The definition of early sexual debut varied across studies and is displayed

for each study in Table 1. Studies that explored the impact of early age at first sexual debut on only high-risk behaviours, perceived risks or determinants of HIV/AIDS risk, such as condom use, multiple partners, other STIs or circumcision, were excluded. The abstracts of all 1572 relevant studies were screened by a single reviewer (NK) after 100% agreement was reached on applying the inclusion criteria to a sample of 250 abstracts between the two reviewers (NK and HS). Following this, the full texts of all studies learn more that could potentially be included in the review were obtained. After removing 13 duplicates, 128 full texts were retrieved and double-screened independently by two reviewers (NK and HS). Any differences in decisions were resolved through discussions. It had been decided that a third reviewer (CW) would be approached if there were differences in decision that could not be resolved; however,

this was not necessary as an agreement was reached on all studies. A total of 26 articles met the inclusion criteria and were included in the review. Two articles this website reported on the same data, and their information was therefore combined in the analysis, and they were treated Epothilone B (EPO906, Patupilone) as one study.[12, 13] Information about sample characteristics, setting,

study design, variables adjusted for and statistical results were extracted from the study by one reviewer (NK), and a confirmatory data extraction and quality appraisal were carried out by a second reviewer (HS). The quality appraisal was conducted using seven appraisal questions adapted from the graphical appraisal tool for epidemiological studies (GATE).[30] Each study could obtain a maximum score of 14, indicating the highest level of evidence. For each of the following components, a maximum score of two was given per study: focus of the study, generalisability of the findings, study design, use of adequate control variables, reliable and sensible outcome measures, sensitive reporting of biases and outcomes and ethics. It was agreed that a total score of 0–4 would imply very low quality, 5–8 low quality, 9–11 medium and 12–14 high. Any differences in scoring were resolved through discussions between the two reviewers. The findings of the systematic search first report the unadjusted bivariate associations that emerged in this review. Due to their equivocal outcomes, the multivariate findings of the review were summarised according to the conceptual framework (Fig. 2). Only studies that found a significant association in the unadjusted analysis are examined in the multivariate analysis.