nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
Neonatal sepsis of vertical transmission constitutes one of the most significant challenges for neonatologists due to its high morbidity and mortality and difficult diagnosis1.
This pathology affects approximately 9-11 per 1000 very low birth weight newborns (VLBWN), and the mortality can be as high as 8%2,3. Also, the pathogenesis is more complicated in preterm infants4-6. Intraamniotic infection probably starts before labor and might be the cause—not the consequence—of premature rupture of membranes. In addition, other causes of immune-mediated inflammation may promote rupture and trigger labor. Therefore, it is implied that the same infectious risk factors do not have the same weight on both populations7,8.
Regarding diagnosis, symptomatology is not specific and is common to other pathologies of prematurity: infection markers are not sensitive and specific enough, and blood culture, considered the gold standard, has a mediocre diagnostic yield, in addition to a long waiting period9,10. Furthermore, during the first days of life, the ongoing metabolic, hormonal, and immunological processes complicate determining the standard values of these markers.
These reasons, such as the often fulminant presentation of this type of sepsis and the great difficulty in its diagnosis, often lead to initiate probabilistic antibiotic treatment at birth and prolong it despite negative cultures11-13. However, the consequences are not trivial.
The use of antibiotics in the neonatal period leads to a decrease in the biodiversity of the intestinal microbiota, among other things, favoring the appearance of pathogenic organisms and delaying the usual colonization of the gastrointestinal tract14-16. In extremely premature infants, this microbiota modification implies an increased risk of late sepsis, ulcerative-necrotizing enterocolitis, and invasive fungal infection17-19.
Therefore, the challenge is to reduce the use of antibiotics without compromising this vulnerable population. One of the means of improvement to achieve this objective is identifying new biological markers of infection. These include interleukins 6, 8, and 10, nCD64, alarmins, and IP-10, among others20-23.
The use of interleukin 6 (IL-6) as a marker was introduced in our Unit in 2009. IL-6 is a pleiotropic interleukin of the innate immunity produced by different cell types and recognized as the central mediator of the acute phase response24. Also, IL-6 has markedly early kinetics in infections. Importantly, there is no physiological elevation of IL-6 in newborns. Numerous studies support its use as an early marker of sepsis, particularly associated with other markers25-36.
This study aimed to determine the use of antibiotics during the first 72 hours in VLBWN and analyze whether the introduction of IL-6 as a marker of sepsis reduced this use37. As secondary objectives, we wanted to compare the morbidity and mortality of both groups during admission and to study other possible factors associated with initiating antibiotic treatment.
Methods
We conducted a retrospective cohort study. The first cohort corresponded to a period before IL-6 was used as a marker of sepsis in the service (group I or Pre IL-6). The second cohort corresponded to a period after introducing IL-6 (group II or Post IL-6).
We included patients admitted from January 1, 2007, to December 31, 2008 (group I) and from January 1, 2011, to December 31, 2012 (group II). This study was a single-center study in a level IIIb unit (Neonatology Department of the Hospital Universitario Central de Asturias, Spain).
Preterm newborns with birth weight < 1500 g born in our hospital or hospitalized during the first 72 hours of life were selected. We considered the day of admission as the moment of enrollment in the cohort (Figure 1).
The primary outcome variable was the administration of antibiotic therapy during the first 72 hours of life. The secondary outcome variables were those related to morbidity during admission: nosocomial sepsis, necrotizing enterocolitis, invasive fungal infection, days of hospital stay, and mortality.
We also collected general variables and those related to the diagnosis and treatment of early sepsis (C-reactive protein (CRP) and IL-6 levels in the admission blood test, blood culture results, antibiotic treatment, and duration) to characterize both populations. IL-6 levels at birth are expressed in pg/mL. In group I, the cut-off value in the first 72 hours was 150 pg/mL, and in group II, 300 pg/mL38,39.
We also collected data on potentially effect-modifying variables such as risk factors for infection or gestational age. Discharge reports, hospitalization clinical course, and nursing charts were used for data collection through the hospital's information software (Cerner Millenium).
Qualitative variables were expressed as percentages. Normal quantitative variables were reported as the mean with a 95% confidence interval (95% CI). In the case of the Apgar test, the median and interquartile ranges were used. As the groups are sufficiently large, we assumed that the distribution of the means followed a normal distribution according to the central limit theorem.
