A systematic review of the prevalence of selected sexually transmitted infections in young people in Latin America

ABSTRACT Objective. To estimate the burden of Chlamydia trachomatis (CT), Neisseria gonorrhoeae (NG), Treponema pallidum (TP), and human papillomavirus (HPV) infections among people aged 10 to 25 in Latin America and the Caribbean. Methods. The MEDLINE, EMBASE, and LILACS databases were searched, as well as documents from regional organizations or national health Institutions. Population-based studies that reported prevalence or incidence of CT, NG, TP, and HPV detected through confirmatory tests in adolescents and young people were included. Two reviewers independently selected studies and extracted data. The quality of studies was assessed using the Newcastle–Ottawa Scale. Pooled estimators were calculated in cases where heterogeneity was <70%; when not feasible, prevalence ranges were reported. Results. Out of a total of 3 583 references, 15 prevalence studies complied with the inclusion criteria. Due to substantial heterogeneity (>70%), it was not possible to pool frequency estimators. Among the general population, the prevalence of CT infection ranged between 2.1% and 30.1% (9 studies, 5 670 participants); for NG, prevalence ranged between 0% and 2.9% (8 studies, 5 855 participants); for TP, prevalence varied between 0% and 0.7% (3 studies, 11 208 participants), and for HPV infection, prevalence ranged between 25.1% and 55.6% (8 studies, 3 831 participants). Conclusions. Reliable, population-based data on sexually transmitted infections (STIs) in adolescents and youth in Latin America and the Caribbean are limited. Additional studies are needed to better understand the burden of STIs in this population. However, given the substantial prevalence of STIs detected, countries need public health policies for prevention, early diagnosis, and treatment of STIs in young people.

of morbidity and mortality due to their contribution to complications such as pelvic inflammatory disease (3), infertility, ectopic pregnancy, miscarriage, and fetal and infant death (4,5). Studies have demonstrated a strong association between HIV infection and ulcerative and non-ulcerative STIs (6). While reliable data are limited, indications are that young people carry a substantial proportion of the STI burden. It is estimated that close to half of STIs are acquired by young people aged 10-24 years in the United States of America (7) and that one in four adolescents is infected with Chlamydia or human papillomavirus (HPV) (8). Behavioral aspects related to the initiation of sexual activity, the number of sexual partners, sexual relations with casual partners, and inconsistent condom use are among the risk factors of this population, in addition to the use of psychoactive substances (9,10). In Latin America and the Caribbean (LAC), an estimated 11% of males and females engage in sexual intercourse before the age of 15 (11). Subgroups identified as having a higher frequency of STIs are sex workers, men who have sex with men, persons living with HIV, and people in correctional facilities (12).
There are a limited number of systematic reviews that estimate the burden of STIs globally or in specific regions like sub-Saharan Africa or LAC, and information concerning STIs in adolescents and youth is scarce, both globally and in LAC (1,2,(13)(14)(15). Limited understanding of the burden of STIs in young people, stigma, and the systemic barriers young people face to access quality health services, contribute to an inadequate response for the prevention, early diagnosis, and treatment of STIs in this population group. Information on the incidence and prevalence of STIs among young people is highly relevant to better understand the magnitude of the problem and to guide national policies for prevention and control specifically targeting this age group. Thus, the objective of this systematic review is to estimate the burden of CT, NG, TP, and HPV infections in people aged 10 to 25 years in the LAC region.

MATERIALS AND METHODS
A systematic literature review was undertaken based on the protocol published in PROSPERO under the code CRD42020155877.

Eligibility criteria of studies
We searched population-based observational studies or data from mandatory STI reporting systems to measure the incidence or prevalence of one of the following STIs: CT, NG, TP, or HPV. We excluded studies that did not report frequency of CT, NG, TP, and HPV separately for each condition, those that did not report the frequency of the respective STI in populations between 10 and 25 years of age, and studies that did not indicate explicitly the denominator (population at risk) from which the reported relative frequencies of CT, NG, TP, and HPV infection were obtained. Population-based studies were defined as those with a random sampling of clusters (e.g., schools or primary care centers) or participants. The age group criterion was that of a population between 10 and 25 years from LAC countries. The outcome considered for analysis was the incidence or prevalence of CT, NG, syphilis, or HPV, diagnosed by a gold-standard test for each condition. For CT and NG diagnosis, this referred to culture or amplification of nucleic acids; for HPV infection, DNA detection; and for syphilis, both treponemal and nontreponemal tests had to be positive (independently of titer) or positive darkfield for TP.

Search of studies
A systematic search was conducted in MEDLINE (Ovid), EMBASE (embase.com), and LILACS (iAHx interface -BIREME) using keywords related with "adolescents," "sexually transmitted infections," and "prevalence or incidence." The search strategies were designed by the Cochrane Sexually Transmitted Infections Group (supplementary material available upon request). Complementary searches were done in the following search engines and websites: Joanna Briggs Institute Library, Google, webpages of the ministries of health of the LAC countries, OpenGrey Repository, Pan American Health Organization, World Health Organization, and UNICEF. Duplicate records were removed using EndNote (Clarivate). The last search was undertaken in February 2021.

