Introduction

Preterm birth is a phenomenon that has reached social dimensions due to growing survival rates of very low birthweight or <1500 grams, and extremely preterm infants (<1000 grams). For instance, in 2010 it was estimated that 15 million of infants were born before 37 gestational weeks worldwide; of whom 13 million survived. From these, 2.7% had moderate or severe neurodevelopmental impairment [1].

Nowadays, survival rate of very low birthweight infants is higher or equal to 90% and between five to 10% of them present motor impairments. Other disabilities occur in between 25 and 50% of cases and involve cognitive and behavior disorders [2]. Cerebral palsy incidence is even higher in extreme preterm infants, with 17 or 19%; according to whether they were less than 26 gestational weeks old or extremely low birthweight [3],[4].

Some of the most important risk factors for the occurrence of neurodevelopmental disorders in very low birthweight infants, and especially in extremely low birth weight infants include male sex, no maternal steroid administration, presence of periventricular leukomalacia, neonatal convulsions, retinopathy of prematurity, bronchopulmonary dysplasia, prolonged mechanical ventilation, persistent ductus arteriosus, postnatal steroids use, malnutrition, extended parenteral nutrition, necrotizing enterocolitis and/or intestinal perforation [5],[6],[7]. 

Diverse studies have found association between presence of maternal infection, specially clinical or histological chorioamnionitis, and presence of white matter damage or cerebral palsy, in both preterm and at term children [8],[9].  

Adverse neurological outcome in preterm infants not only has been associated to the presence of intrauterine maternal infections or chorioamnionitis, but also neonatal sepsis per se. The first two studies which have found a strong association between presence of neonatal infection and worse neurological outcome in extremely preterm infant populations, were published only few years ago [10],[11].    

Bacterial infections continue being an important cause of morbidity and mortality in the neonatal period. About 1.5% of all very low birthweight infants present early-onset sepsis and until about 21% present late-onset sepsis, with a mortality rate between 10 and 30%, even when incidences varies from regions and years [12].

Most studies that have evaluated neonatal sepsis association with neurodevelopmental sequels have been carried out in developed countries [12],[13]. None of the studies published in Cuba about the relationship between factors that take place in the neonatal period and neurodevelopment in very low birthweight infants [14],[15],[16], has been focused in the role of neonatal sepsis as a risk factor of the occurrence of neurodevelopmental disorders, despite the high neonatal sepsis incidence in this neonatal group, as reported by some studies performed in this country [17],[18].

As expressed here, it is inferred that the impact of neonatal sepsis on neurodevelopment in very low birth weight infants discharged from Cuban neonatal intensive care units is unknown, being Cuba a developing country that has achieved low infant mortality rates. 

The Teaching Provincial Gynecological and Obstetrical Hospital of Matanzas Dr. Julio Alfonso Medina is located in the capital city of the same name province, and constitutes a reference center for management of infants born under 1,500 grams. A neurodevelopment follow-up program is carried out in the outpatient clinic of this center, where high neurological risk infants are evaluated during the first years of life with a method described in other works [19].

Beginning with the hypothesis that neonatal sepsis seems to be a risk factor associated with poor neurodevelopmental outcome in very low birth weight infants, we made this research work, with the main aim of determining the impact of neonatal sepsis as a risk factor of neurodevelopmental disorders in very low birthweight infants discharged alive from a Cuban tertiary hospital.

Methods

A cohort study including 89 infants with birth weight <1500 g was carried out. Infants were born during the period 2006-2010 in the Province of Matanzas and were discharged alive from the Teaching Provincial Gynecological and Obstetrical Hospital of Matanzas Dr. Julio Alfonso Medina. All patients were born with less than 37 weeks of gestational age (according to last menstrual period, ultrasound fetal measurements at the beginning of pregnancy or modified Ballard score in absence of the former); and were followed-up in a neurodevelopment program at the outpatient clinic of this center until 24 months of corrected gestational age (months age beginning at term of gestation, for 40 weeks gestation estimation).

Major malformations, hereditary syndromes, neonatal and post neonatal deaths were excluded from the study.

