on 11-Jun-2023 (Sun)

Clinicians commonly refer to a febrile illness without an initially obvious etiology (sometimes called fever without localizing signs) as fever of unknown origin (FUO). This usage is not accurate. Most febrile illnesses either resolve before a diagnosis can be made or develop distinguishing characteristics that lead to a diagnosis.

The definition of FUO derived by Petersdorf and Beeson in 1961 from a prospective analysis of 100 cases has long been the clinical standard [1]:

● Fever higher than 38.3ºC on several occasions

● Duration of fever for at least three weeks

● Uncertain diagnosis after one week of study in the hospital

Prior to concluding that a patient has an FUO, the following evaluation should have been performed and should have been unrevealing:

● History

● Physical examination

● Complete blood count, including differential and platelet count

● Blood cultures (three sets drawn from different sites with an interval of at least several hours between each set; in cases in which antibiotics are indicated, all blood cultures should be obtained before administering antibiotics)

● Routine blood chemistries, including liver enzymes and bilirubin

● If liver tests are abnormal, hepatitis A, B, and C serologies

● Urinalysis, including microscopic examination, and urine culture

● Chest radiograph

Three general categories of illness account for the majority of "classic" FUO cases and have been consistent through the decades. These categories are:

● Infections

● Malignancies

● Systemic rheumatic diseases (eg, vasculitis, rheumatoid arthritis)

Less common diagnoses of fever of unknown origin

Infections Abscesses (especially intra-abdominal) African tick bite fever* Amebic liver abscess* Anaplasmosis/ehrichiosis* Babesiosis* Brucellosis* Castleman's disease Chikungunya* Chronic active hepatitis Culture-negative endocarditis¶ Cytomegalovirus Dental abscesses Dengue* Diskitis Epididymitis Fascioloiasis* Filariasis* Gonococcal arthritis Herpes simplex encephalitis Infectious mononeucleosis Kala azar (visceral leishmaniasis)* Kikuchi's disease Lassa fever* Leptospirosis* Lyme disease* Osteomyelitis Prostatitis Pyelonephritis Pyometria Q fever* Relapsing fever (Borrelia recurrentis)* Rheumatic fever Sinusitis Toxoplasmosis Typhoid fever* Tuberculosis Whipple's disease Zika virus*

Malignancies Aleukemic leukemia Atrial myxoma Colon cancer Hepatocellular carcinoma or other tumors metastatic to the liver Kaposi's sarcoma Leukemia Lung cancer Lymphoma, especially non-Hodgkin's Mesothelioma Multiple myeloma Myelodysplastic syndromes Renal cell carcinoma Sarcoma

Systemic inflammatory diseases Allergic granulomatous angiitis Antiphospholipid syndrome Behçet's disease Cryoglobulinemia Giant cell arteritis Granulomatosis with polyangiitis (formerly Wegener's disease) Granulomatous hepatitis Hypersensitivity vasculitis Inflammatory bowel disease Panaortitis Polyarteritis nodosa

...

The fraction of undiagnosed FUOs dropped from over 75 percent in the 1930s to less than 10 percent in the 1950s. Since then, the fraction of FUOs that go undiagnosed has steadily increased (figure 1) [4,7,14-16].

Infective endocarditis, once a frequent cause of FUO, has become a less common cause with improved techniques for the isolation of organisms. In the current era, when endocarditis is ultimately diagnosed in a patient with FUO, it is more likely to be culture negative or caused by difficult-to-isolate organisms, such as Bartonella quintana.

True FUOs are uncommon. This was illustrated in a report from the Netherlands in which only 73 patients were identified between December 2003 and July 2005 at a 950-bed academic referral hospital and five community hospitals comprising 2800 hospital beds [9]. The authors excluded immunocompromised patients, such as those with AIDS, hypogammaglobulinemia, granulocytopenia, and glucocorticoid therapy. The following distribution of causes was noted:

● Systemic rheumatic disease (eg, vasculitis, systemic lupus erythematosus, polymyalgia rheumatica) – 22 percent

● Infection – 16 percent

● Malignancy – 7 percent

● Miscellaneous – 4 percent

● No diagnosis – 51 percent

The frequent use of empiric antimicrobials, for example, can delay the diagnosis of some occult abscesses and infections and increases the number of drug fevers. Aggressive immunosuppressive regimens, increased exposure to potentially allergenic medications, lengthy intensive care unit admissions, and the increase in multiresistant organisms as resident hospital flora have all altered the types of FUO’s encountered.

