Intervenciones educativas y basadas en aptitudes para la prevención de la violencia de pareja y el noviazgo violento en adolescentes y adultos jóvenes
Resumen
Antecedentes
Las intervenciones educativas y basadas en aptitudes a menudo se utilizan para la prevención de la violencia de pareja y el noviazgo violento entre los jóvenes.
Objetivos
Evaluar la eficacia de las intervenciones educativas y basadas en aptitudes diseñadas para prevenir la violencia de pareja y el noviazgo violento en adolescentes y adultos jóvenes.
Métodos de búsqueda
Se hicieron búsquedas en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials) (CENTRAL), MEDLINE, EMBASE, CINAHL, PsycINFO, en otras seis bases de datos y en un registro de ensayos el 7 de mayo 2012. Se hicieron búsquedas manuales en las listas de referencias de artículos clave y en dos revistas (Journal of Interpersonal Violence y Child Abuse and Neglect). También se estableció contacto con investigadores de este tema.
Criterios de selección
Estudios aleatorios, aleatorios por grupos y cuasialeatorios que compararan una intervención educativa o basada en aptitudes para prevenir la violencia de pareja y el noviazgo violento en adolescentes y adultos jóvenes con un control.
Obtención y análisis de los datos
Dos autores de la revisión evaluaron de forma independiente la elegibilidad de los estudios y el riesgo de sesgo. Para cada estudio incluido en el metanálisis, los datos fueron extraídos de forma independiente por GF y otro revisor (CH, JN, SH o DS). Se realizaron metanálisis para los siguientes resultados: episodios de violencia de pareja, comportamientos, actitudes, conocimiento y aptitudes.
Resultados principales
Se incluyeron 38 estudios (15 903 participantes) en esta revisión, de los cuales 18 eran ensayos aleatorios por grupos (11 995 participantes) y dos eran ensayos cuasialeatorios (399 participantes). Se incluyeron 33 estudios en los metanálisis. Se incluyeron ocho estudios (3405 participantes) en el metanálisis que evaluó los episodios de violencia de pareja. Hubo una heterogeneidad considerable (I2= 57%) para este resultado. El cociente de riesgos fue 0,77 (intervalo de confianza [IC] del 95%: 0,53 a 1,13). Se incluyeron 22 estudios (5256 participantes) en el metanálisis que evaluó las actitudes hacia la violencia de pareja. La diferencia de medias estandarizada (DME) fue ‐0,06; (IC del 95%: ‐0,01 a 0,15). Se incluyeron cuatro estudios (887 participantes) en el metanálisis que evaluó el comportamiento relacionado con la violencia de pareja; la DME fue ‐0,07 (IC del 95%: ‐0,31 a 0,16). Se incluyeron diez estudios (6206 participantes) en el metanálisis que evaluó el conocimiento relacionado con la violencia de pareja; los resultados mostraron un aumento del conocimiento a favor de la intervención (DME 0,44; IC del 95%: 0,28 a 0,60), aunque hubo una heterogeneidad considerable (I2= 52%). Se incluyeron siete estudios (1369 participantes) en el metanálisis que evaluó las aptitudes relacionadas con la violencia de pareja. La DME fue 0,03 (IC del 95%: ‐0,11 a 0,17). Ninguno de los estudios incluidos evaluó la salud física, la salud psicosocial ni los resultados adversos. Los análisis de subgrupos no demostraron diferencias estadísticamente significativas por contexto de intervención o tipo de participantes. La calidad de las pruebas para todos los resultados incluidos en el metanálisis fue moderada debido al riesgo incierto de sesgo de selección y de detección y al riesgo alto de sesgo de realización en la mayoría de los estudios.
Conclusiones de los autores
Los estudios incluidos en esta revisión no mostraron pruebas de la efectividad de las intervenciones en cuanto a los episodios de violencia de pareja o las actitudes, los comportamientos y las aptitudes relacionadas con la violencia de pareja. Se encontró un aumento pequeño del conocimiento, aunque hubo pruebas de una heterogeneidad considerable entre los estudios. Se necesitan estudios adicionales con seguimiento a más largo plazo, y los revisores deben usar instrumentos de medición estandarizados y validados para maximizar la comparabilidad de los resultados.
PICO
Resumen en términos sencillos
Intervenciones para prevenir la violencia de pareja y el noviazgo violento en adolescentes y jóvenes
La violencia de pareja y el noviazgo violento es un problema significativo en los adolescentes y los adultos jóvenes. La violencia de pareja incluye una variedad de comportamientos violentos, desde el abuso verbal a la agresión física y sexual y desde amenazas hasta la violación y el homicidio. Actualmente, hay muchos programas en las escuelas y las universidades y dentro de los contextos comunitarios que procuran prevenir la violencia de pareja. Es importante establecer si dichos programas funcionan y si dan lugar a reducciones a largo plazo en la violencia de pareja. Esta revisión analizó los resultados de 38 estudios. Los resultados no mostraron pruebas convincentes de que los programas redujeran la violencia de pareja, o de que mejoraran las actitudes, los comportamientos y las aptitudes de los participantes en relación con la violencia de pareja. Los resultados indicaron que el conocimiento de los participantes acerca de las relaciones mejoró levemente después de los programas. Estos resultados deben interpretarse con cuidado, debido a que los estudios individuales difirieron en los tipos de participantes y las intervenciones que utilizaron y las formas en las que se midieron los cambios. Ninguno de los estudios consideró el efecto de los programas sobre la salud física y mental. Se necesitan estudios adicionales, que hagan un seguimiento de los participantes durante un período más prolongado y que consideren la relación entre las actitudes, el conocimiento, el comportamiento, las aptitudes y el número de veces que ocurre la violencia de pareja para ayudar a comprender mejor la función de estos programas.
Authors' conclusions
Summary of findings
| Educational and skills‐based interventions compared with control for the prevention of relationship and dating violence in adolescents and young adults | ||||||
| Patient or population: adolescents and young adults (aged 12‐25 years) Settings: any community or educational setting Intervention: educational and skills‐based interventions to prevent relationship and dating violence Comparison: control intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control group | Intervention group | |||||
| Episodes of relationship violence (events) (0‐12 months) | 164 per 10001 | 133 per 10002 (105 to 167) | RR 0.81 (0.64 to 1.02) | 3405 (8) | ⊕⊕⊕⊝ | |
| Physical health | See comments | See comments | Not estimable | None | Not estimable | No studies reported physical health outcomes |
| Psychosocial health | See comments | See comments | Not estimable | None | Not estimable | No studies reported psychosocial health outcomes |
| Attitudes towards relationship violence (0‐12 months) (higher score = less accepting of dating violence) | The mean attitudes score (measured by RMAS 19) ranged across control groups from 37 to 524 | The mean attitudes score in the intervention groups was on average 0.12 higher (95% CI ‐0.02 to 0.27) | 5256 | ⊕⊕⊕⊕ Moderate5 | ||
| Behaviour in relationship violence (0‐12 months) (higher score = more positive behaviour) | The mean behaviour score (measured by DBS) ranged across control groups from 38 to 516 | The mean behaviour score in the intervention groups was 0.07 lower (95% CI ‐0.31 to 0.16) | 887 | ⊕⊕⊕⊕ Moderate7 | ||
| Knowledge of relationship violence (0‐12 months) (higher score = better knowledge) | The mean knowledge score (as measured by a variety of scales) ranged across control groups from 0 to 37 | The mean knowledge score in the intervention groups was 0.43 higher (95% CI 0.25 to 0.61) | 6206 | ⊕⊕⊕⊝ | ||
| Skills related to relationship violence (0‐12 months) (higher score = better skills) | The mean skills score (as measured by SCS) ranged across control groups from 34 to 439 | The mean skills score in the intervention groups was 0.03 higher (95% CI ‐0.11 to 0.17) | 1369 | ⊕⊕⊕⊕ Moderate10 | ||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1. Based on average risk in control groups of 8 studies included in this analysis. 2. Corresponding intervention risk (per 1000) = 1000 * ACR * RR. 3. Majority of studies at risk of bias. Also a high degree of heterogeneity. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in episodes of dating and relationship violence. 4. Based on scores in 8 studies using the RMAS 19 scale. 5. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in acceptance of dating and relationship violence. 6. Based on scores in 3 studies using the DBS scale. 7. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in positive behaviour. 8. Majority of studies at risk of bias. Also a high degree of heterogeneity. Quality of evidence downgraded by 1 level. 9. Based on scores in 6 studies using SCS scale. 10. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in skills. | ||||||
Background
Description of the condition
The term 'intimate partner violence' describes actual or threatened physical, sexual or psychological violence that occurs within a relationship or is perpetrated by a current or former partner or spouse (Saltzman 2002). Saltzman 2002 further defines the components of intimate partner violence as follows:
-
physical violence is the intentional use of physical force with the potential for causing death, disability, injury or harm;
-
sexual violence is divided into three categories:
-
the use of physical force to compel a person to engage in a sexual act against his or her will, whether or not the act is completed,
-
attempted or completed sex act involving a person who is unable to understand the nature or condition of the act, to decline participation, or to communicate unwillingness to engage in the sexual act, for example because of illness, disability, or the influence of alcohols or other drugs, or because of intimidation or pressure,
-
abusive sexual contact;
-
-
threats of physical or sexual violence involves the use of words, gestures or weapons to communicate the intent to cause death, disability, injury or physical harm;
-
psychological/emotional violence involves trauma to the victim caused by acts, threats of acts or coercive tactics.
Physical, sexual and verbal violence can be common responses to conflict within relationships and can have significant effects upon the mental, physical and social well‐being of those involved. Although intimate partner violence is often unreported, prevalence within the adult population is estimated to be high, with prevalence rates varying between countries (WHO 2005). For example, one multi‐country study conducted by the World Health Organization (WHO) found that between 3% and 54% of women report having experienced physical or sexual violence by an intimate partner in the previous year (WHO 2005). In addition, between 10% and 50% of women report having experienced violence from partners or ex‐partners at some point in their lives (Watts 2002).
