Perinatal risk factors for neonatal early-onset sepsis: a meta-analysis of observational studies

Abstract

Objective

Early-onset neonatal sepsis (EONS) remains an important cause of neonatal mortality and has many risk factors, therefore, this study aimed to investigate the perinatal risk factors for EONS.

Methods

We searched CNKI, Wan Fang, VIP, CBM, PubMed, Embase, and Web of Science to compile studies regarding the incidence of neonatal early-onset sepsis, published up to 1 May 2022. To evaluate the quality of the included studies, we used the Newcastle-Ottawa Scale, and the RevMan5.3 software was used for meta-analysis.

Results

A total of 17 studies were included, with 1987 cases in the case group and 4814 cases in the control group. Meta-analysis showed that perinatal asphyxia or intrauterine distress (OR = 3.00, 95% CI: 2.18–4.13), amniotic fluid meconium contamination (OR = 4.51, 95% CI: 2.31–8.81), group B streptococcal (GBS) colonization in pregnant women (OR = 2.13, 95% CI: 1.48–3.05), chorioamnionitis (OR = 4.58, 95% CI: 2.61–8.05), premature rupture of membranes (OR = 2.63, 95% CI: 2.09–3.30), lower gestational age (OR = 1.31, 95% CI: 1.18–1.44), maternal urinary or reproductive tract infection (OR = 3.61, 95% CI: 2.14–6.11), perinatal fever (OR = 3.59, 95% CI: 2.25–5.71), very low birth weight (OR = 3.79, 95% CI: 2.14–6.73), and vaginal examination ≥3 times (OR = 7.95, 95% CI: 4.04–15.64) were the perinatal risk factors for EONS.

Conclusion

Perinatal asphyxia or intrauterine distress, meconium contamination in amniotic fluid, GBS colonization in pregnant women, chorioamnionitis, premature rupture of membranes, lower gestational age, maternal urinary tract or reproductive tract infection, perinatal fever, very low birth weight, and vaginal examinations ≥3 times may increase the risk of EONS.

Keywords:

• Neonatal sepsis
• early onset
• perinatal period
• risk factors
• meta-analysis

1. Introduction

Neonatal sepsis is defined as a systemic inflammatory response caused by pathogens invading the neonatal blood circulation, where they grow, multiply, and produce toxins. Depending on the time of onset, neonatal sepsis can further be divided into early-onset and late-onset sepsis, where early-onset neonatal sepsis (EONS) refers to sepsis that develops within 72 h after birth [1,2].

The incidence of neonatal sepsis is estimated to be between 1 and 12 per 1000 live births in high-income countries [3,4], but the incidence in low- and middle-income countries is higher. The morbidity of EONS is estimated to be 7.1 to 38 per 1000 live births in Asia, 6.5 to 23 per 1000 live births in Africa, and 3.5 to 8.9 per 1000 live births in South America and the Caribbean [5]. Sepsis is probably responsible for 30–50% of the total neonatal deaths each year in developing countries [6]. Although the incidence of EONS is lower than that of late sepsis, it is still an important cause of neonatal morbidity and mortality [5,9].

Even though there are many studies on perinatal risk factors for EONS, the results are inconsistent. EONS is a severe disease, and it is necessary to be familiar with its risk factors. Currently, there are no meta-analyses of perinatal risk factors for EONS; therefore, this study systematically compiled and meta-analyzed studies on perinatal risk factors for EONS.

2. Materials and methods

2.1. Search strategy

We searched CNKI, Wan Fang, VIP, CBM, PubMed, Embase, and Web of Science to compile studies published up to 1 May 2022, regarding the incidence of EONS. A search strategy combining subject words and free words was adopted, and the following keywords were used: ‘Neonatal Sepsis,’ ‘newborn sepsis,’ ‘Neonatal Early-Onset Sepsis,’ ‘early onset,’ ‘Peripartum Period,’ ‘Peripartum,’ ‘perinatal,’ ‘prenatal,’ ‘risk,’ ‘mortality,’ and ‘cohort.’ The keyword search was supplemented by a manual search and literature traceability. This study adopted a secondary research method, and all the data were extracted from published literature; therefore, ethical approval was not necessary.

2.2. Inclusion and exclusion criteria

Inclusion criteria: (1) case-control studies and cohort studies on perinatal risk factors for EONS; (2) multivariate regression odds ratio (OR) values and 95% confidence intervals (CI) can be extracted; (3) in the case-control studies, the case group consisted of neonates with early-onset sepsis, and the control group consisted of neonates without sepsis.

