School-based education programmes for the prevention of child sexual abuse

Knowledge effect sizes in prior reviews ranged from 0.57 to 1.07.

“These effect sizes ranged from 0.57 (Heidotting 1994, 18 studies), through 0.71 (Rispens 1997, 16 studies) and 0.90 (Berrick 1992, 13 studies) to 1.07 (Davis 2000, 27 studies).”

• cites Heidotting 1994Not checked — no full text of the source
• cites Rispens 1997 sourceNot checked — no full text of the source
• cites Berrick 1992 sourceSupported

No studies measured parental anxiety or fear.

“No studies measured parental anxiety or fear.”

• Also stated in: § Data synthesis · plain

  “No studies measured parental anxiety or fear.”

• Also stated in: § Harm -parental or child anxiety or fear · plain

  “No studies measured parental anxiety or fear.”

Protective behaviour skills had an odds ratio of 5.71 (95% CI 1.98 to 16.51) based on two studies with 102 participants.

“OR 5.71 (1.98 to 16.51) 102 (2 studies) ⊕⊕⊕ moderate 1”

Knowledge gains were not significantly eroded one to six months after the intervention for either intervention or control groups.

“Knowledge gains (measured via questionnaires) were not significantly eroded one to six months after the intervention for either intervention or control groups.”

The mean questionnaire-based knowledge score in intervention groups was 0.61 standard deviations higher than in control groups.

“The mean knowledge score in the intervention groups was 0.61 standard deviations higher (0.45 higher to 0. 78 higher)”

Programmes increased applied knowledge measured by vignettes immediately after the intervention.

“Vignette-based knowledge (applied knowledge measured by assessing responses to hypothetical scenarios, immediately post intervention) (higher score = higher knowledge)”

Programmes increased knowledge of sexual abuse prevention concepts, measured by questionnaires immediately after the intervention.

“Questionnairebased knowledge (factual knowledge measured by assessing responses to items on a questionnaire or multi-choice test, immediately post intervention) (higher score = higher knowledge)”

There was no evidence that programmes increased or decreased children's anxiety or fear.

“In terms of harm, there was no evidence that programmes increased or decreased children's anxiety or fear.”

The mean vignette-based knowledge score in intervention groups was 0.45 standard deviations higher than in control groups.

“The mean knowledge score in the intervention groups was 0.45 standard deviations higher (0.24 higher to 0. 65 higher)”

School-based sexual abuse prevention programmes improved participants' protective behaviour skills, measured by simulation tests immediately after the intervention.

“behaviours (self protective events measured using a stranger simulation test immediately post in- tervention) 390 per 1000 795 per 1000 (559 to 914) OR 5.71 (1.98 to 16.51) 102 (2 studies) ⊕⊕⊕ moderate 1”

Children exposed to a child sexual abuse prevention programme had greater odds of disclosing their abuse than children who had not been exposed.

“Children exposed to a child sexual abuse prevention programme had greater odds of disclosing their abuse than children who had not been exposed”

No published intraclass correlation coefficient was found for school-based child sexual abuse prevention interventions.

“No published ICC for school-based child sexual abuse prevention interventions could be found.”

Studies have not yet adequately measured the long-term benefits of programmes in reducing the incidence or prevalence of child sexual abuse in programme participants.

“Studies have not yet adequately measured the long-term benefits of programmes in terms of reducing the incidence or prevalence (or both) of child sexual abuse in programme participants.”

The mean anxiety or fear score across control groups ranged from 2 to 7.

“The mean anxiety or fear score measured using a variety of scales across control groups ranged from 2 to 7”

The review included 24 studies with 5802 participants in primary and secondary schools in the United States, Canada, China, Germany, Spain, Taiwan, and Turkey.

“This review included 24 studies, conducted with a total of 5802 participants in primary (elementary) and secondary (high) schools in the United States, Canada, China, Germany, Spain, Taiwan, and Turkey.”

The effect on disclosure was more uncertain when the analysis was adjusted for the grouping of participants in classes or schools.

“however we were more uncertain about this effect when the analysis was adjusted to account for the grouping of participants in classes or schools.”

The mean questionnaire-based knowledge score across control groups ranged from 3 to 64.

“The mean knowledge score measured using a variety of scales across control groups ranged from 3 to 64”

Disclosures of past or current child sexual abuse during or after programme completion were 4 per 1000 in control groups and 14 per 1000 (5 to 45) in intervention groups.

“Disclosures (of past or current child sexual abuse made during or after programme completion) 4 per 1000 14 per 1000 (5 to 45)”

Some included studies randomised clusters such as classes, schools, or districts.

“some included studies involved cluster-randomisation at the level of the class, school, or district.”

The mean anxiety or fear score in intervention groups was 0.08 standard deviations lower than in control groups (95% CI -0.22 to 0.07).

“The mean anxiety or fear score in the intervention groups was 0.08 standard deviations lower (0.22 lower to 0.07 higher)”

The updated review reports 24 unique trials in 29 papers.

“In total, this updated review reports on a total of 24 unique trials reported in 29 papers”

Interventions did not fit an active/passive dichotomy because many combined active and passive components.

“upon close scrutiny, interventions did not appear to fit an active/passive dichotomy with many having multiple components of both active and passive types”

Eighteen studies assigned participants to two groups: an intervention and a control.

“Eighteen studies allocated participants to one of two groups, the intervention (school-based sexual abuse prevention programme) and a control group (no programme or wait-listed).”

Complete data for both intervention and control groups were reported in only four studies.

“Complete data (for intervention and control groups) were reported in only four studies (Dawson 1987;Hazzard 1991;Kolko 1989;Lee 1998).”

In Kraizer 1991, follow-up data were not reported.