Rates of relationship abuse vary according to age, sex and previous experience of violence (Foshee 1996; Foshee 1998; Archer 2000). The prevalence of relationship violence is higher in adolescents than in adults, with females aged 12 to 18 years having the highest victimisation rate (Home Office 1999; Wolfe 2003). This form of violence is called dating violence and perpetrators are most likely to be peers (Schewe 2006). Approximately 20% of young women have experienced violence from a dating partner (O'Keeffe 1986; Bergman 1992). Additionally, studies on relationship violence have found that first episodes of violence frequently occur in adolescence (Henton 1983). In younger dating samples, relatively higher proportions of aggression by women against men have been described, although results vary according to the measurement methods used and must, therefore, be interpreted with caution (Archer 2000).
Early experiences of dating violence are linked to poor health outcomes such as sexually transmitted infections (Campbell 2002; WHO 2005; Exner‐Cortens 2013), teenage pregnancy (Campbell 2002), substance misuse (Roberts 2003; Tyler 2012), cancer, coronary heart disease, attempted suicide (WHO 2005; Exner‐Cortens 2013), depression (Campbell 2002; Roberts 2003; Wolitzky‐Taylor 2008), and symptoms of post‐traumatic stress disorder (Campbell 2002; Wolitzky‐Taylor 2008). Relationship violence during pregnancy is also associated with adverse maternal and neonatal health outcomes including preterm delivery (Campbell 2002), low birthweight (Murphy 2001; Campbell 2002), and foetal death (Campbell 2002). Moreover, adolescents who have experienced dating violence in the past are more likely to be perpetrators or victims of intimate partner violence as adults (Krug 2002; Loh 2006; Chiodo 2012).
Description of the intervention
This review focuses on educational and skills‐based interventions targeted at young people aged 12 to 25 years. It includes primary preventive interventions, where participants may have never experienced or perpetrated relationship violence, and secondary prevention, where participants have experienced or perpetrated relationship violence in the past. This review focuses only upon interventions that actively provide the participants with knowledge and skills aimed at preventing initial or further relationship violence. It does not include 'screening programmes' that only offer referral to support agencies. We selected the age group 12 to 25 years to include both adolescents and young adults.
A number of environments can be used to deliver educational and skills‐based interventions, including the community, and in particular, within schools and higher education environments (Wolfe 1999). Because schools play an important role in the development of social behaviour, they provide an appropriate environment to target children and adolescents in the prevention of dating violence and subsequently other forms of relationship violence. Previous systematic reviews have focused on the effectiveness of general violence prevention programmes, such as those against aggression and bullying (Mytton 2006; Adi 2007; Park‐Higgerson 2008). However, there is further potential to utilise schools and other settings in preventing relationship violence. Studies from the USA suggest that interventions delivered to college‐based populations may have an effect on reducing incidences of sexual assault and possibly intimate partner violence (Luthra 2006). Programmes can also be delivered within home (Foshee 2012) and community (Wolfe 2003; Salazar 2006) settings to raise awareness about abuse, promote positive relationships, enable help‐seeking and peer support, challenge discriminative viewpoints and encourage the development of protective skills (Wolfe 1999).
How the intervention might work
Educational and skills‐based programmes aiming to prevent or reduce dating and relationship violence may provide participants with the skills to communicate effectively; deal constructively with stress, disappointment and rejection; resolve conflicts and promote healthier relationships (Wolfe 1999). They may also provide young people with skills to protect themselves from the risk of relationship violence and to improve low self‐esteem, which is linked to the likelihood of being a victim of relationship violence (Gidycz 2006).
Why it is important to do this review
The high prevalence of relationship violence and the severity and duration of its health consequences render this area an important public health issue. To date, many review studies have focused either on intimate partner violence or domestic violence in adult populations (Ramsay 2009; Wood 2010), or on the prevention of sexual abuse (Zwi 2009), or general violence (Mytton 2006), in children. We have found only one systematic review of interventions to prevent dating and relationship violence in young people (Whitaker 2006). Whitaker 2006, which reviewed only primary prevention programmes, included non‐randomised (e.g. pre‐ and postinterventions) as well as randomised studies, and summarised results narratively. Our review builds upon these findings by limiting included studies to randomised controlled trials (RCTs) and quasi‐RCTs, including primary and secondary prevention programmes, and summarising results in a meta‐analysis. Given that interventions to prevent relationship violence based in schools and universities are becoming increasingly widespread, this review is important for strengthening the evidence base, providing a clearer idea of what works, and helping to inform future policy, practice and research in this area.
Objectives
To assess the efficacy of educational and skills‐based interventions designed to prevent relationship and dating violence in adolescents and young adults.
Methods
Criteria for considering studies for this review
Types of studies
RCTs, cluster‐RCTs and quasi‐RCTs (in which participants were assigned to intervention or comparison/control groups according to date of birth, day of the week, simple alternation by order of enrolment or other similar methods). Quasi‐RCTs were included for determining intervention effects because it was maintained that delivery of educational and skills‐based interventions in schools and other settings was practice‐based research and that there would be situations where individual randomisation would not be possible.
Types of participants
Adolescents aged 12 to 18 years and young adults aged 19 to 25 years in any setting. We included studies with a wider age range of participants if we could extract or obtain data for these age groups or if more than 80% of the participants included in the study were within the age range of 12 to 25 years.
Types of interventions
Any programme that was applied universally or to specifically targeted high‐risk groups and actively provided adolescents or young adults with educational or skills‐based interventions, or both, aimed at the prevention of dating or relationship violence. We included interventions delivered in any setting and of any duration. We compared all interventions with a control intervention including no intervention, placebo intervention (e.g. provision of first aid classes) or standard care.
Exclusions
-
Any intervention where the prevention of dating or relationship violence was not stated in the aims or objectives, or that involved a multiple intervention programme in which it was not possible to isolate the relative effects of the violence prevention component.
-
Interventions that only screened for the occurrence of dating or relationship violence and then referred to a support agency, unless the intervention actively provided an educational or skills‐based component, or both, following screening.
Types of outcome measures
Primary outcomes
Primary outcome measures were:
-
reduction in the number of episodes of relationship and dating violence experienced, as measured by self reports by victims or perpetrators or as reported by official (e.g. police) records;
-
reduction in injuries resulting from relationship and dating violence experienced, as reported by victims or perpetrators of by official (e.g. police) records;
-
self reported subjective improvement in mental well‐being;
-
adverse events (i.e. an increase in the number of episodes of relationship or dating violence, or both, reported).
Please see the summary of findings Table for the main comparison for details of the primary outcomes. Outcomes measured did not form part of the criteria for inclusion of studies in the review. In other words, we included any study that met our inclusion criteria for type of intervention, study design and participants.
Secondary outcomes
We also considered a number of secondary outcomes that are closely associated with relationship or dating violence behaviour. These secondary outcomes can help to explain how the interventions might work. These were:
-
improvements in behaviour or knowledge about relationship and dating violence (participant‐reported);
-
improvements in access to (or knowledge of) help or support services (participant‐reported);
-
attainment of protective skills (participant‐reported);
-
intervention‐related factors: cost of the programme, time commitment required and acceptability of the programme (as measured by dropout rate).
Measurement scales
Since a variety of measurement scales are available to assess outcomes of educational and skills‐based interventions, we only included data from studies in which a full description of the measurement scale and its scoring system was available. Where further evaluations of the reliability or validity of a measurement scale existed in the literature, we drew upon these to help determine the suitability and applicability of the scale in relation to the given outcome.
Timing of outcome assessment
We divided all primary and secondary outcomes into short‐term outcomes (outcomes assessed immediately following the intervention to six months following the intervention); medium‐term outcomes (outcomes assessed between 6 and 12 months following the intervention) and long‐term outcomes (outcomes assessed more than 12 months following the intervention).
Search methods for identification of studies
We considered both published and unpublished work to be eligible for inclusion in the review. The Cochrane Developmental, Psychosocial and Learning Difficulties Group (CDPLPG) Trials Search Co‐ordinator advised on and carried out the search. There were no restrictions on language or date of publication. We planned to assess articles published in languages other than those spoken by the review authors using the assistance of translators.
Electronic searches
We searched the following electronic databases on 7 May 2012:
-
Cochrane Central Register of Controlled Trials (CENTRAL), Issue 4, 2012
-
Ovid MEDLINE, 1946 to April week 4 2012
-
EMBASE, 1980 to week 18 2012
-
CINAHL, 1937 to current
-
PsycINFO, 1967 to May week 1 2012
-
Sociological Abstracts, 1952 to current
-
Social Sciences Citation Index, 1970 to 4 May 2012
-
ERIC, 1966 to current
-
National Criminal Justice Reference Service Abstracts (www.ncjrs.gov/library.html)
-
metaRegister of Controlled Trials (mRCT) (www.controlled‐trials.com/mrct/)
-
ZETOC, search limited to conference proceedings
-
WorldCat, search limited to theses/dissertations (www.worldcat.org/)
Details of search strategies used are in Appendix 1.
Searching other resources
In order to identify further relevant literature that was not obtained by searching the databases listed above, we carried out additional searches. We handsearched reference lists of key articles included in the review and issues of the Journal of Interpersonal Violence and Child Abuse and Neglect published between 2005 and 2012. Finally, we contacted authors of key studies and asked them to share any published, unpublished and ongoing work relevant to the review.
Data collection and analysis
Selection of studies
Two review authors (GF and CH) independently screened the titles and abstracts of articles identified in the search against the inclusion criteria and decided which reports should be retrieved. We rejected articles at this stage if the title or abstract did not focus on prevention of relationship and dating violence in adolescents or young adults. If there was insufficient information in the title and abstract to make such decisions, we retrieved the full text. When the full text was not readily available or when we needed further clarification to establish eligibility for our review, we contacted the authors by email. If this was unsuccessful, we contacted host universities in the case of doctoral theses, requested interlibrary loans and sought the help of the Cochrane Group Trials Search Co‐ordinator and a local librarian.
We reviewed selection decisions and resolved disagreements by consultation with a third review author. If disagreements were not resolved in conjunction with the third review author, we consulted the CDPLPG editor. We documented the principal reason for exclusion of each study that seemed initially to meet our inclusion criteria but on closer inspection did not in the Characteristics of included studies table.
