Pyrexia of unknown origin: a prospective study of 100 cases.D Kejariwal, N Sarkar, SK Chakraborti, V Agarwal, S Roy
Department of Medicine, Institute of Post Graduate Medical Education and Research and SSKM Hospitals, Calcutta, India. , India
AIM: There are few studies on pyrexia of unknown origin (PUO) from India. The present study was planned to elucidate the causes of in PUO Eastern India and to define the changing patterns of PUO, if any. STUDY DESIGN: Prospective case series. PATIENTS: One hundred patients meeting the classic criteria of pyrexia of unknown origin. MAIN OUTCOME MEASUREMENT: The final diagnosis established at discharge or during follow up. RESULTS: Infections, especially tuberculosis was the most dominant cause (53%), followed by neoplasms (17%), and collagen vascular disorders (11%), Miscellaneous causes were responsible in 5% cases and in 14% the cause of fever remained undiagnosed. CONCLUSION: It is concluded that infections remain the most important cause of PUO in India, confirming the trends found earlier in other studies. The incidence of neoplasms was much higher compared to other studies from India.
Keywords: Adolescent, Adult, Aged, Child, Female, Fever of Unknown Origin, epidemiology,etiology,Human, India, epidemiology,Infection, complications,Male, Middle Age, Neoplasms, complications,Prospective Studies,
Pyrexia of unknown origin (PUO) is a challenging medical problem. While dealing with such problems one has to make critical use of all the available clinical and investigative methods keeping in mind the local epidemiological conditions.
Petersdorf and Beeson defined PUO as an illness with temperature exceeding 38.3 degree centigrade, evolving over a period of at least three weeks, with no diagnosis reached after one week of inpatient investigation. The spectrum of diseases found in several series examining PUO shows some variation, but overall, infections continue to be the most important cause of PUO accounting for about 20-40% of cases, followed by neoplastic lesions, collagen vascular disorders and other rare illness.,,,,,, The spectrum of diseases causing PUO not only seems to be determined by geographical factors, but also appears to change with time. Scanty data on PUO from northern India suggests preponderance of infections as the cause of PUO and infrequency of neoplasms, connective tissue diseases as a calls after factor etc., Accordingly, we conducted this prospective study to find out the spectrum of diseases causing PUO in Eastern India.
Patients with PUO admitted in the department of General Medicine in our tertiary referral centre during a three year period from May 1998 to April 2001 were included in the study. Only those patients who fulfilled the Petersdorf and Beeson criteria were included in the study (illness characterized by temperature exceeding 38.3 degree centigrade evolving during at least three weeks, with no diagnosis after one week of inpatient investigation). Patients with neutropenia (neutrophil count less than 500/ml), patients developing fever after hospital admission and human immunodeficiency virus (HIV) positive patients were excluded from the study. Further, as our department looks after only adults, patients below the age of 10 years could not be included in the study.
After informed consent, patients were included in the PUO protocol, which consisted of a standardized precoded history and standardized thorough physical examination. Routine laboratory investigations such as complete haemogram, urine examination and culture, peripheral blood smear for malarial parasite, liver function tests, sugar, urea & creatinine, Widal test, Blood culture, chest X-ray and ECG were performed before including the patients in the PUO protocol. In the present study much help was obtained in the choice of investigations from potentially diagnostic clues (PDCs) which are defined as all localising signs, symptoms or abnormalities potentially pointing towards possible diagnosis. In few cases, however, PDCs proo red to be false causing inadvertent delay in the diagnosis.
The causes of PUO included five categories- infections, malignancies, Connective tissue diseases, miscellaneous conditions and undiagnosed cases. PUO was classified as undiagnosed if no evidence of the cause of fever was obtained and if there was complete spontaneous recovery even though the fever had persisted for several weeks or months.
A total of 100 patients were included in the study. There were 59 males and 41 females. The mean age of the patients was 32.4 years (range 12-65 years). The mean duration of fever before admission to the hospital was 85.3 days. Sixty-one cases were referred, of which 33 cases were referred by private practitioners and 28 cases from other hospitals. The cases of PUO are listed in [Table - 1]. [Table - 2] shows the procedures by which the diagnoses were established. Non- invasive tests were able to establish a diagnosis of PUO in 67.5% of our patients while invasive methods were required to diagnose the disease in 32.5% of our patients.
Infections were the commonest cause, being responsible for 53% of cases of PUO. Tuberculosis was the single most common infection. The mean duration of fever in patients with tuberculosis was 107.7 days with three patients having fever of more than six months duration. There was no case of pulmonary tuberculosis presenting as PUO. The diagnosis was delayed despite a high index of suspicion because of normal chest roentgenograms, negative skin test, inconclusive imaging and histopathological examination. In seven patients abscesses were the cause of fever. Four abscesses were intraabdominal and one each was located in the lung, prostate and kidney. The delayed diagnosis in these patients was due to presence of false PDCs, faulty interpretation of an abnormal chest roentgenogram, inconclusive ultrasound examination and failure to order pelvic ultrasonography because of the absence of localizing symptoms or signs. Endocarditis was found in five patients. Culture negative endocarditis occurred in three patients and in two patients cultures became positive when empiric antibiotic therapy was stopped. There were cardiac findings in each of these five patients, but was considered insignificant and ignored. This problem of endocarditis conveyed a good lesson to us that one should not depend on negative blood culture to rule out endocarditis and also insignificant murmurs should not be ignored when possibilities of endocarditis are in mind. Echocardiographic examinations were really helpful to us in these cases. Presence of false PDCs contributed to the delay in diagnosis. The diagnosis of a simple disease like enteric fever was missed early on in five patients because of its non-response to ciprofloxacin and negative blood cultures. Repeated blood cultures and rising titres in Widal test was used to arrive at a diagnosis. In four patients urinary tract infection (UTI) turned out to be the cause of the fever. All four patients were males and had received antibiotics for other presumed infections at the time of the first urine culture. In three patients, urine cultures yielded positive results eventually (Klebsiella pneumoniae in two, Proteus in one). In the fourth patient, recurrent prostatitis was found by transrectal sonography and culture of prostatic secretion yielded E.coli . The diagnosis of Kalazar was made in five patients by detecting the presence of LD bodies in bone marrow or splenic aspirate. Malaria was responsible for PUO in three patients. The diagnosis was delayed because of repeatedly negative peripheral blood smears and non-response to empiric chloroquine therapy.
