Assessing the role of school-based sex education in sexual health behaviours: a systematic review

Abstract

Objective

To identify and synthesise evidence on the role of school-based sex education interventions and to use a Behaviour Change Techniques (BCTs) taxonomy to identify behaviour change techniques on sexual health behaviours.

Methods

The systematic review was informed by the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis and included the components of PICOS: participant/population, interventions, comparisons, outcomes and study design. Five electronic databases were searched up to February 2023 including PUBMED, Cochrane Central Register of Controlled Trials, ERIC, Web of Science Core Collection and PsycINFO. Methodological quality was assessed using the Effective Public Health Practice Project, Quality Assessment tool. Results were presented as a narrative synthesis.

Results

Of the 1387 studies identified, twenty-seven studies met the inclusion criteria. Studies examined sexual health behaviours such as condom usage, frequency of sexual activity, initiation of sexual activity and number of sexual partners. More than half studies (56%) reported a statistically significant change on one or more sexual health behaviour outcomes. Nine (out of 93) BCTs were identified with the most used BCTs being information about health consequences and social and emotional consequences, demonstration of behaviour, behavioural practice/rehearsal and instructions on how to perform the behaviour.

Conclusion

School-based sex education interventions can be effective in promoting positive sexual health behaviours. The findings of this review provide understanding of such interventions in shaping sexual health behaviours. These also offer evidence-based knowledge for researchers, educators and policy makers in understanding how they can support future development of school-based sex education programmes.

IMPACT STATEMENT

The findings of this systematic review provide an important understanding of the role of school-based sex education on shaping sexual health behaviours. Findings may be used to guide the design of interventions targeting positive behavioural outcomes and support policymakers in developing informed strategies to shape sexual health outcomes in young people. Ultimately, this can contribute to promoting the overall health and well-being of young individuals.

Keywords:

• Behaviour change techniques
• sexual health
• sexual behaviour
• secondary school, adolescents, interventions

REVIEWING EDITOR:

• Daryl O’Connor
• University of Leeds, UK

SUBJECTS:

• Health Psychology
• Secondary Education
• Health & Society
• Public Health Policy and Practice

Introduction

Sexually Transmitted Infections (STIs) remain a major public health concern across the globe. There are four curable sexually transmitted infections, including syphilis, gonorrhoea, chlamydia and trichomoniasis and four incurable but treatable STIs, including hepatitis B, herpes simplex virus (HSV or herpes), Human Immunodeficiency Virus (HIV), and Human Papillomavirus (HPV) (World Health Organisation, 2022). Approximately, one million (STIs) are acquired each day, with a large amount of these new cases occurring among young people aged between 15–25 years old (World Health Organisation, 2022). High-risk sexual behaviours (e.g. not using a condom) commonly begin in adolescence and can negatively impact young people’s health by exposing them to STIs, which can have detrimental health, social and economic implications (Ritchwood et al., 2015; Ybarra et al., 2016). Untreated STIs, for instance, chlamydia, can have serious consequences on young people such as infertility, fetal and neonatal deaths, cancer and HIV transmission (Denford et al., 2017; Workowski et al., 2021). Given the significance of STIs, it is vital to implement intervention strategies to address STIs, support the sexual health of young people and meet public health goals (Ferreira et al., 2013; Lopez et al., 2016).

School-based sex education interventions

Schools comprise important environments for promoting sexual health and reproductive programmes as they are one of the sources for obtaining sexual health information in young people (Young et al., 2018). School-based sex education interventions are implemented as a part of the formal curriculum and they have the potential to educate young people to make healthy decisions about their sexual lives (Kirby & Laris, 2009). The implementation of sex education interventions can be complex and multi-faceted with schools adopting varying approaches for delivering such interventions (Keating et al., 2018). For instance, schools could adopt an abstinence-only approach, which is informed by the notion that abstinence is the best way to prevent pregnancy and STIs. Abstinence-only sex education interventions include content designed to teach behaviours, attitudes and skills consistent with abstinence and on managing pressures before marriage (Blake et al., 2001; Denford et al., 2017; Keating et al., 2018). Conversely, Comprehensive Sexuality Education (CSE) aims to provide young people with opportunities to acquire comprehensive, accurate and evidence-based as well as age-appropriate information on sexual health and sexuality as part of a continuing sexual health education process. This approach has been found to include content on knowledge about STIs/HIV transmission, prevention and risk perception, attitudes towards safe sex and condom use (Fernandes & Junnarkar, 2019; Kantor & Lindberg, 2020; Leung et al., 2019).

Previous systematic reviews have been conducted to assess the effectiveness of school-based sex education interventions (e.g.Kirby et al., 2005; Kirby et al., 2007; Jones et al., 2009). Kirby and colleagues (Kirby et al., 2005) reviewed 42 CSE programmes in the USA. This review found no evidence to support that CSE programmes reduce self-reported STIs and unwanted pregnancies. Another review of 83 CSE interventions that were conducted in developing countries and implemented in schools, reported that under half of these interventions had a positive effect on at least one of the behavioural outcomes (Kirby et al., 2007). A review that included 75 trials of sex education for young adolescents aged 11 – 19 years showed that abstinence-only programmes were effective in improving knowledge and attitudes, but no effects were demonstrated for sexual health behaviours (Underhill et al., 2008). Another review that included 13 abstinence-only trials with seven of them conducted in a school setting concluded that these interventions were not effective in reducing unprotected sex, frequency of sex or increasing condom use (Underhill et al., 2007). On the contrary, they found that participants reported an increase in STIs and frequency of sex.

Furthermore, Denford and colleagues (Denford et al., 2017) conducted a comprehensive review of reviews on the effectiveness of school-based sex education interventions aimed at improving sexual health. They found that these reviews were of weak study quality and reported inconsistent findings for sexual health behaviours. Most importantly, these authors stated that while school-based sex education interventions can be effective, the development of these interventions should focus on the design, content and implementation characteristics as these aspects appear to be associated with the greater effectiveness of school-based sex education interventions (Denford et al., 2017).

The present systematic review

Previous studies examining the effectiveness of school-based sex education interventions are producing diverse findings. This pattern is continuously observed across the existing literature and studies do not appear to be providing new and meaningful contributions concerning new knowledge about sexual health behaviours. This leaves the existing literature with a series of studies that fail to build upon each other with meaningful findings and to generate new insights concerning understanding school-based sex education on shaping sexual health behaviours. Therefore, new approaches are required to further our understanding of school-based sex education interventions and how they promote positive sexual health behaviours. The aim of this systematic review was to identify and synthesise available evidence on school-based sex education interventions and use a Behaviour Change Technique taxonomy (v1) to identify behaviour change techniques applied in relation to sexual health behaviours.

Methods

The present systematic review was informed by the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and included the components of PICOS: participant/population, interventions, comparisons, outcomes and study design (Page et al., 2021). This study was registered with Open Science Framework (identifier; 10.17605/OSF.IO/86SPZ)

Search strategy

A systematic search of the literature was conducted up to 28^th February 2023. Electronic searches were conducted in the following databases PUBMED, Cochrane Central Register of Controlled Trials (CENTRAL), ERIC, Web of Science Core Collection and PsycINFO, using the strategy presented in supplementary material (S1).

Eligibility criteria

Participants

Sex education interventions targeting adolescents in secondary school were eligible for inclusion. Secondary school is defined as the period after primary education (elementary level) and before college/university (tertiary level) (UNESCO, 2011).

Intervention and control groups

School-based sex education interventions were defined as interventions that were delivered in a secondary school (middle/high school) within a classroom setting. Studies that evaluated these interventions on shaping sexual health behaviours and compared to a control group (receiving the existing school-based sex education programme, a modified version of the intervention, or no sex education programme) were included.

Sexual health outcomes

Studies were included if they measured outcomes of sex education interventions using standardised and non-standardised measures and descriptive/frequencies for sexual health behaviours. Any sexual health behaviour targeted by sex education programmes (e.g. STI and pregnancy rates, delay in onset of intercourse, reduction in the frequency of intercourse and contraception use) was eligible for inclusion. Studies that focused on outcomes of sexual health knowledge, values and attitudes were excluded. However, if these studies included a secondary measure of sexual health behaviours (e.g. condom use, STI testing), then they were deemed eligible for inclusion.