Data extraction and management
GF developed the data extraction forms, performed the first round of data extraction from all included studies and entered results into Review Manager 5 (RevMan 2011). We conducted a second round of data extraction in order to ensure the accuracy of data extraction. This was carried out by dividing all included studies between the remaining authors (CH, SH, JN, DS), each of whom independently extracted data for their allocated studies and compared their results with those extracted by GF. Each included study, therefore, had data extracted by GF and one other review author. We resolved disagreements by a further review of the studies in question. We extracted data concerning population, age, control group, baseline characteristics, intervention characteristics, duration, compliance and outcome measures and have presented this in the Characteristics of included studies table. We requested the specific data relevant to age groups included in the review from authors of trials where there was a wide spread of ages among participants.
Assessment of risk of bias in included studies
One review author (GF) assessed the risk of bias in each study using The Cochrane Collaboration's 'Risk of bias' tool (Higgins 2011), with each of the other review authors (CH, SH, JN, DS) independently conducting a 'Risk of bias' assessment and comparing their results to those of the first review author (GF). There were no disagreements. For each of the six domains listed below, we described what was reported to have happened in the study and gave a judgement of low, high or unclear risk of bias.
Sequence generation
-
Description: the method used to generate the allocation sequence should be described in sufficient detail to enable assessment of whether it should have produced comparable groups.
-
Review authors' judgement: was the allocation sequence adequately generated?
Allocation concealment
-
Description: the method used to conceal the allocation sequence should be described in sufficient detail to determine whether intervention allocations could have been foreseen in advance of, or during, enrolment.
-
Review authors' judgement: was the allocation adequately concealed?
As inadequate allocation concealment can introduce bias into the study results, we performed sensitivity analyses and excluded studies from meta‐analysis where no allocation concealment was used or if there was uncertainty about allocation concealment. Quasi‐RCTs may introduce bias, as the method of allocation to the different groups is not sufficiently rigorous to ensure allocation concealment. We conducted sensitivity analyses to assess the impact of including quasi‐RCT studies.
Blinding
-
Description: any measures used to blind study participants and assessors from knowledge of which intervention a participant was allocated to should be described.
-
Review authors' judgement: was knowledge of the allocated intervention adequately prevented during the study?
Incomplete outcome data
-
Description: the completeness (including attrition and exclusions from analysis) of outcome data for each main outcome should be reported.
-
Review authors' judgement: were complete data for each outcome reported, and, if not, were adequate reasons for incomplete outcome data provided?
Selective outcome reporting
-
Description: the possibility of selective outcome reporting should be examined.
-
Review authors' judgement: were the reports of the study free of suggestion of selective outcome reporting?
Measures of treatment effect
We used risk ratios (RR) to summarise dichotomous data due to ease of interpretation. We reported continuous data as mean differences (MD) where the same scale was used for measurement and standardised mean differences (SMD) where different scales were used to measure the same outcome.
Unit of analysis issues
For the cluster‐RCTs included in this review, we followed the guidance on statistical methods for such trials outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011, Section 16.3). We planned to use a summary measure of effect from an analysis that adequately accounted for the cluster design. If this was not available, we planned to extract or calculate appropriate measures of effect as for a parallel group trial and adjust the standard errors (SE) to account for the effect of clustering. We used an intraclass correlation coefficient (ICC) to describe the relative variability in outcomes within and between clusters. Among our included cluster‐RCTs, only one used an ICC (Wolfe 2009). The ICC used in this study was 0.02, which we felt was insufficiently conservative based on meta‐analyses of other similar subjects. Thus, for our analyses, we adopted a higher ICC of 0.15, based on two similar reviews: first, a review of school‐based programmes to prevent violence that used ICC values of 0.1 and 0.2 (Mytton 2006) and second, a meta‐analysis of multi‐component preventive interventions for children at risk of antisocial behaviour (CPPRG 1999), which used an ICC value of 0.15. We used the ICC to calculate a design effect for each cluster‐RCT. In the meta‐analyses, we calculated SMD effect sizes. These were weighted using the generic inverse variance function, and we used random‐effects assumptions.
Several studies identified for inclusion in the review had multiple intervention groups. In these cases, to avoid double counting and creating unit‐of‐analysis errors, we made single pair‐wise comparisons. In other words, we combined all relevant intervention groups (i.e. educational and skills‐based interventions aiming to prevent relationship or dating violence) into a single group, and all relevant control groups into a single group. We then made comparisons between the combined intervention group and the combined control group. For dichotomous outcomes, we summed the sample sizes and number of outcomes across groups. For continuous outcomes, we combined means and standard deviations (SD) (Higgins 2011, Section 16.5.4). If this would have limited the investigation of potential sources of heterogeneity, we planned to compare each intervention group separately against a proportion of the common control group such that no double‐counting of individuals in the common control group occurred. We carried out these analyses using Review Manager 5 (RevMan 2011).
Dealing with missing data
If significant quantities of participant data were missing and the review authors agreed that a study's conclusions were compromised as a result, we contacted trial authors and asked them to supply missing data (e.g. on subgroup means and SDs, numbers of participants). If we received no reply or if missing data were not available, we excluded studies from the final analysis. We also asked trial authors for ICC values. For each study included in our review, we report the dropout rate (calculated as the number of participants included in the final analysis as a proportion of those who began the intervention) in the 'Risk of bias' table.
We considered conducting an intention‐to‐treat analysis of all randomised participants using imputed values for the missing data but maintained that the size of the missing data was too problematic to impute values. Instead, we excluded participants for whom no outcome data were available (Higgins 2011). This re‐introduces bias previously removed by the randomisation process. We discuss the extent to which the results and conclusions of the review are altered by the missing data.
Assessment of heterogeneity
In order to investigate the extent of variation between studies, we assessed the distribution of relevant participant (e.g. age, gender) and trial (e.g. randomisation, assessor blinding, attrition rate, and type and duration of intervention) factors. We assessed statistical heterogeneity using the I2 statistic, which describes the proportion of variation in point estimates that is due to heterogeneity rather than sampling error (Higgins 2011). We considered an I2 value of greater than 50% to represent substantial heterogeneity (Higgins 2011). We also used the Chi2 test of homogeneity to determine the strength of evidence for genuine heterogeneity.
Assessment of reporting biases
For the meta‐analyses involving 10 or more studies, we produced funnel plots (estimating differences in treatment effects against their SE) to assess the presence of possible publication bias. We assessed these visually, followed by exploratory analyses to investigate possible causes, for example comparison of fixed‐effect and random‐effects estimates. While funnel plot asymmetry may indicate publication bias, this is not inevitably the case, and we consider possible explanations for any asymmetry found in the Discussion section (Egger 1997).
Data synthesis
To make best use of the data, we combined studies in meta‐analysis across all settings and irrespective of duration or intensity. These aspects were explored in subgroup analysis and no differential effect was found. Where there was substantial heterogeneity, we computed pooled estimates only for those trials that could be analysed together and for which the necessary statistical data were available. In our protocol we agreed that if substantial heterogeneity was indicated (i.e. greater than 50%), we would exclude studies from meta‐analysis and report results narratively. However, for two of our outcomes we went ahead with meta‐analysis despite heterogeneity being greater than 50%. This decision was based on a number of reasons: firstly, a pooled result was still deemed to be useful despite high heterogeneity, particularly for the main outcome of episodes of relationship and dating violence; secondly, much of the heterogeneity was due to one or two small, outlying studies; and thirdly, heterogeneity was only marginally above our threshold (57% for episodes of relationship and dating violence and 52% for knowledge of relationship and dating violence). We describe our reasons in more detail under Effects of interventions>Primary outcomes>Episodes of relationship and dating violence and Effects of interventions>Secondary outcomes>Knowledge of relationship and dating violence. We discuss the implications of conducting meta‐analyses on studies with a high degree of heterogeneity in the Discussion.
We conducted analyses according to outcome (e.g. episodes of relationship and dating violence, attitudes towards dating and relationship violence, skills associated with dating and relationship violence). Within each outcome, we created subgroups according to the measurement instrument used to assess the outcome. We adjusted scales where necessary so that high scores across all scales would signify either an improvement or deterioration.
We carried out data synthesis using Review Manager 5 (RevMan 2011). As per our protocol, we used a random‐effects model where there was no severe funnel plot asymmetry. Random‐effects analysis assumes that the treatment effect differs between studies, while fixed‐effect analysis assumes that the studies are estimating the same underlying treatment effect. Given that we were examining educational and skills‐based interventions for preventing relationship and dating violence in both adolescents and young adults, it was likely that the review would be combining data from trials with differences in design, population and interventions, thus resulting in different effects. This made the use of a random‐effects analysis more appropriate. However, where studies had similar interventions measuring the same outcomes, we used a fixed‐effect analysis. Where there was significant funnel plot asymmetry, we used both fixed‐effect and random‐effects models and reported the degree of agreement between the results of the two models. We calculated overall effects using the inverse variance method (Higgins 2011). For dichotomous outcomes, we calculated an overall RR. We calculated MDs for continuous outcomes and similar comparisons and outcome measures and SMDs for continuous outcomes measured with similar, but not identical, instruments across studies. For ease of interpretation, where possible, we expressed results as a RR (or odds ratios (OR)) and included 95% confidence intervals (CI). We described studies in which the combining of data in a meta‐analysis was inappropriate due to substantial heterogeneity (as defined above) individually.
Subgroup analysis and investigation of heterogeneity
We explored the reasons for any evidence of heterogeneity among studies that we included. Irrespective of the degree of heterogeneity found, we carried out subgroup analyses for intervention setting (i.e. school, university or community settings), the target audience (i.e. general population or high‐risk population), the timing of outcome assessment, and the duration of the intervention. We planned to conduct subgroup analyses for age and gender but these could not be carried out for reasons explained in the Effects of interventions section (under 'Subgroup analyses').
Sensitivity analysis
We conducted sensitivity analyses to assess the extent to which results were sensitive to the analysis being restricted to only those studies judged to be at a low risk of bias. We ran sensitivity analyses in which the analyses were restricted to the following:
-
studies with a low risk of selection bias (as determined by the quality of the random sequence generation);
-
studies with a low risk of assessment bias (as determined by the quality of blinding of assessors);
-
studies with a low risk of attrition bias (as determined by the completeness of the data).