When considered relevant, percentages were compared using the χ2 test and the odds ratio (OR) with its confidence interval. The Student's t-test was used to compare quantitative variables and the Mann-Whitney's U-test for the Apgar test.
A binary logistic regression was performed (Wald backward stepwise method with PIN 0.05 and POUT 0.10) to determine which variables had a statistically significant effect after the previous analysis. The dependent variable was the antibiotic treatment, and the independent variables were prolonged rupture of membranes (> 18 h), gestational age (< 28 weeks), spontaneous prematurity, and the study group (I or II).
Statistical analysis was performed with SPSS-17 and R packages.
The Regional Ethics Committee (CEIP-Comité de Ética en Investigación de Asturias) approved this protocol.
Results
General characteristics of both groups
We had two groups of 102 and 98 patients (group I and II, respectively). Table 1 shows the general variables of both groups. In group I, the mean gestational age was 28 weeks (95% CI 28.4-29.3), the mean birth weight was 1086 g (95% CI 1034-1139), the sex distribution was almost equal (52% male), and the median Apgar 7 and 8 at 1 and 5 min, respectively. In group II, the mean gestational age was 29 weeks (95% CI 28.3-29.6), mean birth weight 1085 g (95% CI 1030-1140), 53% were male, and median Apgar 7 and 9 at 1 and 5 min, respectively. No significant differences were found between groups regarding these variables.
Table 1 Comparative analysis of initial characteristics of patients in the studied groups
| Characteristics of the patients | Group I (n = 102) Jan 2007-Dec 2008 n (%) | Group II (n = 98) Jan 2011-Dec 2012 n (%) | p-values |
|---|---|---|---|
| Gestational age (weeks)* | 28 (28.4-29.3) | 29 (28.3-29.6) | 0.738 |
| Birth weight (g)* | 1086 (1034-1139) | 1085 (1030-1140) | 0.983 |
| Spontaneous prematurity | 57 (55.9%) | 62 (63.3%) | 0.288 |
| Rupture of membranes (> 18 h) | 7 (6.9%) | 20 (20.4%) | 0.005 |
| Sex | |||
| Male | 53 (52%) | 52 (53.1%) | |
| Female | 49 (48%) | 46 (46.9%) | |
| Multiple birth | 40 (39%) | 38 (39%) | 0.958 |
| Chorioamnionitis | 3 (2.9%) | 8 (8.2%) | 0.105 |
| Type of delivery | |||
| Cesarean section | 77 (75.5%) | 60 (61.2%) | 0.025 |
| Normal | 25 (24.5%) | 34 (34.7%) | |
| Instrumental vaginal delivery | 0 | 4 (4.1%) | |
| Gestation obtained through in vitro fertilization | 8 (7.8%) | 23 (23.5%) | 0.002 |
| Prenatal corticosteroid therapy | |||
| Partial | 26 (25.5%) | 46 (46.9%) | 0.005 |
| Complete | 61 (59.8%) | 40 (40.8%) | |
| Maternal antepartum antibiotic therapy | 16 (16.3%) | 38 (39.6%) | 0.001 |
| Apgar 1 min** | 7 (4) | 7 (4) | 0.548 |
| Apgar 5 min** | 8 (2) | 9 (2) | 0.883 |
| Intubation in the delivery room | 41 (41.8%) | 33 (32.4%) | 0.132 |
| Received surfactant at some point | 55 (53.9%) | 52 (53%) | 0.972 |
| Diagnosis of hyaline membrane disease | 54 (52.9%) | 56 (57.1%) | 0.492 |
| Persistent ductus arteriosus | 20(19.6%) | 33 (33.7%) | 0.030 |
| Pharmacological treatment of persistent ductus arteriosus (ibuprofen) | 14 (13.7%) | 15 (15.3%) | 0.706 |
| CRP (mg/dL)* | 0.55 (0.39-0.72) | 0.10 (0.06-0.15) | 0.001 |
| IL-6 > 150 pg/mL | — | 18 (37.8%) | |
| IL-6 > 300 pg/mL | — | 15 (31%) | |
| Inotropes hypotension | 22 (21.6%) | 22 (22.4%) | 0.382 |
| Nasal CPAP/MV at 28 days of life | 10 (9.8%) | 13 (13.2%) | 0.707 |
CI, confidence interval; CPAP, continuous positive airway pressure; CRP, C-reactive protein; IL-6, interleukin 6; IQR, interquartile range; MV, mechanical ventilation.
