Article Text

Download PDFPDF

Pyrexia of unknown origin
  1. M Wood1,
  2. M Abinun2,
  3. H Foster3
  1. 1Royal Victoria Infirmary, Newcastle Upon Tyne, UK
  2. 2Newcastle General Hospital, Newcastle upon Tyne, UK
  3. 3Departments of Rheumatology and Child Health, University of Newcastle upon Tyne, Newcastle upon Tyne, UK
  1. Correspondence to:
    Dr Mark Wood
    Royal Victoria Infirmary, Queen Victoria Road, Newcastle Upon Tyne, NE1 4LP, UK;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

One week after receiving the measles, mumps, and rubella (MMR) immunisation, 14 month old Kate developed a non-pruritic “blotchy” erythematous macular rash. The rash was present over the proximal areas of her arms and legs and was most noticeable during the afternoon. Initially she remained otherwise well, but six days later developed coryzal symptoms and lethargy.

Kate’s general practitioner (GP) was consulted. There was no significant past medical or family history. She was taking no regular medication and had no known allergies. On examination, apart from the rash and a pink throat, she appeared normal. He prescribed a course of amoxicillin.

The GP thought that a viral upper respiratory tract infection was the most likely diagnosis but nevertheless prescribed amoxicillin just in case the rash was caused by a bacterial infection. He also considered the possibility of a reaction to the MMR immunisation, but thought this was unlikely in view of the long history.


  • About one week after the first dose of an MMR immunisation it is common to have symptoms of malaise, fever, and a rash, usually lasting 2–3 days. There is also an increased frequency of febrile convulsions at this time. These effects are likely to relate to the measles component and are less common after the second MMR immunisation.1

The next day Kate developed fever and diarrhoea (without blood or mucus). She also became unsettled and more lethargic. The intermittent fever and rash continued. Her mother and grandmother also developed diarrhoea. The GP was again consulted and a referral was made to the local hospital paediatric unit.

In view of the history of diarrhoea in family contacts, the GP considered the most likely diagnosis to be infectious gastroenteritis rather than a side effect of the amoxicillin. He was particularly concerned because Kate looked unwell.

On admission to hospital Kate’s axillary temperature was 38.7°C and a maculopapular rash was noted (fig 1 ). The senior house officer assessed Kate and noted a pink throat but no other signs to indicate a focus of infection. Kate was well perfused with no signs of meningitis, but was crying inconsolably at times. There was no hepatosplenomegaly or lymphadenopathy. Examination was otherwise normal. Initial investigations were arranged (table 1 ).

Table 1

 Initial investigations

Figure 1

 The maculopapular rash that affected the arms and legs but not the face. Reproduced with permission of the child’s parents.

The senior house officer was concerned about the unexplained fever, rash, neutrophilia, and raised C reactive protein (CRP) and so discussed Kate with the registrar on call. They agreed to start intravenous cefotaxime as broad spectrum cover for a possible invasive bacterial infection.


  • In young children intravenous antibiotics should be started promptly if invasive bacterial disease (such as meningitis or septicaemia) is suspected to minimise the risk of serious complications. Without a clear diagnosis, the organisms that need to be covered include Neisseria meningitides, Escherichia coli, Streptococcus pneumoniae, and Haemophilus influenza type b. These are usually all sensitive to cefotaxime.

On the paediatric ward observations every four hours did not suggest any circulatory compromise and it was noted that Kate looked well when apyrexial. The following day the consultant paediatrician reviewed her on the ward round. At that time he found Kate to be apyrexial with no rash.

After reviewing the history, examination and investigation results a viral illness was thought to be the most likely diagnosis, but it was decided to continue the cefotaxime pending receipt of microbiology results due 48 hours after admission.

Later that day Kate had further “spikes” of temperature to over 38°C. These were associated with a pronounced rash over her inner thighs and arms, with sparing of her face. When pyrexial she looked unwell with anorexia and lethargy, and inconsolable crying. However, when apyrexial she looked brighter and the rash became faint.

On the morning ward round three days after admission Kate continued to have spikes of fever and appeared uncomfortable. The blood culture was negative and no pathogenic organisms had been grown from the stool specimen and throat swab. Further investigations were ordered: chest x ray (normal), lumbar puncture (normal cerebrospinal fluid (CSF) results), CRP (94 mg/l), white cell count (17 × 109/l; neutrophil count 12 × 109/l), stool for viral culture, repeat stool for microscopy and bacterial culture, blood for anti-streptolysin titre (ASOT), and serology for Epstein-Barr virus (EBV) and enterovirus. Cefotaxime was continued.

