Prevalence and factors associated with maternal and neonatal sepsis in sub-Saharan Africa: a systematic review and meta-analysis

Objectives This study aimed to determine the prevalence and factors associated with maternal and neonatal sepsis in sub-Saharan Africa. Methods This systematic review and meta-analysis used the PRISMA guideline on sepsis data in sub-Saharan Africa. The bibliographic search was carried out on the following databases: Medline/PubMed, Cochrane Library, African Index Medicus, and Google Scholar. Additionally, the reference lists of the included studies were screened for potentially relevant studies. The last search was conducted on 15 October 2022. The Joanna Briggs Institute quality assessment checklist was applied for critical appraisal. Estimates of the prevalence of maternal and neonatal sepsis were pooled using a random-effects meta-analysis model. Heterogeneity between studies was estimated using the Q statistic and the I2 statistic. The funnel plot and Egger’s regression test were used to assess the publication bias. Results A total of 39 studies were included in our review: 32 studies on neonatal sepsis and 7 studies on maternal sepsis. The overall pooled prevalence of maternal and neonatal sepsis in Sub-Saharan Africa was 19.21% (95% CI, 11.46–26.97) and 36.02% (CI: 26.68–45.36), respectively. The meta-analyses revealed that Apgar score < 7 (OR: 2.4, 95% CI: 1.6–3.5), meconium in the amniotic fluid (OR: 2.9, 95% CI: 1.8–4.5), prolonged rupture of membranes >12 h (OR: 2.8, 95% CI: 1.9–4.1), male sex (OR: 1.2, 95% CI: 1.1–1.4), intrapartum fever (OR: 2.4, 95% CI: 1.5–3.7), and history of urinary tract infection in the mother (OR: 2.7, 95% CI: 1.4–5.2) are factors associated with neonatal sepsis. Rural residence (OR: 2.3, 95% CI: 1.01–10.9), parity (OR: 0.5, 95% CI: 0.3–0.7), prolonged labor (OR: 3.4, 95% CI: 1.6–6.9), and multiple digital vaginal examinations (OR: 4.4, 95% CI: 1.3–14.3) were significantly associated with maternal sepsis. Conclusion The prevalence of maternal and neonatal sepsis was high in sub-Saharan Africa. Multiple factors associated with neonatal and maternal sepsis were identified. These factors could help in the prevention and development of strategies to combat maternal and neonatal sepsis. Given the high risk of bias and high heterogeneity, further high-quality research is needed in the sub-Saharan African context, including a meta-analysis of individual data. Systematic review registration: PROSPERO (ID: CRD42022382050).


Introduction
Maternal and neonatal sepsis is a generalized inflammatory response with systemic manifestations caused by one or more infectious agents (1), which occurs during pregnancy, childbirth, after abortion, or during the postpartum period (42 days) in women or in the first 28 days of life in newborns (2)(3)(4).Sepsis is a major cause of maternal and neonatal morbidity and mortality (2,3,(5)(6)(7)(8)(9).It is the third leading cause of death in women and accounts for a quarter of neonatal deaths (1, [10][11][12][13][14].It is an obstacle to achieving the third Sustainable Development Goal (SDG), which aims to reduce maternal and neonatal mortality and morbidity (15).Low-and middle-income countries (LICs) are particularly affected by maternal and neonatal sepsis (11,16,17).In sub-Saharan Africa (SSA), it is estimated to be responsible for 130,000 maternal deaths and 300,000 neonatal deaths per year, although this may be an underestimation (2,18).These deaths reflect a number of challenges, including policy, poverty, health inequalities, and the health system (19).However, few data are available on the prevalence and factors associated with sepsis across the continuum from pregnancy to postpartum or post-abortion, making it difficult to make a real estimation of maternal and neonatal sepsis in these countries (2,10,11).In the literature, the factors associated with maternal and neonatal sepsis are diverse, including prolonged labor, failure to perform antenatal consultation (ANC), prolonged rupture of membranes, history of infection in the mother, repeated vaginal examinations, intrapartum fever, gestational age, parity, type of delivery, prematurity, chorioamnionitis, meconiumstained amniotic fluid, Apgar score < 7, low birth weight < 2.5 kg, resuscitation of the newborn, age of the newborn <7 days, and male sex (5)(6)(7)(20)(21)(22)(23).
Prevention, early recognition of signs, and rapid and appropriate management of cases are the main factors associated with a reduction in the morbidity and mortality associated with maternal and neonatal sepsis (5,11,(24)(25)(26).A recent meta-analysis carried out in SSA in 2022 identified several risk factors for neonatal sepsis (27) but did not explore the magnitude of neonatal and/or maternal sepsis nor the factors associated with maternal sepsis.Other studies have shown that the prevalence of maternal sepsis was 39% in Ethiopia (28), 12.20% in Keyna (29), and 20% in Tanzania (30); for neonatal sepsis, it was 77.9 in Ethiopia (31), 20.5% in South Africa (32), 49.8 in Tanzania (33), 37.6 in Nigeria (34), and 17.5 in Ghana (35).Although these single studies reported data on the prevalence and factors associated with maternal and neonatal sepsis, there are no regionally representative pooled data on the magnitude and factors associated with maternal and neonatal sepsis in SSA.However, a better understanding of the burden and a synthesis of the evidence on the factors associated with maternal and neonatal sepsis are needed to optimize prevention strategies and management guidelines against this scourge.The aim of this systematic review with metaanalysis was to estimate the prevalence and factors associated with maternal and neonatal sepsis in sub-Saharan Africa.To the best of our knowledge, this is the first systematic review and meta-analysis to examine both the prevalence of and factors associated with maternal and neonatal sepsis in SSA.It aimed to answer the following questions: What is the prevalence of maternal and neonatal sepsis in sub-Saharan Africa?
What are the associated factors with maternal and neonatal sepsis in sub-Saharan Africa?

