Neonatal sepsis and its associated factors in East Africa: a systematic review and meta-analysis, 2019

Background: Neonatal sepsis is one of the most leading causes of inflated death and illness of neonates. Different primary studies in Eastern Africa showed the burden of neonatal sepsis. However, inconsistency among those studies was seen and no review has been conducted to report the amalgamated magnitude and associated factors. Therefore, this review aimed to estimate the national prevalence and associated factors of neonatal sepsis in Eastern Africa. Methods: Using PRISMA guideline, we systematically reviewed and meta-analyzed studies that examined the prevalence and associated factors of neonatal sepsis from PubMed, Cochrane library, and Google Scholar. Heterogeneity across the studies was evaluated using the Q and the I2 test. A weighted inverse variance random-effects model was applied to estimate the national prevalence and the effect size of associated factors. The subgroup analysis was conducted by country, study design, and year of publication. A funnel plot and Egger’s regression test were used to see publication bias. Result: A total of 26 studies with 11239 participants were used for analysis. The pooled prevalence of neonatal sepsis in East Africa was 29.65% (95% CI; 23.36–35.94).Home delivery(AOR =2.67; 95% CI: 1.15-4.00; I2= 0.0%; P=0.996), maternal history of UTI (AOR=2.083; 95% CI :0.24-3.93; I2= 69.1%; P=0.001),gestational age (preterm) (AOR=1.56; 95% CI: 1.04-2.08; I2= 27.8%;P=0.000) ,prolonged labor (AOR=3.23 ;95% CI: -0.04-6.51; I2= 62.7%; P=0.020) and PROM (AOR= 1.95; 95% CI: 0.53-3.37; I2= 43.2%; P=0.062) were identified factors of neonatal sepsis. Conclusions: The prevalence of neonatal sepsis in Eastern Africa remains high. This review may help policy-makers and program officers to design neonatal sepsis preventive following items used appraise cross-sectional studies: criteria, (2) description of study subject and setting, (3) valid reliable measurement of exposure, (4) objective and standard criteria used, (5)identification of confounder, (6) strategies to handle confounder, (7) outcome measurement, and (8)appropriate statistical analysis. Studies were considered low risk when scored 50% and above of the quality assessment indicators.


Background
Neonatal sepsis is a systemic infection occurring in neonatal life and is a major cause of morbidity and mortality in newborns (1). It is categorized as early-onset neonatal sepsis (EONS) and late-onset neonatal sepsis (LONS) based on the onset of clinical features (2).EONS is sepsis occurring within seven days of life after birth and LONS is sepsis from after the seventh day of life (3). Neonatal sepsis contributes considerably to neonatal morbidity and mortality and is an continuing major global public health challenge (4). As documented in different literatures, neonatal sepsis is caused by factors related to both maternal and neonatal factors such as prolonged rupture of membrane (PROM),, urinary tract infection, intrapartum fever, instrumental delivery,prematurity, chorioamnionitis, frequent vaginal examination, never attend antenatal care (ANC), home delivery, meconium-stained amniotic fluid, contaminated foods intake, low birth weight, complicated or instrument-assisted delivery, low appearance pulse grimace activity respiration (APGAR) scores and invasive procedures during hospital admission (5)(6)(7)(8)(9)(10)(11).
Guidelines for the treatment of neonatal sepsis have been formulated and its implementation along with timely initiation of better treatments would satisfactorily decrease morbidity and mortality of neonates by sepsis (3).
Identification of the risk factors for risk-based diagnosis of neonatal sepsis contributes to better interventions and studies that help to reduce the burden of neonatal mortality resulting from these risks.
Worldwide, neonatal infections cause estimated 26% of under-5 deaths, with mortality rates highest in sub-Saharan Africa (4). Globally, sepsis in neonates is still among the principal causes of neonatal mortality and morbidity, especially in the first seven days of life in low and middle-income countries (LMIC) (12,13).
About four million worldwide deaths in neonates per year, from this 98 % is from developing countries particularly in sub-Saharan Africa (14). The risk of neonatal death is estimated to be six times more in the low and middle-income countries compared to developed (15). Timely diagnosis is difficult due to its nonspecific clinical manifestations.
Besides, treating neonates with antibiotics merely by subtle manifestations is likely to over-treat non-infected neonates (16). The ideal approach will be detecting high-risk neonates and steering them for intensive therapy (17).
Incidence of neonatal sepsis is about forty times higher and mortality rates are two times higher in middle-income countries compared with high-income countries (18).
Neonatal sepsis poses a massive public health burden for sub-Saharan Africa with significant associated economic consequences (19).In Africa sepsis accounts 28% neonatal deaths and infectious causes account for 68 deaths per 1000 live births (7).In sub-Saharan Africa, seventeen percent among all neonatal death results from neonatal sepsis as compared to only six percent in developed countries (20).
NS is also one of the most common cause of neonatal death in East Africa; it is the cause for more than one-third of neonatal deaths in Ethiopia particularly (21).
To achieve sustainable development goal (SDG) reducing newborn and under-five mortality as low as12/1000 and 25/1000 respectively is one of the Global strategies of WHO in African countries by 2030. This could be achieved through better prevention and management of preterm births and severe infections as the key (22).
Identification of risk factors and timely initiation of treatments can significantly decrease neonatal mortality and morbidity (23).
In the last two decades, remarkable progress has been shown on maternal and child deaths, but neonatal health is a part of the 'unfinished agenda'. The world is experiencing an increase in the proportion of under-five death occurring in the neonatal period. Yet despite the neonatal deaths are preventable, they are concentrated in the world's poorest countries. And 85% of all the neonatal were occurred in low and middle-income countries (LMICs) even though they are home to only 62% of the world's newborns (6,24).
Indeed, strategies that can prevent and treat neonates with sepsis are essential to accelerate the progress of newborn survival. In many developing country settings, however, the identification and treatment of newborns with infection is unsatisfactory.
Identification of risk factors and early institution of therapy thereby can improve neonatal mortality and morbidity (6,25 Searching strategy and information sources PubMed, Google Scholar, and Cochrane library were accessed. Articles with incomplete reported data were handled through contacting corresponding authors.