In 13 patients haematologic malignancies were found. Non-Hodgkin’s lymphoma was the cause of fever in eight patients. Three patients diagnosed by lymph node biopsy were strongly suspected of having lymphoma on physical examination. In two patients, the diagnosis was delayed by difficulties in interpreting histology. Two patients were diagnosed by the bone marrow aspiration. In the eighth patient, there was only very small abdominal lymph nodes found by abdominal CT scan and a CT guided FNAC was done .In three patients Hodgkin’s lymphoma was the cause of fever. Acute leukaemia was responsible for PUO in two cases. Periampullary carcinoma, colon cancer, ovarian carcinoma and bronchogenic carcinoma were each responsible for one case of PUO. In this connection it should be mentioned that periampullary carcinoma was initially presented with jaundice and intermittent fever and we did not have the initial suspicion that it would turn out to be periampullary carcinoma. So after relevant investigations when the diagnosis was established it became an enlightening experience for us.
SLE was responsible for five cases of PUO. In one case, fever preceded the more typical manifestations, delaying the diagnosis. Two other patients presented with uncommon manifestations like generalized seizures and pleural effusion. There were two male SLE patients presenting with fever and pericardial effusion respectively. SLE was not suspected in either, delaying the diagnosis. Other collagen vascular disorders in our series were Takayasu’s aortitis, mixed connective tissue disease ankylosing spondylitis and polyarteritis nodosa.
A large number of diseases of unclassified nature are associated with PUO. The miscellaneous group in the present study encompassed sarcoidosis, granulomatous hepatitis, autoimmune hepatitis, atrial myxoma and drug fever found in one patient each.
In 14 patients, no diagnosis could be made. Of these, 11 patients recovered spontaneously whereas in three patients the fever persisted. Of patients with spontaneously resolved undiagnosed PUO, five patients had an illness, which best fits self-limited prolonged viral infection. Six other patients had undiagnosed PUO, which eventually resolved, in three possibly in response to antibiotic therapy and in the other three spontaneously.
Comparison between series of patients with PUO is difficult because of the large number of possible causes and the influence of numerous factors on the relative proportion of the various diagnostic categories. Geographic factors, referral patterns, time of the study and age of the patient have been shown to influence the distribution of the various diagnostic categories.,, Our study represents the disease spectrum in a tertiary referral centre in Eastern India, and we limit the detailed comparison to other studies from India.,,,
The percentage of patients with PUO in whom it was possible to establish a final diagnosis in the present study was 86. As in previous reports,,,, infection such as tuberculosis, bacterial endocarditis and localized abscesses were frequently the cause of PUO in our patients. It is to be noted that visceral leishmaniasis which is endemic in this part of the country was also a frequent cause of PUO in our patients. Tuberculosis still is the commonest cause of PUO in our set up. The diagnosis of tuberculosis was based mainly on repeated chest roentgenograms, culture, histopathological tissue diagnosis and therapeutic trial. Therapeutic trial was used as a diagnostic tool in five patients. The diagnosis of simple disease such as UTI and enteric fever was delayed primarily due to negative cultures on account of empiric antibiotic usage. The diagnosis of malaria was missed in three cases because of repeatedly negative blood smears and non-response to chloroquine.
Neoplasms were the second most frequent cause of PUO. This is contrary to other reports from North India, where collagen vascular diseases were a more important cause of PUO than neoplasms. This may be an early trend towards increasing importance of neoplasms as cause of PUO, but it needs further confirmation. As in other series, lymphomas were the most common neoplasm in our patients also. Acute aleukaemic leukaemia qualified as PUO because initial bone marrow aspiration failed to reveal the diagnosis that was eventually established by bone morrow biopsy.
Our findings on the incidence of collagen vascular diseases in PUO agree with previous studies, where it was found in about 8 to 15 % of the patients. SLE was the major collagen vascular disease in our patients. Despite a low threshold for serologic tests, SLE was missed because of its atypical presentation. Five percent of our patients with PUO had diseases other than infections, neoplasia and collagen vascular diseases, which agrees with previous findings. The incidence of PUO of indeterminate cause ranged from 4.7% to 19% in previous series,, and in the present study it was 14%. The prognosis of our patients with idiopathic fever was relatively good, in agreement with previous finding.
Infection especially tuberculosis remains the most important cause of PUO in India, despite the decreasing importance of infections as cause of PUO in the western literature. Though that study included HIV positive and neutropenic patients and comparison on that ground should not be a fair one. Almost all the infections in the current study were due to common microorganisms. It is important for us to realize that such common diseases and microorganisms can cause PUO. The incidence of neoplasm in the present study was much higher, compared to other studies from northern India., This trend needs further confirmation. The diagnosis was established by non-invasive means in more than two-third of the cases. This is perhaps the first study on this issue from Eastern India.
[Table - 1], [Table - 2]