Approach to pyrexia of unknown origin in low- and middle-income countries

Definition

Pyrexia of unknown origin (PUO) is defined as a persistent febrile illness without an apparent initial cause despite reasonable evaluation. The definition is based on three essential criteria:

Fever – Documented temperature ≥38.3°C on multiple occasions.

Over time, the definition of PUO has evolved, with key differences primarily in the third criterion: the first, proposed in 1961, required ≥1 week of in-patient evaluation. ¹ A revision was made in 1991, reducing the in-patient evaluation period to ≥3 days and adding ≥3 out-patient visits. Four PUO subtypes were introduced: classical, nosocomial, neutropenic, and HIV-related. ² Recent qualitative definitions have shifted the focus from duration of evaluation to a standardised panel of investigations, including: (1) Full blood count (with differential) and comprehensive biochemical testing (including calcium and liver function tests); (2) Inflammatory markers: erythrocyte sedimentation rate [ESR], C-reactive protein [CRP], and ferritin; (3) Immunological tests: antinuclear antibody [ANA], rheumatoid factor [RF], and antineutrophil cytoplasmic antibody [ANCA]); (4) Microbiological tests (blood cultures ×3, urinalysis [with culture if significant pyuria], tuberculin skin test or interferon-gamma release assay; and (5) Imaging (chest radiography and abdominal ultrasound or CT scan of chest, abdomen, and pelvis).^3,4

Causes

PUO is categorised into five principal categories: (1) Infectious, (2) Non-infectious Inflammatory, (3) Neoplastic, (4) Miscellaneous, and (5) Undiagnosed.^1–4 Its aetiology varies significantly depending on geographic location, host factors (age, immune status), and available diagnostic resources.^3–5 While high-income countries initially reported infections as the leading cause (at 40–50%), a gradual shift has been observed, with inflammatory disease and malignancy becoming more prevalent.^1–3 In contrast, data from low- and middle-income countries (LMICs) continue to indicate infectious diseases as the leading cause (at 40–50%).^5–7

Common causes in LMICs

Tuberculosis (TB) remains the leading cause worldwide, contributing c. 25% in endemic regions, with extrapulmonary TB accounting for 70% of these, gastro-intestinal and lymph nodal being the most frequently involved sites.^3–7 Pulmonary and combined pulmonary-extrapulmonary TB contribute an additional 10–20% each.

Vector-borne and zoonotic infections, however, remain significant contributors. A meticulous evaluation therefore of epidemiological risk factors – including social, travel, and occupational exposure history – is imperative, as infections linked to environmental exposures (e.g. raw dairy consumption for brucellosis) are frequently overlooked.^5,6,8 Brucellosis, a multi-systemic zoonotic disease with diverse clinical presentations, can closely mimic TB or malignancy, further complicating diagnosis.^5,9

Enteric fever is another common cause due to its indolent course and lower fatality rate, with diagnosis primarily relying on blood culture, which has a high false-negative rate especially after antibiotic exposure.^5,6

In contrast, tropical acute febrile illnesses such as scrub typhus, leptospirosis, malaria (including tropical splenomegaly syndrome), and dengue are now rarer as PUO aetiologies. This is both because of a rapid progression to life-threatening complications prompting earlier diagnosis and intervention, and the widespread availability of rapid diagnostic kits and serological tests to facilitate the diagnoses.

Further, a declining prevalence of rheumatic heart disease and increased access to echocardiography have made cases of endocarditis easier to diagnose.^5,10

Nonetheless, although IgM-based serological tests are widely used, they carry a high false-positive rate and should always be interpreted with caution. Appropriate history taking, clinical examination, assessment of epidemiological risk factors, and confirmatory antibody titres remain essential to prevent misdiagnosis.

