Clinical features and outcomes of patients with fever of unknown origin: a retrospective study

In 1961, Petersdorf and Beeson formally proposed the definition of classic fever of unknown origin (FUO) by observing and summarizing a series of patients with unexplained fever as follows: a temperature > 38.3 °C on several occasions over a period of more than 3 weeks without a diagnosis despite 1 week of inpatient investigation [1]. At present, the aetiological classification of FUO includes infectious diseases, non-infectious inflammatory diseases, tumour diseases, other diseases and unknown diagnoses. Despite advances in medicine, the proportion of patients discharged with undiagnosed FUO after systematic examination has not decreased. Currently, the cause of febrile illness is not identified in approximately 9–51% of patients [2‐5]. The reason may be that with the improvements in the quality of medical care and diagnostic tools, such as the popularization of advanced imaging technology, improvement in pathogen culture technology, newly developed serological detection projects, and application of polymerase chain reaction (PCR) technology, diagnosis rates of common diseases are improved. However, cases that meet the classic definition of FUO are becoming increasingly complex [5, 6]. Only a few studies to date have reported the outcome of patients who are discharged with undiagnosed FUO [2, 6‐10]. In this study, patients who were diagnosed with FUO at admission and discharged without a final diagnosis in the department of infectious disease at Peking Union Medical College Hospital between 2004 and 2010 were followed for the purpose of investigating the clinical outcome of undiagnosed FUO.

Since the definition of FUO was first proposed in 1961, FUO has remained a challenge in the diagnosis and treatment of difficult diseases. At present, few studies have reported the outcome of patients with undiagnosed FUO. The conclusions of these studies are consistent; that is, the majority of patients without a diagnosis after discharge have a good prognosis. Knockaert et al. [7] reported the outcome of 61 patients with undiagnosed FUO, and 41 patients were finally cured. Fever was relieved in 31 patients during hospitalization or shortly after discharge; 10 patients with prolonged fever for several months or years finally recovered and were considered to be cured. In Vanderschueren’s study [8], including 80 cases of undiagnosed FUO, 63 cases were finally cured but still undiagnosed. Zenone et al. [10] reported the outcome of 37 patients who were discharged with FUO, and 29 patients eventually recovered. In this study, 58 patients who were discharged with undiagnosed FUO at Peking Union Medical College Hospital were followed. Approximately 60% of patients recovered after discharge and were eventually considered to be cured.

For patients whose clinical manifestation is mild or relatively stable and whose aetiology is still unclear after detailed examination, long-term follow-up can be used in outpatient clinics to observe the disease duration. Both previous studies and our study have found that the clinical course of most FUO patients without diagnosis is self-limiting and that they are ultimately considered to be cured after long-term follow-up. For patients whose clinical manifestations may be in the early stages of the disease, follow-up and observation are essential means for obtaining the final diagnosis. In Li’s study [11], which included 70 cases of FUO, a patient with prolonged hyperpyrexia and liver damage was positive for anti-SSA antibody at 46 months of follow-up, and he was finally diagnosed with Sjogren’s syndrome.

It is important to search for the cause of disease in patients whose condition has deteriorated or has not been alleviated. Additional biopsies must be performed in difficult cases, especially when malignant diseases are suspected, to improve the positive rate of diagnosis. In a case series study [12] that included 105 cases of FUO, a total of 186 biopsies (including bone marrow, liver, lymph nodes) were performed in 86 patients. Finally, the diagnosis of 40 patients was confirmed by biopsy. Vanderschueren et al. [8] reported the diagnostic methods of 192 cases of FUO. Biopsy pathology was the most valuable diagnostic method. Approximately 24.5% of the patients (47/192) were diagnosed by biopsy.

Lymphoma is the most common malignant disease-causing FUO and accounts for 50–70% of cases [13, 14], and lymphoma with fever has always been a difficult condition to diagnose. Shi et al. [15] reported changes in the aetiological composition of FUO in our hospital from 1985 to 2010, with the proportion of lymphoma gradually increasing (from 36.4 to 68.4%). In 2003, Mao et al. [16] summarized the diagnostic methods of 50 cases of non-Hodgkin’s lymphoma with fever in our hospital. Multiple bone marrow biopsies were performed in all 50 patients, and a final diagnosis was established in 10 cases. Twenty-seven patients received multiple lymph node biopsies, and 14 cases received a definitive diagnosis. Of the 14 cases, one was diagnosed at the 7th lymph node biopsy, and pathological evidence was obtained from 6 patients by biopsy in two ways. In 2006, Li et al. [17] retrospectively analysed the clinical data from 53 cases of lymphoma with fever in our hospital. Fifty-three patients underwent at least 154 pathological examinations (including biopsy of bone marrow, superficial and deep lymph nodes, nasopharynx, subcutaneous nodules, liver, and spleen) before the final diagnosis was established. Nine cases were confirmed by multiple bone marrow biopsies; 44 biopsies of superficial or deep lymph nodes were conducted; and 13 patients had definitive diagnoses. Three patients were diagnosed by nasopharyngeal mass, 3 were diagnosed by subcutaneous nodules, and 25 were diagnosed by laparotomy. In our study, 3 patients diagnosed with lymphoma received 10 biopsies. A patient with prolonged hyperpyrexia, sore throat, purulent sinonasal mucus and swollen tonsils was diagnosed with lymphoma after multiple biopsies, including bone marrow, tonsil and two pharyngeal masses. A patient with hyperpyrexia, cough, lymphadenectasis, liver damage, blurred vision and hearing loss was finally diagnosed with non-Hodgkin’s lymphoma after biopsy of bone marrow, liver and pharynx.