The consultant paediatrician still considered a viral illness to be the most likely diagnosis, but in view of the prolonged pyrexia, rash, and history of irritability he wanted to exclude other infections. He knew that meningitis and pneumonia can be difficult to diagnose by clinical examination alone in a young child. At the same time he was reassured that some of the time Kate looked well and by the fact that the CRP and white cell count had not risen. He decided that continued observation was the best course of action.

Six days after admission no specific infection or other cause of the illness had yet been identified. Kate continued to have spikes of fever (fig 2 ) and diarrhoea. The opinion of a consultant in paediatric infectious diseases and immunology was requested.

Figure 2

 Temperature chart.

This consultant reviewed the history including details of immunisations, travel, pets, parents’ occupations, family, and any contact with infectious diseases. The irritability, diarrhoea (improving) and temperature chart were noted. On examination (while she was apyrexial and looking well) he noted the very faint generalised erythematous macular rash. There was no hepatosplenomegaly or lymphadenopathy and no heart murmur. Areas of mild skin peeling were noted on the right side of Kate’s neck, right thumb and the tips of her left fingers.

The consultant reviewed the results and noted the raised erythrocyte sedimentation rate (ESR), mildly raised CRP, and thrombocytosis with no evidence as yet of any infectious organisms. He advised that the cefotaxime could be stopped since there was now little evidence of a bacterial infection.

He could not make a firm diagnosis, but drew up a list of possible causes and relevant investigations designed to distinguish between them (table 2). The clinical features were unchanged during the next 48 hours. These results were normal except for a serum ferritin concentration of 762 μg/l (normal < 40 μg/l).

Table 2

 Differential diagnoses considered and the investigations requested

The most likely diagnosis was thought to be either incomplete Kawasaki disease or systemic onset juvenile idiopathic arthritis (SOJIA). The pros and cons of each possibility were considered (table 3) and on balance it was thought that incomplete Kawasaki disease was the most likely.

Table 3

 Evidence considered in the differential diagnosis


  • A raised serum ferritin concentration is a non-specific finding in several conditions (table 4). A concentration > 1000 μg/l is unusual and may indicate haemophagocytic syndromes, SOJIA, haemochromatosis, liver disorders, and malignancy. A normal serum ferritin concentration would make SOJIA very unlikely.

Table 4

 Common causes of a raised serum ferritin concentration

Oral aspirin (100 mg/kg/day in four doses) was started and Kate received immunosuppressive intravenous immunoglobulin (IVIG) at a dose of 2 g/kg eight days after admission. Counselling was given regarding the risks associated with IVIG and consent to treatment obtained. Blood was taken for storage before IVIG was administered in case future testing was necessary.


  • IVIG is a blood derived product and, despite stringent testing of donors and the product, there remains a very small risk of transmitting blood borne viruses. The patient’s family should also be aware of the theoretical risk of contracting blood borne infections that may, as yet, be unrecognised.

  • For several months after the administration of IVIG serology testing of the patient will be altered by the treatment.

During the next five days spikes of temperature over 38.5°C continued, usually in the evenings.

The consultant knew this was unusual for Kawasaki disease and so reviewed the evidence in the literature to see if any particular treatment would be more appropriate. He found a retrospective study in which over 85% of children treated with IVIG within 10 days of onset of symptoms of Kawasaki disease remained pyrexial 48 hours after treatment started.2 He found anecdotal reports of re-treatment with IVIG and pulses of steroid, but also read a recent review that stated that no comparative trials have been reported.3 He decided to repeat the IVIG.

The IVIG (2 g/kg) was repeated 15 days after admission. The next day Kate was brighter and for the first time since admission she remained apyrexial for 12 hours. However, on day 17 the spikes of fever and rash returned.

The diagnosis of incomplete Kawasaki disease was still thought to be the most likely, but there was concern that occult malignancy may have been missed. A bone marrow aspirate was therefore arranged which was normal. In view of the poor response to the second dose of IVIG, it was decided to commence corticosteroids.


  • A bone marrow aspirate should be considered in the investigation of pyrexia of unknown origin (PUO) even if the peripheral blood film is normal, especially before corticosteroids are given as this treatment can mask leukaemia.

Kate received intravenous methylprednisolone (IVMP) infusions of 30 mg/kg/day for three days starting on day 19. Her fever and rash rapidly subsided, she became more comfortable, and her inflammatory markers fell. The aspirin dose was reduced to 5 mg/kg/day.

