Surveillance during a 1-year period.

    The male/female ratio was 1.4; males represented the 58.3% (35/60) and females 41.7% (25/60). The majority of patients were from northern Lima at 42% (25/60), mainly from the districts of San Martín de Porres and Los Olivos, followed by districts of eastern Lima at 15% (9/60). Figure 2 shows that the cases came from areas surrounding the hospitals.

HCH: Hospital Cayetano Heredia
HSB: Hospital San Bartolomé
HNAL: Hospital Nacional Arzobispo Loayza
HDAC: Hospital Daniel Alcides Carrión
INMP: Instituto Nacional Materno Perinatal

Figure 2. Distribution of neonatal meningitis cases by district of origin (Metropolitan Lima)

    From the total, 34% (18/53) were early NM cases and 66% (35/53) were late NM cases. Cases of confirmed NM were 58.5% (31/53), of which 25.8% (8/31) were early and 74.2% (23/31) late. Bacterial NM occurred in 87.1% (27/31) of the confirmed cases and viral NM in 12.9% (4/31). Probable NM was 22.6% (12/53) of the total of cases and possible NM was 18.9% (10/53). Regarding outpatients, 4 had confirmed NM, 2 had probable NM and 1 had possible NM.

For early NM, the associated prenatal factors were meconial amniotic fluid (38.9%), urinary tract infection (33.3%), maternal fever (27.8) and chorioamnionitis (22.2%). However, in the late NM, these factors did not seem to have a major influence (Table 2).

Table 2. Prenatal, natal and postnatal characteristics, according to the type of meningitis

 NM: neonatal meningitis

^a Mean (SD)

    Sepsis was the most important factor related to NM; and according to the NM types, 50% (9/18) were early meningitis cases, 20% (7/35) were late meningitis cases and 42.9% (3/7) were of out-of-hospital meningitis cases. The average age for the onset of symptoms in early NM cases was 0.9 days; 18.6 days in late NM; and 11.9 days in the out-of-hospital cases. Symptoms such as respiratory distress, were more common in the early NM. In late NM, fever, irritability and hypoactivity predominated (Table 2). Table 5 shows the risk factors associated with early NM in relation to late NM.

Table 5. Factors associated with early neonatal meningitis compared to late neonatal meningitis.

OR: Odds Ratio, 95% CI: 95% confidence interval

Cytochemical and bacteriological characteristics of the CSF

On average, 2.5 and 2 LPs were performed for early and late NM, respectively. In most cases of early NM, the LP was performed on the first day of hospitalization; and in cases of late NM, it could take until the third day of illness.

    In Table 3, the cytochemical characteristics of the CSF are presented. The median value for pleocytosis was 225 leukocytes/μL for early NM and 202 leukocytes/μL for late NM, the median value for polymorphonuclears (PMN) was of 57% and 30%, respectively. Hypoglycorrhachia was similar in both types of meningitis and hyperproteinorrhachia was higher in early NM. In the out-of-hospital type, there was less pleocytosis and more glycorrhachia ( Table 3).

Table 3. Cerebrospinal fluid characteristics, according to the type of meningitis.

NM: neonatal meningitis, IQR: interquartile range (25th and 75th percentiles), PMN: polymorphonuclears

    A total of 35 germs were identified, including bacteria, viruses and a case of Candida albicans. In all clinical types, Escherichia coli and Listeria  monocytogenes predominated. In 17.1% (6/35) of the cases the germ was isolated both in blood culture and in the CSF. Escherichia coli was found in four of those cases, group B Streptococcus and coagulase negative Staphylococcus, in one case each. Cases of influenza B, coronavirus and adenovirus were identified by indirect immunofluorescence (IIF). In a single case, PCR was performed, isolating herpes virus VI (Table 4).

Table 4. Isolation of the infectious agent according to the type of meningitis.

NM: neonatal meningitis; CSF: cerebrospinal fluid; PCR: polymerase chain reaction; IIF: indirect immunofluorescence

^a Isolation of the germ in both blood and CSF

The total represents the burden of disease addressed by all the hospitals.

Treatment schedules were highly variable. The mean duration for cases of early NM was 21 days and 19.5 days for late meningitis. Before the first LP was performed, 62% of children received antibiotics. The most commonly used drugs were ampicillin (60%), cefotaxime (38%), vancomycin (28%), meropenem (33%) and gentamicin (22%), in different schedules.

    In the specific analysis of late meningitis, not including outpatients, two groups were differentiated (Figure 1). In the first group, 86.7% (13/15) of patients were pre-term infants, 46.6% (7/15) had respiratory distress, most were diagnosed at seven days of age, and their CSF was characterized by increased pleocytosis. In the second group, 75% (15/20) were full-term infants, fever and irritability were the most frequent symptoms and the diagnosis was made within the first two days of hospitalization.

