Approach

Fever of unknown origin (FUO) is defined according to the qualitative FUO criteria, first described in 1997 (‘classic FUO’, Table 1).[3][4]​​ Subsequent investigators have modified these criteria to reflect contemporary practice changes in medicine, given the increased outpatient evaluation of patients and in an era of advances in molecular and radiological technologies.[4][27][28]

​​​​​[Figure caption and citation for the preceding image starts]: Contemporary qualitative criteria for classic FUO​Adapted from Wright WF et al. Am J Med. 2022 Feb;135(2):173-78; used with permission [Citation ends].com.bmj.content.model.assessment.Caption@3803e391

Two retrospective studies from Japan and Italy reported that the prevalence of classic FUO ranged from 2.0% to 2.9%, based on admission diagnosis.[29][30]​​​​ Despite revised definitions and the introduction of improved serological, laboratory, and imaging technologies, eliciting the aetiology of FUO remains a challenge. The most important lesson learned from the ‘classic FUO’ concept is that, in many instances, information obtained from the history and physical examination has the highest yield towards securing a diagnosis.

Potential diagnostic clues (PDCs) are derived from the history, physical examination, laboratory studies, or imaging findings. Clinicians must be aware that an individual PDC is often misleading. In one prospective study by de Kleijn et al, PDCs led to a diagnosis in 62% of patients.[3] However, a prospective study by Bleeker-Rovers et al, reported that an average of 15 PDCs were noted per patient, of which 19% contributed to the final diagnosis.[31] This study found that 81% of PDCs were misleading, a percentage substantially higher than that reported by de Kleijn et al, in which approximately 48% of PDCs were misleading.[3][31]​ Despite the poor outcomes of chasing clues, the authors emphasise that searching for clues remains a cornerstone of FUO evaluation.

Many algorithms have been proposed to assist with evaluation of FUO.​[31][32][33][34]​​​​​​​ Although useful for general guidance, ordering numerous tests without thoughtful development of a differential diagnosis-directing selection may delay diagnosis, yield misleading results and raise costs. Among 19 prospective trials (n=2627 )in a systematic review and meta-analysis among FUO series that varied across and within WHO geographical regions, the use of a structured diagnostic protocol was not significantly associated with higher odds of yielding a diagnosis compared with non-structured protocols in aggregate (odds ratio [OR], 0.98; 95% CI, 0.65 to 1.49).[35] Researchers have, therefore, proposed an alternative approach to the diagnostic evaluation of patients with FUO based on 3 steps: 1) comprehensive history and exam, 2) ordering minimal obligatory studies, 3) repeat comprehensive history and exam if no PDCs or patients remains ill without a diagnosis other than FUO.[4]

Regular clinical evaluations throughout the course of FUO can elicit new findings.[5]​​ For example, the appearance of a new-onset rash or skin lesion will influence/change the diagnostic approach.

History

It is axiomatic that a comprehensive history is a foundation for evaluating complex illnesses such as FUO. The findings from the history can be significant in determining the choice of laboratory and radiographical investigations. Particular attention should be given to eliciting risk factors for tuberculosis, the patient’s residence, recent travel, exposure to pets, other animals or unusual foods, work environment, and recent contact with people exhibiting similar symptoms or potentially carrying diseases.[24][32]

The family history should enquire as to the patient’s ethnic background and possible hereditary causes of fever, such as familial Mediterranean fever. The past medical history must cover any prior episodes of FUO and any previously diagnosed conditions, such as lymphoma, rheumatic fever, as well as any intra-abdominal disorders, complications or reactivation of which might account for the source of fever. Rare disorders (e.g., Hyper IgD syndrome, Rosai-Dorfman disease, or cryopyrin-associated periodic syndromes) are more common in recurrent FUO than in non-recurrent cases and are more likely to have been undiagnosed.[36]​ Finally, a complete list of the patient's medications and duration of use must be obtained so that each may be evaluated as a potential source of drug-induced fever.

Physical examination

In the investigation of FUO, several aspects of the physical examination should be accorded closer scrutiny than generally given when evaluating other illnesses.[37]​ Key physical abnormalities in patients with FUO are often subtle and require repeated examinations to be performed (Table 2).[37] Examples include the nodular or weakly pulsatile temporal artery of temporal arteritis; the tell-tale oral ulcers of disseminated histoplasmosis, systemic lupus, inflammatory bowel disease or Behçet's syndrome; the choroid granuloma or epididymal nodule of extrapulmonary tuberculosis; the testicular nodule of polyarteritis nodosa; and the vague rectal fluctuance of a perirectal abscess. All regional lymph node groups and any persistent or recurrent rashes should be evaluated. The diagnostic yield of the physical examination alone in evaluating FUO has not been adequately studied to render specific numbers.