Results
Description of studies
Results of the search
The literature search identified 22,184 articles. Of these, 95 appeared to meet our inclusion criteria based on titles and abstracts. For these 95 articles, we sought to obtain the full text to establish eligibility for our review. Correspondence with authors yielded an additional three eligible studies that had not been identified by our search strategy. Of the 98 articles identified, full texts were unavailable for 17 articles despite the use of all the methods listed in the Selection of studies section. The Characteristics of studies awaiting classification table lists the missing studies. For nine of these (Bernardo 1994, Brown 2002, Chrappa 1991, Holcomb 1993, Lawson 2006, Murphy 1997, Northam 1997, Sanchez 2011, Walther 1986) no email or postal details for authors could be found. Of the remaining eight studies for whom authors' contact details were found, five responded but did not have a copy of their study available (Bond 1995, Deiter 1994, Heimerdinger 2006, Hill 1995, Layman‐Guadalupe 1996) and three (Abrams 1992, Avina 2005, Halvorson 2007) did not reply. A more detailed evaluation of the full text of the remaining 80 articles revealed 41 articles (representing 38 studies, details listed in the Characteristics of included studies table) that were eligible for inclusion in our review and 40 that were excluded (with reasons provided in the Characteristics of excluded studies table). Of the 38 studies included in our review, 33 were included in the meta‐analysis. Figure 1 shows a study flow diagram.
Study flow diagram.
Included studies
Types of study
Of the 38 studies included, 18 were RCTs, 18 were cluster‐RCTs and two were quasi‐RCTs.
Settings
With the exception of one study conducted in the Republic of Korea (Yom 2005), all included studies were carried out in the USA. The majority of studies were conducted in educational settings (25 in universities, 10 in high schools). Three studies were conducted in community settings: Florsheim 2011 studied young pregnant women and their partners attending health clinics, Salazar 2006 studied adjudicated youth in a prison and courtroom setting, and Wolfe 2003 studied teenagers with a history of maltreatment recruited from community centres.
Participants
Most interventions were aimed at general audiences rather than targeted at individuals at high risk of experiencing or committing relationship violence. Five studies targeted high‐risk individuals such as adjudicated adolescent males (Salazar 2006), individuals or couples known to be at high risk of dating aggression (Schewe 1996; Stephens 2009; Woodin 2010), and individuals with a history of maltreatment and therefore at risk of relationship violence (Wolfe 2003).
Interventions
Interventions were predominantly educational, although five provided an additional component on self defence (Wolfe 2003; Gidycz 2006; Orchowski 2008; Wolfe 2009; Florsheim 2011) and one provided a component on communication skills (Macgowan 1997). The duration of interventions ranged from a single, 50‐minute session to 18 sessions delivered over four months (Wolfe 2003). In the majority of studies, the control group received either no intervention or standard care, or were 'wait list controls' (i.e. receiving the intervention after completion of data collection). A number of studies provided the control groups with placebo interventions, which generally took the same format as the intervention but with different and unrelated content, such as presentations, videos or plays on career development (Davis 1997), multi‐cultural issues (Lanier 1998), sexually transmitted infections (Pinzone 1998), stress management (Saberi 1999) and other similar issues (Yom 2005). One study showed the control group an episode of a situation comedy television programme (Kuffel 2002).
Outcomes
Of the outcomes we listed, the included studies reported one primary outcome (episodes of relationship or dating violence) and three secondary outcomes (improvements in behaviour or knowledge, improvement in access, and attainment of protective skills). Included studies did not report injuries, mental well‐being, adverse events and intervention‐related factors.
Seventeen studies reported episodes of relationship or dating violence. Of these, eight used the Sexual Experiences Survey (SES), five used the Revised Conflict Tactics Scale (CTS) and the remaining four used other scales. Eight studies had sufficient data to be included in the meta‐analysis.
Of our secondary outcomes, the majority of studies included in our review assessed improvements in behaviour, knowledge and skills related to dating and relationship violence. Several studies included measures of changes in attitudes, and we included this outcome in our results even though we did not stated it separately in our predefined outcomes because we had assumed this to be a subcomponent of behaviour and knowledge. Sixteen studies assessed attitude changes; the scale used most frequently was the Rape Myth Acceptance Scale (RMAS) (Burt 1980) and variations thereof (a shortened or modified form of the RMAS; Illinois Rape Myth Acceptance; Rape Myth Scale). Other scales used to assess attitudes included the Rape Empathy Scale, the Date Rape Attitudes Survey, the Relationship Expectations Scale and the General Attitudes Towards Rape scale. Six studies assessed behaviour change, of which three used the Dating Behaviour Survey to measure outcomes. Twelve studies assessed the change in knowledge following the intervention. Many studies used their own measurement tools, often using multiple‐choice questions, to assess participants' knowledge. Finally, seven studies assessed improved skills to prevent relationship and dating violence, of which five used the Sexual Communication Survey (SCS) to measure changes in participants' ability to communicate effectively with dating partners.
Individual studies measured outcomes at different timepoints. Thirty studies assessed short‐term outcomes (0‐6 months following intervention); four studies assessed medium‐term outcomes (6‐12 months following intervention); and four studies assessed long‐term outcomes (more than 12 months following intervention). Because such a significant majority of studies assessed short‐term outcomes, we analysed all studies together and conducted subgroup analyses to assess whether effects differed when medium‐term, long‐term, or both, were excluded from the analyses. If a study had multiple follow‐up points, we chose the longest duration of follow‐up.
The Characteristics of included studies table summarises details of each included study.
Excluded studies
We excluded 40 studies because they did not meet our inclusion criteria:
-
not an RCT, cluster‐RCT or quasi‐RCT (16 studies: Hanson 1993; Jensen 1993a; Frazier 1994; Kilian 1996; Krajewski 1996; Michener 1997; Echols 1998; Proto‐Campise 1998; Weisz 2001; Sanchez‐Cesareo 2002; Baumann 2004; Wolf 2004; Rothman 2006; De Gannes 2009; Foubert 2010; Wolfe 2012);
-
no intervention (observational or descriptive study) (10 studies: Holcomb 1993a; Gidycz 1995; Callahan 2003; Hendy 2003; Jaycox 2006; Gidycz 2007; Gidycz 2008a; Gidycz 2008b; Foshee 2009; Foshee 2011);
-
review of other studies or preliminary reports (five studies: Indermaur 1998; Foshee 2000; Pittman 2000; Foshee 2007; Jouriles 2009);
-
participants not aged 12 to 25 years (four studies: Carpentier 2006; Taylor 2010; Testa 2010; Foshee 2012);
-
not addressing dating violence (two studies: Edwards 2000; Melendez 2003);
-
no control group (two studies: Lavoie 1995; Lonsway 2000);
-
measurement instrument not adequately described (one study: Gray 1990).
We summarise the details of all excluded studies in the Characteristics of excluded studies table.
Risk of bias in included studies
We summarise the risk of bias for each study in the Characteristics of excluded studies table. We have presented authors' assessments of the six domains of bias as a percentage across all included studies in Figure 2 and by each individual study in Figure 3. These figures show that for the majority of studies, the risk of selection bias (due to inadequate random sequence generation or allocation concealment) is unclear, the risk of attrition and reporting bias is low, and the risk of performance bias (due to inadequate blinding of participants and personnel) is high.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Allocation
Random sequence generation
All included studies were RCTs (18 studies), cluster‐RCTs (18 studies) or quasi‐RCTs (two studies). Eight studies indicated how the random sequence was generated: by computer in three studies (Andersen 1992; Pacifici 2001; Miller 2012), by drawing lots in two studies (Forst 1993; Boulter 1997), by alternation in two studies (Davis 1997; Bradley 2009) and by coin toss in one study (Wolfe 2009). We classed all of these studies except Davis 1997 and Bradley 2009 as being at low risk of selection bias. We deemed Davis 1997 and Bradley 2009 to be at high risk of selection bias as both used an unconcealed alternation sequence in which it would have been possible for the authors to choose which of the individuals or classes to allocate to the first group. Andersen 1992 used a computer‐generated table of random numbers to determine the assignment of the first athletic organisation to intervention or control. Following this, we selected the remaining two athletic organisations based on their size in order to keep total numbers in experimental and control groups similar. This introduces a high risk of selection bias. All other studies gave no further details other than that individuals or clusters were "randomly" assigned to groups, and we classed the risk of bias in these studies as unclear. Five studies matched clusters or individuals before randomisation (Macgowan 1997; Foshee 1998; Bradley 2009; Wolfe 2009; Gidycz 2011). In summary:
-
six studies had low risk of bias;
-
29 studies had unclear risk of bias;
-
three studies had high risk of bias.
Allocation concealment
No studies provided information on allocation concealment. We classed studies that used alternation (Davis 1997; Bradley 2009) or coin toss (Wolfe 2009) as being at high risk of bias. We classed Salazar 2006, which used sealed envelopes, and Forst 1993, which used the drawing of lots, as being at low risk of bias as neither participants nor investigators would have been able to foresee allocation. For all other studies, we classed bias as unclear. In summary:
-
two studies had low risk of bias;
-
33 studies had unclear risk of bias;
-
three studies had high risk of bias.
Blinding
Blinding of participants and personnel (performance bias)
The nature of the interventions evaluated in these trials made blinding of participants and personnel delivering the interventions virtually impossible. Interventions were delivered by study authors, established teaching staff in the institutions being studied or members of a third‐party organisation specialising in the delivery of such interventions (e.g. Miller 1999). Most studies provided training (to varying degrees) for the personnel delivering the interventions. Of these, some described ways of minimising the potential for performance bias, such as providing personnel with a script or detailed guidance to follow. However, only six studies (Davis 1997; Gidycz 2001; Wolfe 2003; Jaycox 2006; Orchowski 2008; Gidycz 2011) described monitoring performance objectively through methods such as observation or recording the interventions being delivered to ensure adherence to the study protocol. If any form of monitoring adherence to the study protocol was mentioned, the we classed the study as being at low risk of performance bias. We deemed all other studies to be at high risk. In summary:
-
six studies had low risk of bias;
-
no studies had unclear risk of bias;
-
32 studies had high risk of bias.
Blinding of outcome assessment (detection bias)
None of the included studies specified whether outcome assessors were blinded. With one exception, we classed all studies as being at unclear risk of bias. The exception was Florsheim 2011, where part of the outcome was assessed subjectively through coding of discussions with participants. We felt that a lack of mention of assessor blinding rendered results highly susceptible to detection bias. In summary:
-
no studies had low risk of bias;
-
37 studies had unclear risk of bias;
-
one study had high risk of bias.