Selection of studies and data extraction and management
Eligible studies were selected independently by two researchers based on title and abstract. Subsequently, two researchers independently verified the eligibility criteria and extracted the information in the available full texts. In case of discrepancies, researchers met to resolve disagreements. Information was collected in an Excel spreadsheet designed for the purpose. Information regarding each study was collected: design, setting, population, age group, sampling, type of STI assessed, case definition, laboratory method of detection, confirmation used in the diagnosis of infection, type of frequency estimator used, and frequency estimated by type of infection.

Risk of bias assessment of the included studies
All studies were assessed independently by two researchers applying the Newcastle-Ottawa Scale adapted for crosssectional studies (16,17). This tool evaluates selection bias, performance bias, bias caused by inadequate control of confounding, and selective reporting bias.

Measurement of the effect of the interventions and unit of analysis
To establish the frequency estimators, we reported the percentages of prevalence defined by the included studies. The unit of analysis was individuals.

Data analysis
To determine the possibility of generating pooled frequency estimators, the following steps were undertaken: (a) ascertainment of compliance with criteria for the development of meta-analysis of proportions (18) -three studies per infection were established as the minimum required to consider performing a quantitative analysis; (b) assessment of heterogeneity, undertaken qualitatively by evaluating possible sources of heterogeneity related to the condition and the population (type of infection, type of population, special populations, and coinfection with HIV) and epidemiological aspects such as the design and type of estimator reported. The quantitative assessment of heterogeneity was done by estimating the I 2 statistic (19), obtained by direct method, and the double arcsine square root transformation was compared to get the best adjustment (20); frequency pooled estimators were calculated by type of STI when the I 2 value was lower than 70% through a randomeffects model. In cases where this was not possible, prevalence ranges were reported.

Summary of changes to the protocol (PROSPERO: CRD42020155877)
The age groups were changed from 10-19 and 20-24 to 10-20 and 21-25 according to the findings of the included studies. Studies that were focused on high-risk populations for HIV/ STI or based on diverse populations were excluded. Disagreements were resolved during the selection of studies and data extraction not by a third evaluator but by consensus. The unit of analysis included only individuals, as episodic prevalence was not found. Adjustments were made to the risk-of-bias assessment tool based on studies with a descriptive objective (measurement of prevalence). Studies were not excluded due to missing data. Sensitivity analysis was not performed due to the quality of the studies.

Characteristics of the included studies
Regarding gender, 12 studies were reported among women, and three were conducted exclusively in pregnant or postpartum women (24,30,32). One study included pregnant and non-pregnant patients; however, because it did not specify whether random sampling was used in pregnant women, only the non-pregnant women group was included in the analysis (25). Three studies reported male and female populations (29,33,34). All 15 studies registered approval by an ethics committee. Two studies did not declare the sources of funding (21,32). Table 1 presents the characteristics of the included studies. Information about detailed search strategies and the list of excluded studies are available from the authors upon request.

Risk of bias
Selection bias. Thirteen studies used a random sampling design (21)(22)(23)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34), and two applied recruitment strategies that ensured that the entire population of interest was included (24,35). Eight studies reported sufficient information to justify   All: Non-pregnant women aged 18 to 49 years without contraindication for pap smear examination. Group 1 -sex worker population: sex workers who had been visited as part of a local program to prevent STIs. Group 2 -general population: women who attended selected health centers regardless of the reason for the visit. the sample size (22,24,29,30,(32)(33)(34)(35). Two studies provided adequate justification for the percentage of non-responders and established comparability of the study population with the non-included population (28,35).

Measurement bias.
All included studies reported that confirmatory tests were processed in reference centers, thus all studies were classified as independent measurements of the outcome of interest.  Note: * The study fulfilled the assessed criteria; ** The study measured the outcome through independent blind assessment or by record linkage. Source: Prepared by the authors based on published data.
Syphilis. The prevalence range of syphilis infection in women was reported in three studies (11 208 participants; quality range was between 4/8 and 6/8; I 2 85.14%) (22,24,33), which used the FTA-ABS treponemal test as screening and confirmatory test, one study used VDRL (22), another RPR (33), and another ELISA (24). The prevalence varies between 0% and 0.7% ( Figure  3), with the frequency of 0.7% corresponding to the prevalence of infection in pregnant women (24). No information was found on prevalence in men separately. Regarding the distribution of prevalence by age group, the three studies provided information on prevalence of infection in the population aged <20 years, and the range was between 0% and 0.7% (3 studies; 10 858 participants; I 2 85.8%), but only one study reported the frequency of infection in the population 21 to 25 years, which was 0.3% (1 study; 350 participants) (22).

