Search conducted on 12 March 2020 in PubMed, Embase and Literatura Latino Americana em Ciências daSaúde (LILACS).

Search strategies included exploded MeSH/Emtree terms and broad terms with no language or time restrictions.

The definition of Latin America and the Caribbean included 47 countries classified into three subregions:

• Central America: Belize, Costa Rica, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama.

• South America: Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, French Guiana, Guyana, Paraguay, Peru, Surinam, Uruguay, Venezuela.

• Caribbean: Anguilla, Antigua and Barbuda, Aruba, Bahamas, Barbados, Bermuda, British Virgin Islands, Cayman Islands, Cuba, Curacao, Dominica, Dominican Republic, Grenada, Guadeloupe, Haiti, Jamaica, Martinique, Montserrat, Puerto Rico, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, St. Barthelemy, St. Martin, Trinidad and Tobago, Turks and Caicos.

Search results were imported into the reference manager EndNote (Thomson Reuters, USA).

Screening was performed in four stages:

1. Duplicate publications were identified and excluded.

2. Titles and abstracts were screened for relevant and potentially relevant publications.

3. Full texts of relevant and potentially relevant publications were retrieved and screened for relevance.

4. Bibliographies of relevant publications and reviews were checked for additional potentially relevant publications.

Inclusion criteria were any publication, including a study with a minimum sample size of 10, reporting primary data on any of the following outcome measures:

1. HSV-2 antibody incidence as detected by a type-specific diagnostic assay.

2. HSV-2 antibody prevalence (seroprevalence) as detected by a type-specific diagnostic assay.

3. Proportion of HSV-2 in GUD as detected by standard viral detection and subtyping methods.

4. Proportion of HSV-2 in genital herpes (as opposed to HSV-1), as detected by standard viral detection and subtyping methods.

Exclusion criteria were

• Case reports, case series, reviews, editorials, commentaries and qualitative studies.

• Measures reporting seroprevalence in infants <6 months old as their antibodies can be maternal in origin.

In this study, the term ‘publication’ refers to a document reporting one or several outcome measures. ‘Study’ or ‘measure’ refers to a specific outcome measure and its details.

Extracted variables included author(s), publication title, year(s) of data collection, publication year, country of origin, country of survey, city, study site, study design, study sampling procedure, study population and its characteristics (eg, sex and age), sample size, HSV-2 outcome measures, and diagnostic assay.

Overall outcome measure and their stratified measures were extracted, provided the sample size in each stratum is ≥10.

For studies including overall sample size, but no individual strata sample sizes, the sample size of each stratum was assumed equal to overall sample size divided by the number of strata in the study.

Stratification hierarchy for incidence and seroprevalence in descending order of preference were

1. Population type as defined in online supplemental box S1.

2. Sex.

3. Age group classified as (groups optimised to best fit reported data):

   • <20 years old.

   • 20–30 years old.

   • 30–40 years old.

   • >40 years old.

Stratification hierarchy for GUD and genital herpes included genital herpes episode status and study site:

1. Genital herpes episode status classified as

   • First episode genital herpes.

   • Recurrent genital herpes.

2. Study site stratification classified as

   • Hospital.

   • STD clinic.

Measures reporting any HSV-2 outcome among children <15 years old were only reported but not included in the analyses.

Study’s precision classification into low versus high based on the sample size (<200 vs ≥200).

Study’s appraisal into low vs high ROB was determined using two quality domains:

• Sampling method: probability-based vs non-probability based.

• Response rate: ≥80% vs <80% or unclear.

Meta-analyses were conducted using DerSimonian-Laird random-effects models with inverse variance weighting. The variance of each outcome measure was stabilised using the Freeman-Tukey arcsine square-root transformation.

Pooled mean HSV-2 seroprevalence was estimated for each population type by sex, and for general populations by country, subregion, year of data collection range and year of publication range.

Pooled proportions of HSV-2 detection in GUD and in genital herpes cases were estimated.

Heterogeneity assessment was based on three complementary metrics:

• Cochran’s Q statistic to assess existence of heterogeneity in effect size (p value of <0.1 indicated heterogeneity).

• I² heterogeneity measure to assess the percentage of between-study variation in effect size that is due to actual differences in effect size rather than chance.

• Prediction interval to describe the distribution of true outcome measures around the pooled mean.

Univariable and multivariable random-effects meta-regression analyses using log-transformed proportions were carried out to identify predictors of HSV-2 seroprevalence.

Factors in the univariable model with a p value of <0.1 were included in the multivariable analysis.

Factors in the multivariable model with a p value of ≤0.05 were deemed to be significant predictors.

Variables included in the univariable metaregression model for HSV-2 seroprevalence were

• Population type.

• Age group.

• Sex.

• Country.

• Subregion.

• Country’s income: LIC, LMIC, UMIC, and HIC according the World Bank classification.

• Assay type (western blot, ELISA, and monoclonal antibody).

• Sample size.

• Sampling method.

• Response rate.

• Year of data collection.

• Year of publication.

• Year of data collection category (<2000, 2000–2010, >2010).

• Year of publication category (<2000, 2000–2010, >2010).

The year of data collection had a few missing variables that were imputed by adjusting the year of publication using the median difference with the year of data collection.