“In Kraizer 1991, data were not reported.”

All 24 trials focused specifically on child sexual abuse prevention.

“In all 24 trials, interventions focused specifically on child sexual abuse prevention.”

Individual-randomisation trial sample sizes ranged from 46 to 382 students.

“The total number of students randomised in trials with individuals as the unit of randomisation ranged from 46 (Chen 2012) to 382 (Del Campo Sanchez 2006).”

Seventeen trials compared prevention programmes with a wait-listed control.

“In 17 trials, the effectiveness of prevention programmes was compared to that of a wait-listed control group.”

Ten studies reported short-term knowledge outcomes one to three months after the intervention.

“Ten studies also reported short-term knowledge outcomes one to three months following intervention”

The 24 trials included 5802 school-aged participants.

“A total of 5802 school-aged participants were included in the 24 trials.”

Three studies measured behaviour change with a simulated abuse situation and scored the child's response.

“Three studies measured change in behaviour using a simulated abuse situation and scored the child's response to the situation (Fryer 1987a;Kraizer 1991;Poche 1988).”

Those three studies used a stranger simulation test to assess self-protective skills.

“All three studies used a version of a stranger simulation test to assess children's self protective skills”

All 21 studies that measured post-test questionnaire-based knowledge did so within two weeks after the intervention.

“All of the 21 studies measuring post-test questionnaire-based knowledge did so within a two-week period following intervention.”

There were insufficient data to analyse retention of protective behaviours over time.

“There were insufficient data to proceed with analysis for retention of protective behaviours over time.”

No included study took place in a secondary school.

“None of the included studies were conducted in secondary (high) school settings.”

One study measured long-term knowledge outcomes at 12 months.

“One study measured long-term outcomes at 12 months (Hazzard 1991).”

Five studies reported children's disclosures of child sexual abuse during or following the intervention.

“Children's disclosures of child sexual abuse during or following intervention were reported by five studies (Blumberg 1991;Del Campo Sanchez 2006;Hazzard 1991;Kolko 1989;Oldfield 1996).”

There were too few studies for meaningful subgroup comparisons.

“there were insufficient numbers of studies to allow for meaningful comparisons.”

In the three studies included in the disclosure meta-analysis, there was no heterogeneity (I2=0%, P=0.84).

“There was no heterogeneity (I² = 0%, P value = 0.84).”

Three of the 24 included studies measured retention of protective behaviours over time.

“Three of the 24 included studies measured retention of protective behaviours over time.”

One study found that children with higher self-esteem showed better protective behaviours after intervention.

“The results of one study suggest that children with greater self esteem (Fryer 1987a), as measured by the Harter Perceived Competence Scale (HPCS) (Harter 1982), exhibited better protective behaviours following intervention.”

• cites Fryer 1987aNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Harter 1982 sourceNot checked — no full text of the source

Hazzard 1991 found no important difference between intervention and control children in parental reports of anxiety or fear.

“Hazzard 1991 (n = 399) reported no important differences between intervention and control group children on parental perceptions of anxiety or fear (summary data not provided).”

Sensitivity analysis with ICC 0.2 produced an OR of 2.95 (95% CI 0.69 to 12.61).

“and an ICC of 0.2 produced an OR of 2.95, 95% CI 0.69 to 12.61; see Analysis 5.3).”

Two studies found no improvement in knowledge from post-test to follow-up (Blumberg 1991 at five months; Hazzard 1991 at one year).

“two studies reported no knowledge gains between post-test and follow-up (at five months, Blumberg 1991; at one year, Hazzard 1991)”

In Hazzard 1991 and Oldfield 1996, STAIC scores showed no important difference between intervention and control groups.

“Using child self report measures, Hazzard 1991 and Oldfield 1996 reported no important differences in STAIC scores between intervention and control groups (Hazzard 1991, treatment mean 29.7, control mean 29.9; Oldfield 1996, F(1, 593) = 0.05, P value = 0.825).”

All three studies used a version of a stranger simulation test involving staging a simulated abuse or grooming situation with each individual child.

“All used a version of a stranger simulation test involving staging of a simulated abuse or grooming situation with each individual child where a research assistant, posing as a stranger, requested the child's help with a task that required them to go with the stranger”

Visual inspection of fixed-effect forest plots revealed no discernable trend towards greater effect sizes in smaller studies.

“Visual inspection of fixed-effect forest plots revealed no discernable trend towards greater effect sizes in smaller studies.”

Two studies found that knowledge gains were maintained at two months (Hébert 2001; Lee 1998).

“two studies reported maintenance of knowledge gains at two-month follow-up (Hébert 2001;Lee 1998)”

Disclosures were more frequent among intervention participants (OR 3.56, 95% CI 1.13 to 11.24).

“Disclosure occurred more often in the intervention group (OR 3.56, 95% CI 1.13 to 11.24).”

In the four studies with follow-up, knowledge scores declined slightly from immediately post-intervention to one-to-three-month follow-up (SMD 0.78 to 0.69).

“Results suggest that knowledge appeared to deteriorate slightly over time as demonstrated by a decline in the SMD from 0.78 (95% CI 0.38 to 1.17; I² = 84%, Tau² = 0.13, P value = 0.0003) immediately post-intervention to SMD 0.69 (95% CI 0.51 to 0.87; I² = 25%; Tau² = 0.01, P value = 0.26) at one to three months follow-up (see Analysis 3.1).”

When adjusted for clustering with an ICC of 0.2, knowledge scores decreased slightly from immediate post-intervention to follow-up (SMD 0.86 to 0.72).

“When adjusted with an ICC of 0.2, knowledge decreased slightly over time as demonstrated by a small decline in the SMD from SMD 0.86 (95% CI 0.53 to 1.20) immediately post-intervention to 0.72 (95% CI 0.32 to 1.11) at follow-up (Analysis 3.3).”