Human papillomavirus.
Eight studies on HPV infection were identified in women (3 831 subjects; quality range 3/8 to 6/8; I 2 94.6%) (21-23, 26, 27, 31, 33, 35), and all of them used PCR for detection. The prevalence of HPV infection among women ranged between 25.1% and 55.6% (27,31). Three studies reported infection frequencies in women from the general population between 10 and 20 years (643 women), finding a prevalence between 27.9% and 42.9% (33,35). One study reported an infection prevalence of 35.6% among women in the general population aged 21-25 (35); four studies did not provide information that allowed including data in subgroups by age (21,26,27,31).  reported prevalence of CT, NG, TP, or HPV infections in populations aged 10 to 25 years in LAC were identified. We found substantial heterogeneity between results, which did not allow us to pool the prevalence estimators by sex, age group, or country. In adolescent and young women, CT infection ranged from 2.1% to 30.1%, NG varied from 0% to 2.9%, syphilis from 0% to 0.7%, and HPV from 25.1% to 55.6%.

Completeness and applicability
The study searches included the main databases of medical literature in LAC, with emphasis on the Latin American and Caribbean Health Sciences Literature (LILACS) electronic library, as well as reports from the epidemiological surveillance of communicable infections units of the ministries of health of the countries of the region. Consequently, it is unlikely that we have missed population-based studies that include the target population of this systematic review. Regarding applicability, the samples were obtained from population-based random   Finally, with respect to HPV prevalence, our data are similar to those reported by Brunni et al. for LAC, which are between 15% and 35% in a systematic review on HPV infection in women (44). However, this study does not provide data by age group. In Mexico, Smith et al. reported a prevalence of 14% in men under 24 years of age, but most studies are conducted on special populations (45).

Quality of evidence
We consider that the body of evidence included in this review is qualified as low due to limitations from the moderate risk of bias and substantial heterogeneity in the informed prevalence. Based on this appraisal, new studies should be undertaken to confirm these results.

Limitations and strengths of the review
The strength of this review is the selection of studies that guaranteed an adequate representation of the population of interest and the use of gold-standard diagnostic tests to confirm cases, in addition to the use of a scale to assess the methodological quality of the studies. The main limitation was a high heterogeneity in the results of the studies, which limited the generation of pooled estimators. Also, due to the absence of frequencies in at least 10 studies, it was not possible to perform an objective analysis of publication bias through funnel plots.

Conclusion
Reliable, population-based data on STIs in adolescents and youth in LAC are limited. Additional studies are needed to better understand the burden of STIs in this age group, especially in males and females from 10 to 14 years of age. However, given the substantial prevalence of STIs found in this review, countries need to formulate and implement national public health policies for prevention, early diagnosis, and treatment of young people, considering the early sexual initiation of adolescents in this setting.
Author contributions. All authors conceived the original idea, prepared the protocol, prepared the final document, interpreted the results, and reviewed the manuscript. MTV and HGD collected the data, analyzed the data, and wrote the paper. All authors reviewed and approved the final version.

Conflict of interest. None declared.
Financial support. This review was supported by the Pan American Health Organization. The sponsor did not influence in any way in the design, data collection, analysis, writing, or the decision to publish these results.
Disclaimer. Authors hold sole responsibility for the views expressed in the manuscript, which may not necessarily reflect the opinion or policy of the Revista Panamericana de Salud Pública/Pan American Journal of Public Health and/or those of the Pan American Health Organization. sampling; however, data could be inferred more to the young people aged 15-24 (considering the scarcity of specific information for adolescent people aged from 10 to 15).

Risk of bias of this systematic review
We followed the criteria for selection of studies according to the protocol; those were clear and non-ambiguous. Additionally, risk of bias in selection is considered low because searches were conducted in the main databases in Spanish, Portuguese, and English, the most frequently spoken languages in LAC. Regarding study appraisal, we used the Newcastle-Ottawa Scale modified for prevalence studies. We were not able to report data on STI incidence in young people because there were no studies that complied with the selection criteria.

Comparison with similar studies
Regarding the prevalence of CT, our results are similar to the CT prevalence of 11.3% (95% CI [7.3, 17.1]) among pregnant women of all ages reported by Davey and that reported by Korenromp,9.5% (95% CI [4.4, 15.4]), by means of modeling (Spectrum STI), which was also carried out on the general population (15,36 (40); however, they did not describe the age range of the participating women or the diagnostic methods used.
Concerning syphilis, our results are similar to those reported by Korenromp et al., who reported a global prevalence of syphilis with an interquartile range of 0.16% to 1.62% (41). The epidemiological bulletin of Chile reported a "rate" of 21 per 100 000 in the 15-19 age group and a rate of 55 per 100 000 in the 20-25 age group (42); these are reported cases, but it is unknown whether they are new cases or new and old cases (latent syphilis). In Costa Rica, the 2015 epidemiological bulletin reported a rate of 2.7 per 100 000 live births in the 10-14 age group, 31.6 per 100 000 live births in the 15-19 age group, and 74.6 per 100 000 live births in the 20-25 age group, which are cases reported by year (43). These data do not indicate whether they are new or old, and are data found in this study. Additionally, most epidemiological bulletins focus on the prevalence of gestational syphilis and do not differentiate by age group.