Advertisement

Prevention Science

, Volume 19, Issue 4, pp 490–506 | Cite as

Do School-Based Programs Prevent HIV and Other Sexually Transmitted Infections in Adolescents? A Systematic Review and Meta-analysis

  • Ali Mirzazadeh
  • M. Antonia Biggs
  • Amanda Viitanen
  • Hacsi Horvath
  • Li Yan Wang
  • Richard Dunville
  • Lisa C. Barrios
  • James G. Kahn
  • Elliot Marseille
Article

Abstract

We systematically reviewed the literature to assess the effectiveness of school-based programs to prevent HIV and other sexually transmitted infections (STI) among adolescents in the USA. We searched six databases including PubMed for studies published through May 2017. Eligible studies included youth ages 10–19 years and assessed any school-based programs in the USA that reported changes in HIV/STI incidence or testing. We used Cochrane tool to assess the risk of bias and GRADE to determine the evidence quality for each outcome. Three RCTs and six non-RCTs, describing seven interventions, met study inclusion criteria. No study reported changes in HIV incidence or prevalence. One comprehensive intervention, assessed in a non-RCT and delivered to pre-teens, reduced STI incidence into adulthood (RR 0.36, 95% CI 0.23–0.56). A non-RCT examining chlamydia and gonorrhea incidence before and after a condom availability program found a significant effect at the city level among young men 3 years later (RR 0.43, 95% CI 0.23–0.80). The remaining four interventions found no effect. The effect on STI prevalence was also not significant (pooled RR 0.83 from two non-RCTs, RR 0.70 from one RCT). Only one non-RCT showed an increase in HIV testing (RR 3.19, 95% CI 1.24–8.24). The quality of evidence for all outcomes was very low. Studies, including the RCTs, were of low methodological quality and had mixed findings, thus offering no persuasive evidence for the effectiveness of school-based programs. The most effective intervention spanned 6 years, was a social development-based intervention with multiple components, rather than a sex education program, and started in first grade.

Keywords

HIV Sexually transmitted infections Adolescents School-based programs USA 

Introduction

Each year, nearly 2000 adolescents ages ≤19 years in the USA become infected with HIV (Centers for Disease Control and Prevention 2015a). Youth who are sexually active, have sex with multiple partners, do not use condoms, and who live in marginalized and urban communities are at higher risk of becoming infected with human immunodeficiency virus (HIV) or other sexually transmitted infections (STI) compared with either those who are not sexually active or who are consistently using condoms. In addition to their own health risks, infection with some STIs creates higher risk of HIV infection (Centers for Disease Control and Prevention 2015b). Thus, increasing rates of HIV/STI testing among young people is a public health priority.

There are several HIV/STI prevention interventions aimed at reducing the risk of HIV and other STIs among youth (Chin et al. 2012; Goesling et al. 2014). Such interventions, including but not limited to school- and community-based programs, include a range of components such as education, screening and counseling, prevention services, and treatment. Examples are abstinence education, sexual risk reduction, service learning, and peer leadership-based interventions (Chin et al. 2012). Other school-based interventions included STI screening (Cohen et al. 1999) and condom availability programs (Kirby and Brown 1996) that targeted adolescents.

Understanding the effectiveness of existing school-based programs is critical to informing policy and funding decisions. The global evidence suggests that some interventions may be efficacious in reducing sexual risk behavior in adolescents (Chin et al. 2012; Fonner et al. 2014; Mavedzenge et al. 2014; Underhill et al. 2008). Schools may be an important setting to implement interventions since over 90% of youth in the USA are in school, during critical developmental stages, and for over 6 h a day (National Center for Education Statistics 2015).

There has not yet been a systematic review and meta-analysis focused specifically on school-based programs among young people in the USA that have reported the primary outcomes of interest, HIV/STI incidence, and testing. The published literature reviews to date have been narrative only; have not focused on schools; and have focused on knowledge, attitudes, and behavioral outcomes but not on biological outcomes (Hale et al. 2014); excluded one or more of our primary outcomes (Chin et al. 2012; Goesling et al. 2014; Johnson et al. 2011; Mason-Jones et al. 2012). In addition, they were restricted to studies that provided positive evidence of effectiveness (Goesling et al. 2014); were not restricted to the USA (Hale et al. 2014; Mavedzenge et al. 2014; Oringanje et al. 2016; Oringanje et al. 2009; Picot et al. 2012) or excluded the USA (Fonner et al. 2014; Paul-Ebhohimhen et al. 2008). To better understand the potential of school-based programs to reduce HIV/STI incidence, we searched for studies in the USA that assessed changes in HIV/STI incidence and testing. We quantitatively assessed the impact of identified programs as well as quality of evidence by outcome across the literature.

Methods

Our review methods were based on those recommended in the Cochrane Handbook (Higgins and Green 2011). Our review protocol was registered in the PROSPERO international prospective register of systematic reviews (registration number CRD42016033007) (PROSPERO 2016).

Inclusion and Exclusion Criteria

Our population of interest was youth who received an intervention between the ages of 10 and 19 in the USA. Studies needed to report on sexual risk reduction or avoidance interventions delivered to young people in school settings, including after school hours, and analyze one or more of the primary outcomes. All types of behavioral and clinical interventions, single or multi-component, focused only on risk avoidance (i.e., abstinence-focused programs) or risk reduction, were included, provided they did not meet the exclusion criteria (see below).

Our primary outcomes included either changes in HIV/STI incidence, changes in rates of HIV testing, or changes in rates of STI testing. Secondary outcomes included changes in self-reported sexual risk behavior and HIV/STI knowledge and skills. Eligible study designs included randomized controlled trials (RCTs), observational cohort (prospective or retrospective), serial cross-sectional studies, and other longitudinal analyses with a comparison group. We limited our analysis to US studies, but studies published in any language were eligible for inclusion. We required a control condition; eligible comparators included no intervention or another intervention not specifically designed to reduce sexual risk.

We excluded studies not reporting our primary outcomes; studies with one or more components external to the school context if they did not report the outcomes of interest by the intervention components or by the target setting (i.e. at school); studies focused on secondary prevention (e.g., STI screening programs); studies conducted primarily outside school settings; and studies without comparators. We excluded interventions only focused on STI screening because their primary target outcome was STI detection and diagnosis, not STI incidence/prevalence reduction and because they had no control groups.

Searches and Study Selection

Using relevant keywords and Medical Subject Heading (MeSH) terms, we developed a comprehensive search strategy (Appendix K). We searched PubMed, the Cochrane Central Register of Controlled Trials, Education Resource Information Center (ERIC), PsycINFO, Scopus, and Web of Science. To identify unpublished studies, we searched conference abstract archives of the Conference on Retroviruses and Opportunistic Infections (CROI), the International AIDS Conference (IAC), the International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention (IAS), the American Public Health Association (APHA), and the US National HIV Prevention Conference. We also searched for ongoing studies in the National Institutes of Health’s trials registry at ClinicalTrials.gov.

We conducted searches of the “gray literature,” by the following strategies: (a) searching gray literature archive at the New York Academy of Medicine, (b) using advanced search syntax in Google to search the web sites of important agencies and organizations relevant to adolescent health, (c) searching archived conference abstracts from the American Public Health Association, the International AIDS Conference, and the International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, and (d) searching doctoral dissertations (via ProQuest Dissertations). We also reviewed bibliographies of all included and other highly relevant studies, and sought the guidance of experts and researchers in the field who may have known of unpublished or inaccessible research.

We imported all resulting records into EndNote version X7 (EndNote Corp 2013). One reviewer removed duplicate records and those that were clearly irrelevant according to their titles. Following this, two reviewers working independently applied inclusion criteria in screening citations by titles, abstracts, and keywords and identified records for full-text review. A third reviewer reconciled any disagreement. Two reviewers then examined the full text of each article to identify articles which satisfied our inclusion criteria.

