Study design and search strategy

A systematic review was conducted, and the project was registered in PROSPERO (ID: CRD42021286725). The research question involved a population of pregnant adolescent women in Latin America and the Caribbean, with the exposure being adolescent pregnancy (≤19 years), the comparison being pregnant adult women (20-35 years), and the outcome being the risk of specific maternal complications (preeclampsia, eclampsia, puerperal hemorrhage, puerperal sepsis).

The search strategy was carried out by an experienced medical research librarian. Searches were conducted in the following databases: PubMed, SCOPUS, Web Of Science, EMBase, LILACS, and SciELO. The search strategy was broad, without restrictions by study type or year, except in PubMed and SCOPUS where it was limited to studies from the past 10 years. The detailed search strategy can be found in the Supplementary Material. Additionally, the search was limited to studies written in English, Spanish, and Portuguese.

Eligibility criterio

Articles were included if they contained numerical data from Latin American or Caribbean countries addressing the research question. Articles published from 2012 onwards were included, specifically case-control and cohort studies with a group of pregnant adolescents (≤19 years) and a control group of pregnant adults (20-35 years) or groups within these age ranges.

Case reports, literature reviews, editorials, conference abstracts, experimental, and qualitative studies were not eligible for inclusion. Studies that did not present separate findings for Latin America and/or the Caribbean, those without full-text availability, and those focused on patients with specific diseases were excluded.

Selection process

The search strategy for all the aforementioned databases was downloaded into EndNote 20 software to remove duplicate articles. Subsequently, the articles were exported to the online software Rayyan (https://www.rayyan.ai/) where two researchers (MAOM and MAPCA) independently reviewed the titles and abstracts. All included articles proceeded to the full-text evaluation phase by the same two authors (MAOM and MAPCA). In cases of disagreement, a third reviewer (VERC) was consulted. Finally, the references of the included articles were reviewed to identify any missing articles.

Data extraction, risk of bias assessment, and qualitative análisis

Information from the selected articles in the final phase was independently collected by two authors (MAOM and MAPCA) into a data collection table created in Microsoft Excel for Mac version 16.6. The table included the following domains: author, year of publication, country of execution, study design, data collection period, total number of participants in the adolescent group and their age range, total number of participants in the control group and their age range, inclusion and exclusion criteria, and outcomes (preeclampsia, eclampsia, puerperal hemorrhage, and puerperal sepsis).

The Newcastle-Ottawa Scale (NOS) was used to assess the risk of bias for each included article, rating the quality of the study as strong, moderate, or low. The NOS consists of 8 items divided into 3 risk of bias domains: selection of study groups, comparability of groups, and verification of exposure and outcome. Each item was awarded 1 point, except for the comparability item, which could receive 2 points if warranted, with a maximum score of 9 points. The NOS considers a score of ≥7 as low risk of bias, 4-6 as moderate risk, and <4 as high risk of bias⁹. This assessment was independently conducted by two authors (MAOM and MAPCA) and compared and discussed to reach a consensus. To illustrate the quality assessment of the studies, risk of bias graphs were created using Review Manager version 5.4 software.

For the qualitative analysis, a narrative evaluation of all the included articles was conducted to compare their methodological similarities and differences. The analysis included the following data: author, year of publication, country of execution, study design, data collection period, total number of participants in the adolescent group and their age range, total number of participants in the adult group and their age range, inclusion and exclusion criteria, and outcomes. All these data were compiled into a Microsoft Excel for Mac version 16.60 table.

Data collection flowchart

All evaluated and reviewed articles were outlined using a PRISMA flow diagram, illustrating the selection process and the total number of included studies, as well as those that were excluded¹⁰.

Summary measure and synthesis method

The summary measure used was the Odds Ratio (OR). The OR with a 95% confidence interval was extracted from each study. Statistical heterogeneity was evaluated using the I² statistic. A meta-analysis was conducted comparing the raw ORs using a random-effects model with inverse variance. The meta-analysis results were presented using Forest plot. The meta-analysis was developed using Review Manager 5.4 software.

Publication bias

The study included fewer than 10 articles, so a funnel plot was not performed to assess publication bias.

Ethical considerations

This study received ethical approval from the Research Ethics Committee of the Faculty of Human Medicine at Universidad Ricardo Palma - Lima, Peru on 10/21/2021 (code PG-50-021). Additionally, it has been drafted in accordance with the "PRISMA 2020 Statement for Systematic Reviews and Meta-Analyses"¹¹.

Selection of results

The search strategy identified 1,501 potential studies, of which only 4 were included. No articles meeting the inclusion criteria were found upon reviewing the references. After reviewing the titles and abstracts, 845 studies were excluded, and 10 studies were excluded after reading the full articles. One article met the inclusion criteria but was a duplicate with the title in English¹². Additionally, another article could have been included, but it was incomplete, missing tables with information on the variables of interest (¹³). The exclusions at each phase with their explanations are described in Figure 1.

