Fever of Unknown Origin
Every clinician can name the cause of most fevers within a day or two: a positive urine culture, a chest full of crackles, a red hot joint. Fever of unknown origin (FUO) is the frustrating remainder — the patient who keeps spiking temperatures for weeks while the obvious tests come back clean. It is one of the great diagnostic puzzles of internal medicine, and learning to work through it teaches something more valuable than any single disease: how to reason methodically when the answer is hiding.
FUO matters because the differential is enormous — well over two hundred causes have been described — yet the correct diagnosis usually turns out to be an unusual presentation of a common disease rather than a rare disease itself. The skill is not memorising exotic infections; it is disciplined, iterative investigation guided by clues from the history and examination that are easy to miss on a first pass.
Learning Objectives
- State the classic and revised definitions of fever of unknown origin and why the definition deliberately excludes short febrile illnesses.
- Distinguish the four categories of FUO (classic, nosocomial, immune-deficient, HIV-associated) and how each reshapes the differential.
- Group the major causes into infections, malignancies, non-infectious inflammatory diseases, and miscellaneous, with the leading examples in each.
- Apply a staged, clue-directed diagnostic workup rather than a scattergun of tests.
- Recognise when no diagnosis is reached and how such patients are safely managed.
Quick Answer
Fever of unknown origin is defined classically as a temperature above 38.3°C (101°F) on several occasions, lasting more than three weeks, that remains undiagnosed after appropriate initial investigation. The three-week floor is deliberate: it filters out the self-limiting viral illnesses that account for most short fevers. Causes fall into four buckets — infections, malignancies, non-infectious inflammatory diseases, and miscellaneous — and the balance shifts with geography, age, and immune status. In wealthy countries, non-infectious inflammatory conditions (such as giant cell arteritis and adult-onset Still disease) now rival infections as the top group, while a growing share of cases stay undiagnosed and simply resolve on their own. The workup is staged and driven by potential diagnostic clues from a repeated history and examination rather than by ordering every test at once. Empiric antibiotics or steroids should generally be withheld until a diagnosis is reasonably established, because they blur the picture without curing it.
Where It Came From
Fever itself was the central object of medicine for millennia — Hippocratic physicians charted fever curves and named quotidian, tertian, and quartan patterns long before anyone knew what caused them. But the specific concept of a prolonged, unexplained fever as a defined clinical entity is modern. It crystallised in 1961, when Robert Petersdorf and Paul Beeson published a landmark prospective study of 100 patients at Yale, proposing the criteria that still anchor the field: temperature over 101°F lasting more than three weeks, undiagnosed after a week of in-hospital investigation.
Why invent such a precise definition? Because the pre-antibiotic and early-antibiotic eras were awash in prolonged fevers, and physicians needed a shared framework to study them systematically rather than chase each case blindly. Petersdorf and Beeson's genius was to insist on a duration threshold, which excluded the flood of ordinary viral infections and concentrated attention on the smaller set of serious, treatable, or diagnosable diseases. Their study also delivered the enduring lesson that FUO is usually a common disease behaving atypically — tuberculosis, endocarditis, abscess, lymphoma — not a zebra.
The definition was revised in 1991 by Durack and Street, who recognised that the one-week-in-hospital rule no longer fit an era of rapid outpatient testing. They replaced it with a requirement for a defined set of investigations and, crucially, split FUO into four categories reflecting the changing landscape of medicine — the rise of hospital-acquired infection, immunosuppressive therapy, and the HIV pandemic. That four-category scheme is how the topic is taught today.
The Definition and Why Its Boundaries Are Deliberate
The classic (Petersdorf-Beeson) criteria have three parts, and each earns its place:
- Temperature above 38.3°C (101°F) on several occasions. Requiring documented, repeated elevation weeds out normal circadian variation and single spurious readings, and distinguishes genuine fever from factitious or fraudulent temperature reports.
- Duration greater than three weeks. This is the load-bearing criterion. Most acute febrile illnesses are viral and self-resolve; three weeks selects for the diagnostically meaningful minority.
- No diagnosis after appropriate investigation. The fever must be unexplained despite a reasonable workup — not simply unexplained because nobody has looked yet. This prevents lazy labelling.