Data Extraction

Two reviewers independently extracted data into a standardized, pre-piloted data extraction form, and then reconciled any discrepancies. The following characteristics were extracted from each included study:
  • Study details: Complete citation, study location, study design characteristics, and other relevant details.

  • Details of participants: Age range, ethnicity, socioeconomic status, and other relevant details.

  • Details of schools: Type and level of school, urban or rural context, and other relevant details.

  • Outcome details: Numerators and denominators associated with each outcome; definitions and descriptions of outcomes provided in papers; details of how outcomes were assessed.

  • Methodology: Methods, of recruitment and randomization if an RCT, numbers of participants entering the study, comparability of groups, study inclusion and exclusion criteria, length of follow-up, losses to follow-up, withdrawals, or drop-outs.

  • Bias assessment: Other details necessary to perform a bias risk assessment using the Cochrane tool described below, or the criteria for observational studies described below.

Risk of Bias Assessment

We used the Cochrane Collaboration tool (Higgins et al. 2011) for assessing the risk of bias for each included study. For RCTs, the risk of bias in individual studies includes seven domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other potential biases. For non-RCTs, we additionally used indices recommended by the Cochrane Collaboration(Cochrane 2017b), checking whether the intervention and comparison groups were comparable, exposures and outcomes appropriately measured, evaluating the adequacy of measures to adjust for confounding, and adequacy of follow-up time. For all sources of bias (e.g., random sequencing generation) that a non-RCT was not able to address, we assigned a “high” risk of bias.

Additional Data Inquiries

We contacted eight corresponding authors of the included studies to inquire about any unpublished data or sub-group analyses that might be relevant to our systematic review, and about any additional studies that we may have missed. We received responses from four authors, none of which led to the generation of additional effectiveness estimates or amendments of existing ones.

Quality of the Evidence

We used the GRADE approach (Guyatt et al. 2011) to assess the quality of evidence by outcome across the body of literature. We used GRADEpro software version 3.2 to perform analyses and create evidence profiles (Brozek et al. 2008). GRADE ranks the quality of evidence on four levels: “high,” “moderate,” “low,” and “very low.” Evidence quality from RCT data is initially presumed to be “high,” but can be downgraded based on study limitations, inconsistency of results, indirectness of evidence, imprecision, or for reporting bias. Evidence quality from non-RCT study data starts at “low” but can be upgraded if the magnitude of treatment effect is very large, if there is a significant dose-response relation, or if all possible confounders would decrease the magnitude of an apparent treatment effect (Guyatt et al. 2011). Evidence from non-RCT studies can also be downgraded.

Measures of Intervention Effect

For binary outcomes, we reported the risk ratio (RR) as the effect size measure. Wherever the outcome was reported as a proportion in intervention and control group, we directly calculated the RR and 95% confidence interval (CI). For outcomes reported as an odds ratio (OR), we used the OR, sample size, and proportion of the outcome reported for the control group at follow-up time to back-calculate the 2 × 2 table (outcome by groups) and then calculated the RR and 95% CI.

For continuous outcomes, we reported the mean ratio (MR) or the standard mean difference (SMD). In cases where the mean and standard deviation of the outcome was reported at follow-up in the intervention and control group, we used the RevMan calculator (Cochrane 2011) to calculate the SMD and 95% CI. If the paper reported the effect size as mean ratio and 95% CI, we also reported the effect size as MR.

Meta-analyses

The effect measures and 95% CI were entered into Stata® (version 13) for meta-analysis and generation of forest plots using Stata’s Metan command. We used random effect models to pool the studies’ effects on measures for similar outcome, population, intervention, comparison, and follow-up time categories. The choice of a random model was an a priori decision to support extrapolation of findings with plausible confidence intervals to other school settings and youth populations beyond those analyzed in the included studies. We reported the results for each outcome in subcategories by study design (RCT and non-RCT). We assessed the publication bias for STI incidence/prevalence by subgroups of RCTs and non-RCTs visually using a Funnel Plot (the standard normal deviation of intervention effect estimate against its precision). We only assessed the symmetry of the funnel plot visually because we had an insufficient number of studies (<10) to test for symmetry per the methods described in the Cochrane handbook Ch.10.4.3.1 (Cochrane 2017a).

Results

Results of Search

We found 4583 unique records through our searches. Screening only titles, one reviewer removed 2091 records that were clearly irrelevant. Working independently, two reviewers screened titles, abstracts, and indexing terms of 2491 records. After excluding 2401 records based on title and abstract screening, we reviewed 90 full-text articles for eligibility, of which nine met our inclusion criteria. We had no disagreements in screening (Fig. 1).
Fig. 1

School-based programs to prevent HIV/STI in adolescents in the USA: flowchart for systematic review

Included Studies

Nine studies evaluating seven separate interventions met our inclusion criteria and are described in Table 1. Three of the studies evaluated the same intervention, the Seattle Social Development Project (SSDP) but at different follow-up periods (Hawkins et al. 2008; Hill et al. 2014; Lonczak et al. 2002). Three of the studies were RCTs (two randomized classrooms (Coyle et al. 2001; Kirby et al. 1997), one randomized schools (Spoth et al. 2017)), and six were non-RCT (Hawkins et al. 2008; Hill et al. 2014; Lemieux et al. 2008; Lonczak et al. 2002; Walter and Vaughan 1993; Wretzel et al. 2011). The length of follow-up varied from 3 months to 18 years.
Table 1

Description of studies included in the review (9 studies, 7 interventions)

First author, year

Methods

Participants

Interventions

Outcomes

Intervention: safer choices

Coyle, 2001

Design: randomized controlled study

Unit of randomization/study: restricted randomization process.

Randomization was by school.

Duration: fall 1993–Spring 1996

City/state/country: Texas and California, USA

Setting: 20 high schools; 10 in an urban area in Northern California and 10 in an urban area in Texas.

Inclusion criteria: students had to remain at the school through both years of the program and be in 9th grade when program began. Parental consent needed.

Participants at baseline: 3869 students

Age range at baseline: 14–15 years old

Sex (% female) at baseline: 50.3% (intervention) | 53.6% (control)

Race/ethnicity at baseline:

Hispanic: intervention 28.7% | control 26.3%;

Black: intervention 19.6% | control 14.5%; White: intervention 28.7% | control 30.9%; Asian: intervention 13.7% | control 22.4%; Other: intervention 7.7% | control 5.9%

Intervention description: 2-year, school-based HIV, other STD, and pregnancy prevention program. Five component curriculum: (1) 20 classroom sessions, (2) parental education, (3) school-community linkages, (4) peer resources (events hosted by a “Safer Choices” student club), and establishment of a “School Health Promotion Council” (made up of teachers, students, parents, community representatives, and administrators.)

Control: standard knowledge-based HIV prevention program curriculum

Primary: tested for HIV and STI among all youth regardless of baseline sexual activity.

Secondary: ever had intercourse, condom use at last intercourse, frequency of intercourse without a condom in last 3 months, number of times had sex in last 3 months, number of sexual partners in last 3 months

Assessed at: 7, 19, and 31 months follow-up

Lost to follow-up:

At 7 months: 5%

At 19 months: 17%

At 31 months: 21%

Intervention: Seattle Social Development Project

Lonczak, 2002

Design: non-randomized controlled study

Unit of randomization/study: none. Schools were non-randomly assigned to the intervention or control condition.

Duration: fall 1981–1996

City/state/country: Seattle, WA, USA

Setting: fifteen public elementary schools in Seattle, Washington serving diverse communities including high-crime neighborhoods

Inclusion criteria: first to sixth grade student at participating schools that received parental written consent to participate were assigned to one of the three groups: (1) those already assigned to full intervention arm in first grade and still in the school district in fifth-sixth grade, (2) those assigned to the late intervention in 5-6th grade, and (3) those assigned to the control arm. In this paper, only full intervention and control group are presented.

