Etiology of Classic Fever of Unknown Origin (FUO) among Immunocompetent Indian Adults

Abstract

Background: Fever of unknown origin (FUO) has been a vexing problem for physicians for decades. The advent of imaging, functional scans, guided procedures and advanced molecular techniques has made many of the hitherto undiagnosed diseases easily diagnosable. FUO epidemiology can be geographically unique varying from country to region. Studies done in India are scarce, with variable definitions. Methods: This prospective observational cohort study recruited 300 consecutive patients presenting with classic FUO as defined by Durack and Street. Potential diagnostic clues (PDCs) were identified and workup proceeded towards establishing a confirmatory diagnosis. Results: Among the 300 classic FUO in our series, infections, neoplasms and NIIDs contributed to 48%, 21.6% and 20.6% of the cases. Tuberculosis and Melioidosis were the most important infections. Hematological malignancies like Non Hodgkins’ lymphoma, Hodgkins’ lymphoma and Leukemia contributed to 78% of neoplasms causing FUO whereas solid organ malignancies contributed to 18% of the cases. Among the NIIDs, Systemic lupus erythematosus, Granulomatous diseases and Vasculitis contributed to 26%, 18% and 14.5% respectively. Diagnostic tests of utility included image guided biopsies (100%); CT scan of abdomen and or thorax (92.4%) and Lymph node biopsies at 72%. Mortality was 5%. A boot strapping analysis was done on PDCs contributing to each specific diagnostic category and algorithms were developed. Conclusions: This is the largest series of FUO from South India. Systematic sequence of investigations without start of empirical therapy led to a diagnosis in 99.4% which is the highest in described literature. © 2019, Journal of Association of Physicians of India. All rights reserved.

Full text

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019 21
Etiology of Classic Fever of Unknown Origin (FUO) among
Immunocompetent Indian Adults
Priscilla Rupali1*, Divyani Garg2, Surekha Viggweswarupu3, Thambu David Sudarsanam3,
Visalakshi Jeyaseelan4, Ooriapadickal Cherian Abraham3
1Professor of Medicine, 2Post Graduate Student, 3Professor, Department of Geriatrics, 4Associate Professor, Department of
Biostatistics, Christian Medical College and Hospital, Vellore, Tamil Nadu; *Corresponding Author
Received: 11.08.2017; Accepted: 23.08.2018
ORIGINAL ARTICLE
Abstract
Background: Fever of unknown origin (FUO) has been a vexing problem
for physicians for decades. The advent of imaging, functional scans, guided
procedures and advanced molecular techniques has made many of the
hitherto undiagnosed diseases easily diagnosable. FUO epidemiology can be
geographically unique varying from country to region. Studies done in India are
scarce, with variable definitions.
Methods: This prospective observational cohort study recruited 300 consecutive
patients presenting with classic FUO as defined by Durack and Street. Potential
diagnostic clues (PDCs) were identified and workup proceeded towards
establishing a confirmatory diagnosis.
Results: Among the 300 classic FUO in our series, infections, neoplasms and NIIDs
contributed to 48%, 21.6% and 20.6% of the cases. Tuberculosis and Melioidosis
were the most important infections. Hematological malignancies like Non
Hodgkins’ lymphoma, Hodgkins’ lymphoma and Leukemia contributed to 78% of
neoplasms causing FUO whereas solid organ malignancies contributed to 18%
of the cases. Among the NIIDs, Systemic lupus erythematosus, Granulomatous
diseases and Vasculitis contributed to 26%, 18% and 14.5% respectively.
Diagnostic tests of utility included image guided biopsies (100%); CT scan of
abdomen and or thorax (92.4%) and Lymph node biopsies at 72%. Mortality was
5%. A boot strapping analysis was done on PDCs contributing to each specific
diagnostic category and algorithms were developed.
Conclusions: This is the largest series of FUO from South India. Systematic
sequence of investigations without start of empirical therapy led to a diagnosis
in 99.4% which is the highest in described literature.
