The prevalence of pregnancy among adolescent girls and young women across the Southern African development community economic hub: A systematic review and meta-analysis

Background: Despite the high rate of HIV infections, there is still high rate of early unprotected sex, unintended pregnancy, and unsafe abortions especially among unmarried adolescent girls and young women (AGYW) 10-24 years of age in sub Saharan Africa. AGYW face challenges in accessing health care, contraception needs, and power to negotiate safer sex. This study aimed to estimate the rate of pregnancy among AGYW aged 10-24, 10-19 and 15-19 years in the Southern African Development Community (SADC) economic region. Methods: A systematic review and meta-analysis was used to describe the prevalence of pregnancy among AGYW in 15 SADC member countries between January 2007 and December2017. The articles were extracted from PubMed/MEDLINE, African Index Medicus, and other reports. They were screened and reviewed according to PRISMA methodology to fulfil study eligibility criteria. Results: The overall regional weighted pregnancy prevalence among AGYW 10-24 years of age was 25% (95% CI: 21% to 29%). Furthermore, sub-population 10-19 years was 22% (95% CI:19% to 26%) while 15-19 years was 24% (18% to 30%). There was a significant heterogeneity detected between the studies (I=99.78%, P < 0.001), even within individual countries. Conclusion: The findings revealed a high pregnancy rate among AGYW in the SADC region. This prompts the need to explore innovative research and programs expanding and improving sexual and reproductive health communication to reduce risk and exposure of adolescents to early planned, unplanned and unwanted pregnancies, SRHR challenges, access to care, HIV/STIs, as well as other risk strategies.


Types of studies and participants
This study focused only on AGYW pregnancy, looking for studies reporting "ever pregnant" (percentage of AGYW aged 10-24, 10-19 and 15-19 who reported ever pregnant) using a collection of search terms including "adolescent pregnancy, youth pregnancy, young women pregnancy, teen pregnancy" in SADC region.
The type of studies included in this systematic review were cross sectional, baseline data from cohort or randomized controlled trial (RCT) study designs, and from national Demographic and Health Survey (DHS) studies published as full papers, conference proceedings/ abstracts, policy or reports. The eligibility screening criteria for the studies were: (1) population: all "ever pregnant" studies describing AGYW, aged between 10-24 years; (2) setting: AGYW studies from the 15 SADC economic hub countries; (3) Studies having the number of ever pregnant AGYW (n) or ever pregnant proportion (%) and 95% confidence interval (CI) of ever pregnant in addition to the total AGYW population (N). The time frame for the review was restricted from 2007 to 2017. Studies describing general population rate of pregnancy were ineligible for the study as well as studies without clear pregnancy rate among AGYW.

Search methods for identification of studies Electronic searches
We carried out a comprehensive literature search of quantitative published studies reporting on pregnancy rate among AGYW according to Moher et al. 41 We limited studies to those published in English/French dating from 1 January 2007 to December 2017. We searched the following electronic databases: PubMed/MEDLINE, African Index Medicus and other African Journal online, EMBASE, Web of Science, Google Scholar, and Cochrane, using the following keywords: "adolescent pregnancy", "youth pregnancy", and "young women pregnancy" "teen pregnancy". Similarly, we searched using the following French key words: "grossesse et adolescents", "grossesse et jeunes'' , "grossesse et jeunes femmes". For example the Pubmed search English strategy used was "("Young"[Journal] OR "young"[All Fields]) AND ("women"[MeSH Terms] OR "women"[All Fields]) AND ("pregnancy"[MeSH Terms] OR "pregnancy"[All ])" Fields]) OR "country specific" (tw))". The search strategy was modified to include other online electronic databases such as BMC, Science Direct and PLoS One. Some of these studies were published as abstracts, conference posters and conference proceedings. Other data sources were national surveillance systems reporting on the rate of pregnancy among AGYW in some SADC countries. We embraced these sources as there were no specific pooled factor ratio describing the rate of pregnancy among AGYW between the ages of 10-24; 10-19 and 15-19 years in the 15 member states of the SADC Economic hub -a region with the highest unabated HIV reduction when compared to other region of the world. 18

Data collection and analysis
Selection of studies Two independent review authors TM and SN extracted, screened the title and abstract results from the search online databases and applied pre-piloted checklist eligibility criteria to identify eligible studies. The articles were screened for duplicates, multiple publications, and other irrelevant studies under guidance of CY.