Participants at baseline (initiation of late intervention): 376 students

Mean age at baseline: 10.8 years

Sex (% female) at 9 years follow-up: intervention (49%) | control (48%)

Race/ethnicity at 9 years follow-up: White intervention (47%) | control (46%)

Intervention description: this was not a sexual education program. Guided by the Social Development Model, the intervention sought to engage students by providing opportunities to actively become involved in their community and help them form bonds with their parents and teachers. The intervention had three components: (1) Classroom instruction and management—5 days of in service training for teachers annually; teachers in the full intervention arm received additional training on interpersonal cognitive problem solving. (2). Child Skills Development—first grade teachers (in the full intervention arm) received additional training on interpersonal cognitive problem solving to facilitate cooperative learning groups for the students. In fifth-sixth grade, students in the full intervention group received 4 h of training from project staff to develop skills to resist peer pressure and stay out of trouble. (3) Parent Intervention—voluntary ongoing parent training classes were offered from grades 1–6 (16 sessions total). Students in the full intervention arm participated in grades 1–6. Those in the late arm participated in 5–6 grade.

Control: no intervention

Primary: STD prevalence among all youth regardless of baseline sexual activity.

Secondary: condom use at last intercourse, condom use (% of time)

Assessed at: 9 years follow-up (age 20)

Lost to follow-up:

At 9 years: 7%

Hawkins, 2008

Design: non-randomized controlled trial

Unit of randomization/study: none. Schools were non-randomly assigned to the intervention or control condition based on their preference

Duration: fall 1981–2003

City/state/country: Seattle, WA, USA

Setting: fifteen public elementary schools in Seattle, Washington serving diverse communities including high-crime neighborhoods

Inclusion criteria: first to sixed grade student at participating schools that received parental written consent to participate were assigned to one of the three groups: (1) those already assigned to full intervention arm in first grade and still in the school district in fifth-sixth grade, (2) those assigned to the late intervention in 5–6th grade, and (3) those assigned to the control arm

Participants at 12 and 15 years follow-up: 598 students

Mean age at baseline: 10.8 years

Sex (% female at 12 and 15 years follow-up: 50%

Race/ethnicity at 12 and 15 years follow-up:

African American (26%); Asian American: (22%); White: (46%), Native American: (6%)

Intervention description:

two intervention conditions were examined: a full-intervention condition implemented throughout grades 1 through 6 and a late-intervention condition implemented only in grades 5 and 6. In this paper, results from both intervention groups as well as control group are presented. For details, see Lonczak et al. (2002)

Control: no intervention

Primary: lifetime STD index among all youth regardless of baseline sexual activity.

Secondary: condom use (% of time), sex partners in last year

Assessed at: 12 and 15 years follow-up (age 24 and 27 years, respectively)

Lost to follow-up:

At 12 years: 7%

At 15 years: 7%

Hill, 2014

Design: non-randomized controlled study

Unit of randomization/study: none. Schools were non-randomly assigned to the intervention or control condition

Duration: fall 1985–2003

City/state/country: Seattle, WA, USA

Setting: fifteen public elementary schools in Seattle, Washington serving diverse communities including high-crime neighborhoods

Inclusion criteria: first to sixth grade student at participating schools that received parental written consent to participate were assigned to one of the three groups: (1) those already assigned to full intervention arm in first grade and still in the school district in fifth-sixth grade, (2) those assigned to the late intervention in 5–6th grade, and (3) those assigned to the control arm

Participants at baseline (initiation of late intervention): 608

Mean age at baseline: 10.8 years

Sex (% female) at baseline: intervention (48%) | control (47%)

Race/ethnicity at baseline: White: intervention (49%) | control (46%)

Intervention description: see Lonczak et al. (2002)

Control: no intervention

Primary: STI onset among all youth regardless of baseline sexual activity.

Secondary: none

Assessed at: 18 years follow-up (age 30)

Lost to follow-up

At 18 years: 9%

Intervention: Project SNAPP: skills and knowledge for AIDS and pregnancy prevention

Kirby, 1997

Design: randomized controlled study

Unit of randomization/study: randomization was done by classroom

Duration: sessions delivered over a 2-week period

City/state/country: Los Angeles, California, USA

Setting: 102 seventh grade classrooms at 6 Los Angeles middle schools serving the Hollywood-Wilshire and Central areas

Inclusion criteria: not reported

Participants at 17 months follow-up: 1616

Mean age at baseline: 12.3 years

Sex (% female) at baseline: 54%

Race/ethnicity at baseline:

Latino (64%); Asian (13%); African American: (9%); Non-Latino White (5%)

Intervention description: Project SNAPPs primary objective is to prevent unintended pregnancies and HIV infection by delaying the initiation of sexual intercourse. The curriculum, which consists of 8 sessions, is grounded in social learning theory and incorporated principles of the Health Belief Model. Through small and large group activities, role-playing and guided discussions, students learn to recognize social influences, and practice assertive communication and peer pressure resistance skills. The program was implemented by trained peer leaders that reflected the diversity of the sample.Control: conventional sex education

Primary: diagnosed with STD among all youth regardless of baseline sexual activity

Secondary: sex initiation, condom use at last intercourse, number of times had sex in last 3 months

Assessed at: 5 and 17 months follow-up

Lost to follow-up:

At 5 months: 27%

At 17 months: 23%

Intervention: a music-based HIV prevention intervention for urban adolescents

Lemieux, 2008

Design: non-randomized controlled study

Unit of randomization/study: none. Schools were non-randomly assigned to the intervention or control condition

Duration: 5 months

City/state/country: not reported

Setting: health classes at three diverse urban high schools

Inclusion criteria: musically talented students identified as “opinion leaders”; needed to be well respected, influential, able to credibly promote safer sex or abstinence, and previous musical experience. Participants needed to participate in health classes

Participants at baseline: 306 students

Mean age at baseline: 16 years

Sex (% female) at baseline: 49%

Race/ethnicity at baseline:

Latino (43%); African American (37%); White (4%); Asian (2%); Native American (1%); and Other (13%)

Intervention description: intervention was based on the Natural Opinion Leader (NOL) model, which asserts that peers viewed as popular heavily influence adolescents. Based on this assertion, HIV prevention interventions implemented by popular/leader peers should increase student receptivity. It is also based on the idea that popular music influences teens. Over a 4-month period, musically talented NOLs wrote, recorded HIV prevention-themed music, and recorded a music video. NOLs then present content in health classes and 1 month later set up a table to promote and distribute the recorded CDs and other educational materials

Control: standard of care (health class without the intervention)

Primary: new HIV test among youth who were sexually active at baseline

Secondary: none

Assessed at: 3 months follow-up

Lost to follow-up:

At 3 months: 27%

Intervention: special AIDS-preventive curriculum

Walter, 1993

Design: non-randomized controlled study

Unit of randomization/study: classroom

Duration: 1990–1991

City/state/country: New York City, New York, USA

Setting: four academic high schools in a New York City borough.

Inclusion criteria: participants had to be in 9th or 11th grade at a participating school enrolled in required general education classes.

Participants at baseline: 1201

Mean age at baseline: 15.7 years

Sex (% female) at baseline: 58%

Race/ethnicity at baseline: Black (37%); Hispanic 35%; Other (28%)

Intervention description: the intervention consisted of six lessons, implemented on consecutive school days. The first two focused on transmission of HIV. The middle two lessons focused on understanding one’s own personal values and dispelling common misconceptions about the prevalence of HIV and risk. The final two lessons focused on empowering the students with skills necessary to practice safer sex and tools to negotiate and resist peer pressure to have sex

Control: no intervention

Primary: STD incidence among all youth regardless of baseline sexual activity

Secondary: consistent condom use in the last 3 months

Assessed at: 3 months

Lost to follow-up:

At 3 months: intervention (28.5%), control (27%)

Intervention: condom distribution

Wretzel, 2011

Design: Observational Double-Arm Retrospective Cohort Cluster

Duration: 2002–2008

Participants at baseline: not reported

Age range: 15–19 years

Sex (% female): NR.

Race/ethnicity: (Holyoke, Springfield) 3.7, 21.0% Black; 41.4, 27.2% Latino

This study compared changes in publicly reported STI rates in a city that began offering condoms in all high schools to another city that did not.