Introduction
Fever of unknown origin (FUO)
has perplexed physicians for
generations. The causes of FUO are
more than 200 and detailed knowledge
of various medical conditions is
required to reliably make a diagnosis.
Petersdorf and Beeson in 1961 in their
original paper defined FUO as fever
more than 38.3°C (101°F) on several
occasions with a duration of greater
than 3 weeks and uncertain diagnosis
after 1 week of inpatient hospital
investigations.2 With the advent of HIV
infection, organ transplantation and
improvement of intensive care facilities
this was subsequently revised by
Durack and Street et al into 4 categories:
Classic FUO, Neutropenic FUO,
Nosocomial FUO and HIV associated
FUO.3 However despite the advances
in diagnostic techniques and facilities
the proportion of undiagnosed entities
has continued to be substantial in the
case of classic FUO.4 The diagnosis and
spectrum of FUO has been elucidated
from the developed countries, but
data from India and other developing
countries is limited. The main causes
of classic FUO include infections,
neoplasms and Non infectious
Inflammatory Diseases (NIID).1
However in developing countries,
infections are a prominent cause of
FUO unlike in developed countries,
where all three play an important role.5
Therefore diagnostic approach in India
to FUO has to be distinctly different
from that in the developed countries
considering the different spectrum
and costs. FUO series from India have
included data from East (Kolkata), West
(Mumbai), Central (Wardha) and North
India (Delhi) with no data from South
India, hence this study was designed
to prospectively evaluate classic FUO
in a tertiary care hospital with a view
to elucidating various causes of FUO,
identifying potential diagnostic clues
(PDC) and using these to develop an
algorithmic approach applicable in a
resource limited setting.
Material and Methods
This was a prospective observational
cohort study performed over a 20
month period from December 2010
to July 2012, at Christian Medical
College (CMC) Vellore, Tamil Nadu,
India. CMC Vellore is a tertiary care
hospital with patients across the
country accessing care.
We enrolled patients (age > 15 years)
who fulfilled the following criterion for
classical PUO as defined by Durack and
Street et al.3
Temperature of >38.3 degree C
(101 degree F) on several occasions
as documented by a health care
practitioner for > 3 weeks duration
and in whom there was a failure to
establish a diagnosis with appropriate
investigations after 3 outpatient visits
or 3 days as an inpatient.
We excluded patients with HIV
associated FUO, Neutropenic (< 500

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019
22
cells/mm3) FUO and nosocomial FUO
and those on steroids (> 10 mg/day) or
other immunosuppressants for at least
2 weeks or those on chemotherapy for
a malignancy. Moribund patients with
PUO who were unlikely to survive the
duration of diagnostic investigations
were also excluded. Preliminary tests
were done to exclude an acute febrile
illness and then were recruited into
the study.
Based on localizing clinical features,
patients were then subjected to second
rung of tests usually an imaging -
Ultrasound Abdomen, CT Abdomen
and Thorax (contrast-enhanced),
CT or MRI brain/ spine. Analysis,
aspiration, cultures and biopsies of
fluid, collections and tissue were done
if clinically indicated. Endoscopies with
biopsies and cultures and PET scanning
were also considered if required. All
demographic and clinical variables
were collected in a structured data form
by the principal investigator. Repeated
and detailed physical examinations
were done every two days. We provided
assistance to the investigation of
FUO but no rigid protocol/algorithm
was followed. The final diagnosis
established at discharge or during
follow-up comprised the main outcome
of the study. Only diagnoses confirmed
by a diagnostic test or sufficiently
validated by a therapeutic trial with
reasonable certainty were accepted. The
laboratory test or diagnostic method
that diagnosed the cause of fever first
was also recorded. Tests were halted as
soon as the diagnosis was established,
and appropriate treatment initiated.
Patients with empirical therapy were
strictly monitored and followed up to
ensure a sustained clinical response.