Study design

Quantitative studies that employed randomised control trials (RCTs), non-randomised control trials (non-RCTs) (e.g. cohort studies and case-control studies) and quasi-experimental designs were included.

Publication date

The timeframe for searching the literature (2006 onwards) was informed by the updated definition of sexual health provided by the World Health Organization (World Health Organization, 2006). Supplementary information (S2) summarises the inclusion and exclusion criteria.

Screening

Articles that were retrieved through the searches were imported into the Covidence Systematic Review software (Veritas Health Innovation, 2019)and duplicates were removed. Title and abstracts of identified articles (n = 1387) were screened by the first reviewer, and 100% checked by the second reviewer. Disagreements were resolved through discussion. Inter-reviewer agreement for the title and abstract screening was 90%. Full-text review of the included articles (n = 107) was carried out independently by two reviewers. Disagreements were resolved via discussion. Inter-reviewer agreement for the full-text review phase was 80%. The studies selected at the end of this process (n = 27) were included for data extraction and the reference section of these studies was reviewed for potentially eligible studies to be included. Figure 1 presents the PRISMA flow diagram.

Figure 1. PRISMA flow diagram depicts study identification via databases.

A flow chart showing the flow of information through the different stages of a systematic review.

Data extraction

Data extraction was conducted using the Cochrane Developmental, Psychosocial and Learning data collection form for intervention reviews: RCTs and non-RCTs (Higgins, 2011). This tool includes different sections for data extraction (e.g. characteristics of included studies, methods, participants) and provides a form to meet the needs of all systematic reviews. Data extraction and quality assessment were conducted by the first reviewer, with a random 20% extracted by a second reviewer.

Quality appraisal

The methodological quality of included studies was assessed by two reviewers who independently rated the methodology by using the Effective Public Health Practice Project, Quality Assessment tool (EPHPP) (Armijo-Olivo et al., 2012). This tool is useful for systematic reviews aiming to assess the quality of studies with the capacity to evaluate both RCTs and non-RCTs (Armijo-Olivo et al., 2012). It consists of six component ratings that assess the quality of each study on selection bias, study design, confounders, blinding data collection method, withdrawals and dropouts. A quality rating of strong (1), moderate (2) or weak (3) was assigned to each study by two reviewers.

Effectiveness assessment

For each included study, school-based sex education interventions were deemed effective if there was a statistically significant difference between the intervention and comparison groups for at least one of the objective measures of sexual health behaviours (e.g. scores etc). To assess statistical significance, this review selected p < 0.05 as the criterion for assessing or reporting statistically significance.

Behaviour change Technique (BCT) taxonomy coding

The Behaviour Change Technique (BCT) taxonomy (Michie et al., 2011) was used to identify key aspects of the interventions that were designed to enhance sexual health behaviour outcomes. To our knowledge this is the first time that this taxonomy is applied to evaluate the effectiveness of sex education interventions for sexual health behaviours. The BCT taxonomy includes 93 behaviour change techniques and these techniques are categorised into 16 clusters based on their functions and characteristics. BCTs are defined as an observable, replicable and irreducible component of an intervention designed to alter or redirect causal processes that regulate behaviour. The BCT categorises the smallest identifiable components (active ingredients) of interventions that are designed to change behaviour. Descriptions of interventions and control groups were independently coded by two coders into BCT components as per the BCT taxonomy V1 (Michie et al., 2011). Both coders had completed the online course for using the BCT Taxonomy V1 (https://www.bct-taxonomy.com). A BCT had to be explicitly present to be simply coded as present or absent. Coding differences were resolved through discussion and unresolved agreements were sought from a third party.

Data synthesis

Because the intervention designs and the outcome measures of the included studies were highly heterogeneous, a narrative rather than a meta-analytical approach was deemed appropriate (Popay et al., 2006).

Results

Characteristics of included studies

In total, the studies included 36,364 participants (N = 36,364), with the sample size ranging from 24 to 9,372 participants. Table 1 summarises the 27 included studies. Interventions were heterogeneous, differing in delivery period, agent and mode. Of the included studies the number of intervention sessions delivered ranged from 2 to 25. The interventions were delivered by teachers (n = 9, 33%), followed by the researcher of the project (n = 6, 22%), health educator (n = 5, 18%) and peer educator (n = 4, 15%). For the remaining three studies (11%) these interventions were delivered by volunteers and nurses and one study did not specify who delivered the intervention. All studies compared a school-based sex education intervention with a control group receiving no intervention (n = 14, 52%) or receiving an existing school-based sex education programme (n = 13, 48%). The majority of studies were RCTs (n = 13, 48%). The remaining studies employed quasi-experimental (n = 6, 22%), pre and post-test (n = 4, 15%), longitudinal pre and post (n = 1, 4%), and cross-sectional (n = 1, 4%) research designs. One study did not report the design apart from conducting a survey to evaluate a three-year sex education programme using intervention and control groups. Most studies were conducted in the United States (n = 12, 44%) followed by South Africa (n = 3, 11%), Nigeria (n = 2, 7%). Each remaining study (n = 10, 37%) was conducted in Mexico, China, Northern Ireland, Scotland, Ethiopia, Uganda, Bangkok, Swaziland, Northern Malawi, and Mongolia.

Table 1. Characteristics of studies included in the review.