Exclusion criteria: (1) study designs that included animal experiments, case reports, guidelines, reviews, systematic reviews, and review articles; (2) repeated publications; (3) articles without a control group; (4) unable to extract relevant data or there were errors in the data; (5) the ratio of the number of samples in the control group to the case group was >10.

2.3. Data extraction and quality evaluation

Two investigators independently conducted literature screening, literature quality evaluation, and data extraction according to the inclusion and exclusion criteria. When the results of the two investigators were different, a senior investigator was asked to determine the final result. We extracted the name of the first author, year of publication, study area, study time, study design, number of case groups, number of control groups, risk factors involved, OR value of multivariate regression, and 95% confidence interval from the included studies. The Newcastle-Ottawa Scale (NOS) was used to evaluate the quality of the included studies, and a NOS score ≥6 was considered a high-quality study.

2.4. Statistical analysis methods

RevMan5.3 software was used for meta-analysis and forest plot construction. The heterogeneity was evaluated according to the Q test; the I2 statistic and the test level were set to 0.1. If the P value was ≥0.1 and I2 was <50%, then the fixed-effect model was used to merge effect size; if the P value was <0.1 and I2 was ≥50%, the random effect model was used to combine effect size; if the P value and I2 results were opposite, the random effect model was used to combine effect size. Sensitivity analysis was conducted by eliminating studies individually to find the source of heterogeneity; the meta-analysis was performed again after the heterogeneity source elimination. The Funnel plot and Egger’s test were used to evaluate publication bias; the bias correction was done by trimming. The funnel plot was drawn using Stata 14.0.

3. Results

3.1. Literature screening results

In a preliminary screening according to the inclusion and exclusion criteria, 559 studies were obtained. In total, 17 studies (nine Chinese and eight English) were included in the meta-analysis. Full-text articles excluded 56 studies, as the research content or study design did not match, or the relevant data was unavailable. The flow diagram for literature selection is shown in Figure 1. The included studies consisted of 15 case-control and two cohorts with 1987 cases in the case group and 4,814 cases in the control group. The study area covered seven countries, such as China, the United States, the Netherlands, India, Bangladesh, Greece, and Tanzania. Studies included in this review had an NOS score of at least six points. The basic information of the included studies is shown in Table 1.

Figure 1. Flow diagram.

Table 1. The basic information of the included studies.

Study	Research area	Research time	Study design	Cases	Controls	NOS
Xu 2012 [7]	Fujian, China	2010.1–2011.12	Case control study	153	150	8
Dai 2018 [8]	Wenzhou, China	2013.12–2017.6	Case control study	129	159	7
Tian 2020 [9]	Zhejiang, China	2018.1–2019.8	Case control study	216	350	7
Zhuang 2020 [10]	Wenzhou, China	2018.12–2019.12	Case control study	80	80	9
Rong 2017[11]	Shijiazhuang, China	2014.10–2015.10	Case control study	100	100	7
Jia 2021 [12]	Taizhou, China	2018.1–2020.1	Case control study	119	357	7
Rong 2015 [13]	Shijiazhuang, China	2013.9–2014.8	Case control study	82	356	7
Wu 2021 [14]	Jiangsu, China	2016–2019	Case control study	168	168	7
Sheng 2018 [15]	Shanghai, China	2016–2017	Case control study	32	88	6
Nayeri 2018 [16]	United States	Not available	Cohort study	92	279	6
Van Kempen 2016 [17]	Netherlands	2002.3–2015.4	Case control study	68	346	8
Dutta 2010 [18]	India	Not available	Case control study	85	516	7
Santhanam 2018 [19]	India	2004–2014	Case control study	71	284	7
Rafi 2020 [20]	Bangladesh	2019.1–2019.12	Case-control study	56	116	8
Rallis 2021 [21]	Greece	2017–2018	Cohort study	81	185	7
Masanja 2020[22]	Tanzania	2017.5–2017.7	Case control study	105	217	8
Puopolo 2011 [23]	United States	1993–2007	Case control study	350	1063	7