    Three patients had special presentations. One had minor pleocytosis and urinary-related bacteremia by Escherichia coli; another had normal initial cytochemistry, but with a CSF culture positive for Escherichia coli that later developed pleocytosis; and the third one had three episodes of meningitis (recurrent) by extended-spectrum beta-lactamase-producing Escherichia coli.

At least one neurological complication was observed in 25% (15/60) of the cases, from which, 73.3% (11/15) were pre-term infants. Early and late neurological complications were ventriculitis and hydrocephalus respectively.

From the cases with late NM, 95.2% (40/42) were discharged with favorable evolution, early NM had a favorable evolution in 77.8% (14/18) of the cases. Four cases were referred to a hospital with a higher complexity level. Two neonates died (3.3%), one presenting early NM and the other, the late type.

NM had a hospital incidence of 1.4 cases per 1,000 live births, with a higher risk in pre-term than in full-term infants. The wide variability in incidence leads to the suspicion that diagnostic protocols for NM were not standarized across hospitals. A variety of causal germs, mostly bacteria, were identified, with the frequency of Eschericha coli and Listeria monocytogenes being particularly noteworthy.

    We present a larger NM clinical scenario than those known. In this scenario the early type is related to birth conditions and the late type is related to prolonged stay of pre-term infants in neonatal units ⁽⁹⁾. We provide a new viewpoint, derived from the community, which occurs more in full-term infants near the second week, related to a higher proportion of viral agents.

    Neonatal meningitis is an under-diagnosed and under-recorded prevalent disease in our country ^(5,13). In 2016, Zea et al., noted that LP is often deferred in confirmed sepsis ⁽¹⁴⁾. Likewise, in a similar population it has been observed that medical criteria may vary depending on the level of medical specialization ⁽¹⁵⁾.

    We present an epidemiological surveillance study according the management protocols of each hospital. The incidence of 1.4 per 1,000 live births is an average value worldwide ⁽³⁾, and is initially taken as a reference. This value will have to be adjusted in the future when the diagnostic criteria are standardized. However, the high incidence in premature infants alerts us about the need for vigilance in neonatal units ^{(2, 16)}.

    This study was characterized by the inclusion of cases with defined pleocytosis. This was made to meet the inflammatory criteria for meningitis, classified as confirmed, probable and possible, according to the definition of neonatal sepsis. Therefore, more positive isolates were found in blood cultures. Less were found to be positive in blood and CSF cultures, and only a few cases were observed solely in CSF culture. We believe that PCR could have helped to reduce the number of probable and possible cases, and to identify cases of pleocytosis as just an inflammatory phenomenon ^(8,9,17).

    All known risk factors for neonatal sepsis are related to NM ^(12,18). For the early type, peripartum fever and incomplete prenatal controls were found to be risk factors. These factors suggest the risk of microbial invasion from the vaginal flora, the subsequent placental inflammatory response, initiation of labor and consequently sepsis and meningitis ⁽⁸⁾. However, other clinical and sociocultural factors may not be considered.

    Classically, NM is divided into early and late according to its mechanism of contamination ^(12,16,19). However, we have identified a third group of patients who come from their homes, from the community environment, are term infants, febrile and irritable with less pleocytosis, contaminated with common respiratory tract agents, both bacterial and viral, and in some cases by germs that colonize maternal secretions.

    The age of symptoms onset for both types of NM was found to be within the expected ranges, 0.9 days for early NM cases and 18.6 days for late NM cases. This was found to be in accordance with other series, and clearly associated with the type of birth and neonatal unit stay ^(12,16,20). The group of children of out-of-hospital origin also behaved as late NM at 11.9 days.

    The symptoms were more frequent in early NM than in the late type, being very nonspecific and related to sepsis. Among them, respiratory difficulty stood out in 70% of early cases, perhaps due to lung immaturity in the premature group or to respiratory acidosis ^(3,12,16). In late NM more neurological symptoms were observed ^(1,3). However, identification of these symptoms depends on the experience of the examiner ^(14,21). Maternal fever, sepsis, and respiratory distress were three factors found to be more likely to develop in early NM than in the late type. These were probably generated by maternal infections, urinary tract infections and chorioamnionitis ^(1,8). It will remain for future studies to ensure the diagnosis of chorioamnionitis by pathological examination of the placenta.

    In most cases, more than one LP was carried out, following international guidelines. Given that NM is a difficult to diagnose multi-symptomatic disease caused by many aggressive agents, the guidelines recommend that the LP be performed prior to the use of antibiotics. It is also recommended that a new control should be performed within 48-72 hours, especially if there is no clinical improvement, with the purpose of reducing the bacterial load or achieving sterilization of the CSF ^(14,21,22).

