In the univariable analyses, at-risk population, assay type, sampling methodology, sample size, year of publication, year of data collection, country, response rate, and sex were associated with prevalence at p≤0·2. Alignment with meta-regression underlying assumption of normal random effects ...