Nosocomial infections in a pediatric intensive care unit of a developing country : NHSN surveillance

Introduction: This study aimed to determine the epidemiology of the three most common nosocomial infections (NI), namely, sepsis, pneumonia, and urinary tract infection (UTI), in a pediatric intensive care unit (PICU) in a developing country and to define the risk factors associated with NI. Methods: We performed a prospective study on the incidence of NI in a single PICU, between August 2009 and August 2010. Active surveillance by National Healthcare Safety Network (NHSN) was conducted in the unit and children with NI (cases) were compared with a group (matched controls) in a case-control fashion. Results: We analyzed 172 patients; 22.1% had NI, 71.1% of whom acquired it in the unit. The incidence densities of sepsis, pneumonia, and UTI per 1,000 patients/day were 17.9, 11.4, and 4.3, respectively. The most common agents in sepsis were Enterococcus faecalis and Escherichia coli (18% each); Staphylococcus epidermidis was isolated in 13% of cases. In pneumonias Staphylococcus aureus was the most common cause (3.2%), and in UTI the most frequent agents were yeasts (33.3%). The presence of NI was associated with a long period of hospitalization, use of invasive devices (central venous catheter, nasogastric tube), and use of antibiotics. The last two were independent factors for NI. Conclusions: The incidence of NI acquired in this unit was high and was associated with extrinsic factors.

Nosocomial infection (NI) constitutes a major health problem associated with high morbidity, mortality, and increase of healthcare costs, especially in pediatric intensive care units (PICU).Patients in these units, despite representing a small percentage of inpatients, contribute to over 20% of NI 1 .Bloodstream infections (BSI) are the most common NI in these units (28-52% of all) 2,3 , followed by pneumonia and urinary tract infection (UTI) 3,4 .The first two are responsible for approximately 50% of NIs, with UTI causing an additional 12% to 22% 5 .A previous study developed in a Brazilian PICU showed sepsis as a major cause (18.6%) of death 6 .
In Europe, incidence ranges from 1% in general pediatric wards to 23.6% in PICUs 7 .In Brazil, despite the lack of information about this issue in children there are many studies on adult patients, including NI risk factors and measures to prevent a patient from acquiring NI, and the work is extremely heterogeneous as far as number of beds, services available, and patients are concerned 8 .Studies have reported that less than 10% of beds are available to intensive care units in Brazilian hospitals, despite the fact that they represent 40% of NIs 9,10 .In this study, we investigated the epidemiological profile of the three most common NIs (sepsis, pneumonia, and UTI) in a PICU of a Brazilian university hospital and the risk factors associated with these infections.

Hospital
The Hospital das Clínicas of the Universidade Federal de Uberlândia, State of Minas Gerais, Brazil, is a public teaching hospital of more than 500 beds, with the PICU presenting eight beds.The hospital offers tertiary care.

Design of the study
This survey was divided into two stages.The first stage entailed a prospective cohort study in the PICU

RESULTS
Porto JP et al -Nosocomial infection in PICU of developing country: NHSN surveillance of the Hospital das Clínicas of the Universidade Federal de Uberlândia by the National Health Care Safety Network (NHSN) system 11 .The children were followed until discharged or death.Every inpatient hospitalized from August 2009 to August 2010 was initially considered a potential participant, and those with stays of >48h or longer were included.After the first phase, a nested case-control study was conducted to identify the associated factors for NI.The cases were children with NI, whereas the controls were children without infections.The medical records of patients were reviewed for demographic and risk factor data.Data collected included gender, length of stay, underlying diseases, invasive procedures, infection on admission, and PICU-acquired infection.

Definitions
PICU-acquired NIs were defined according to the Centers for Disease Control and Prevention (CDC) 12 .Infections that commenced at or after 48h after admission to the PICU were included as PICU-acquired infections.
Urinary tract infection: the patient must have at least one of the following signs or symptoms with no other recognized cause: fever (>38°C), urgency, frequency, dysuria, and positive urine culture with counts ≥105 colony-forming units per milliliter (CFU/ml).
Pneumonia: the criteria for the definition of pneumonia were chest radiograph with new pulmonary infiltrate or progression of an existing one, accompanied by two of the following signs or symptoms: leukocytosis (>0,000/mm 3 ) or leukopenia (<4,500/ mm 3 ), hyperthermia (>38 o C) or hypothermia (<35 o C), purulent sputum, tracheal aspirate bacterial count of ≥106CFU/ml.
Bacteremia: this was defined as the biological documentation of infection, i.e., the result of a positive blood culture.
Sepsis: this was defined as a systemic response to an infection, followed by one or more of these conditions: a) temperature >38 o C or <36 o C; b) heart rate >90 beats/min; c) respiratory rate >20 breaths/ min or PaCO2 <32mmHg; d) leukocyte count >12,000cells/mm 3 , or >10% of juveniles; e) sepsis with hypotension, associated with the presence of perfusion abnormalities that may include lactic acidosis, oliguria, or acute alteration in mental status defined as septic shock; total patients per day was defined as the somatory of total time of hospitalization by the patients in PICU; epidemiological investigations' incidence rate was defined as the number of each infection as the numerator and the number of days those patients were at risk as the denominator per 1,000 13 ; device-associated incidence rate was defined by dividing the number of each infection by the number of days those patients were exposed to risk factors per 1,000 13 .