Among neoplastic causes, haematological malignancy (viz. lymphoma, leukaemia) are the most frequently encountered as PUO.^3–7 Lymphoma, in particular, presents a diagnostic challenge, as its common manifestations, including hepato-splenomegaly and lymphadenopathy, closely mimic infections such as TB.^5–7

In contrast, solid tumours associated with PUO typically present in advanced stages, often due to distant metastases or post-obstructive phenomena such as cholangitis, pneumonia, or urinary tract infections.^5,7

Among inflammatory disease, Still's disease remains a common but under-diagnosed cause, primarily due to its diagnosis being based on exclusion criteria.^5,7,11,12 It presents with non-specific systemic features, including arthralgia, skin rash, hepato-splenomegaly, lymphadenopathy, and neutrophilic leucocytosis. A markedly elevated serum ferritin level (>2000μg/L) should suggest Still's disease in young adults (<40 years) with PUO. Delayed diagnosis is, however common, increasing the risk of its progression to macrophage activation syndrome or secondary haemophagocytic lymphohistiocytosis.^5,11,12

Diagnostic evaluation: a stepwise approach

The challenge is to differentiate one of the above from an atypical presentation of a common disease or a rare disorder.^3–7,11 While structured diagnostic protocols have been proposed, their superiority over non-structured approaches remains debatable. ¹² I propose a stepwise approach, integrating local epidemiology, potential diagnostic clues, and systematic testing to improve diagnostic accuracy while avoiding unnecessary investigations.^{3–6,11–15}

OPEN IN VIEWER

Flowchart Algorithm: Stepwise Protocol for PUO

Step 1: Clinical Evaluation Comprehensive history. Full physical examination. Temperature chart (to assess fever patterns and exclude factitious fever). Medication history, including complementary and alternative medication (to rule out drug-induced fever).. Environmental or occupational exposure and travel history (to identify zoonotic, vector-borne or occupational infections). N.B. Discontinue antimicrobial agents and glucocorticoids, except in rapidly deteriorating cases.

Step 2: Basic Minimal Investigations Full blood count (including differential) Biochemical testing (including calcium and liver function tests). ESR, CRP, ferritin. ANA, RF, ANCA. Blood cultures (x3). Urinalysis (culture only if significant pyuria is present). Mantoux (TST) or IGRA. Chest radiography & abdominal ultrasound. If clues are identified, further laboratory testing (e.g. serology for inflammatory or infectious disease) should be pursued accordingly.

Step 3: Whole-Body Imaging Particularly useful for occult infections, inflammation, or malignancy. FDG-PET/CT is the preferred imaging modality, which has a higher diagnostic yield than CECT and identifies hypermetabolic activity, guiding targeted biopsies. CECT of the thorax, abdomen, and pelvis is more widely available than PET; and detects characteristic abnormalities. If no clue is detected with one imaging modality, consider alternative approach to localise disease activity for targeted tissue sampling.

Step 4: Invasive Procedures (a) Image (radiology or endoscopy)-guided tissue sampling (if imaging suggests localised pathology). (b) Bone marrow examination (if cytopenia, splenomegaly, or lymphadenopathy, or immunocompromised or older patients). Diagnostic in haematological malignancy, enteric fever, histoplasmosis, leishmaniasis, TB, cytomegalovirus, and HLH. Useful in auto-inflammatory conditions (e.g. Still's disease) where infection and malignancy must be excluded. (c) Splenic aspiration (if splenomegaly) or liver biopsy (if hepatomegaly with elevated alkaline phosphatase).

Step 5: Empirical Therapy Reserved for select cases with strong clinical suspicion or progressive deterioration; e.g. ATT in TB-endemic regions where diagnostics remain inconclusive. A positive Mantoux (TST) may aid decision-making. If no response within 6–8 weeks of ATT, reassess for an alternative aetiology.

Step 6: Advanced Testing Metagenomic next-generation sequencing (mNGS)   - Broad pathogen detection, including fastidious or unculturable organisms missed by traditional methods.   - Rapid turnaround time aids early detection, but high cost and risk of false positives.   - Useful excluding infections, so guiding immunomodulatory/anti-inflammatory therapy.

Step 7: Watchful Waiting If all investigations remain inconclusive and the patient is clinically stable, close monitoring and periodic reassessment should be performed. 10–20% of PUO cases remain undiagnosed, many resolve spontaneously.

Note: The ‘Stop, Look, and Go’ approach is essential. (a) Pause at each step to avoid premature empirical treatments or unnecessary testing. (b) Reassess history, clinical findings, and investigations for missed clues. (c) Proceed stepwise based on the most probable differential diagnoses.