A novel method for identifying settings for well-motivated ecologic studies of cancer.

A low within-country variability and a large between-country variability in cancer incidence may indicate that ecologic factors are involved in the etiology of the disease. The aim of this study is to explore the within- and between-country variability of cancer incidence to motivate high-quality ecologic studies. We extracted age-standardized incidence rate estimates (world standard population) from 135 regions for the ten most frequent invasive cancers in Europe for non-Hispanic white populations from Cancer Incidence in Five Continents, Volume X. We fitted weighted multilevel Poisson regression models with random country effects for each cancer and sex. We estimated intraclass correlation coefficients (ICCs) and 95% confidence intervals (95% CIs). A high ICC indicates a low within- and a high between-country variability of rates. The two cancer sites with the highest ICC among men were prostate cancer (0.96, 95% CI: 0.92-0.99) and skin melanoma (0.78, 0.64-0.93). Among women, high ICCs were observed for lung cancer (0.84, 0.73-0.95) and breast cancer (0.80, 0.69-0.91). The two most prominent sex differences for ICC occurred for cancers of the head and neck (men: 0.70, 0.55-0.85, women: 0.19, 0.08-0.30) and breast cancer (men: 0.04, 0.01-0.07, women: 0.80, 0.69-0.91). ICCs were relatively low for pancreatic cancer (men: 0.23, 0.10-0.35; women: 0.13, 0.04-0.21) and leukemia (men: 0.12, 0.04-0.21; women: 0.08, 0.02-0.14). For cancers with high ICC for which systematic factors of the health care system, screening and diagnostic activities are not plausible explanations for between-country variations in incidence, cross-country sex-specific ecologic studies may be especially promising.