3.2. Meta-analysis results

Risk factors included in ≥3 articles were selected for meta-analysis and included perinatal asphyxia or intrauterine distress, meconium contamination of amniotic fluid, group B streptococcal (GBS) colonization in pregnant women, chorioamnionitis, premature rupture of membranes, lower gestational age, maternal urinary or reproductive tract infection, perinatal fever, very low birth weight, and vaginal examination ≥3 times. The results of the heterogeneity test showed that there was no heterogeneity in the four risk factors of perinatal asphyxia or intrauterine distress, maternal urinary or reproductive tract infection, perinatal fever, and vaginal examination ≥3 times (I2 < 50%), so the fixed-effect model was selected to combine the effect size. The remaining six risk factors had heterogeneity (I2 > 50%), so the random effect model was selected to combine the effect size. Meta-analysis results showed that perinatal asphyxia or intrauterine distress (OR = 3.00, 95% CI: 2.18–4.13); meconium contamination in amniotic fluid (OR = 3.61, 95% CI: 1.97–6.60); GBS colonization in pregnant women (OR = 3.45, 95% CI: 1.45–8.21); chorioamnionitis (OR = 4.58, 95% CI: 2.61–8.05); premature rupture of membranes (OR = 3.25, 95% CI: 2.25–4.69); lower gestational age (OR = 1.55, 95% CI: 1.23–1.96); maternal urinary tract or reproductive tract infection (OR = 3.61, 95% CI: 2.14–6.11); perinatal fever (OR = 3.59, 95% CI: 2.25–5.71); very low birth weight (OR = 3.79, 95% CI: 2.14–6.73); and vaginal examination ≥3 times (OR = 7.95, 95% CI: 4.04–15.64) were perinatal risk factors for EONS. The results of the heterogeneity test and meta-analysis are summarized in Table 2.

Table 2. The results of heterogeneity test and meta-analysis.

Risk factors	Literature sources	Heterogeneity test			Effect model	OR(95%CI)	P-value
		Q	I²(%)	P
Perinatal asphyxia or intrauterine distress	[7,8,12,14,20,22]	2.58	0	0.76	Fixed	3.00(2.18 ∼ 4.13)	<0.01
Meconium contamination of amniotic fluid	[7,9,12,15,19]	13.80	71	<0.01	Random	3.61(1.97 ∼ 6.60)	<0.01
GBS colonization in pregnant women	[10,11,13,23]	14	79	<0.01	Random	3.45(1.45 ∼ 8.21)	<0.01
Chorioamnionitis	[8,10,11,13,15,18,21,22]	19.51	64	<0.01	Random	4.58(2.61 ∼ 8.05)	<0.01
Premature rupture of membranes	[7–9,11–14,17,22]	20.67	61	<0.01	Random	3.25(2.25 ∼ 4.69)	<0.01
Lower gestational age	[11,16–20,20,23]	44.43	89	<0.01	Random	1.55(1.23 ∼ 1.96)	<0.01
Maternal urinary or reproductive tract infection	[12,19,20]	1.20	0	0.55	Fixed	3.61(2.14 ∼ 6.11)	<0.01
Perinatal fever	[12,14,19]	3.25	38	0.20	Fixed	3.59(2.25 ∼ 5.71)	<0.01
Very low birth weight	[12,14,18]	4.29	53	0.12	Random	3.79(2.14 ∼ 6.73)	<0.01
Vaginal examination ≥ 3 times	[18,19,22]	0.45	0	0.80	Fixed	7.95(4.04 ∼ 15.64)	<0.01

3.3. Sensitivity analysis

The sensitivity analysis was conducted by excluding the literature successively. The heterogeneity of amniotic fluid meconium pollution was strong (I² = 71%, p < 0.01); after excluding the study by Tian et al. 2020 [9], the heterogeneity decreased significantly (I² = 59%, p = 0.06) and a random effect model was used to perform the meta-analysis again. The heterogeneity of GBS colonization in pregnant women was very strong (I² = 79%, p = 0.03) and disappeared after excluding Rong et al. 2015 [13] (I² = 45%, p = 0.16), indicating that this article was the source of heterogeneity. After excluding the source, a fixed-effect model was used to perform the meta-analysis again. The heterogeneity of lower gestational age was very strong (I² = 85%, p < 0.01) and disappeared after excluding Rong et al. 2017 [11] and Puopolo et al. 2011 [23] (I² = 43%, p = 0.15), indicating that the two studies were the source of heterogeneity. The fixed-effect model was used to perform the meta-analysis again after excluding those studies. Premature rupture of membranes had moderate heterogeneity (I² = 61%, p < 0.01) that disappeared after excluding Rong et al. 2015 [13] (I² = 39%, p = 0.12), indicating that this article was the source of heterogeneity. After exclusion, the meta-analysis was performed again using a fixed-effect model. There was moderate heterogeneity for chorioamnionitis (I² = 64%, p < 0.01), and this factor involved a cohort study, but a sensitivity analysis did not identify the source of heterogeneity.

