eonatal sepsis is frequently due to organisms colonizing the skin and mucosal surfaces, such as Coagulase-negative Staphylococci and Candida (1). Blood stream infection (BSI) due to Candida species in the neonatal intensive care unit (NICU) is less frequent than that due to Gram-positive or Gram-negative bacteria, but it has higher morbidity and mortality rates (2). Candida is the third most common etiologic agent in late-onset neonatal sepsis and is responsible for 8 to 15% of hospital-acquired infections (3). Preterm infants have high Candida colonization rates compared to term infants, and it is well established that colonization with Candida is inversely proportional to gestational age (4,5). Colonization precedes invasive Candida infection, and the number of colonization sites and density of skin colonization with Candida correlate with candidemia (6)(7)(8). Even though Candida albicans is the most prevalent yeast pathogen, BSIs caused by Candida non-albicans, particularly Candida parapsilosis complex and Candida glabrata complex has been increased in recent years (9,10). Newborns who survived from invasive candidiasis frequently have a long-term neurological impairment, including cerebral palsy, blindness, hearing impairment, cognitive deficits, and periventricular leukomalacia (9). Candida infections are responsible for an 'attributable mortality' of 18-25%, significant morbidity, and healthcare costs (11). The incidence and associated mortality due to candidemia can be influenced by several factors, including the population at risk, healthcare facility standards, Candida spp. involved and anti-fungal resistance (12).

Clinical presentation of candidemia resembles sepsis syndrome and to establish a clinical diagnosis of candidemia is difficult. (13). Here, in this study, we are trying to identify the clinical and laboratory profile of neonatal candidemia and the risk factors associated with it.

Ethical consideration: Permission to conduct this study was obtained from the institutional research committee and ethical committee of Government T. D Medical College, Alappuzha.

This study was a retrospective case-control study. Sex matched cases and controls were selected through consecutive sampling by file review of newborns of NICU of Govt. T.D. Medical College from 1 st January 2010 to 31 ST December 2014.Cases were identified through the review of a meticulously maintained NICU logbook in which relevant details N (including the IP number, name, age, sex, date of admission, discharge diagnosis, date of discharge/death, and status at discharge) of all admitted cases were entered. We also reviewed the blood culture record of all cases and controls for final case confirmation from Microbiology Laboratory. All babies in whom blood culture yielded candida were considered as cases. Two newborns of the same-sex as that of the case and admitted within one week prior or after the case but negative for candida blood culture were considered for controls. New born babies who were on antibiotics before sampling for blood culture and whose case books incomplete or not available were excluded.

Case records of all these newborns were retrieved from the medical record library. After applying exclusion criteria, cases and controls were finalized. Details of risk factors (maternal, neonatal, and nosocomial), clinical features, investigation results, and outcome of all and controls were recorded in the proforma and analyzed.

Candidemia: The blood culture yields candida species organisms. Silent hypoxemia: Low Spo2 with no tachypnea/ bradypnea/apnoea/ respiratory distress/ hypothermia and with no evidence of shock.

Feed intolerance: The enteral milk feeding has to be stopped due to presence of one or more of the following-gastric residuals >50% of previous feed, vomiting, abdominal distension, visible bowel loops, and blood in the stool (occult/ gross).

The gastric residue is 20-50% of previous feed consecutively for at least three times without features of feed intolerance.

For statistical analysis, continuous variables were summarized using mean and standard deviation and categorical variables using frequencies and percentages. To identify risk factors of candida infection, Chi-square test was used, and for the strength of association odds ratio was used. Binary logistic regression was performed using the enter method to identify independent predictors of neonatal candidemia and mortality associated with candidemia. All analysis was performed using SPSS V18.

IV.