A patient febrile for four days with a normal chest X-ray does not have FUO; they have an undifferentiated acute febrile illness, which is a different problem with a different (largely watchful) approach.
The Four Categories
Durack and Street's categorisation is the single most useful teaching device, because the category dictates the differential.
| Category | Typical setting | Leading concerns |
|---|---|---|
| Classic FUO | Community patient, fever over 3 weeks, undiagnosed after appropriate outpatient or brief inpatient workup | Infection, malignancy, non-infectious inflammatory disease |
| Nosocomial FUO | Fever arising after admission (not present or incubating on arrival), undiagnosed after 3 days | Device and line infection, C. difficile colitis, drug fever, venous thromboembolism, sinusitis in intubated patients |
| Immune-deficient (neutropenic) FUO | Neutropenia (neutrophils below 500), fever undiagnosed after 3 days | Bacterial and fungal infection (Candida, Aspergillus); needs urgent empiric therapy |
| HIV-associated FUO | Confirmed HIV, fever over 3–4 weeks (outpatient) or 3 days (inpatient) | Opportunistic infection (disseminated TB, MAC, cryptococcus, PCP), lymphoma, HIV itself |
Note the shortened time frames outside the classic category: in a neutropenic or nosocomial patient, waiting three weeks would be dangerous, so the definition compresses to days and the response is far more aggressive.
The Big Four Cause Groups
For classic FUO, causes are traditionally grouped into four categories, and knowing the leading members of each is high-yield.
Infections. Historically the largest group, and still dominant in older and lower-income populations. The classic culprits are those that hide: tuberculosis (especially extrapulmonary or miliary), intra-abdominal or pelvic abscesses, infective endocarditis (particularly culture-negative or after partial antibiotic treatment), osteomyelitis, and complicated urinary infections. Geography matters enormously — think malaria, typhoid, brucellosis, leishmaniasis, and Q fever depending on travel and exposure.
Malignancies. Lymphoma (Hodgkin and non-Hodgkin) is the classic fever-producing cancer, sometimes with drenching night sweats and the periodic Pel-Ebstein pattern. Also consider renal cell carcinoma, hepatocellular carcinoma, leukaemia, and atrial myxoma (a benign tumour that mimics endocarditis with fever, emboli, and constitutional symptoms).
Non-infectious inflammatory diseases (NIID). In modern high-income series this group has risen to rival or surpass infection. Key entries: giant cell arteritis and polymyalgia rheumatica (essential to consider in anyone over 50 — a missed diagnosis risks blindness), adult-onset Still disease (spiking fevers, evanescent salmon-coloured rash, arthralgia, very high ferritin), systemic lupus erythematosus, polyarteritis nodosa, and sarcoidosis.
Miscellaneous. Drug fever (a common and easily reversed cause — often with a well-looking patient despite high temperatures), venous thromboembolism, subacute thyroiditis, factitious fever, familial Mediterranean fever and other periodic fever syndromes, and hyperthyroidism.
An important fifth outcome deserves emphasis: undiagnosed FUO is now a large and growing category. In modern series, 20–50% of classic FUO cases resolve spontaneously without a diagnosis ever being made — and these patients generally have an excellent prognosis.
A Staged, Clue-Directed Workup
The cardinal error in FUO is the reflex to order every test at once. The evidence-based approach is iterative: repeat the history and examination, identify what the modern literature calls potentially diagnostic clues (PDCs), and let those clues direct the next test.
Step 1 — Confirm the fever and stop harmful noise. Document temperatures objectively. Review the full medication list for drug fever. Take a meticulous history: travel, animal and tick exposure, sexual history, occupation, family history (periodic fever syndromes), prior surgery and prosthetic material, and a full systems review. Re-examine daily — a new murmur, a temporal artery that is tender, splenomegaly, a subtle rash, or lymphadenopathy can appear over days.
Step 2 — Baseline investigations. Full blood count with differential, blood film, comprehensive metabolic panel, liver enzymes, urinalysis and culture, at least three sets of blood cultures ideally off antibiotics, inflammatory markers (ESR and CRP), lactate dehydrogenase, ferritin, HIV test, chest X-ray, and tuberculosis testing (interferon-gamma release assay). ESR that is strikingly high in an older patient points toward giant cell arteritis, abscess, or malignancy.