Ease of travel has the potential to bring back to the United States and other resource-rich countries more geographically restricted illnesses that may not be familiar to clinicians, such as malaria, brucellosis, kala azar, filariasis, schistosomiasis, African tick bite fever, relapsing fever, Q fever, dengue virus, chikungunya virus, Zika virus, or Lassa fever [17,18]

Zoonoses like babesiosis, ehrlichiosis, anaplasmosis, and Lyme disease may present as FUOs in new ecologic niches.

Despite the wide variety of exotic diseases, the most common infections among series of FUOs have not changed over the past half century. These continue to include typhoid fever, tuberculosis, amebic abscesses, and malaria. Fever of unknown origin is more often caused by an atypical presentation of a common entity than by a rare disorder.

The causes of FUO in HIV-infected patients reflect the degree of immunosuppression, best measured by CD4 counts and viral load determinations [21].

Acute HIV infection can present as an FUO when telltale symptoms and signs (rash and diffuse adenopathy) are absent. In one series of 79 episodes of FUO among HIV-infected patients with CD4 counts ranging from 0 to 790/microL and a median of 40/microL, 79 percent were due to infections and 8 percent to malignancies; only 9 percent had no definite diagnosis [ 6]. Over one-half were due to mycobacteria, two-thirds of which were atypical, most commonly Mycobacterium avium complex (MAC). Only lymphomas were highly represented among malignancies, particularly non-Hodgkin lymphomas. Disseminated Kaposi's sarcoma was relatively infrequent.

As with FUO in HIV-uninfected patients, the country in which the study is performed influences the prevalent etiologies detected. In one study from Spain, for example, 137 HIV-infected patients with fevers for at least 10 days and no diagnosis after 1 week in hospital underwent bone marrow biopsy [23]. The three most common diagnoses were mycobacterial infection (18 patients with Mycobacterium tuberculosis and 14 with MAC), non-Hodgkin lymphoma, and visceral leishmaniasis.

Neutropenia-associated febrile episodes without a source are most frequently linked to bacteremia. Fungal infections replace bacterial infections in prominence after the acute period (after seven days) [12].

Fever of unknown origin in this population is particularly confounding. It occurs in the context of a serious underlying condition that may itself cause fever. In addition, patients are often taking a number of medications (including antimicrobials), are receiving blood products, have varying degrees of immunosuppression, and, in transplant recipients, may experience allograft rejection. Finally, febrile neutropenia usually mandates empirical antimicrobial therapy, which may confound the subsequent detection of bacterial infections. (

Fever, even if unexplained, usually abates with return of neutrophils. When fever persists or returns after the patient is no longer neutropenic, hepatosplenic candidiasis should be strongly considered [12].

The most critical feature of the evaluation of a patient with FUO is to take a careful history and to reassess the patient frequently. It is important to look for uncommon presentations of common diseases and to perform a detailed physical examination.

The art of diagnosis is one of discrimination, as the clinician must determine which data to glean and which clues to pursue. In the series of 73 patients from the Netherlands cited above, the authors found an average of 10.5 potential diagnostic clues per patient through their careful history and physical examination and three per patient through laboratory testing [9]. Eighty-one percent of these clues were misleading.

A thorough history should include the following information:

● Travel

● Animal exposure (eg, pets, occupational, living on a farm)

● Immunosuppression (with the degree noted)

● Drug and toxin history, including antimicrobials

● Localizing symptoms

S ubtle findings may be elicited through a careful history. Examples include subtle changes in behavior or cognition consistent with granulomatous meningitis, jaw claudication consistent with giant cell arteritis, tooth sensitivity to cold or gum tenderness consistent with dental abscesses, and nocturia consistent with prostatitis. Revisiting the history on several occasions may provide new clues in difficult cases.

The degree of fever, nature of the fever curve, apparent toxicity, and response to antipyretics has not been found to provide enough specificity to guide the diagnosis of FUO [24]. Fever may be attenuated in older patients and moderated by use of steroids and nonsteroidal anti-inflammatory drugs. However, the course of the fever curve may be helpful in determining whether the disease is escalating or waning.

A normal ESR or CRP also suggests that a significant inflammatory process, of whatever origin, is absent; however, there are exceptions. As an example, some patients with giant cell arteritis have a normal ESR [26]

In addition to basic testing that establishes an FUO (see 'Establishing that a patient has an FUO' above), we typically perform the following minimum diagnostic evaluation for patients who have true FUO. We individualize other laboratory tests based on clinical and historical findings. (See 'Additional testing for specific circumstances' below.)