Primary: chlamydia and gonorrhea incidence.

Assessed at: 36 months

Intervention: PROSPER partnership delivery system for brief universal preventive interventions

Spoth, 2017

Design: randomized controlled study

Unit of randomization/study: randomization was done by public school district

Random allocation method: public school districts were blocked on the basis of size and location, then randomized and informed of their assignment to the intervention condition

Duration: sessions for the two components were delivered over a 2-year period. The family-based component had seven sessions, delivered over the first year (6th grade). The number of sessions for the school-based component, delivered in year 2 (7th grade), varied depending on which program the district selected: Life skills training (LST) had 15 sessions, Project ALERT had 11 sessions, and All Stars had 13 sessions

City/state/country: Iowa and Pennsylvania states, USA

Setting: 14 rural public school district in each of the two states

Inclusion criteria: for a school district to be eligible for inclusion, it had to serve between 1300 and 5200 students; a team of community and school stakeholders had to agree to be randomly assigned to intervention or control group; and have the capacity and willingness to support PROSPER model implementation if assigned to the intervention group. Participants had to be in 6th grade. A subset of participants was selected by random for long-term follow-up (outcomes reported in this paper). Students in the same school district for 9th grade who have completed the 6th grade baseline assessment were eligible for inclusion. Subsample selection was stratified by school district, gender and risk behavior status

Participants: two successive cohorts of 6th graders: At baseline, N intervention = 5515, N control = 5334. Given that only a subsample followed into adulthood (age 19), at follow-up, N intervention = 1004, N control = 981

Mean age at baseline: 11.8 years; mean age at follow-up: 19.5 years

Race/ethnicity at baseline: White (85%)

Sex (% female) at baseline: 51%

Intervention description: teams of community stakeholders selected two interventions from a menu of evidence-based programs: one family-based program and one school-based intervention. All 14 community teams choose Strengthening Families Program: for parents and youth 10–14 (SFP 10–14) for the family-based program. SFP 10–14’s curriculum is focused on enhancing parenting skills and improving youth substance refusal skills and prosocial skills

For the school-based component, four districts chose Life Skills Training (LST); four selected Project Alert; and six selected the All Stars program. Each school-based program was delivered during a class typically taken by the 7th graders and typically implemented by a trained classroom teacher. LST’s intervention focused on promoting skills around resisting peer pressure, and self-management and developing strategies to avoid the use of substances. Project ALERTs curriculum was social influence-focused and aimed to change student perceptions around substance use norms and strengthen their refusal self-efficacy. All Stars’ curriculum focused on student’s beliefs about violence and substance use, sought to increase the accuracy of their beliefs around peer norms and asked them to make a personal commitment to avoiding negative behaviors.

Control: “usual programming conditions”

Primary: Lifetime STI

Assessed at: 8 years (at the age of 19 years).

Lost to follow-up: not reported

Participants: A total of nearly 19,000 baseline participants were represented in the included studies, ranging from 242 to 10,849 participants per study. Participants included first to six graders (Hawkins et al. 2008; Hill et al. 2014; Lonczak et al. 2002), children in middle school (Kirby et al. 1997), and high school students (Coyle et al. 2001; Lemieux et al. 2008; Walter and Vaughan 1993; Wretzel et al. 2011) (Spoth et al. 2017). Participants were evenly divided by sex. The racial/ethnic distribution of participants varied by study. In Kirby et al. (Kirby et al. 1997), over half of participants were Hispanic/Latino, and in Coyle et al. (Coyle et al. 2001), approximately one quarter were Hispanic/Latino. Whites made up nearly half of one intervention (Hawkins et al. 2008; Hill et al. 2014; Lonczak et al. 2002; Spoth et al. 2017), and 5% or less of two (Lemieux et al. 2008; Walter and Vaughan 1993). African American represented over one third of the participants in two interventions (Lemieux et al. 2008; Walter and Vaughan 1993), and about one fourth of them in one intervention (Hawkins et al. 2008; Hill et al. 2014; Lonczak et al. 2002).

Settings: All studies, except one (Spoth et al. 2017), were conducted in schools located in urban centers throughout the USA. The number of schools ranging from 3 to 28 schools per study.

Interventions: The intervention reported by Coyle et al. (Coyle et al. 2001) was a 2-year HIV, STD, and pregnancy educational curriculum with five components including 20 classroom sessions, parental education, school-community linkages, peer resources, and a health promotion council at school. SSDP intervention, which was assessed by Lonczak et al. (Lonczak et al. 2002), Hawkins e al. (Hawkins et al. 2008), and Hill et al. (Hill et al. 2014), was not a sexual education program per se but focused on social development, engaging students in community activities and helping them form bonds with their parents and teachers. It had three components including classroom instruction and management, child skills development, and parent intervention. Skills and Knowledge for AIDS and Pregnancy Prevention (SNAPP) intervention (Kirby et al. 1997) was a curriculum-based training program to prevent unintended pregnancies and HIV infection among students. It consisted of eight sessions and was implemented through group activities, role-playing, and guided discussions led by trained peer leaders. The intervention assessed by Lemieux et al. (Lemieux et al. 2008) was based on the natural opinion leader model, which asserts that peers viewed as popular heavily influence adolescents. It was delivered by writing and recording HIV prevention-themed music and a music video. Special AIDS-preventive curriculum intervention, which was assessed by Walter and Vaughan (Walter and Vaughan 1993), consisted of six lessons (two on HIV transmission, two on understanding one’s personal values, and two on building skills to practice safer sex), implemented on consecutive school days. Wretzel et al. (Wretzel et al. 2011) evaluated a school-based condom availability program. Spoth et al. (Spoth et al. 2017) evaluated the PROSPER intervention in which teams of community stakeholders selected two programs from a menu of evidence-based interventions addressing multiple risk and protective factors, designed to reduce substance misuse; a family-based program was delivered over the first year to 6th grade and a school-based intervention was delivered over the second year to 7th grade.

Comparisons: Four interventions were compared to standard sexual education or standard care (Coyle et al. 2001; Kirby et al. 1997; Lemieux et al. 2008; Spoth et al. 2017). Two interventions were compared to nothing (Hawkins et al. 2008; Hill et al. 2014; Lonczak et al. 2002; Walter and Vaughan 1993). One intervention compared the change in the rates of gonorrhea and chlamydia infection in 15–19 years old in a city with a condom availability program to the rates of teens in a similar city that did not have such a program (Wretzel et al. 2011).

Outcomes: Seven studies assessed STI incidence/prevalence (Hawkins et al. 2008; Hill et al. 2014; Kirby et al. 1997; Lonczak et al. 2002; Spoth et al. 2017; Walter and Vaughan 1993; Wretzel et al. 2011) and two studies assessed HIV and/or STI testing outcomes (Coyle et al. 2001; Lemieux et al. 2008). Three studies examined secondary outcomes such as sexual initiation (Coyle et al. 2001; Kirby et al. 1997), frequency of sex (Coyle et al. 2001; Kirby et al. 1997), condom use (Coyle et al. 2001; Walter and Vaughan 1993), and number of sexual partners (Coyle et al. 2001).

Excluded Studies

Of the 81 studies that we excluded after full-text review, 50 did not report on primary outcomes, 14 were not school-based programs, 5 were reviews, 3 did not meet other inclusion criteria, 2 were reviews of reviews, 4 did not report outcomes at pre/post intervention, 2 had insufficient data, and 1 was a database that did not contain any additional studies meeting our inclusion criteria. The list of excluded studies and justification is presented in Appendix L.

