I. INTRODUCTION

Healthcare-associated infection (HAI) is an acquired complication by the patient after his hospital admission and it is an important preventable cause of morbidity and mortality among hospitalized patients. Among HAIs causes, the Primary Bloodstream Infection (PBSI) is considered one of the most relevant nosocomial infections in pediatrics. Around 10 to $23\mathrm{\%}$ of hospitalized children have hospital-acquired bacteremia and, despite all the advances made in hospital care, PBSI is associated with a high mortality in pediatric intensive care unit (PICU). In addition, PBSI increases hospitalization time and cost.

PBSI risk factors are related to severity of underlying illnesses, age, prolonged hospitalization, invasive procedures, medications, parenteral nutrition, steroid exposure, use of central venous catheter (CVC), among others. ${}^{3,4,5,6}$ However, the main risk associated with PBSI is the use of intravascular devices, accounting approximately for $60\mathrm{\%}$ of nosocomial bacteria. ${}^7$ In the pediatric population, according to the National Nosocomial Infections Surveillance System (NNIS), PBSI rates can reach 7.3 cases per 1000 catheter-days. ${}^8$

PBSI can be diagnosed from laboratory and clinical criteria. Those with positive blood culture have more objective diagnostic criteria. Nevertheless, blood culture sensitivity is variable due to technique variation among different hospitals and laboratories.9PBSIs caused mainly by coagulase-negative Staphylococcus, accounting for $31\mathrm{\%}$.3 As PBSI have important consequences for both the patient and the health service, the analysis of incidence, of epidemiological profile and of the inpatient outcomes with PBSI are important to evaluate the need for prevention, identify possible improvements to be made in healthcare and achieve better and rationalized treatment, aiming for lower complications, morbidity, mortality, and healthcare-related costs.

II. METHODS

Descriptive, observational and cross-sectional study between March 2016 and March 2020, with retrospective data collection from medical records on the PHILIPS Tasy system (Philips Healthcare, Cambridge, MA, USA) at Dr. Jeser Amarante Faria Children's Hospital (HJAF) in Joinville-state of Santa Catarina, Brazil.

In this study were included patients diagnosed from laboratory or clinical criteria with PBSI admitted to the general pediatrics and surgical PICU of the HJAF. The laboratory diagnostic criteria for PBSI was laboratory confirmed infections, with at least one positive blood culture not related to infection at another site. The clinical diagnostic criteria for PBSI used was clinical sepsis, more precisely, when treatment for sepsis was instituted and there was no apparent infection at another site, and the blood culture was negative.

Were excluded from the study children under 28 days of age, patients whose diagnosis has changed during the period of hospitalization and patients whose medical records were missing data or with incorrect or insufficient completion.

The collected variables are age, sex, diagnosis on admission, use of invasive devices, type of device - CVC or PICC (Peripherally Inserted Central Catheter), puncture site, previous comorbidities, and length of PICU stay.

To build the database Microsoft Excel software was used. The collected data were tabulated and its analysis was performed using IBM SPSS, version 21.0. Descriptive (absolute and relative frequency) and inferential analysis (Spearman Correlation Test, Kruskal Wallis Comparison Test and Mann-Whitney Comparison Test) were used.

The choice of non-parametric tests was based on the result of the Kolmogorov Smirnov normality test, in which the variables showed a tendency towards nonnormality. In this way, data referring to the median and interquartile range were displayed. For all tests, a significance level of $5\mathrm{\%}$ (p-value ) was adopted.

This study complies with the guidelines for research involving human beings, contained in resolution 466/2012 of the National Health Council and was approved by the Ethics Committee in Research with Human Beings of the Hans Dieter Schmidt Hospital under number 4.365.075.

III. CRITICAL ANALYSIS

As it is an observational and descriptive study, it does not pose a risk to the patients. Likewise, there was no need for informed consent forms. The study also did not influence any change in patients' conduct and/or treatment. All collected data regarding patients are the sole responsibility and confidentiality of the researchers.

PBSI incidence analysis and PICU admitted patients' profile make it possible to assess the need for improvement in healthcare, aiming to reduce the mortality and morbidity resulting from disease complications and, consequently, the reduction of hospitalization time and hospital costs.

IV. RESULTS

Absolute and relative frequencies of PICU admission diagnoses and comorbidities can be seen in Table 2. Heart disease is the most common diagnosis at admission and/or comorbidity. Diagnoses such as multiple traumas, oncological pathologies, encephalitis, acute abdomen, among others, accounts for $19.67\mathrm{\%}$ of PICU admission diagnoses. Other comorbidities, such as encephalopathy and chronic lung diseases, added up to $17.21\mathrm{\%}$.

