Fever of unknown origin (FUO): a 7-year clinical experience, etiological distribution, and diagnostic approaches.

backgroundFever of unknown origin (FUO) remains a significant diagnostic challenge. Changes in patient populations and diagnostic technologies may shift the spectrum of underlying etiologies. This study aimed to examine recent FUO cases at a single tertiary center to identify changes in etiology compared with previous reports and between the pre- and post-COVID-19 pandemic periods, assess the diagnostic value of laboratory and imaging findings, and examine FUO characteristics in underrepresented subpopulations, including people living with HIV (human immunodefficiency virus), those on immunosuppressive therapy, and individuals with recurrent fever, with the goal of informing a center-specific diagnostic approach.

methodsA retrospective analysis was performed on 100 patients hospitalized with FUO between 2017 and 2024, classified according to Durack and Street’s criteria. Subgroups included classical (n = 85), HIV-associated (n = 13), neutropenic (n = 1), and nosocomial FUO (n = 1). Clinical, laboratory, imaging, and biopsy findings were reviewed, and subgroup-specific analyses were conducted.

resultsAmong 100 FUO patients (mean age 45.1 years; 52 males), recurrent fever was observed in 14 patients within the classical FUO group. The median fever duration at admission was 6 weeks. The average time to diagnosis was 14 days, and the mean hospital stay was 21 days. Final diagnoses were: collagen vascular diseases (CVD, 39%), infections (19%), malignancies (17%), miscellaneous (12%), and undiagnosed (13%). Vasculitis predominated among CVDs (25.6%), extrapulmonary tuberculosis (TB) among infections (36.8%), lymphomas among malignancies (52.9%) and, subacute thyroiditis among “miscellaneous” category (25%). In subgroup analysis, people living with HIV had no CVD, while no infections were found in recurrent fever cases. Among people living with HIV, infections (46%) and malignancies (38%) were most common. Immunosuppressed patients had diverse etiologies, including drug-induced pneumonitis and disease flares. CRP ≥140 mg/L, low hemoglobin, and anti-HIV positivity were independent predictors of malignancy; ferritin ≥875 µg/L was significantly associated (p = 0.01). PET-CT (Positron Emission Tomography – Computed Tomography) had the highest diagnostic yield, especially in infections and malignancies (p = 0.004). Microbiological studies were diagnostic in > 50% of infectious cases; autoantibodies aided diagnosis in ~50% of CVDs. During follow-up, 11 patients died, including 7 from the malignancy group.

conclusionOur findings indicate a shift toward non-infectious inflammatory causes of FUO, with CVD surpassing infections. PET-CT and targeted biopsies are valuable diagnostic tools, particularly in suspected malignancies or infections. Subgroup-specific approaches are essential, especially in people living with HIV and immunosuppressed patients. Despite advances, a proportion of FUO cases remain undiagnosed, underscoring the need for stepwise, individualized diagnostic strategies. CLINICAL TRIAL NUMBER: Not applicable.