If no diagnosis was obtained despite
detailed and invasive evaluation on
first admission and patients were
clinically stable, they were counselled
against empirical therapy and advised
to come back for a re-evaluation after 6
weeks if symptoms persisted with the
intention that the disease would have
progressed enough for us to make a
diagnosis.
Denition for disease states
Tuberculosis (TB) was diagnosed
when M. tuberculosis grew in culture
on tissue/tissue fluid or granulomas
were seen on histopathology with
a compatible clinical picture and a
clinical response to antituberculosis
therapy (ATT).
Occult tuberculosis was diagnosed
when we were unable to obtain a
tissue or culture diagnosis during that
admission but was proven later by
positive cultures for M. tuberculosis
or a dramatic response to empirical
ATT with clinical and radiological
resolution.
Infective endocarditis was diagnosed
when Modified Duke’s Criteria were
fulfilled.
Melioidosis was diagnosed based on
growth of Burkholderia psuedomallei
on culture from appropriate sample
(pus, blood, urine).
Enteric fever was diagnosed
when blood/bone marrow culture
grew Salmonella typhi/paratyphi or
WIDAL positivity with rising titres
in a compatible clinical setting with
response to treatment for enteric fever.
Rheumatoid arthritis diagnosis was
made based on the 2010 American
College of Rheumatology (ACR) and the
European League Against Rheumatism
(EULAR) revision of the 1987 ACR
classification criteria for RA.
Various rheumatological diseases
were diagnosed based on their
established criteria.
Lymphomas or haematological
malignancies were diagnosed
based on histopathology and
Immunophenotyping from appropriate
tissue.
Diseases as defined above with
relevant clinical and laboratory features
were noted and grouped into the three
major etiological groups i.e., infections,
neoplasms and non-infectious
inflammatory diseases (NIIDs).
Sample size was calculated to be 96
based on the prevalence of tuberculosis
of 40% (25-50%) among FUO studies
in India
Sample size= (1.962 x P x [1-P])/D2
P=prevalence of the disease (%),
D= confidence interval (taken as
10%). Summary statistics and tests
of significance (Chi square test for
categorical variables and student t-test
for continuous variables) were done
using statistical software package SPSS
version 16.
Institutional review board approval
and funding
IRB approval was obtained prior
to starting the study and funding was
provided through an internal fluid
research grant, to a post graduate
student in Department of General
Medicine and was subsequently
submitted to the Tamilnadu Dr MGR
Medical University as a PG dissertation.
Results
A total of 300 consecutive patients
were recruited into the study. Almost
65% were male with an average age of
40 years (median 40; range 15-76 years).
Two-thirds of our patients were from
Tamil Nadu (31%) and West Bengal
(32%) with the rest from South and
North Indian states. The mean duration
of fever prior to presentation was 148
days (median: 90; range- 21 to 1460).
Overall the striking clinical features
included anorexia (58%) and weight loss
(61%) in two-thirds, followed closely by
pallor (41%) hepatosplenomegaly (33%)
and lymphadenopathy (30%). A third
(32%) of them had symptoms localizing
to the respiratory tract i.e., cough and
dyspnea and 12% localizing to the CNS
- headache, seizures or focal deficits
with the rest to the musculoskeletal
system. Important localizing lab tests
included elevated alkaline phosphatase
(56.5%), and an abnormal chest X-ray
(25%). Median Hb was 12.45. (Range:
3.6 -15.3), median ESR and CRP were
46 (range: 7-140) and 7 (range: 2-218),
respectively.
In our cohort, infection was the
most common cause of classic FUO
accounting for 48% of all cases, followed
by neoplasms in 21.6% cases and non-
infectious inflammatory diseases in
20.6% of the cases. Miscellaneous and
undiagnosed causes accounted for the
remaining 8.6% and 1.6% of the cases.