Materials and methods
This systematic review was reported in accordance with PRISMA (Preferred Reporting Item for Systematic Review and Meta-analysis) guidelines (36).The protocol for this review was developed and registered in the "International prospective register of systematic reviews PROSPERO" (ID: CRD42022382050).

Search strategies
To identify eligible studies, Medline/PubMed, Cochrane Library, African Index Medicus, and Google Scholar databases were searched.We also conducted manual searches of the bibliographic references of included studies and meta-analyses.The adapted PECO format was used for this systematic review.This PECO included population (P), exposure (E), comparison (C), and outcome (O), as shown in Table 1.It consisted of using all the identified keywords and indexing terms to search different databases.All the search terms used and the MeSH terms for the search were added, as well as the Boolean operators, to guarantee the exhaustiveness of the search process.Key terms defining the same concept were introduced using the "OR" operator, and the "AND" operator was used to introduce different concepts.

Selection of studies/eligibility criteria
The result of bibliographic searches carried out on the various search tools was exported to Zotero, where duplicates were identified and removed using the Duplicates command and manually also removed during the screening.The study selection process followed two evaluation stages (37).The first evaluation was based on an examination of the titles and abstracts of the articles.The titles and abstracts that were outside the scope of the study were excluded.The second assessment consisted of examining the full text of eligible studies.The entire process was carried out by two reviewers (FBT and NHD).They independently performed abstract screening and fulltext study selection, where both authors had to approve the inclusion of the study in the systematic review.The reference lists of the included studies were screened for potentially relevant studies (38).
Studies reporting on the prevalence and/or at least one factor associated with maternal and/or neonatal sepsis in SSA in pregnant women from 28 weeks of amenorrhea (SA), postpartum women up to 42 days after delivery, post-abortion women, and newborns within 28 days of birth were included.Cross-sectional, cohort, and casecontrol studies on the prevalence, frequency, and factors associated with maternal and neonatal sepsis in SSA published in French and English between January 2012 and October 2022 were included.Qualitative studies, systematic reviews, and case series were excluded from the analysis, but the reference lists of these were screened.