*Mean (95%CI);
**Median (IQR).
We found significant differences between the groups (p = 0.025) regarding the birth route: 24.5% of normal deliveries and 75.5% of cesarean sections in group I versus 34.7% of normal deliveries, 60% of cesarean sections, and 4.1% of instrumental delivery in group II.
Concerning infectious risk factors, no statistically significant differences were found in the percentage of spontaneous prematurity or chorioamnionitis. However, significant differences in prolonged rupture of membranes were observed (6.9% in group I vs. 20.4% in group II).
Regarding the antibiotherapy, 76.5% of patients received antibiotics in the first 72 hours of life in group I versus 57% in group II. This difference was statistically significant (p = 0.004). When we analyzed the percentage of antibiotic therapy in patients with sterile blood culture, the difference was 17%, but not statistically significant (75% group I and 57.7% group II; p = 0.052) (Table 2). We were able to collect this variable in all preterm infants except for four patients from group II. Figure 2 shows the percentage of antibiotic therapy and antibiotic therapy in patients with sterile blood culture by study group.
Table 2 Comparative analysis of the primary and secondary outcome variables of both groups
| Primary and secondary outcome variables | Group I Jan 2007-Dec 2008 (n = 102) n (%) | Group II Jan 2011-Dec 2012 (n = 98) n (%) | p- values |
|---|---|---|---|
| Antibiotic therapy in the first 72 h of life | 78 (76.5%) | 56 (57.1%) | 0.004 |
| Antibiotic therapy in the first 72 h of life in patients with sterile blood culture | 69 (75%) | 52 (57.7%) | 0.052 |
| Result of blood culture at birth | |||
| Gram-negative | 4 (3.9%) | 1 (1%) | 0.348 |
| Coagulase-negative Staphylococcus | 2 (1.9%) | 2 (2%) | |
| Another microorganism | 4 (3.9%) | 1 (1%) | |
| Sterile | 92 (90%) | 90 (91.8%) | |
| Invasive fungal infection | 4 (3.9%) | 0 | 0.066b |
| Exitus during admission | 22 (21.6%) | 21 (21.4%) | 0.981 |
| Antibiotic cycles | |||
| Median (IQR) | 1 (1) | 1 (5) | 0.041 |
| Median (95% CI) | 1.5 (1.3-1.7) | 1.25 (1.0-1.4) | 0.086 |
| Blood culture resulta | n = 50 | n = 44 | |
| Gram negative | 18 (36%) | 9 (20%) | 0.119 |
| Coagulase negative Staphylococcus | 15 (30%) | 23 (51%) | |
| Another microorganism | 5 (10%) | 6 (13%) | |
| Sterile | 12 (24%) | 7 (16%) | |
| In-hospital stay days in survivorsc | |||
| Mean (95%CI) | 89.8 (66.7-112.8) | 51.1 (44.7-57.5) | 0.002 |
aBlood culture extracted in the first episode of nosocomial sepsis;
bFisher's exact test;
cExcluding those who died in the first week of life.
CI, confidence interval; IQR, interquartile range.
Figure 2 Percentage of antibiotic therapy by study group and percentage of antibiotic therapy in patients with sterile blood culture by study group. Comparison of the percentage of antibiotic therapy between both groups, p = 0.002. Comparison of the percentage of antibiotic therapy between both groups in patients with sterile blood culture, p = 0.006.
Mean CRP at birth was 0.55 mg/dL (95% CI 0.39-0.72) in group I and 0.10 mg/dL (95% CI 0.06-0.15) in group II (p = 0.001). We were able to collect IL-6 levels in 79 of 98 patients. The mean IL-6 at birth was 501 pg/mL (95% CI -52-1054), and the median was 36 pg/mL. Both markers do not appear to be correlated strongly with CRP at birth (Spearman's correlation coefficient of 0.3) in most patients.
