Pyrexia of Unknown Origin: Investigation and Management

Key points

  • Classic PUO: temperature above 38.3°C on several occasions, lasting more than 3 weeks, with no diagnosis after a reasonable initial inpatient or outpatient work-up.
  • Four categories of cause: infection, malignancy, autoimmune/inflammatory disease, and miscellaneous - with a growing proportion left undiagnosed as modern imaging resolves the easier cases early.
  • Approach: a thorough, repeated history and examination generates more diagnostic leads than any single test - travel, occupation, animals, drugs, family history and a full systems review all matter.
  • Investigation strategy: staged from simple, non-invasive first-line tests through cross-sectional imaging and echocardiography to PET-CT and directed biopsy.
  • PET-CT: has become central to modern PUO work-up, identifying a focus of infection, malignancy or vasculitis in a substantial proportion of otherwise unexplained cases.
  • Empirical treatment: generally avoided before a diagnosis is reached, because antibiotics or steroids can mask the very findings needed to make it - important exceptions exist, such as suspected giant cell arteritis with visual symptoms.
  • Outcome: many cases resolve spontaneously without ever reaching a diagnosis, and this itself carries a good prognosis.

Introduction

Pyrexia of unknown origin (PUO) is fever without an identified cause despite a reasonable period of assessment. The classic definition, first proposed by Petersdorf and Beeson in 1961 and still the one used in exams, requires a temperature above 38.3°C on several occasions, persisting for more than 3 weeks, with no diagnosis reached after at least 1 week of inpatient investigation (or an equivalent outpatient work-up).1

Durack and Street later split PUO into four categories that are worth knowing, because the differential and the appropriate work-up differ between them: classic PUO (the definition above), nosocomial PUO (fever developing in a hospitalised patient with no infection present or incubating on admission), neutropenic PUO (fever in a patient with a neutrophil count below 0.5 x 10⁹/L), and HIV-associated PUO (fever persisting more than 4 weeks in an outpatient, or 3 days as an inpatient, with confirmed HIV infection).2 This article focuses on classic PUO in an immunocompetent adult; neutropenic sepsis and HIV-related opportunistic infection are managed differently and are covered in their own articles.

PUO is a favourite exam topic precisely because it resists a single algorithm: it tests whether you can build a structured differential from a good history, and whether you know which investigations to escalate to and in what order, rather than testing recall of one disease.

Causes

The differential for PUO is traditionally divided into four groups. Their relative frequency has shifted over the decades as imaging has become more sensitive: infection and malignancy are diagnosed earlier and more often now, so a growing share of modern PUO series end without a cause ever being found.

Infection (the largest group, historically 20-40%)

  • Tuberculosis - especially extrapulmonary or disseminated disease, which is easy to miss on a normal chest X-ray
  • Infective endocarditis, including culture-negative endocarditis from prior antibiotic exposure or fastidious organisms (the HACEK group, Coxiella, Bartonella)
  • Intra-abdominal or pelvic abscess - subphrenic, hepatic, splenic, diverticular, tubo-ovarian
  • Osteomyelitis or discitis, particularly in older adults or after instrumentation
  • Urinary tract infection or pyelonephritis, especially with an obstructed or complicated urinary tract
  • Viral infection - EBV, CMV, and primary HIV seroconversion, which can cause a prolonged undulating fever
  • Travel- or exposure-related infection - brucellosis, Q fever, typhoid, malaria (always ask about travel and take a thick and thin film)

Malignancy (historically around 20%)

  • Lymphoma, especially Hodgkin lymphoma - the classic Pel-Ebstein fever cycles over several days, though this pattern is seen infrequently in practice
  • Leukaemia and myelodysplastic syndrome
  • Renal cell carcinoma - a recognised paraneoplastic cause of fever, sometimes with no other symptoms
  • Hepatocellular carcinoma and hepatic metastases
  • Atrial myxoma - not malignant, but presents with fever, embolic phenomena and a variable murmur that mimics endocarditis

Autoimmune and inflammatory disease (historically 15-20%)

Miscellaneous and undiagnosed

  • Drug fever - a diagnosis of exclusion; classically causes relative bradycardia for the degree of fever and eosinophilia, and resolves within days of stopping the culprit
  • Venous thromboembolism, which can present with fever alone
  • Familial Mediterranean fever and other periodic fever syndromes - ask about family history and ethnicity, and look for a pattern of recurrent short episodes
  • Factitious or self-induced fever
  • No diagnosis reached - in modern series this is now the single largest outcome, and usually carries a good prognosis

Clinical assessment

A repeated, detailed history and examination is the single most productive step in PUO and should be revisited at intervals, since a new sign or a detail recalled later often redirects the whole work-up.

History

  • Full characterisation of the fever - pattern, height, response to antipyretics, and any associated rigors
  • Complete systems review - weight loss, night sweats, rash, joint pain, headache, visual change, cough, gastrointestinal symptoms, urinary symptoms
  • Travel history in detail, including dates, itinerary, malaria prophylaxis, freshwater exposure and vaccinations
  • Occupational and animal exposure - farming, abattoir work, contact with cats (Bartonella), livestock (brucellosis, Q fever), ticks (Lyme disease)
  • Full drug history, including recently stopped drugs and over-the-counter and herbal preparations - a drug started weeks earlier can still be the cause
  • Sexual history and HIV risk
  • Dental work, procedures or prosthetic material that could seed endocarditis
  • Family history - periodic fever syndromes, early vasculitis, or a relevant ethnic background
  • IV drug use, recent surgery, and any known malignancy or immunosuppression

Examination

A genuinely complete, top-to-toe examination, repeated on subsequent days, since findings can evolve or only become apparent later.