Incomplete outcome data
Attrition rate varied greatly, ranging from 0% to 70%. We report attrition rates for individual studies in the 'Risk of bias' tables. We found the highest loss to follow‐up in studies evaluating long‐term outcomes. We classed the risk of bias as low if attrition was low, moderate but equal across both study arms, moderate for long‐term follow‐up or if the reasons were unlikely to be related to the outcome being assessed. We classed six studies as being at high risk of attrition bias due to high rates of loss to follow‐up or significantly different rates of attrition between the intervention and control groups. Lanier 1998 provided no information on attrition bias. In summary:
-
30 studies had low risk of bias;
-
one study had unclear risk of bias;
-
six studies had high risk of bias.
Selective reporting
Although most studies reported all their outcomes fully (providing number of events or percentages for dichotomous outcomes and number of participants, means and SDs for continuous outcomes), a number of studies provided summary statistics only (e.g. stating that the differences between groups was "significant" without providing data, or presenting F, t‐test or P values only). These were considered to be incomplete reports of the results. We contacted the authors of these studies to request further information but we received only a few replies and none provided additional data. Therefore, we classed these studies as being at high risk of reporting bias. In summary:
-
33 studies had low risk of bias;
-
no studies had unclear risk of bias;
-
five studies had high risk of bias.
Other potential sources of bias
We assessed the risk of publication bias by drawing funnel plots for the two meta‐analyses involving 10 or more studies: attitudes towards relationship violence (Figure 4) and knowledge of relationship violence (Figure 5). We found no significant asymmetry and we therefore assume that there is no significant publication bias in the studies included in our meta‐analysis. Based on the results of the funnel plots, we compared results of fixed‐effect and random‐effects models only for attitudes towards relationship violence, which displayed moderate heterogeneity. We present results of both models in the Effects of interventions section (under 'Attitudes towards relationship and dating violence').
Funnel plot of comparison of studies included in meta‐analysis for outcome 1.3: attitudes towards relationship violence
Funnel plot of comparison of studies included in meta‐analysis for outcome 1.5: knowledge of relationship violence
Effects of interventions
See: Summary of findings for the main comparison
The objective of this review was to assess the efficacy of educational and skills‐based interventions designed to prevent relationship and dating violence in adolescents and young adults. We included 38 studies in the review, of which we included 33 in the meta‐analyses. We excluded five studies from the meta‐analyses: Andersen 1992 was excluded because results were analysed using non‐parametric (Mann‐Whitney U) analyses; Foshee 1998 did not report the number of participants in each arm or any tests of significance; Shultz 2000 did not report the number of participants in each group; and Holcomb 2002 and Woodin 2010 reported F‐statistics, which could not be used to extract evidence of effect as the statistic was not associated to a direct comparison of the intervention under study (and therefore not equivalent to a t‐test).
Where a study used more than one scale to measure a particular outcome, we included the scale that was most widely used across other studies and the most validated, such as the RMAS (Burt 1980) and the SES (Koss 1982). If a study measured more than one outcome (e.g. episodes of dating violence and change in attitudes towards dating violence), we included all outcomes but ensured each outcome was included in a separate meta‐analysis. If studies had more than one follow‐up period, we selected the longest period of follow‐up (up to one year) for inclusion in our meta‐analysis.
We applied an ICC value of 0.15 to results of all cluster‐RCTs with the exception of Wolfe 2009, in which the authors had already accounted for the effect of clustering using two‐level hierarchical models. In three studies, the number of participants in each study arm was not presented but we considered it was reasonable to make the following assumptions.
-
Bradley 2009 described an even split of their 196 participants into two groups. We interpreted this as meaning there were 98 participants in each arm.
-
Lanier 1998 stated that the 436 students were assigned to groups "in approximately equal numbers". For our calculations, we assumed there were 218 participants in each group.
-
Macgowan 1997 reported 440 students in total but did not provide the number of students in each class. Based on the size of classes in other studies included in this review, we assumed the mean number of students per class to be 30.
We describe the effects of the interventions on the outcomes reported below. Table 1 summarises the outcomes that we planned to assess, the outcomes reported in included studies and the outcomes used in this review.
| Planned outcomes | Outcomes reported in included studies | Outcomes included in review |
| Primary | ||
| Episodes of relationship or dating violence | YES | YES |
| Physical health | NO | NO |
| Psychosocial health | NO | NO |
| Adverse events | NO | NO |
| Secondary | ||
| Improvement in behaviour or knowledge, or both | YES | YES1 |
| Improvement in access to/knowledge of support services | NO | NO |
| Improvement in skills | YES | YES |
| Cost of programme | NO | NO |
| Time commitment of programme | NO | NO |
| Acceptability of programme | NO | NO |
1. 'Improvement in behaviour and/or knowledge' was assessed in our review as three separate outcomes: behaviour, knowledge and attitudes.
Primary outcomes
Episodes of relationship and dating violence
Seventeen studies measured episodes of relationship and dating violence. We conducted analysis as two separate meta‐analyses: one of categorical data and one of continuous data.
The analysis of categorical data included eight studies (3405 participants), of which seven used the SES. The remaining study assessed the number of students experiencing physical dating violence only (Wolfe 2009). Within this analysis, there was evidence of substantial heterogeneity (I2 = 57%). The high Chi2 statistic (16.35; degrees of freedom = 7) and low P value (0.02) provide further evidence of variation of effect estimates beyond chance. Despite this, we felt it important to conduct a meta‐analysis because episodes of relationship and dating violence was our most important outcome, because this was the only outcome for which all included studies except one used the same measurement scale and because the I2 value of 57% was only marginally above our originally defined threshold of 50%. Indeed, the level of heterogeneity between studies, as shown by the forest plot, suggests that pooling results is unlikely to be problematic and that much of the heterogeneity may be the influence of two outlying studies (Stephens 2009; Gidycz 2011). Using a random‐effects model, the RR was 0.77 (95% CI 0.53 to 1.13). The RR suggests a 23% reduction in the episodes of relationship violence experienced by participants receiving the intervention, but the CI does not exclude the possibility that the interventions had no effect or were associated with an increase in episodes of violence experienced. (See Analysis 1.1.)
The second analysis for this outcome was conducted on the five studies (3171 participants) that assessed the occurrence of relationship and dating violence using continuous scoring systems such as the Interpersonal Violence Scale (IPV), the Revised CTS2 and the Conflict in Adolescent Dating Relationships Inventory (CADRI). Foubert 2000 used the SES but presented results as mean scores and was therefore included in this second analysis rather than the first analysis. Within this second analysis, there was no heterogeneity (I2 = 0%) and a fixed‐effect model was used. The SMD was ‐0.05 (95% CI ‐0.19 to 0.09). Although the point estimate suggests a mean 0.05‐point reduction in relationship violence experienced by those exposed to the interventions, the CI includes the possibility of the intervention having no effect or even increasing participants' experiences of violence as compared to the control group. (See Analysis 1.2.)
Episodes of physical injury
None of the included studies reported episodes of physical injury.
Improved mental well‐being
None of the included studies reported improved mental well‐being.
Adverse events
None of the included studies reported adverse events.
Secondary outcomes
Attitudes towards relationship and dating violence
The meta‐analysis for attitudes towards relationship and dating violence included 22 studies. Thirteen of these studies used the RMAS to measure outcomes and nine used other scales. A total of 5256 participants were included. There was evidence of moderate heterogeneity (I2 = 48%). Using a fixed‐effect model, the SMD was 0.06 (95% CI ‐0.03 to 0.15). Because of the moderate level of heterogeneity, a random‐effects model was also conducted, which produced an SMD of 0.08 (95% CI ‐0.06 to 0.22). Point estimates from both the random‐effects and fixed‐effect models suggest slightly improved (i.e. less accepting) attitudes towards relationship violence in participants receiving the intervention. However, CIs from both models include the possibility of the interventions having no effect or even worsening participants' attitudes towards relationship violence. (See Analysis 1.3.)
Behaviour in relationship and dating violence
We included four studies (887 participants) in the meta‐analysis for behaviour in relationships and dating violence. Three studies used the Dating Behaviour Survey and one study used the Behavioural Intent to Rape Survey. There was no heterogeneity (I2 = 0%) and, therefore, we used a fixed‐effect model. The SMD was ‐0.07 (95% CI ‐0.31 to 0.16). The point estimate of ‐0.07 suggests a slight deterioration in behaviour towards relationship violence among those exposed to the intervention. However, the CI cannot rule out the possibility of the intervention having no effect or having a beneficial effect on behaviours. (See Analysis 1.4.)
Knowledge of relationship and dating violence
We included 10 studies (6206 participants) in the meta‐analysis for knowledge of relationship and dating violence. Each study used different scales to measure participants' knowledge. Salazar 2006 measured students' knowledge using the Seventh Grade Inventory of Knowledge and Attitudes. The authors did not state the direction of the scoring in their scale. The other nine studies included in the meta‐analysis of knowledge change had scales in which a higher score indicated better knowledge, and we assumed this rule to apply to the Inventory used by Salazar 2006. The I2 of 52% for this outcome suggests that there may be substantial heterogeneity. The high Chi2 statistic (18.81; degrees of freedom 9) and low P value (0.03) provide further evidence of variation of effect estimates beyond chance. This heterogeneity is likely to be the result of the differences in scales used to assess knowledge (nine different scales used in 10 studies). As with our first outcome measure, we proceeded with meta‐analysis despite the I2 value of 52% being above our threshold. We present results here and discuss the implications of the high level of heterogeneity in the Discussion. Using a random‐effects model, the SMD was 0.44 (95% CI 0.28 to 0.60), suggesting a mean increase in knowledge of 0.44 as assessed by these scales. (See Analysis 1.5.)
Skills related to relationship and dating violence
We included seven studies (1369 participants) in the meta‐analysis for skills related to relationship and dating violence. Of these, five used the SCS to assess outcomes, one (Wolfe 2003) used the Adolescent Interpersonal Competence Questionnaire and one (Foubert 1998) used the Behavioural Intent to Rape Survey. All of these scales assessed respondents' abilities to communicate effectively. The I2 was low at 0%, with a Chi2 statistic of 5.26 (degrees of freedom 6) and a non‐significant P value (0.51). Although the I2 suggests low heterogeneity, care must be taken when assessing Chi2 tests on a meta‐analysis of a small number of studies, as the test has low power. A non‐significant result should, therefore, not be interpreted as evidence of no heterogeneity (Higgins 2011). Using a fixed‐effect model, the SMD was 0.03 (95% CI ‐0.11 to 0.17). The point estimate suggests slight improvement in skills following exposure to the intervention but the CI does not exclude the possibility of the intervention having no effect or causing a deterioration in skills. (See Analysis 1.6.)