Risk of Bias in Included Studies

All included studies were at high risk of bias (Appendix A, Fig. 2). Two RCT (Kirby et al. 1997; Spoth et al. 2017) did not report sequence generation methods, so it was not clear whether these methods were adequate to prevent selection bias. Another RCT (Coyle et al. 2001) used adequate sequence generation methods.
Fig. 2

Risk of bias item for included studies, per review team’s assessment

Allocation of participants to intervention or control conditions was not concealed in all trials, nor were participants, study personnel, or outcome assessors blinded to participant assignment. Further, outcome assessors in all trials observed outcomes only indirectly, relying on participant self-report. In addition to the indirectness, bias risk in all studies may be higher because adolescents may not accurately report sexual behavior outcomes (Brown et al. 2012; Clark et al. 1997; Glassman et al. 2014).

Participant loss to follow-up was high in two RCTs, reaching >20%, with inadequate reporting of reasons for attrition. The loss to follow-up was not reported in one RCT (Spoth et al. 2017). It was not possible to discern whether selective outcome reporting may have been an issue in these trials, as their original protocols were not available for review. Finally, it was unclear whether the intervention and control populations in Kirby et al. 1997 were socio-demographically equivalent at baseline (Kirby et al. 1997). No additional risks of bias were identified in Coyle et al. (Coyle et al. 2001).

The non-RCTs lacked the methodological benefits of randomization and were thus at inherently high bias risk. Walter and Vaughan 1993 was a quasi-randomized trial but quasi-randomization methods do not adequately prevent potential selection bias (Walter and Vaughan 1993). As with the RCTs, allocation was not concealed in these studies, nor were participants, personnel, or outcome assessors blinded to participant status. Groups were socio-demographically comparable in SSDP and Walter and Vaughan (Walter and Vaughan 1993) but this was unclear in Lemieux et al. (Lemieux et al. 2008). Again, outcomes could only indirectly be assessed through participant self-report. Losses to follow-up were high (>20%) in Walter and Vaughan (Walter and Vaughan 1993) without adequate reporting of participant disposition. While attrition was also >20% in Lemieux et al. (Lemieux et al. 2008), investigators report that it was comparable across groups. Losses to follow-up were fairly low (7–9%) in SSDP and were balanced across groups. As with the RCTs, we could not assess the risk of selective outcome reporting because study protocols were not available. With regard to other types of bias risk, Lemieux et al. (Lemieux et al. 2008) did not adequately control for confounding variables by statistically adjusting study analyses. In their first report (Lonczak et al. 2002), the SSDP study did not clearly describe adjusting analyses, though subsequent reports explained this process adequately (Hawkins et al. 2008; Hill et al. 2014). Walter and Vaughan (1993) did not adequately adjust for confounders. The Wretzel et al. (Wretzel et al. 2011) assessed the exposure and outcomes both at the community level leaving the possible contamination effect of non-exposure participants and did not adjust for possible confounding effects (racial differences). All non-RCTs followed participants for an adequate length of time for appropriate outcome assessment.

Effect of Interventions

Primary Outcomes

HIV incidence or prevalence: No study reported this outcome.

STI incidence or prevalence (Fig. 3): One RCT and one non-RCT reported STI incidence. In 3 to 5 months’ follow-up, both Kirby et al. (1997) (RR 1.27, 95% CI 0.47, 3.38) and Walter and Vaughan (1993) (RR 0.63, 95% CI 0.28, 1.42) reported non-statistically significant effects. Kirby et al. (1997) also assessed the STI incidence at 17 months’ follow-up, and found no significant effect (RR 0.72, 95% CI 0.23, 2.25). In Hill et al. (2014), a non-RCT with 18 years of follow-up, the pooled effect of the intervention on STI incidence was 0.60 (95% CI 0.12, 1.08), not statistically significant. The group that had received the full intervention showed a significant reduction in STI incidence (RR 0.36, 95% CI 0.23, 0.56) when compared with the control group; however, the effect was not statistically significant for the late intervention group (RR 0.85, 95% CI 0.66, 1.09). The long-term effect of school-based interventions on STI prevalence was reported by two non-RCT studies (Hawkins et al. 2008; Lonczak et al. 2002); the pooled RR was 0.83 (95% CI 0.56, 1.09) and one RCT (Spoth et al. 2017). The group that received the full intervention in Hawkins et al. (2008) study had a significant reduction (RR 0.67, 95% CI 0.47, 0.95) in STI prevalence at 27 years, whereas the group that received late intervention did not have a significant effect. Spoth et al. (2017), an RCT that assessed the effect of intervention after 8 years (at the age of 19) on lifetime STI prevalence, had and RR of 0.70 (95% CI 0.42, 1.16). Wretzel 2011, the only study that assessed STI outcomes by biological testing, examined chlamydia and gonorrhea incidence before and after the condom availability program. Three years after the program, the incident chlamydia and gonorrhea decreased more among boys in a city where schools began distributing condoms compared to a nearby city where they did not (RR 0.43, 95% CI 0.23, 0.80), an effect that was not observed among girls (RR = 1.02, 95% CI 0.92–1.13; Wretzel et al. 2011).
Fig. 3

School-based sexually transmitted infection (STI) prevention programs effects on STI incidence/prevalence, by study design and follow-up time and reported outcome

HIV testing (Fig. 4): One RCT (Coyle et al. 2001) and one non-RCT (Lemieux et al. 2008) assessed HIV testing uptake in less than 12 months’ follow-up. Although HIV testing increased threefold (RR 3.19, 95% CI 1.24, 8.24) in the Lemieux et al. (2008) study, no statistically significant effect was observed in Coyle et al. (2001; RR 0.78, 95% CI 0.41, 1.47). Coyle et al. (2001) also assessed the effect of intervention on HIV testing at 19 and 31 months’ follow-up; and again, no statistically significant effect was reported. The corresponding RRs were 1.35 (95% CI 0.85, 2.14) and 1.18 (95% CI 0.73, 1.92).
Fig. 4

School-based sexually transmitted infection (STI) prevention programs effects on HIV testing, by study design and follow-up time

STI testing (Fig. 5): Coyle et al. (2001) assessed the effect of intervention on STI testing at 7, 19, and 31 months’ follow-up. STI testing gradually increased in the intervention over these intervals compared with the control arm (RR ranging from 1.04 to 1.51) but was not statistically significant.
Fig. 5

School-based sexually transmitted infection (STI) prevention programs effects on STI testing, by follow-up time

Secondary Outcomes

Frequency of intercourse (Appendix D and E): Only two RCTs (Coyle et al. 2001; Kirby et al. 1997) assessed the effect of interventions on the frequency of sex and sex without using condoms. In Coyle et al. (2001), at 7 months’ follow-up, there was no difference in the average frequency of sex in the last 3 months by MR 1.16 (95% CI 0.85, 1.58). Although a significant reduction in the frequency of sex without a condom was observed (MR 0.50, 95% CI 0.27, 0.94), at 19 and 31 months’ follow-up, both frequency of sex and sex without a condom were similar in the intervention and control groups. Similarly, in the Kirby et al. (1997) study at 5 and 17 months’ follow-up, participants in the intervention group reported similar frequency of sex in the last 3 months to the control group (SMD 0.06 and 0.08).

Number of partners (Appendix F and G): Three studies reported this outcome. There were no significant differences in the average number of sex partners (which ranged from 0.91 to 1.02 at 7 and 31 months’ follow-up, respectively) or number of sexual partners (SMD −0.06, 95% CI −0.26, 0.13).

Initiation of sexual intercourse (Appendix H): Two RCTs showed no statistically significant difference in the proportion of participants that initiated sexual intercourse in the year following the intervention (RR 1.14, 95% CI 0.90, 1.45), 12–23 months’ follow-up (RR 0.97, 95% CI 0.79, 1.19), and 31 months’ follow-up (RR 1.04, 95% CI 0.94, 1.16).