Study/Country/Quality Rating	Sample characteristics	Intervention characteristics	Sexual health behaviour outcomes	Length of follow up & Results	Effective Y/N
1. Menna et al. (2015)Design: Quasi-experimental Country: Ethiopia Quality: Moderate	Students/Adolescents (Grade 11) No mean age reported Age range 15 to 18 Pre-test intervention group: (n = 240, 85.7%); Pre-test control group: (n = 234, 83.6%) Post-test intervention group (n = 232, 82.9%); Post-test control group (n = 223, 79.6%)	Setting: Secondary school, Group level, face to face, classroom setting Duration: 2 sessions per week, 40 mins for both intervention and control Delivery agent: Peer educators Intervention: No information reported; Control: No information reported.	Target behaviour(s): Consistent condom usage in previous 12 months	Time points: Baseline and 3 months (End of Intervention) and 3 months (follow up) Length of follow up: Short-term <12 months	Yes
2. Musiimenta (2012) Design: Controlled pre & post test Country: Uganda Quality: Moderate	Secondary school students, all years recruited No mean age reported Age range 11 – 16 years Pre-test (N = 300) Intervention group (n = 152); Control group (n = 148) Post-test (N = 292) Intervention. Group (n = 146); Control group (n = 146)	Setting: Secondary school, Group level, face to face, classroom setting Duration: 14 lessons, no time allocation reported Delivery agent: Teachers Intervention: World Starts With Me, sexuality and HIV/AIDS intervention; Control: Not reported	Target behaviour(s): Abstinence, Number of sexual partners, Condom use	Timepoints: Baseline and 10 months (End of intervention) Length of follow up: Short-term <12 months	Yes
3. Berglas et al. (2016)Design: cRCT Country: USA Quality: Moderate	High school students, Grade 9 M age 14.2 (SD = 0.62) No age range reported N = 972 Intervention group (n = 412); Control group (n = 329)	Setting: High school, Group level, face to face, classroom setting Duration: 12 sessions, during regular class times Delivery agent: Health educators Intervention: The Sexuality Education Initiative intervention.Control: Received sexual health services and the control classroom curriculum.	Target behaviour(s): Primary outcomes: i) Pregnancy risk, ii) STI risk, iii) Multiple sexual partners, iv) Ever using sexual health services Secondary outcomes: (1) Ever engaging in vaginal or anal sex, (2) Ever engaging in oral sex, (3) Engaging in vaginal or anal sex in the previous 3 months, (4) Engaging in oral sex in the previous 3 months, (5) Using a condom at last vaginal or anal sex, (6) Using birth control or condoms at last vaginal or anal sex, and (7) Currently carrying a condom (e.g. in a pocket)	Timepoints: Baseline, intervention and 12 months follow up Length of follow up: Long-term: ≥12 months Results: Condom use effect size OR = 2.71, 95% CI = 1.44 – 5.09	Yes
4. Denny and Young (2006) Design: Quasi-experimental Country: USA Quality: Weak	High school students, Grades 9 or above No mean age and age range reported Pre-test: Treatment group (n = 226); Control group (n = 111) Post-test: Treatment group (n = 195); Control group (n = 92) 18 month follow up: Treatment group (n = 168); Control group (n = 76)	Setting: High school, group level, face to face, classroom setting Duration: 5 weeks, 24 sessions, no time allocation reported Delivery agent: Teachers (with 3.5 hours training provided) Intervention: Sex can wait curriculum; Control: Schools had existing sexuality education programmes.	Target behaviour(s): Yes/No items asking whether students had ever participated in sexual intercourse and whether they had done so in the last 30 days.	Timepoints: Baseline, 5 weeks (End of Intervention) and 18 months follow up Length of follow up: Long-term: ≥12 months	Yes
5. Esere (2008) Design: Quasi- experimental Country: Nigeria Quality: Moderate	Secondary school students, all school years recruited M age = 15.5 Age range 13–19 years Treatment Group A: Male (n = 6) & Female (n = 6) Control Group B: Male (n = 7) & Female (n = 5)	Setting: Secondary school, group level, face to face, classroom setting Duration: Eight sessions intensity, one session per week, no time allocation reported Delivery agent: Researcher of project Intervention: Sex Education Programme; Control: No treatment provided	Target behaviour(s): Self-reported exposure to 1) sexually transmitted diseases, 2) multiple sex partners, 3) anal sex, 4) oral sex, 5) non-use of condom. (Cronbach’s 0.73).	Timepoints: Baseline, 8 weeks (End of Intervention) Length of follow up: Not reported	Yes
6. Li et al. (2011) Design: Quasi-experimental Country: China Quality: Moderate	High school students, all school years recruited (M = 16.3, SD = 0.66) Age range 15–19 years Total sample: N = 304 students. Baseline: Intervention group (n = 140); Control group (n = 164) Follow up: Intervention group (n = 131); Control group (n = 156) 94.4% of the baseline cohort participated in the 6-month post-intervention survey (N = 287).	Setting: High school, Group level, face to face, classroom setting Duration: 8 session, delivery time of 12 hours (90 minutes per session) Delivery agent: Mix delivery agent - Four graduate students and seven faculty members from University in Nanjing Intervention: Cultural adaptation of the FOK HIV prevention programme.Control: Not reported	Target behaviour(s): Sexual intercourse in the past 6 months	Timepoints: Baseline and 6 months follow up Length of follow up: Short-term <12 months	No
7. Thato et al. (2008) Design: Quasi-experimental Country: Bangkok Quality: Moderate	High school students, Grades 9, 10, 11 Intervention group M age = 15.67; Control group M age = 15.93 Age range 13–18 years Total sample (N = 552) 3 month follow up: Intervention group (n = 230); Control group (n = 261) 6 month follow up: Intervention group (n = 230); Control group (n = 189)	Setting: High school, Group level Face to face sessions, classroom setting Duration: 6 weeks, one-hour modules (5 module total), divided equally into two consecutive 3-hour classes taught weekly by the first author Delivery agent: Researcher of project Intervention: CSCSEP Curriculum.Control: Not reported	Target behaviour(s): Condom use among sexually active and consistent condom use and frequency of sexual activity.	Timepoints: Baseline, 6 weeks (End of Intervention) and 3, 6 month follow up Length of follow up: Short-term <12 months Results: Frequency of sexual activity effect size OR 0.80, 95% CI = 0.84 – 0.95	Yes
8. Burnett et al. (2011)Design: RCT Country: Swaziland Quality: Moderate	High school students, Grades 9 & 11 No mean age and age range reported N = 204 (101 males and 103 females) 66.5% completed surveys at pre and post-test (N = 135).	Setting: High school, group level, face to face, workshop setting Duration: 13 (weeks) half-day Saturday sessions, with 1 hour per module (4 modules per week) (4 hours total each week) Delivery agent: Health educators Intervention: It’s Our Future Too! Intervention. Control: Not reported	Target behaviour(s): 1) Ever had sex 2) Condom use at last sex	Timepoints: Baseline and 13 weeks (End of Intervention) Length of follow up: Short-term <12 months	No
9. LaChausse (2006) Design: Longitudinal experimental design Country: USA Quality: Moderate	High school students, Grade 9 No mean age or age range reported Pre-test: Intervention group (n = 216); Control group (n = 137) 6 month follow up: Intervention group (n = 174); Control group (n = 113)	Setting: High school setting, group level, face to face, classroom setting Duration: Six sessions, 45 minutes time allocation Delivery agent: Teachers Intervention: The Positive Prevention HIV/STD Curriculum.Control: Not reported	Target behaviour(s): Frequency of sexual behaviours, condom use, initiation of sexual intercourse	Timepoints: Baseline, 1 month and 6 month follow up Length of follow up: Short-term <12 months	Yes
10. Walker et al. (2006)Design: cRCT Country: Mexico Quality: Moderate	High school students, 1st year from 40 public high schools No age range reported N = 10,954 Post Intervention group (n = 9372) N = 7308 One year follow up (June 2003) Condom promotion group: Females (M age = 16.8) Males (M age = 16.9) Condom promotion group with emergency contraception: Females (M = 16.7) Males (M = 16.9) Control group: Females (M age = 16.7) Males (M age = 16.8)	Setting: High school, group level, face to face, classroom setting Duration: 15 weeks consisting of 30 hours and the contraception arm received two extra hours over 15 weeks resulting in a total of 32 hours Delivery agent: Teachers Intervention: HIV prevention course that promoted condom use and emergency contraceptionControl: Standardised sex education course implemented by the Ministry of Education.	Target behaviour(s): Primary outcome measure was reported condom use.	Timepoints: Baseline, 15 weeks (end of intervention), 12 and 16 month follow up Length of follow up: Long-term: ≥12 months	No
11. Daley et al. (2019)Design: Pair Matched cRCT Country: USA Quality: Moderate	High school students, 9^th grade, No mean age reported Age range 14–16 yrs. Total sample N = 4,789 Intervention group (n = 2,198); Control group (n = 2,590)	Setting: High school, group level, face to face, classroom setting Duration: 25 weekly sessions (over 9 months; 20 hours total) Delivery agent: Health educators Intervention: TOP Level 2 curriculum Control: Standard school sexual health curriculum.	Target behaviour(s): 1) Ever had sex (ever sex), 2) Ever having been pregnant or caused a pregnancy (ever pregnant), 3) Sexual intercourse in the past 3 months (recent sex), & 4) Condom use in the past 3 months (recent condom use).	Timepoints: Baseline, 12 months follow up Length of follow up: Long-term: ≥12 months	No
12. Tucker et al. (2007) Design: Cross sectional before & after Country: Scotland Quality: Moderate	Secondary school students, Year 3 and Year 4 Median age 14 years and 6 months. No age range reported Breakdown of sample not reported	Setting: Secondary schools, group level, face to face, classroom setting Duration: 20 sessions total across 2 school years (10 sessions in 3rd year and 10 sessions in 4th year) and time allocation not specified Delivery agent: Nurses Intervention: Healthy Respect’s Sexual Health And Relationships Education (SHARE). Control: Standard sexual health education programme	Target behaviour(s): Self-reported sexual intercourse at age <16 years.	Timepoints: Not reported Length of follow up: Not reported	No
13. Mwale and Muula (2019)Design: Quasi-experimental Country: Northern Malawi Quality: Moderate	Secondary school students, all years recruited. Intervention group (M age = 16.2); Control group (M age = 16.1) No age range reported N = 360 (Intervention group (n = 158); Control group (n = 202)).	Setting: Secondary school, group level, face to face, classroom setting Duration: 5 sessions over 3 months. No time allocation reported Delivery agent: 6 peer educators trained to facilitate and allowed to practice their roles before the intervention was implemented. Intervention: Peer education risk reduction intervention Control: Standard-as-usual Life skills regimen already provided to adolescents and young people in schools.	Target behaviour(s): Primary outcome: HIV risk reduction Secondary outcomes: Abstinence, Delayed debut, Reductions in unprotected sex, Correct and consistent condom use, Faithfulness to one partner, Reductions in multiple partnerships.	Timepoints: Baseline, 3 months (End of Intervention), 8 month follow up Length of follow up: Short-term <12 months Results: Frequency of sex effect size OR = 0.04, 95%, CI = 0.01–0.20Condom use effect size OR = 0.02; 95%, CI = 0.01–0 .04 Number of sexual partners effect size OR = 0.04, 95%, CI = 0.02–0.11	Yes
14. James et al. (2006)Design: Pre-test and post-test experimental control Country: South Africa Quality: Moderate	Secondary school students, two 9^th grade classes M age = 15.5 Age range 12 – 21 years Total sample: N = 1141 Baseline: N = 844. 6 month follow up (dropout rate 26%) and 4 month follow up: (N = 768)	Setting: Secondary school, group level, face to face, classroom setting Duration: 20-week duration. No time reported Delivery agent: Teachers Intervention: Life Skills Programme focusing on HIV and AIDS prevention. Control: Received odd lessons about aspects of HIV and AIDS education in a nonstructured format and in some cases celebrated awareness days on the topic.	Target behaviour(s): Self-report 1) whether they were ever sexually active (0 = no, 1 = yes), 2) sexually active in the recall period (6 months at T1 and T2, 3 months at T3), and 3) whether they used a condom the last time they had sex (0 = no, 1 = yes; only measured at T2 and T3).	Timepoints: Baseline, 20 weeks (End of Intervention), 6 and 10 month follow up Length of follow up: Short-term <12 months Results: Frequency of sexual activity effect sizes (B = –.05, SE = .26, Wald (1, 657) = 0.03, p = .86, OR = 1.04)Condom use effect sizes (B = –.21, SE = .41, Wald (1, 157) =0.27,p= .60,OR=.81).	No