Data extraction and management
Using Microsoft Excel designed data extraction spreadsheet, TM and SN reviewed the extracted "ever pregnant" data among AGYW aged between 10-24 years on adolescent pregnancy", "youth pregnancy", and "young women pregnancy" "teen pregnancy". The major data fields extracted were: country, age group and years, study design and recruitment methods, sample size, number pregnant, sample population as shown in Table 1. 6,24,34,38, The included age sub-groupings were: 10-24 years (25 studies), 10-19 years (22 studies) and 15-19 years (12 studies). Only Somba et al 42 had restricted data on subgrouping 19-23 years age (youth group). This was further reviewed by CY and AM for consistency. The rate of pregnancy from each study was extracted, together with the total number of participants enrolled, and the total number ever pregnant.

Statistical methods
The AGYW pregnancy was defined as "ever been pregnant within the ages 10-24 years'' . We used the prevalence of pregnancy to measure the rate of pregnancy. We calculated prevalence for each included study by dividing the number of AGYW ever pregnant (n) by the total number of AGYW in the study sample and expressed it as percent. 63,78 The sampling distribution for the prevalence statistic was assumed to have a normal distribution since the sample sizes were large enough to assume Central Limit Theorem. 63,78 Using STATA 13.1 (StataCorp, Texas, USA), the prevalence of pregnancy from the different studies were pooled in a random effects meta-analysis since we anticipated heterogeneity owing to the studies in different countries and settings. We applied the I 2 test statistic which estimates the percentage of variation that is due to heterogeneity rather than the chance occurrences, where values exceeding 50% indicate significant heterogeneity.
We also applied the chi-square test to infer the extent of heterogeneity. 39 In addition, the Egger's regression test was used to estimate the study publication bias using metabias command. 79 Results were displayed using forest plots. We investigated sources of heterogeneity through subgroup analysis with respect to the country from which the study was done and with respect to whether the study was from DHS.

Study search results and characteristics of included studies
We identified a total of 7627 citations (7620 from electronic and 7 from other sources) reporting AGYW pregnancy from 2007 to 2017. We removed 7444 citations which were either not relevant or from non-SADC countries or had not reported quantitative data for the number of ever-pregnant AGYW. From the remaining 184 citations, 144 were out of the study age range (study age eligibility), duplicates and irrelevant articles. The remaining 40 articles were further subjected to review and 14 articles were non eligible due to study design (case control) or not reporting on study denominator (sample size) ( Figure 1). Of the 7627 citations, only 25 studies were found eligible for age group 10-24 years of age, 6 Exploratory studies and DHS that reported only point prevalence without denominators or confidence intervals were also excluded. Qualitative studies were also excluded. Similarly, the non-experimental study from Zimbabwe had no denominator and was excluded. 75 The following studies had no specific sample sizes: reporting pregnancy among adolescent in Botswana 76 the Angola MIHS (2015-2016), 43 and data from Zambia. 61 The Mauritius teenage pregnancies of approximately 2000 cases per year arising from unsafe abortions 77 was also excluded for the same reason. All the case control studies were ineligible and were excluded from the study. 59,61 The female sex workers data from DRC was included because it met the inclusion criteria, having a sample size, prevalence and the population was 12-21 years of age 45 (Table 1) despite the limitations.
Of the 15 countries in SADC economic hub: 13 countries had 25 eligible studies of pregnancy among AYGW aged 10-24 years freely available online-internet. The described 25 eligible studies were used to pool the national and regional pregnancy rate among AGYW. Of the 25 studies, 81 692 AGYW were enrolled, and of those 14 089 reported ever pregnant translating to a crude prevalence of 17.3% (95% CI, 12% to 18%).