Sex without a condom (Appendix I and J): Two RCTs and two non-RCTs reported this outcome. Walter and Vaughan (1993) had insufficient data to calculate an effect size (RR). At the 7 months’ follow-up, one RCT (Coyle et al. 2001) showed a significant reduction in no condom use at last sex (RR 0.65, 95% CI 0.46, 0.93) between the intervention and control groups. The other study (Kirby et al. 1997) had no statistically significant effects. The pooled effect was RR 0.82 (95% CI 0.46, 1.46), which was also not statically significant. Similarly, at 17 months’ follow-up, there was no significant reduction in sex without a condom among participants in the intervention group when compared to the control group (RR 0.92, 95% CI 0.71, 1.21). One non-RCT (Lonczak et al. 2002) showed a significant decrease (RR 0.72, 95% CI 0.54, 0.97) in non-condom use sex at last sex between participants in the intervention (36/89) and control groups (86/154). One non-RCT (Hawkins et al. 2008; Lonczak et al. 2002) showed no difference in the percent of sexual episodes in which a condom was used in the last year (SMD 0.11, 95% CI −0.06, 0.28).

HIV/STI knowledge, attitude, and behavior (Appendix B): Four studies reported these outcomes. Two studies (Coyle et al. 2001; Walter and Vaughan 1993) reported a favorable and statistically significant effect of the intervention on increasing HIV knowledge. Significant improvements in attitudes toward condoms (Coyle et al. 2001; Lemieux et al. 2008), other STI knowledge, HIV, and other STI risk perception (Coyle et al. 2001), benefits of engaging in AIDS prevention programs and norms (Walter and Vaughan 1993) were also reported by three studies. Three studies (Kirby et al. 1997; Lemieux et al. 2008) showed no significant difference for HIV risk perception between the intervention and control group, and two studies (Coyle et al. 2001; Lemieux et al. 2008) reported no significant effect on normalizing condom beliefs. No study reported significant effects of the intervention on attitudes and beliefs about sexual intercourse.

Quality of the Evidence: GRADE Analysis

The quality of evidence for each outcome across studies is presented in Appendix C. For all outcomes, the quality of evidence was very low primarily due to risk of bias and serious or very serious imprecision for the reported outcomes. STI prevalence was reported in four studies; STI incidence was reported in two studies. Other subcategories as shown in Appendix C were only reported in one study.

Discussion

We evaluated three RCTs and four non-RCT studies that reported on one or more of the primary outcomes of STI incidence or uptake of HIV/STI testing. We found no studies that assessed the effect of school-based prevention programs on HIV incidence among adolescents. Of the included studies, we extracted nine RRs on STI incidence or prevalence, of which two (Hill et al. 2014; Hawkins et al. 2008) that involved both teachers and parents in helping pre-teen students to develop skills to resist peer pressure and stay out of trouble showed a favorable, statistically long-term significant effect in adulthood.

Only one intervention provided to pre-teens and assessed in a non-RCT (Hill et al. 2014; Hawkins et al. 2008) found that a school-based program might reduce STI incidence in adulthood (RR 0.35–0.67). The intervention was unique as it was not a sex education program but aimed to engage students in community service and bonding them with teachers and parents. This effect was not observed in any of the other studies. A randomized trial with a curriculum-based peer-delivered intervention that targeted middle school youth with 5 and 12 months’ follow-up did not show a significant effect on STI incidence (Kirby et al. 1997). One study, Wretzel et al. (2011), assessed STI rates as reported by the departments of public health in two cities, one with school-based condom availability program and the other without. The differences in the racial composition of the two cities, old census data for population denominators, the inclusion of non-program participants, and the small number of STI cases and high variability of observed disease incidence make it hard to attribute changes in STI rates to the condom availability program.

Two studies (Coyle et al. 2001; Lonczak et al. 2002) showed a favorable effect of the intervention on condom use at last sex. These interventions did not reduce frequency of intercourse, frequency of intercourse without condoms, number of partners, or likelihood of sexual initiation.

In the four studies that reported on changes in HIV/STI knowledge and attitude, two showed favorable and statistically significant effects on HIV/STI knowledge and STI risk perception; but none reported a significant effect on HIV risk perception and normalizing condom beliefs.

Most of the studies included in this review featured an educational curriculum and were delivered mostly by teachers (Walter and Vaughan 1993) (Coyle et al. 2001; Lonczak et al. 2002; Hawkins et al. 2008; Hill et al. 2014; Walter and Vaughan 1993), but some were delivered by peers (Kirby et al. 1997; Lemieux et al. 2008). Similarly, recent reviews (Oringanje et al. 2016; Kirby 2002) have suggested that classroom-based education alone is not effective, regardless of who delivers the intervention.

Across different outcomes and studies, the effect of school-based interventions was heterogeneous. The I 2 for STI incidence/prevalence ranged from 63.6 to 91.7%. Overall, non-RCTs reported bigger effects than RCTs; a pattern that could be due to residual and unmeasured confounding (Fewell et al. 2007) or publication bias (Kicinski 2013). The funnel plot for the outcome of STI incidence/prevalence shows that the distribution of non-RCTs is asymmetrical which indicates possible publication bias (Fig. 6). This may mean a bias toward more favorable results in the published literature since non-significant findings tend disproportionately not to be published (Kicinski 2013).
Fig. 6

Funnel plot for STI incidence/prevalence by subgroups of RCTs and non-RCTs studies. The distribution of non-RCTs is asymmetrical which indicates possible publication bias

The SSDP intervention was unique in that its most successful component spanned 5 years and started as early as first grade (full), and had an extensive focus on teacher and parent training. Thus, it was longer, aimed at earlier ages, and involved more of the important adults in the youth’s lives than the other interventions we examined. Although no conclusions can be drawn from such a small subsample, and our study was not designed to differentiate between more effective and less effective intervention models, the efficacy of interventions of longer duration may be an important focus of further research.

One limitation of the studies reviewed is that all relied on self-reported STI prevalence, with the exception of one which used STI surveillance data in the city where the condom availability program was run. HIV prevalence was not assessed even by self-report. To the extent possible, they should include biological tests for disease incidence, rather than relying on self-report. The body of research we examined is of limited quality. These are largely due to the nature of the programs, rather than from investigator’s errors or poor design choices. For example, the measurement of biological outcomes rather than reliance on students’ self-report would be desirable but probably infeasible on a sufficient scale. Similarly, certain desirable design features such as lack of blinding and allocation concealment may be impossible to fully implement in the context of these types of studies.

Given these inherent limitations, future research should emphasize use and reporting of as many GRADE criteria as possible, and include measure biological outcomes when possible. The next round of research should assess the program types that content and technical experts believe are the best candidates for showing benefit. The limited available research funds can thus be focused on evaluating school-based interventions that are most likely to benefit young people.

Another general issue confronting effectiveness research in this area is that the low prevalence of STIs and particularly of HIV in many school-aged populations requires very large (and thus expensive) samples in order to yield statistically meaningful results. Nevertheless, prevalence estimation, particularly when based on biological outcomes is a desirable goal, and may be attainable in some high prevalence settings, especially when focused on students who are sexually active, practicing unprotected sex, and older.

The studies we reviewed reported the frequency or history of intercourse, but the type of intercourse, vaginal, anal, oral, or a combination, was unspecified. This could be a source of the observed heterogeneity for the frequency of intercourse. Future studies would be strengthened if the type of sex or intercourse were asked in school survey instruments and their report of outcomes.

For two studies and three outcomes (Walter and Vaughan 1993—consistent condom use outcome, Coyle 2006—HIV testing, number of partners, and frequency of intercourse outcomes), the reported data were insufficient to calculate the standard error of the study effect measures so we could not include them all in our meta-analysis. In some cases (Coyle et al. 2001), only the numbers of participants recruited to the study was reported. However, for the meta-analysis, we required the final sample size that was used for the analysis (Higgins 2011). Such numbers even were not available to the authors of the papers whom we were able to contact. By sensitivity analysis, we imputed the analytical sample size that produced the same precision for the original effect measure reported in the paper. This strategy was used to maintain the maximize number of eligible studies in meta-analysis, that otherwise were excluded. Following CONSORT (for trials) (Kane et al. 2007) and STROBE (for observational studies) (Von Elm et al. 2014) guidelines will strengthen the reporting of epidemiological studies and prevent from incomplete data reporting.