Among the infections, tuberculosis
in varied forms was the leading cause
of infectious FUO accounting for 61%
of the cases, followed by Melioidosis
(10%), subacute bacterial endocarditis
(4%) and visceral abscesses (4.8%),
invasive fungal infections (3.4%),
disseminated histoplasmosis (2.7%)
and fungal endocarditis in 0.7%. Other
causes accounted for 10% and included
varied causes like delayed diagnosis of
Enteric fever, Visceral leishmaniasis,
Type II Lepra reaction, Brucellosis etc.
Among the tuberculosis patients,
disseminated (45%), extra-pulmonary
(33%) and pulmonary in 14% were
the commonest clinical presentations
with occult tuberculosis in 8% of the
cases. Among the extra-pulmonary

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019 23
Table 1: Distribution of specic categories of FUO across the age
groups
Age group Infection Neoplasm NIID Miscellaneous Total
<20 years
% within category
% within age group
9
6.3
39.1
3
4.6
13
7
11.3
30.5
4
13.8
17.4
23
100
20-29 years
% within category
% within age group
33
22.9
47.1
8
12.3
11.4
23
37.1
32.9
6
20
8.6
70
100
30-39 years
% within category
% within age group
24
16.7
48
9
14
18
10
16.1
20
7
23
14
50
100
40-49 years
% within category
% within age group
38
26.3
51.4
20
30.8
27
13
21
17.6
3
10
4
74
100
50-59 years
% within category
% within age group
20
13.9
44.4
15
23
33.3
3
4.9
6.7
7
23
15.6
45
100
Above 60 years
% within category
% within age group
20
13.9
55
9
13.8
23
6
9.7
15
3
10
7
38
100
Total 144 64 62 30 300
Table 4: Potential diagnostic clues in each category
Infection Neoplasm NIID P Value
Symptoms
Cough 64.5% 24.8% 9.7% 0.019
Diarrhea 50% 0 50% 0.028
Headache 84.6% 0 15.4% 0.050
Arthritis 22.2% 2.8% 75% 0.000
Rash 9.5% 23.8% 66.7% 0.000
Signs
Pallor 42.5% 38.1% 19.5% 0.000
Lymphadenopathy 38.4% 36% 25.6% 0.001
Hepatosplenomegaly 43.4% 45.3% 11.3% 0.000
Arthritis 22.2% 2.8% 75% 0.000
Basic investigations
Anemia (Hb <8 g%) 58.1% 18.6% 23.3% 0.000
Thrombocytopenia 39.2% 45.1% 15.7% 0.000
Leukopenia (WBC count <4000 cells/
mm.cu)
43.9% 39.0% 17.1% 0.044
Diagnostic tests
Blood culture 100% 0 0 0.001
Bone marrow studies 34.9% 62.8% 2.3% 0.000
CT thorax 58.6% 13.8% 27.6% 0.019
CT thorax with abdomen 56.8% 32.4% 10.8% 0.006
Lymph node biopsy 50.0% 41.3% 8.7% 0.000
Table 2: Dierential diagnoses of FUO in India Table 3: Diagnostic yield of various tests in
the diagnosis of FUO
Diagnostic test Diagnostic yield (%)
Chest X-ray 25.3
Ultrasound abdomen 66.8
CT thorax and abdomen 83.2
Bone marrow aspiration,
biopsy, cultures
18.5
Lymph node biopsies 69.6
PET scan 66.7
Liver biopsy 50
Diagnostic splenectomy 100
Diagnosis No.
of pts.
(%)
% of
specic
diagnostic
category
Infections 144 (48)
Tuberculosis 88 (29) 61
Melioidosis 16 (5) 10
Infective endocarditis 6 (2) 4
Visceral abscesses 7 (2) 4.8
Disseminated histoplasmosis
Fungal endocarditis
4 (1)
1 (0.3)
2.7
0.7
Enteric fever
Visceral leishmaniasis
Lepra reaction
Brucellosis
Others (1 case each)*
6 (2)
3 (1)
3 (1)
2
8
4
2
Neoplasms 64 (22)
Non hodgkin’s lymphoma
Hodgkin’s lymphoma
Lymphoma unclassied
Leukemia
20 (7)
17 (6)
2
6 (2)
31
26
3
9
Multiple Myeloma
Solitary plasmacytoma
Myelodysplastic syndrome
6 (2)
1
1
9
Solid organ tumours 12 (4) 18
Diagnosis No.
of pts.
