Tracing the tumors: navigating challenges in mapping cancer trends across twentieth-century Belgium

This article delves into the challenges and methodologies involved in mapping cancer trends across Belgium during the twentieth century. Disease mapping, pioneered by John Snow and Alfred Haviland, has been instrumental in understanding spatial disease patterns. However, comprehensive spatial cancer studies for Belgium are notably scarce before the 1960s, complicating comparisons over time. This research addresses the gap by mapping cancer mortality across the twentieth century, using various mortality indicators, geographical units, and visualization approaches together with a tailored historical geographic information system (HISGIS). We identify five major challenges and examine how these affected the analysis of mortality patterns over time: changing administrative boundaries, small sample sizes, evolving registration and classification practices, data availability and compatibility, and visual presentation over time. It appears that while historical boundaries mainly reveal local clusters, current boundaries and smoothed maps are more effective for detecting regional differences. Overall, our findings indicate important regional disparities in cancer mortality as over the course of the century the highest crude cancer death rates and percentages shifted from Wallonia (southern Belgium) to Flanders (northern Belgium). When adjusted for age and population size, high rates in the early to mid-twentieth century are observed in urban as well as in rural areas on both sides of the language border, with no apparent link to industrialization. From 1990s onward, notably higher rates are observed in the province of East Flanders and the southeastern Belgian districts near the French and German borders. Locally, we see early cancer clustering in large cities such as Antwerp, Liège and Ghent. However, at the end of the century, the clusters with the highest cancer rates show no clear association with urbanization. Lastly, we demonstrate that these changes in cancer mortality patterns are strongly influenced by diagnostic and registration practices, suggesting that more insight into these issues can lead to a better understanding of cancer trends over time.