Regarding central device usage, $96.97\mathrm{\%}$ of patients had at least one catheter, of which $75.0\mathrm{\%}$ were CVC and $21.97\mathrm{\%}$ were PICC. Only $3.03\mathrm{\%}$ of the evaluated patients did not have any type of central catheter and $8.2\mathrm{\%}$ had both devices simultaneously. Insertion sites and its statistically significant relation with central device type can be observed in Table 3.

Regarding microbiological profile, 101 blood cultures $(82.79\mathrm{\%})$ showed microorganisms growth, these are shown Table 4.

V. DISCUSSION

When evaluating the characteristics of patients diagnosed with PBSI in the PICU, it was observed that most of them are infants. This age predominance is in line with other studies, it is due to the fact that this group has an immune system still in formation and many have underlying medical conditions that increase the risk of acquiring an infection.[8,10] There is not a specific pattern regarding sex as it is not a determining characteristic for infection.[2,4,8,11]

Patients with complex surgical conditions, mainly from heart diseases, have a great propensity to acquiring PBSI. This is due to the risk increase intrinsic to this population, such as: prolonged hospitalization, major surgical procedure, pro-inflammatory status, need for invasive devices, among others.[3,12,13] As evidenced by Hatachi et al., PBSIs were associated with a longer length of PICU stay, a fact that demonstrates the severity of the patient condition and, as consequence, a greater need for healthcare.[14]

Hospitalization time until infection also varies significantly according to the population under study, as some groups have aggravating characteristics for acquiring PBSI. There are reports in literature of groups with similar profiles that converged to a median of 13.4 days of length of PICU stay until PBSI, a result slightly below the median of 14.76 days found in the sample studied in this present work.15 There were no correlations found of age and duration of CVC use until infection, therefore this comparison could not be performed.

Regarding the place where PBSI diagnoses were made, a study carried out in a cardiac PICU found most PBSI diagnoses were made in postoperative surgical patients.[12] These data corroborate the findings of this sample, since most surgical PICU patients have a heart disease.

Central device use is an important risk factor for PBSI. ${}^{2,5,7,10,15}$ The majority of the patients assessed in this study had a central device (96.97%) and, of these, 75% had a CVC. In relation to the device site, unlike this study, two studies show there is an infection predominance in femoral vein inserted catheters. ${}^{16,17}$ This discrepancy may occur due to unusual performance of the procedure in this site at HJAF.

Woods-Hill et al. described a significant risk increase of infection in patients with multiple central devices.[11] This present work, in contrast with the literature, found only $8.2\mathrm{\%}$ of patients with more than one central venous catheter, this fact is probably due to the population studied, in which only one central catheter was sufficient.

Gram-negative bacteria were the predominant etiological agents in this study (48.36%), specially Serratia marcescens, responsible for 10.66% of all PBSIs. This is different from many studies, mainly from developed countries, that have shown gram-positive bacteria prevail.[3,8,15,17,18]

Infections by gram-negative bacteria are closely related to healthcare assistance since these pathogens colonize the gastrointestinal tract and oropharynx. Several Serratia marcescens outbreaks have been described in literature, many of them related to devices and solutions contamination, being the hands of health professionals a major vehicle of transmission.[19]

Jang et al. published a study in which, by performing pulsed-field gel electrophoresis in the hands of health professionals from a neonatal Intensive Care Unit, strains of Serratia marcescens involved in an infectious outbreak were isolated, showing that the absence of basic care, such as hands hygiene, can negatively impact on several sectors of hospital care.[20] In conclusion, it was observed that the profile of children with PBSI admitted to the HJAF PICUs from 2016 to 2020 was of infants with complex surgical conditions, mostly being heart diseases. Most of the patients had a CVC, and the jugular vein was the most prevalent puncture site. Regarding the PBSI microbiological profile, gram-negative bacteria were responsible for the highest number of infections recorded, and the median length of PICU stay until PBSI were 14.76 days.

VI. LIMITATIONS

This present work objective was to study the PBSI prevalence in HJAF PICUs, the infected patients' characteristics and its microbiological profile, being the incidence rate and final outcomes of the patients not analyzed.

Author's contributions

Declaration: The data underlying the research text are contained in the manuscript in the form of tables. However, the database contains patient records and personal data, which is why they will be available on demand with the corresponding author.

Conflict of interests

The authors declare that they have no conflict of interest.

Funding

No funds, grants, or other support was received.

Total number of words: text 1551; abstract 150; tables 4.