The study sample was constituted by 89 very low birthweight infants out from a total of 134 newborns with birth weight <1500 grams, born in the Province of Matanzas during the period 2006-2010. Ninety of them were born in the Teaching Provincial Gynecological and Obstetrical Hospital of Matanzas Dr. Julio Alfonso Medina. The rest were referred to this center during the neonatal period from other hospitals of the province. Out of 134 newborns with birth weight <1500 grams, 26 resulted in neonatal deaths. Thus, 108 were discharged alive. Four of them were excluded from the study due to major malformations, hereditary syndromes and post neonatal deaths; and 104 entered the follow-up program. Eighty nine of them completed the follow-up until two years of corrected gestational age, so they constituted the study sample.

Study groups
With the 89 patients who met the study criteria, two groups were conformed: one group of neonatal sepsis (exposed), composed by those who had been diagnosed with clinical sepsis during neonatal period (n=19); and a second group without sepsis (not exposed), composed by patients who did not have sepsis during neonatal period (n=70).

For neonatal sepsis diagnosis Haque KN criteria were taken into account [20], which include clinical, hemodynamic, perfusion tissue and inflammatory variables, with the exception that the last ones included neither C-reactive protein values, procalcitonin nor interleukines 6 or 8, because they are not routinely used in the center where the research wotk took place. 

Sepsis was considered proven when any germ species was isolated from blood cultures, but when these cultures were not carried out or were negative; it was considered not proven sepsis. In this work were included both, proven and not proven neonatal sepsis.

Besides variables involved on Haque KN criteria (leukocytosis>34,000x10⁹/l, leukopenia<5,000x10⁹/l, immature neutrophils >10%, ratio of immature over total neutrophils ≥0.2 and thrombocytopenia <100,000x10⁹/l), there were other contained on Rodwell et al criteria [21] that were taken into account for including patients in the neonatal sepsis group. Those were: presence of leukocytosis (≥25,000x10⁹/l at birth or ≥30,000x10⁹/l between 12-24 hours of postnatal period, or ≥21,000x10⁹/l beyond 48 hours of postnatal period), neutrophilia or neutropenia, increased neutrophilic index, ratio of immature over segmented neutrophils ≥ 0.3 and presence of neutrophils with toxic granulation and vacuolization.

The neonatal sepsis group involved not only patients without localized neonatal sepsis, but also those who presented a localized infection plus two or more signs of systemic inflammatory response, established by international consent as criteria for neonatal systemic inflammatory response syndrome. These signs were: presence of heart rate abnormalities (>180 or <100 beats per minute), respiratory rate abnormalities (>50 breaths per minute in neonates during the first week of life or <40 per minute between eight and 30 days of life), temperature disturbances (<36 o >38.5 °C), total leukocyte count abnormalities (>34 000 in the first week of life or >19,500 o <5,000 between eight and 30 days of life), and presence of band cells (>10% of total leukocyte count) [22].

Research stages:

This research had three stages:

• 1st stage: A cohort study was performed, with the objective of determining association between neonatal sepsis and neurodevelopmental disorders.
• 2nd stage: By means of a logistic regression analysis, it was determined if neonatal sepsis was associated to the occurrence of neurodevelopmental disorders when considering other variables which potentially could constitute risk factors for neurodevelopmental disorders. These variables were previously chosen from a bivariate risk analysis.    
• 3rd stage: The association between time of onset of sepsis and presentation of neurodevelopmental disorders was analysed, as well as a “dose-response” association between exposition levels to neonatal sepsis and the occurrence of neurodevelopmental disorders. 

Studied variables
Any deviation from normal neurodevelopment during the first 24 months of corrected gestational age, demonstrated by neurological exam, Bayley scale II [23] or neurophysiologic tests was considered a neurodevelopmental disorder.

In the neurodevelopmental outcome classification during the first two years of corrected age, other categories were used. They were:

• Major disorders. Presence of one or more of the following abnormalities: moderate-severe cerebral palsy, psychomotor development index or mental development index <70, bilateral deafness or blindness, epilepsy, hydrocephaly or microcephaly.
• Minor disorders. Presence of one or more of the following abnormalities: mild or transient muscular hypotonia, mild motor insufficiency with hypertonia, mild or transient reflex abnormalities, transient or mild psychomotor delay (psychomotor development index 70-84), mild mental delay (mental development index 70-84), mild language delay, hyperactivity, mild or unilateral neurosensory abnormality.
• Normal. When none of the disorders mentioned above were present.

