Invasive devices as risk factors for neonatal sepsis in neonatal intensive care units

This study aimed to evaluate the impact of invasive devices as risk factors for the development of neonatal sepsis in Neonatal Intensive Care Units. Hospital-based retrospective cohort study performed in two Neonatal Intensive Care Units in Ponta Grossa, Paraná, Brazil. Documentary data were collected through consultation of electronic medical Research, Society and Development, v. 9, n. 12, e27891211048, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i12.11048 2 charts of all patients admitted to two hospitals and of the patients with diagnosis of sepsis in another hospital. Health conditions at admission and outcomes were evaluated. Frequencies of the reasons for admission and the outcomes were calculated. In the association analysis, exposure variables were calculated with odds ratio and confidence intervals (95%). The frequency of sepsis was 39%, and 45.7% of the cases were of early-onset sepsis and 54.3% of late-onset sepsis. The mortality rate associated with sepsis was 9.9%. The use of invasive devices was observed to increase by 6 times the risk of neonatal sepsis. Peripherally inserted central catheter and phlebotomy were the devices causing higher risk. The high incidence of late-onset sepsis, its association with the use of invasive devices and the higher mortality rate among newborns with sepsis suggest the presence of fragilities in neonatal care and the need to seek alternatives of neonatal approach to avoid new cases of neonatal sepsis and consequent death.

charts of all patients admitted to two hospitals and of the patients with diagnosis of sepsis in another hospital. Health conditions at admission and outcomes were evaluated. Frequencies of the reasons for admission and the outcomes were calculated. In the association analysis, exposure variables were calculated with odds ratio and confidence intervals (95%). The frequency of sepsis was 39%, and 45.7% of the cases were of early-onset sepsis and 54.3% of late-onset sepsis. The mortality rate associated with sepsis was 9.9%. The use of invasive devices was observed to increase by 6 times the risk of neonatal sepsis. Peripherally inserted central catheter and phlebotomy were the devices causing higher risk. The high incidence of late-onset sepsis, its association with the use of invasive devices and the higher mortality rate among newborns with sepsis suggest the presence of fragilities in neonatal care and the need to seek alternatives of neonatal approach to avoid new cases of neonatal sepsis and consequent death.

Introduction
About 36% of the 4 million neonatal deaths estimated annually are caused by infections (Shane & Stoll, 2014). One of the major concerns in Neonatal Intensive Care Units (NICU) is the development of neonatal infections and sepsis, the latter being the main cause of mortality and morbidity in these units. Characterized by signs and symptoms of infection and bacteremia in the first 28 days of life (Silveira & Procianoy, 2012), sepsis has a direct impact on neonatal mortality and can implicate neurodevelopmental disorders, especially in cases of extremely low birth weight (Shindler et al., 2017). Incidence and mortality rates are higher in developing countries such as Brazil.
The neonatal period is characterized by vulnerability due to the immaturity of the immunological system and the first major exposures to microorganisms, starting from the Research, Society andDevelopment, v. 9, n. 12, e27891211048, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i12.11048 4 moment the newborn leaves the intra uterine cavity and gets in contact with maternal blood, the birth canal, people and eventual extreme situations such as admission to intensive care units and use of invasive devices (Rosa et al., 2020;Shane & Stoll, 2014). The incidence of neonatal sepsis varies according gestational age and the risk is higher in preterm infants, in whose cases can be around 60% (Wynn, 2016).
Much progress has taken place in neonatal care in the last 30 years, such as the development of successful ventilation strategies for the management of pulmonary hypertension and hyaline membrane disease, therapeutic hypothermia for hypoxic-ischemic encephalopathy, and improvement in parenteral nutrition care. However, the treatment of neonatal sepsis and outcomes of neurodevelopmental disorders remain the same despite all efforts to decrease the burden of infection (Wynn et al., 2014).