Step 3 — Directed second-line testing. Guided by clues: echocardiography (start transthoracic, escalate to transoesophageal) if endocarditis is plausible; cross-sectional imaging (CT chest/abdomen/pelvis) to hunt for abscess, lymphadenopathy, or tumour; serologies for suspected zoonoses; autoantibodies if a rheumatic cause is suspected.
Step 4 — Advanced imaging and biopsy. FDG-PET/CT has become a powerful tool because it lights up sites of infection, inflammation, and malignancy simultaneously, frequently redirecting the workup toward a biopsy target. Temporal artery biopsy is warranted in older patients with high ESR and any cranial symptoms. Bone marrow or lymph node biopsy follows suggestive clues.
A Worked Vignette
A 68-year-old woman has six weeks of daily low-grade fever, fatigue, a 4 kg weight loss, and a new dull headache with scalp tenderness when she brushes her hair. Basic labs show ESR 92 mm/hr and normocytic anaemia; blood cultures are negative and CT is unremarkable. The clue is the combination of age over 50, high ESR, and cranial symptoms. The next test is not another round of cultures but a temporal artery biopsy — and, because the risk of sudden blindness is real, high-dose corticosteroids are started before the biopsy result returns. Biopsy confirms giant cell arteritis. This case captures the whole discipline: the answer lay in the history, not in ordering more of the same test.
Real-World Applications
In everyday hospital medicine, the FUO framework prevents two opposite failures: the panic of ordering hundreds of low-yield tests, and the complacency of anchoring on the first plausible explanation. It structures ward rounds — asking each day "what new clue has emerged?" On the wards, the nosocomial category reminds teams to examine every line, catheter, and surgical wound, and to consider drug fever before escalating antibiotics. In oncology and transplant units, the immune-deficient category converts fever into a medical emergency demanding empiric broad-spectrum therapy within an hour. In HIV and global-health settings, the framework foregrounds tuberculosis and opportunistic infection. For students, FUO is the classic examination case precisely because it rewards a systematic history and examination over reflexive test-ordering.
Common Mistakes
Mistake 1: Calling a four-day fever an FUO. FUO requires roughly three weeks (or the compressed frames for the other categories). Short fevers are usually viral and self-limiting. Applying the FUO machinery to an acute febrile illness leads to needless, low-yield, sometimes harmful testing. The correction: honour the duration criterion, and manage undifferentiated short fevers with targeted testing and watchful waiting.
Mistake 2: Giving empiric antibiotics or steroids to "cover" the patient. Blind antimicrobials in a stable classic-FUO patient can suppress and delay diagnosis (for example, partially treating endocarditis into a culture-negative state) without curing anything, while blind steroids can mask malignancy or worsen occult infection. The correction: withhold empiric therapy until a working diagnosis exists — the important exceptions being neutropenic and haemodynamically unstable patients, and the specific case of suspected giant cell arteritis with visual threat, where steroids cannot wait.
Mistake 3: Reaching for exotic diagnoses first. The enduring lesson since 1961 is that FUO is usually a common disease presenting atypically — TB, abscess, endocarditis, lymphoma, giant cell arteritis — not a rare tropical zebra. The correction: work the common causes thoroughly before chasing rarities, and let exposure history (not imagination) trigger the exotic serologies.
Comparison and Connections
| Concept | What it means | Key distinction |
|---|---|---|
| Undifferentiated acute febrile illness | Fever of days, no clear source yet | Too short to be FUO; usually viral and self-limiting |
| Fever of unknown origin | Fever over ~3 weeks, undiagnosed after appropriate workup | The defined diagnostic puzzle |
| Sepsis | Life-threatening organ dysfunction from dysregulated infection response | Acute and unstable; not a diagnostic-delay problem |
| Factitious fever | Self-induced or fabricated fever | A cause of FUO; suspect with normal pulse, no diurnal pattern, discordant readings |
| Hyperthermia | Raised body temperature from failed thermoregulation (heat stroke, malignant hyperthermia) | Not mediated by the hypothalamic set-point, unlike true fever |
FUO connects tightly to several branches: it is a showcase for Infective Endocarditis as a classic culture-negative cause, and it draws on microbiology, rheumatology, and haematology-oncology in equal measure.