(%)
% of
specic
diagnostic
category
Non infectious inammatory
diseases
61 (21)
SLE 16 (5) 26
Vasculitis 9 (3) 14.5
Inammatory bowel disease
Sarcoidosis
Kikuchi’s disease
4
2
5
Mixed connective tissue
disorder
5
Still’s disease 5
Rheumatoid Arthritis
Seronegative
spondyloarthropathy
4
2
Others** 9
Miscellaneous 26 (8.6)
Self limited 17 (5)
Hyperthyroidism
Others***
6 (2)
3 (1)
Undiagnosed 5 (2)
tuberculosis, meningeal and lymph
nodes were the most common sites,
accounting for 70% of the cases.
Melioidosis caused by B. pseudomallei
emerged as the second most important
cause of infection contributing to
a PUO accounting for 10% of all
infections and 5% of overall cases of
FUO. The predominant risk factor for
acquisition of Melioidosis was diabetes.
The diagnosis was made from blood
culture (septicemic melioidosis) in
37.5% and aspiration from visceral
abscesses, bone or other sites (Chronic
Granulomatous Melioidosis) in
62.5%. Infective Endocarditis (IE) and
pyogenic abscesses (liver, spleen and
renal) accounted for 2.3% of all cases.
Among invasive fungal infections,
disseminated histoplasmosis and fungal
(Candida) endocarditis contributed to
1.6% of the causes of PUO.
Neoplasms caused FUO in 22% (n=64)
of cases. Haematological malignancies
were the commonest neoplasms
accounting for 78% with lymphoma
accounting for 60%, leukemias 9% and
multiple myeloma 9%. Solid organ
malignancies like colon, lung, prostate
accounted for 18% (n=12) of FUO. Less
common causes were hepatocellular
and poorly differentiated carcinoma.
Non-infectious inflammatory
diseases (NIID) caused classic FUO
in 21% of the cases and among these,
systemic lupus erythematosus (SLE)
was the most common ( 26%) followed
by vasculitis (14.5%), Mixed Connective
Tissue Disease and Adult onset Still’s
disease at (8%). Other infrequent
causes included Rheumatoid Arthritis,
Kikuchi-Fujimoto disease, Seronegative
Spondyloarthropathy, Inflammatory
bowel disease and, Sarcoidosis.
Rarer causes of PUO, like
hyperthyroidism, accounted for 2% of
all cases. Among those fevers in which
no cause could be found, 8.6% were
self-limited and truly undiagnosed
cases were seen in only 1.6% of the

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019
24
from developing countries show that
infections are the most common cause
of FUO,6 and they accounted for 48% of
cases in our series. Classic FUO in India
is often also due to delayed diagnosis
of acute febrile illnesses with lack of
confirmation either through culture
or molecular techniques and further
compounded by lack of specificity of
serology in an endemic setting. The
previous studies elucidating the causes
of classic FUO have used inconsistent
and different definitions of disease
processes thus questioning their
validity and reliability.7 In addition
there can be varying causes of FUO
depending on the geographical location
and prevalence of local diseases and
hence causes of fever are often different
in East, West, North and South India.7-
10 We used very strict definitions of
disease in our study validated in a
previous published FUO study.11
Infections as the major category
Fig. 1: Algorithm for diagnosis of classic FUO in India
Classic Pyrexia of Unknown Origin
Document Temperature, exclude HIV, Steroid use
Carefully look for localising symptoms and examine for signs
Basic investigations: CBC, Malarial Parasite x 2, Blood culture x3,
Urine Microscopy, ESR, CRP
Suspect
Infection
Neoplasm
Non-infectious inflammatory
diseases (NID)
Symptoms: Oral ulcers, Rash,
Arthritis
Signs: Joint swelling
Likely NID
Proceed to collagen vascular
diseases workup
Classify into SLE / Vasculitis /
Sarcoidosis
If Unclassified
Label as NID and initiate
and treat with
antiinflammatory drugs
Wait for
patient to
evolve if stable
CT Abdomen / Thorax
(To look for nodes)
Bone marrow biopsy
Lymph node biopsy
Likely Neoplasm
Symptoms: Anorexia, weight loss
Signs: Anemia, Lymphadenopathy,
Hepatosplenomegaly
Symptoms: Anorexia, Weight Loss / Signs: Anemia, Nodes
+ Cough Likely
Infection
Sputum
AFB x 3
Negative
CT
Radiology guided
procedures with
biopsy and
cultures
