Background The associations between diabetes and mortality from a number of causes, such as cardiovascular and renal disease, are clearly understood. Concurrently, there is a growing body of evidence demonstrating an increase in cancer mortality among those with diabetes, although there is heterogeneity regarding the strength of the association and uncertainty around confounding by overweight/obesity. Around 2.9 million individuals in the UK are currently living with diabetes; this figure is expected to increase to 5 million by 2025, caused by rising rates of obesity and population ageing. The high prevalence of diabetes means that even a small increased risk of cancer mortality could equate to a large number of deaths among the diabetic population. This large cohort study utilised data from the Health Survey for England and the Scottish Health Survey, linked to mortality data, to explore overall and site-specific cancer mortality.

Methods Nationally-representative, cross-sectional data from the Health Survey for England (HSE: 1994-2008) and Scottish Health Survey (SHeS: 1995, 1998 and 2003) linked with mortality records up to the first quarter of 2011 and 2008 respectively. Odds ratios (OR) and 95% confidence intervals (CI) adjusted for age-group, sex and smoking status and additionally for BMI category were estimated for all and site-specific cancer mortality using logistic and multinomial logistic regression respectively. To allow for a consideration of the time sequence of any association between the two diseases, individuals with cancer at baseline were excluded from the analyses.

Results The study sample included 204,537 participants (6,258 with diabetes) and 5,562 cancer deaths. After adjustment for age, sex and smoking status, those with diabetes were at greater risk of dying from cancer compared with those without diabetes (OR 1.16 [1.02, 1.31]). This increase remained after adjustment also included BMI (1.19, 1.04-1.36). For site-specific cancer mortality, there were statistically significant increases for pancreatic (1.77 [1.16, 2.72], breast (1.62 [1.02, 2.65]) and lung (1.34 [1.05, 1.72]) cancer when adjustment included age, sex, smoking and BMI; as well as point estimate increased risk for cancers of the colorectum, bladder, stomach, lymphoid and haematopoietic, kidney and oesophagus.

Conclusion The high and rising prevalence of diabetes is of public health concern. Those with diabetes have increased overall cancer mortality and site-specific cancers of the pancreas, breast and lung, unaffected by measurements of overweight/obesity. Although some cases of pancreatic cancer may be reverse causality, health services for people with diabetes need to be aware of the increased cancer mortality found among this cohort.