15. Mba et al. (2007)Design: Pre and post intervention Country: Nigeria Quality: Weak	Secondary school students, all school years recruited M age = 14.3 years Age range 10 – 20 years Total sample: N = 360	Setting: Secondary school, group level, face to face, classroom setting Duration: 6 weeks, 3 hours Delivery agent: Not specified Intervention: Sexually transmitted diseases, HIV/AIDS and family planning intervention Control: Received no programme	Target behaviour(s): Sexual behaviour	Timepoints: Baseline, 6 weeks (End of Intervention) Length of follow up: Short-term <12 months	No
16. Gelfond et al. (2016)Design: Secondary analysis 3-year quasi experimental study Country: USA Quality: Weak	High school students, grades 9, 10 & 11 No mean age or age range reported Baseline: N = 1437 (Intervention group (n = 747); Control group (n = 690)). Before matching: N = 1113 students (Intervention group (n = 578); Control group (n = 535)). Matched: N = 964 (Intervention group: (n = 482); Control group: (n = 482)).	Setting: High school, group level, face to face, classroom setting Duration: Intervention consisted of 16 sessions, over three years (Autumn 2012–Spring 2015) and 25 minutes per session Delivery agent: Health educator Intervention: Need to Know (N2K), intervention Control: Received no programme	Target behaviour(s): Primary outcome: Pregnancy rate within the first 3 years of high school measured with Yes/No response items by 1) Pregnancy event, 2) Sex in past 3 months, 3) Sex in past 3 months without condom and 4) Sex in past 3 months without birth control.	Timepoints: Baseline, 4 years (End of Intervention) Length of follow up: Long-term: ≥12 months	No
17. Mason-Jones et al. (2013) Design: cRCT Country: South Africa Quality: Weak	High school students, Grade 10 No mean age or age range reported Total sample: N = 728 (Intervention group (n = 295); Control group (n = 433)).	Setting: High school, group level, face to face, classroom setting Duration: Not reported Delivery agent: Peer educators Intervention: Peer educators’ intervention on sexual health behavioursControl: Received no programme	Target behaviour(s): Primary outcomes were age of sexual debut and for those who were sexually active and use of a condom at last sex.	Timepoints:Baseline, 18 month follow up Length of follow up: Long-term: ≥12 months Results: Age of sexual debut effect size, adjusted Odds Ratio, 0.17, 95%, CI = −0.20 – 0.55Condom use effect size, adjusted Odds Ratio, 1.19, 95%, CI = 0.56 - 2.52	No
18. Lan et al. (2019)Design: Pre & post test Country: USA Quality: Weak	High school students, 9^th grade No mean age or age range reported Total sample: N = 300	Setting: High school, group level, face to face, classroom setting Duration: 60-minute sessions delivered over 8 weeks Delivery agent: Health educators Intervention: AMP live intervention. Control: Virtual intervention	Target behaviour(s): 1) whether they had ever had sexual intercourse, 2) whether they or their partner had used a condom at their last sexual intercourse.	Timepoints: Baseline, 8 week (End of Intervention) and 3 week follow up Length of follow up: Short-term <12 months	No
19. Cartagena et al. (2006) Design: Survey of 3-year intervention Country: Mongolia Quality: Weak	High school students, grade 10 No mean age reported Age range 15 to 19 Total sample: N = 647 (Intervention (n = 320); Control group (n = 327)).	Setting: High school, group level, face to face, classroom setting Duration: Not specified Delivery agent: Peer educators, trained for 3 days Intervention: Peer education prevention programme. Control: Not specified	Target behaviour(s): Sexual health practices - Condom use during the last three months of sexual intercourse.	Timepoints: Not reported Length of follow up: Not reported	No
20. Rohrbach et al. (2015) Design: cRCT Country: USA Quality: Moderate	High school students, 9^th grade Intervention M age = 14.15 (SD = .60); Control M age = 14.16 (SD =.55) Age range 12–18 years Total sample: N = 1447 (Intervention group (n = 769); Control group (n = 678)).	Setting: High school, group level, face to face, classroom settings Duration: Intervention: 53 days (12 sessions). No time reported; Control: Average span of 9 days Delivery agent: Volunteers (1 day training) Intervention: The SEI curriculum Control: Basic sex education control curriculum.	Target behaviour(s): Primary outcomes: 1) Pregnancy risk, 2) STI risk, 3) Multiple sexual partners, and 4) Use of sexual health services. Secondary outcomes: (1) Ever engaging in vaginal or anal sex; (2) Ever engaging in oral sex; (3) Engaging in vaginal or anal sex in the previous 3 months; (4) Engaging in oral sex in the previous 3 months; (5) Using a condom at last vaginal or anal sex; (6) Using birth control or condoms at last vaginal or anal sex; and (7) Currently carrying a condom	Timepoints: Baseline, 53 days (End of Intervention) and 12 month follow up Control, 9 days (No follow up) Length of follow up: Long-term: ≥12 months Results: Condom use effect size, Odds ratio, 1.97; 95% confidence interval, 1.39 - 2.80Frequency of sexual activity effect size, Odds ratio, 1.13; 95% CI, .82 - 1.56	Yes
21. Taylor et al. (2014) Design: RCT Country: South Africa Quality: Moderate	High school students, Grade 8 Mean age: Males (M = 14.6, SD= 1.4); Females (M = 13.9, SD = 1.1) No age range reported Total sample: N = 818 (Intervention group (n = 432); Control group (n = 386)). 8 month follow up: (Intervention group (n = 383); Control group (n = 298)).	Setting: High school, group level, face to face, classroom setting Duration: 12 weeks, no time reported Delivery agent: Researcher of project Intervention: School life skills programme.Control: Regular sexual health education established in school	Target behaviour(s): 1) ‘Have you ever had sex? 2) Do you use condoms? 3) Have you ever been pregnant/caused a pregnancy?’ (yes/no response) Condom use consistency ‘How often do you use condoms with your regular partner?’ (Four-point Likert response: 1 (never) - 4 (always)).	Timepoints: Baseline, 12 weeks (End of Intervention) and 4 and 8 months follow up Length of follow up: Short-term <12 months	Yes
22. Lohan et al. (2018)Design: cRCT Country: Northern Ireland Quality: Moderate	Secondary school students, no years reported M age = 14 No age range reported Total sample: N = 831	Setting: Secondary School, group level, face to face, classroom setting Duration: 12 months, no time reported Delivery agent: Teacher Intervention: If I Were Jack intervention. Control: Usual school-based sex education programme	Target behaviour(s): Unprotected sex	Timepoints: 12 months (End of Intervention), 9 month follow up Length of follow up: Short-term <12 months Results: Effect size (95% CI 0.5 to 2.2%)	Yes
23. Markham et al. (2014) Design: RCT Country: USA Quality: Moderate	Middle school students, 7^th grade M age = 12.6 No age range reported Follow up surveys completed by N = 1,233 (29.2% attrition), 10^th grade students Excluded 46 students leaving sample of N = 1,187	Setting: Middle school, group level, face to face, classroom setting Duration: 24 sessions (12 in 7th grade & 12 in 8th grade), 50 minutes duration Delivery agent: Researcher of project Intervention: It’s Your Game. Keep It Real (IYG), Risk Avoidance and Risk Reduction programmes. Control: Regular sexual health education established in school	Target behaviour(s): Primary outcome: Delayed sexual initiation by 10th grade Secondary outcomes: 1) unprotected sex at last vaginal intercourse, 2) number of lifetime sexual partners, and 3) other recent (i.e. in the past 3 months) behaviours including frequency of oral, vaginal, and anal sex, 4) frequency of vaginal or anal sex without a condom and 5) number of sexual partners.	Timepoints: Not reported Length of follow up: Not reported Results: Initiation of sex effect sizes Risk Avoidance: adjusted odds ratio [AOR], .64, 95% confidence interval [CI], .42 −.99 Risk Reduction: AOR, .65, 95% CI, .50 - 84	Yes
24. Markham et al. (2012) Design: RCT Country: USA Quality: Moderate	Middle school African American and Hispanic students, 7^th grade Mean age = 12.6 (59.8% Female) No age range reported Total sample: N = 1,258	Setting: Middle school, group and individual level, face to face, classroom setting Duration: 24 lessons, duration of 50-minutes, with 12 lessons in 7th grade and 12 lessons in 8th grade. Delivery agent: Researcher of project Intervention: It’s your game (IYG), keep it real intervention.Control: Regular sexual health education established in school.	Target behaviour(s): Primary outcome: Delayed sexual initiation (oral, vaginal, or anal). Secondary outcome included 1) delayed initiation of oral, vaginal, and anal sex, and on 2) reduced sexual risk behaviour in 9th grade Secondary outcomes: (1) unprotected sex at last vaginal intercourse; (2) frequency of oral, vaginal, and anal sex in the past 3 months; (3) frequency of vaginal or anal sex without a condom in the past 3 months; (4) number of lifetime sexual partners; and (5) number of sexual partners in the past 3 months.	Timepoints: Not reported Length of follow up: Not reported Results: Adjusted odds ratio (AOR) .65, 95% CI = .54-.77	Yes
25. Rohrbach et al. (2019) Design: Quasi-experimental Country: USA Quality: Moderate	Middle school students, Grades 7 and 8 No mean age or age range reported	Setting: Middle school, group level, face to face, classroom setting Duration: 7th grade (18 sessions) 8th grade (6 sessions), no time reported Delivery agent: Teachers Intervention: HIV/STI/teen pregnancy prevention curriculum.Control: No intervention	Target behaviour(s): Primary outcomes: 1) ever engaged in oral sex (performed or received), 2) vaginal intercourse, or 3) anal intercourse (yes/no response) Secondary outcomes: Recent sexual behaviours: 1) ever engaged in oral, vaginal, or anal sex in the past 3 months (yes/no response) 2) condom use at last vaginal or anal sex (two items; yes/no) and 3) whether the subject had vaginal or anal sex with- out the use of condoms in the past 3 months (yes/no response).	Timepoints: Not reported Length of follow up: Not reported	Yes
26. Tortolero et al. (2010) Design: RCT Country: USA Quality: Moderate	Middle school students, 7^th grade students followed through to 9^th grade No mean age and age range reported Total sample: N = 907	Setting: Middle school, group level, face to face, classroom setting Duration: 24 sessions (12 in 7th grade & 12 in 8th grade), 45 minutes duration Delivery agent: Teachers Intervention: IYG intervention. Control: No IYG intervention	Target behaviour(s): Primary outcome: Delayed sexual initiation at 9th grade follow up Secondary outcomes: 1) delayed initiation of specific types of sex (oral, vaginal, anal); 2) delayed sexual initiation by gender and racial/ethnicity; and 3) reduced risk behaviour at the ninth-grade follow-up for those who reported being sexually active in the last 3 months.	Timepoints: Baseline, 24 month follow up Length of follow up: Long-term: ≥12 months	Yes
27. Peskin et al. (2015) Design: RCT Country: USA Quality: Moderate	Middle school students, 8^th grade into 9^th grade No mean age or age range reported Total sample: N = 1,374 Students received a $10 incentive when they completed the 1 year follow up survey	Setting: Middle school, individual level, computer based, classroom setting Duration: 13 sessions, 35–45 minutes Delivery agent: Teachers Intervention: IYG curriculum to simulate student skills practice in real-world situations.Control: No IYG	Target behaviour(s): Primary outcome: Delayed sexual initiation Secondary outcomes: 1) delayed initiation of specific types of sexual activity; 2) number of lifetime sexual partners; 3) current sexual activity (defined as sexual activity in the past 3 months); 4) number of occasions students had sex without a condom; 5) number of partners in the past 3 months; 6) number of partners without a condom in the past 3 months; and 7) use of a condom during last sex.	Timepoints: Baseline, 12 month follow up Length of follow up: Long-term: ≥12 months	No

