Epidemiology of antenatal depression in Africa: a systematic review and meta-analysis

Background Antenatal depression is a serious problem worldwide that has devastating consequences not only for the mother but also for the child and family. The pooled evidence regarding the prevalence and associated factors of antenatal depression is rare in Africa. Hence this review aimed to investigate the prevalence and associated factors of antenatal depression in Africa. Methods We searched CINHAL, MEDLINE, PsycINFO, Psychiatry online, PubMed, SCOPES, and Emcare databases for English written observational studies conducted in Africa from 2007 to 2018.Quality of studies was assessed using the Newcastle Ottawa Scale (NOS), and studies with good quality were included in the final review. Heterogeneity across studies was assessed using the I2 and Higgins test. Publication bias was checked using Funnel plot symmetry, and Egger’s regression test and adjustment was made by using Duval and Tweedie’s Trim and Fill analysis. A random effect Meta-analysis was employed to determine the pooled estimates with 95% confidence interval (CI). Stata 14 was used for analysis. The review protocol has been registered in PROSPERO number CRD42018106717. Result Of the 175 studies identified, 28 studies with an overall sample size of 17,938 were included. According to the random effect model following trim and fill analysis, the pooled prevalence of antenatal depression in Africa was 26.3% (95%CI: 22.2, 30.4%). Economic difficulties [POR = 1.87;95%CI:1.25,2.78,I2 = 88.1%], unfavorable marital condition [POR = 4.17;95% CI:1.75, 9.94, I2 = 81.2%], poor support from relatives [POR = 1.36;95% CI:1.18, 1.56, I2 = 78.0%], bad obstetric history [POR = 2.30;95% CI:1.81, 2.92), I2 = 81.7%], and history of mental health problem [POR = 2.97; 95% CI:1.74, 5.06, I2 = 92.0%]were the factors associated with antenatal depression. Conclusion The prevalence of antenatal depression is high in Africa, which showed that one in four pregnant women had depression. Pregnant mothers who had economic difficulties, bad obstetric history, poor support from relatives, previous mental health problems, and unfavorable marital conditions were at higher risk of antenatal depression. Therefore these factors should be considered while designing mental health care services for pregnant mothers.


Background
Depression is one of the types of mood disorders characterized by markedly decreased interest or pleasure in almost all activities, significant weight loss or gain, disturbed sleep, feeling of fatigue, loss of appetite, feeling of hopelessness, reduced self-esteem and confidence, diminished ability to think or concentrate, and recurrent thoughts of death [1,2]. According to the World Health Organization (WHO) 2017 estimate, 322 million people are living with depression, and29.9 million (9%) of these are living in Africa. Depressive disorders are ranked among the top five contributors to the global disease burden [3]. The prevalence of depression increased by 18.4% between 2005 and 2015 worldwide [4].
Antenatal depression is a non-psychotic depressive episode ranging from mild to severe symptoms that occur while the woman is pregnant [5,6]. Women are known to be at higher risk of mental disorders like depression than males [7]. The mental health of women of reproductive age is becoming a significant public health problem both in developing and developed countries, and depression is the most prevalent mental disorder during pregnancy [8,9]. A systematic review and meta-analysis conducted in developed countries showed the prevalence of depression to be 7. 4, 12.8, and 12% at the first, second, and third trimesters of pregnancy, respectively [10]. Another metaanalysis conducted worldwide also reported the prevalence of antenatal depression that ranges from 0.5 to 51% [11]. Moreover, a similar study from low-and middleincome countries also showed a mean weighted prevalence of common mental disorders during pregnancy to be 15.6% [12]. The magnitude of antenatal depression varies across different countries in Africa, and studies showed the prevalence to be between 8.3 and 78.2% [8,[13][14][15][16][17][18].
Depressive disorders during pregnancy may have devastating consequences not only for the mother but also for the child and family [19]. Antenatal depression is identified to be a risk factor for adverse obstetric and birth outcomes like fetal growth retardation, low Apgar score, preterm birth, low birth weight, and stillbirth [13,[20][21][22][23]. Antenatal depression is also associated with increased smoking, alcohol consumption, and unhealthy behaviors [24]. These factors, together with depression, may predispose the mothers to obstetric complications [19] such as preterm labor [25], preeclampsia, abruption placenta [26,27]. Furthermore, depression during pregnancy is also associated with postnatal depression [28], which negatively affects child development, motherinfant interaction, and the family at large [9].
Compared with women in developed countries, women in developing countries are more exposed to the risk factors for the development of antenatal depression, such as; younger age of mothers [18], low level of education, exposure to domestic violence [8] or relationship conflicts [18], history of obstetric complications, history of depression [15,29,30], unplanned pregnancy, lack of social support [27,31], and low economic status [30].
Despite variations in the magnitude and associated factors of antenatal depression across different countries of Africa, pooling the available evidence and reporting the extent of the problem in a more precise way might help policymakers to prioritize the problem more than ever. Therefore, the objective of the current review is to assess the epidemiology of antenatal depression in Africa.