CSF studies
including
cultures
CT / MRI
Likely
infection
+ Headache
Diarrhea
Stool occult
blood, stool
parasites
CT
Abdomen
Establish
diagnosis
Other
symptoms or
significant
Syndrome
directed
evaluation
Abbreviation: HIV: human immunodeficiency virus; CBC: complete blood count; ESR: erythrocytes sedimentation rate; CRP: C-reactive protein;
CT: computerised tomography; MRI: magnetic resonance imaging; CSF: cerebro spinal fluid; SLE: systemic lups erythematous; AFB: acid fast bacilli
entire cohort.
Infections were the most common
cause of FUO across all age groups.
In patients < 50 years, NIIDs were
second most common (occurring in
the third decade, between 20 to 29
years of age) but in patients > 50 years
neoplasms were second (occurring in
the 5th decade, between 40 -60 years).
Infections were uncommon in the very
young (< 20 years) (Table 1).
Invasive procedures were needed
to make a diagnosis in 69% of the
cases and image guided biopsies had
the highest diagnostic yield 100%
followed by lymph node and bone
marrow biopsies with cultures at 63%
and 19%. Non-invasive tests including
CT scans, revealed a diagnosis in 72.4%.
Diagnosis was made based on clinical
picture alone in 3% of the cases.
We noted every symptom, sign
or abnormal laboratory test that
contributed to a specific diagnostic
category and the final diagnosis. The
variables which were seen in maximum
proportion in a particular diagnostic
category i.e., infections, neoplasms and
NIIDs were then subjected to univariate
analysis and the significant variables
were noted. Potential diagnostic clues
(PDCs) both clinical and laboratory in
each specific diagnostic category (Table
4) were used to construct an algorithm
for evaluation of classic FUO (Figure 1).
The mortality in this FUO series
was 5% (n=15) and the commonest
diagnosis among patients who expired
was lymphoma or disseminated
tuberculosis. Two patients died without
a diagnosis ever being made.
Discussion
This is the largest prospective
observational cohort study of classic
FUO from South India. Most studies

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019 25
caused FUO. This is unlike what has
been described from other case series.
In fact, in the series by Bandyopadhyay
et al,10 solid organ cancers did not
present as PUO.
Infections were the commonest
cause across all age groups. There
seemed to be a clear distinction between
<50 years and >50 years for the second
most common cause. NIID were
common in the <50 years age group
vs. neoplasms in >50 years age group.
This seems to be different compared
to what is seen in Western Literature
where NIIDs were the commonest
cause of FUO as compared to infections
and neoplasms.14 The number of
undiagnosed cases in our series in one
of the lowest in published literature
and we attribute this to the fact that
we are a tertiary care centre with all the
facilities available on site and inclusion
of only established and health care
documented cases strictly documenting
temperatures rather than just based
on patient history, thus fulfilling
the criteria of classic FUO. This low
percentage of undiagnosed cases (1.6%)
is in sharp contrast to all other case
series so far, which have described
undiagnosed cases between 7 to 51%.5,15
Every attempt was made to obtain
unequivocal confirmation of diagnosis
by invasive or non-invasive means for
e.g., diagnosis of an infection was based
only on cultures in the background
of a compatible clinical picture;
diagnosis of a neoplasm was based on
confirmatory histopathological and/
or immuno-histochemical evidence. A
diagnosis based on clinical judgement
without conclusive microbiological or
histopathological evidence was made
in only 3.7 % of the cases and empirical
therapy was instituted in these patients
with close follow up to ensure that
the initial diagnosis was correct. Only
patients who had complete clinical and
radiological resolution to empirical
therapy consistent with the original
clinical diagnosis were deemed to
have that disease process. This was
mostly limited to occult tuberculosis
and undifferentiated collagen vascular
disease.