Assessment of study quality and risk of bias
The quality of the included studies was assessed by two authors (FBT and NHD) using the Joanna Briggs Institute (JBI) quality assessment checklist (39).For cross-sectional studies, the following criteria were used: (1) conformity between target population and source population; (2) appropriate sampling technique; (3) representativeness of the sample; (4) description of the subject and context of the study; (5) data analysis with sufficient sample coverage; (6) valid methods for identifying the condition; (7) standard and reliable way of measuring the condition for all participants; (8) appropriate statistical test; and (9) adequate response rate.When items received a score ≥ 6 out of 9, they were considered to be of high quality.The following were used to assess cohort studies: (1) similarity of groups, (2) similarity of exposure measurement, (3) validity and reliability of measurement, (4) identification of confounders, (5) strategies for dealing with confounders, (6) adequacy of groups/ participants at study entry, (7) validity and reliability of measured outcomes, (8) sufficient duration of follow-up, (9) completeness of follow-up or description of reasons for loss to follow-up, (10) strategies for dealing with incomplete follow-up, and (11) adequacy of statistical analysis.The criteria used to evaluate case-control studies are (1) comparable groups, (2) appropriateness of cases and controls, (3) criteria for identifying cases and controls, (4) standard exposure measurement, (5) similarity of exposure measurement for cases and controls, (6) treatment of confounding factors, (7) strategies for treating confounding factors, (8) standard evaluation of results, (9) appropriateness of duration of exposure, and (10) appropriateness of statistical analysis (see Table 2).

Data extraction
Two authors (FBT and NHD) independently extracted data using a tested form on Microsoft Excel.If discrepancies between data extractors continued, a third reviewer (EMD) was involved.Data extracted included author name and year of publication, country, study period, study setting, study design, study population, sample size, type of sepsis (maternal or neonatal), prevalence of neonatal sepsis, prevalence of maternal sepsis, and risk bias.
For the associated factors, data were extracted on the age of the newborn, Apgar score, gestational age, birth weight, resuscitation of the newborn at birth, sex of the newborn, antenatal consultation (ANC), presence of prolonged rupture of membranes, repeated vaginal examinations, intrapartum fever, gestational age, parity, type of delivery, prematurity, meconium amniotic fluid, history of infection in the mother, maternal fever, type of delivery, and prolonged labor.

Statistical analysis
Data were entered into Microsoft Excel and then exported into Stata version 17 software.We pooled maternal and neonatal sepsis prevalence estimates using a random-effects meta-analysis model because it accounts for variability between studies.We examined the heterogeneity of effect size using the Q statistic and the I 2 statistic (40).An I 2 value ≥ 50% was considered strong heterogeneity, and a random-effects model was used (41,42).The random-effect RELM method was mainly used.The funnel plot and Eggers's regression test were used to check for publication bias (43).To complete the tests for publication bias, we added the Begg and Thompson tests on R software.Forest plots were used to display the results graphically.A subgroup analysis was performed according to study design (prospective or retrospective), sepsis diagnostic criteria (clinic or clinic and biology), risk of bias (low or moderate), and different regions of SSA (East, West, or South Africa).

Results
A total of 2,638 titles were screened (2,033 from PubMed, 58 from the African Index of Medicus, 41 from Cochrane, and 506 from Google Scholar).After removing duplicates, 2,145 studies remained.Evaluation of titles and abstracts led to the exclusion of 2,019 studies.Nine studies were excluded because the full text was not available.The full-text review involved 126 studies.The number of articles retained for inclusion was 39.The reasons for the exclusion of 87 studies were the lack of availability of the variable of interest and the difference in the target population (Figure 1).

Subgroup analysis of the prevalence of maternal sepsis
A subgroup analysis of the prevalence of maternal sepsis was carried out according to the diagnostic criteria for sepsis (clinical or clinical and biological).It was 25.33 (9.28-41.38)for sepsis based on clinical signs and 15.43 (8.28-22.58)for sepsis based on clinical signs and biology (22,29,30).See Table 4 for details.

Prevalence of neonatal sepsis
The pooled prevalence of neonatal sepsis in SSA was 36.02% (95% CI, 26.68-49.36).A significant heterogeneity between the included Eligibility Full text articles assessed for eligibility (n=126) Full text articles excluded (n= 87) x Did not report variable of interest or other population (n= 87) Studies included in the review (n=39)

Identification Screening
Records screened (n=2145) Records excluded (n= 2019) x According to title (n=1801) x By abstract (n=209) x No full text available (n=9) PRISMA flow diagram describing the results of study selection.studies was observed (I2 = 99.1%,p < 0.000).Therefore, a random effects model was used to estimate the pooled prevalence.The prevalence ranged from 11.1 (56) up to 77.9% (31) (see Figure 3).

