Intensive Care Med

DOI 10.1007/s00134-014-3603-2 ORIGINAL

Arunaloke Chakrabarti
Prashant Sood
Incidence, characteristics and outcome
Shivaprakash M. Rudramurthy of ICU-acquired candidemia in India
Sharon Chen
Harsimran Kaur
Malini Capoor
Deepinder Chhina
Ratna Rao
Vandana Kalwaje Eshwara
Immaculata Xess
Anupama J. Kindo
P. Umabala
Jayanthi Savio
Atul Patel
Ujjwayini Ray
Sangeetha Mohan
Ranganathan Iyer
Jagdish Chander
Anita Arora
Raman Sardana
Indranil Roy
B. Appalaraju
Ajanta Sharma
Anjali Shetty
Neelam Khanna
Rungmei Marak
Sanjay Biswas
Shukla Das
B. N. Harish
Sangeeta Joshi
Deepak Mendiratta

M. Capoor
Received: 23 August 2014 A. Chakrabarti ())
P. Sood

VMMC and Safdarjang Hospital,
Accepted: 4 December 2014 S. M. Rudramurthy
H. Kaur
New Delhi, India
Department of Medical Microbiology,
Ó Springer-Verlag Berlin Heidelberg and Postgraduate Institute of Medical Education
ESICM 2014 D. Chhina
Research, Chandigarh 160012, India
Dayanand Medical College and Hospital,
e-mail: arunaloke@hotmail.com
For the SIHAM Candidemia Network. Ludhiana, India
Tel.: ?91-172-2755173
Take home message: This multicentric R. Rao
study from India on ICU-acquired A. Chakrabarti
Apollo Hospital, Hyderabad, India
candidemia highlights the unique Center of Advance Research in Medical
epidemiology of this country with its vast Mycology, WHO Collaborating Center for
V. K. Eshwara
spectrum of Candida species and high rate Reference and Research on Fungi of
of C. tropicalis isolation. The disease Kasturba Medical College,
Medical Importance, National Culture
occurred comparatively early after ICU Manipal, India
Collection of Pathogenic Fungi, Chandigarh
admission, even in patients with less severe 160012, India
physiology scores. I. Xess
All India Institute of Medical Sciences,
Electronic supplementary material S. Chen
New Delhi, India
The online version of this article Centre for Infectious Diseases and
(doi:10.1007/s00134-014-3603-2) contains Microbiology Laboratory Services,
A. J. Kindo
supplementary material, which is ICPMR-Pathology West, Westmead,
Sri Ramachandra Medical College,
available to authorized users. New South Wales, Australia
Chennai, India
P. Umabala R. Marak 4–15 days), even infecting patients
Nizam’s Institute of Medical Sciences, Sanjay Gandhi Postgraduate Institute of with lower APACHE II score at
Hyderabad, India Medical Sciences, Lucknow, India admission (median 17.0; mean ± SD
J. Savio S. Biswas 17.2 ± 5.9; interquartile range
St. John’s Medical College, Bangalore, Tata Memorial Hospital, Mumbai, India 14–20). The important finding of the
India study was the vast spectrum of agents
S. Das (31 Candida species) causing candi-
A. Patel University College of Medical Sciences, demia and a high rate of isolation of
Sterling Hospital, Ahmedabad, India Delhi, India Candida tropicalis (41.6 %). Azole
and multidrug resistance were seen in
U. Ray B. N. Harish
Apollo Gleneagles Hospital, Kolkata, India Jawaharlal Institute of Postgraduate 11.8 and 1.9 % of isolates. Public
Medical Education and Research, sector hospitals reported a signifi-
S. Mohan Pondicherry, India cantly higher presence of the
Christian Medical College and Hospital, relatively resistant C. auris (8.2 vs.
Ludhiana, India S. Joshi 3.9 %; p = 0.008) and C. rugosa (5.6
Manipal Hospital, Bangalore, India vs. 1.5 %; p = 0.001). The 30-day
R. Iyer crude and attributable mortality rates
Global Hospitals, Hyderabad, India D. Mendiratta
Mahatma Gandhi Institute of Medical of candidemia patients were 44.7 and
J. Chander Sciences, Wardha, India 19.6 %, respectively. Logistic
Government Medical College, Chandigarh, regression analysis revealed signifi-
India cant independent predictors of
Abstract Purpose: A systematic mortality including admission to
A. Arora epidemiological study on intensive public sector hospital, APACHE II
Fortis Escorts Heart Institute, New Delhi, care unit (ICU)-acquired candidemia score at admission, underlying renal
India across India. Method: A prospec- failure, central venous catheterization
R. Sardana
tive, nationwide, multicentric, and steroid therapy. Conclu-
Indraprasth Apollo Hospital, New Delhi, observational study was conducted at sion: The study highlighted a high
India 27 Indian ICUs. Consecutive patients burden of candidemia in Indian ICUs,
who acquired candidemia after ICU early onset after ICU admission,
I. Roy admission were enrolled during April higher risk despite less severe physi-
Calcutta Medical Research Institute, 2011 through September 2012. Clin- ology score at admission and a vast
Kolkata, India ical and laboratory variables of these spectrum of agents causing the dis-
B. Appalaraju
patients were recorded. The present ease with predominance of C.
PSG Institute of Medical Sciences and study is an analysis of data specific tropicalis.
Research, Coimbatore, India for adult patients. Results: Among
1,400 ICU-acquired candidemia cases Keywords Candidemia

A. Sharma (overall incidence of 6.51 cases/1,000 Intensive care unit

Gauhati Medical College, Guwahati, India ICU admission), 65.2 % were adult. Candida tropicalis
Risk factor