Subgroup analyses
The meta‐analysis for each outcome was re‐run to assess whether there was any effect of the delivery setting or type of audience (general or high‐risk). Tests for subgroup differences used random‐effects models due to the risk of false‐positive results when comparing subgroups in a fixed‐effect model (Higgins 2011). We summarise subgroup differences in Table 2. For delivery setting, there was a statistically significant difference in subgroups when university‐based interventions were compared with community‐ and school‐based interventions for the outcome of knowledge of relationship violence (Chi2 6.27, P value = 0.01; see Analysis 2.1). For all other outcomes, we found no significant differences when analyses were conducted separately for the three possible intervention settings. For audience type, we found significant subgroup differences between interventions aimed at general audiences and those aimed at high‐risk audiences in the episodes of relationship violence experienced (Analysis 2.2) and attitudes towards relationship violence (Analysis 2.3). The decision to conduct subgroup analyses by audience type was made post‐hoc upon finding that a number of included studies targeted their interventions specifically at individuals deemed to be at high risk of experiencing relationship violence, and who might differ systematically from general (lower risk) audiences. To assess whether the timing of outcome assessment affected results, we conducted subgroup analyses in which studies assessing medium‐term outcomes, long‐term outcomes or both were excluded. Exclusion of these studies made no significant difference to pooled results across any of the outcomes.
| Outcomes | Setting Chi2 (P value) | Audience Chi2 (P value) | ||
| Community1 | School2 | University3 | General vs. high risk | |
| Episodes of relationship violence (events) | N/A4 | 0.02 (0.90) | 0.02 (0.90) | 4.89 (0.03) |
| Episodes of relationship violence (scores) | 0.55 (0.46) | 0.21 (0.65) | 0.20 (0.66) | 0.10 (0.75) |
| Attitudes towards relationship violence | N/A4 | 0.11 (0.74) | 0.11 (0.74) | 9.36 (0.002) |
| Behaviour towards relationship violence | N/A4 | N/A5 | N/A6 | N/A7 |
| Knowledge of relationship violence | 1.23 (0.27) | 0.88 (0.35) | 6.27 (0.001) | 1.01 (0.31) |
| Skills related to relationship violence | 1.00 (0.32) | N/A5 | 1.00 (0.32) | 1.00 (0.32) |
N/A: not available.
1. Comparison of community‐based interventions vs. school‐based and university‐based interventions.
2. Comparison of school‐based interventions vs. community‐based and university‐based interventions.
3. Comparison of university‐based interventions vs. school‐based and community‐based interventions.
4. There were no community‐based studies to exclude from the original analysis for this outcome.
5. There were no school‐based studies to exclude from the original analysis for this outcome.
6. There were no university‐based studies to exclude from the original analysis for this outcome.
7. There were no high‐risk audience studies to exclude from the original analysis for this outcome.
To assess whether the duration of the intervention affected results, we ran two subgroup analyses. In the first subgroup analysis for duration of intervention, we categorised studies into three groups according to total contact time (≥1 hour; 1‐5 hours; ≥5 hours). In the second subgroup analysis for duration of intervention, we categorised studies into three groups according to total number of sessions (1 session; 2‐5 sessions; >5 sessions). For the first, 13 studies were excluded from subgroup analysis as the duration of each session was unclear. For the second, 5 studies were excluded from subgroup analysis as the total number of sessions was unclear. We found no significant differences for any of the outcomes when analyses were conducted separately for each category of total contact time or total number of sessions.
We did not carry out the remaining planned subgroup analyses for the following reasons:
-
age: participant age was directly correlated with the delivery setting, we did not feel it was necessary to re‐run subgroup analyses by age;
-
gender: a number of studies included only males (Andersen 1992; Schewe 1996; Boulter 1997; Davis 1997; Foubert 1997; Foubert 1998; Foubert 2000; Salazar 2006; Stephens 2009; Miller 2012) or only females (Breitenbecher 1998; Gidycz 2001; Gidycz 2006; Orchowski 2008; Senn 2011). However, we were unable to run a subgroup analysis by gender as there were insufficient studies that presented results by gender for any given outcome.
Sensitivity analyses
As assessment bias and attrition bias was deemed to be low in the majority of studies (37 and 30 studies, respectively), we did not run sensitivity analyses to assess the extent to which results were sensitive to the analysis being restricted to only those studies judged to be at a low risk of bias in these areas. Selection bias, as assessed by random sequence generation, was considered to be low in only six studies. Limiting the analysis to those studies at low risk of selection bias made no significant difference to the pooled result for the episodes of relationship violence experienced, attitudes towards relationship violence or behaviour towards relationship violence. In the analysis of knowledge about relationship violence, only one study was at low risk of selection bias (Miller 2012). The individual results of Miller 2012 showed an SMD of ‐0.01 (95% CI ‐0.40 to 0.38), compatible with either an increase or decrease in knowledge for the intervention group, compared to the pooled result of 0.43 (95% CI 0.25 to 0.61), which suggested an increase in knowledge for the intervention group. We conducted sensitivity analyses to assess the impact of including quasi‐RCTs, which are at high risk of allocation bias. Excluding quasi‐RCTs from the analysis made no significant difference to the pooled result for any of the outcomes assessed.
Studies not included in meta‐analysis
Five studies included in our review did not contribute to the meta‐analysis. In these studies only summary statistics (such as F‐tests not equivalent to t‐tests), non‐parametric tests, or lack of information on the numbers of participants, means and SDs in each group were available. One study excluded from the meta‐analysis was the 'Safe Dates' study by Foshee et al (Foshee 1998). This was a large cluster‐RCT involving 1886 students aged 11 to 17 years from 14 schools in North Carolina, USA. The intervention group received 10 sessions, each lasting 45 minutes, addressing dating violence at school, as well as community workshops within emergency departments and social services. Follow‐up was conducted one month, one year and four years following the intervention (Foshee 1998). At one month, there was 25% less psychological abuse perpetration (P value < 0.05); 60% less physical violence perpetration (P value < 0.05) and 60% less sexual violence perpetration (P value < 0.10). At one year, there was no significant difference in behaviour. At four years, there was a significant reduction in the perpetration of physical (P value < 0.02) and sexual (P value = 0.04) dating violence, and less victimisation of physical (P value < 0.05) and sexual (P value = 0.01) dating violence. A four‐year booster intervention made no further improvements to the original intervention. The fact that results of this important study could not be included in our meta‐analysis (due to insufficient raw data provided in results) is a significant limitation of our results.
We excluded four other studies from our meta‐analyses. Andersen 1992 was a cluster‐RCT in which athletic organisations at a university received a prevention programme aimed at increasing awareness of acquaintance rape and creating a safer college experience. There was a statistically significant greater increase in scores in one intervention group compared to control (U = 196.5, z = ‐3.06, P value < 0.05), but no statistically significant difference in scores between a second intervention group compared to control (U = 117, z = ‐0.87, P value > 0.05). Holcomb 2002 was a cluster‐RCT of freshman athletes who received a gender date rape prevention programme consisting of case scenarios and discussions lasting 50 minutes. Following the prevention programme, the intervention group showed significantly more disapproving attitudes towards date rape than the control group (F(1,35) = 47.089, P value = 0.0001). Shultz 2000 was a small RCT of university students who participated in an interactive drama programme on the topic of date rape. Results showed statistically significant differences in post‐test scores between the control (mean 74.25) and intervention groups (means 83.18 and 81.73 for pre‐tested and unpre‐tested groups, respectively) (P value < 0.0002), suggesting that the programme successfully improved students' attitudes. Woodin 2010 was an RCT involving 50 university dating couples who reported at least one episode of physical aggression in their current relationship. Couples attended an interview and received motivational feedback individually as well as in couples. There was a significant overall reduction in physical aggression perpetration (effect size d = 0.58, P value < 0.05) among participants in the intervention group, and the intervention group also reduced physical aggression at a significantly greater rate than participants in the control group (d = 0.56, P value < 0.05).
Discussion
Summary of main results
The effectiveness of interventions to prevent relationship and dating violence can be quantified by a number of different outcomes. We assessed the effectiveness of these interventions as measured by changes in the number of episodes of relationship violence, changes in behaviours, attitudes and knowledge, and protective skills attained. For all outcomes apart from knowledge change, our meta‐analyses show no evidence of a statistically significant effect. In the meta‐analysis for knowledge of relationship violence, interventions appeared to have a beneficial effect. However, there was substantial heterogeneity (I2 = 57%) between these studies, which is likely to be due to the variation between studies. Furthermore, when we excluded the studies at moderate or high risk of selection bias, only one study remained, which showed no evidence of an effect of the intervention on knowledge. Overall, therefore, this review has found no evidence of an effect of interventions on the outcomes reported.
The outcomes addressed by the studies included in our review can be categorised into two groups: direct measures, which we included as our primary outcomes, and proxy measures, which we included as our secondary outcomes. Examples of direct measures include the number of episodes of relationship violence and the physical and psychosocial health outcomes occurring as a direct consequence of violence. Importantly, health outcomes, which constituted two of our primary outcomes, were not assessed by any of the studies included in our review. The noticeable absence of health outcomes, especially when the associations between violence and health have been extensively documented, is an important finding in itself. We regard the lack of findings on these outcomes as a significant gap in the existing literature rather than a poor choice of indicators on our part. Assessing the frequency of dating and relationship violence is another direct measure. This outcome is challenging to measure for a number of reasons. The stigma associated with relationship violence in some settings may lead to fear of disclosing or reporting episodes of violence. In other situations, violent behaviour may be long‐standing such that it has become the 'norm', with victims blaming themselves for the violent behaviour of their partner and not seeking support. Peer pressure or the fear of retaliation may prevent victims from seeking help, and especially among adolescents, there may be poor knowledge about support services available. When cases are reported, information may not be available to researchers due to the non‐disclosure of personal and sensitive information by officials. Finally, large studies with long follow‐up periods are required in order to capture all events reliably.