Discussion
High AGYW pregnancy and HIV/STIs new infections is of concern in the SADC region. This study was conducted to estimate the prevalence of pregnancy among AGYW aged 10-24, 10-19 and 15-19 years in the SADC region. The pregnancy evaluation was to enhance interventions Removal of articles without quantitative measurement (Sample size and prevalence) and non-SADC Countries: n=7443 Records that meet the eligibility criteria of sample size/age appropriateness. Reviewed by two independent reviewers: n=40 Removal of articles without age appropriateness, duplicates articles, and other irrelevant articles: n= 144  83 Also, mental health is one of the factors limiting AGYW attending PMTCT. 84 According to a study by Nyamukoho et al, 84 HIV positive AGYW attending PMTCT were 3.2 times more likely to suffer from depression, compared to older women. Therefore, detailed age appropriate epidemiological data is needed to provide an accurate estimate of pregnancy rate among AGYW which can be used to enhance interventions towards mental health interventions.
Our results showed a varied national pregnancy rate among AGYW in SADC countries. This result was similar to other low-income countries such as Latin America and the Caribbean where adolescent fertility remains high. 37 The same report 37 documents measures to reduce AGYW pregnancy, such as joint regional commitment task force, sub regional and national action plans targeted age appropriate disaggregation by engaging and empowering youth to contribute and drive the design, implementation, and monitoring of interventions. Our results show the pooled pregnancy rate for AGYW aged 15-19 years at 24% and national variation from 11% in South Africa to 38% in Zambia. The national variation for Latin America and the Caribbean was between 11.6% in Uruguay to 30.7% in Panama. 37 To reduce AGYW pregnancy by 20%, UNFPA 85 has instigated the leveraging of proven and effective interventions such as access to effective contraceptives, comprehensive sexual education (CSE), age-appropriate counselling, access to available, equitable and acceptable information, building and creating gender equality, adolescent SRHR education at targeted settings. In addition, reports by Yakubu et al 13 and UNICEF 80 also highlight awareness and knowledge regarding sexual education, health and family planning, and parenting skill development that include, contraception methods, and SRH misconceptions. Although, contraception misconceptions are likely to differ in some settings due to differences between modern and traditional attitudes towards contraceptive methods. [86][87][88][89] In some settings, the majority of AGYW do not make  The data from Mauritius and Botswana will be updated whenever the data is available use of contraception methods. 80 For example, in settings such as India, IIPS 90 conducted a household survey in 2008 among AGYW aged 15-24 years and found that more than half of the AGYW population had never received any formal education about sex or family planning. In the same study 77% of the girls had no formal contraception education. 90 Sexual and gender-based violence (SGBV) and gender inequalities are some of the root causes of the high pregnancy rate in the region. 34 Based on these limitations, the SADC Maputo Protocol of 2016 was held to harmonize national abortion and pregnancy laws and policies in Southern and Eastern Africa. 91 The meeting proceedings advocated for more research and interventions by expanding women comprehensive sexual and reproductive education and health services. With respect to AGYW accessing health care for contraceptives and SRHR information, many SADC countries have rolled-out adolescent friendly youth services including SRHR information. 75,92 However, AGYW are still confronted with negative and stigmatizing attitudes of health care providers when seeking health care. 8,13,93 Due to these challenges and according to report by Panday, 94 AGYW delay access to health care services during pregnancy 94,95 resulting in five times SRH consequences among adolescents girls less than 15 years of age when compared to older women 96 The pooled estimate of pregnancy rate among AGYW in South Africa and Zimbabwe was estimated at 16% and 22%. This was slightly less than the 30% reported by Flanagan et al 97 and Willan et al 98 among AGYW who had 'ever been pregnant' . This number decreased but still high. 97,98 According to reports by Lillian et al, 99 high AGYW pregnancy in Namibia is influenced by educational background, socio-economic status, and cultural beliefs. In response to these challenges the Namibian government has geared intervention programs and policy towards youth CSE regardless of the socio-economic and/or cultural contexts and status. In Mozambique, Decree 39/ GM/2003 has been initiated where pregnant schoolgirls are initiated on extra mural studies in order to complete their education. 100 This review showed that DRC has the highest pregnancy rate, estimated at 62%: this correlates with the country's high birth rate. DRC is the highest populated country in SADC region and the 3rd populated country in sub Saharan Africa with an estimated population of over 80 million people and a fertility rate of 6.1% per woman. 101 Swaziland has the highest rate of HIV among women of 15-49 (35.1%) and adolescents (16.7%) globally 102 and the pregnancy rate was estimated at 22% among adolescents aged 10-19 years.
AGYW who desire to be pregnant are less likely to practice safer sex. This was highlighted by findings from the Mozambique DHS showing AGYW who want to get pregnant were less likely to use condoms and other prevention methods with non-marital partners than those who want to delay childbearing. 103 Furthermore, according to reports by Neal et al, 104 approximately 2.5 million AGYW give birth before the age of 16 each year in low resource settings and around 50%-65% of them before the age of 15 especially in Chad, Guinea, Mali, and Mozambique. Due to the increased risks of engaging in unsafe sexual intercourse among unmarried AGYW: identifying and scaling up evidence-based programs for sexually active AGYW is critical. Positive cost-effective ways of addressing and reshaping AGYW CSE, access to SRHR, and use of pre exposure prophylaxis (PrEP). Pregnancy itself does not increase the risk of HIV/ STI acquisition, however unprotected sex, a singular incident of pregnancy acquisition plays a significant role of HIV/STI transmission. It should be noted that AGYW pregnancy has been a neglected research area despite the risks associated with HIV/AIDS, sexual abuse, infant and maternal mortality, school drop-out, and loss of self- According to a report by Higgins et al, 40 there is always potential biases in study reviews, including agreements and disagreements, applicability of evidence and quality of the evidence. 40,105 Only English and French were used as the language of study search. There was no relevant data from Botswana and Mauritius and the sex worker data from DRC skewed the outcome of 10-24 years of age. The studies were not also aggregated as urban and rural areas limiting setting generalizability. Due to publication bias, findings with favourable outcomes may not have been available especially those of Population Council, Guttmacher and WHO. Despite these limitations, the search generated real time generalizable effect size estimates. The majority of the cross-sectional studies included had adjusted key variables. In addition, adolescent pregnancy data between those aged 15-19 years were mostly from DHS ( Figure 5), increasing the power and generalization of the findings. The regression analysis showed that the combined constant was 0.900 and the P value at 0.900 indicating no publication bias for those 10-24 years of age.