Though the study confirmed the Mortality
A. Shetty
PD Hinduja Hospital, Mumbai, India
already known risk factors for candi-
demia, the acquisition occurred early
N. Khanna after admission to ICU (median
Batra Hospital, New Delhi, India 8 days; interquartile range

Introduction Furthermore many regions of the world are witnessing a

surge in non-albicans Candida species [6, 7], which have
Candida bloodstream infection (candidemia) is a life- diverse virulence and susceptibility profiles [8]. Geo-
threatening affliction in intensive care unit (ICU) patients. graphic variation in causative species, ethnically diverse
Its incidence varies from 0.24 to 34.3 patients/1,000 ICU at-risk population and varying diagnostic and manage-
admissions [1–4]. With a high mortality rate of 35–75 % ment capabilities of ICUs further add to the complexity.
[2], early antifungal treatment is essential for survival. As there are very few multicentric studies on candidemia
ICU-acquired candidemia patients characteristically have from Asian countries, we undertook this study to com-
several underlying medical and surgical risk factors, and prehensively elucidate the disease burden, epidemiology,
are frequently exposed to high-risk medications [5]. microbial circulation, resistance pattern and management
challenges of adult ICU-acquired candidemia in India. proforma. These included demographic details; clinical
We aim to utilize this information for better management presentations at admission; severity scores including
and survival of these patients. Acute Physiology and Chronic Health Evaluation II
(APACHE II) and Glasgow Coma Scale (GCS); comorbid
medical conditions; surgical and invasive procedures;
exposure to antibiotics, corticosteroids, anticancer che-
Methods motherapeutics, prior antifungals; secondary bacterial
infections; candidemia-associated complications; anti-
Study design fungal treatment instituted in the ICU; treatment adverse
effects; Candida species isolated; antifungal susceptibility
This was a prospective, multicentric, observational study and final outcome. All data from participating centres
undertaken from April 2011 to September 2012 at 27 were collected and analysed at the coordinating centre.
medical and surgical ICUs across India. The Postgraduate
Institute of Medical Education and Research, Chandigarh,
India was the coordinating centre. Ethics approval was Microbiological methods
obtained from each institution’s review board. Being an
observational study, it entailed no additional risk to the Candida isolates were identified and tested for antifungal
patients. Prior to commencing work, the study was reg- susceptibility at the coordinating centre. Isolates were
istered as ‘‘SIHAM (Society for Indian Human and identified by a Vitek YBC system, and ITS2 and D1/D2
Animal Mycologists) Candidemia Network: an observa- region sequencing was performed when concordance was
tional study’’ (identifier NCT01281345) with less than 98 %. Antifungal susceptibility was performed
ClincalTrials.gov. using the Clinical and Laboratory Standards Institute’s
(CLSI) broth microdilution M27-A3 method [9]. MIC
breakpoints were interpreted from the CLSI M27-S4
Definitions supplement for yeasts [10]. Isolates with a MIC greater
than 1 lg/ml for amphotericin B were considered resis-
A patient older than 18 years of age was considered an tant. Multidrug resistance (MDR) was defined as
adult. Candidemia was defined as the isolation of Candida resistance to two or more classes of antifungals.
species from blood cultures. A case was defined as ICU-
acquired candidemia if it occurred more than 48 h after
ICU admission or less than 48 h after discharge from Statistical analysis
ICU. Patients already diagnosed with candidemia before
ICU admission or those transferred from another ICU Descriptive analysis is presented as frequencies and
with a positive culture for any yeast infection were confidence intervals (CI); medians and range; or means
excluded from the study. A patient was defined to have and standard deviation (SD), as appropriate. Pearson’s
contracted more than one episode of Candida blood- Chi square test and Fisher’s exact test were used to
stream infection if the subsequent episode occurred more compare categorical variables, while the Student t test,
than 30 days after the previous one. Date of onset of ANOVA, Mann–Whitney or Kruskal–Wallis tests were
candidemia was taken as the date when the first positive employed for continuous data. A predictive model for
blood culture was drawn from the patient. A surgery or 30-day mortality was developed using logistic regression.
invasive procedure undertaken within 30 days of con- Its predictive power was evaluated by the Hosmer–
tracting candidemia was considered a potential risk factor. Lemeshow goodness-of-fit test and area under the recei-
Outcome was determined at the end of 30 days after onset ver operating characteristic (ROC) curve. Kaplan–Meier
of candidemia and cure was considered complete micro- statistics was used for survival analysis. Two-tailed
biological resolution. A death attributed to candidemia p values \0.05 were taken as significant.
was ascertained by the treating physician independently
on the basis of clinical judgement and microbiology
reports.
Results
Data collection Demography and burden