Neither of the two studies reported follow-up assessment of protective behaviours.

“Follow-up assessment of protective behaviours was not reported in either of the studies.”

Of the 24 included studies, three reported collecting data on protective behaviours.

“Of the 24 included studies, three studies reported collecting data on protective behaviours (Fryer 1987a;Kraizer 1991;Poche 1988).”

With an ICC of 0.1, the standardised mean difference was -0.04 (95% CI -0.42 to 0.33).

“For an ICC of 0.1, the SMD was -0.04 (95% CI -0.42 to 0.33; Analysis 4.2)”

Three studies found minor, non-important extra knowledge gains from post-test to follow-up (Kolko 1989 at six months; Saslawsky 1986 and Wurtele 1986 at three months).

“three studies reported small, but unimportant additional knowledge gains between post-test and follow-up (six months, Kolko 1989; three months, Saslawsky 1986; Wurtele 1986).”

Oldfield 1996 reported no important difference in RCMAS anxiety scores between experimental and control groups among younger participants.

“Oldfield 1996 also found no important differences between experimental and control group anxiety scores using the Revised Children's Manifest Anxiety Scale (RCMAS) with younger participants, F(1, 653) = 1.40,P value = 0.248.”

The meta-analysis of three studies found an SMD of -0.08 (95% CI -0.22 to 0.07) for anxiety or fear.

“The SMD was -0.08 (95% CI -0.22 to 0.07) (see Analysis 4.1).”

In both Fryer 1987a and Kraizer 1991, children with high self esteem who had improved knowledge scores post-intervention were more likely to exhibit protective behaviours.

“In both studies, children with high self esteem who had improved knowledge scores post-intervention were more likely to exhibit protective behaviours.”

Sensitivity analysis with ICC 0.1 produced a nonsignificant OR of 3.04 (95% CI 0.75 to 12.33).

“Using this method and an ICC of 0.1 produced a nonsignificant OR of 3.04, 95% CI 0.75 to 12.33; Analysis 5.2)”

When adjusted for clustering with an ICC of 0.1, knowledge scores decreased slightly from immediate post-intervention to follow-up (SMD 0.86 to 0.73).

“When adjusted with an ICC of 0.1, knowledge decreased slightly over time as demonstrated by a small decline in the SMD from 0.86 (95% CI 0.53 to 1.20) immediately post-intervention to 0.73 (95% CI 0.41 to 1.06) at follow-up (Analysis 3.2).”

A meta-analysis of two studies found significant gains in protective behaviours in simulated at-risk situations assessed immediately (within two weeks) after intervention.

“Meta-analysis of data from two studies showed significant improvements in protective behaviours in simulated at-risk situations, measured immediately (up to two weeks) post-intervention.”

When studies at high risk of bias for blinding of outcome assessment were excluded, the SMD was reduced to 0.47 (95% CI 0.29 to 0.66).

“When studies at high risk of bias were excluded (Chen 2012;Crowley 1989;Dake 2003;Dawson 1987;Hazzard 1991;Hébert 2001;Snyder 1986), the SMD was reduced to 0.47 (95% CI 0.29 to 0.66).”

The test for subgroup differences was not significant (Chi2 = 0.29, df = 1; P = 0.59).

“However, the test for subgroup differences was not significant (Chi² = 0.29, df = 1; P value = 0.59).”

The test for subgroup differences was just below the statistically significant cut-off (Chi2 = 4.04, df = 1; P = 0.04).

“The test for subgroup differences was just below the statistically significant cut-off of 0.05 (Chi² = 4.04, df = 1; P value = 0.04).”

The high Chi2 statistic (34.25, df = 10) and low P value (< 0.0002) provide further evidence of variation in effect estimates beyond chance.

“The high Chi² statistic (34.25, df = 10) and low P value (< 0.0002) provide further evidence of variation in effect estimates beyond chance.”

Meta-analyses of 18 questionnaire-based and 11 vignette-based studies indicated immediate (within two weeks) gains in factual and applied knowledge.

“Meta-analysis of data from 18 studies for questionnaire-based knowledge and 11 studies for vignette-based knowledge suggested gains in factual and applied knowledge immediately (up to two weeks) post-intervention.”

• AssumptionRests on an assumption: The knowledge scales used in the included studies validly measure children's knowledge relevant to child sexual abuse prevention.

With an ICC of 0.2, the standardised mean difference was -0.03 (95% CI -0.46 to 0.40).

“and for an ICC of 0.2, the SMD was -0.03 (95% CI -0.46 to 0.40; Analysis 4.3).”

Sensitivity analyses adjusting for cluster-randomisation produced ORs of 5.43 (95% CI 1.88 to 15.65) with ICC 0.1 and 5.16 (95% CI 1.81 to 14.70) with ICC 0.2.

“Using this method and an intraclass correlation coefficient (ICC) of 0.1 produced an OR of 5.43 (95% CI 1.88 to 15.65; Analysis 1.2) and an ICC of 0.2 produced an OR of 5.16 (95% CI 1.81 to 14.70; Analysis 1.3).”

Protective behaviours were greatly enhanced in intervention groups compared to control groups immediately post-intervention (OR 5.71, 95% CI 1.98 to 16.51; two studies; n = 102).

“Protective behaviours were greatly enhanced in intervention groups compared to control groups immediately post-intervention (odds ratio (OR) 5.71, 95% confidence interval (CI) 1.98 to 16.51; two studies; n = 102)”

In the three studies included in the harm meta-analysis, there was no heterogeneity (I2=0%, P=0.79).

“There was no heterogeneity (I² = 0%, P value = 0.79).”

