Determinants of preterm birth among mothers who gave birth in East Africa: systematic review and meta-analysis

Background Preterm birth (PTB) can be caused by different factors. The factors can be classified into different categories: socio demographic, obstetric, reproductive health, medical, behavioral and nutritional related. The objective of this review was identifying determinants of PTB among mothers who gave birth in East African countries. Methods We have searched the following electronic bibliographic databases: PubMed, Google scholar, Cochrane library, AJOL (African journal online). Cross sectional, case control and cohort study published in English were included. There was no restriction on publication period. Studies with no abstracts and or full texts, editorials, and qualitative in design were excluded. Funnel plot was used to check publication bias. I-squared statistic was used to check heterogeneity. Pooled analysis was done by using fixed and random effect model. The Joanna Briggs Critical Appraisal Tools for review and meta-analysis was used to check the study quality. Results A total of 58 studies with 134,801 participants were used to identify determinants of PTB. On pooled analysis, PTB was associated with age < 20 years (AOR 1.76, 95% CI: 1.33–2.32), birth interval less than 24 months (AOR 2.03, 95% CI 1.57–2.62), multiple pregnancy (AOR 3.44,95% CI: 3.02–3.91), < 4 antenatal care (ANC) visits (AOR 5.52, 95% CI: 4.32–7.05), and absence of ANC (AOR 5.77, 95% CI: 4.27–7.79). Other determinants of PTB included: Antepartum hemorrhage (APH) (AOR 4.90, 95% CI: 3.48–6.89), pregnancy induced hypertension (PIH) (AOR 3.10, 95% CI: 2.34–4.09), premature rupture of membrane (PROM) (AOR 5.90, 95% CI: 4.39–7.93), history of PTB (AOR 3.45, 95% CI: 2.72–4.38), and history of still birth/abortion (AOR 3.93, 95% CI: 2.70–5.70). Furthermore, Anemia (AOR 4.58, 95% CI: 2.63–7.96), HIV infection (AOR 2.59, 95% CI: 1.84–3.66), urinary tract infection (UTI) (AOR 5.27, 95% CI: 2.98–9.31), presence of vaginal discharge (AOR 5.33, 95% CI: 3.19–8.92), and malaria (AOR 3.08, 95% CI: 2.32–4.10) were significantly associated with PTB. Conclusions There are many determinants of PTB in East Africa. This review could provide policy makers, clinicians, and program officers to design intervention on preventing occurrence of PTB.


Background
Preterm birth (PTB) is birth occurs between 20 weeks of pregnancy and 37 weeks of pregnancy. It is a concern because babies who are born too early may not be fully developed. They may be born with serious health problem. Some problems like cerebral palsy, can last a life time. Other problems like learning disabilities may appear later in childhood or even adulthood [1].
Each year 15 million babies in the world, more than one in 10 births is born too early. More than 1 million of those babies die shortly after birth; countless other suffer some type of life long physical, neurological or educational disabilities often at great cost to families and society [2]. The survival chances of the 15 million babies born preterm each year vary dramatically depending on where they are born. South Asia and sub-Saharan Africa account for half the world's births, more than 60% of the world's preterm babies and over 80% of the world's 1.1 million deaths due to PTB complications. Around half of these babies are born at home. Even for those born in a health clinic or hospital, essential newborn care is often lacking. The risk of a neonatal death due to complications of PTB is at least 12 times higher for an African baby than for a European baby. Yet, more than three-quarters of PTB could be saved with feasible, cost-effective care, and further reductions are possible through intensive neonatal care [3]. PTB has multiple causes; therefore, solutions will not come through a single discovery but rather from an array of discoveries addressing multiple biological, clinical, and socio-behavioral risk factors. Age of mother [4], household income [5], educational status of mother [4,6], place of residence and employment status [7] were associated with PTB. Many studies in different settings of the world revealed the contributing factors of PTB as physical activity [8], maternal cardiovascular disease [9], delivering by previous cesarean section [10], had history of miscarriage [11], and history of PTB [11][12][13]. The contributing factors for PTB also include pregnancy interval [14], body mass Indexes(BMII) [11], antenatal care(ANC) [12,15,16], multiple pregnancy [17], antepartum hemorrhage(APH) [15],urinary tract infections(UTI) [11], premature rupture of membrane(PROM) [15,18], and pregnancy induced hypertension(PIH) [17,19]. Moreover, marital status [12], polyhydramnios or oligohydramnios and genitourinary infections [20], periodontal disease [11], ascending infection (bacteriuria) [21] and exposure to intimate partner violence (IPV) [15,22] are included in contributing factors of PTB.
So far different researches are done and published on determinants of PTB among mothers who gave birth in East Africa countries. However, the results of the studies were inconsistent, factors that had direct association in some studies may be inversely associated or had no association in other studies and vise-versa. Moreover, to the best of our knowledge, there is no pooled data on determinants of PTB in East Africa. Hence, this systematic review and meta-analysis was conducted to identify determinants of PTB among mothers who gave birth in East Africa countries. This will help to make conclusions based on best available scientific evidence. Moreover, the result of this review could support policy makers, clinicians, and programmers to design intervention on preventing PTB.

