Oudega et al's neat study of malignancy and deep vein thrombosis (DVT) in the September issue found that 4.4% of patients with a DVT went on to have a malignancy diagnosed in the next 2 years. We have performed a similar — albeit retrospective — study of 813 patients diagnosed with either a primary colorectal (n = 349), prostate (n = 217) or lung (n = 247) cancer during 1998–2002.1–4 Each case was matched with five controls without the cancer of interest for age, sex and doctor's surgery. We coded the records for 2 years before the diagnosis for both cases and controls. Only 10 of the 813 cases had had a DVT confirmed (nine in the last year), and a further nine of 4059 controls had a DVT (four in the last year). In addition, three cases (one in the last year) and five controls (two in the last year) had had a pulmonary embolism. Combining the two conditions gives a likelihood ratio for thrombo-embolic disease in the year before a new cancer of 8.5 (95% confidence interval = 3.1 to 22%). These thrombo-embolic conditions occurred in the age group at risk from cancer. The UK population over 40 years of age is approximately 30 million, and these people have approximately 250 000 new cancers, giving an annual risk of developing a new cancer of approximately 0.83%. Using Bayes' theorem, the risk of a new cancer being identified following thrombo-embolism can be estimated to be 6.6%, which is not dissimilar to Oudega's figure.
The key clinical decision is whether to investigate an apparently spontaneous thrombo-embolism for an underlying cancer. On the face of it, a risk of 4.4% (Oudega) or 6.6% (this study) appears to warrant investigation, at least by simple measures.