The PubMed database search identified 236 potential citations. An additional 170 possible citations were identified through related citation searches and citations from co-authors. Of the 406 combined citations, 356 citations were excluded as they were either duplicates or did not meet the inclusion criteria based on the title screen. The title and abstracts of the 50 remaining potential citations were then examined for inclusion. Overall, 31 studies did not meet the inclusion criteria based on their abstracts; hence the remaining 19 full text articles were retrieved and then assessed for full inclusion. Of the 19 potentially eligible studies, 10 did not meet the full inclusion criteria and were excluded. The remaining 9 studies met the inclusion criteria and were included in this review (Fig. 3). The search was updated in March 2017 to identify new studies and to include both USA and UK English spelling of key terms (behavior/behavior, utilisation/utilization), but no new studies were identified.
Study Setting and Demographics of Participants
Of the nine included studies, eight were conducted in the USA and evaluated CAPs based in schools in urban areas within Los Angeles, Seattle, Philadelphia, and New York [14,15,16,17,18,19,20] (Table 1). The remaining study was conducted in urban schools in Tijuana, Mexico . All nine studies were conducted in public secondary schools with students in the 13–18 years old age range. All the USA-based studies had approximately equal distributions of male and female students; in the Mexico study, approximately 60% of the student population were female .
Study Design Characteristics
The majority of studies (8/9) were outcome evaluations of school-based CAPs. Three studies used cross-sectional study designs to compare schools with CAPs to those without CAPs [17, 19, 20]. Four studies used quasi-experimental study designs . Two studies used a longitudinal design with pre- and post-CAP implementation surveys [14, 15]. Studies with longitudinal elements had follow-up times that varied from 6 months to 5 years in length. Outcomes in all studies were ascertained using self-reported surveys.
Four studies examined the effects of school-based CAPs on condom acquisition by adolescent participants, and all studies reported an increase in condom acquisition in schools with a CAP [16,17,18, 21]. The largest effect was observed in a quasi-experimental study in Mexico which evaluated a school-based CAP delivered with HIV prevention workshops . This study found that the odds of acquiring condoms among students in the intervention schools were 20 times more than students in the control schools (odd ratio (OR) 20.28, p < 0.001), albeit the sample size was small (n = 320) . The smallest effect size was observed in a quasi-experimental study in the USA in which the odds of acquiring condoms among students was 1.8 times higher (OR 1.81, 95% CI 1.32–2.49) in intervention schools where CAP implementation was strengthened . Furthermore, this study found that the odds of acquiring condoms among sexually active students in the intervention group were three times more than the control group (OR 3.08, 95% CI 1.77–5.36) .
Seven studies examined the effects of school-based CAPs on condom use. Two observational studies from the USA showed no significant differences in condom use [14, 19], whereas one school survey showed a 6% decrease in condom use in last 3 months in schools with CAPs (57–51% p = 0.042) . The authors of this study hypothesize that a substitution effect may have been present as students may have substituted condoms accessed from their school for condoms they acquired previously from the community, which may have led to this marginal decrease in condoms use . The remaining three studies showed increases in condom use [11, 17, 20]. The largest effect was observed in a cross-sectional study in the USA which showed that the odds of using condoms (OR 2.1, 95% CI 1.5–2.9) or using condoms as a form of contraception (OR 2.1, 95% CI 1.5–2.8) were two times higher among students that attended schools with CAPs . A similar effect was also observed in a cross-sectional study conducted in the USA (OR 1.36; p < 0.01) . Moreover, this study showed that the odds of using a condom during the last sexual intercourse within the past 6 months among sexually active students from schools with CAPs almost doubled (OR 1.85, p < 0.01) .
Eight studies assessed the impact of school-based CAPs on sexual behavior. Overall, five studies showed no significant differences in sexual behavior outcomes as measured by exposure to sexual intercourse within the last 3 months, frequency of sexual intercourse, and prevalence of multiple partners [11, 14,15,16, 20]. No study reported a significant increase in sexual activity among students attending schools with CAPs. However, three studies reported a significant decrease in some of their sexual behavior outcomes because of the CAPs [17, 18, 21]. The largest effect was observed in a quasi- experimental trial conducted in urban schools in Mexico . This study found that students in the intervention group who were exposed to school-based CAPs and HIV prevention workshops had an 86% decreased risk of initiating sexual practices compared with students who attended schools in the control group (HR 0.14, p < 0.001) . It is unclear what factors may have led to this finding in this study as key program attributes (anonymity and accessibility) were judged to be “low”  compared to other studies in the USA with programs that we considered to have medium to high anonymity and accessibility.
A cross-sectional study found that the odds of ever having sexual intercourse or having sexual intercourse in the past 3 months were 20% lower among students in schools with CAPs, (OR 0.8, 95% CI 0.6–0.9, p = 0.0037) and (OR 0.8, 95% CI 0.6–0.9, p = 0.0252), respectively . Furthermore, 2 years’ post-implementation of a school-based CAP in public high schools in the USA was associated with a 4% decrease in the percentage of students having sex in the last 3 months (32–28%) .
School-Based CAP Components
None of the included studies assessed factors facilitating or impeding delivery of the school-based CAP. Overall, four studies were judged to have implemented school-based CAPs with medium levels of anonymity (Table 1). The majority of studies (n = 5) were judged to have implemented school-based CAPs with low accessibility as condoms were not available at multiple locations or could not be accessed independently by the student. Two studies from the USA indicated that parental consent was required for students to access CAPs [14, 19]. There was no distinct pattern identified between condom acquisition or use and the anonymity or accessibility of the CAP or parental consent as almost all studies found that CAPs were positively associated with condom acquisition and/or condom use.
Risk of Bias
Five studies were judged to have moderate risk of bias as authors employed rigorous study designs and adjusted for confounders or reported adequate survey completion rates (60–80%) [14,15,16,17,18] but were prone to potential selection bias (Table 2). Three studies were judged to have high risk of bias as there was a lack of information regarding selection of participants and participant completion rate was low [11, 19, 20]. One study was judged to have low risk of bias as the study did not report information regarding the validity or reliability of their measurement tool .