Intervention exposure was associated with up to 3.5 times higher odds of disclosure.

“The odds of disclosure were as much as 3.5 times higher in participants exposed to the intervention (see Analysis 5.1).”

The result indicates no increases or decreases in anxiety or fear among intervention participants.

“This result reveals evidence of no increases or decreases in anxiety or fear in intervention participants.”

Complete data were available for four studies (956 participants).

“Complete data were available for four studies (956 participants) (Dawson 1987;Hazzard 1991;Kolko 1989;Lee 1998).”

The high Chi2 statistic (104.76; df = 17) and low P value (< 0.00001) indicated variation of effect estimates beyond chance.

“The high Chi² statistic (104.76;df = 17) and low P value (< 0.00001) indicated variation of effect estimates beyond chance.”

There was evidence of substantial heterogeneity (I2 = 71%; Tau2 = 0.08).

“There was evidence of substantial heterogeneity (I² = 71%; Tau² = 0.08) in the meta-analysis.”

In the meta-analysis, there was evidence of substantial heterogeneity (I2 = 84%; Tau2 = 0.10).

“In the meta-analysis, there was evidence of substantial heterogeneity (I² = 84%; Tau² = 0.10).”

Seven of 29 records included in this review were unpublished theses.

“Seven of 29 records included in this review were unpublished theses (Blumberg 1987;Chadwick 1989;Crowley 1989;Dawson 1987;Grendel 1991;Kraizer 1991;Snyder 1986).”

Only Fryer 1987a and Poche 1988 could be included in the meta-analysis for protective behaviours, as Kraizer 1991 did not report a breakdown of pass or fail scores for intervention and control groups.

“Only the Fryer 1987a (n = 48; randomised controlled trial (RCT)) and Poche 1988 (n = 74; cluster-RCT) studies could be included in the meta-analysis for protective behaviours, as Kraizer 1991 (n = 670; cluster-RCT) did not report a breakdown of pass or fail scores for intervention and control groups.”

A meta-analysis of three studies found higher odds of disclosure among children who received the intervention.

“Our meta-analysis of three studies showed greater odds of disclosures by children receiving interventions.”

Every previous review reported medium-to-large knowledge effects favouring intervention groups.

“All previous reviews have found medium to large effects for knowledge outcomes in favour of intervention groups.”

The narrative synthesis found that interventions were multidimensional in content, methods, and delivery.

“Our narrative synthesis of included studies documented multidimensionality in intervention contents, methods, and delivery.”

No study in the review tested a programme delivered electronically.

“None of the included studies evaluated programmes delivered via electronic means.”

In the included studies, many demographic and other independent variables were reported poorly.

“Many demographic and other independent variables were poorly reported in the included studies.”

This review found an SMD of 0.61 (95% CI 0.45 to 0.78; 18 studies, n=4657) for questionnaire knowledge.

“The current review found a SMD of 0.61 (95% CI 0.45 to 0.78; 18 studies, n = 4657).”

No included study tested a web-based or online programme.

“None of the included studies investigated the effectiveness of a web-based or online programme.”

Eighteen of 24 studies included multiple grade levels together.

“most (18 of 24) studies were undertaken with multiple grade levels together.”

For children in the intervention groups, knowledge scores remained stable from one to six months after taking part in a programme.

“Knowledge scores did not deteriorate for intervention participants one to six months after programme participation, signalling that booster sessions or other maintenance strategies for reinforcement of key messages remain appropriate follow-up strategies.”

The studies took place in high-income and upper-middle-income countries, per World Bank categories.

“Studies were conducted in countries with high and upper-middle income economies according to the World Bank's analytical income categories (The World Bank 2013).”

• cites The World Bank 2013Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Programmes might lead to more disclosures of past or current sexual abuse by participating children.

“may result in greater odds of disclosures of past or current sexual abuse from children who have been programme participants”

The review authors could not replicate the meta-analyses of participation and number of sessions because the study data were reported inadequately.

“Due to inadequate reporting of study data, we were unable to replicate these meta-analyses, and would caution against using the broad variable of participation as the only marker for programme variation.”

Most studies reported disclosures poorly or not at all.

“Disclosures were poorly reported or not reported in most studies.”

Trial methodological quality has not improved much.

“Nevertheless, the methodological quality of trials has not improved substantially.”

Moderate quality evidence indicates that the school-based child sexual abuse prevention programmes reviewed here improve protective behaviours and knowledge in elementary school children immediately after the intervention.

“Our overall interpretation is that there is moderate quality evidence that school-based programmes for the prevention of child sexual abuse, of the types described in this review, are effective in increasing primary (elementary) school-aged children's protective behaviours and knowledge immediately post-intervention.”

The earlier review reported an SMD of 0.59 (95% CI 0.44 to 0.74; nine studies, n=3022) for questionnaire knowledge.

“Our previous review found a SMD of 0.59 (95% confidence interval (CI) 0.44 to 0.74; nine studies, n = 3022) for the questionnaire-based knowledge outcome, which is the outcome most comparable to the outcomes reported in previous reviews.”

Included studies reported baseline characteristics of intervention and control groups inconsistently and poorly.

“Baseline characteristics of intervention and control groups were inconsistently and poorly reported in the included studies.”

The review found no evidence that these programmes lower the incidence of child sexual abuse.

“We do not have evidence that these programmes reduce the incidence of child sexual abuse.”

Fourteen studies randomised children by classroom, school, or district for practical implementation.

“In 14 of the included studies children were randomised in groups of classrooms, schools, or school districts for ease of implementation.”

The available data were insufficient to assess the specific effects of programme type, duration, frequency, or setting.

“Insufficient data were provided to evaluate the specific effects of programme type, duration, frequency, or setting.”