In our study, 2 patients were diagnosed with AID. The duration from febrile illness until diagnosis was 9 and 15 years. The clinical manifestations included prolonged intermittent fever, rash, arthralgia, myalgia, splenomegaly and advanced inflammatory markers. No evidence of infection, tumours or other connective tissue diseases was obtained, and they were finally diagnosed with AID by genetic screening. AID is a group of hereditary non-invasive inflammatory diseases that are mainly caused by gene mutation in the coding protein, leading to inherent immune disorders and causing a systemic inflammatory response. AID is characterized by fever, rash, arthralgia, arthritis, and eye lesions and can affect multiple organ systems. Inflammatory markers, such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), are elevated, and no autoantibodies or only low titres of autoantibodies exist during disease flares [18]. AID is being increasingly recognized, and the genetic diagnosis of AID is being used more widely. However, the known mutation sites of AID-related genes still cannot be detected in approximately 60% of patients [19]. For patients whose clinical characteristics are consistent with AID and cannot be explained by infection or tumours, AID should be suspected, and genetic screening should be considered to assist diagnosis.

Clinical diagnoses can be performed when patients’ clinical characteristics are consistent with some specific diseases, the treatment is effective, no evidence of other diseases has been obtained after systematic examination, and the condition is stable after long-term follow-up. In this study, a patient with hyperpyrexia, headache, arthralgia, myalgia and lymphadenectasis had a 5-week febrile period before admission. Considering that he was more likely to have a viral infection according to the clinical manifestation and auxiliary examinations, NSAIDs were given for treatment, and his febrile illness was relieved. He had a self-limited clinical course (7.4 w) after long-term observation and was finally diagnosed with viral infection. Three patients whose clinical manifestations met ASD criteria demonstrated no evidence of infection or tumour after detailed examination. Glucocorticoid and antirheumatic drugs were effective, their condition was stable after prolonged follow-up, and they were finally diagnosed with ASD. In a retrospective study [20] of 153 patients with FUO, 53% of patients diagnosed with connective tissue diseases were clinically diagnosed. After exclusion of infections and tumours, these patients were eventually diagnosed with connective tissue disease based on their clinical features and response to diagnostic treatment.

In this study, only 19% of patients (11/58) received a definitive diagnosis after long-term follow-up. The cause may be that most patients without a final diagnosis have a good prognosis. Furthermore, the logistic regression method was used to evaluate whether age, sex, time of fever before admission, different system damage and the amount of damage were risk factors related to death in 58 patients discharged with undiagnosed FUO; however, positive results were not obtained. Then, 11 patients with definite diagnosis were removed, and whether the above factors were related to death was evaluated in 47 patients without a definite diagnosis; again, positive results were not obtained. The limited sample size may have resulted in the negative results.

Several limitations were observed in this study. First, this is a retrospective study with a small sample size in a single centre. Patients were followed by telephone, and more detailed information about diagnosis and treatment could not be obtained. Second, of the 200 patients discharged with undiagnosed FUO between 2004 and 2010, follow-up data were available for only 58. Thus, the results were not representative, and some bias occurred. The reason for the high rate of missing cases may be that our institution is a centre for complicated and severe diseases in China and that most of the patients are non-local residents. Patients in our study were enrolled between 2004 and 2010, and a complete follow-up model was not established at the time except for the telephone number. More than 10 years had passed and most of the patients or their families changed their contact information; thus, they could not be followed by telephone. Currently, only a few studies [2, 7‐10] have reported the outcomes of patients discharged with undiagnosed FUO, and they have a small sample size and shorter follow-up duration (As shown in Table 4). In our study, the sample size was relatively large, and the information obtained was more accurate for each patient with follow-up data because of the long follow-up duration.

In conclusion, the illness is self-limited in most patients, and their prognosis is positive. Identifying the cause of disease in patients whose condition is deteriorated or not alleviated is important. In difficult diseases, multiple repeated invasive examinations are necessary to obtain a diagnosis.