Behaviour change techniques (BCTs)

Nine BCTs were identified out of a possible 93 available in the taxonomy. Table 2 summarises the identified BCTs. The following BCTs appeared more frequently within effective interventions: 5.1. Information about health consequences (11), 5.3 Information about social and emotional consequences (10), 6.1. Demonstration of the behaviour (10), 8.1 Behavioural practice/rehearsal (10), followed by 4.1 Instructions on how to perform the behaviour (6). For credible source (code 9.1), this BCT was identified across all interventions.

Table 2. The commonly used Behaviour Change Techniques and examples from interventions.

BCT	Cluster	Description	Example of coded text
4.1 Instruction on how to perform a behaviour (6) 9%	Shaping knowledge	Advise or agree on how to perform the behaviour (includes ‘Skills training’)	‘All eighth graders received computerized skill-based activities practicing steps for correct condom use’
5.1 Information about health consequences (11) 17%	Natural consequences	Provide information (e.g. written, verbal, visual) about health consequences of performing the behaviour	‘Curriculum included modules with information on HIV and Sexually Transmitted Infection Basics; Prevention, Treatment, and Testing of HIV’
5.3. Information about social and environmental consequences (10) 16%	Natural consequences	Provide information about social and environmental consequences of performing the behaviour	‘Curriculum included modules with information on stigma and discrimination; and Living with HIV’
6.1 Demonstration of the behaviour (10) 16%	Comparison of behaviour	Provide an observable sample of the performance of the behaviour, directly in person or indirectly e.g. via film, pictures, for the person to aspire to or imitate (includes ‘Modelling’)	‘The peers are encouraged to: observe and imitate the behaviours of their peer educators and of others’
8.1 Behavioural practice/rehearsal (10) 16%	Repetition and substitution	Prompt practice or rehearsal of the performance of the behaviour one or more times in a context or at a time when the performance may not be necessary, in order to increase habit and skill	‘Role playing, were used to help students to build confidence to negotiate delay of sexual initiation, condoms use, asking a partner’s HIV status, and getting an HIV test done’
9.1 Credible source (15) 24%	Comparison of outcomes	Present verbal or visual communication from a credible source in favour of or against the behaviour	‘Having a doctor present in the lessons helped in giving accurate information about medical issues but the researcher controlled the process of delivery’.

Quality appraisal

According to the EPHPP tool, studies that receive no weak rating in each of the six components are considered strong. Studies that receive one weak rating is considered moderate and those that receive two or more weak ratings are considered weak. For the present review, the quality appraisal of the included studies using these rating criteria provided no studies that scored a rating of strong, 21 studies scored a rating of moderate and six studies scored a rating of weak. The main reasons for the weak quality rating were selection bias, failure to mention whether the researcher or participants were blind to the study aims, lack of control for possible confounding variables and failure to report reliabilities and validities on outcome measures. Table 1 presents the quality ratings of each study.