The evidence base mainly examines changes in participants' skills and knowledge, and less often examines harm such as child anxiety or fear, or disclosures of past or current abuse.

“The current evidence is primarily focused on improvements in participants' skills (protective behaviours) and knowledge (both factual and applied knowledge), and to a lesser extent on assessing harm (child anxiety or fear) and disclosures of past or current child sexual abuse.”

Participants' disclosures of past or current abuse after intervention were the only direct measure of programme effects.

“The only direct measure of programme effects was participants' disclosures of past or current sexual abuse that were made following interventions.”

The evidence overall does not show that programmes increase or decrease children's fear or anxiety.

“On balance of evidence, programmes do not appear to increase or decrease children's fear or anxiety, and may result in greater odds of disclosures of past or current sexual abuse from children who have been programme participants, however results are uncertain because of inappropriate data analysis in individual studies.”

Ten of the 24 studies reported ethnicity data poorly or not at all.

“Ethnicity data were poorly or not reported in 10 of the 24 studies.”

All included studies except one took place in primary school settings.

“All but one of the included studies was conducted in primary (elementary) school settings.”

Programme dose ranged from a single 45-minute session to eight 20-minute sessions.

“The duration and frequency (dose) of programme interventions varied from one single 45-minute session to eight 20-minute sessions.”

There are five earlier meta-analyses of sexual abuse prevention programmes, including the original review.

“Five previous meta-analyses of sexual abuse prevention programmes exist as noted in Table 1, including the original version of this review (Zwi 2007).”

• cites Zwi 2007 sourceNot checked — no full text of the source

With an intraclass correlation of 0.2, the pooled standardised mean difference for vignette-based knowledge was 0.60 (95% CI 0.31 to 0.89; 11 studies).

“6 Vignette-based knowledge, ICC = 0.2 11 Std.Mean Difference (Random, 95% CI) 0.60 [0.31, 0.89]”

With an intraclass correlation of 0.1, the pooled standardised mean difference for questionnaire-based knowledge at post-test was 0.86 (95% CI 0.53 to 1.20; 4 studies).

“2.1 Questionnaire-based knowledge (post-test) 4 Std.Mean Difference (Random, 95% CI) 0.86 [0.53, 1.20]”

Without correction for clustering, the pooled standardised mean difference for anxiety or fear was -0.08 (95% CI -0.22 to 0.07; 3 studies, 795 participants).

“1.1 Anxiety or fear 3 795 Std.Mean Difference (IV, Random, 95% CI) -0.08 [-0.22,0.07]”

With an intraclass correlation of 0.1, the pooled standardised mean difference for anxiety or fear was -0.04 (95% CI -0.42 to 0.33; 3 studies).

“2.1 Anxiety or fear 3 Std.Mean Difference (Random, 95% CI) -0.04 [-0.42, 0.33]”

With an intraclass correlation of 0.1, the pooled standardised mean difference for vignette-based knowledge was 0.53 (95% CI 0.32 to 0.74; 11 studies).

“5 Vignette-based knowledge, ICC = 0.1 11 Std.Mean Difference (Random, 95% CI) 0.53 [0.32, 0.74]”

Without correction for clustering, vignette-based knowledge was higher in intervention than control (SMD 0.45, 95% CI 0.24 to 0.65; 11 studies, 1688 participants).

“4 Vignette-based knowledge, no correction for clustering 11 1688 Std.Mean Difference (IV, Random, 95% CI) 0.45 [0.24, 0.65]”

With an intraclass correlation of 0.1, the pooled standardised mean difference for questionnaire-based knowledge at follow-up was 0.73 (95% CI 0.41 to 1.06; 4 studies).

“2.2 Questionnaire-based knowledge (follow-up) 4 Std.Mean Difference (Random, 95% CI) 0.73 [0.41, 1.06]”

In the meta-analysis without correction for clustering, protective behaviours were higher in the intervention than control (OR 5.71, 95% CI 1.98 to 16.51; 2 studies, 102 participants).

“1 Protective behaviours, no correction for clustering 2 102 Odds Ratio (M-H, Random, 95% CI) 5.71 [1.98, 16.51]”

With an intraclass correlation of 0.2, the pooled standardised mean difference for anxiety or fear was -0.03 (95% CI -0.46 to 0.40; 3 studies).

“3.1 Anxiety or fear 3 Std.Mean Difference (Random, 95% CI) -0.03 [-0.46, 0.40]”

With an intraclass correlation of 0.1, the pooled odds ratio for protective behaviours was 5.43 (95% CI 1.88 to 15.65; 2 studies, 102 participants).

“2 Protective behaviours, ICC=0.1 2 102 Odds Ratio (Random, 95% CI) 5.43 [1.88, 15.65]”

For questionnaire-based knowledge at post-test without correction for clustering, the pooled standardised mean difference was 0.78 (95% CI 0.38 to 1.17; 4 studies, 956 participants).

“1.1 Questionnaire-based knowledge (post-test) 4 956 Std.Mean Difference (IV, Random, 95% CI) 0.78 [0.38, 1.17]”

With an intraclass correlation of 0.2, the pooled odds ratio for protective behaviours was 5.16 (95% CI 1.81 to 14.70; 2 studies, 102 participants).

“3 Protective behaviours, ICC=0.2 2 102 Odds Ratio (Random, 95% CI) 5.16 [1.81, 14.70]”

Without correction for clustering, disclosures were more likely in intervention than control (OR 3.56, 95% CI 1.13 to 11.24; 3 studies, 1788 participants).

“1 Disclosures, no correction for clustering 3 1788 Odds Ratio (M-H, Random, 95% CI) 3.56 [1.13, 11.24]”

Without correction for clustering, questionnaire-based knowledge was higher in intervention than control (SMD 0.61, 95% CI 0.45 to 0.78; 18 studies, 4657 participants).