Subgroup analysis of the prevalence of neonatal sepsis
A subgroup analysis of prevalence was performed according to the design (prospective and retrospective), subdivision of SSA (East, West, or South Africa), definition of sepsis (clinical when diagnostic based on clinical signs or clinical and biological when diagnostic based on clinical signs and confirmed by biological tests), and risk of bias (low or moderate).
Depending on the design, the prevalence of neonatal sepsis was 32.88 (22.35-43.41)

Publication bias
We assessed publication bias using the funnel plot and Egger's regression test (76).Compared with the maternal sepsis studies, a visual asymmetrical distribution was observed, and Egger's regression test (value of p = 0.01) indicated the presence of a publication bias (see Figure 4).We did not use the Begg and Thompson tests due to the number of studies less than 10.
For the study concerning neonatal sepsis, we assessed publication bias using the funnel plot, Egger's regression test, Begg's test, and Thompson test.The diagrams showed asymmetry, and the result of Egger's test showed the presence of bias (p = 0.04; see Figure 5).However, Begg's test (value of p = 0.06) and the Thompson test (value of p = 0.14) did not reject the city Ho.Therefore, the asymmetry on the funnel plot is reflected much more by a small study effect than by a publication bias.

Discussion
Knowledge of the burden and risk factors of maternal and neonatal sepsis is crucial for developing preventive measures and reducing maternal and neonatal mortality.This systematic review with meta-analysis filled the gaps in the literature on the prevalence of maternal and neonatal sepsis and associated factors in SSA.Based on 39 studies included in the final analysis, we found a pooled prevalence of 19.21% for maternal sepsis and 36.02% for neonatal sepsis.However, considerable heterogeneity was observed.
Factors such as rural residence, parity, prolonged labor, and multiple vaginal examinations significantly increased the risk of maternal sepsis in our study.The factors associated with neonatal sepsis in this review are classified into maternal and neonatal factors.Maternal factors such as prolonged rupture of membranes > 12 h, intrapartum fever, and history of maternal urinary tract infection and neonatal factors such as Apgar score < 7, presence of meconium in amniotic fluid, and male sex were significantly associated with neonatal sepsis.Our estimate confirms that maternal and neonatal sepsis is a major public health problem in SSA.The pooled prevalence of maternal sepsis in our review is in line with other reviews conducted in 2009 (77) and 2021 (20).This high prevalence can be explained by various factors, such as the coverage of childbirth in health facilities, the asepsis and hygiene of surfaces and materials used, the personal hygiene of pregnant women, and certain harmful practices.Our study is the first to examine both the extent and the factors associated with maternal and neonatal sepsis, which sets it apart from other reviews.It identified other factors, such as rural residence and parity, which were not identified in the 2022 review or the WHO guidelines (78).
Our results could serve as a broader database for the development of interventions for the prevention and management of maternal and neonatal sepsis.All the maternal sepsis studies included were from East Africa.The subgroup analysis of the prevalence of maternal sepsis ranged from 15.43% based on clinical signs and biology to 25.33% based on clinical signs.
The subgroup analysis of the prevalence of neonatal sepsis varied from one subdivision to another.East Africa recorded the highest prevalence, 42.96%, which was higher than the pooled prevalence.This could be explained by the weakness of the health system, the quality of services offered, and socio-cultural factors.The underanalysis of sepsis in other parts of Africa, due to the small number of studies carried out there, masks a probable high prevalence in these countries, as areas of high prevalence in these countries have probably not been covered by the small number of studies published.The prevalence of neonatal sepsis was 44.11% for the diagnostic criterion based on clinical signs and 31.25% for clinical signs and biology.Despite the various subgroup analyses, they did not make it possible to explore the source of the heterogeneity.This could be explained by the heterogeneous definition of sepsis in the different studies.The clinical signs differed from one study to another, as did the laboratory tests, which is why a meta-analysis of the individual data was necessary to identify a single algorithm.
The factors associated with maternal sepsis in our review are partly in accordance with a review of the literature in other systematic reviews in 2009 and 2022 (27, 77), where the risk factors identified were intrapartum maternal fever, multiple vaginal examinations, foulsmelling vaginal discharge, prematurity, prolonged rupture of membranes, prolonged labor, and multiple vaginal touches.
Living in a rural area was significantly associated with maternal sepsis.Women living in rural areas were 2.3 times more likely to develop sepsis than those living in urban areas.The possible explanation could be poor hygiene levels, lack of water sources, poor cord care, overcrowding in homes, and low education level of mothers in rural areas (73,75,79) (80), which reported maternal factors such as intrapartum maternal fever, prolonged rupture of water membranes, gestational age < 37 weeks, prolonged labor, multiple vaginal touches, history of maternal urinary tract infection and low socioeconomic status, neonatal factors such as resuscitation at birth, low birth weight < 2.5 kg, Apgar score < 7, absence of crying immediately after birth, meconium-stained amniotic fluid, and male sex.
Intrapartum fever was significantly associated with maternal sepsis.The probability of maternal sepsis was 2.4 times higher in women with a fever during pregnancy than in those without a fever.Intrapartum fever is suggestive of a maternal infection that is transmitted to the baby in utero or during passage through the genital tract, leading to sepsis.
Membrane rupture > 12 h significantly increased the risk of neonatal sepsis, and a history of urinary tract infection in the mother was significantly associated with neonatal sepsis.The water membrane protects the upper genital tract; when it is ruptured, bacteria can proliferate through the dilated cervix into the upper internal genital tract, causing infection.These pathogens also colonize the birth canal, which could contaminate the newborn during passage through the canal.Intrapartum antibiotic prophylaxis has been recommended as an effective practice for at-risk mothers to reduce sepsis worldwide (81).Having an Apgar score < 7 significantly increased the risk of neonatal sepsis.As a result, the probability of neonatal sepsis was 2.4 times higher in neonates with an Apgar score < 7 compared with those with an Apgar score > 7.This finding is consistent with studies from Iraq (82) and Indonesia (83).Neonates with low Apgar scores tend to have a poor adaptation to extrauterine life due to the stress experienced during labor and therefore are more prone to infection.In addition, resuscitation procedures following birth asphyxia tend to expose newborns to pathogenic microbes.In our study, however, we did not assess birth asphyxia as a risk factor for neonatal sepsis.
Male sex has been identified as a factor associated with neonatal sepsis, and this finding has been made in the journals in India and SSA.However, we suggest further research into this factor to give a more rational explanation.