Methods and materials
Reporting The report was written by using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)) guideline [23]. This review was registered in PROSPERO database (PROSPERO 2019: CRD42019127645).

Inclusion and exclusion criteria
Cross sectional, case control and cohort studies done in East African countries were included in this study. East African countries include (Sudan, South Sudan, Kenya, Uganda, Djibouti, Eritrea, Ethiopia, Somalia, Tanzania, Rwanda, Burundi, Comoros, Mauritius, Seychelles, Mozambique, Madagascar, Zambia, Malawi, Zimbabwe, Reunion, Mayotte) [24]. Studies reported the determinants of PTB and published in English were incorporated. There was no restriction on publication period. Studies with no abstracts and or full texts, editorials, and qualitative studies were excluded.

Searching strategy and information sources
We have searched the following electronic databases: PubMed, Google scholar, Cochrane library, AJOL (African journal online). Furthermore, we have searched bibliographies and contacted authors. There was no restriction on publication period. To conduct a search of the literature databases, we have used Boolean Logic, connectors "AND", "OR" in combinations [25]. The search strategy for PubMed database was done as following: preterm OR "preterm birth" OR "premature birth"(MeSH terms) AND

Study selection
Studies retrieved from database were exported to reference note manager, endnote version 7 to remove duplicate studies. Title and abstract was screened by two reviewers. The full text of these potentially eligible studies was retrieved and independently assessed for eligibility by two review team members. Disagreement between them over the eligibility of particular studies was resolved through discussion with a third reviewer.

Risk of bias in individual studies
To evaluate the quality of the papers, the Joanna Briggs Critical Appraisal Tools for review and meta-analysis was used [26]. Two independent reviewers assessed the quality of the study. Differences was reconciled by a third reviewer.
The following items was used to appraise case control studies:(1) comparable groups, (2) cases and controls matched, (3) the same criteria used for identification, (4) exposure measured in a standard, valid and reliable way, (5) exposure measured in the same way for cases and controls, (6) confounding factors identified, (7) strategies to deal with confounding factors, (8) outcomes assessed in a standard, valid and reliable way for cases and controls, (9) The exposure period of interest long enough, and (10) appropriate statistical analysis. Cohort studies were appraised by using the following items:(1) the two groups similar and recruited from the same population, (2) the exposures measured similarly to assign people to both exposed and unexposed groups, (3) the exposure measured in a valid and reliable way,(4)confounding factors identified, (5) strategies to deal with confounding factors, (6) participants free of the outcome at the start of the study, (7) the outcomes measured in a valid and reliable way, (8) follow up time reported and sufficient to be long enough for outcomes to occur, (9) follow up complete, and if not, were the reasons to loss to follow up described and explored, (10) strategies to address incomplete follow up, and (11) appropriate statistical analysis. For cross sectional studies the following items were used to appraise the quality: (1) criteria for inclusion, (2) study subjects and the setting described, (3) exposure measured in a valid and reliable way, (4) standard criteria used for measurement, (5) confounding factors identified, (6) strategies to appropriate statistical analysis deal with confounding factors, (7) outcomes measured in a valid and reliable way, and (8) appropriate statistical analysis.
Studies scored 50% and above in the quality assessment indictors were considered as low risk and included in the analysis.

Data collection process
Two independent reviewers extracted data by using structured data extraction form. The name of the first author and year, country, study design, sample size, determinants of PTB, AOR (95% CI), events and total in experimental and control groups were extracted. Whenever variations of extracted data observed, the phase was repeated.