“1 Questionnaire-based knowledge, no correction for clustering 18 4657 Std.Mean Difference (IV, Random, 95% CI) 0.61 [0.45, 0.78]”

With an intraclass correlation of 0.1, the pooled standardised mean difference for questionnaire-based knowledge was 0.66 (95% CI 0.51 to 0.81; 18 studies).

“2 Questionnaire-based knowledge, ICC = 0.1 18 Std.Mean Difference (Random, 95% CI) 0.66 [0.51, 0.81]”

For questionnaire-based knowledge at follow-up without correction for clustering, the pooled standardised mean difference was 0.69 (95% CI 0.51 to 0.87; 4 studies, 929 participants).

“1.2 Questionnaire-based knowledge (follow-up) 4 929 Std.Mean Difference (IV, Random, 95% CI) 0.69 [0.51, 0.87]”

With an intraclass correlation of 0.2, the pooled odds ratio for disclosures was 2.95 (95% CI 0.69 to 12.61; 3 studies, 1788 participants).

“3 Disclosures, ICC=0.2 3 1788 Odds Ratio (Random, 95% CI) 2.95 [0.69, 12.61]”

With an intraclass correlation of 0.1, the pooled odds ratio for disclosures was 3.04 (95% CI 0.75 to 12.33; 3 studies, 1788 participants).

“2 Disclosures, ICC=0.1 3 1788 Odds Ratio (Random, 95% CI) 3.04 [0.75, 12.33]”

The historical controversy over school-based child sexual abuse prevention programmes is concentrated on two outcomes: programmes' actual effectiveness in preventing child sexual abuse, and concerns over negative programme effects.

“This is important because the historical controversy over schoolbased child sexual abuse prevention programmes is concentrated on two outcomes: programmes' actual effectiveness in preventing child sexual abuse, and concerns over negative programme effects (Finkelhor 2007).”

• cites Finkelhor 2007 sourceNot checked — no full text of the source

The review used the extremes 0.1 and 0.2 to calculate a design effect for each cluster-RCT.

“We reasoned that a suitably conservative approach would be to use the extremes of ICC 0.1 and 0.2 to calculate a design effect for each cluster-RCT according to the formula given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011, Section 16.3.4)”

• cites Higgins 2011Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

The quality of evidence for all outcomes included in the meta-analyses was moderate.

“The quality of the evidence for all outcomes included in the meta-analyses (combining of data) was moderate.”

Findings on disclosures, harm, and retention of knowledge over time were inconclusive in the original review.

“Findings on disclosures, harm, and retention of knowledge over time were inconclusive.”

• cites Zwi 2007 sourceNot checked — no full text of the source

In a follow-up study, the same individuals were more likely to use protective strategies they had been taught when confronted with threats and assaults.

“In a followup study, the same individuals were more likely to use the protective strategies they had been taught when confronted with threats and assaults (Finkelhor 1995b).”

• cites Finkelhor 1995bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Girls under 12 effectively used six strategies to avoid abuse: demanding to be left alone, saying they would tell someone, crying, saying they were scared, saying they did not want to, and saying 'no'.

“Girls under the age of 12 years effectively used six strategies to avoid abuse: demanding to be left alone, saying they would tell someone, crying, saying they were scared, saying the they did not want to, and saying”

• cites Leclerc 2011 sourceNot checked — no full text of the source

A review of school-based violence prevention had used 0.1 and 0.2, based on a published 0.15 for similar trials, and this was judged plausible but conservative for classroom clustering.

“estimates of 0.1 and 0.2 had been used in a review of school-based violence prevention programmes (Mytton 2006), based on the rationale for a published ICC of 0.15 for similar trials (CPPRG 1999bin Mytton 2006), and was considered a plausible yet conservative estimate for the impact of clustering at the classroom level (Schochet 2008).”

• cites Mytton 2006 sourceNot checked — no full text of the source
• cites CPPRG 1999b in Mytton 2006Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Schochet 2008 sourceNot checked — no full text of the source

Two studies with high-school and college students showed programmes were associated with reduced incidence of child sexual abuse.

“Two studies with high-school (Ko 2001) and college students (Gibson 2000) showed programmes were associated with reduced incidence of child sexual abuse.”

• cites Ko 2001Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Gibson 2000 sourceNot checked — no full text of the source

Research with sexual offenders found the most effective self-protection strategy, reported by three-quarters of offenders, was to tell the offender they did not want to participate in sexual activities.

“Research with sexual offenders on their perceptions of the efficacy of children's self protection strategies in actual abuse situations has found the most effective strategy, reported by three-quarters of offenders, was to tell the offender they did not want to participate in sexual activities.”

The original Cochrane review found improvements in knowledge and protective behaviours among children who received school-based programmes.

“Consistent with previous reviews, the original Cochrane review found improvements in knowledge and protective behaviours (skills) among children who had received school-based programmes (Zwi 2007).”

• cites Zwi 2007 sourceNot checked — no full text of the source

Study quality was compromised in about half of the included studies because of suboptimal data collection methods for outcomes and inappropriate data analysis.

“Study quality was compromised in about half of the included studies, due to suboptimal data collection methods for study outcomes and inappropriate data analysis.”

This combining allowed pair-wise comparisons without double-counting participants.

“This enabled us to make comparisons between groups using pair-wise comparisons without risk of double-counting participants.”

No studies provided information on methods used to conceal allocation.

“No studies provided information on methods used to conceal allocation.”

In three studies, participants were not informed that the outcome assessment was related to the intervention.

“In three studies, authors noted that participants were not informed that the outcome assessment was related to the intervention”

In these 14 studies, contamination or exposure may have biased results towards underestimation of programme effects, particularly on knowledge outcomes.