Procalcitonin, a serum biomarker that is elevated with certain bacterial infections, has no clear role in distinguishing between bacterial infections and other causes of FUO, and we do not recommend checking it as part of the FUO evaluation.

Many tests are often necessary to determine the etiology for ongoing fever and the results require critical interpretation and synthesis. The diagnostic performance of any single test is limited by several factors:

● The final diagnosis is often not independent of the test being studied.

● Studies evaluating test performance are generally retrospective and rarely standardized.

● True- and false-positive rates are highly dependent on the timing of the test in relation to the course of illness and available clinical information.

● The differential diagnosis for FUO is broad, and tests do not reliably distinguish between infections, malignancy, and inflammatory conditions or identify specific conditions.

● Tests with the highest sensitivity often identify conditions that may not be the cause of fever (eg, hamartomas, thromboses).

Furthermore, during an FUO work-up, a positive result on imaging often does not change management because the diagnosis would have been made even without the imaging study (ie, the test was not helpful). This was illustrated in the series cited above of 73 patients from the Netherlands seen between December 2003 and July 2005 [9]:

● Chest radiograph – Performed in 73 patients: helpful in 6 and false positive in 8 (8 and 11 percent, respectively)

● Chest CT – Performed in 46 patients: helpful in 9 and false positive in 8 (20 and 17 percent, respectively)

● Abdominal CT – Performed in 60 patients: helpful in 12 and false positive in 17 (20 and 28 percent, respectively)

● Positron emission tomography scan – Performed in 70 patients: helpful in 23 and false positive in 10 (33 and 14 percent, respectively)

CT for all patients — We recommend an abdominal and chest CT as part of the routine evaluation of FUO. However, the rate of false-positive tests with these modalities may be similar to the rate of a true-positive result [9].

CT scanning of the abdomen has nearly replaced exploratory laparotomy and other radiographic tests in the search for occult abscesses or hematomas in patients with FUO. The finding of abdominal lymphadenopathy can be a clue to lymphoma or a granulomatous process. The usefulness of CT has resulted in this examination being used in nearly all patients with FUO. While magnetic resonance imaging scan can be more sensitive in certain settings (eg, the diagnosis of spinal epidural abscess), it is rarely required in the initial evaluation of FUO.

For similar reasons, CT scanning of the chest is invaluable in the identification of small nodules (indicative of fungal, mycobacterial, or nocardial infection or malignancy). The identification of hilar or mediastinal adenopathy may prompt biopsy by mediastinoscopy, providing a diagnosis of lymphoma, histoplasmosis, or sarcoidosis.

owever, more recent studies suggest that modern CT scans are more helpful than WBC scans [29,32].

However, recent studies have suggested that the negative predictive value for these scans is high because the scans are typically ordered at a point in the work-up when a source for fever is unlikely to be found. For example, in one study, the negative predictive value for WBC scans in 132 patients with FUO was 89 percent, compared with a negative predictive value of 88 percent for a conventional work-up had the WBC never been ordered [29].

FDG-PET appears to be more sensitive than WBC scans in identifying anatomic sites of inflammation, infection, and malignancy. This modality may find a valuable place in the evaluation of FUO, but additional data are needed to determine its added value beyond repeated clinical evaluation over time and routine CT [34-37]. As noted above, a positive result on FDG-PET may be as likely to be a false positive as it is to be a true positive [9].

A history of trauma, adjacent infection or intravenous drug use may suggest thrombophlebitis of the legs, arms, or pelvic vessels. Venous duplex imaging can be diagnostic. Fever usually responds to anticoagulation within several days.

Subtle central nervous system symptoms or signs should prompt a lumbar puncture and imaging of the head and/or spine.

In the United States, a travel history to the Midwest or the deserts of the West should raise the question of a fungal process like histoplasmosis or coccidioidomycosis, respectively. Testing for the suspected fungal pathogen in individuals who have resided in an endemic area can be useful.

Other appropriate tests for returning travelers are discussed separately. (See "Evaluation of fever in the returning traveler".)

There are no studies evaluating these tests' operating characteristics in clinical use, early versus later use in FUO, ability to differentiate colonization from infection, or sensitivity when performed after antimicrobial therapy.

In several case reports, molecular diagnostic assays have detected infectious etiologies in individuals with FUO for whom cultures and other workup yielded no diagnosis [37,38]. The use of these methodologies has led to such diagnoses as culture-negative tuberculosis, bacteremia due to probiotic supplements, non-tuberculous mycobacterial splenic infection, and Abiotrophia endocarditis.