Data analysis
To identify determinants of PTB, the analyses were divided in to six parts: Socio economic and demographic factors, reproductive health (RH), obstetric factors, medical condition, nutrition and behavioral factors. The Meta-analysis was done by using RevMan 5.3 software [28]. Heterogeneity of the studies was done by I-squared statistic (I 2 ). A values of 25, 50, and 75% represented low, moderate, and high I 2 , respectively [29]. In this study I-squared value less than 50% was considered to interpret the combined effect size. Publication bias was checked by funnel plot. As the studies included in each outcome was less than 10, funnel plot was not presented [30]. Sensitivity analysis could investigate whether any indication of bias (such as different sizes of estimates from studies with individual participant data and from those without, or evidence of funnel plot asymmetry) remains when studies with individual participant data are standardized to match those lacking individual participant data [31]. We have conducted sensitivity analysis to see the effects of a single study on determinants of PTB. For small number of studies, it may be impossible to estimate the between studies variance with any precision. Therefore, we used fixed effect model for less than five studies and random effect model for five and above studies [32]. Pooled analysis was done using mantelhaenszel (M-H) statistical methods and effect measure was computed by odds ratio by using fixed and random effect model [30].

Risk of bias within studies
Joanna Briggs Critical Appraisal Tools for review and meta-analysis for case control studies, cross sectional studies and cohort studies were used. We included studies that had low risk (Table 1).

Conceptual frame work
The conceptual framework of the study showing determinants of PTB in East Africa (Fig. 6).

Discussion
The objective of this systematic review and metaanalysis was identifying determinants of PTB in East Africa. Age of women less than 20 years was correlated with PTB. This is comparable with other studies [22,87]. The increased risk PTB in younger age can be linked to the fact that their reproductive organs are not yet fully developed. The current study depicted that history of still birth/ abortion was significantly associated with PTB. This is related with systematic review and meta-analysis study [11,88]. History of PTB was significantly associated with PTB. This is in agreement with other studies [11-13, 19, 89]. The reason for this could be the likelihood of having PTB with the women with prior spontaneous labor as well as those with inducing PTB rising. PROM was significantly associated with PTB. This is similar with systematic review and meta-analysis study [18].
Shorter pregnancy interval was significantly associated with PTB. This is in line with studies done in Egypt and other places [11,14,90]. This may be because mothers do not have time to recover from the physical stress and nutritional burden of the pregnancy.
Women who attended ANC < 4 times had higher probability to have PTB. This finding is comparable with systematic review and meta-analysis study conducted in Iran [16]. Likewise, the absence of ANC was significantly associated with PTB. The reason for this might be when women had no chance to attend ANC, she cannot be informed of early identification of risk factors associated with PTB. Consequently, the probability of having PTB will increase.
Having multiple pregnancy increased the probability of PTB. It is in line with other studies [17,91].This may be due to overstretching of uterus and deciding to complete pregnancy before term. Furthermore, it may be due to spontaneous labor or PROM. Maternal UTI was associated with PTB. This is in agreement with other studies [11,15,19]. UTI can weaken the membranes of the amniotic sac around the baby. This could lead to PROM and preterm labor [92]. Presence of malaria increased the risk of PTB. This is in line with other studies [15]. Presence of anemia was associated with PTB. This is in agreement with other study [15]. Anemia can decrease blood flow to placenta and this can causes placental insufficiency, finally results in PTB. Presence of vaginal discharge during pregnancy increased the chance of having PTB. This is alike with other study [37].
Women who were already on HAART preconception had higher probability to have PTB. This is in line with other studies [93,94]. HIV positive women had more probability to give PTB than HIV negative women. This is in agreement with study done in South Africa [95]. Presence of periodontal disease was significantly associated with PTB as depicted systematic reviews of studies. Study done in Egypt showed similar result [11]. This may be due to periodontal can results in an increase of pro-inflammatory molecules that can directly or indirectly lead to uterine contractions and cervical dilatation.
The strength of this study is that this review seems to be the first done in East Africa, indicating the various determinants of PTB from 58 studies. This study has the following limitations. The search strategy was limited to studies published in English language, this can cause reporting bias. Data were not found for 10 East African countries, this can cause representativeness problems. Thus, further studies on determinants of PTB should be conducted in all East African countries by using standard WHO definition of PTB.

Conclusions
There are many determinants of PTB in East Africa. The determinants can be categorized into socio economic and demographic factors, RH, obstetric complications, medical condition, and behavior related factors. This review could provide policy makers, clinicians, and program officers to design intervention on preventing the occurrence of PTB.