Figure 1 PRISMA 2020 flow diagram of the literature review 

Characteristics of the studies

Of the total included studies, two (50%) were conducted in Peru^14,15, one (25%) in Mexico¹⁶, and one (25%) in Brazil¹⁷. Additionally, two (50%) of these studies used medical records of the participants as their source of information^14,15, one (25%) utilized data from a demographic survey followed by personal interviews¹⁷. One study (25%) did not specify how the information was obtained; however, it was the only one to specify the exclusion criteria it used¹⁶. The characteristics of the included studies are detailed in Table 1.

Risk of bias assessment

The risk of bias for the 4 included articles was evaluated using the NOS for case-control studies. The data were recorded in Review Manager 5.4 software, which was modified according to the NOS with 3 sections: selection, comparability, and outcome (Table 2).

The category with the highest rating was selection (Figures 2a and 2b). None of the studies used the checklist to describe their studies.

Table 2. Risk of bias assessment using the NOS

Figure 2 Risk of Bias. A: Risk of bias for each included study. B: Summary of risk of bias for the included studies. 

Preeclampsia

Two of the included studies have results on preeclampsia. Flores-Mendoza et al. divided preeclampsia into two groups: with severe signs and without severe signs¹⁶; for data processing in the present study, the variable used was preeclampsia without severe signs. On the other hand, Okumura et al. based their diagnosis of preeclampsia on the ICD-10 (International Classification of Diseases) code records available for each patient¹⁵. The ORs of both studies contain the unit: 0.76 (95% CI 0.53 - 1.07) and 1.04 (95% CI 0.98 - 1.10) respectively, indicating no difference between the study group and the control group regarding preeclampsia. Despite not all studies having the same definition of preeclampsia, a meta-analysis was conducted. The summary OR calculation yielded 0.93 (95% CI 0.69 - 1.25) with high heterogeneity (I² = 67%) (Figure 3 A ).

Eclampsia

Three of the included studies reported results on eclampsia. Oliveira Jr. et al. defined eclampsia as “seizures during pregnancy, childbirth, and postpartum,” whereas Okumura et al. relied on ICD-10 code records available for each patient to diagnose eclampsia¹⁵. The third study did not specify the definition used for eclampsia. Flores-Mendoza et al. reported a high risk of eclampsia in adolescents compared to adults (OR 4.05, 95% CI 2.57 - 6.38)¹⁶, similar to Okumura et al. (OR 1.93, 95% CI 1.50 - 2.49)¹⁵. However, Oliveira Jr. et al. found no difference between the groups¹⁷. Despite the varying definitions of eclampsia, a meta-analysis was conducted. The summary OR calculation yielded 2.43 (95% CI 1.29 - 4.58), which was statistically significant, but with high heterogeneity (I² = 76%) (Figure 3 B ).

Puerperal hemorrhage

Three of the included studies have results on puerperal hemorrhage. Oliveira Jr. et al. defined puerperal hemorrhage as “heavy bleeding that soaks clothes in the first 3 days postpartum”¹⁷. Okumura et al., on the other hand, based their diagnosis of puerperal hemorrhage on ICD-10 code records available for each patient¹⁵.

Bendezu et al. did not define the variable and analyzed it according to the clinical records¹⁴. The latter study reported no significant difference between the study and control groups regarding puerperal hemorrhage (OR 1.32, 95% CI 0.69 - 2.54), similar to Oliveira Jr. et al. (OR 0.90, 95% CI 0.69 - 1.18). However, Okumura et al. showed statistical significance (OR 0.82, 95% CI 0.70 - 0.95)¹⁵. Despite the varying definitions of puerperal hemorrhage, a meta-analysis was conducted. The summary OR calculation yielded 0.86 (95% CI 0.74 - 0.99), which was not statistically significant and presented low heterogeneity (I² = 10%) (Figure 3.C).

Figure 3 Forest plot.  

Puerperal sepsis

Only one study included puerperal sepsis in its analysis. Bendezu et al. reported one case of puerperal sepsis in pregnant adolescents compared to no cases in pregnant adults, with an OR of 3.02 (95% CI 0.12 - 74.56), which was not statistically significant.

Synthesis of Results

All studies included pregnant adolescents aged ≤19 years and a control group of pregnant adults aged 20 years or older. Not all studies included all variables of interest or their definitions. Three of the included studies had a moderate risk of bias, and one had a low risk. The included studies originated from Latin American countries due to the limited scientific evidence from the Caribbean.

No risk of preeclampsia was found in pregnant adolescents (OR = 0.93, 95% CI 0.69 - 1.25) nor for puerperal hemorrhage (OR = 0.86, 95% CI 0.74 - 0.99). However, an increased risk of eclampsia (OR = 2.43, 95% CI 1.29 - 4.58) was found in pregnant adolescents, with high heterogeneity among studies (I² = 76%).