Aggregated findings of included studies

Target sexual health behaviour outcome

Table 1 summarises the sexual health behaviours targeted by each intervention. Twenty studies (74%) examined multiple sexual health behaviours and seven studies (26%) examined single sexual health behaviours. The most frequently targeted sexual health behaviours were condom use (n = 21, 78%) (e.g. how often do you use condoms/consistent condom use in the past three months or did you have sexual intercourse in the past 6 months) and frequency of sexual activity (n = 15, 55%), followed by the number of sexual partners (n = 8, 30%) (multiple/number of sexual partners) and initiation of sexual activity (n = 4, 15%).

Measures employed

In terms of measures employed to assess outcomes, all studies relied on self-report measures developed by the researchers. Only one study employed a standardised measure and reported the reliability coefficient (Esere, 2008).

Length of follow up

Of the included studies, sexual health behaviour data was collected at different time points ranging from 4 weeks to four years post intervention. Table 1 provides a detailed summary. Twelve studies (44%) reported a short-term follow-up of fewer than 12 months with nine studies (33%) reporting long term follow-up of equal to and more than 12 months. Only one study reported a short and long-term follow-up period and six (22%) studies reported no follow-up information.

Intervention effectiveness

As most of the studies did not report effect sizes (n = 18, 66%), but only p-values, determining the effectiveness of each intervention solely relied on the reported p-values. School-based sex education interventions were deemed effective if there was a statistically significant difference between the intervention and comparison groups for at least one or more sexual health behaviour outcome. A total of 15 (56%) studies were identified as effective, reporting a statistically significant positive difference on one or more sexual health behaviour outcomes. Twelve studies (44%) were identified as not effective, reporting no statistically significant impact on sexual health behaviours.

Frequency of sexual activity

Of the included studies, frequency of sexual activity was measured by 15 studies (55%). Two studies (13%) reported that participants were less likely to engage in sexual activity in the past 3 months (Mwale & Muula, 2019; Thato et al., 2008). Two studies (13%) reported a reduction in the frequency of engaging in oral and vaginal sex in the past 3 months (Rohrbach et al., 2015; 2019), one study reported a reduction in having sex in the past 3 months for anal sex only (Markham et al., 2014), one study reported a reduction in the frequency of oral, vaginal and anal sex in the past 3 months (Markham et al., 2012). Finally, one study reported a reduction in the frequency of ever having sex in the past 30 days (Denny & Young, 2006). Eight of the 15 studies (53%) reported no change in the frequency of sexual activity after the delivery of the intervention. Specifically, one study reported no reduction for engaging in vaginal, anal and oral sexual activity (Berglas et al., 2016), two studies reported no decrease in the frequency of sexual intercourse in the past 6 months (James et al., 2006; Li et al., 2011), one study reported no decrease in ever engaging in sexual intercourse in the past 3 months (Daley et al., 2019) and four studies reported no decrease in the frequency of sexual activity (Burnett et al., 2011; LaChausse, 2006; Mba et al., 2007; Tucker et al., 2007) after the delivery of the intervention.

Condom use

Most studies (n = 14, 52%) reported no significant findings based on p-values, with only seven studies reporting significant findings when compared to the control group. For the studies that reported a significant change in condom use due to the intervention on condom use, five (18%) found an increase in the likelihood to use a condom when engaging in sexual activity (Esere, 2008; Markham et al., 2012; 2014; Menna et al., 2015; Taylor et al., 2014); one found an increase in condom use at sexual debut (Mwale & Muula, 2019); and one found an increase in condom use when engaging in anal sex only (Rohrbach et al., 2019). For the studies that did not report significant findings, eight (39%) reported no increase of condom use at last sex (Burnett et al., 2011; James et al., 2006; LaChausse, 2006; Lan et al., 2019; Mason-Jones et al., 2013; Musiimenta, 2012; Peskin et al., 2015; Thato et al., 2008); three studies reporting no increase in condom use when engaging in sexual activity in the past 3 months (Cartagena et al., 2006; Daley et al., 2019; Gelfond et al., 2016), two studies reporting no increase of condom use at last vaginal and/or anal sex (Berglas et al., 2016; Rohrbach et al., 2015) and one study reporting no condom use at sexual debut (Walker et al., 2006) after the delivery of the intervention.

Number of sexual partners

Eight (30%) of the total included studies measured the number of sexual partners after the delivery of the intervention. Three of these studies reported a decrease in the number of sexual partners post-intervention (Esere, 2008; Musiimenta, 2012; Mwale & Muula, 2019) and four studies reported no decrease in the number of sexual partners in the three months (Berglas et al., 2016; Markham et al., 2012; Peskin et al., 2015; Rohrbach et al., 2015) following the delivery of the intervention. One study that examined a three-arm intervention reported an increase (AOR, 2.80, 95% CI, 1.52–5.14) in the number of sexual partners for vaginal sexual activity in the past three months in the risk avoidance arm, with a decrease (AOR, .34, 95% CI, .20 - .56) in the risk reduction intervention arm when compared with the control group (Markham et al., 2014).

Initiation of sexual activity

Of the included studies, four (15%) measured initiation of sexual activity. Two of these studies (50%) reported that participants in the intervention group were less likely to initiate sexual activity (LaChausse, 2006; Tortolero et al., 2010). Two (50%) studies that examined a three-arm intervention risk avoidance, risk reduction and control reported delayed sexual initiation for anal, vaginal and oral sex in the risk reduction group (Markham et al., 2012); and delayed sexual initiation for anal sex only in the risk avoidance group (Markham et al., 2014) when compared with the control group after receiving the intervention.

Application of a theoretical framework

Of the 27 studies included in the present review, 14 studies (52%) did not rely on a theoretical framework. Of those that did, the most frequently reported framework was the Social Cognitive Learning Theory (n = 7, 50%). Three studies were underpinned by multiple theoretical frameworks, including Social Cognitive Theory, Theory of Planned Behaviour, Sociocultural Learning Theory, Health Belief Model and Theory of Reasoned Action (Lan et al., 2019; Markham et al., 2012; Mwale & Muula, 2019). Finally, one study reported using the Theory of Planned Behaviour (Gelfond et al., 2016) and one study reported using the I Change model theory (Taylor et al., 2014).

Methodological considerations

With regard to methodological quality, most studies included in this review obtained a moderate rating. Despite the majority of included studies being characterised as RCTs, limited information on the randomisation process was provided, there was a lack of blinding, and no evidence was provided to control for confounders and selection bias. The majority of studies used self-reported items to measure outcomes of sexual health behaviours that were devised by the study authors, apart from one study that employed a standardised questionnaire (Esere, 2008). All studies but one (Esere, 2008) did not report internal consistency for the scales/items utilised.

Discussion

The present review provided a synthesis of evidence on the effectiveness of school-based sex education interventions on sexual health behaviours from 27 studies. Identified sexual health behaviours were condom usage, frequency of sexual activity, initiation of sexual activity and number of sexual partners. Interventions to shape sexual health behaviours are complex and BCT taxonomy provides an approach to examine this complexity. This is the first review to apply the BCT taxonomy to further our understanding of school-based sex education interventions in adolescents. The most used BCTs of school-based sex education interventions were information about health, social and environmental consequences and demonstration of behaviour and behavioural practice/rehearsal. Our findings highlighted that school-based sex education interventions can be used to improve young peoples’ skills to build and sustain positive sexual health behaviours, and thereby complement their sexual health and well-being.

BCTs identified in the interventions

Using a BCT taxonomy (Michie et al., 2011) this review identified nine BCTs for school-based sex education interventions on sexual health behaviours. Our findings highlighted for interventions containing specific BCTs such as information about health consequences, information of social and environmental consequences, demonstration of behaviour and behavioural practice/rehearsal have a greater likelihood of success. These findings are consistent with a previously conducted systematic review that examined brief (less than 60 minutes) interventions to prevent HIV/STIs as well as unintended pregnancies (De Vasconcelos et al., 2018), and provide support to the identified BCTs which may promote positive sexual health behaviours.

Providing information about health consequences was a most frequently identified BCT. This demonstrates that the design and development of most school-based sex education interventions is not extending beyond providing information (e.g. information on reproductive health and negative health consequences) as well as failing to include cognitive and behavioural strategies (e.g. specific goal setting, prompt barrier identification, self-monitoring) to promote positive sexual health behaviours. Although providing information on sexual health behaviours is important, relying on imparting knowledge about sexual health behaviours to prevent engagement in unnecessary sexual risk-taking could be a problematic approach (Kirby & Laris, 2009).