Besides direct measures of effect, there are also proxy measures such as changes in attitudes, behaviour, knowledge and skills. These differ from direct measures in that they are important contributors to violence but require translation into action in order for a reduction in violence to occur. For example, even if attitudes towards relationship violence are improved and knowledge is increased, participants may not necessarily be able to apply this new information when faced with real‐life incidents. Further research is required to assess whether (and how) changes in these proxy measures translate into reduced rates of relationship violence.
The aim of our review was to provide a comprehensive and unbiased summary of the existing evidence on interventions to prevent dating and relationship violence in adolescents and young adults. However, our results should be interpreted with caution for two main reasons. First, the studies included in our review varied greatly on several aspects. Participants ranged from general (low‐risk) high school and university students to individuals with risk factors for experiencing or perpetrating relationship violence, such as a history of exposure to violence (Wolfe 2003) or adjudication (Salazar 2006). Our subgroup analyses comparing general to high‐risk audiences revealed mixed results. Interventions appeared to be more effective at reducing the episodes of relationship violence experienced in high‐risk audiences: the RR was not statistically significant in general audiences, but indicated a small but statistically significant reduction in risk in high‐risk audiences. However, when assessing changes in attitudes, the SMD was not statistically significant for general audiences, and favoured the control group in high‐risk audiences. Notably, for both outcomes, the high‐risk 'subgroup' consisted of a single study. It is possible therefore that there were factors other than the type of participants that made these particular studies differ from others, and their results may not be generalisable to all studies with high‐risk audiences.
Other studies targeted pregnant women, in whom exposure to relationship violence poses additional risks to their own health and that of the foetus. Furthermore, the focus was on couples rather than individuals in two studies (Woodin 2010; Florsheim 2011). The relationship violence experienced by these couples, particularly the pregnant adolescent women and their co‐habiting parenting partners participating in the study by Florsheim 2011, may differ from the violence experienced by individuals who are not in an established relationship. For example, the latter group might experience violence in the context of dating and include cases of acquaintance rape. In more established relationships, partner violence may be of a more long‐standing nature. The differences between 'dating violence' and 'relationship violence' require further research, especially as there may be important implications for prevention efforts. Finally, the adjudicated youth included in Salazar 2006 represent a group with complex social histories and backgrounds, which may affect their engagement with interventions. On a related note, there is an established link between alcohol and drug use and likelihood of dating or relationship violence. This was not explicitly controlled for within the studies and, while the focus of the review was on universal delivery of educational and skills‐based interventions, drug and alcohol use is likely to have a negative impact on outcomes and this was not adequately described in the studies. While it can be argued that outcomes were still attainable even where drugs and alcohol are indicated, caution is still needed in interpreting the results.
Other variation between studies interventions arose in delivery settings, types of outcomes assessed, duration of the intervention, and duration of follow‐up. The subgroup analysis by intervention setting was significant only for one outcome (knowledge of relationship violence) in one setting (university vs. school and community). Given that for the remaining 17 subgroup and outcome combinations the results were non‐significant, we feel that the single significant outcome has arisen by chance. Intervention duration ranged between single 50‐minute sessions (Bradley 2009; Holcomb 2002; Schewe 1996; Stephens 2009) to 21 75‐minute sessions (Wolfe 2009). Subgroup analyses showed no significant differences in effect by total contact hours or total number of sessions. The duration of follow‐up varied greatly. The majority of studies assessed outcomes immediately following the intervention, with a further assessment between 1 and 12 months later. In order to determine more reliably whether interventions are effective and if so, whether their effects are long lasting and reduce the incidence of relationship violence in later adulthood, longer follow‐up is required.
The second reason for interpreting our results with caution relates to methodological aspects of our review. Individual studies used a wide range of measurement scales to assess outcomes. We presented our results as SMDs, which assumes that each of the assessment measures can be standardised and has comparable SDs. This method also renders the translation of results into a quantifiable improvement or deterioration in outcomes challenging. For example, it is difficult to define what a 0.03‐point improvement in the combined skills score means in practice and what level of statistically significant change is clinically significant. Combining results in a meta‐analysis requires a number of methodological decisions to be made. We have tried to be as transparent as possible by detailing our methods and providing rationales for excluding any studies from meta‐analyses.
Overall completeness and applicability of evidence
All of the studies included in this review came from high‐income countries. With the exception of Yom 2005, all studies were conducted in North America. There are a number of possible reasons for this. First, the term 'dating' may be less frequently used outside of North America. However, our search strategy included the term 'relationship' and other synonyms, which should have identified a broader range of titles. Second, the concept of dating itself may be less common outside of North America. In some lower‐income countries marriage occurs at a younger age, so that if violence in a relationship occurs it is more likely to be classed as intimate partner violence rather than dating or relationship violence. Finally, it is possible that relationship violence is less commonly reported and less researched in other cultures.
With the exception of the Safe Dates study (Foshee 1998), which followed participants up to four years post‐intervention, Wolfe 2009, which followed participants for 2.5 years post‐intervention, and Florsheim 2011, which followed participants up to 18 months post‐intervention, most studies followed participants up to a maximum of 12 months. A few studies only conducted immediate post‐test assessment. There is, therefore, little evidence on the long‐term effectiveness of these interventions. As described above, the majority of studies assessed changes in proxy measures such as attitudes and knowledge, rather than episodes of violence and behavioural change. Further studies are required to explore the interaction between knowledge and attitudes on the one hand, and behaviour, skills and episodes of relationship violence on the other hand.
Quality of the evidence
The quality of most studies in this review was limited by unclear methods of random sequence generation, allocation concealment and assessor blinding, as illustrated in the 'Risk of bias' graph (Figure 2) and 'Risk of bias' summary (Figure 3). While questionnaire scores (as used by most studies) may be less affected by lack of assessor blinding than other outcomes, we felt that the fact that none of the studies raised this point as a consideration was a significant shortcoming. Blinding of personnel delivering the interventions was not possible. However, some studies implemented ways of minimising the effects of this by providing training sessions for staff, scripts and monitoring a selection of sessions to ensure adherence to the study protocol. Studies not implementing these activities were considered to be at high risk of performance bias. We downgraded the evidence from 'high' to 'moderate' for all outcomes in our summary of findings Table for the main comparison.
Another significant concern was the cluster or quasi‐cluster randomisation methods used in 19 of the included studies. Participants were randomised by schools or classes to decrease the risk of cross‐contamination and for practicality. Only one study used an ICC to account for the effect of clustering, though the ICC used was low compared to ICCs of other similar studies. The results of the cluster‐RCTs included in this reviews are, therefore, likely to overestimate the effects of the interventions. Finally, many included studies had small numbers of participants (Andersen 1992; Forst 1993; Schewe 1996; Boulter 1997; Davis 1997; Miller 1999; Shultz 2000; Yom 2005; Salazar 2006; and Woodin 2010 had fewer than 100 participants each) and short periods of follow‐up (maximum follow‐up periods were two weeks or less in Andersen 1992; Forst 1993; Schewe 1996; Avery‐Leaf 1997; Macgowan 1997; Foubert 1998; Lanier 1998; Pinzone 1998; Miller 1999; Saberi 1999; Pacifici 2001; Holcomb 2002; Yom 2005; Bradley 2009), which is likely to affect their quality adversely by decreasing their statistical power and increasing the risk of type I and type II errors.
Potential biases in the review process
Although a large number of articles were initially identified by our search strategy, it is still possible that some relevant studies were missed. Correspondence with authors yielded three further eligible studies that had not been identified by our search strategy. Many authors that we contacted did not reply and it is possible that further relevant studies could have been missed. Two review authors (GF and CH) independently screened titles and abstracts, which minimised the potential for bias in selecting studies for inclusion.
Eighteen studies that appeared by title or abstract to meet our eligibility criteria could not be assessed because full texts were not available. These studies have therefore been listed as 'awaiting classification'. Some of the missing studies are doctoral theses, which we were unable to retrieve despite requesting interlibrary loans, contacting authors and corresponding with universities where the doctoral title was awarded. Other studies were published in journals that we were unable to access. We do not know whether these studies are eligible for inclusion in our review and, if so, how they would have affected our results. Funnel plots for the outcomes that included 10 or more studies showed no evidence of publication bias (Figure 4 and Figure 5). However, it is possible that the studies that we were unable to retrieve are in themselves the result of a publication bias that exists in the general literature; in other words, it may be that the very reason we found these studies difficult to access is that they are studies with inconclusive or negative findings that are less likely to be published or more difficult to retrieve (Scargle 2000). It is possible that the failure to include these studies overestimated the effect sizes of interventions and introduced bias into this review (Higgins 2011).
As outlined in the Methods section, we conducted single pair‐wise comparisons to avoid double‐counting and unit of analysis issues. In some cases, this may have diluted the effect of interventions, for example in studies where high‐risk and low‐risk participants, or those with and those without a history of sexual victimisation were combined into single groups. Combining all outcome assessments up to 12 months into one outcome in our analysis may also have had an effect on our results. For example, it is possible that the effectiveness of interventions was greatest immediately following the intervention, with the effects gradually wearing off with time. However, the substantial differences in outcome assessments made it difficult to assess the effects separately. There was an insufficient number of studies to perform a separate analysis of outcomes assessed between 6 and 12 months as we had planned.
The ICC of 0.15 that we used to adjust for the effect of clustering is likely to lead to conservative estimates of effect. However, we felt this to be the most appropriate ICC based on a review of ICCs used in meta‐analyses of similar topics. It would have been preferable if individual studies had ascertained their own ICCs, or if different ICCs were available depending on whether randomisation occurred at class or school level. An insufficient quantity of examples of ICCs was found to enable us to conduct ICC sensitivity analyses.
The decision to include quasi‐RCTs in our review was based on an assumption that this could be a common method of allocating participants to groups for logistic reasons, particularly in the case of cluster‐RCTs. The inclusion of quasi‐RCTs may have introduced bias into our results as they are at high risk of allocation bias. However, sensitivity analyses showed that including these studies made no significant difference to overall pooled results.