Practice Questions
Recall
Q: State the three components of the classic Petersdorf-Beeson definition of FUO. A: Temperature above 38.3°C (101°F) on several occasions; duration greater than three weeks; and no diagnosis reached after appropriate investigation.
Understanding
Q: Why does the classic definition insist on a three-week duration? A: Because most acute fevers are self-limiting viral illnesses. The three-week threshold filters these out and concentrates attention on the smaller subset of serious, diagnosable, or treatable diseases, making systematic study and workup worthwhile.
Application
Q: A neutropenic patient (neutrophils 300) develops fever 3 days into a chemotherapy admission with no source found. How does the approach differ from a classic FUO? A: This is immune-deficient (neutropenic) FUO, defined over days rather than weeks. It is a medical emergency: rather than staged outpatient investigation, the patient needs urgent empiric broad-spectrum antibiotics (with antifungal cover if fever persists), because occult bacterial and fungal infection can be rapidly fatal in neutropenia.
Analysis
Q: Modern high-income FUO series show non-infectious inflammatory disease rivalling infection, and a large "no diagnosis" group with good prognosis. What explains this shift, and how should it change clinical behaviour? A: Rapid outpatient imaging and cultures now identify common infections early, so they less often persist into FUO territory, while an ageing population raises the incidence of conditions such as giant cell arteritis. The growing undiagnosed group reflects both better exclusion of dangerous diseases and the reality that many self-limiting inflammatory processes never get named. Clinically, this justifies a measured approach: exclude the treatable and dangerous causes rigorously, but avoid endless invasive testing in a stable, improving patient who may simply be resolving.
FAQ
Is FUO a diagnosis in itself? No. It is a situation — a fever that has resisted initial explanation — that frames a search for the true underlying diagnosis. The goal is always to move from "FUO" to a specific cause (or to a safe conclusion of self-limiting, undiagnosed fever).
How often is a cause never found? In modern series, roughly a fifth to a half of classic FUO cases remain undiagnosed. Reassuringly, these patients usually do well, with the fever settling spontaneously over weeks to months.
Should I just start broad antibiotics while I investigate? Not for a stable classic-FUO patient. Empiric antibiotics rarely cure and often obscure — for example, converting endocarditis into a culture-negative puzzle. Empiric therapy is reserved for neutropenic patients, the haemodynamically unstable, and specific threats such as giant cell arteritis (steroids) with impending visual loss.
What single test has changed FUO workups most? FDG-PET/CT. Because it highlights infection, inflammation, and malignancy simultaneously, it frequently identifies a biopsy target and shortens the diagnostic journey when first-line tests are unrevealing.
Why does the same fever get worked up so differently in different patients? Because category and context reshape the differential and the urgency. A returning traveller, a neutropenic oncology patient, a person living with HIV, and a well 40-year-old with three weeks of fever face entirely different probable causes and time pressures, even with an identical temperature chart.
Quick Revision
- FUO (classic): temperature over 38.3°C on several occasions, more than 3 weeks, undiagnosed after appropriate workup.
- Four categories: classic, nosocomial, immune-deficient (neutropenic), HIV-associated — the last three defined over days, not weeks.
- Four cause groups: infections, malignancies, non-infectious inflammatory disease, miscellaneous — plus a large "no diagnosis" group.
- Usually a common disease presenting atypically (TB, abscess, endocarditis, lymphoma, giant cell arteritis), not a rare zebra.
- Workup is staged and clue-directed: repeat history and exam, baseline labs and cultures, then targeted imaging, then FDG-PET and biopsy.
- Avoid empiric antibiotics/steroids in stable patients; exceptions are neutropenia, instability, and visual-threat giant cell arteritis.
- Defined by Petersdorf and Beeson (1961); four-category revision by Durack and Street (1991).
Related Topics
Prerequisites
Related Topics
Next Topics
- Antimicrobial stewardship and the logic of empiric versus targeted therapy (see the branch overview for the medicine of infections)