“This is likely to have biased the results towards an underestimation of programme effects, particularly on knowledge outcomes, which would be more susceptible to such contamination and exposure.”

The school-based nature of the interventions made blinding of participants and personnel impossible.

“The school-based nature of the interventions made blinding of participants receiving the intervention and personnel delivering the intervention impossible.”

These 14 studies were classified as high risk of bias.

“We classified these 14 studies as high risk of bias.”

Three further studies described a classic experimental design but did not report details about random assignment.

“Three further studies described a classic experimental design, but did not report details about random assignment (Dake 2003;Kolko 1989;Kraizer 1991).”

In eight studies, outcome assessors were reported to be blind to group membership, study hypotheses, or both.

“In eight studies, authors reported that outcome assessors were blind to group membership, study hypotheses, or both”

Subgroup analyses were not conducted because included studies lacked sufficient information on hypothesised relevant issues, such as gender breakdown by intervention group.

“We did not conduct subgroup analyses because there was insufficient information provided in the included studies about issues that were hypothesised as being relevant for subgroup analysis, for example, studies did not always provide a breakdown of student gender by intervention group.”

All of these studies were classified as unclear risk of bias.

“We classified all of these studies as unclear risk of bias.”

Seven studies were classified as high risk of bias for blinding of outcome assessment.

“We classified seven studies as high risk of bias.”

This risk was further heightened when the outcome assessors were the same individuals as those delivering the programme.

“This risk was further heightened when the outcome assessors were the same individuals as those delivering the programme (e.g.Dawson 1987).”

In 14 studies, intervention and control groups were located within the same school.

“In 14 studies intervention and control groups were located within the same school.”

These twelve studies were classified as unclear risk of bias because the method of concealment was not described in sufficient detail.

“However, we classified these studies as unclear risk of bias because the method of concealment was not described in sufficient detail for an adequate assessment to be made.”

None of these 14 studies described a means by which programme fidelity or integrity was addressed or measured.

“None of these 14 studies described a means by which programme fidelity or integrity was addressed (e.g. via the use of scripts or standardised lesson plans) or measured (e.g. via observation, audio, or video recordings).”

In Fryer 1987a, no control group comparison was possible at follow-up because the control groups had already received the intervention.

“In Fryer 1987a, no comparison with the control group was available at follow-up because the control groups had been exposed to the intervention.”

Twelve studies reported tests of baseline imbalances showing no statistical differences between groups.

“Twelve studies reported tests of baseline imbalances showing no statistical differences between groups, potentially indicating successful randomisation.”

Twenty studies stated that individuals or groups were randomised but gave no detail on how the random sequence was generated.

“Twenty studies stated that individuals or groups (classes, schools, or districts) were”

For most studies, no control group comparison was available at follow-up because the control groups had been exposed to the intervention by then.

“For most studies, no comparison with the control group was available at follow-up because the control groups had been exposed to the intervention by then.”

In all instances, procedures were concluded to be potentially unconcealed such that assignment could reasonably have been predicted.

“In all instances we concluded that procedures were potentially unconcealed such that assignment to groups could reasonably have been predicted prior to or during the process.”

Ten studies were classified as low risk of bias for blinding of outcome assessment.

“We classified 10 studies as low risk of bias (Blumberg 1991;Daigneault 2012;Fryer 1987a;Grendel 1991;Krahé 2009;Kraizer 1991;Oldfield 1996;Poche 1988;Saslawsky 1986;Wurtele 1986).”

These analyses indicate that adjusting for clustering has very minimal effects on results.

“These analyses indicate that adjusting for clustering has very minimal effects on results.”

Sensitivity analyses using estimated ICCs produced similar point estimates with slightly wider confidence intervals.

“To assess the impact of unit of analysis errors, we conducted sensitivity analyses for estimated ICCs as above, showing little change in point estimates and slightly widening CIs.”

These results indicate that knowledge may be better gained immediately after the intervention by older children.

“These results indicate that knowledge may be better gained immediately after the intervention by older children.”

Three quasi-RCTs also used groups as the unit of randomisation.

“Three quasi-RCTs also used groups as the unit of randomisation (Crowley 1989;Daigneault 2012;Hébert 2001).”

Only four studies reported results in such a way that loss to follow-up for intervention and control groups could be differentiated.

“Only four studies reported results in such a way that loss to followup for intervention and control groups could be differentiated”

Outcome reporting was incomplete in five studies.

“On closer inspection, however, we noted that outcome reporting was incomplete in five studies.”

Taken together, results of the more conservative adjustment for clustering show short-term superiority of the interventions over control group effects.

“Taken together, results of the more conservative adjustment for clustering show the short-term (i.e.immediately post-intervention) superiority of the interventions over control group effects.”

Older children, on average, may score somewhat better than younger children when they complete these measures of applied knowledge immediately after the intervention.

“Thus, older children, on average, may score somewhat better than younger children when they complete these measures of applied knowledge immediately after the intervention.”

No study reported analysis on an intention-to-treat basis.

“No study reported analysis on an intention-to-treat (ITT) basis.”

Eleven of these were cluster-RCTs, where the unit of allocation was a group.

“Eleven of these were cluster-RCTs (Blumberg 1991;Dake 2003;Dawson 1987;Grendel 1991;Hazzard 1991;Kolko 1989;Krahé 2009;Kraizer 1991;Oldfield 1996;Poche 1988;Wolfe 1986), where the unit of allocation was a group (e.g.classroom or school).”

The intervention effect on disclosure changed depending on assumptions about clustering.

“These analyses, adjusted for unit of analysis errors, indicate that the effect of intervention programmes on disclosure was sensitive to different assumptions regarding the effect of clustering on the results.”