Conclusion and Recommendation
The study revealed a high pregnancy rate among AGYW in the SADC region. The high pooled country specific and regional rate among AGYW has highlighted the need for informed SRHR policies and programmes to be tailored towards AGYW. This includes policy changes that will improve the collaboration between adolescents, health care providers, parents, and teachers to address adolescent sexuality education. In addition, increased research is needed to explore innovative ways to expand and improve sexuality communication and sexuality risk reduction strategies that expose adolescents to unsafe sexual practices, unwanted pregnancies, HIV/STIs and improve health seeking behaviour. The reduced focus on the pregnancy rate among AGYW affects the social, political and economic development of the region.

Acknowledgments
We acknowledge the postgraduate and intern students that assisted in the extraction of the data especially Tetelo Maakamedi (TM) our intern student who helped in the the data extraction and Dr. Alfred Musekiwa (AM) who opted and carefully reviewed the meta-statistics. The study had no funding support.

Competing interests
None declared.

Ethical approval
Since all data used in this systematic review have already been published or available in the public domain, there was no ethical clearance required for this review.
Authors' contributions CSY conceptualized, designed, reviewed the data, analysed and wrote the initial draft of the article. TM, SN extracted and reviewed the eligibility criteria. Differences or disagreements between the reviewing authors were resolved through a discussion meeting by a third party (CSY). CSY conducted and performed the metaanalysis statistics which was further reviewed by AM and advised on the general methods and structure. CSY, SN, NN, TM, AK and SM gave further interpretation and critical appraisal.

Disclaimer
The findings, interpretations, and conclusions expressed in this publication do not necessarily reflect the views of any regional or national governments but represent those of the researchers.