All consecutive patients who contracted ICU-acquired A total of 215,112 patients were admitted to 27 ICUs
candidemia during April 2011 through September 2012 from April 2011 to September 2012. Of these 1,400
were enrolled after informed consent. A total of 398 patients contracted ICU-acquired candidemia, amounting
variables were recorded for every patient in standardized to a burden of 6.51 cases/1,000 ICU admissions (95 % CI
6.18–6.86). Eleven ICUs were from north, three each Circulating Candida species
from east and west, four from central and six from south
India (Fig. 1). Eleven were public sector institutions, The majority of patients experienced only a single epi-
while 16 were private/corporate hospitals. The incidence sode (n = 855; 93.6 %) of candidemia, though 58
varied significantly across regions (p \ 0.001), with the (6.4 %) suffered from more than one episode and five
highest burden from the north (8.95/1,000 ICU admis- (0.5 %) of these were infected with two different Candida
sions; 95 % CI 8.29–9.66) and lowest from west (3.61/ species during different candidemia episodes. Of the 918
1,000 admissions; 95 % CI 2.82–4.63). Of 1,400 patients, Candida strains thus isolated, Candida tropicalis
913 (65.2 %) were adults with a median age of 50.0 years (n = 382; 41.6 %) was the most prevalent followed by
(interquartile range 34–63; mean ± SD 49.7 ± 17.7). Candida albicans (n = 192; 20.9 %) and Candida par-
Further analysis pertains only to adult patients. apsilosis (n = 100; 10.9 %) (Electronic supplementary
material Fig. 1). Eight species caused 92.8 % (852) of the
infections while the remaining 7.2 % were inflicted by 22
ICU presentation less common species. Non-albicans Candida infections
affected 79.1 % (726) of cases. While C. tropicalis was
The majority of our adult patients were non-neutropenic significantly higher (44.8 % vs. 37.7 %; p = 0.035) in
(98.7 %) and were admitted to medical ICUs (n = 464; private/corporate ICUs, C. auris (8.2 % vs. 3.9 %;
50.8 %). The median duration of onset of candidemia in p = 0.008) and C. rugosa (5.6 % vs. 1.5 %; p = 0.001)
ICU was 8.0 (interquartile range 4–15) days. Adult can- were significantly more common in public sector ICUs.
didemia patients had a median APACHE II score of 17.0 There was no significant difference between times of
(interquartile range 14–20; mean ± SD 17.2 ± 5.9) at onset of candidemia when all species were compared. The
admission. GCS was measured in 34 patients with a emerging multidrug-resistant C. auris was isolated from
median score of 15.0 (interquartile range 7–15). 19 of 27 ICUs and comprised 5.2 % of all Candida

Fig. 1 Map of India depicting

the 27 intensive care units
which participated in this study.
The red dot represents the
coordinating centre for this
study
isolates. The species was also significantly more common Antifungal exposure before onset of candidemia was
in patients with central venous catheterization (6.8 % vs. recorded in 143 (15.7 %) cases for a median duration of
2.9 %; p = 0.035). 8.0 (interquartile range 5–13) days (Electronic supple-
mentary material Table 1). Prior antifungal exposure
significantly increased the propensity for subsequent non-
Antifungal resistance albicans Candida infection (85.5 vs. 77.9 %; p = 0.045)
and azole-resistant Candida infection (7.1 vs. 3.3 %;
The antifungal susceptibility profiles of the commonest p = 0.045).
species are summarised in Table 1. Overall 46.6 % of
isolates were sensitive to all antifungals. Resistance to
amphotericin B was noted in 19 (2.1 %), azoles 108 Treatment
(11.8 %), and echinocandin 63 (6.9 %) isolates, respec-
tively. MDR was noted in 17 (1.9 %) isolates, with three Antifungal therapy after definitive diagnosis of ICU-
(0.3 %) pan-resistant to all antifungal classes. Common acquired candidemia could be instituted in 59.9 % (547)
MDR isolates were C. tropicalis (n = 4; 23.5 %), cases with a median duration of 9.0 (interquartile range
C. auris (n = 4; 23.5 %) and C. krusei (n = 3; 17.6 %). 5–15) days (Electronic supplementary material Table 2).
Azoles (n = 394; 72.0 %) were preferred, followed by