The final results of the meta-analysis showed that perinatal asphyxia or intrauterine distress (OR = 3.00, 95% CI: 2.18–4.13); amniotic fluid meconium contamination (OR = 4.51, 95% CI: 2.31–8.81); GBS colonization in pregnant women (OR= 2.13, 95% CI: 1.48–3.05); chorioamnionitis (OR = 4.58, 95% CI: 2.61–8.05); premature rupture of membranes (OR = 2.63, 95% CI: 2.09–3.30); lower gestational age (OR = 1.31, 95% CI: 1.18–1.44); maternal urinary or reproductive tract infection (OR = 3.61, 95% CI: 2.14–6.11); perinatal fever (OR = 3.59, 95% CI: 2.25–5.71); very low birth weight (OR = 3.79, 95% CI: 2.14–6.73); and vaginal examinations ≥3 times (OR = 7.95, 95% CI: 4.04–15.64) were the perinatal risk factors for EONS (Table 3, Figure 2: perinatal asphyxia or intrauterine distress as an example forest plot).

Table 3. The final results of meta-analysis.

Risk factors	Literature sources	Heterogeneity test			Effect model	OR (95%CI)	P-value
		Q	I² (%)	P
Perinatal asphyxia or intrauterine distress	[7,8,12,14,20,22]	2.58	0	0.76	Fixed	3.00(2.18 ∼ 4.13)	<0.01
Meconium contamination of amniotic fluid	[7,12,15,19]	7.28	59	0.06	Random	4.51(2.31 ∼ 8.81)	<0.01
GBS colonization in pregnant women	[10,11,23]	3.61	45	0.16	Fixed	2.13(1.48 ∼ 3.05)	<0.01
Chorioamnionitis	[8,10, 11,13,15,18,21,22]	19.51	64	<0.01	Random	4.58(2.61 ∼ 8.05)	<0.01
Premature rupture of membranes	[7–9,11,12,14,17,22]	11.40	39	0.12	Fixed	2.63(2.09 ∼ 3.30)	<0.01
Lower gestational age	[16–18,20]	5.30	43	0.15	Fixed	1.31(1.18 ∼ 1.44)	<0.01
Maternal urinary or reproductive tract infection	[12,19,20]	1.20	0	0.55	Fixed	3.61(2.14 ∼ 6.11)	<0.01
Perinatal fever	[12,14,19]	3.25	38	0.20	Fixed	3.59(2.25 ∼ 5.71)	<0.01
Very low birth weight	[12,14,18]	4.29	53	0.12	Random	3.79(2.14 ∼ 6.73)	<0.01
Vaginal examination ≥ 3 times	[18,19,22]	0.45	0	0.80	Fixed	7.95(4.04 ∼ 15.64)	<0.01

3.4. Publication bias test

Funnel plot and Egger’s test were used to assess publication bias, and the results showed chorioamnionitis (t = 2.65, p = 0.03), premature rupture of membranes (t = 2.49, p = 0.04) and lower gestational age (t = 16.34, p < 0.01) had publication bias (Table 4, Figure 2 and Figure 3: perinatal asphyxia or intrauterine distress as an example funnel plot). After correction by trimming, the effect size of chorioamnionitis (OR = 3.25, 95% CI: 1.77–5.98), premature rupture of membranes (OR = 2.15, 95% CI: 1.55–2.98) and lower gestational age (OR = 1.28, 95% CI: 1.04–1.57) changed slightly, but the direction did not change, indicating that the meta-analysis results were stable and reliable.

Figure 2. Forest plot of perinatal asphyxia or intrauterine distress.

Figure 3. Funnel plot of perinatal asphyxia or intrauterine distress.

Table 4. The results of Publication bias test.

Risk factors	Egger's test
	t	P-value
Perinatal asphyxia or intrauterine distress	2.15	0.09
Meconium contamination of amniotic fluid	1.81	0.21
GBS colonization in pregnant women	0.51	0.70
Chorioamnionitis	2.65	0.03
Premature rupture of membranes	2.49	0.04
Lower gestational age	16.34	<0.01
Maternal urinary or reproductive tract infection	–6.87	0.09
Perinatal fever	0.67	0.62
Very low birth weight	<0.01	0.99
Vaginal examination ≥3 times	–1.43	0.38

4. Discussion

The clinical manifestations of EONS are usually atypical or asymptomatic, making its early diagnosis difficult [24]. The occurrence of this disease is closely related to perinatal factors, so accurately and quickly identifying the risk factors contributes to its prevention, and timely diagnosis prevents the delay of treatment. The final results showed that perinatal asphyxia or intrauterine distress, amniotic fluid meconium contamination, pregnant women with GBS colonization, mothers suffering from chorioamnionitis, premature rupture of membranes, lower gestational age, maternal urinary or reproductive tract infection, perinatal fever, very low birth weight, and vaginal examination ≥3 times were perinatal risk factors for EONS.