Study selection / Eligibility criteria
Retrieved studies were exported to reference manager software, Endnote version 8 to remove duplicate studies. Two investigators (BB and AM) independently screened the identified studies using their titles and abstracts before retrieval of full-text papers. We used pre-specified inclusion criteria to further screen the full-text articles. Disagreements were discoursed during a consensus meeting with a third reviewer (MW) for the final selection of studies to be included in the systematic review and meta-analysis.

Inclusion and exclusion criteria
This systematic review and meta-analysis included Cross-sectional, case-control, and cohort studies. Those studies had reported the prevalence and/or at least one associated factors of neonatal sepsis and published in English language were considered. There was no restriction of the researches study period. Citations lacking abstract and/or full-text, anonymous reports, editorials, and qualitative studies were excluded from the analysis.

Quality assessment
The qualities of the studies were appraised by three independent authors. The Joanna Briggs Institute (JBI) quality appraisal checklist was used (27). The disagreement was resolved by the interference of the third reviewer. The following items were used to appraise cohort studies: (1) similarity of groups, (2) similarity of exposure measurement, (3) validity and reliability of measurement, (4) identification of confounder,(5) strategies to deal with confounder, (6) appropriateness of groups/participants at the start of the study, (7) validity and reliability of outcome measured, (8) sufficiency of follow-up time, (9) completeness of follow-up or descriptions of reason to loss to follow-up, (10) strategies to address incomplete follow-up, and (11) appropriateness of statistical analysis. The items used to appraise case-control studies were: (1) comparable groups, (2) appropriateness of cases and controls, (3) criteria to identify cases and controls, (4) standard measurement of exposure, (5) similarity in measurement of exposure for cases and controls, (6) handling of confounder (7), strategies to handle confounder, (8) standard assessment of outcome, (9) appropriateness of duration for exposure, and (10) appropriateness of statistical analysis. Studies got 50% and above of the quality scale were considered low risk. The following items were used to appraise cross-sectional studies: (1) inclusion criteria, (2) description of study subject and setting, (3) valid and reliable measurement of exposure, (4) objective and standard criteria used, (5)identification of confounder, (6) strategies to handle confounder, (7) outcome measurement, and (8)appropriate statistical analysis. Studies were considered low risk when it scored 50% and above of the quality assessment indicators.

Data extraction
Two independent reviewers extracted data using a structured data extraction form.
Whenever variations of extracted data observed, the phrase was repeated. If discrepancies between data extractors continued, the third reviewer was involved. The name of the first author and year, the study country, the study design, the target population, the sample size, prevalence of neonatal sepsis, and AOR of associated factors were collected.