As for the duration of antibiotic treatment, no significant differences were observed in the mean and median between groups (group I, mean 5.9 days (95% CI 5.4-6.5); median of 5.5; group II, mean 5.3 days (95% CI 4.7-5.9); median of 5).
The antibiotics used in both groups were ampicillin and an aminoglycoside in > 95% of the patients. In group I, vancomycin was used in three patients and cefotaxime in one patient. In group II, vancomycin was used in two patients and meropenem in one patient with abdominal surgical pathology (neonatal appendicitis).
Regarding the results of blood cultures collected at birth, 90.2% were sterile in group I and 95.9% in group II (p = 0.348). Of the remaining, 1.9% of group I and 2 % of group II were positive for coagulase-negative Staphylococcus, 3.9% of group I and 1% of group II were positive for a Gram-negative microorganism, and 3.9% of group I and 1% in group II were positive for another microorganism.
In the second study period, we analyzed the percentage of patients who had an IL-6 value above the cut-off level (within the group that had received antibiotics) to find out the importance of IL-6 in the decision to initiate an antibiotic treatment: 37.8% had an IL-6 > 150 pg/mL and 31% had an IL-6 > 300 pg/mL (Table 1). The percentages of patients with IL-6 above and below the cut-off values (150 pg/mL and 300 pg/mL) in the group of patients receiving antibiotic treatment are shown in Figures 3 and 4.
Figure 3 Percentage of patients with IL-6 above and below the cut-off value (150 pg/mL) in the group of patients receiving antibiotic therapy. IL-6, interleukin 6.
Figure 4 Percentage of patients with IL-6 above and below the cut-off value (300 pg/mL) in the group of patients receiving antibiotic therapy. IL-6, interleukin 6.
Table 2 shows the variables related to evolution during admission and mortality. No statistically significant differences were found in the percentage of nosocomial sepsis (first episode), necrotizing enterocolitis, invasive fungal infection, or mortality. However, all were slightly higher in group I. In-hospital stay in those who survived beyond the first week of life was significantly longer in group I patients (mean 89.8 days vs. 51.1 days, p = 0.002). The median number of antibiotic cycles was 1.5 (95% CI 1.3-1.7) in group I and 1.25 (95% CI 1.0-1.4) in group II.
The result of the blood culture obtained during the first episode of nosocomial sepsis was documented: 24% were sterile in group I and 16% in group II (p = 0.304). When analyzing those positive for a Gram-negative microorganism, we found 36% in group I and 20% in group II.
Table 3 shows the bivariate analysis between the patients who received antibiotics (groups I and II) and those who did not. Lower gestational age, lower birth weight, spontaneous prematurity, chorioamnionitis, and belonging to study group I were more frequent in the patients receiving antibiotics, with statistically significant differences.
Table 3 Comparative analysis of variables related to the risk of developing sepsis of vertical transmission between the patients who received antibiotics in the first 72 hours of life and those who did not receive antibiotics
| Variables related to risk | Patients receiving antibiotic therapy in the first 72 h of life (n = 129) n (%) | Patients NOT receiving antibiotic therapy in the first 72 h of life (n = 71) n (%) | p- values |
|---|---|---|---|
| Gestational age (weeks)* | 27.9 (27.5-28.4) | 30.9 (30.4-31.5) | < 0.001 |
| Weight (g)* | 1022 (978-1067) | 1215 (1155-1275) | < 0.001 |
| Spontaneous prematurity | 96 (71.6%) | 23 (34.8%) | < 0.001 |
| Chorioamnionitis | 11 58.2%) | 0% | 0.017 |
| Rupture of membranes > 18 h | 22 (16.4%) | 5 (7.6%) | 0.085 |
| Belonging to study group I | 78 (58.2%) | 24 (36.4%) | 0.004 |
*Mean (95%CI).
CI, confidence interval.
The results of the multivariate analysis are shown in Table 4. Spontaneous prematurity would multiply the risk of receiving antibiotics by 3, extreme prematurity by 16, and belonging to study group I by 4.