Early-onset sepsis arises in the first 72 hours of life, is mostly caused by Group B
Streptococci or Escherichia coli, and accounts for about 8% of neonatal deaths. On the other hand, late-onset sepsis starts after 72 hours of life, is caused by healthcare-associated pathogens, and has a four-fold greater impact on neonatal mortality (Global Maternal and Neonatal Sepsis Initiative Working Group, 2017).
Late-onset neonatal sepsis appears from the fourth day of life onwards and although it can be associated with childbirth-related factors, in general, postnatal factors have a greater influence (Silveira & Procianoy, 2012). Cross-contamination via the hands of health care professionals plays an important role in increasing the risk of infection, which reinforces the importance of proper hand hygiene Sánchez;Stoll, 2017). Premature infants are at greater risk of developing late-onset sepsis in view of their immunological immaturity and the multiple procedures they are subjected to during their stay in the NICU and prolonged length of hospitalization (Hornik et al., 2012). Prolonged parenteral nutrition, intravascular catheters, orotracheal intubation and invasive mechanical ventilation are predisposing factors to the onset of infection (Alcock et al., 2017).
In view of the national and worldwide relevance of neonatal sepsis, this study aimed to evaluate the impact of invasive devices as risk factors to the development of neonatal sepsis at two NICU in Ponta Grossa, Paraná, Brazil. Both are tertiary care hospitals, the first being a reference to high risk pregnancy, the second did not have an obstetric ward at the time of the study, and newborns came from other Hospitals in the region. Research, Society and Development, v. 9, n. 12, e27891211048, 2020 (CC BY 4. Regarding the years encompassed in this study, a limited period of time was adopted for data collection because it represented the short period of study in one of the hospitals. Electronic charts of patients were consulted for a retrospective analysis of gestational age, birth weight, maternal pathologies, maternal age, maternal use of antibiotics and steroids, use and time of use of invasive devices, prenatal care, method of delivery, late rupture of membranes,-and chorioamnionitis.
This study was performed in South Brazil, state of Paraná, which is the fifth-largest economy of the country. The South region has the lowest maternal mortality and child mortality rates in the country.
In this study, three criteria were considered for the diagnosis of sepsis: 1 -Abnormal laboratorial analysis (Rodwell hematologic score ≥ 3 and/or elevated C-reactive protein); 2 -Presence of any non-specific sign/symptom of infection (respiratory distress, apnea, lethargy or irritability, thermic instability, hypotonia, persistent gastric stasis, pulmonary hemorrhage); 3 -Prescription of antimicrobial treatment by the medical team. All newborns who presented the three diagnostic criteria in the first 28 days of life were classified as presenting neonatal sepsis. Automatized blood cultures were not available at the time; thus, less than 5% of the blood cultures tested positive and they were not enough to diagnose neonatal sepsis. Aiming to increase specificity to the diagnosis, we adopted the three abovementioned criteria, i.e. laboratorial, clinical, and medication parameters.
Crude frequencies and 95% confidence intervals (95% CI) of early-and late-onset neonatal sepsis in all individuals, and in individuals from each hospital, were estimated.
Crude frequencies estimate the real absolute number of the events and the rates at which they happened in the patients.
The frequency of presence and absence of sepsis was calculated for each one of the risk factors. The Chi-square test was used to check differences in proportions for all the risk factors. In order to evaluate the association between neonatal sepsis and variables of a binary outcome, the odds ratios (95% CI) were calculated. The Chi-square statistic is tool designed to analyze group differences between categorical variables and reveals if the observed distribution occurred at random, when compared to a theoretical distribution.

Means, medians, standard deviations and interquartile ranges of numerical variables
were calculated to investigate the differences between presence versus absence of sepsis, and early-versus late-onset neonatal sepsis according exposure variables (birth weight, gestational age, maternal age, time of membranes rupture, and number of prenatal medical consultations).
The association between numerical independent variables and neonatal sepsis were tested with the Student's t-test or Mann-Whitney test, depending on the compliance with normal distribution. The Student's t-test is a tool for quantitative variables that compares the sample mean with the population, and the Mann-Whitney test is a non-parametric version of the Student's t-test when the variables do not have a normal distribution.
All variables with p-value <= 0.20 in the bivariate analysis were included in a logistic regression model. The logistic regression is a statistical multivariate method that allows the construction of a predictor model to explain the outcome based on a series of binary explanatory variables. Data were analyzed with SPSS®, 20 th version. All the notes of this study were extracted by the main researcher in both hospitals through consultation of the medical charts of the patients. Research, Society and Development, v. 9, n. 12, e27891211048, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i12.11048