Search strategy and selection criteria
The study conforms to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [32]. The reviewed articles were sourced from the following databases: MEDLINE (via Ovid), Psy-cINFO, CINAHL (EBSCO), Psychiatry Online, Emcare, PubMed, Scopus. Besides, google scholar, snowballing, and retrieving references from a list of eligible studies were employed. A search strategy was developed for each database by using a combination of free texts and controlled vocabularies (i.e., Mesh terms).
Example of the search strategy for PubMed: We included all observational (cross-sectional, casecontrol, prospective, and retrospective cohort) studies conducted in Africa, which conducted from 2007 up to 2018 and included antenatal depression and associated factors as a primary outcome. Additionally, studies included must have used a validated tool to screen depression, and they must be available in the English language. Studies that were reviewed, studies with a poor quality based on NOS [33],and studies that were conducted on high-risk population groups were excluded from the review. Articles were independently screened in two stages: firstly, the titles and abstracts were screened, and secondly, the full-text articles that met the eligibility criteria mentioned above were retrieved and screened further for possible inclusion by two reviewers (AFD and HFW). Where there was disagreement between the two reviewers, further discussion was made until a consensus reached.

Data extraction
The data extraction sheet was prepared to collect information on the name of the author, year of publication, country, study setting(i.e., population-based versus institution-based), study design, sample size, time of screening for the depression, the tool used to screen depression, and the prevalence estimates of antenatal depression. Data were extracted by two reviewers (AFD and AGB) from publications, and HFW crosschecked for accuracy.

Data quality and risk of bias assessment
The quality of evidence and risk of bias for studies (casecontrol and cohort) was assessed using the Newcastle-Ottawa Scale (NOS). Crossectional studies were also assessed using an adapted version of NOS. The criteria include 3 categories with a maximum score of 9 points. The first is the "selection" category, which accounts for a maximum of 4 points, the second is the "comparability" category, which accounts for a maximum of 2 points, and the third is "outcome," which accounts for a maximum of 3 points. Based on the composite score from these three categories, the studies were classified as good quality if the score is ≥7 points, fair quality 2 to 6 points, and poor quality ≤1 point [33]. Only studies with good quality were included in this review.

Data analysis
Meta-analysis was conducted to synthesize the pooled prevalence of antenatal depression and the odds ratio of the associated factors using the random effect model. To determine the extent of variation between the studies, we did a heterogeneity test using the Higgins method, which was quantified by I 2 values [34]. Publication bias was assessed by using the symmetry of the Funnel plot and statistically by Egger's regression test. Duval and Tweedie's Trim and fill analysis was done to correct for publication bias [34,35]. Sub-group analysis for the pooled prevalence was done, and sensitivity analysis was also carried out to detect any effect of outlier study that significantly affected the pooled estimate. Odds ratio  with a 95% confidence interval was used to assess the association between antenatal depression and associated factors. The analysis was done using Stata version 14 software [36].

Protocol registration
The review protocol has been registered in PROSPERO with protocol number CRD42018106717.

Search
We conducted an electronic literature search and identified 175 unique records of journal articles and 124 duplicates that were removed. After meticulous review of the titles and abstracts, we excluded 22 articles because of difference in population under study [37][38][39][40][41][42] and outcome of the study [43][44][45], full text not found and study area not known [46][47][48][49][50], reviews [5,19,[51][52][53] and conducted in restricted population [54,55]. We obtained full-text copies of 29 records for further review. Of those full-text articles, one article [56] was excluded because of poor quality. Finally, 28 articles with an overall sample size of 17, 938 were included in the narrative review and Meta-analysis (Fig. 1).
Besides, subgroup analysis was done based on the income of countries, time of depression measurement, study setting, study design, year of publication, and tool used to screen depression. Based on the time of screening for depression in the course of pregnancy, the highest pooled prevalence of antenatal depression was observed in the second trimester, 32.20% (95% CI: 26.13, 38.28%). On the other hand, a significantly lowest prevalence of was found from a single study with a screening time in the first trimester, 9.90% (95% CI: 8.60, 11.20%). Depending on the study setting, the pooled prevalence of antenatal depression for 19 health institution based studies was 26.77% (95%: 22.54, 30.99%).
Based on the study design, the pooled prevalence of antenatal depression for seven studies, which applied longitudinal study design was higher, at 28 Table 2).
Sensitivity analysis was completed after excluding the study with the highest prevalence [16], and it showed that omission of any of the incorporated studies did not change the pooled prevalence of antenatal depression (Fig. 5) (Figs. 6 and 7).