    In both clinical types, moderate pleocytosis without PMN predominance was noted, they also presented hypoglycorrhachia and proteinorrhachia. This particular characteristic has already been observed in other national studies ^(6,7), perhaps, bacteriological factors, sample processing and patient’s immunological conditions are involved. In bacterial NM, hypoglycorrhachia and proteinorrhachia are common findings. These are explained by glucose consumption and increased detritus, their persistence for more than two weeks has been associated with poor prognosis ⁽²³⁾. However, these indicators may be aggravated by the presence of intracranial hemorrhage. Also, on rare occasions, the first LP may not demonstrate pleocytosis, and a second sample may be required within 12 to 24 hours ⁽²¹⁾.

    The microbiological behavior of NM has varied regarding time and different geographical areas ^{(2,3,16,24-26)}. Streptococcus agalactie stands out mainly in developed countries and gram‑negative bacteria in non‑developed countries ^(3,24). In this series, Eschericha coli and Listeria monocytogenes were the prevalent germs in both types of NM, followed by a variety of gram‑negative and gram‑positive bacteria, fewer virus cases and one case of Candida albicans, all described in different case series.

    NM by Eschericha coli has been known for many decades ^(1,2) to be a part of early neonatal sepsis cases. It can also cause late NM, usually associated with severe acute and mid‑term complications such as hydrocephalus, subdural effusions, cerebral infarctions and abscesses ^(2,27). In recent years, the increased frequency of beta‑lactamase strains and their antimicrobial resistance has being notable ^(24,26). Therefore, their presence in this series alerts about early identification and treatment.

    Listeria monocytogenes is a pathogen that has become more important in Peruvian series in recent years ^(5,7). It has been observed to be 5-20% ⁽²⁶⁾ of the early and late types reported, and it usually produces a moderate or severe disease, according to some international and national reports. However, its infectious mechanism is not clearly identified, but it is understood that the invasion is by genitourinary route and related to the maternal intestinal flora.

    The lethality rate by NM in national reports has been decreasing over time. In 1993, Oliveros et al. reported 20% death in a series of 24 cases ⁽⁶⁾, and in 2017, Lewis reported 3.8% in a series of 53 patients ⁽⁵⁾. Such decrease may be related to early diagnosis and treatment. However, the frequency of neurological complications was 25%, and the high morbidity in premature infants (75%) was noteworthy ^(28,29). Consequently, the use of cerebral ultrasound as a diagnostic tool for hydrocephalus, ventriculitis and cerebral infarction is very important in premature infants ⁽³⁰⁾.

    Not including certain variables such as prenatal steroid use, intrauterine infections, histological chorioamnionitis, invasive procedures, recording of sepsis cases without meningitis, antimicrobial sensitivity and resistance, community contacts, and not involving more hospitals are among the main limitations of this study. However, the strengths of the study were to demonstrate that NM is frequent, that pre-term infants are at greater risk, that the disease can present itself in different ways and that a wide spectrum of causal infectious agents exists. With these considerations we contribute to the national knowledge of this disease.

    In conclusion, the hospital incidence of NM was 1.4 cases per 1,000 live births, and even higher in premature infants. Respiratory distress was the most frequent symptom for early NM, while fever and irritability were the most frequent symptoms of late NM. Moderate pleocytosis, with hypoglycorrhachia and proteinorrhachia, was noted in the CSF. The most frequent pathogens isolated were Eschericha coli and Listeria monocytogenes. The most common neurological complications were ventriculitis and hydrocephalus. A new pathogenic scenario for NM is proposed, it consists of three infection types: vertical infection, by vaginal flora germs; nosocomial infection, by contamination in neonatal units; and infection from the community by common germs.

    A national epidemiological surveillance study of NM is recommended. This study should standardize diagnostic criteria (clinical, cytochemical, culture, PCR), neuroimaging criteria (ultrasound and resonance) and criteria for identification of perinatal risk factors.

Acknowledgement: To all physicians and nurses in the neonatal departments of the participating hospitals.

1. Bell WE, McCormick WF, Murillo PL. Meningitis neonatales. Infecciones neurológicas en el niño. 2a ed. Barcelona: Salvat; 1979.

2. Ziai M, Haggerty RJ. Neonatal meningitis. N Engl J Med. 1958;259(7):314-20. doi: 10.1056/NEJM195808142590702.

3. Ku LC, Boggess KA, Cohen-Wolkowiez M. Bacterial Meningitis in the Infant. Clin Perinatol. 2015;42(1):29–45. doi: 10.1016/j.clp.2014.10.004.

4. Holt DE. Neonatal meningitis in England and Wales: 10 years on. Arch Dis Child - Fetal Neonatal Ed. 2001;84(2):85F-89.

Funding: This research was entirely self-financed.

Authors’ contributions: DGP participated in the conception of this study. DGP, BM and JY participated in the study design, writing and data analysis. BM and JY participated in the enrollment and data collection. MLR, AM, MR, MLS, SV, OL, AT, LC, LF, LV, OE, CD and PM participated in the data collection. All authors reviewed and approved the article.

Conflict of interest: The authors declare no conflict of interest.

Received: 30/08/2019

Approved: 29/04/2020

Online: 15/06/2020