Statistical analysis
Significant differences between groups were determined using the X 2 -test or Fisher , exact test, where appropriate.Statistical significance was defined by p ≤ 0.05.Association between variables and nosocomial infections were expressed as odds ratios (OR) with their respective 95% confidence intervals (95%CI) obtained by means of conditional logistic regression models.Statistical analyses were performed using Epi-Info (version 3.3.2) and Bioestat 5.0 (CDC, Atlanta).

Ethical considerations
Approval for the study was granted by the research ethics committee of the hospital.Patient names were not disclosed, and all the information about them was kept confidential.All the samples were analyzed by the hospital's microbiology laboratory and recovered to another's tests.
From August 2009 to August 2010 a total of 172 patients were hospitalized in the PICU with a mean stay of 8.1 days.Of these patients, 38 (22.1%) were infected at least once in the PICU, with 60 episodes, and 8.7% had community-acquired infection.The PICU mortality rate for the population during the study period was 8.1%.Table 1 shows the demographic features, co-morbidities, use of devices, length of stay, and mortality of the patients.
The mean overall patient NI rate was 27.2 per 1,000 patients/day, with an incidence of 22.1% (patients with NI).The rates of primary BSI, pneumonia, and UTI per 1,000 device-days were 18.2, 17.8, and 7.0, respectively.Table 2 shows the potential risk factors of NI associated with unvaried analyses.Risk factors significantly associated with NI included use of central venous catheter (CVC) [p=0.001,odds ratio (OR) = 8.77], length of stay (p<0.001),nasogastric tube (p=0.0002,OR=5.00), and use of antibiotics (p=0.0004,OR=7.89).These same factors, with the exception of length of hospitalization, were also independently associated with infection by multivariate analyses.Gram-negative bacilli were the most common pathogens identified in the NI in our PICU (47%), followed by Gram-positive cocci bacteria (44.1%).Sepsis was more commonly reported, and 60% of cases had at least one microorganism isolated in blood culture.The most common agents were Enterococcus faecalis (18%) and Escherichia coli (18%), and among non-fermenting bacilli only Acinetobacter baumannii was detected (4%).Staphylococcus epidermidis was isolated in 13% of cases.The pathogens isolated from pneumonia were Staphylococcus aureus as the most common cause (3.2%), and among isolates of Gram-negative bacilli, 1.5% were Pseudomonas aeruginosa.In three children, more than one microorganism was isolated.In UTI cases (83.3%) at least one microorganism was isolated.Yeasts were more commonly reported in UTI (33.3%), and S. aureus and P. aeruginosa were uncommon (Figure 1).

DISCUSSION
It is well established that NI is associated with high rates of morbidity, mortality, and significant economic cost that tend to represent a larger problem in hospitals in developing countries than in developed ones 14 .Although there are abundant data from epidemiological studies about these infections in PICU, most of the research has not come from developing countries such as Brazil, and has mainly been expressed in infection rates per hospital/day or device/day 15,16 .Analysis of data obtained in our study suggests that surveillance systems, like the NHSN, are important for providing feedback from individual PICUs, allowing comparison with other literature data.
The authors declare that there is no conflict of interest.