Discussion
The current review assessed the prevalence of antenatal depression and its associated factors in Africa. Our review showed the pooled prevalence of depression among African pregnant women was 26.3%, and it is significantly associated with economic difficulties, poor support from relatives, bad obstetric history (such as  previous pregnancy loss and complications), unfavorable marital condition, and history of mental health problems.
The result was consistent with a review conducted in low-and middle-income countries, which showed a pooled prevalence of 25.3% [74]. The pooled prevalence of antenatal depression from this review was found to be higher than other reviews done in Africa and low-and middle-income countries, which showed that the weighted mean prevalence of common mental health disorders to be 11.3 and 15.6%, respectively [12,75]. This might indicate that antenatal depression is increasing over time. Our estimate was also higher than another systematic review and meta-analysis done in developed countries, which showed the pooled prevalence at the first, second, and third trimesters to be 7.4, 12.8, and 12%, respectively [10]. The higher pooled prevalence in our review could be mothers in Africa are exposed to additional socio-economic problems and stressful life events than mothers from developed countries, which increases their risk of developing depression.
The odds of having antenatal depression among women who have economic difficulties was found to be 1.87. This result was consistent with other systematic reviews published in Ethiopia, low-and middle-income countries, and worldwide [5,12,76,77]. The finding was also supported by another large scale prospective cohort study conducted to assess the association between low socioeconomic status and antenatal depression [78]. Mothers in such economic difficulties would worry about the family and the coming babies' basic needs, and this can also be related to family food insecurity, which is highly associated with antenatal depression [31,79].
This review showed that unfavorable marital condition (explained as divorce, relationship difficulties, being single at the time of pregnancy, marital conflict) was significantly associated with increased risk of antenatal depression by4.17 times. A similar association was found from other systematic reviews done in Ethiopia and other low-and middle-income countries [5,12]. This might be because of a lack of support from husband at the time of pregnancy as those women who receive husbands' support during their pregnancy may be well empowered to deal with their pregnancy and their home responsibility. Another reason could be those women with unfavorable marital conditions might live alone, practice more loneliness, and low self-confidence that may predispose them to depression. This review also showed that there was an increased risk of antenatal depression among women who had poor support from relatives, which is supported by other previous reviews [12,76,80]. The possible reason could be because social support reduces stressful life events by providing informational, instrumental, and emotional support during pregnancy [80,81]. Perhaps, the objective evaluation of social support that women received during pregnancy may be challenging as it has been noticed that depressed women tend to feel less supported than they objectively are [82]. Women with a bad obstetric history were found to be at an increased risk of antenatal depression by an odds of 2.30 times. This result was similar to other reviews done in Ethiopia [5,77] and worldwide [80]. This is directly associated with the mother's fear of facing similar complications in her current pregnancy like the previous one.
Women with a history of mental health problems also had three-fold increased odds of developing antenatal depression in the current pregnancy as compared to those who did not report such history. This result was supported by other reviews done in Ethiopia [5,77] and worldwide [80]. Having a history of mental disorders, including depressive episodes, may indicate the mother's biological vulnerability to the disorder, which may indirectly cause pregnancy mood changes in the current pregnancy [83].
The strength of this review is that it included only high-quality studies that scored ≥ seven based on the NOS criteria, and this may increase the reliability and validity of the findings. Besides, all available observational studies in Africa that satisfied our inclusion criteria were included, and this would increase the review generalizability. However, the use of screening instruments in prevalence studies relies on symptom identification rather than diagnosis, and this affects the validity of the review as screening tools over or underestimate the actual effect estimate. Though a subgroup and random-effect meta-analysis were conducted to minimize the effect of heterogeneity, the prevalence estimates might still be affected by the inherent heterogeneity of included studies, which may be due to the difference in the study area, methodology, study period and the type of screening tool used in the studies. Therefore, policymakers should make the interpretations of these results with caution by considering this inherent heterogeneity. Further standardization of screening tool that would be used in different countries might help to minimize the observed heterogeneity. Moreover, excluding works of literature that are not published in the English language may create selection bias. The clinical importance of this review is that it clearly showed that depression during pregnancy is highly prevalent and needs to be targeted for screening and early treatment.

Conclusion
We found that one in four mothers had antenatal depression, and it is independently associated with a mother's economic difficulties, poor support from relatives, unfavorable marital condition, previous history of