We included studies where more than 80% of the sample was within the age range of 12 to 25 years in anticipation that some community‐based studies may target a wider range of age groups. The only studies that included participants higher than our stated age range were those set in universities, which captured a number of mature students (Forst 1993; Anderson 1998; Stephens 2009). The percentage of students outside of the age range in these studies was very low. Only one included study had participants younger than our age range (Miller 1999): students were aged 10 to 14 years, with a mean of 12.8 years and the majority were aged 13 years. It is unlikely that these small numbers would have influenced our results. Furthermore, subgroup analyses showed no significant differences in effect by intervention setting. As participant age is directly correlated with setting, we feel that there is little risk of the inclusion of these studies impacting our results.
Non‐RCT studies on the topic of relationship and dating violence and those that were excluded from our review were not systematically reviewed.
Agreements and disagreements with other studies or reviews
Whitaker et al conducted a systematic review of primary prevention programmes for the perpetration of partner violence (Whitaker 2006). The authors found that nine of the 11 included studies reported at least one positive intervention effect for knowledge, attitude or behaviour and concluded that such prevention programmes are promising. This finding differs from the lack of evidence of an effect found in our review. However, the review by Whitaker et al included observational (non‐randomised) trials, which may have influenced results. Results were summarised narratively, rather than by means of a meta‐analysis. To our knowledge, no other systematic reviews on this topic have been conducted.
Study flow diagram.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Funnel plot of comparison of studies included in meta‐analysis for outcome 1.3: attitudes towards relationship violence
Funnel plot of comparison of studies included in meta‐analysis for outcome 1.5: knowledge of relationship violence
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 1 Episodes of relationship violence (events).
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 2 Episodes of relationship violence (scores) (higher score = greater frequency of violence).
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 3 Attitudes towards relationship violence (higher score = less accepting of dating violence).
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 4 Behaviour towards relationship violence (higher score = more positive behaviour).
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 5 Knowledge of relationship violence (higher score = better knowledge).
Comparison 1 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults, Outcome 6 Skills related to relationship violence (higher score = better skills).
Comparison 2 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults: subgroup analyses, Outcome 1 Knowledge of relationship violence (higher score = better knowledge).
Comparison 2 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults: subgroup analyses, Outcome 2 Episodes of relationship violence (events).
Comparison 2 Educational and skills‐based interventions for preventing relationship and dating violence in adolescents and young adults: subgroup analyses, Outcome 3 Attitudes towards relationship violence (higher score = less accepting of dating violence).
| Educational and skills‐based interventions compared with control for the prevention of relationship and dating violence in adolescents and young adults | ||||||
| Patient or population: adolescents and young adults (aged 12‐25 years) Settings: any community or educational setting Intervention: educational and skills‐based interventions to prevent relationship and dating violence Comparison: control intervention | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control group | Intervention group | |||||
| Episodes of relationship violence (events) (0‐12 months) | 164 per 10001 | 133 per 10002 (105 to 167) | RR 0.81 (0.64 to 1.02) | 3405 (8) | ⊕⊕⊕⊝ | |
| Physical health | See comments | See comments | Not estimable | None | Not estimable | No studies reported physical health outcomes |
| Psychosocial health | See comments | See comments | Not estimable | None | Not estimable | No studies reported psychosocial health outcomes |
| Attitudes towards relationship violence (0‐12 months) (higher score = less accepting of dating violence) | The mean attitudes score (measured by RMAS 19) ranged across control groups from 37 to 524 | The mean attitudes score in the intervention groups was on average 0.12 higher (95% CI ‐0.02 to 0.27) | 5256 | ⊕⊕⊕⊕ Moderate5 | ||
| Behaviour in relationship violence (0‐12 months) (higher score = more positive behaviour) | The mean behaviour score (measured by DBS) ranged across control groups from 38 to 516 | The mean behaviour score in the intervention groups was 0.07 lower (95% CI ‐0.31 to 0.16) | 887 | ⊕⊕⊕⊕ Moderate7 | ||
| Knowledge of relationship violence (0‐12 months) (higher score = better knowledge) | The mean knowledge score (as measured by a variety of scales) ranged across control groups from 0 to 37 | The mean knowledge score in the intervention groups was 0.43 higher (95% CI 0.25 to 0.61) | 6206 | ⊕⊕⊕⊝ | ||
| Skills related to relationship violence (0‐12 months) (higher score = better skills) | The mean skills score (as measured by SCS) ranged across control groups from 34 to 439 | The mean skills score in the intervention groups was 0.03 higher (95% CI ‐0.11 to 0.17) | 1369 | ⊕⊕⊕⊕ Moderate10 | ||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1. Based on average risk in control groups of 8 studies included in this analysis. 2. Corresponding intervention risk (per 1000) = 1000 * ACR * RR. 3. Majority of studies at risk of bias. Also a high degree of heterogeneity. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in episodes of dating and relationship violence. 4. Based on scores in 8 studies using the RMAS 19 scale. 5. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in acceptance of dating and relationship violence. 6. Based on scores in 3 studies using the DBS scale. 7. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in positive behaviour. 8. Majority of studies at risk of bias. Also a high degree of heterogeneity. Quality of evidence downgraded by 1 level. 9. Based on scores in 6 studies using SCS scale. 10. Majority of studies at risk of bias. Quality of evidence downgraded by 1 level. The confidence interval crosses no difference and is compatible with either an increase or a decrease in skills. | ||||||
| Planned outcomes | Outcomes reported in included studies | Outcomes included in review |
| Primary | ||
| Episodes of relationship or dating violence | YES | YES |
| Physical health | NO | NO |
| Psychosocial health | NO | NO |
| Adverse events | NO | NO |
| Secondary | ||
| Improvement in behaviour or knowledge, or both | YES | YES1 |
| Improvement in access to/knowledge of support services | NO | NO |
| Improvement in skills | YES | YES |
| Cost of programme | NO | NO |
| Time commitment of programme | NO | NO |
| Acceptability of programme | NO | NO |
| 1. 'Improvement in behaviour and/or knowledge' was assessed in our review as three separate outcomes: behaviour, knowledge and attitudes. | ||
| Outcomes | Setting Chi2 (P value) | Audience Chi2 (P value) | ||
| Community1 | School2 | University3 | General vs. high risk | |
| Episodes of relationship violence (events) | N/A4 | 0.02 (0.90) | 0.02 (0.90) | 4.89 (0.03) |
| Episodes of relationship violence (scores) | 0.55 (0.46) | 0.21 (0.65) | 0.20 (0.66) | 0.10 (0.75) |
| Attitudes towards relationship violence | N/A4 | 0.11 (0.74) | 0.11 (0.74) | 9.36 (0.002) |
| Behaviour towards relationship violence | N/A4 | N/A5 | N/A6 | N/A7 |
| Knowledge of relationship violence | 1.23 (0.27) | 0.88 (0.35) | 6.27 (0.001) | 1.01 (0.31) |
| Skills related to relationship violence | 1.00 (0.32) | N/A5 | 1.00 (0.32) | 1.00 (0.32) |
| N/A: not available. 1. Comparison of community‐based interventions vs. school‐based and university‐based interventions. 2. Comparison of school‐based interventions vs. community‐based and university‐based interventions. 3. Comparison of university‐based interventions vs. school‐based and community‐based interventions. 4. There were no community‐based studies to exclude from the original analysis for this outcome. 5. There were no school‐based studies to exclude from the original analysis for this outcome. 6. There were no university‐based studies to exclude from the original analysis for this outcome. 7. There were no high‐risk audience studies to exclude from the original analysis for this outcome. | ||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Episodes of relationship violence (events) Show forest plot | 8 | 3405 | Risk Ratio (Random, 95% CI) | 0.77 [0.53, 1.13] |
| 2 Episodes of relationship violence (scores) (higher score = greater frequency of violence) Show forest plot | 5 | 3171 | Std. Mean Difference (Fixed, 95% CI) | ‐0.05 [‐0.19, 0.09] |
| 3 Attitudes towards relationship violence (higher score = less accepting of dating violence) Show forest plot | 22 | 5256 | Std. Mean Difference (Random, 95% CI) | 0.08 [‐0.06, 0.22] |
| 3.1 Rape Myth Acceptance Scale | 13 | 1931 | Std. Mean Difference (Random, 95% CI) | 0.14 [‐0.04, 0.32] |
| 3.2 Other scales | 9 | 3325 | Std. Mean Difference (Random, 95% CI) | 0.00 [‐0.21, 0.22] |
| 4 Behaviour towards relationship violence (higher score = more positive behaviour) Show forest plot | 4 | 887 | Std. Mean Difference (Fixed, 95% CI) | ‐0.07 [‐0.31, 0.16] |
| 5 Knowledge of relationship violence (higher score = better knowledge) Show forest plot | 10 | 6206 | Std. Mean Difference (Random, 95% CI) | 0.44 [0.28, 0.60] |
| 6 Skills related to relationship violence (higher score = better skills) Show forest plot | 7 | 1369 | Std. Mean Difference (Fixed, 95% CI) | 0.03 [‐0.11, 0.17] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Knowledge of relationship violence (higher score = better knowledge) Show forest plot | 10 | Std. Mean Difference (Fixed, 95% CI) | 0.48 [0.38, 0.57] | |
| 1.1 University‐based interventions | 5 | Std. Mean Difference (Fixed, 95% CI) | 0.60 [0.46, 0.74] | |
| 1.2 School‐ or community‐based interventions | 5 | Std. Mean Difference (Fixed, 95% CI) | 0.36 [0.22, 0.49] | |
| 2 Episodes of relationship violence (events) Show forest plot | 8 | Risk Ratio (Random, 95% CI) | Subtotals only | |
| 2.1 General audience | 7 | Risk Ratio (Random, 95% CI) | 0.86 [0.61, 1.22] | |
| 2.2 High‐risk audience | 1 | Risk Ratio (Random, 95% CI) | 0.25 [0.09, 0.70] | |
| 3 Attitudes towards relationship violence (higher score = less accepting of dating violence) Show forest plot | 22 | Std. Mean Difference (Random, 95% CI) | 0.08 [‐0.06, 0.22] | |
| 3.1 General audience | 21 | Std. Mean Difference (Random, 95% CI) | 0.11 [‐0.02, 0.24] | |
| 3.2 High‐risk audience | 1 | Std. Mean Difference (Random, 95% CI) | ‐0.68 [‐1.17, ‐0.19] | |