The nine (out of 93) BCTs for school-based sex education interventions identified by our systematic review are relatively homogenous and lack scope in utilising different aspects of the BCT taxonomy. To identify a BCT, the intervention description needs to be explicit and indicate the applied techniques aimed at changing the target behaviour (Michie et al., 2011). As some of the interventions evaluated by the included studies did not provide enough information regarding the intervention, it was difficult to determine whether a BCT was present or not. For the included studies in the present review because the main delivering agent of these interventions were teachers, the interventions were specifically designed for teachers as facilitators. It could be argued that the lack of information pertaining to the BCTs may not have been included in the intervention design, due to the interventions being adjusted to meet the needs of the teacher facilitating the intervention (Buston et al., 2002; Goldfarb & Lieberman, 2021). Previous research has reported that while the adoption of school-based sex education interventions is welcome, schools face difficulties (e.g. timetable constraints) in fully implementing these, which results in schools amending the intervention to suit their teaching needs. Subsequently, this highlights that interventions may be unlikely to achieve their desired goal unless they are implemented originally as intended (Buston et al., 2002; Vanwesenbeeck et al., 2016). In a similar vein, the focus of the interventions included in the present review was on building awareness through promoting information of factors (knowledge, values and attitudes, perceived risk, communication about sex, self-efficacy and skills) that may contribute to behaviour change at individual level. However, it would be also important to understand how these factors operate at school level. For instance, the feasibility of the BCT taxonomy provides codes such as restructuring the social environment and physical environment that could support the identification of BCTs within the school level (Michie et al., 2011). Understanding the extent to which BCTs are identified within this context could further our understanding of the effectiveness of school-based sex education interventions as well as of known determinants of sexual health (i.e. social and cultural norms) (Kok et al., 2016). This could provide a comprehensive account that takes into consideration behaviour change at a school level, while at the same time supporting the development and successful implementation of future school-based sex education interventions within the intended setting of delivery.

The role of sex education interventions on sexual health behaviours

The findings of this review indicated that most of the school-based sex education interventions of the included studies resulted in a positive change of sexual health behaviour outcomes. Of the 27 included studies, 15 demonstrated effectiveness for at least one sexual health behavioural outcome at either intervention endpoint and/or at follow up. The most common sexual health behaviours were frequency of sexual activity, condom use, number of sexual partners and initiation of sexual activity. Our findings are consistent with previous research indicating that school-based sex education interventions have a positive effect on at least one self-reported sexual health behaviour outcome (e.g. delay of sexual initiation, frequency of sex and condom use) (Denford et al., 2017; Kirby et al., 2007; Robin et al., 2004). However, also in keeping with previous research (Denford et al., 2017; Shepherd et al., 2010; Underhill et al., 2007), the present review showed that some school-based sex education interventions may result in negative changes of sexual health behaviours for young people. It should be also noted that some studies reported only one change in sexual health behaviour while other studies did not report changes in any of the included measures of sexual health behaviours.

This finding is consisted with previous research that has found that sex education does not result in positive outcomes for sexual health behaviours (e.g. decrease the rates of STIs and unplanned pregnancies) (Kirby et al., 2007; Lindberg & Maddow-Zimet, 2012). Additionally, using data from the National Longitudinal Study of Adolescents in the United States (Sabia, 2006),measured outcomes of virginity status, contraceptive use, frequency of intercourse, the likelihood of pregnancy and contracting an STI. The findings of this research showed that sex education delivered in school does not affect sexual health behaviours. This evidence demonstrates that sex education interventions continue to present mixed findings of the effectiveness concerning sexual health behaviours, with some reporting positive effects, and other reporting negative effects.

There was a general lack of conducting follow up post-intervention in many of the interventions of the included studies. These interventions did not demonstrate sustained effects because they either measured changes in behaviours at a short term follow up period post-intervention (n = 11), or at a long term follow up period (more than 12 months) post-intervention (n = 9), or they did not include a follow up (n = 6). However, it is noteworthy, that most of the interventions that included a long term follow up post-intervention (longer than 12 months) did not demonstrate sustained effects. This finding is in line with previous research indicating that although school-based sex education programmes can be often effective in promoting knowledge and changing attitudes, their effectiveness on sexual health behaviours is inconsistent and often short-term (Denford et al., 2017).

Nevertheless, it has been argued that measuring the effectiveness of school-based sex education interventions on the sexual health behaviours in school-aged adolescents could be problematic due to low incidence of sexual intercourse within this cohort (Denford et al., 2017). For instance, three studies included in the present review could not measure specific sexual health behaviour outcomes, such as condom use, as the number of adolescents who identified as sexually active at baseline was too small (Burnett et al., 2011; Cartagena et al., 2006; Thato et al., 2008). However, while some of the interventions fail to show a pattern of effectiveness, these could still be generating changes (Denford et al., 2017) and promoting positive sexual health behaviours even after the study completion. Therefore, it is important to consider the influence that these interventions have on shaping sexual health behaviours into adulthood, as these could carry delayed effects in furthering our understanding of sex education interventions.

The studies included in the present review aimed to measure sexual risk outcomes that focus on preventing negative consequences associated with sexual health behaviours. These include sexual health behaviours such as frequency of sexual activity, condom use, number of sexual partners and initiation of sexual activity. Our findings are in line with previous research reporting that too much attention is being given to sexual risk behaviours without considering other and more positive aspects of sexual health behaviours (Janssens et al., 2020). In the present review, the most significantly reported outcome was the frequency of sexual activity, with little evidence to support intervention effectiveness on condom use. However, previous research showed that promoting pleasure in male and female condom use, including safer sex information was linked to increased and consistent use of condoms and to practising safe sex (Philpott et al., 2006). This suggests that a positive sex approach to sexual health can enhance well-being without focusing merely on preventing negative experiences. While these high-risk sexual health behaviour outcomes are important, sexual health behaviours are dominated by a public health discourse of risk and danger. A potential consequence of this is that sexual health and education are considered mostly by a medically oriented stance, which conceals the fact that sexual health also encompasses a social practice, occurring in specific socio-cultural contexts with embedded traditional values, norms and beliefs as well as dimensions of well-being and quality of life.

Application of a theoretical framework

The observed links between the applied theory and the selection of BCTs for the reviewed studies were infrequently and often insufficiently reported. Because applying a theoretical framework that informs intervention design and evaluation is considered an essential aspect of successful interventions (Rosa & Tudge, 2013), this suggests that inadequate consideration was given to the development and design of these interventions. Of the 27 included studies in this review, only 14 reported that the intervention was underpinned by a behaviour change theory, most commonly the Social Cognitive Theory (Bandura, 2004) The theories so far employed for the development of sex education interventions tend to focus on individual aspects of behaviour change (e.g. Theory of Planned Behaviour, Social Cognitive Theory) than on more ecological approaches of behaviour change. Future directions for the development of sex education interventions could be implemented through applying an ecological approach [i.e. Bioecological Theory P-P-C-T model; Bronfenbrenner Ecological Systems Theory, 1979; Rosa & Tudge, 2013; Bronfenbrenner, 1979). Considering the development of one’s sexual health is a social process and dyadic in nature, incorporating an ecological approach in the design of these interventions could further enhance our understanding from where the individual interacts to how and with whom they interact (Baranowski et al., 2003; Neal & Neal, 2013). This could provide more evidence for the effectiveness of school-based sex education interventions on sexual health behaviours.

Recommendations for practice

Our findings provide important evidence-based knowledge to support the improvement and implementation of school-based sex education interventions on shaping sexual health behaviours. The BCT taxonomy (Michie et al., 2011) allowed to investigate the process by which evidence is translated into guideline recommendations for the implementation of school-based sex education interventions. Evidence suggests that school-based sex education interventions that include interactive and participatory educational strategies (Robin et al., 2004) as well as diverse content and activities (Gallant & Maticka-Tyndale, 2004) can be effective in reducing sexual risk behaviours. The findings of this review have shown that school-based sex education interventions are giving greater attention to information relevant to risk reduction strategies (e.g. information on reproductive health and negative health consequences). Additionally, these interventions are based on individual-level factors overlooking other key aspects of sexual health behaviours including cognitive and behavioural strategies (e.g. specific goal setting, prompt barrier identification, self-monitoring) to promote positive sexual health behaviours This has important practical implications as it is not taking into consideration that sexual health behaviours are socially negotiated and subject to moral and sociocultural standards. Therefore, school-based sex education interventions need to focus on behaviours and skills that aim to enhance positive sexual health behaviours through incorporating active learning strategies, such as decision making, planning, motivational control and goal prioritisation (Kirby, 2008; Schaalma et al., 2004).