We suggest performing a biopsy when the clinical data support a diagnosis that can be confirmed from a tissue sample, especially when the risks are low (e.g., skin biopsy). Blind biopsies rarely yield helpful results.

The following examples include data from the Netherlands study of 73 patients cited above on the utility of biopsy of different sites [9]:

● Liver biopsy for possible miliary tuberculosis, granulomatous hepatitis, or other granulomatous diseases such as sarcoidosis – Liver biopsy was performed in 7 patients; it was helpful in 1 and false positive in 3.

● Lymph node biopsy for possible malignancy, especially lymphoma, or infections such as cat-scratch disease – Lymph node biopsy was performed in 11 patients; it was helpful in 5 and false positive in 3.

● Temporal artery to look for giant cell arteritis or biopsy of an affected tissue to diagnose a vasculitic process such as polyarteritis nodosa – Temporal artery biopsy was performed in 14 patients; it was helpful in 1 with no false positives.

● Pleural or pericardial biopsy in the evaluation of extrapulmonary tuberculosis.

● Bone marrow biopsy was performed in 19 patients; it was helpful in 2 and false positive in 1.

Two retrospective reviews of bone marrow biopsies to evaluate FUOs demonstrated high diagnostic yields and high prevalence of hematologic malignancies [39,40]. The authors did not define the prevalence of infectious diseases at the hospitals under study nor did they define the decision-making that led physicians to request bone marrow biopsies. However, the observations in both studies were similar: lymphomas constituted >40 percent of diagnoses, whereas infections were detected in <15 percent of patients. Other causes of FUO included acute myeloid leukemia, myelodysplastic syndromes, sarcoidosis, systemic mastocytosis, and disseminated granulomatosis. In both studies, hematologic malignancies were strongly predicted by the presence of leukoerythroblastic changes in peripheral blood and a greatly elevated ferritin level (>1000 ng/mL); in one of the studies, hematologic malignancies were also predicted by the presence of splenomegaly [40].

Therapeutic trials of antimicrobials or glucocorticoids, while tempting in the effort to "do something," rarely establish a diagnosis. In addition, the diagnostic yield of blood cultures and cultures of biopsy material will be reduced following the initiation of antibiotics.

A therapeutic trial of glucocorticoids for an inflammatory process should not replace relevant biopsies for steroid-responsive diseases such as sarcoidosis, other granulomatous diseases, or vasculitis. A careful evaluation for infection should precede such a trial.

When morbid illnesses like endocarditis, temporal arteritis, central nervous system tuberculosis, and leptospirosis present as FUOs, they rarely progress rapidly enough to warrant urgent treatment. However, when illnesses like these are suspected, we speed diagnostic testing and treat empirically.

Antipyretics improve patients' comfort, reducing headache, myalgias, arthralgias, and fatigue. However, drugs with antipyretic effects may delay or obscure early symptoms and signs of specific diseases. Thus, we try to avoid prescribing acetaminophen, nonsteroidal anti-inflammatory drugs, or glucocorticoids.

The use of antipyretics is discussed in greater detail separately. (See "Pathophysiology and treatment of fever in adults", section on 'Treatment of fever and hyperpyrexia'.)

The outcome of patients with an FUO depends upon the underlying diagnosis and any comorbid conditions. The rate of no diagnosis in studies of FUO published since 1990 has varied from 9 to 51 percent [4-9,29].

Most adults who remain undiagnosed after an extensive evaluation have a good prognosis [7,8,41]. This was illustrated in a study of 199 patients with FUO, 61 of whom (30 percent) were discharged from the hospital without a diagnosis [41]:

● A definite diagnosis was established in 12 (20 percent), usually within two months after discharge.

● Thirty-one (51 percent) became symptom-free during hospitalization or shortly following discharge.

● Eighteen (30 percent) had persisting or recurring fever for several months or even years after discharge, 10 of whom were considered to be finally cured.

● Four were treated with glucocorticoids and six required intermittent therapy with nonsteroidal anti-inflammatory drugs.

● Six died, but the cause of death was considered to be related to the disease that caused FUO in only two cases.

Similar findings were noted in the Netherlands series of 73 patients cited above [9]. Among the 37 patients with no diagnosis who were followed for at least six months, 16 spontaneously recovered, 5 recovered with nonsteroidal anti-inflammatory drugs or glucocorticoids, 15 had persistent fever, and 1 died.