Results
The number of newborns analyzed in the study was 520. These children were hospitalized in both institutions in the period of study, and 226 had neonatal sepsis (Table 1).
Hospital 2 had a higher number of cases of late-onset sepsis (Table 2).  In the crude analysis, risk factors for the development of neonatal sepsis were vaginal delivery, birth weight, admission of the mother to the ICU, use of invasive devices such as endotracheal intubation, bi-level positive airway pressure (BiPAP), peripherally inserted central catheter (PICC), phlebotomy, and total parenteral nutrition (TPN) ( Table 3).

Discussions
Neonatal sepsis causes 3 million deaths per year. Early detection and treatment of maternal conditions during prenatal care may avoid 1/3 of these deaths (Black et al., 2016).
Almost half of the patients analyzed in this study had sepsis. Late-onset sepsis was more frequent. The higher incidence of late-onset sepsis worldwide point to the impact of the fragilities in healthcare, because this condition is mostly caused by healthcare-associated pathogens and is particularly related with NICU care (Giannoni et al., 2018).
Low, very low and extremely low birth weight preterms are more prone to develop sepsis due to their immunological immaturity, constant handling by healthcare professionals, prolonged hospital stay, and use of invasive devices. At this study, VLBW and LBW had statistically significant higher risk of neonatal sepsis.
Different approaches can be successful to prevent this severe condition. A first measure is to implement interventions during prenatal and delivery care for adequate detection and treatment of the mother's infections (Silveira & Procianoy, 2012).
Regarding the method of delivery, vaginal delivery showed a higher risk of neonatal sepsis. However, the maternal information available in the charts was not complete and as neonatal sepsis has a multifactorial etiology, the association of sepsis with method of delivery has not enough support.
Since 2011, a healthcare strategy called the "Rede Cegonha" was introduced in Brazil aiming to provide assistance to family planning of reproductive health and humanized care during gestation, labor and puerperium. In the state of Paraná, this program is called "Mãe Providing proper assistance in rural areas and extremely poor regions can be very challenging, and better governmental strategies are needed for a comprehensive national coverage (Frank et al., 2016).
Regarding late-onset sepsis, admission to NICU is necessary to avoid this condition.
This study shows evidence of flaws in local hospital care: individuals using invasive devices, PICC and with phlebotomy had 6-fold, 8-fold and 11-fold higher risk for sepsis. The impact of invasive devices on the increase of risk of sepsis and, consequent mortality is noteworthy.
Septic patients were 3-foldmore prone to die.
Bundle implementation to control the use of invasive devices is likely to be effective to reduce morbidity and mortality. Gocke et al. (2018)  The limitations of this study include the retrospective cohort approach and the analysis of medical charts which can provide incomplete information. On the other hand, these methods chosen made the research easier to follow up the individuals. Chart analysis was also considered to reduce the bias of non-response and loss of follow up. Moreover, information bias was reduced because information loss was equally distributed among patients with and without neonatal sepsis. At the end, to minimize analysis bias, data were examined by a statistical expert blinded for the neonatal sepsis risk factors already described by the literature.
Also, as this was a hospital-based study, the newborns had more severe infectious conditions.
Further studies in NICU should be performed, such as clinical trials, and should compare intervention strategies aiming at reducing exposure to the risk factors found in the present study. Finally, we stress that researchers of neonatal sepsis, especially in developing countries, should investigate the role of educational practices for the healthcare team and families so as to reduce neonatal sepsis.

Conclusion
The high incidence of late-onset sepsis, its association with the use of invasive devices and the higher mortality rate among newborns with sepsis suggest the presence of fragilities in the neonatal care and the need to seek alternatives of neonatal approach to avoid new cases of neonatal sepsis and consequent death. Other studies in NICU should be performed to compare intervention strategies aiming to reduce neonatal sepsis.