To develop and promote positive elements of sexual health behaviours, specialist training is required to equip teachers or other school personnel with necessary skills to teach active strategies within the implementation of sex education interventions. In the present review, only four studies associated with effective interventions (Denny & Young, 2006; LaChausse, 2006; Lohan et al., 2018; Rohrbach et al., 2019) reported that training was provided, with the main delivering agent being teachers. Previous research examined an intensive five-day training programme provided to teachers aiming to equip them with skills and methods of how to enhance student participation, through active strategies of role-playing and the development of self-efficacy related to sexual health behaviours (Schaalma et al., 2004). Teachers’ reports of this training programme were overwhelmingly positive, stating that the programme increased their confidence in being able to select methods to engage with as well as teach sex education. This highlights the importance of educator training when it comes to delivering various aspects of sex education targeting sexual health behaviours.

While these practical recommendations are important for improving sex education curricula, research showed that teachers, even when presented with these recommendations, can be reluctant to deliver key aspects of these interventions (Ngabaza & Shefer, 2019; Schaalma et al., 2004). This suggests that when it comes to the implementation of these interventions within the school context, their aim may not be achieved (Buston et al., 2002). For instance, it has been suggested that teachers who do not feel comfortable delivering active strategies, may tailor their teaching method to another activity they are more comfortable with (Finelli et al., 2018; Shi, 2017). This ultimately reduces the opportunity for young people to practice specific skills that are relevant to promoting positive sexual health behaviours. While active strategies and teacher training can be put into practice, it is important to highlight that the development of future school-based sex education interventions should be informed by the organisational structure of the school in addition to individual behaviour change to promote positive change in sexual health behaviours.

Future directions for research

First, future research studies could utilise the BCT taxonomy when designing and developing school-based sex education interventions for sexual health behaviours. This will provide the opportunity to identify common BCTs that are included for school-based sex education interventions targeting sexual health behaviours, which will allow to identify effective or ineffective interventions. Furthermore, future research should provide comprehensive and accurate descriptions of school-based sex education interventions that would enable the identification of included BCTs.

Second, research conducted to further our understanding of school-based sex education interventions targeting sexual health behaviours typically focuses on school-aged adolescents. Yet, research has not frequently considered older cohorts and specifically young adults aged between 18 to 25 years old. This age period is also defined as emerging adulthood, which is a transitional developmental period usually spanning from late adolescence through to mid-twenties (Arnett, 2000). This developmental stage is marked with exploration of sexual identity, experimentation with different sexual roles and engaging in various sexual experiences. It is also characterised by rapid transitions into new social contexts that involve greater freedoms and less social control than experienced during adolescence (Arnett, 2007). Outcomes of sexual health behaviours (e.g. condom use) peak at this time (Arnett, 2007), while prevalence of sexual activity increases during this period (Owen et al., 2010). This makes this cohort of young people susceptible in experiencing negative consequences related to sexual health behaviours such as high STI infection rates and unwanted pregnancies (Owen et al., 2010). Hence, it would be beneficial for future research to examine the role school-based sex education interventions in longitudinally shaping sexual health behaviours in this age group.

Third, the studies included in this review had methodological weaknesses and provided no information for confounders and selection bias. Future studies should place more importance on the designing and development of these interventions through employing robust methodological designs and rigorous research practices. These may include using validated measures as well as transparent and thorough reporting of the research. In addition, most of the included studies did not rely on a theoretical framework, which is an important element of intervention design. Future research should include a theoretical framework to guide the intervention and provide important information for the underlying mechanisms and factors influencing outcomes (Moullin et al., 2020). The effectiveness of included studies was determined only by reporting a statistically significant (p-value) as 18 out of the 27 studies did not report effect sizes. Considering that the included studies had an experimental and a control group when examining the effectiveness of sex education interventions, it would have been more appropriate to report effect sizes. This could allow understanding the actual magnitude of differences (Aarts et al., 2014) as well as support replication by future studies.

Finally, studies included in the present review did not indicate whether the interventions were delivered as planned (termed fidelity of delivery) (Walton et al., 2020). Thus, it is difficult to clarify whether the intervention was indeed conducted as planned and/or whether the findings were attributed to the intervention or other confounding factors (Rojas-Andrade & Bahamondes, 2019). For instance, environmental, organisational and individual factors have been found to impact the fidelity of interventions (Walton et al., 2020). Thus, it is important for future studies to provide comprehensive descriptions on the implementation process including fidelity of delivery.

Limitations

Despite its strengths, the present review has some limitations that should be considered. First, the included studies were written in English language. Therefore, some findings relevant to this topic may not have been included in this review. Second, we did not include grey literature and relevant studies for inclusion could have been missed, which may have introduced publication bias (Brown et al., 2017). Nevertheless, while the inclusion of grey literature can add another dimension, it has been argued that it can also pose challenges. For instance, as grey literature comes in varied formats, the methodological quality can be difficult to assess, and the inclusion of the grey literature may reduce effect sizes and influence decision making (Bell, 2018; Gul et al., 2021; McAuley et al., 2000). Third, because included studies provided descriptions of interventions that were not always detailed or clear, further BCTs could have been present, but were not coded. Fourth, because the intervention designs and outcome measures were heterogenous, it was not possible to conduct a meta-analysis; hence, we conducted a narrative synthesis on the effectiveness of the findings. Future research on sex education interventions should aim to co-ordinate and standardise outcomes so that quantitative comparisons through meta-analysis can be conducted. Finally, it is important to highlight that the screening of the title and abstract (Cohen’s k = .45, 90% agreement between reviewers) and full-text review (Cohen’s k = .51, 80% agreement between reviewers) both yielded a moderate kappa coefficient despite the high percentage of agreement between reviewers. This is referred as the kappa paradox where high percentage of agreement between reviewers create a symmetrical imbalance, resulting in a low kappa coefficient (e.g. see Dettori and Norvell (Dettori & Norvell, 2020) for review). This paradox was present in this review due to high percentage of agreement and the exclusion of more papers than those included between reviewers (Delgado & Tibau, 2019; Feinstein & Cicchetti, 1990).

Conclusions

The present review addresses a gap in the literature by providing a timely and systematic examination of school-based sex education interventions for shaping sexual health behaviours. Through using the Behaviour Change Technique taxonomy, this has provided a standardised framework for identifying and categorising techniques for school-based sex education interventions and sexual health behaviour outcomes. This review provides vital evidence-based knowledge for educators, policymakers and researchers as an important guide for the future development of school-based sex education programmes.

Acknowledgements

The authors would like to acknowledge Dr Daráine Murphy for her contributions to the literature search and reviewing process.

Disclosure statement

The authors report there are no competing interests to declare.

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials.

Additional information

Notes on contributors

Rachel Niland

Rachel Niland is a Research Fellow with Global Brain Health Institute, Trinity College Dublin and St James’s Hospital, Dublin. Rachel’s research interests focus on improving and promoting the lives of individuals across society.

Clodagh Flinn

Clodagh Flinn is a PhD candidate on the Ad Astra Fellow PhD Studentship programme in University College Dublin School of Psychology. Clodagh’s research interests include various aspects of health in young people including mental health and sexual health. Further aspects of her research involve resilience and chronic skin conditions.

Finiki Nearchou

Finiki Nearchou is the Director of the Resilience &amp; Health Laboratory and the Research Director of the Doctoral Programme in Clinical Psychology at University College Dublin. Finiki is a Chartered Member and Associate Fellow of the British Psychological Society. Her research programme focuses on understanding the complex mechanisms underpinning social, environmental, psychological and biological factors associated with sustainable resilience and quality of life. Finiki is particularly interested in promoting youth health and wellbeing.