[Figure caption and citation for the preceding image starts]: Examples of physical findings with particular significance in patients with fever of unknown originAdapted by BMJ Knowledge Centre from Cunha BA et al. Infect Dis Clin North Am. 2007 Dec;21(4):1137-87, xi [Citation ends].com.bmj.content.model.assessment.Caption@5099223b

Laboratory investigations

Basic laboratory tests are indicated in the initial evaluation of all patients, including full blood count with differential, urea/creatinine, liver function tests, erythrocyte sedimentation rate, and urine and blood cultures.[24]​​​

Subsequent laboratory studies, including additional cultures obtained from affected areas, should be guided by any abnormal laboratory or clinical findings.

Examination of blood smears is occasionally diagnostic, especially in patients with tick-borne or louse-borne relapsing fevers. Paradoxically, the advent of enhanced microbial culture systems has had less impact on the proportion of successful diagnoses than anticipated. Modern culture systems have become so proficient at recovering fastidious bacteria, some mycobacteria, and fungi from the blood, that they provide the diagnosis promptly before the time required for FUO.

Imaging

Chest radiography should be done in the initial round of tests in all patients.​​[24]​ Magnetic resonance imaging (MRI) is indicated in the initial investigation of osteomyelitis alongside basic laboratory workup.[38]

Of the commonly employed diagnostic imaging methods, one multi-centre study reported diagnostic sensitivities of 60% for plain chest x-ray, 82% for chest computed tomography (CT), 86% for an abdominal ultrasound, and 92% for abdominal CT.[31]​ False-negative CT studies have occasionally been encountered, even in cases of abscesses or infiltrating malignancies in solid organs, due to distortions of normal anatomy, small lesion size, or failure to use both oral and intravenous contrast agents. CT pulmonary angiograms can diagnose pulmonary emboli. 

Other imaging and biomarker analysis techniques, such as 18F-fluorodeoxyglucose–positron emission tomography (18FDG-PET) and diffusion-weighted magnetic resonance imaging (MRI), provide opportunities to identify focal inflammatory or infectious processes, mainly when CT scans are unrewarding, and fevers are otherwise without apparent source.

Diffusion-weighted MRI is beneficial for evaluating the central nervous system and, within the abdomen, the spleen and lymph nodes. It has been reported that MRI of the aortic arch and proximal cervical arteries improved the diagnosis of large vessel vasculitis by 20%.[38]

Several meta-analyses have examined the diagnostic yield of 18FDG-PET integrated with CT (18FDG-PET/CT), reporting sensitivities of 85% to 98% and specificities of 52% to 85%.[28][39][40][41][42][43]​​​​​​​​​​​ The overall diagnostic yield of 18FDG-PET/CT appears to be greater than 50%.[28] In a subgroup analysis of five studies that addressed the contribution of 18FDG-PET/CT compared with conventional CT, the diagnostic yield was 32%.[40]​ Additionally, negative 18FDG PET-CT results are associated with a higher rate of spontaneous resolution of fever.[41]

FDG-PET/CT has been reported to have a higher sensitivity and similar specificity to labeled leukocyte scans (i.e., either technetium-99m or indium-111 labelled) for localisation of infection or cancer.[44]​ 18FDG-PET/CT has been found to be particularly helpful for patients with FUO and anaemia or unintentional weight loss.[45][46]​​​

Several studies have suggested that 18FDG-PET/CT, primarily when used early in the investigation of FUO, has helped establish an earlier diagnosis, reduce length of hospital stay, and avoid unnecessary tests.[28][47]​​​[48][49]​​​​​​​​ One study found that 18FDG-PET/CT never contributed to the final diagnosis when both inflammatory parameters and body temperature were normal, supporting its use in earlier diagnosis. Diagnostic considerations for which this technology may be most helpful included localised abscesses, osteomyelitis, sinusitis, sarcoidosis, vasculitis, adult-onset Still's disease, Crohn's disease, and subacute thyroiditis. If 18FDG-PET/CT is unavailable, experts have suggested 67Ga-citrate and labeled leukocyte scintigraphy as potential alternative nuclear medicine methods.[50][51]​​​​​

Invasive diagnostic investigations

If a diagnosis remains elusive, invasive testing may be indicated. Directed biopsies should be performed as clinical suspicion indicates, particularly when the source of fever remains unidentified after extensive evaluation. Examples of other invasive investigations to consider include endoscopy, laparotomy or laparoscopy.

Bone marrow and liver biopsies are potentially useful techniques for diagnosing prolonged fever in immunocompetent patients, although not always diagnostic.[3][31]​​​[35][52]​​​​

Molecular diagnostic investigations

The drawbacks of conventional culture-based testing, such as a lack of detection of fastidious organisms, may be overcome using molecular diagnostic techniques.[5][27]​​ One such technique is next-generation sequencing, which entails the objective sequencing of every genetic component in a specimen.[27]​ Other approaches have used broad-range or pathogen-specific PCR assays. However, at this time there is little data on the regular use of these molecular techniques in cases of FUO, therefore these tests cannot be considered first-line, but may be considered in appropriately selected circumstances.[27][53]

Use of this content is subject to our disclaimer