This meta-analysis showed that perinatal asphyxia or intrauterine distress and meconium contamination of amniotic fluid may increase the risk of EONS. Asphyxia leads to immune damage and immune function decline [25], which further damages the mucosal barrier function in newborns, and this results in defensive ability reduction and pathogenic infection resistance [26]. In addition, when the fetus is hypoxic, meconium may enter the amniotic fluid. The amniotic fluid itself has a protective effect on the fetus, but meconium can act as a growth factor, reducing its bacteriostatic ability, leading to the reproduction of bacteria, and antagonizing the host’s defense system [27,28].

This meta-analysis showed that chorioamnionitis and premature rupture of membranes are perinatal risk factors for EONS. Premature rupture of membranes ≥18 h has been reported to be an independent risk factor for acute chorioamnionitis, and neonates born to mothers with premature rupture of membranes and chorioamnionitis are approximately twice as likely to have EONS compared to neonates born to mothers with premature rupture of membranes alone [29–31]. This reminds clinicians to evaluate and prevent chorioamnionitis in a timely manner. However, the evaluation of fetal inflammatory response should also be considered because fetal involvement is more predictive of neonatal outcome than isolated maternal or uterine inflammation. In addition, clinically meaningful early disease biomarkers can provide more rapid and accurate risk prediction [32]. All stages of funisitis are strongly associated with neonatal early-onset sepsis [33]. Due to the study design and the number of study limitations, a meta-analysis of this factor was not performed in this study.

This meta-analysis showed that lower gestational age and very low birth weight are perinatal risk factors for EONS. Newborns have an immature immune system and cannot produce an adequate immune response, which increases their risk of developing the disease. Moreover, very low birth weight and preterm infants have a poor ability to pump and reserve milk, possibly leading to hypoglycemia, which is one of the reasons for the high mortality rate of EONS [34].

EONS is usually caused by vertical transmission, so there is no doubt that maternal urinary or reproductive tract infection may increase neonatal morbidity risk [35–37]. Further research found that the pathogenic bacteria were mainly Gram-negative bacilli obtained before and after childbirth, and these bacteria were found to be strongly associated with unhygienic practices and suboptimal sterilization [38,39]. Therefore, we have reason to suspect that nonstandard vaginal examinations of pregnant women ≥3 times may increase the risk of EONS. Furthermore, GBS is a major pathogen in EONS in developed countries, but the incidence of GBS sepsis in Chinese neonates is lower than that in developed countries, which may be due to the lower rate of GBS colonization in Asian pregnant women [39].

This meta-analysis showed that perinatal fever is a risk factor for EONS, and a previous study reported that the higher the temperature, the higher the risk of neonatal infection [40]. Additionally, research by Towers et al. showed that Peripartum maternal fever is a risk factor for EONS, and if present, requires prophylactic antibiotic treatment for infants, even for clinically healthy newborns [41].

This study had several limitations. First, some of the factors involved were found in less than three studies, so no meta-analysis was performed for them, which may affect the comprehensiveness of the results. Second, the small sample size of some studies may affect the accuracy of their findings and be a source of heterogeneity. Many of the newborns in the control group came from hospitals and may have been affected by other diseases. Finally, the studies included in this meta-analysis came from seven countries; while enriching the sample sources, there may also be heterogeneity due to race and living environment differences. In addition, nine studies came from China and covered most of China, so there may be a potential geographical bias. Therefore, we need more high-quality studies to verify and enrich the conclusions of this study.

In summary, perinatal asphyxia or intrauterine distress, meconium contamination in amniotic fluid, GBS colonization in pregnant women, chorioamnionitis, premature rupture of membranes, lower gestational age, maternal urinary or reproductive tract infection, perinatal fever, very low birth weight, and vaginal examination ≥3 times may increase the risk of EONS. Therefore, we must strengthen the health care of pregnant women, actively identify and prevent perinatal risk factors, and rationally use antibiotics to reduce the incidence of EONS.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.