Background: There is controversy about whether diabetes mellitus is a risk factor for pancreatic cancer or an epiphenomenon of the cancer. The present study aims to determine if long-term diabetes is a risk factor for pancreatic cancer.
Methods: The study undertook to determine the prevalence of diabetes among three matched (age/gender) patient groups (pancreatic cancer (PaC), colorectal cancer (CRC), and fracture neck of femur (NOF)) at the date of diagnosis of cancer or fracture as well as 1 and 5 years prior to this. A retrospective review of the medical records of the three groups of patients was undertaken. Patients identified with PaC in the period July 1994 to February 1998 were age (+/- 5 years)- and gender-matched to patients identified in the same time period with NOF and with CRC. The data were then analysed using McNemar's test for discordant pairs.
Results: Over a 44-month period 116 patients with PaC were identified of which 24% had diabetes at the time of diagnosis of their malignancy (NOF, 8%; CRC, 9.5%). There was a statistically significant difference (PaC and NOF, P < 0.01; PaC and CRC, P < 0.01). For a duration of diabetes of > 5 years the prevalence of diabetes fell to 7.8% in the PaC group, to 6% in the NOF group and to 6.9% in the CRC group, with no significant difference between the groups.
Conclusion: There is no increase in the prevalence of long-standing diabetes mellitus in patients with PaC compared to age- and gender-matched controls with NOF and CRC. The relationship of PaC and diabetes may be an epiphenomenon, rather than diabetes being a risk factor for pancreatic malignancy.