On evaluation of various diagnostic
tests, the following were found to have
a diagnostic yield of >50% - diagnostic
splenectomy (100%); CT thorax and
Abdomen (83%), lymph node biopsy
(70%), Ultrasound abdomen and PET
scan (67% each) (Table 3). We found
of FUO has predominated over
the decades and this has remained
consistent9,10 in India. In our study,
Tuberculosis comprised 1/3 (29%) of
all cases and 2/3 (61%) of the infections
similar to previous Indian studies. In
the latest case series described from
Kolkata,10 28 % of the patients had
tuberculosis with 72% having extra-
pulmonary tuberculosis. In the series
by Kejariwal et al,9 tuberculosis was
again the commonest diagnosis but
pulmonary tuberculosis was not seen
presenting as PUO unlike our series,
where pulmonary tuberculosis was
seen in 14% of the cases. In our studies,
diagnosis was most often established
from specimens other than sputum AFB
smear or culture e.g., bronchoalveolar
lavage, pleural fluid or molecular
techniques.
Melioidosis was the second most
common infection contributing to
a classic FUO and this has not been
described before as an important
cause in previous studies. We feel
that this is because though abscesses
have been described in previous
studies9 as a common cause of FUO
in previous studies, the etiological
agent of Melioidosis i.e., Burkholderia
Pseudomallei was probably not
identified in these patients. Limited
experience, lack of validated diagnostic
strategies and dependence on
automated blood systems often leads
to misdiagnosis of this organism.10 The
Microbiology laboratory in our hospital
has a standard protocol for identification
based on typical morphology (closed
safety pin appearance), appearance of
culture plate (metallic sheen), oxidase
negativity, testing with polyclonal
antiserum (in house preparation of
antiserum in rabbits) and resistance
to Gentamicin and Polymyxin B on
antimicrobial susceptibility. This
organism is often dismissed as a
contaminant as it is a non fermenting
gram negative bacillus and on isolation
from specimens from non sterile sites
it may be overgrown by commensal
organisms.12,13 There has been a decrease
in prevalence of endocarditis as a cause
of FUO, probable to due to earlier
recognition of the same due to better
culture techniques and availability of
Transoesophageal Echocardiography
(TOE).
Among the neoplastic causes of FUO
apart from haematological causes of
malignancies, solid organ cancers also
that imaging with guided biopsies were
the mainstay of diagnosis. Our study
suggested lymph node biopsies should
be done early if lymphadenopathy is
detected, with repeated screening for
the same as these have a high diagnostic
yield. Bone marrow biopsies though
often done were found to have a fairly
low diagnostic yield of 19%, similar
to other series of FUO up to 25%16,17
suggesting that they should only be a
third rung of investigations. We found
rare causes of PUO in our case series,
likely due to the large sample size and
diagnostic abilities.
Limitations include a possible
referral bias; patients referred to us
were usually evaluated elsewhere
and referred after non-response to a
therapeutic trial, leading us to suspect
an alternate diagnosis. We were unable
to conclusively establish a causal
spectrum of FUO from South India
alone, as our centre sees a large number
of cases from Eastern India.