Outcome measurement
Neonatal sepsis was considered, neonates with presence of at least one clinical sign plus at least two laboratory results which are suggestive for neonatal sepsis (CRP,WBC,ANC, ESR, Platelet count, and Blood glucose) or neonates who are diagnosed as sepsis by attending physician and fulfill sepsis criteria within 0-28 days of life

Statistical analysis
We pooled the overall prevalence estimates of neonatal sepsis by a random effect metaanalysis (28).We examined the heterogeneity of effect size using Q statistic and the I 2 statistics (28). The Q-test measures whether the observed effect size is considerably different from one another than expected by chance. When Q test is higher than the degree of freedom it indicates significant heterogeneity (also supplemented by P-value).
The I2 statistics assess the proportion of total variance across the included studies contributed to the observed heterogeneity. In this study, the I2 statistic value of zero indicates true homogeneity, whereas the value 25, 50, and 75% represented low, moderate and high heterogeneity respectively (29,30).
For the data identified as heterogeneous, we conducted our analysis by random-effects model analysis.
When statistical pooling is not possible, non-pooled data was presented in table form.
Subgroup analysis was done by the study country, design, and year of publication.
Sensitivity analysis was employed to see the effect of a single study on the overall estimation.
Publication bias was checked by funnel plot and more objectively through Egger's regression test (31).

Result Study selection
A total of 4931 studies were identified; 3282 from PubMed, 12 from Cochrane Library, 1610 from Google Scholar and 27 from other sources. After duplication removed, 1235 remained. Finally, 301 studies were screened for full-text review and finally, 26 (n=11,239) were selected for the prevalence and/ or associated factors analysis (Fig. 1).
Characteristics and quality status of the studies  (5) (34)). From those studies, the pooled prevalence of neonatal sepsis in East Africa was 29.65 %( 95%CI; 23.36-35.94). We found significant heterogeneity among the studies (I 2 =98.8%; p<0.001).We analyzed by random-effects model analysis and we did subgroup analysis ( Figure 2).

Test of heterogeneity
Subgroup analysis of the prevalence of neonatal sepsis in Eastern Africa The subgroup analysis was done based on the country, study design, and year of publication. Based on this, the prevalence of neonatal sepsis found to be 38.31 % in

Sensitivity analysis
We employed a leave-one-out sensitivity analysis to identify the potential source of  Figure 6).

Publication bias
A funnel plot showed asymmetrical distribution .Egger's regression test p-value was 0.010, which indicated the presence of publication bias.

Prevalence of neonatal sepsis
The estimated overall prevalence of neonatal sepsis is presented in a forest plot (Fig. 4).

Factors associated with neonatal sepsis
In Eastern Africa context neonatal sepsis is associated with socio-economic, obstetric and maternal behavior, infant, and environmental-related factors (Table 3).  (5) revealed that neonates who delivered at home were 6.36 times at risk of being neonatal sepsis compared to those who delivered at the health institution. Gebremedhin et al (6) found that the odds of neonatal sepsis was higher among newborns who delivered at home (AOR=19, 95% CI: 1.74, 4.41) compared to those who delivered at the health institution.
Getabelew et al revealed that neonates who delivered at home were 6 times at risk of being neonatal sepsis compared to those who delivered at the health institution. Alebachew et al revealed that the odds of neonatal sepsis was higher among neonates whose mother have a history of UTI(AOR=2.9,95% CI: 1.48, 5.52) compared to those whose mother has no history of UTI. Yirga et al revealed that neonates whose mother have a history of UTI were 10.8 times at risk of being neonatal sepsis (95% CI: 3.44, 33.97) compared to those who delivered at the health institution. G/eyesus et al revealed that neonates whose mothers have a history of UTI were 7.06 times at risk of being neonatal sepsis compared to those whose mother has no history of UTI. Gebremedhin et al (6)found that the odds of neonatal sepsis was higher among neonates whose mother have a history of UTI (AOR=15.04, 95% CI: 1.65, 3.38) compared to those whose mother has no history of UTI. Getabelew et al revealed that neonates whose mothers have a history of UTI were 6.45 times at risk of being neonatal sepsis compared to those whose mother has no history of UTI. Okaba et al revealed that the odds of neonatal sepsis was higher among neonates whose mother have a history of UTI (AOR=6.28, 95% CI: 1.62, 7.38) compared to those whose mother has no history of UTI. Bua John et al revealed that the odds of neonatal sepsis was higher among neonates whose mother have a history of UTI (AOR=3.37,95% CI: 1.23, 9.22) compared to those whose mother has no history of UTI.
Four studies (J Mugalu, Mersha, et al.) found no significant association between maternal history of UTI and neonatal sepsis.
Test of heterogeneity for the maternal history of UTI Galbraith plot showed moderate heterogeneity and the forest plot showed the overall estimate of AOR of a place of birth was 2.083( 95%C I: 0.24-3.93;I 2 = 69.1%;P=0.001). I-Squared (I 2 ) and P-value also showed substantial heterogeneity. Main meta-analysis was done with random effect models( Figure 11).
The pooled estimate of UTI Publication bias maternal history of UTI A funnel plot showed a symmetrical distribution. Egger's regression test p-value was 0.928, which indicated the absence of publication bias( Figure 12).  found no significant association between gestational age and neonatal sepsis.