REFERENCES
In our investigation, the data showed a high frequency of these infections (27.2 per 1,000 patients/day; 22.1%), with a significant proportion (71.1%) of PICU-acquired infection, as also reported in the PICU of Hospital São Paulo, which showed an NI incidence rate of 18% 6 .Sepsis was considered as community-induced if it manifested up to 72h after the patient ' s confinement, and if the patient did not come from another hospital environment 12 .Due to underlying diseases of individuals and their need for long hospitalization, these patients can develop infection from some pathogens restricted to hospital environments, making the recovery of patients harder and making these patientesan infection source that spreads these infections to the community 17 .In our study only 8.7% of infections were acquired in the community.
In intensive care units where there is usually a high density of antibiotic use, resultant bacteria tend to be more common, isolates are often resistant and multi-resistant, and horizontal dissemination is common 18 .In this group the risk for acquiring NI is from five to ten times greater 19 .The major risk factors for the acquisition of NI in pediatric patients are the severity of the underlying disease, the presence of an invasive device, longer time of hospitalization in PICU, high population density, and use of antibiotics 20 .In this series, we verified as statistically significant the following risk factors: use of invasive devices such as CVC or nasogastric tubes, length of stay in PICU, and antibiotics use.However, only the use of antibiotics and the use of CVCs and nasogastric tubes were independently associated with NI, similar to results reported elsewhere [20][21][22] .The length of stay in our PICU was found to be the risk factor for NI by unvaried analyses, similar to other studies, and long stay reflects the severity of the underlying disease, requiring greater care 6,23 .
As in adults, most BSI in PICU are associated with the use of a CVC 2,15,20 .Our data showed that the incidence of sepsis was 17.9/1,000 patient days, accounting for 65.8% of all NI.A higher proportion of all BSI was related to CVCs, with the incidence density rate (18.2/1,000CVC days) observed in the present study higher than the NHSN data for a clinical-surgical PICU 27 .
Enterococcus faecalis and Escherichia coli as etiological agents of infections (18% each), followed by Enterobacter cloacae and S. epidermidis (13% each), were the main agents of sepsis.Bloodstream infection is classified as primary or secondary based on the presence or the absence of knowledge focus outside the vascular system.The main focus of secondary infections is the lung, responsible for 50% of all cases 28 .Primary sepsis is usually (85%) related to a CVC, and shows a mortality rate between 12% and 25%, lengthening the patient ' s hospitalization time from ten to forty days 29,30 .In our investigation, primary and secondary sepsis rates were similar (53.3% and 46.6%, respectively).The CVC was the main origin in the first at about 50% of cases, with the lung the main origin in secondary sepsis at about 50% of cases.
Pneumonia associated with mechanical ventilation is considered the most common infection in adult diagnosis, but isolating a single type in children is difficult [31][32][33] .There are few studies on pediatric patients, and incidence rates range from 2% to 68% 34 .Surveillance studies in PICU patients have reported that pneumonia accounts for 2% to 17% of NI 27,35,36 .In this study pneumonia was the second most common infection (30.2%) with an incidence rate of 11.4 per 1,000 patients a day and 17.8 per 1,000 ventilation days.
According to the literature, Gram-negative bacilli are the most frequent pathogens in these infections, either in adults or in children, in developing countries 2,26,27,[36][37][38] .Our results differed, with S. aureus as the most frequent agent.
In our study the frequency of UTI/1,000 urinary catheter (UC) days was greater than in studies realized in developing countries 46 , and although these infections are responsible for almost 40% of all NI, they continue to be unrepresentative in Brazilian hospitals, considering that the majority of them (80%) are associated with the use of urinary catheters 47,48 .Fifty-eight percent of our patients were using this invasive device, with an infection rate of 7.0/1,000 UC days.
Our data suggest that frequent surveillance systems, like the NHSN, were important to evaluate the association of these wellknown risk factors with PICU risk factors and causative organisms, as a high frequency of these infections was associated with extrinsic risk factors and the predominance of Gram-negative bacilli followed by Gram-positive cocci.
This study has some limitations, particularly when it is to be compared with others.First, we studied only three types of infections.Second, we had a relatively small number of patients in the case group, thus reducing the statistical power and the ability to study subsets of patients.Furthermore, information on the management of the condition and on the severity of the underlying illness was unviable, and because all our study patients where hospitalized in a single tertiary hospital, our results may not be generalized to other institutions.

FIGURE 1 -
FIGURE 1 -Pathogens identified in nosocomial infections developed by children hospitalized in the PICU of the Hospital das Clínicas of the Universidade Federal de Uberlândia, Brazil, between August 2009 and August 2010.

TABLE 1 -Characteristics of patients admitted to the PICU of the Hospital das Clínicas of the Universidade Federal de Uberlândia, Brazil, between August 2009 and August 2010. Patients (n = 172) Variables n %
PICU: pediatric intensive care unit; CVC: central venous catheter; MV: mechanical ventilation; NT: nasogastric tube; *hypertension, cerebral palsy, sickle cell disease, and short bowel syndrome.

TABLE 2 -Univariate analysis of associated factors presented by patients admitted to PICU of the Hospital das Clínicas of the Universidade Federal de Uberlândia, Brazil, between August 2009 and August 2010. Patients with NI (n=38) Patients without infection (n=119)
PICU: pediatric intensive care unit; SD: standard deviation; OR: odds ratio; CVC: central venous catheter; MV: mechanical ventilation; NT: nasogastric tube; *cystic fibrosis, cerebral palsy, sickle cell disease, and short bowel syndrome.