Modelling Evolving Cancer Risk During Epidemiological Transition Using Economic Data.

Epidemiological projections sizing patient populations are fundamental to budget impact analyses and market forecasting. When disease risk evolves over time, as in the case of epidemiological transition, using historical estimates in epidemiological projection becomes unjustified. Incorporation of additional variables that model evolving risk may allow for more reliable forecasts. The hypothesis that gross domestic product per capita (GDP) is correlated with disease risk was tested for a variety of cancers using global epidemiological and economic datasets. Age-standardized incidence for 18 cancer sites across 188 countries was retrieved from the International Agency for Research against Cancer for the years 1993-1997. Corresponding country-specific GDP data was collected from the World Bank. For each site, correlation between GDP and incidence was measured using R2and linear co-efficient values. Risk is strongly correlated with GDP for all of the sites studied, with the exception of the stomach (R2=0; p>0.05). Correlation was strongest for those sites associated with diet/lifestyle factors prevalent in higher-income countries, namely: colorectal (R2=0.61; p<0.001), breast (R2=0.61; p<0.001) and lung (R2=0.42; p<0.001). Strong association was also seen for prostate (R2=0.47; p<0.001), although this may be explained by better case-detection in higher-income countries. Conversely, those cancer sites associated with infection showed a negative correlation, namely: cervix (R2=0.33; p<0.01), liver (R2=0.09; p<0.001) and oesophagus (R2=0.06; p<0.01). GDP is strongly associated with cancer risk and varies by organ site in a manner concordant with the evolving exposures to known pathogens that characterize epidemiological transition. On the assumption that such correlation is a marker for various causal relationships, and that robust economic methods underlie GDP forecasts, it is reasonable to conclude that these correlations could be used to make epidemiological projections more accurate than projections assuming constant disease risk.