Cerebral palsy was considered by the presence of a permanent or chronic motor function disorder, originated on brain or cerebellum, with early onset, attributed to non-progressive disturbances in a child with a minimum of two years of age at the time of evaluation [24]. Severe and bilateral visual disability was defined as the impossibility for useful vision on both eyes. Mild or unilateral blindness was considered as a visual acuity diminished in one or both eyes, not making impossible the total visual function. Deafness was considered bilateral and severe when audition by both ears was impossible and patients required amplification in order to hear; and unilateral and mild, when the hearing impossibility affected only one ear, or diminished hearing capacity did not require any amplification equipment to reach the audition function.

Independent variables included in bivariate analysis were: male sex (versus female), birth weight (grams), gestational age (according to last menstrual period, or by default, foetal ultrasound at the beginning of pregnant or neonatal exam by modified Ballard score), low birthweight for gestational age (birth for gestational age less than the 10th percentile according to Dueñas curves et al. [25]), low Apgar score at first and fifth minute (score 0-6), mode of delivery (vaginal vs. caesarean section), respiratory distress syndrome, bronchopulmonary dysplasia, (present when, at least at 28 days of postnatal age, patient requires oxygen therapy) [26], not physiologic hyperbilirubinemia (presence of total serum bilirubin levels ≥15 mg/dl or 256.5 μmol/l), antenatal steroids use (one or more doses) and mechanical ventilation (conventional or high frequency ventilation).

Neonatal sepsis were classified, according to the onset of symptoms and signs, in early (onset during the first 72 hours of life), late (onset beyond 72 hours after birth), and early sepsis with late re-infection (when sepsis appear during the first 72 hours of life and after that another sepsis emerges). 

On the final analysis, corresponding to exposition level to sepsis and its association with presentation of neurodevelopmental disorders, the variable “neonatal sepsis episodes” was classified in three categories: without sepsis, one sepsis episode and more than one sepsis episode (independently of time of its onset).    

Follow-up
All infants underwent a neurological exam at term (when they reached 38 to 42 weeks, corrected gestational age), by means of the methods of Amiel Tison et al., Prechtl and Fenichel [14],[19]. At that moment they were classified in one of the following categories:

• Normal: neurological exam without abnormalities.
• Suspect: presence of axial hypotonia, abnormal primary and deep reflexes, as well as adaptive capacity.
• Pathologic: presence of global hypotonia (axis and extremities) or axis and/or extremities hypertonia, integrated and deep reflexes diminished or increased, and poor adaptive capacity. 

All patients were examined during the first year of corrected age at the first, second, third, sixth, ninth and twelfth month, and during each trimester of the second year. Each one underwent neurologically assessment by Amiel-Tison method [14],[19] during the first year. Classic neurological exam was used during the second year.

Psychomotor and mental assessment was carried out by means of Bayley scale II [23], which was applied at the end of the first and the second year of corrected age. Clinical follow-up was completed using brain ultrasound series at first month, end of the third and sixth month of corrected age; as well as brainstem auditory evoked potentials.

Data collection
Data about perinatal evolution of patients were obtained from maternal and neonatal charts. That information and the one obtained during follow-up were summarized in a data collection form.

Statistical analysis
All the above information was entered in a database using statistic package SPSS version 17.0, with which all the statistical processing was performed. To achieve the objective of the first stage of this research, relative risk (RR) and its 95% confidence interval was the main measure. In a similar way, the association between neonatal sepsis and specific neurodevelopmental disorders was evaluated.

In the second stage of the study, to select candidate variables to be included further in the logistic regression model, media and standard deviation were used as summary measures of continuous quantitative variables. Absolute and relative frequencies were used for qualitative variables. In the bivariate analysis, Chi square or Fisher exact tests were applied for nominal qualitative variables, the latter when expected frequencies were less than five or if they were equal to cero. With quantitative variables, t-test was applied for comparison of media between two populations. For all the cases, p value <0.05 was considered significant.

Afterward, a logistic regression analysis was performed. In this analysis, odds ratios (OR) and their 95% confidence intervals were calculated by means of the omnibus test. All p values <0.05 were considered significant. Previous to logistic regression analysis, a collinearity analysis among independent covariates candidates to be included in the former was performed, with the Pearson correlation coefficient. Covariates included in the logistic regression analysis were chosen under the principle that there was not a high correlation between them. Thus, when a Pearson correlation coefficient of any pair of variables was over 0.5, only the most important of both was selected, according to literature review and objectives of the study. Hence, covariate mechanical ventilation was selected and respiratory distress syndrome and use of surfactant were excluded, due to the high correlation between the last two and the former.          