echinocandins (n = 100; 18.3 %) and amphotericin B
Underlying diseases and risk factors (including dexoycholate and lipid preparations) (n = 79;
14.4 %). While amphotericin B deoxycholate was used
Underlying respiratory illness was noted in 228 (25.0 %) more commonly in public sector ICUs (11.5 vs. 5.3 %;
patients. Pneumonia (n = 75; 32.9 %), acute respiratory p \ 0.001), azole (47.1 vs. 37.7 %; p = 0.006) and ech-
distress syndrome (n = 41; 17.9 %) and chronic inocandin (13.4 vs. 7.4 %; p = 0.005) use was more
obstructive pulmonary disease (COPD) (n = 35; 15.4 %) common in private/corporate hospitals. Antifungal regi-
were the major respiratory afflictions. Underlying renal men was altered during the course of therapy in 81
disease was equally prevalent (n = 209; 22.9 %), with (14.8 %) patients, regimen escalation and antifungal
the majority in acute (n = 128; 61.2 %) or chronic resistance being the major reasons. Only eight (1.5 %)
(n = 63; 30.1 %) renal failure. Malignancy was present patients on treatment suffered any adverse reactions
in 117 (12.8 %) cases, 82.9 % being solid organ malig- (Electronic supplementary material Table 2). Central
nancies and 17.1 % haematological. Nearly half of the venous catheter (CVC) was removed from 32.1 % (217/
malignancies (47.9 %) were gastrointestinal while 60.7 % 676) catheterized patients, with the majority (54.4 %)
were intraperitoneal. More than a third (n = 341; 37.3 %) withdrawn within 48 h of diagnosis. Kaplan–Meier ana-
of patients had undergone one or more surgical proce- lysis revealed that CVC removal significantly increased
dures in the 30 days or less prior to the onset of 30-day survival (mean ± standard error 19.0 ± 0.9 vs.
candidemia. Nearly half (n = 165; 48.4 %) of these 15.9 ± 0.6 days; p = 0.002). Regression models of the
comprised gastrointestinal, hepatobiliary and pancreatic effect of the antifungal therapy and CVC removal on the
surgeries (Electronic supplementary material Table 1). outcome after adjusting for the comorbidities (APACHE
Central venous catheterization was instituted in 74.0 % of II) showed significant survival (p = 0.016) only when
patients for a median duration of 8.0 (interquartile range treated with antifungal agents with simultaneous removal
7–26) days. Parenteral nutrition was instituted in 122 of the catheter within 48 h (Table 2). Survival was poor
(13.4 %) patients for a median of 9.0 (interquartile range in patients treated at public sector hospitals with respi-
5–16) days before the onset of candidemia. Median ratory and renal disease, invasive ventilation, CVC,
duration of any dialysis performed prior to development dialysis, corticosteroid and polymyxin therapy (Fig. 2).
of candidemia was 8.0 (interquartile range 2–15) days.
Nearly all candidemia patients were receiving antibi-
otics (n = 849; 93.0 %) with the majority on broad- Outcome and mortality predictors
spectrum agents (n = 784; 92.3 %) before the onset of
candidemia. The median duration of any antibiotic ther- Candidemia microbiologically resolved in 251 (27.5 %)
apy prior to development of candidemia was 16.0 cases. The 30-day crude mortality was 44.7 % (408/913)
(interquartile range 7–36) days and that for broad-spec- and 30-day attributable mortality was 19.6 % (179/913)
trum agents was 11.0 (interquartile range 4–23) days (Electronic supplementary material Table 1). Univariate
(Electronic supplementary material Table 1). Corticoste- analysis between survivors and non-survivors revealed 31
roids were given to 164 cases (18.0 %), at a median dose key factors associated with 30-day crude mortality
of 50.0 mg per day (interquartile range 25–100) and (Electronic supplementary material Table 3). Logistic
median duration of 7.0 (interquartile range 4–12) days regression was used to construct a predictive model for
before the onset of candidemia. 30-day crude mortality (R2 = 0.399, p \ 0.001,
Table 1 Antifungal susceptibility profile of highest burden Candida species circulating among adult candidemia intensive care patients from India, based on the CLSI M27-S4
(2012) guidelines

