First-Time Exploration into Breast Cancer Incidence and Mortality Across Small-Areas in Cyprus: Spatial Patterning and Associations with Rurality Indicators
Journal
International Journal of Epidemiology
Date Issued
2015
DOI
10.1093/ije/dyv096.423
Abstract
INTRODUCTION: At 77 per 100,000, female breast cancer incidence rates in Cyprus are consistent with other countries in Southern Europe. However, with complete absence of GIS from the Public Health arena, the extent of geographic inequalities in mortality and incidence across communities on the island, if any, remain unknown. We investigated the geographical patterning of breast cancer across small-areas in Cyprus and its association with rurality indicators.METHODS: Standardised Mortality and Incidence Ratios (SMRs/
SIRs) across 370 communities were calculated based on latest available registry data for period 2004–11 and 2003–08, respectively to ensure sufficient numbers. Bayesian hierarchical Poisson models with spatially unstructured and/or structured random effects were used to smooth maps and investigate the association with population density (rurality), population potential (remoteness from major population centers) and percentage of retired population. RESULTS: SMRs (range 0–6) and SIRs (range 0–4) were unreliable at such a small level of aggregation (median population 129, IQR: 47–416, 10% of areas > 1500).Nevertheless, up to two-fold differences remained across smoothed maps, with as much as 80% of the variation explained locally. Mortality and incidence demonstrated a similar geographic pattern with higher rates in and around metropolitan areas and lower rates in rural and mountainous areas.Population density (1.13 95% CI
¼ 1.04–1.23, per SD increase) and population potential (1.09 95% CI¼ 1.01–1.19) were both significantly
associated with increased mortality rates. Similar associations were observed with incidence rates which were also inversely associated with the percentage of retired population (0.92, 95%CI¼ 0.84–0.99). Interestingly, more than three-fold differences were observed in Mortality-to-Incidence ratios, suggesting differences in survival and/or registration quality.CONCLUSIONS: Even on a small island like Cyprus there is substantial small-area variation in breast cancer mortality and incidence rates, more likely to suggest urban-rural differences in reproductionrelated factors, even though the influence of other lifestyle and environmental factors cannot be ruled out.
SIRs) across 370 communities were calculated based on latest available registry data for period 2004–11 and 2003–08, respectively to ensure sufficient numbers. Bayesian hierarchical Poisson models with spatially unstructured and/or structured random effects were used to smooth maps and investigate the association with population density (rurality), population potential (remoteness from major population centers) and percentage of retired population. RESULTS: SMRs (range 0–6) and SIRs (range 0–4) were unreliable at such a small level of aggregation (median population 129, IQR: 47–416, 10% of areas > 1500).Nevertheless, up to two-fold differences remained across smoothed maps, with as much as 80% of the variation explained locally. Mortality and incidence demonstrated a similar geographic pattern with higher rates in and around metropolitan areas and lower rates in rural and mountainous areas.Population density (1.13 95% CI
¼ 1.04–1.23, per SD increase) and population potential (1.09 95% CI¼ 1.01–1.19) were both significantly
associated with increased mortality rates. Similar associations were observed with incidence rates which were also inversely associated with the percentage of retired population (0.92, 95%CI¼ 0.84–0.99). Interestingly, more than three-fold differences were observed in Mortality-to-Incidence ratios, suggesting differences in survival and/or registration quality.CONCLUSIONS: Even on a small island like Cyprus there is substantial small-area variation in breast cancer mortality and incidence rates, more likely to suggest urban-rural differences in reproductionrelated factors, even though the influence of other lifestyle and environmental factors cannot be ruled out.
Subjects