Table 4 Binary logistic regression. Dependent variable: initiation antibiotic treatment in the first 72 hours of life in very-low-birth-weight infants
| Risk factors | B | Significance | OR | 95%CI | |
|---|---|---|---|---|---|
| Inferior | Superior | ||||
| Spontaneous prematurity | 1.11 | 0.002 | 3.28 | 1.52 | 7.04 |
| Study group | 1.44 | < 0.001 | 4.22 | 1.96 | 9.08 |
| Gestational age < 28 weeks | 2.80 | < 0.001 | 16.59 | 4.65 | 59.23 |
| Constant | 1.28 | < 0.001 | 0.27 | ||
CI, confidence interval; OR, odds ratio.
Discussion
The use of antibiotics during the first 72 hours of life in VLBWN reached 76.5% of the patients in our Unit, similar to those found in the literature. After implementing IL-6 testing as a biological marker of sepsis in 2009, we observed a 20% reduction in antibiotic therapy. This reduction was almost maintained in the group of patients with negative blood cultures. We have specified this last point because it seems that the negative consequences of this antibiotic treatment would be more pronounced in preterm infants with sterile blood cultures40. These data confirm our working hypothesis; however, due to the type of study performed, we cannot conclude that this decrease is exclusively due to the use of IL-6. To analyze the question in more detail, we have examined some of the factors related to initiating antibiotic treatment. As already discussed, this decision is based on three pillars that are not easy to interpret in preterm infants: clinical features, clinical risk factors, and biological markers of infection.
Regarding the general data that can determine the status of the newborn after delivery (gestational age, weight, Apgar, and others), both groups had similar characteristics.
Although CRP levels at birth differed significantly, no clinical differences were observed. There is no single cut-off value accepted worldwide, but the figures in both groups are clearly below levels suggestive of sepsis. As procalcitonin is only used to diagnose nosocomial sepsis in our department, it was not included in the analysis.
Regarding clinical risk factors, neonatologists do not usually distinguish between term and preterm newborns, although the pathophysiology in both conditions is not the same. Our data support this fact, as prolonged rupture of membranes, one of the most important infectious risk factors in the term newborn, predominates significantly in the second study period without a higher percentage of confirmed sepsis. In these cases, other non-infectious causes of stress might be leading to the initiation of labor and rupture of membranes41,42.
Following the reasoning of the document published by the American Academy of Pediatrics in 2018, preterm infants with the lowest risk of presenting an early bacterial neonatal infection are those born by cesarean section and those born due to induced prematurity6. However, we found a higher percentage of spontaneous prematurity and vaginal deliveries in group II.
We can cite two reasons that might play a role: the change in the Director of Obstetrics in 2010, who clearly wanted to reduce the percentage of cesarean sections, and the progressive change in obstetrics practices—better control of pregnant women and a limitation of the indications for cesarean section in the case of premature delivery43,44. After analyzing these items in our sample, no other variable was found that clearly explained the decrease in the use of antibiotic treatment in the second period, except for IL-6.
We also performed a binary logistic regression to determine which variables were significantly related. Our results were consistent with other studies6: gestational age < 28 weeks would multiply the risk of receiving antibiotics by 16, spontaneous prematurity (versus induced prematurity) by 3, and belonging to group I of the study by 4. It should be noted that the percentage of extreme prematurity in both samples was similar (35.3% in group I and 37.7% in group II).
Regarding the evolution during admission, we found no statistically significant differences in the percentage of nosocomial sepsis, invasive fungal infection, necrotizing enterocolitis, or mortality. However, the mean in-hospital stay was significantly longer in group I. The complexity of the patients does not seem to justify this fact a priori since it was similar in both groups. Conversely, no significant changes were identified between the two periods in the management protocols of these patients in our Unit, especially in the management of neonatal sepsis of vertical transmission or those aspects related to the incidence of nosocomial sepsis, such as catheter insertion, its duration, and the type of feeding, among others.
After analyzing the clinical and paraclinical variables associated with the decision to initiate antibiotic treatment, it seems that the use of IL-6 as an early marker of sepsis in very low birth weight preterm newborns significantly reduces the use of antibiotics in the first 72 hours of life. It is more difficult to conclude subsequent evolution, but it seems that morbidity, length of hospital stay, and mortality are lower in the second period.