Antifungal AFST All species C. tropicalis C. albicans C. parapsilosis C. auris C. glabrata C. rugosa C. krusei C. guilliermondii
(n = 918) (n = 382) (n = 192) (n = 100) (n = 52) (n = 65) (n = 29) (n = 16) (n = 16)

Amphotericin B MIC50 (lg/ml) – 0.50 0.25 0.50 1.00 0.25 0.25 0.50 0.25
MIC90 (lg/ml) – 1.00 1.00 1.00 2.00 1.00 1.00 1.00 1.00
Resistant (%) 2.1 % 4 (1.0) 1 (0.5) 2 (2.0) 7 (13.5) 2 (3.1) 1 (3.4) 0 (0.0) 0 (0.0)
MIC percentile (25–75) 0.25–1 0.12–1 0.25–1 0.25–1 0.12–0.5 0.12–0.5 0.25–0.7 0.22–1
Fluconazole MIC50 (lg/ml) – 0.50 0.50 1.00 8.00 0.50 0.50 6.00 0.50
MIC90 (lg/ml) – 2.00 2.00 4.00 64.00 2.00 8.00 8.00 4.00
Resistant (%) 6.2 % 10 (2.6) 10 (5.2) 4 (4.0) 16 (30.8) 1 (1.5) 0 (0.0) 16 (100.0) 0 (0.0)
SDD (%) 11.0 % 9 (2.4) 8 (4.2) 9 (9.0) 7 (13.5) 64 (98.5) 1 (3.4) 0 (0.0) 0 (0.0)
MIC percentile (25–75) 0.25–1 0.12–1 0.25–1 1–64 0.25–1 0.5–2 3.25–8 0.5–1
Itraconazole MIC50 (lg/ml) – 0.06 0.06 0.06 0.03 0.03 0.03 0.12 0.12
MIC90 (lg/ml) – 0.12 0.25 0.12 0.50 0.12 0.12 0.50 1.00
Resistant (%) 1.2 % 1 (0.3) 1 (0.5) 1 (1.0) 2 (3.8) 0 (0.0) 0 (0.0) 1 (6.3) 1 (6.3)
SDD (%) 9.3 % 27 (7.1) 22 (11.5) 2 (2.0) 11 (21.2) 4 (6.2) 2 (6.9) 4 (25.0) 4 (25.0)
MIC percentile (25–75) 0.03–0.12 0.03–0.12 0.03–0.06 0.03–0.18 0.03–0.06 0.03–0.06 0.05–0.25 0.03–0.25
Posaconazole MIC50 (lg/ml) – 0.03 0.03 0.03 0.06 0.03 0.03 0.12 0.06
MIC90 (lg/ml) – 0.25 0.25 0.12 0.50 0.12 0.25 0.50 0.25
MIC percentile (25–75) 0.03–0.12 0.03–0.06 0.03–0.06 0.03–0.18 0.03–0.06 0.03–0.12 0.03–0.12 0.03–0.15
Voriconazole MIC50 (lg/ml) – 0.12 0.06 0.06 0.50 0.06 0.06 0.25 0.06
MIC90 (lg/ml) – 0.50 0.50 0.25 1.00 0.50 1.00 0.50 1.00
Resistant (%) 5.6 % 31 (8.1) 15 (7.8) 3 (3.0) 2 (3.8) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)
SDD (%) 22.9 % 128 (33.5) 58 (30.2) 18 (18.0) 1 (1.9) 1 (1.5) 2 (6.9) 0 (0.0) 0 (0.0)
MIC percentile (25–75) 0.06–0.25 0.03–0.25 0.03–0.12 0.12–1 0.03–0.12 0.03–0.25 0.12–0.25 0.03–0.07
Anidulafungin MIC50 (lg/ml) – 0.03 0.03 0.25 0.12 0.03 0.12 0.06 0.12
MIC90 (lg/ml) – 0.25 0.25 1.00 1.00 0.25 2.00 0.50 2.00
Resistant (%) 1.7 % 8 (2.1) 2 (1.0) 0 (0.0) 0 (0.0) 4 (6.2) 0 (0.0) 0 (0.0) 0 (0.0)
Intermediate (%) 1.6 % 8 (2.1) 3 (1.6) 0 (0.0) – 3 (4.6) – 1 (6.3) 0 (0.0)
MIC percentile (25–75) 0.03–0.06 0.03 0.06–1 0.06–0.04 0.03 0.03–0.5 0.03–0.09 0.03–0.62
Caspofungin MIC50 (lg/ml) – 0.25 0.25 0.50 0.50 0.25 0.50 0.50 0.50
MIC90 (lg/ml) – 0.50 0.50 1.00 2.00 0.50 2.00 1.00 1.00
Resistant (%) 5.6 % 16 (4.2) 7 (3.6) 0 (0.0) 4 (7.7) 15 (23.1) 2 (6.9) 3 (18.8) 0 (0.0)
Intermediate (%) 10.1 % 50 (13.1) 19 (9.9) 0 (0.0) – 19 (29.2) – 5 (31.3) 0 (0.0)
MIC percentile (25–75) 0.12–0.25 0.12–0.25 0.25–0.5 0.25–1 0.12–0.25 0.5–1 0.25–0.5 0.25–0.62
Micafungin MIC50 (lg/ml) – 0.03 0.03 0.25 0.12 0.03 0.06 0.05 0.12
MIC90 (lg/ml) – 0.12 0.25 1.00 1.00 0.12 2.00 0.50 1.00
Resistant (%) 1.7 % 5 (1.3) 2 (1.0) 0 (0.0) 0 (0.0) 4 (6.2) 1 (3.4) 0 (0.0) 0 (0.0)
Intermediate (%) 2.2 % 11 (2.9) 4 (2.1) 1 (1.0) – 3 (4.6) – 1 (6.3) 0 (0.0)
MIC percentile (25–75) 0.03 0.03 0.06–0.5 0.03–0.25 0.03 0.03–0.5 0.03–0.12 0.3
AFST antifungal susceptibility testing, MIC minimum inhibitory concentration, SDD susceptible dose dependent
Table 2 Logistic regression model depicting the effect of anti- younger and comparatively less serious patients acquiring
fungal therapy and catheter removal (within 48 h of diagnosis) on candidemia may be due to prior exposures to broad-
the outcome of the candidemia patients
spectrum antibiotics and steroids in large numbers of our
Treatment modalities OR P 95 % CI patients and compromised health care.
Our study revealed high prevalence of C. tropicalis
Lower Upper (41.6 %) while C. albicans and C. parapsilosis affected
No antifungal ? no catheter 1.000 – – –
only 20.9 and 10.9 % of cases, respectively. This is in
removal contrast to the developed world, where C. tropicalis is
No antifungal ? catheter removal 1.006 0.996 0.125 8.064 uniformly less common (5.6–12.0 %) [11, 17–20], and
Antifungal ? no catheter removal 0.655 0.171 0.358 1.200 C. albicans (45.0–74.0 %) and C. glabrata (16.7–22.6 %)
Antifungal ? catheter removal 0.389 0.016 0.180 0.839 are more prevalent [1, 15–17, 20]. We encountered
APACHE II at admission 1.117 \0.001 1.058 1.180
C. glabrata candidemia in only 7.1 % of patients. The rise
OR odds ratio, CI confidence intervals, APACHE II Acute Physi- in C. glabrata in the Western world can be linked to
ology and Chronic Health Evaluation II increased use of azole prophylaxis [1, 11, 15, 21]. Though
