Fever of Unknown Origin and Incidence of Cancer.

Abstract

Diagnostic tools for determining causes of fever of unknown origin (FUO) have improved over time. We examined if cancer incidence among these patients changed over a 20-year period. Population-based cohort study using nationwide Danish registries. We identified individuals diagnosed with FUO (1998-2017) to quantify their excess risk of cancer compared with the general population. Follow-up for cancer started 1 month after FUO. We computed absolute risks and standardized incidence ratios (SIRs) of cancer, and mortality rate ratios adjusted for age, sex, and cancer stage. Among 6620 patients with FUO (46.9% male; median age: 39 years), 343 were diagnosed with cancer (median follow-up: 6.5 years). The 1- to <12-month risk was 1.2%, and the SIR was 2.3 (95% CI, 1.8-2.9). The increased 1- to <12-month SIR was mainly due to an excess of Hodgkin lymphoma (SIR = 41.7) non-Hodgkin lymphoma (SIR = 16.1), myelodysplastic syndrome/chronic myeloproliferative diseases (SIR = 6.0), lower gastrointestinal cancer (SIR = 3.3), and urinary tract cancer (SIR = 2.9). Beyond 1-year follow-up, malignant melanoma, hepatobiliary tract/pancreatic cancer, and brain/CNS/eye cancer were diagnosed more often than expected. The 1- to <12-month cancer SIR attenuated over time, and for the 2013-2017 period we found no excess risk. Patients diagnosed with cancer ≤1 year after FUO had similar mortality to cancer patients without this diagnosis. Patients with FUO have a higher 1- <12-month cancer SIR; thereafter, the incidence for most cancers equals that of the general population. Decreasing SIRs over time suggests improvements in the initial diagnostic workup for FUO.