The lack of evidence of effectiveness likely also lies in the complexity of factors that can influence behaviors and behavior change. While this review was not designed to address the question of what program components and other features may be linked with effectiveness, it is suggestive that the most successful intervention reviewed was of several years duration, and targeted teachers, parents, and peers to involve students in community service and bond them with teachers and parents, rather than focusing on school-based only sex education programs (Lonczak et al. 2002). Successful interventions may require lasting and multi-component approaches (Kirby 2002; Oringanje et al. 2016).

Quality of the Evidence

The research agenda going forward should address the problems we reported in the quality of evidence section by assessing, via well-powered RCTs, a number of the key program types for which there is consensus among content and technical experts for likely benefit. To the extent possible, they should include biological tests for disease incidence, rather than relying on self-report. Certain desirable design features such as lack of blinding and allocation concealment may be impossible to fully implement in the context of these types of studies.

Conclusion

This review is the first to assess the effectiveness of school-based interventions in reducing STIs and HIV in the USA. We found very few studies that met our inclusion criteria. The lack of published studies on this topic may be due the high costs required to conduct rigorous evaluations in this field, especially with sample sizes adequate to detect changes in HIV incidence. It is also possible that some programs with non-statistically significant results are less likely to be published. Of the eight studies reviewed, we found only two studies published from one intervention that had an effect on the primary outcome of interest, STI incidence, and none that reported HIV incidence. While we did find some positive effects on changes in STI-related knowledge and attitudes, we found little evidence that these changes decrease STI. The only effective intervention seemed to be one that covered multiple years, started early, and had multiple components. The variability in the interventions, study populations, settings, and outcomes reviewed make it difficult to identify the specific aspects of an intervention that may be most effective at reducing STIs and HIV among young people.

Notes

Compliance with Ethical Standards

Funding

This project was completed as part of Consortium for Assessment of Prevention Economics (CAPE), of Cooperative Agreement (Grant No: U38PS004649), NEEMA, with the Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Conflict of Interest

The authors declare that they have no conflict of interest.

Research Involving Human Participants and/or Animals

This article does not contain any studies with human participants or animals performed by any of the authors.

Informed Consent

As no human subjects were involved in the research undertaken to produce this article, no informed consent was required.

Supplementary material

11121_2017_830_MOESM1_ESM.docx (1.8 mb)
ESM 1 (DOCX 1806 kb)