Unfortunately her symptoms flared again five days later causing great anxiety to her parents who could see no end to the problems. Further IVMP pulses over three consecutive days were given and produced symptomatic relief. However, after a further five days the pyrexia returned.

Options considered at this time were to give further IVMP pulses or to start oral corticosteroids. In view of difficulties with venous access and the lack of a sustained response, the latter option was taken.

Oral prednisolone was started (35 days after admission) at 2 mg/kg/once a day. On day 38 Kate refused to weight bear and her parents noted an intermittent swelling of her hands and feet. Examination revealed synovitis of the right wrist, both ankles and the first, second, and third proximal interphalangeal joints of the right hand. At this point she was referred to a paediatric rheumatologist.

The paediatric rheumatologist considered the following salient clinical features:

  • The chronic nature (six weeks) of the systemic symptoms and the timing of onset of arthritis was not typical of Kawasaki disease, in which any arthritis usually occurs in the acute phase.

  • The temperature pattern with a return to the baseline in between “spikes” of fever occurring on a daily basis (quotidian pattern) is typical of SOJIA.

  • The rash was more pronounced when she was pyrexial and was more evident on her inner thighs and arms and spared her face. This pattern is more typical of SOJIA than Kawasaki disease.

  • The raised serum ferritin concentration, neutrophil leucocytosis, absence of autoantibodies, and the lack of evidence suggesting an infection or malignancy were all consistent with a diagnosis of SOJIA.

A diagnosis of SOJIA was made and Kate was continued on the prednisolone and started on regular ibuprofen (40 mg/kg/day in divided doses). From day 40 she became more comfortable and less miserable, with an improvement in inflammatory markers including serum ferritin (fig 3 ).

Figure 3

 Serum ferritin concentrations.

Kate was referred to the paediatric rheumatology multidisciplinary team (physiotherapy, occupational therapy and a clinical nurse specialist), which provided a programme of physical therapy and support for her and her family. She was examined for uveitis by a paediatric ophthalmologist and this was normal. No IgG to varicella was detected in the serum saved since before the IVIG and therefore advice was given regarding the need for intravenous aciclovir should spots develop after chickenpox exposure.

An echocardiography (planned when the working diagnosis was Kawasaki disease) was performed eight weeks after initial admission and was normal (no coronary artery aneurysms). By this time she had developed mild Cushingoid features but no glycosuria or hypertension.

Kate remained on prednisolone (1 mg/kg on alternate days) and received regular ibuprofen. Oral methotrexate (steroid sparing agent) and folic acid were prescribed three months after the original admission. Chickenpox vaccination was considered but not given since the level of immunosuppression was likely to interfere with the response. Regular blood tests (full blood count, inflammatory markers and liver function tests) were arranged to monitor for efficacy and features of methotrexate toxicity.


The arthritis and systemic symptoms were well controlled on the prednisolone and methotrexate. Corticosteroids were gradually tapered and stopped over six months. Kate was unable to tolerate methotrexate by mouth because of nausea so this was changed to the subcutaneous route. This was well tolerated and improved the control of her disease, with the return of her inflammatory markers (including serum ferritin) to normal values. Liver function tests and neutrophil counts remained normal throughout.

Over the subsequent two years Kate had three flares of systemic symptoms which were controlled with IVMP pulses and tapering doses of oral steroids. Her joints have shown little evidence of active arthritis. She attends a playgroup where she joins in activities with her peers. Her parents understand that she has a rare form of JIA with a variable prognosis and requires long term follow up.


There is controversy regarding diagnostic criteria for PUO.4–,6 Essentially this is a term for a prolonged (more than one week) pyrexial illness when the history, examination and all “standard investigations” (table 5) fail to indicate the diagnosis. The history needs to include details of immunisations, travel, pets, occupations of family members, family contact with infectious diseases and, if appropriate, sexual history. A list of the common and rare causes of PUO and their relevant clinical features and discriminatory investigations are shown in table 6.