our patients had a significantly higher propensity of
predictive accuracy = 69.5 %, area under ROC developing candidemia due to non-albicans Candida
curve = 0.75). The most significant independent predic- species after azole exposure, such a link is not possible in
tors of 30-day crude mortality include admission to a our patients with C. tropicalis candidemia. Prior anti-
public sector hospital, APACHE II score at admission, fungal exposure (p = 0.09), prior azole therapy
underlying renal failure, central venous catheterization (p = 0.24) or prior fluconazole therapy (p = 0.26) was
and steroid therapy (Table 3). not significantly high in C. tropicalis candidemia. More-
over, the majority of our C. tropicalis isolates were
sensitive to amphotericin B (n = 378, 99.0 %), azoles
(n = 344, 90.1 %), fluconazole (n = 372, 97.4 %) and
Discussion echinocandins (n = 360, 94.2 %), thereby indicating
interplay of other undetermined factors for their high
This is the first and largest multicentre observational prevalence. In the present study, we did not attempt to
study of ICU-acquired candidemia from India. We found find the possible reasons for the high prevalence of
an overall incidence of 6.51 cases per 1,000 ICU admis- C. tropicalis candidemia as it was purely an observational
sions, though the incidence varied significantly across the study. A study form Paris and surrounding areas noted
country. The variation may be due to difference in patient pre-exposure to the fluconazole as an independent risk
groups and management protocol of respective ICUs. In factor for acquiring C. tropicalis candidemia [22],
addition, significant variation in the rate of candidemia whereas the present study did not find this to be an
between private/corporate and public sector hospitals was independent risk factor for C. tropicalis candidemia. A
noted. A large number of patients attend public sector detailed case–control and environmental study is essential
hospital (in the coordinating centre 10,000 new patients to clarify the reason for the high prevalence of C. tropi-
attend outpatient departments every working day) as the calis candidemia in India. In an earlier study at the
hospital charges are subsidized, which may lead to com- coordinating centre, 82 % of health care providers were
promise in the standard of the health care and higher rate found to carry yeast on their hands and 80 % were
of candidemia. In comparison to the nationwide survey C. tropicalis [23]. Therefore, horizontal transmission and
reports from other countries, the incidence in the present compromise of infection control systems are distinct
study approaches the higher end of values from Australia, possibilities for the high rate of C. tropicalis candidemia
France, Germany and the EPIC II study (0.24–6.9/1,000 in India. The high rate of C. tropicalis candidemia was
ICU admission) [1, 11–13], yet is much lower than observed in other Asian countries as well [24]. But no
reported in Spain and Argentina (34.3/1,000 ICU admis- study has evaluated the exact reasons for the high rate of
sion) [2]. C. tropicalis candidemia and hand carriage in these
Our adult candidemia patients were considerably countries. We noted a low prevalence of C. glabrata
younger (mean 49.7 years) than in other countries (mean despite prior exposure to antifungal agents in 15.7 % of
59.0–66.2 years) [2, 14]. Notably our patients contracted patients. The higher rate of non-albicans Candida infec-
ICU-acquired candidemia significantly earlier (8 days) tions in patients exposed to prior antifungals was
than in other series (11–15 days) (p = 0.03) [1, 15]; influenced by significantly the higher prevalence of
however, Playford et al. from Australia and Leroy et al. C. auris (32.7 vs. 14.8 %, p = 0.002) and C. krusei (37.5
from France have also reported similar early onset of vs. 15.4 %, p = 0.029). In another study Fournier et al.
infection [11, 16]. Our patients had lower mean APACHE reported a significantly higher proportion of C. parapsi-
II scores (17.2) than studies from Spain (20.1), Argentina losis due to prior exposure to caspofungin [25]. In
(20.1) and the USA (18.6) [2, 17]. The reasons for such contrast, Blot et al. reported no change in Candida species
Fig. 2 Kaplan–Meier survival curves (with confidence intervals) according to risk factors for mortality. Comparison of the survival curve
was performed by Log-rank (Mantel–Cox) test