References

  1. Brown, J., Sales, J., DiClemente, R., Salazar, L., Vanable, P., Carey, M., & Stanton, B. (2012). Predicting discordance between self-reports of sexual behavior and incident sexually transmitted infections with African American female adolescents: Results from a 4-city study. AIDS and Behavior, 16, 1491–1500. doi: 10.1007/s10461-012-0163-8.CrossRefPubMedPubMedCentralGoogle Scholar
  2. Brozek, J., Oxman, A., & Schünemann. (2008). GRADEpro [Computer software]: GRADE Working Group.Google Scholar
  3. Centers for Disease Control and Prevention. (2015a). HIV Surveillance Report, 2014; vol. 26. Retrieved from http://www.cdc.gov/hiv/library/reports/surveillance.
  4. Centers for Disease Control and Prevention. (2015b). STDs and HIV—CDC Fact Sheet. Retrieved from http://www.cdc.gov/std/hiv/stdfact-std-hiv.htm.
  5. Chin, H. B., Sipe, T. A., Elder, R., Mercer, S. L., Chattopadhyay, S. K., Jacob, V., & Santelli, J. (2012). The effectiveness of group-based comprehensive risk-reduction and abstinence education interventions to prevent or reduce the risk of adolescent pregnancy, human immunodeficiency virus, and sexually transmitted infections: Two systematic reviews for the Guide to Community Preventive Services. American Journal of Preventive Medicine, 42, 272–294. doi: 10.1016/j.amepre.2011.11.006.CrossRefPubMedGoogle Scholar
  6. Clark, L., Brasseux, C., Richmond, D., Getson, P., & D'Angelo, L. (1997). Are adolescents accurate in self-report of frequencies of sexually transmitted diseases and pregnancies? The Journal of Adolescent Health, 21, 91–96.CrossRefPubMedGoogle Scholar
  7. Cochrane (2011). Retrieved from http://training.cochrane.org/calculator.
  8. Cohen, D. A., Nsuami, M., Martin, D. H., & Farley, T. A. (1999). Repeated school-based screening for sexually transmitted diseases: A feasible strategy for reaching adolescents. Pediatrics, 104, 1281–1285.CrossRefPubMedGoogle Scholar
  9. Coyle, K., Basen-Engquist, K., Kirby, D., Parcel, G., Banspach, S., Collins, J., et al. (2001). Safer choices: Reducing teen pregnancy, HIV, and STDs. Public Health Reports, 116, 82–93.CrossRefPubMedPubMedCentralGoogle Scholar
  10. Coyle KK, Kirby DB, Robin LE, Banspach SW, Baumler E, Glassman JR. All4You! A randomized trial of an HIV, other STDs, and pregnancy prevention intervention for alternative school students. AIDS Education and Prevention 2006 Jun;18(3):187-203. PubMed PMID:16774462.Google Scholar
  11. EndNote Corp. (2013). [Computer software] EndNote X7: Thomson Reuters. Retrieved from http://endnote.com/product-details/X7
  12. Fewell, Z., Davey Smith, G., & Sterne, J. (2007). The impact of residual and unmeasured confounding in epidemiologic studies: A simulation study. American Journal of Epidemiology, 166, 646–655.CrossRefPubMedGoogle Scholar
  13. Fonner, V. A., Armstrong, K. S., Kennedy, C. E., O’Reilly, K. R., & Sweat, M. D. (2014). School based sex education and HIV prevention in low- and middle-income countries: A systematic review and meta-analysis. PloS One, 9, e89692. doi: 10.1371/journal.pone.0089692.CrossRefPubMedPubMedCentralGoogle Scholar
  14. Glassman, J., Coyle, K., Anderson, P., & Baumler, E. (2014). Accuracy of self-report measures of sexual behaviors among middle school students: Association between lying on surveys and peer norms. Abstract retrieved from Society for Prevention Research: 22nd Annual Meeting Abstracts database.Google Scholar
  15. Goesling, B., Colman, S., Trenholm, C., Terzian, M., & Moore, K. (2014). Programs to reduce teen pregnancy, sexually transmitted infections, and associated sexual risk behaviors: A systematic review. The Journal of Adolescent Health, 54, 499–507. doi: 10.1016/j.jadohealth.2013.12.004.CrossRefPubMedGoogle Scholar
  16. Guyatt, G. H., Oxman, A. D., Schunemann, H. J., Tugwell, P., & Knottnerus, A. (2011). GRADE guidelines: A new series of articles in the journal of clinical epidemiology. Journal of Clinical Epidemiology, 64, 380–382. doi: 10.1016/j.jclinepi.2010.09.011.CrossRefPubMedGoogle Scholar
  17. Hale, D. R., Fitzgerald-Yau, N., & Viner, R. M. (2014). A systematic review of effective interventions for reducing multiple health risk behaviors in adolescence. American Journal of Public Health, 104, e19–e41. doi: 10.2105/ajph.2014.301874.CrossRefPubMedPubMedCentralGoogle Scholar
  18. Hawkins, J. D., Kosterman, R., Catalano, R. F., Hill, K. G., & Abbott, R. D. (2008). Effects of social development intervention in childhood 15 years later. Archives of Pediatrics & Adolescent Medicine, 162, 1133–1141. doi: 10.1001/archpedi.162.12.1133.CrossRefGoogle Scholar
  19. Higgins, J. (2011). Cochrane handbook for systematic reviews of interventions, Version 5.1.0 [updated March 2011]. The Cochrane Collaboration. Available from www.cochrane-handbook.org .
  20. Higgins, J., & Green, S. (2011). Cochrane handbook for systematic reviews of interventions (J. Higgins & S. Green Eds. 5.1.0 ed.): The Cochrane Collaboration.Google Scholar
  21. Higgins, J., Altman, D., & Sterne, J. (2011). Assessing risk of bias in included studies In J. Higgins & S. Green (Eds.), Cochrane handbook for systematic reviews of interventions (Version 5.1.0 ed.): The Cochrane Collaboration.Google Scholar
  22. Hill, K. G., Bailey, J. A., Hawkins, J. D., Catalano, R. F., Kosterman, R., Oesterle, S., & Abbott, R. D. (2014). The onset of STI diagnosis through age 30: Results from the Seattle Social Development Project intervention. Prevention Science, 15, S19–S32. doi: 10.1007/s11121-013-0382-x.CrossRefPubMedGoogle Scholar
  23. Johnson, B. T., Scott-Sheldon, L. A., Huedo-Medina, T. B., & Carey, M. P. (2011). Interventions to reduce sexual risk for human immunodeficiency virus in adolescents: A meta-analysis of trials, 1985-2008. Archives of Pediatrics & Adolescent Medicine, 165, 77–84. doi: 10.1001/archpediatrics.2010.251.Google Scholar
  24. Kane, R. L., Wang, J., & Garrard, J. (2007). Reporting in randomized clinical trials improved after adoption of the CONSORT statement. Journal of Clinical Epidemiology, 60, 241–249. doi: 10.1016/j.jclinepi.2006.06.016.CrossRefPubMedGoogle Scholar
  25. Kicinski, M. (2013). Publication bias in recent meta-analyses. PloS One, 11, e81823.CrossRefGoogle Scholar
  26. Kirby, D. (2002). The impact of schools and school programs upon adolescent sexual behavior. Journal of Sex Research, 39, 27–33.CrossRefPubMedGoogle Scholar
  27. Kirby, D. B., & Brown, N. L. (1996). Condom availability programs in U.S. schools. Family Planning Perspectives, 28, 196–202.CrossRefPubMedGoogle Scholar
  28. Kirby, D., Korpi, M., Adivi, C., & Weissman, J. (1997). An impact evaluation of project SNAPP: An AIDS and pregnancy prevention middle school program. AIDS Education and Prevention, 9, 44–61.PubMedGoogle Scholar
  29. Lemieux, A. F., Fisher, J. D., & Pratto, F. (2008). A music-based HIV prevention intervention for urban adolescents. Health Psychology, 27, 349–357. doi: 10.1037/0278-6133.27.3.349.CrossRefPubMedGoogle Scholar
  30. Lonczak, H. S., Abbott, R. D., Hawkins, J. D., Kosterman, R., & Catalano, R. F. (2002). Effects of the Seattle social development project on sexual behavior, pregnancy, birth, and sexually transmitted disease outcomes by age 21 years. Archives of Pediatrics & Adolescent Medicine, 156, 438–447.CrossRefGoogle Scholar
  31. Mason-Jones, A. J., Crisp, C., Momberg, M., Koech, J., De Koker, P., & Mathews, C. (2012). A systematic review of the role of school-based healthcare in adolescent sexual, reproductive, and mental health. Systematic Reviews, 1, 1–12. doi: 10.1186/2046-4053-1-49.CrossRefGoogle Scholar
  32. Mavedzenge, S. N., Luecke, E., & Ross, D. A. (2014). Effective approaches for programming to reduce adolescent vulnerability to HIV infection, HIV risk, and HIV-related morbidity and mortality: A systematic review of systematic reviews. Journal of Acquired Immune Deficiency Syndromes, 66, S154–S169. doi: 10.1097/QAI.0000000000000178.CrossRefPubMedGoogle Scholar
  33. National Center for Education Statistics. (2015). Current population survey. Retrieved from https://nces.ed.gov/programs/coe/.
  34. Oringanje, C., Meremikwu, M. M., Eko, H., Esu, E., Meremikwu, A., & Ehiri, J. E. (2009). Interventions for preventing unintended pregnancies among adolescents. The Cochrane Database of Systematic Reviews, (4), Cd005215. doi: 10.1002/14651858.CD005215.pub2
  35. Oringanje, C., Meremikwu, M., Eko, H., Esu, E., Meremikwu, A., & Ehiri, J. (2016). Interventions for preventing unintended pregnancies among adolescents. The Cochrane Database of Systematic Reviews, 2, Cd005215.Google Scholar
  36. Paul-Ebhohimhen, V. A., Poobalan, A., & van Teijlingen, E. R. (2008). A systematic review of school-based sexual health interventions to prevent STI/HIV in sub-Saharan Africa. BMC Public Health, 8, 4. doi: 10.1186/1471-2458-8-4.CrossRefPubMedPubMedCentralGoogle Scholar
  37. Picot, J., Shepherd, J., Kavanagh, J., Cooper, K., Harden, A., Barnett-Page, E., et al. (2012). Behavioural interventions for the prevention of sexually transmitted infections in young people aged 13-19 years: A systematic review. Health Education Research, 27, 495–512. doi: 10.1093/her/cys014.CrossRefPubMedGoogle Scholar
  38. PROSPERO. (2016). International prospective register of systematic reviews. Retrieved from http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42016033007
  39. Spoth, R., Redmond, C., Shin, C., Greenberg, M. T., Feinberg, M. E., & Trudeau, L. (2017). PROSPER delivery of universal preventive interventions with young adolescents: long-term effects on emerging adult substance misuse and associated risk behaviors. Psychological Medicine, 1–14. doi: 10.1017/s0033291717000691
  40. Underhill, K., Montgomery, P., & Operario, D. (2008). Abstinence-plus programs for HIV infection prevention in high-income countries. The Cochrane Database of Systematic Reviews, 1, CD007006. doi: 10.1002/14651858.CD007006.Google Scholar
  41. Von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gotzsche, P. C., & Vandenbroucke, J. P. (2014). The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. International Journal of Surgery, 12, 1495–1499. doi: 10.1016/j.ijsu.2014.07.013.CrossRefGoogle Scholar
  42. Walter, H. J., & Vaughan, R. D. (1993). AIDS risk reduction among a multiethnic sample of urban high school students. The Journal of the American Medical Association, 270, 725–730.CrossRefPubMedGoogle Scholar
  43. Wretzel, S. R., Visintainer, P. F., & Pinkston Koenigs, L. M. (2011). Condom availability program in an inner city public school: Effect on the rates of gonorrhea and chlamydia infection. The Journal of Adolescent Health, 49, 324–326. doi: 10.1016/j.jadohealth.2010.12.011.CrossRefPubMedGoogle Scholar

Copyright information

© Society for Prevention Research 2017

Authors and Affiliations

  • Ali Mirzazadeh
    • 1
    • 3
  • M. Antonia Biggs
    • 2
  • Amanda Viitanen
    • 3
  • Hacsi Horvath
    • 1
    • 3
    • 4
  • Li Yan Wang
    • 5
  • Richard Dunville
    • 5
  • Lisa C. Barrios
    • 5
  • James G. Kahn
    • 3
    • 4
    • 6
  • Elliot Marseille
    • 7
  1. 1.School of Medicine, Department of Epidemiology and BiostatisticsUniversity of California, San FranciscoSan FranciscoUSA
  2. 2.Department of Obstetrics, Gynecology and Reproductive Sciences, Advancing New Standards in Reproductive Health (ANSIRH), Bixby Center for Global Reproductive HealthUniversity of California, San FranciscoOaklandUSA
  3. 3.Global Health SciencesUniversity of California, San FranciscoSan FranciscoUSA
  4. 4.Philip R. Lee Institute for Health Policy StudiesUniversity of California, San FranciscoSan FranciscoUSA
  5. 5.Division of Adolescent and School HealthU.S. Centers for Disease Control and Prevention, Office of Infectious Diseases, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB PreventionAtlantaUSA
  6. 6.Global Health Economics ConsortiumUniversity of California, San FranciscoSan FranciscoUSA
  7. 7.Health Strategies InternationalOaklandUSA

Personalised recommendations