In conclusion, infections are most
important cause of classical FUO with
extrapulmonary tuberculosis being the
most frequent, in India. Melioidosis is
an emergent cause of FUO seen often in
diabetics. Lymphoma is the commonest
neoplasm FUO and SLE was the most
common non-infectious inflammatory
disease causing FUO. Invasive tests
especially lymph node biopsy have
a high diagnostic yield in FUO and
hence patients should be referred to
centres where these can be done in case
of diagnostic dilemmas. The number
of cases in our series who remained
undiagnosed was extremely small as
compared to previous studies, probably
due to a strict documentation of fever
in hospital, an aggressive diagnostic
approach to FUOs and avoidance of
empirical therapy as much as possible.
Acknowledgements
We gratefully acknowledge, the
Departments of Medicine, Infectious
Diseases, Biostatistics and our patients
who have helped us with this study.
References
1. Arnow PM, Flaherty JP. Fever of unknown origin. Lancet
1997; 350: 575-580.
2. Petersdorf RT, Beeson PB. Fever of unexplained origin:
Report on 100 cases. Medicine 1961; 40: 1-30.
3. Durack DT, Street AC et al. Fever of unknown origin: re-
examined and re-dened. Curr Clin Topics Infect Dis 1991;
11: 35-51.
4. Horowitz HW. Fever of unknown origin or too many origins.
N Engl J Med 368; 3:197-9.

Journal of The Association of Physicians of India ■ Vol. 67 ■ January 2019
26
5. Bleeker-Rovers CP, Vos FJ, de Kleijn EHMA, et al. A prospective
multicenter study on fever of unknown origin: the yield of
a structured diagnostic protocol. Medicine 2007; 86:26-38.
6. Newsholme W, Brown M. Pyrexia of unknown origin: a review
of studies from the developing world. Tropical Doctor 2005;
35:68-71.
7. Jung A, Singh MM, Jajoo U. Unexplained fever-analysis of
233 cases in a referral hospital. Indian J Med Sci 1999; 53:535-
44.
8. Mir T, Dhobi GN, Koul AN, Saleh T. Clinical prole of classical
FUO. Caspian J Intern Med 2014; 5:35-39.
9. Kejariwal D, Sarkar N, Chakraborti SK, Agarwal V, Roy S.
Pyrexia of unknown origin: A prospective study of 100 cases.
Journal of Postgraduate Medicine 2001; 47:104-107.
10. Bandyopadhyay D, Bandyopadhyay R, Paul R, Roy D.
Etiological study of Fever of unknown origin in patients
admitted to medicine ward of a teaching hospital of eastern
India. J Glob Infect Dis 2011; 3:329–33.
11. Rupali P, Abraham OC, Zachariah A et al. Aetiology
of prolonged fever in antiretroviral naive human
immunodeciency virus infected adults. Natl Med J India
2003; 16:189-195.
12. Inglis TJ, Merritt A, Chidlow G, Aravena-Roman M, Harnett
G. Comparison of diagnostic laboratory methods for
identication of Burkholderia pseudomallei. J Clin Microbiol
2005; 43:2201.
13. Lowe P, Engler C, Nor ton R. Comparison of automated and
nonautomated systems for identication of Burkholderia
pseudomallei. J Clin Microbiol 2002; 40:4625.
14. Knockaert DC, Vanneste LJ, Bobbaers HJ. Fever of unknown
origin in elderly patients. J Am Geriatr Soc 1993; 41:1187.
15. Naito T, Mizooka M, Mitsumoto F, et al. Diagnostic workup
for fever of unknown origin: a multicenter collaborative
retrospective study. BMJ Open 2013; 3:e003971. doi:10.1136/
bmjopen-2013-003971.
16. Hot A, Jaisson I, Girard C, French M, Durand DV, Rousset H,
et al. Yield of bone marrow examination in diagnosing the
source of fever of unknown origin. Arch Intern Med 2009;
169 :2018–23.
17. David Rosario. The histo-pathological and microbiological
examination of bone marrow in the diagnostic evaluation
of Fever of Unknown Origin. PG Dissertation submitted to
the T.N Dr MGR Medical University in the year 2000.