  • Cardiovascular: careful, repeated auscultation for a new or changing murmur, and examination for stigmata of endocarditis - splinter haemorrhages, Osler's nodes, Janeway lesions, Roth spots on fundoscopy
  • Lymphoreticular: all lymph node groups, and the abdomen for hepatosplenomegaly
  • Skin: rash (evanescent in Still's disease), nodules, needle marks
  • Musculoskeletal: synovitis, spinal tenderness suggesting discitis
  • Temporal arteries: tenderness or a thickened, pulseless vessel in anyone over 50
  • Eyes: fundoscopy for Roth spots or choroidal tubercles
  • Pelvic and rectal examination, and testicular examination in men, where relevant to the differential

Investigations

Investigation proceeds in stages, moving from simple and non-invasive tests towards more specific and invasive ones, guided at every step by whatever the history, examination and preceding results suggest.

First line

  • FBC with differential and blood film - cytopenias or blasts point towards a hae­matological cause
  • U&Es, LFTs, bone profile, CRP and ESR - a very high ESR (>100) narrows the differential towards infection, malignancy, vasculitis or Still's disease
  • Ferritin - a markedly elevated level supports adult-onset Still's disease or haemophagocytic lymphohistiocytosis
  • Urinalysis and midstream urine culture
  • At least three sets of blood cultures, from different sites, ideally before any antibiotics are given
  • Chest X-ray
  • ANA and rheumatoid factor, and an HIV test - offered to everyone regardless of perceived risk

Second line

  • CT chest, abdomen and pelvis with contrast - to find an occult abscess, lymphadenopathy or a mass
  • Transthoracic echocardiogram, proceeding to transoesophageal echocardiography if endocarditis remains suspected after a negative TTE
  • Further serology or PCR directed by exposure history - Q fever, brucella, Bartonella, EBV/CMV
  • ANCA and complement levels if vasculitis is suspected
  • Mantoux or interferon-gamma release assay if tuberculosis is possible, recognising that both can be falsely negative in disseminated disease

Third line

  • PET-CT - increasingly used earlier rather than last, because it images metabolic activity throughout the body and can identify a focus of infection, malignancy or large-vessel vasculitis that anatomical imaging alone misses
  • Directed biopsy - lymph node, bone marrow, liver, or temporal artery, guided by whatever the preceding assessment has pointed towards
  • Lumbar puncture if there are any neurological features
  • Diagnostic laparoscopy in highly selected, otherwise undiagnosed cases
How specific findings redirect the work-up.
FindingConsider
New or changing murmurInfective endocarditis - echocardiography and repeat blood cultures
Temporal headache, jaw claudication, visual symptomsGiant cell arteritis - urgent steroids, temporal artery biopsy or ultrasound
Very high ferritin, evanescent rash, sore throatAdult-onset Still's disease
Weight loss, night sweats, lymphadenopathyLymphoma - lymph node biopsy, not just imaging
Travel to an endemic areaMalaria, typhoid, brucellosis - directed films and serology
Relative bradycardia, eosinophilia, recent new drugDrug fever - trial withdrawal of the likely culprit

Management

Management is directed entirely by the diagnosis once one is reached. The harder question, and the one most often examined, is what to do while a cause is still being sought.

There are important exceptions where the risk of withholding treatment outweighs the risk of clouding the diagnosis:

  • Suspected giant cell arteritis with visual symptoms - start high-dose corticosteroids immediately; biopsy can follow within about 2 weeks without significantly reducing its yield
  • Clinical deterioration or haemodynamic instability - manage as sepsis of unknown source, per standard sepsis protocols, once cultures have been taken
  • Strong clinical suspicion of tuberculosis with a deteriorating patient in whom biopsy is not feasible - empirical anti-tuberculous therapy may be started, accepting the diagnostic trade-off
  • Neutropenic fever - always treated as an emergency regardless of diagnostic uncertainty

Outside these situations, symptomatic treatment (antipyretics, hydration) and active, iterative investigation - repeating the history, examination and appropriate tests over time - is the right approach, together with clear safety-netting and a defined review plan.

Prognosis

Prognosis depends almost entirely on the underlying cause where one is found: malignancy and endocarditis carry the outcomes associated with those diagnoses, while many infective and inflammatory causes respond well to targeted treatment.

A substantial proportion of classic PUO - up to half in some modern series - resolves spontaneously without a cause ever being identified. This group generally has a favourable prognosis: the absence of a diagnosis after thorough, modern investigation is itself reassuring evidence against an aggressive underlying process, though ongoing follow-up is still appropriate in case symptoms recur or evolve.

References

  1. Petersdorf RG, Beeson PB. Fever of unexplained origin: report on 100 cases. Medicine (Baltimore). 1961. Available here
  2. Durack DT, Street AC. Fever of unknown origin - reexamined and redefined. Current Clinical Topics in Infectious Diseases. 1991.
  3. NICE Clinical Knowledge Summaries. Fever of unknown origin (unexplained fever) - general approach. Available here
  4. Mulders-Manders C, Simon A, Bleeker-Rovers C. Fever of unknown origin. Clinical Medicine (London). 2015. Available here
  5. Takeuchi M, Nihashi T, Gafter-Gvili A et al. Association of 18F-FDG PET or PET/CT results with spontaneous remission in classic fever of unknown origin: a systematic review and meta-analysis. Medicine (Baltimore). 2018. Available here
  6. British Society for Rheumatology. Giant cell arteritis: diagnosis and management guideline. 2020. Available here

This article is written for revision and education. It is not clinical guidance and must not be used to make decisions about the care of a patient. Always check current NICE guidance and local protocols.

← All Infectious Diseases notes