At first sight, today’s social inequalities in cancer mortality may seem self-evident. Disadvantaged groups die more often from cancer, while better-off groups tend to live longer. But history tells a more complicated story. 

To understand long-term patterns in cancer mortality, we need to look beyond cancer as a single disease and consider cancer by type and the determinants of cancer mortality.   

First, incorporating cancer by type is easier said than done as site-specific cancer data only become available from the early 1960s onward. Once these data appear, however, clear contrasts emerge. Stomach cancer, for example, is common in the early postwar years but falls sharply thereafter. This decline is linked to long-term improvements in living conditions: better food preservation through refrigeration, less reliance on salted and smoked foods, improved nutrition, and falling infection rates with Helicobacter pylori. Lung cancer follows the opposite trajectory. From the 1960s to the early 1990s, mortality rose steeply, especially among men, reflecting the delayed effects of widespread smoking combined with occupational and environmental exposures. 

These diverging trends are often described as part of the cancer transition: a long-term shift from cancers closely associated with infections and poor living conditions (stomach, liver, or cervical cancer) towards cancers linked to lifestyle, ageing, and non-infectious processes (e.g. lung, breast, colorectal, and prostate cancer). In most Western countries, this transition unfolded gradually between the 1930s and the 1960s. 

Interestingly, local research suggests that this transition did not occur everywhere at the same pace. In the city of Antwerp, the shift towards non-infectious, lifestyle-related cancers appears to have taken place already in the late 1930s, roughly two decades earlier than at the national level. There, recorded mortality from non-infectious cancers was initially higher among more advantaged groups. The reason was not better protection due to knowledge, but the opposite: wealthier groups adopted risky behaviours, such as smoking, earlier. By contrast, cancers linked to infections and poor living conditions showed little social differentiation, largely because no one yet knew how to prevent them. 

Secondly, direct data on historical risk behaviours such as smoking or alcohol consumption are largely absent. To fill this gap, we rely on socioeconomic indicators such as literacy, income, and occupational structure of the population. Doing so reveals another striking finding: early in the twentieth century, social gradients in cancer mortality were weak, inconsistent, or even absent. This means that who died from cancer depended less on social position than on chance or local registration practices. Without effective prevention, diagnosis, or treatment, social advantage offered little protection. 

That began to change after the Second World War. From the late 1940s onward, socioeconomic inequalities in cancer mortality became more pronounced, particularly among men. This suggest gendered access to information, prevention, and care. Between 1948 and 1970, the direction of the social gradient reversed, in line with international evidence and our own hypothesis: as knowledge improves, social resources increasingly translated into survival advantages. 

One caveat remains crucial. Our most telling results are related to registration quality. Areas with high proportions of deaths recorded as “old age” or “ill-defined” consistently show lower recorded cancer mortality. This pattern reveals substantial under-registration. In other words, part of the observed inequalities are based on who was diagnosed and counted, and who was not. These practices differed enormously between regions. 

Finally, a methodological note: our analyses are conducted at the district level. At the municipal level, cancer deaths were often too few to produce stable rates. Aggregating to districts allows meaningful comparisons while still capturing long-term spatial variation.