Table 5

 Suggested “standard investigations” to be performed before the situation is regarded as a PUO

Table 6

 Causes of pyrexia of unknown origin


Kawasaki disease is an important cause of cardiovascular morbidity and mortality both in the acute phase (myocarditis or pericarditis) and long term (thrombosis of coronary artery aneurysms and atherosclerosis). Management with high dose aspirin and immunosuppressive IVIG reduces the incidence of cardiac complications if given within 10 days of the onset and probably if given later.7 The diagnosis of Kawasaki disease is made when the described criteria8 are met (table 7). The pyrexia is generally unremitting (that is, not returning to the baseline) and rarely lasts more than 15 days. Characteristic features usually occur from the third day of illness. Other features can include diarrhoea, sterile pyuria, arthritis, uveitis, aseptic meningitis, hepatitis, and hydrops of the gall bladder. Children are typically miserable. Acute phase reactants are raised as is the ESR, and typically there is a neutrophil leucocytosis and thrombocytosis. An echocardiogram in the acute phase can show evidence of myocarditis or pericarditis, which can cause heart failure or arrhythmias. Cardiac aneurysms take longer to develop and echocardiography should be performed at three weeks and repeated at 8–12 weeks after the onset of illness.

Table 7

 Diagnostic criteria for Kawasaki disease

Over recent years it has been recognised that cardiac complications can also occur after illnesses that do not fulfil the prerequisite number of criteria. This is termed “incomplete Kawasaki disease” and it has been shown that immunosuppressive IVIG and high dose aspirin, given in the same regimen as for classical Kawasaki disease,9 can reduce associated cardiovascular morbidity and mortality. Features most commonly occurring in incomplete Kawasaki disease (other than the classical criteria) are irritability, diarrhoea, sterile pyuria, and coryzal symptoms.10



Chronic arthritis in children is a heterogeneous group of diseases of mainly unknown aetiology. They are mainly clinically and genetically distinct from causes of chronic arthritis in adults. The latest classification uses the unifying term juvenile idiopathic arthritis (JIA) to describe arthritis of no known cause which persists for at least six weeks in a child under 16 years old.11,12 This replaces the old terms “juvenile chronic arthritis” and “juvenile rheumatoid arthritis”.

The classification lists the following seven subtypes:

  • Oligoarticular onset JIA (persistent or extended)—Arthritis affecting up to four joints during the first six months of disease. If subsequently more than four joints are affected the term “extended oligoarthritis” is used, otherwise the term “persistent oligoarthritis” is used. This is the most common pattern (50% of all JIA) and usually involves large joints of the lower limbs, especially knees. These children have the best prognosis but are at high risk of asymptomatic uveitis and therefore must be screened regularly.

  • Polyarthritis (rheumatoid factor negative)—Arthritis affecting five or more joints during the first six months. If IgM rheumatoid factor is not detected in blood then the child’s disease falls in this group, which makes up 17% of all JIA. Severity is very variable.

  • Polyarthritis (rheumatoid factor positive)—This subtype accounts for 7% of cases and is characterised by a symmetrical polyarthritis, nodules, and IgM rheumatoid factor being identified at least twice three months apart. Patients are typically adolescent girls and the prognosis is guarded as early joint damage often occurs.

  • Systemic onset—SOJIA accounts for 11% of cases of JIA. It can occur at any age and often affects pre-school children but rarely in infancy. Similar incidence rates are seen in males and females.

  • Enthesitis related arthritis—This term is used for arthritis and enthesitis (inflammation of tendon insertions). The group also includes arthritis or enthesitis with at least two of:

    • tenderness of the sacroiliac joint and/ or inflammatory spinal pain

    • HLA B27

    • family history in a first or second degree relative of HLA B27 related disease

    • anterior uveitis (usually symptomatic with redness, pain and blurred vision)

    • arthritis after 8 years of age in a boy (usually affecting large lower limb joints.

The distal large joints in the lower limbs are commonly affected with an asymmetrical presentation and there is a high risk of these patients developing ankylosing spondylitis in early adulthood.

  • Psoriatic arthritis—Children with psoriasis and arthritis. This group also includes children with arthritis or psoriasis and at least two of:

    • dactylitis

    • pitting or onycholysis of nails

    • psoriasis in a first degree relative

  • Other arthritis—This group is for children with idiopathic arthritis that does not fit the other groups.

Systemic onset juvenile idiopathic arthritis

The main features of SOJIA are summarised in table 8. Patients can be systemically very unwell and potentially life threatening complications may occur early in the disease course (for example, pericarditis, macrophage activation syndrome, sepsis). Prompt treatment with high dose corticosteroids can therefore be life saving. The decision to start immunosuppressive treatment has to be made after careful exclusion of other diagnoses, especially infection, Kawasaki disease, and malignancy, and is particularly difficult when definite evidence of arthritis is absent.