The unit of randomisation in 14 studies was clusters.

“The unit of randomisation in 14 studies was clusters.”

Two studies were classified as high risk of bias because not all measures discussed in the methods section were also reported in the results.

“We classified two studies as high risk of bias (Fryer 1987a;Wolfe 1986), because not all measures discussed in the methods section of the paper were also reported in the results.”

Clustering-adjusted sensitivity analyses produced nearly identical findings.

“Sensitivity analyses, adjusting for clustering yielded very similar results.”

Four further studies were classified as low risk of bias, reporting attrition rates of less than 10%.

“We classified four further studies as low risk of bias, reporting attrition rates of less than 10% (Dawson 1987;Kraizer 1991;Lee 1998;Snyder 1986).”

Due to poor reporting of variables that may be responsible for heterogeneity, it was not possible to further explore sources of variation via meta-regression.

“Due to poor reporting of variables that may be responsible for heterogeneity, it was not possible to further explore the sources of variation, for example, via the use of meta-regression.”

None of these studies reported appropriate analyses accounting for clustering effects.

“None of these studies reported appropriate analyses accounting for clustering effects.”

These studies did not report effect sizes to enable assessment of the magnitude of the relationships between self esteem, knowledge, and protective behaviours.

“These studies did not report effect sizes to enable assessment of the magnitude of the relationships between self esteem, knowledge, and protective behaviours”

Children who received a school-based sexual abuse prevention programme were substantially more likely to demonstrate protective behaviours in a simulated situation administered immediately after the programme ended.

“That is, children who received a school-based sexual abuse prevention programme were substantially more likely to demonstrate protective behaviours in a simulated situation that was administered immediately after the programme ended.”

These analyses indicate that adjusting for clustering has minimal effects on the results.

“These analyses indicate that adjusting for the effect of clustering have minimal effects on our results.”

One study did not provide a breakdown of data for intervention and control groups.

“One study did not provide a breakdown of data for intervention and control groups (Kraizer 1991).”

These analyses suggest that adjusting for clustering has only slight effects on results.

“These analyses suggest that adjusting for clustering has only slight effects on results.”

The choice of knowledge measure is an important source of between-study heterogeneity.

“An important source of heterogeneity across studies is the knowledge measure used.”

In four studies, outcomes were reported as summary statistics without means and SDs for continuous outcomes.

“In four studies, outcomes were reported as summary statistics (e.g.F-tests or T-tests) without including means and SDs for continuous outcomes”

The uncertainty in these results comes from inappropriate data analysis in some of the included studies.

“however results are uncertain because of inappropriate data analysis in individual studies”

The initial analyses ignored unit-of-analysis errors present in at least half of the studies in each meta-analysis.

“Initial analyses do not take account of unit of analysis errors that occurred in at least half of the studies in each meta-analysis.”

Trials that allocated classes within one school to intervention or control had a high risk of contamination.

“Studies which randomised classes within a single school to intervention and control groups were at high risk of contamination effects owing to the interaction of children in school playgrounds, friendship groups and families, and also from chance exposure to programme concepts via teachers and other school staff familiar with programme contents.”

Using one ICC for studies that randomised at class, school, and district levels could further inflate effect estimates.

“Furthermore, the same ICC was used for studies that had undertaken cluster-randomisation at class, school, and district level, which could further overestimate the magnitude of the findings.”

Evidence quality was rated moderate because of risk of bias, imprecision, or clustering adjustment in some studies.

“We downgraded the quality of evidence to moderate quality either due to risk of bias, imprecision, or because of the impact of adjusting for the effect of clustering within some of the studies.”

The descriptive analysis suggests that dichotomous labels such as active/passive or behavioural/instructional are artificial, since most programmes had multiple components.

“Our descriptive analysis shows this categorisation to be somewhat artificial as most programmes in this review were multifaceted with multiple components.”

Findings could have changed if true ICCs or author-provided cluster-adjusted results had been available.

“Therefore, results might have differed had the true ICCs from these studies been available, or had cluster-adjusted results been provided by the authors.”

In Dake 2003, incomplete outcome data were judged at high risk because attrition was 24% due to absenteeism and unmatchable questionnaires.

“Attrition was reported as 24% due to”

Even large cohort studies might not give definitive evidence about changes in the incidence of child sexual abuse.

“However, even large cohort studies may not provide definitive evidence for changes in child sexual abuse incidence, as it is under-identified and difficult to prove.”

• AssumptionRests on an assumption: Child sexual abuse is under-identified and hard to prove.

Most studies had unclear selection bias risk because information on random sequence generation and allocation concealment was inadequate.

“Most studies in this review were at an unclear risk of selection bias as illustrated in Figure 1 and Figure 2, due to inadequate information regarding methods of random sequence generation and allocation concealment.”

For Poche 1988, blinding of outcome assessment was judged at low risk of detection bias for all outcomes.

“Blinding of outcome assessment (detection bias) All outcomes Low risk”

• Also stated in: § Blinding of outcome assessment (detection bias) All outcomes · plain

  “This observer was blind to the experimental condition of each subject”

For Snyder 1986, blinding of outcome assessment was judged at high risk of bias because group administration made blinding impossible and assessors were also programme presenters.

“Blinding of outcome assessment (detection bias) All outcomes High risk Group administration of outcome assessment meant that outcome assessors would have to be blinded to the condition of entire classes.This was not possible as outcome assessors were also programme presenters”

For Kraizer 1991, allocation concealment was judged at high risk of selection bias.

“Allocation concealment (selection bias) High risk”

• Also stated in: § Blinding of outcome assessment (detection bias) All outcomes · plain

  “Method of concealment was not reported.”