From1st January 2000 to 31 st December 2014, there were 2754 admissions to the newborn nursery. Out of this, 107(3.9%) had candidemia. Seven babies who were started on fluconazole before blood culture and 6 with incomplete case records were excluded from the study. Finally, 94 cases and 188 controls were included in this study. 92(97.9%) cases were Candida nonalbicans, and the rest were candida albicans. Cases and controls were comparable concerning sex, route of delivery, and gravidity. The frequency of Gestational diabetes mellitus, Gestational hypertension, PROM, and 3 rd trimester UTI was also comparable (Table -1). There was a significant difference in the mean age of admission between cases and controls. Symptoms &signs such as lethargy ( 67%), respiratory distress (67%), apnoea (35.1%), silent hypoxia (21.3%), seizure(14.9%), feed intolerance (33%), sub-threshold feed tolerance(16%), weak cry (86.2%), prolonged CFT (48.9%), mucosal candida infection (5.3%), abdominal distension (54.3%) and hepato-splenomegaly (22.3%) was significantly higher in cases (table-2). Newborn babies with candida BSI showed a significantly higher proportion of thrombocytopenia (75.5%), positive Creactive protein (47.9%), and abnormal CSF (9.6%). Urine fungal hyphae (14.9%) were isolated only from neonates with candidemia. The mean WBC count was significantly lower in cases (10858) compared to controls (14191) (table-2). As many of these risk factors are interdependent, inorder to identify the independent risk factors of candidemia, we did binary logistic regression. The independent risk factors were; more than two antibiotics, amino acid infusion, neonatal hyperglycemia before blood culture, and maternal vaginal candidiasis (table-4). The mortality rate among admitted newborns with candidemia and without candidemia was 18.1% and 8.5%,respectively. This finding was statistically significant. The mean duration of hospital stay among cases was also significantly higher. (table-5) On univariate analysis, we found that the following were predictors for mortality in neonatal candidiasis; vaginal delivery, preterm <32 weeks GA, birth weight <1.5 kg, lethargy, respiratory distress, apnoea, prolonged CFT, skin mottling, endo tracheal intubation, assisted ventilation, administration of caffeine or surfactant ,and abdominal surgery. (Table -6)

Multivariate regression analysis found out only three independent predictors of mortality. These were apnoea, abdominal surgery, and active resuscitation.(Table - V.

We report a prevalence of 3.89%for neonatalcandidemia during the study period. This prevalence rate is comparable to the prevalence reported from developed countries. (14-16) Our study found a significant difference in the age of admission between cases and controls contrary to a study from China (17).This may be due to the difference in indications for admission among the hospitals.

The clinical features associated with neonatal candidemia reported in our study namely lethargy, respiratory distress, apnoea, silent hypoxia, seizure, feed intolerance, sub-threshold feed tolerance, weak cry, prolonged CFT, mucosal candida infection, abdominal distension, and hepatosplenomegaly were similar to that reported in the literature (18,19).The two new signs we identified were silent hypoxemia (specificity-99.46% and sensitivity-21%) and subthreshold feed tolerance (specificity-100% and sensitivity-15.9%),which is not reported in other studies. In some of the cases, these early signs were the reason for suspecting candidiasis.

We found that neonatal candidemia was significantly associated with thrombocytopenia, positive CRP, a relatively low total WBC count, and abnormal CSF, which is consistent with reports elsewhere.

Most of the significant riskfactors for neonatal candidemia we obtained were documented in the literature (18,19,22,23). Many of these factors are interdependent. Regression analysis revealed only four independent risk factors that are significantly associated with neonatal candidemia. Though the administration of ranitidine and abdominal surgery are reported to be significant risk factors, we could not find this association.

The death rate of neonates with candidemia was significantly higher than neonates without candidemia (18.1% vs. 8.5%). Other studies also report a higher death rate among systemic candidiasis compared to other inpatients of the NICU (24).The duration of hospitalization was also high for cases.

The independent risk factors of mortality in neonatal candidemia were apnoea, abdominal surgery, and invasive resuscitation. The increased mortality in the babies who underwent abdominal surgery may be linked to large inoculum size of candida organisms, which is already colonized in GIT and use of broadspectrum antibiotics after surgery, which may facilitate its growth. The anaerobic environment during apnoea and before resuscitation may facilitate adhesion, tissue invasion, and disruption of host immune function by candida through increased production of secretory aspartyl proteinases (SAP), which may increase the severity of the infection.

Our study has few limitations. We could not perform species differentiation among the candida organisms and antifungal susceptibility. Assessments of complications using USG abdomen & CT of the head could not be performed in many cases because of logistic problems.

In conclusion, in developing countries, Candidemia is a common cause of BSI among neonates. Candida nonalbicansis the predominant type identified. Two new signs (silent hypoxemia and subthreshold feed tolerance) identified were more specific for neonatal candidemia. Management of maternal vaginal candidiasis during pregnancy, judicious use of antibiotics in newborns, maintaining euglycemia in the newborn period, early enteral feed, and avoiding amino acid infusion may help in the reduction of neonatal candidiasis. Early identification of this problem, especially among babies who had apnoea, active resuscitation, and abdominal surgery and managing them, may help to reduce mortality.

What is already known: Clinical diagnosis of neonatal candidemia is difficult What this study adds: Silent hypoxemia and subthreshold feed tolerance are two most specific features of neonatal candidiasis.