Table 8

 Key features of systemic onset juvenile idiopathic arthritis (SOJIA)

Systemic features may predate the arthritis by several weeks and occasionally longer. The arthritis typically involves the small joints of the hands and wrists, ankles, hips, knees, and cervical spine and about a third of children ultimately develop a severe polyarthritis. The physical signs of arthritis may be masked by corticosteroids given for acute systemic features (for example, carditis) and are often overlooked in the initial assessment. Musculoskeletal assessment does not have a high profile in undergraduate or postgraduate teaching and many paediatricians lack confidence in their ability to perform a competent musculoskeletal assessment.13

There are no pathognomic tests or agreed diagnostic criteria for SOJIA and so it remains a diagnosis based on the identification of recognised features and exclusion of other conditions. As in Kate’s case, the initial differential diagnosis may be wide and there can be a delay of several weeks before the diagnosis is established. However, repeated careful examination of the temperature chart is helpful, especially in the early stages when arthritis is absent. The presence of a recurring evening spiking temperature, that returns to or falls below baseline in between spikes, is classical for SOJIA. Pyrexial episodes are also the most fruitful times for eliciting the rash, which is erythematous and maculopapular, often asymptomatic, and typically spares the face and extremities and can therefore be easily missed. The rash is typically on the inner thigh and arms, especially on areas of trauma or pressure (Koebner phenomenon). In the assessment of PUO, nursing staff and paediatricians must be encouraged to examine the undressed child when pyrexial. A photograph of the rash can be very useful as it may be transient.

Management of JIA

JIA is a not a benign disease and outcome is variable.14 At least a third of patients have ongoing active disease into adulthood and many have sequelae of the disease or treatment—for example, joint damage requiring joint replacement, short stature from chronic disease compounded by steroid toxicity, localised growth problems (micrognathia or leg length inequality), visual loss from uveitis, and psychosocial morbidity with high unemployment. Optimal management requires early intervention by an experienced and coordinated multidisciplinary team. This should include input from a nurse specialist, physiotherapist, occupational therapist, paediatric rheumatologist, ophthalmologist, clinical psychologist, social worker, dentist, orthodontist, and orthopaedic surgeon. Transitional care for the adolescent and young adult needs to be well coordinated.

Corticosteroids are important agents, although there is increasing concern regarding their side effects including growth retardation, osteoporosis, and cataracts. Therefore exposure to systemic corticosteroids is minimised where possible by increasing use of intra-articular steroids and early use of DMARDs—disease modifying anti-rheumatic drugs (especially methotrexate). Calcium and vitamin D supplements are often given to children receiving corticosteroid treatment. Intra-articular corticosteroids (triamcinolone hexacetonide/acetonide) are usually given under sedation or general anaesthetic in young children or with entonox in older children. Joint injections with corticosteroid are a highly effective treatment, appear to be very safe, and increasingly multiple joints are injected at one session with ultrasound imaging used to deliver the drug more accurately to relatively inaccessible joints.

Intravenous methylprednisolone is used for severe polyarthritis or active SOJIA and is very useful as a disease remitting agent while starting methotrexate treatment which can take several months to be effective.

Methotrexate is the disease modifying drug of choice and there is evidence that early use helps to reduce joint damage and minimise the exposure to, and side effects of, corticosteroids. Long term studies show that methotrexate is beneficial and well tolerated in most children with side effects generally limited to occasional nausea, mild (reversible) elevations of serum liver enzymes, and mild bone marrow suppression. Doses up to and even beyond 1 mg/kg (30 mg/m2) are used with very few reports of complications. The theoretical risk of malignancy and infertility has not so far been borne out in long term outcome studies. Folic acid supplementation is usually given and improves tolerability. Methotrexate is given once a week and increasingly by the subcutaneous route to improve bioavailability. Regular blood tests are needed to monitor for efficacy and side effects. Live vaccines should be avoided and in the non-immune patient intravenous aciclovir should be administered promptly at the first sign of chickenpox.

Patients who are refractory to high dose parenteral methotrexate are considered for novel immunosuppressive agents such as biologics (for example, soluble anti-TNF α receptor etanercept), autologous stem cell transplantation, or very high dose immunosuppression.


  • SOJIA should be considered as a cause of PUO

  • Arthritis can be minimal or even absent initially (and for several weeks) in SOJIA

  • Persistent pyrexia returning to baseline is suggestive of SOJIA

  • A rash that is more pronounced when pyrexial is typical of SOJIA

  • Involvement of a paediatric rheumatologist in the investigation of a PUO is recommended