So, in the logistic regression model were included four covariates resulting from the above explained selection: neonatal sepsis, mechanical ventilation, not physiologic hyperbilirubinemia and bronchopulmonary dysplasia; and also male sex. Presence of a neurodevelopmental disorder was included in the model as the dependent variable.

To determine association between the moment of sepsis onset and the presentation of neurodevelopmental disorders, the categories early neonatal sepsis, late neonatal sepsis and early sepsis with late re-infection were managed as dummy variables.

Finally, to determine a possible dose-response relation between neonatal sepsis episodes and occurrence of neurodevelopmental disorders, RR and 95% confidence intervals were calculated for each category.

Ethical aspects
Patients’ enrollment on the follow-up study was performed under informed parental consent. Parents were free to abandon the study whenever they wanted. The research work was approved by institutional board and ethics committee on research of the center.

Results

Table 1 presents general characteristics of 89 very low birthweight infants included in this study, in which female sex was predominant. The mean value for birthweight was 1,278 grams (minimum 700 and maximum 1,490 grams) and for gestational age was 31.4 weeks (minimum 26 and maximum 36 weeks). Neonates with <1000 grams were nearly 8%. Most infants had a normal Apgar score and about 40% received some mode of mechanical ventilation. Neonatal sepsis was present in 21.3% of studied infants.

Table 1: General characteristics of very low birthweight infants included in the study (N=89).

When analyzing neonatal sepsis association with neurodevelopmental disorders occurrence on the first two years of corrected age, it was observed that risk of neurodevelopmental disorders was greater in very low birthweight infants presenting neonatal sepsis than in those who did not present sepsis (47.4 versus 17.1%; RR 2.7 95% CI: 1.3-5.5; p=0.005).

Results about neurodevelopment of studied infants, according to having or not sepsis, are shown in detail in Table 2. According to this analysis, neonatal sepsis was associated to a higher risk of major neurodevelopmental disorders, even when this association was not significant. Meanwhile, neonatal sepsis association with increased risk of minor neurodevelopmental disorders resulted significant.

Table 2 also shows the frequency and risk of occurrence of specific neurodevelopmental disorders in studied patients. Patients with neonatal sepsis presented an increased risk of cerebral palsy, moderate to severe motor developmental delay (PDI<70) and moderate to severe mental developmental delay (MDI<70), but these associations were not statistically significant. About minor disorders, the same table highlights neonatal sepsis was significantly associated to a higher risk of mild or transitory hypotonia and mild or transitory motor developmental delay, while it was not significantly associated to mild or unilateral visual disability.

Table 2: Developmental disorders during the first two years of corrected age found in the studied infants (N=89).

Table 3 shows the main results of bivariate risks analysis, with which the association between some variables and neonatal sepsis was ascertained. From analyzed variables, only neonatal respiratory distress syndrome, bronchopulmonary dysplasia, not physiologic hyperbilirubinemia, exogenous surfactant use and mechanical ventilation had a significant association with neonatal sepsis.    

Table 3: Association between studied variables with neonatal sepsis, according to bivariate analysis (N=89). 

Results of logistic regression analysis, performed in order to determine independent association between variables included in the model and the presence of neurodevelopmental disorders, are shown in Table 4. This analysis showed that neonatal sepsis was the only significant variable associated to a four-fold increase in the risk of neurodevelopmental disorders. Mechanical ventilation in neonatal period and bronchopulmonary dysplasia were associated to an increased risk of neurodevelopmental disorders, the same as male sex, but these associations were not significant.

Table 4: Association of neonatal sepsis with the presence of neurodevelopment disorders, according to logistic regression analysis (N=89; cases=21, controls=68).

When the above risk analysis was carried out, but now including the type of neonatal sepsis instead of the covariate neonatal sepsis, result was early neonatal sepsis with late re-infection was associated to significant increase on the risk of neurodevelopmental disorders (odds ratio 4.5 95% CI: 1.02-20.3), while late neonatal sepsis was associated to an increased risk, but not significant (odds ratio 7.5, 95% CI: 0.82-69.5). Early neonatal sepsis, when occurred alone, was not associated to neurodevelopmental disorders.

Finally, the association between exposure number to neonatal sepsis episodes and neurodevelopmental disorders occurrence was analyzed. As a result, it was found that the higher the number of exposures to sepsis episodes the greater the risk of neurodevelopmental disorders, as shown in Figure 1.

Figure 1: Risk of neurodevelopmental disorders according to neonatal sepsis episode number. Relative risk and confidence interval.