So what is a PUO? How should it be managed? Why does the management of PUO often go wrong?
What is pyrexia of unknown origin?
There is a specific definition of PUO:
A temperature >38°C for >3 weeks
PLUS >2 visits to hospital OR >3 days of investigation in hospital
- Some doctors do not actually know what a PUO is
- The investigations of PUO are often haphazard and chaotic with common tests being missed and uncommon tests being repeated many times!
What are the common causes of PUO?
Infection is only part of a list of differential diagnoses which can cause PUO. Infection causes 27-50% of PUO e.g. abdominal or pelvic abscess, infective endocarditis, tuberculosis and urinary tract infection.
Some doctors forget about the other causes of a fever and continue to try and prove a diagnosis of infection despite all results coming back negative. This is a dangerous strategy for 2 reasons:
- If you do enough tests eventually 1 will be positive, unless you are targeted in your approach you will run the risk of a false positive result and miss the actual diagnosis e.g. a test with a 97% positive predictive value is considered good but 3 in 100 people will have a positive test even though they don’t have the disease
- Failing to investigate for other causes of fever leads to missed diagnoses and missed opportunities to treat the patient correctly. The common causes of PUO after infection include:
- Malignancy (13-25%) – lymphoma, leukaemia, renal cell carcinoma, hepatocellular carcinoma or liver metastases
- Connective tissue disorder (9-17%) - Still’s disease, Systemic Lupus Erythematosus (SLE), Rheumatoid arthritis, Temporal arteritis, Polymyalgia rheumatic
- Other diagnosis (15-21%) – Drug reaction, venous thrombosis, haematomas, hyperthyroidism
- Undiagnosed (5-23%)
So, infection, malignancy and connective tissue disorders account for up to 80% of PUOs and therefore investigations should be targeted at these diagnoses.
Diagnostic approach for the investigation of PUO
It is really important that patients are not routinely started on antibiotics just because they have a fever. Antibiotics will prevent bacteria growing in cultures and make diagnosing infections very difficult, including infection with M. tuberculosis against which many commonly used antibiotics are active e.g. beta-lactamase inhibitor combinations, carbapenems, fluoroquinolones, aminoglycosides and macrolides. It is also important to try and withhold antipyretics as these can delay or cover up symptoms and signs that may aid the diagnosis. Septic or febrile neutropaenic patients do need treating but almost all other patients can be safely investigated first before rushing in with treatment.
My personal approach to the investigation of PUO is split it into stages (a handy postcard summary here to print and put into your copy of Microbiology Nuts and Bolts). I would first investigate for common infections as most of the time PUOs are caused by common things presenting in uncommon ways. Essentially this is an initial investigation of pyrexia rather than specifically PUO and aims to rule out an easily treatable infection that might have been missed. Acute phase reactants such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) are used to help predict the likelihood of infection and inflammation but they are not fool proof. A normal CRP and ESR makes significant inflammation unlikely however 10-20% of patients can have normal inflammatory markers and still have serious inflammatory diseases causing a fever.
My second line investigations target the more specific causes of PUO listed above. Tuberculosis, chronic viral infections, hepatitis, abscesses, malignancies and connective tissue disorders are the main focus of my investigations. A CT scan of the chest, abdomen and pelvis is considered to look for TB, abscesses and malignancies which can then be biopsied. Along with this is a series of blood tests looking for TB, chronic viral infections, hepatitis and connective tissue disorders (see diagram below for specifics).
My third line investigation is a CT PET scan. Most of the time you can get a diagnosis before you get to this stage. This scan looks for metabolically active tissue which includes inflammation (infection or connective tissue disease) and malignancy. Whilst this scan doesn’t give a definitive diagnosis it is very useful for allowing biopsy of the most metabolically active target and therefore gives the best chance of finding the cause. CT PET is not routinely done at present in the UK, as it is not widely available and takes time to arrange.
Biopsies of possible target lesions in the body are often helpful, but it is important to remember that even this type of investigation can still give false positive results. In some studies the use of biopsy in PUO gave a diagnosis in about 33% of patients however the false positive rate was also 33%... that’s huge! Bone marrow biopsies were only helpful in 10% of patients who underwent this painful procedure which makes me think it is not worthwhile unless imaging suggests the site of the biopsy is abnormal in the first place e.g. a lytic lesion on CT scan or a bright signal on CT PET.
1st line investigations
- History & examination including - localizing symptoms e.g. right upper quadrant pain indicating possible liver pathology, travel, animal contact
- Drug history (prescribed and not prescribed)
- Blood tests – Daily FBC, U&Es, LFTs, CRP (+/- ESR)
- Midstream urine (MSU) – MC&S
- Blood cultures – 3 sets over 24 hours for bacteraemia or infective endocarditis
- Chest X-ray
- If localizing symptoms or signs consider CT chest, abdomen and pelvis +/- biopsy
- DO NOT give empirical antibiotics until a diagnosis is made
2nd line investigations if no diagnosis after 3-7 days
- Repeat history & examination
- Tuberculosis – Sputum +/- early morning urine cultures, IGRA (for latent TB)
- Viral infections – Blood for HIV, CMV, EBV, and Hepatitis A, B and C (if LFTs abnormal)
- Infectious mononucleosis – Blood for Monospot or Paul Bunnell in under 30 year olds
- Haematological malignancy – LDH, Ferritin
- Connective tissue disorder – Rheumatoid factor, antinuclear antibodies (ANA), antineutrophil cytoplasmic antibodies (ANCA), Ferritin
- Localization of tissue for biopsy – CT chest, abdomen and pelvis +/- biopsy
3rd line investigations if no diagnosis after 2 weeks
- Consider CT PET scan to look for potential diagnostic biopsy target
- Consider echocardiography if infective endocarditis strongly suspected
- If foreign travel or unusual occupational/recreational exposure discuss further investigations with a Microbiologist
So what do you do if you can’t find a cause of a PUO?
Some doctors worry if they are unable to find a cause for a fever in a patient, but really they shouldn’t. If all of the above tests are normal and the patient is otherwise well then they can be followed up as an outpatient to see if there is a pattern of the fever or any change in their symptoms and signs which might lead to a diagnosis. The risk of the patient coming to harm with this strategy is low. In a recent study from the Netherlands 50% of patients with PUO were discharged from hospital without a diagnosis; of these 60% settled spontaneously or with steroids, 39% had ongoing fevers and 1% died. It is likely that the risk of death could be reduced even further if the investigations above are repeated at some stage during the follow up period.
So our patient with a fever had his antibiotics stopped. Over the next three days he had the 1st line investigations in the flow chart above. His white blood cells and CRP were raised. His chest X-ray was normal. His blood cultures did not grow any bacteria but his urine had a high white cell count and grew an E. coli. He was given targeted therapy for his UTI with Trimethoprim for 5 days and made a full recovery. Simple really…