Level of Evidence for spatial variation
This was assessed using Tango’s Maximised Excess Events Test (MEET) global clustering test (Tango 2000), which has been shown to perform well across a variety of datasets (Kulldorff et al. 2006). The input data required for this test is the modelled counts (i.e. from the model results, the number of diagnoses or excess deaths per area) and the expected counts (as input into the Bayesian model). As it is expected to consider up to half the total area, our maximum distance of examination was 2000 km.
The p-value from Tango’s MEET was divided into 4 categories, consistent with previous analyses (Cramb et al. 2011),
- Strong (3 bars; p-value <0.01)
- Moderate (2 bars; p-value 0.01 to <0.05)
- Weak (1 bar; p-value 0.05 to <0.10)
- None (No bars; p-value 0.10+)
Tango’s MEET uses Monte Carlo methods, so estimates may have slight differences on subsequent runs. To ensure our categorisation was appropriate, we calculated Tango’s MEET three times, with the most conservative category being used. Only 1 cancer and sex combination had different categories for diagnoses, and 4 for excess deaths.
If there is one bar or less for the level of evidence, this indicates that overall there is NO meaningful evidence that the estimate for this cancer type and sex varies across the country. There may still be some individual areas that differ from the national average, but given the lack of evidence for overall variation, these individual differences should be interpreted with greater caution.
Cramb, S. M., K. L. Mengersen and P. D. Baade (2011). Developing the atlas of cancer in Queensland: methodological issues. International Journal of Health Geographics 10: 9.
Kulldorff, M., C. Song, D. Gregorio, H. Samociuk and L. DeChello (2006). Cancer map patterns: are they random or not? American Journal of Preventive Medicine 30 (2 Suppl): S37-S49.
Tango, T. (2000). A test for spatial disease clustering adjusted for multiple testing. Statistics in Medicine 19 (2): 191-204.