distribution despite long-term exposure to fluconazole candidemia was more common in private/corporate sector
[26]. Therefore, more studies are required to understand ICUs, while C. auris and C. rugosa were more prevalent
the interplay of the factors responsible for any shift in in public sector ones. The emergence of MDR C. auris in
Candida species distribution. In our study, C. tropicalis India is a matter of concern, as this fungus was isolated
Table 3 Logistic regression predictive model summarizing the risk catheterization (74.0 %), invasive mechanical ventilation
factors for 30-day mortality in adult candidemia intensive care (52.9 %), urinary catheterization (75.9 %), haemodialysis
patients
(17.3 %) and total parenteral nutrition (13.4 %) prior to
Predictor OR P 95 % CI onset of candidemia were also encountered at lower
proportions in our cohort as compared to other series
Lower Upper (88.5–100 % [2, 13], 72.1–97.4 % [2, 13], 86.7–97.4 %
Public sector hospital 2.062 0.006 1.235 3.446
[2, 16], 17.5–32.5 % [11, 15] and 43.7–71.1 % [2, 11]).
APACHE II at admission 1.114 \0.001 1.061 1.169 Antifungal therapy after microbiological diagnosis of
Renal failure 2.237 0.016 1.161 4.312 ICU-acquired candidemia could be evaluated in only
Central venous catheterization 2.343 0.003 1.330 4.129 59.9 % of cases because of delayed diagnosis and early
Steroid therapy 5.425 \0.001 2.627 11.201 death of our patients unlike other studies (74.5–94.1 %)
Antifungals treatment for 0.494 0.007 0.295 0.827
candidemiaa [4, 16–18]. In certain situations the patient left for another
hospital or could not afford the therapy. The delay in
OR odds ratio, CI confidence intervals, APACHE II Acute Physi- diagnosis at a few centres was due to use of conventional
ology and Chronic Health Evaluation II biphasic media for Candida isolation instead of com-
a
Therapy instituted after definitive microbiological diagnosis of
candidemia in the ICU mercial systems. Removal of CVCs and start of targeted
antifungal therapy within 48 h of onset of candidemia
improved 30-day survival in our patients (p = 0.016) like
from 19 of 27 ICUs and comprised 5.2 % of all Candida other reports [2, 16, 29, 30], yet the catheter was removed
isolates in the present study. only in one-third of our patients. The reason is difficult to
Overall, 46.6 % of our isolates were susceptible to all explain. We did not interfere with the existing protocol of
antifungals. Species-specific resistance rates against any ICU and some critical care specialists believe those
fluconazole were 2.6 % in C. tropicalis, 5.2 % in studies where a lack of benefit of CVC removal was
C. albicans, 4.0 % in C. parapsilosis and 1.5 % in demonstrated [4, 30].
C. glabrata. Corresponding figures from across the Our 30-day crude mortality (44.7 %) is similar to the
globe include C. tropicalis (4.5–14.3 %), C. albicans EPIC II international series (42.6 %) [13]. Mortality rates
(1.4–4.4 %), C. parapsilosis (2.7–10.5 %) and C. glab- vary across the globe (35–75 %) [2]. Candidemia-attrib-
rata (5.9–93.8 %) [11, 14, 16, 20]. Likewise, resistance to utable mortality also varies widely (5.0–49.0 %) between
voriconazole (1.2–5.9 %), itraconazole (4.7 %) and ech- centres [2, 14, 18, 31] possibly because of lack of stan-
inocandins (0.3–2.2 %) from international studies are dardized criteria. Lower attributable mortality
different from our isolates [14, 20]. Low resistance of our (10.9–21.7 %) has been reported in some studies [3, 20,
C. glabrata isolates emphasizes the need to compare our 32], as compared to the present one (19.6 %).
isolates with MDR isolates from other centres, as cryptic Comparison of independent mortality predictors
species are already known under C. glabrata [27]. In modelled in our and other series highlight common and
contrast to C. glabrata, we observed more MDR C. trop- divergent patterns, though the large number (398) of
icalis, C. auris and C. krusei. variables studied might act as a limitation of the study by
Co-morbidities of our patients are similar to a certain increasing the likelihood of false positive results. How-
extent to earlier reports. Like in our study, underlying ever, we have factored in the higher alpha error expected
respiratory, renal and gastrointestinal illnesses were sig- out of multiple testing owing to the large number of
nificant in other studies as well. But they have also variables tested. A stringent Bonferroni’s test was used to
reported sizable proportions of diabetes (10.7–28.0 %), filter out only those significant variables which stood the
cardiovascular (15.8–48.3 %) and neurological (13.2 %) test. APACHE II scores in our study proved useful for
illnesses in their patients, which were much less common developing a robust predictive model as has been previ-
in our cohort, 0.1 %, 15.1 % and 19.5 %, respectively [2, ously described [30]. Other severity scores like SAPS II
11–16, 19]. Underlying malignancy was present in and SOFA, central venous catheterization, corticosteroid
12.8 % of our patients, compared to 24.6–36.1 % in other therapy, total parenteral nutrition, antibiotic use, sepsis/
studies [13, 14]. HIV/AIDS (0.4 %) and neutropenia shock and antifungal treatment are common predictors
(1.3 %) were rare, unlike in other series, 4.0–6.0 % and [15–18, 33]. In contrast, admission to a public sector ICU
6.6–19.7 %, respectively [15–17, 19]. A third of our and renal failure appear additional significant predictors
patients (37.3 %) had undergone recent surgery as com- for our patients. Similarly, increasing age, underlying
pared to 44.7–66.1 % seen in other series [2, 11]. Among diabetes, immunosuppression, neutropenia, C. glabrata
types of surgery, abdominal surgery is considered a and C. parapsilosis infection, mechanical ventilation,
leading risk factor for candidemia [28]. Half of our sur- haemodialysis, gastrointestinal surgery and underlying
geries (18.1 % of total) also were gastrointestinal, similar malignancy were unique to other countries, but did not
to other series (17.0–31.6 %) [2, 12, 28]. Central venous stand out in our study [4, 15–18, 33].
Conclusions (Apollo Hospital, Hyderabad, India); Aroma Oberoi, Ashu Sara
Mathai (Christian Medical College and Hospital, Ludhiana, India);
Shweta Sharma (Fortis Escorts Heart Institute, New Delhi, India);
This is the largest prospective multicentre study of ICU- DC Thamke (Mahatma Gandhi Institute of Medical Sciences,
acquired candidemia from India. The overall incidence of Wardha, India); A Krishna Prasad (Nizam’s Institute of Medical
ICU-acquired candidemia is high and marked by an early Sciences, Hyderabad, India); Camilla Rodrigues, Mahesh Lakhe,
onset infection after ICU admission. Our patients are Mehul Panchal, Niyati Desai (PD Hinduja, Mumbai, India); Ga-
gandeep Singh, Ashutosh Nath Aggarwal, Neerja Bhardwaj, L N
comparatively young and have lower APACHE II scores Yaddanapudi, Joseph Jillwin, A Shamnath (Postgraduate Institute
at admission. C. tropicalis is the leading pathogen in our of Medical Education and Research, Chandigarh); Pradeep Kumar
patients, though its predominance is not explained by Verma, Harish Chand Sachdeva (Safdarjang Hospital, New Delhi,
higher exposure to prior azoles or antifungals. Emergence India); Sriram Sampath (St John’s Medical College, Bangalore,
India) are also acknowledged for their help. This work was sup-
of MDR C. auris in the majority of ICUs is a matter of ported by the MSD Pharmaceuticals Pvt. Ltd Educational Grant
concern. Our 30-day mortality prediction model can help through the Society for Indian Human and Animal Mycologists, an
triage and manage high-risk candidemia patients better. affiliate of the International Society of Human and Animal
Mycology. MSD did not play any role in study design, data analysis
Acknowledgments We wish to acknowledge Prof. Niranjan Na- or manuscript writing.
yak, President SIHAM for providing us invaluable logistic support
and continuous encouragement to accomplish this study. Other Conflicts of interest The authors declare that they have no con-
members of the SIHAM Candidemia Network include (participat- flicts of interest and no financial relationship with the funding
ing centres in parenthesis; in alphabetical order): Purva Mathur (All agency.
India Institute of Medical Sciences, New Delhi, India); Ratnamani