Strengths and limitations
Our review used an appropriate search strategy, with the combination of global and regional databases reducing the risk of missing relevant regional studies.Duplicate screening and data extraction, as well as rigorous quality assessment of included data and subgroup analysis, was also performed.The relatively high number of included studies for neonatal sepsis is a strength.The number of articles included was small for maternal sepsis (7), which may limit the generalizability of the results.Most of the studies were cross-sectional, which could be a limitation.The majority of studies included in this systematic review were from East Africa, which may affect the generalizability of our results to sub-Saharan Africa.There were also publication biases between studies and a high degree of heterogeneity.We believe that this is   an important subject that has not been sufficiently explored and that a meta-analysis of individual data will be necessary.

Conclusion
In our review, the prevalence of maternal and neonatal sepsis was high.Several factors were significantly associated with this prevalence, which could help prevent maternal and neonatal sepsis by developing appropriate standard infection prevention techniques, reducing certain harmful practices, and reducing susceptibility to infection by improving maternal health through nutritional supplementation and treating infections during pregnancy.However, there is a need for evidence on other important risk factors for maternal and neonatal sepsis, including in the community.Given the high risk of bias and high heterogeneity, further high-quality research is needed in the sub-Saharan African context.
error by odds ratio

FIGURE 5 Funnel
FIGURE 5Funnel plot showing publication bias d in neonatal sepsis studies.

TABLE 1
PECOT framework for the review objective.

TABLE 2
Quality assessment results of included studies in sub-Saharan Africa from January 2002-October 2022 using the Joanna Briggs Institute (JBI) quality appraisal checklist.

TABLE 3
Characteristics of included studies.

TABLE 4
Sub-group analysis of the prevalence of maternal sepsis in sub-Saharan Africa.

TABLE 5
Sub-group analysis of the prevalence of neonatal sepsis in sub-Saharan Africa.

TABLE 6
Risk factors included in the meta-analysis for maternal sepsis.

TABLE 7
Risk factors included in the meta-analysis for neonatal sepsis.