginal partners in developing and proposing targeted initiatives to prevent cancer, and treat and provide support for cancer patients of Aboriginal and Torres Strait Islander background:
Improve ongoing data access:
The authors would like to acknowledge the significant work of the Public Health Information Development Unit at the University of Adelaide, upon which this article is based. In particular, the authors would like to thank Associate Professor John Glover, Dr Diana Hetzel and Ms Su Gruszin.