References
1. Meyer E, Geffers C, Gastmeier P et al 8. Maubon D, Garnaud C, Calandra T et al 15. Chow JK, Golan Y, Ruthazer R et al
(2013) No increase in primary (2014) Resistance of Candida spp. to (2008) Risk factors for albicans and
nosocomial candidemia in 682 German antifungal drugs in the ICU: where are non-albicans candidemia in the
intensive care units during 2006–2011. we now? Intensive Care Med intensive care unit. Crit Care Med
Euro Surveill 18(24):20505 40(9):1241–1255 36(7):1993–1998
2. Gonzalez de Molina FJ, Leon C, Ruiz- 9. CLSI (2008) Reference method for 16. Leroy O, Gangneux JP, Montravers P
Santana S et al (2012) Assessment of broth dilution antifungal susceptibility et al (2009) Epidemiology,
candidemia-attributable mortality in testing of yeasts, approved standard. management, and risk factors for death
critically ill patients using propensity CLSI document M27-A3, 3rd edn. of invasive Candida infections in
score matching analysis. Crit Care Clinical and Laboratory Standards critical care: a multicenter, prospective,
16(3):R105 Institute (CLSI), Wayne, PA observational study in France
3. Nolla-Salas J, Sitges-Serra A, Leon-Gil 10. CLSI (2012) Reference method for (2005–2006). Crit Care Med
C et al (1997) Candidemia in non- broth dilution antifungal susceptibility 37(5):1612–1618
neutropenic critically ill patients: testing of yeasts, fourth informational 17. Pappas PG, Rex JH, Lee J et al (2003)
analysis of prognostic factors and supplement. CLSI document M27-S4. A prospective observational study of
assessment of systemic antifungal Clinical and Laboratory Standards candidemia: epidemiology, therapy, and
therapy. Study group of fungal infection Institute, Wayne, PA influences on mortality in hospitalized
in the ICU. Intensive Care Med 11. Playford EG, Marriott D, Nguyen Q adult and pediatric patients. Clin Infect
23(1):23–30 et al (2008) Candidemia in Dis 37(5):634–643
4. Charles PE, Doise JM, Quenot JP et al nonneutropenic critically ill patients: 18. Chen S, Slavin M, Nguyen Q et al
(2003) Candidemia in critically ill risk factors for non-albicans Candida (2006) Active surveillance for
patients: difference of outcome between spp. Crit Care Med 36(7):2034–2039 candidemia, Australia. Emerg Infect
medical and surgical patients. Intensive 12. Leroy G, Lambiotte F, Thevenin D et al Dis 12(10):1508–1516
Care Med 29(12):2162–2169 (2011) Evaluation of Candida score in 19. Leroy O, Mira JP, Montravers P et al
5. Guery BP, Arendrup MC, Auzinger G critically ill patients: a prospective, (2010) Comparison of albicans vs. non-
et al (2009) Management of invasive multicenter, observational, cohort albicans candidemia in French intensive
candidiasis and candidemia in adult study. Ann Intensive Care 1(1):50 care units. Crit Care 14(3):R98
non-neutropenic intensive care unit 13. Kett DH, Azoulay E, Echeverria PM 20. Pfaller MA, Messer SA, Moet GJ et al
patients: part I. Epidemiology and et al (2011) Candida bloodstream (2011) Candida bloodstream infections:
diagnosis. Intensive Care Med infections in intensive care units: comparison of species distribution and
35(1):55–62 analysis of the extended prevalence of resistance to echinocandin and azole
6. Lass-Florl C (2009) The changing face infection in intensive care unit study. antifungal agents in intensive care unit
of epidemiology of invasive fungal Crit Care Med 39(4):665–670 (ICU) and non-ICU settings in the
disease in Europe. Mycoses 14. Bassetti M, Merelli M, Righi E et al SENTRY antimicrobial surveillance
52(3):197–205 (2013) Epidemiology, species program (2008–2009). Int J Antimicrob
7. Pfaller MA, Diekema DJ (2007) distribution, antifungal susceptibility, Agents 38(1):65–69
Epidemiology of invasive candidiasis: a and outcome of candidemia across five
persistent public health problem. Clin sites in Italy and Spain. J Clin
Microbiol Rev 20(1):133–163 Microbiol 51(12):4167–4172
21. White MH (1997) The contribution of 26. Blot S, Janssens R, Claeys G et al 30. Nucci M, Anaissie E, Betts RF et al
fluconazole to the changing (2006) Effect of fluconazole (2010) Early removal of central venous
epidemiology of invasive candidal consumption on long-term trends in catheter in patients with candidemia
infections. Clin Infect Dis candidal ecology. J Antimicrob does not improve outcome: analysis of
24(6):1129–1130 Chemother 58(2):474–477 842 patients from 2 randomized clinical
22. Lortholary O, Renaudat C, Sitbon K 27. Lockhart SR, Messer SA, Gherna M trials. Clin Infect Dis 51(3):295–303
et al (2014) Worrisome trends in et al (2009) Identification of Candida 31. Gudlaugsson O, Gillespie S, Lee K et al
incidence and mortality of candidemia nivariensis and Candida bracarensis in (2003) Attributable mortality of
in intensive care units (Paris area, a large global collection of Candida nosocomial candidemia, revisited. Clin
2002–2010). Intensive Care Med glabrata isolates: comparison to the Infect Dis 37(9):1172–1177
40(9):1303–1312 literature. J Clin Microbiol 32. Das I, Nightingale P, Patel M et al
23. Chakrabarti A, Chatterjee SS, Rao KL 47(4):1216–1217 (2011) Epidemiology, clinical
et al (2009) Recent experience with 28. Leon C, Ostrosky-Zeichner L, Schuster characteristics, and outcome of
fungaemia: change in species M (2014) What’s new in the clinical candidemia: experience in a tertiary
distribution and azole resistance. Scand and diagnostic management of invasive referral center in the UK. Int J Infect
J Infect Dis 41(4):275–284 candidiasis in critically ill patients. Dis 15(11):e759–e763
24. Mootsikapun P, Hsueh PR, Talwar D Intensive Care Med 40(6):808–819 33. Colombo AL, Guimaraes T, Sukienik T
et al (2013) Intravenous anidulafungin 29. Blot SI, Vandewoude KH, Hoste EA et al (2014) Prognostic factors and
followed optionally by oral et al (2002) Effects of nosocomial historical trends in the epidemiology of
voriconazole for the treatment of candidemia on outcomes of critically ill candidemia in critically ill patients: an
candidemia in Asian patients: results patients. Am J Med 113(6):480–485 analysis of five multicenter studies
from an open-label phase III trial. BMC sequentially conducted over a 9-year
Infect Dis 13:219 period. Intensive Care Med
25. Fournier P, Schwebel C, Maubon D 40(10):1489–1498
et al (2011) Antifungal use influences
Candida species distribution and
susceptibility in the intensive care unit.
J Antimicrob Chemother
66(12):2880–2886