Original ResearchSpatial variation in cancer incidence and survival over time across Queensland, Australia☆
Introduction
With an estimated 14.1 million cancer cases diagnosed globally in 2012 (Ferlay et al., 2013), the impact of cancer is felt worldwide. With wide variation in cancer incidence and survival not only between countries (Ferlay et al., 2013, Allemani et al., 2015), but also within countries (Siegel et al., 2016, Australian Institute of Health and Welfare, 2014), there are important disparities depending on where people live.
Quantifying and understanding the extent of small-area variation in cancer incidence and survival is becoming increasingly important, with government and other policy makers needing to make evidence-based decisions on resource allocation and planning interventions to address any known disparities. Consistent with this, an increasing number of small-area cancer atlases have been published, including those in Australia (Public Health Information Development Unit, 2012, Cramb et al., 2011, Bois et al., 2007), USA (National Cancer Institute, 2015) and the UK (Quinn et al., 2005).
There is great variation in the statistical approaches used in these atlases. These methods range from direct estimation of area-specific age-standardised incidence rates (Public Health Information Development Unit, 2012) through to modelling approaches incorporating smoothing such as Poisson kriging (Goovaerts, 2005), empirical Bayes (Benach et al., 2001) or fully Bayesian methods (Bois et al., 2007). While each method has various benefits and disadvantages, some form of smoothing is often preferred to reduce spurious variation associated with very small area-specific counts (Best et al., 2005).
We have previously demonstrated the extent of small area variation in incidence and survival across the state of Queensland, Australia for around 20 of the most commonly diagnosed cancers (Cramb et al., 2011). This cancer atlas highlighted the extent of the geographical variability in incidence across Queensland, and how the survival outcomes were poorer in many of the more remote areas of the state.
However, it was unclear how these geographical patterns in cancer incidence and survival have changed over time. Since the ability to understand whether the spatial patterns are changing over time and in what direction is critical to guide efforts to reduce existing disparities, we have examined how the geographical variation in cancer incidence and survival in Queensland has changed over time for the five most commonly diagnosed cancers.
Section snippets
Methods
Ethical approval to conduct this study was obtained from the Darling Downs Hospital and Health Service Human Research Ethics Committee (HREC/15/QTDD/57).
Incidence
During 2005–2012, there were almost 112,000 new diagnoses of the five most common cancers among our study cohort (Table 1). This was an increase of almost 31,000 new cases diagnosed compared to 1997–2004. Across total Queensland, incidence rates were lower for colorectal cancer in the later time period but higher for prostate cancer (Table 1). The age-standardised incidence rates for breast, lung and melanoma remained reasonably similar over the two time periods, with overlapping 95% confidence
Discussion
In Queensland, the risk of a cancer diagnosis or cancer-related death varies by residential location. This is true for all five of the most commonly diagnosed cancers, whether diagnosed during 1997–2004, or during 2005–2012. The general improvement in survival over most areas between the two time periods means that geographical disparities have remained. This suggests that it is not sufficient to just ensure that diagnostic and management strategies are equivalent across the state, rather
Conflict of interest
The authors declare no potential conflicts of interest. All funding sources have been acknowledged.
References (46)
- et al.
Global surveillance of cancer survival 1995–2009: Analysis of individual data for 25 676 887 patients from 279 population-based registries in 67 countries (CONCORD-2)
Lancet
(2015) - et al.
The international epidemiology of lung cancer: latest trends, disparities, and tumor characteristics.
J Thorac Oncol
(2016) - et al.
Spatio-temporal relative survival of breast and colorectal cancer in Queensland, Australia 2001–2011
Spat Spatiotemporal Epidemiol
(2016) - et al.
Reliability of collecting colorectal cancer stage information from pathology reports and general practitioners in Queensland.
Aust N Z J Public Health
(2008) - et al.
Use of space–time models to investigate the stability of patterns of disease
Environ Health Perspect
(2008) - et al.
Progress in adjuvant chemotherapy for breast cancer: An overview
BMC Med
(2015) - Australian Bureau of Statistics. Unit record mortality data for Queensland by state of usual residence, 1982–2005....
- Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS): Volume 1 – Main structure and...
- Australian Coordinating Registry. Registries of births, deaths and marriages, the coroners and the national coronial...
- Australian Institute of Health and Welfare. Cancer in Australia: an overview, 2014. Canberra: AIHW;...
Urban-rural differences in prostate cancer outcomes in Australia: What has changed?
Med J Aust
Geographic remoteness, area-level socio-economic disadvantage and advanced breast cancer: A cross-sectional, multilevel study
J Epidemiol Commun Health
Atlas of mortality in small areas in Spain (1987–1995)
Bayesian analysis of space–time variation in disease risk
Stat Med
A comparison of Bayesian spatial models for disease mapping
Stat Methods Med Res
Space–time mixture modelling of public health data
Stat Med
Cancer maps for New South Wales 1998 to 2002
Atlas of Cancer in Queensland: Geographical variation in incidence and survival, 1998 to 2007
A flexible parametric approach to examining spatial variation in relative survival
Stat Med
Identifiability and convergence issues for Markov chain Monte Carlo fitting of spatial models
Stat Med
Cited by (23)
Access to diagnostic imaging and incidental detection of differentiated thyroid cancer in Ontario: A population-based retrospective cohort study
2022, Spatial and Spatio-temporal EpidemiologyCitation Excerpt :The Geweke diagnostic is a test for significant difference in means for the first (10%) and latter (50%) iterations of the chains divided by the (time-series) standard error of the difference obtained for each parameter (Geweke, 1991). For the Geweke diagnostic, we used a p-value <0.01 over all parameters as evidence of convergence (Cramb et al., 2017). Accuracy was assessed using Monte Carlo error as a percent of the posterior standard deviation using the <5% rule of thumb (Spiegelhalter et al., 2014).
Artificial intelligence for melanoma diagnosis: How can we deliver on the promise?
2019, Annals of OncologyArtificial intelligence for melanoma diagnosis: How can we deliver on the promise?
2018, Annals of OncologyExtended excess hazard models for spatially dependent survival data
2024, Statistical Methods in Medical ResearchBayesian spatiotemporal modelling for disease mapping: an application to preeclampsia and gestational diabetes in Florida, United States
2023, Environmental Science and Pollution Research
- ☆
KLM acknowledges support from the ARC Centre of Excellence in Mathematical and Statistical Frontiers. The views expressed in this paper are those of the authors and not of any funding body.