Test of heterogeneity gestational age
Galbraith plot showed moderate heterogeneity and the forest plot showed the overall estimate of AOR of the place of birth was 1.56 (95% CI: 1.04-2.08; I 2 = 27.8%; P=0.000). I-Squared (I 2 )and P-value also showed moderate heterogeneity.

Publication bias gestational age
A funnel plot showed an asymmetrical distribution. Egger's regression test p-value was 0.000, which indicated the presence of publication bias.

Trim and fill analysis gestational age
Trim and fill analysis was done and 2 studies were added and the total number of studies become 12 .the pooled estimate of AOR of preterm becomes 4.69( Figure 13).

Preterm pooled estimate Publication bias preterm
Begg's test shows there is publication bias among studies regarding gestational age of respondents ( Figure 14).

Trim and fill
After trim and fill analysis two studies were added and the pooled effect size changed from 1.56 to 4.69 ( Figure 15). .I-Squared (I 2 ) and P-value also showed substantial heterogeneity ( Figure 16).

Publication bias
Publication bias prolonged labor A funnel plot showed a symmetrical distribution. Egger's regression test p-value was 0.770, which indicated the absence of publication bias (Figure 17).  )and P-value also showed moderate heterogeneity( Figure 18).

Publication bias Publication bias PROM
A funnel plot showed an asymmetrical distribution. Egger's regression test p-value was 0.030, which indicated the presence of publication bias ( Figure 19).

Trim and fill PROM
Trim and fill analysis PROM Trim and fill analysis was done and 4 studies were added and the total number of studies become 15 .The pooled estimate of AOR of preterm becomes 5.86 ( Figure 20).

Discussion
In this systematic review and meta-analysis, we explored the prevalence and determinants of neonatal sepsis in Eastern Africa. 26 studies were included in the final analysis. Based on the meta-analysis a significant proportion (more than 1 in 4) of neonates had neonatal sepsis in Eastern Africa. This shows that neonatal sepsis is a significant public health problem in Eastern Africa. We also identified factors that were significantly associated with neonatal sepsis in Eastern Africa. In this study, the pooled prevalence of neonatal sepsis in Eastern Africa was. The pooled prevalence of neonatal sepsis in East Africa was 29.65 %( 95%CI; 23. 36-35.94).The results of this meta-analysis were higher than in studies conducted in other low and middle-income countries (LMICs), 17.2% (53).
These differences might be due to the socioeconomic and cultural differences between the countries. Moreover, the other obvious reason for the various might be the sample size, a collection of data from different settings (community and institution setting) as well as different study periods. Home delivery, maternal history of UTI, being preterm, prolonged labor and PROM were identified factors which significantly increase the risk of neonatal sepsis. A similar finding was also reported from the meta-analysis (54-56).

Conclusion
The prevalence of neonatal sepsis in Eastern Africa remains high. Home delivery, maternal history of UTI, being preterm, prolonged labor and PROM were identified factors which significantly increase the risk of neonatal sepsis. This review may help policy-makers and program officers to design neonatal sepsis preventive interventions.

Strength And Limitations
This study has several strengths: First, we used a pre-specified protocol for search strategy and data abstraction and conducted quality assessment two independent investigators to lessen the possible assessor bias; Second, we employed subgroup and sensitivity analysis based on study country, study design, and publication year to identify the small study effect and the risk of heterogeneity in; third, The quality of the included studies was evaluated by two authors. Nevertheless, our systematic review and meta- Prevalence of neonatal sepsis  Subgroup analysis by study design Subgroup analysis by country The pooled effect of place of birth Publication bias for the place of birth Trim and fill analysis place of birth The pooled estimate of UTI Figure 12 publication bias of gestational age (preterm) Figure 13 Preterm pooled estimate preterm Figure 14 Publication bias preterm Figure 15 Trim and fill preterm Figure 16 Pooled estimate of prolonged labor Figure 17 test of publication bias prolonged labor Publication bias PROM