A systematic review of universal campaigns targeting child physical abuse prevention

The purpose of this review was to better understand the impact of universal campaign interventions with a media component aimed at preventing child physical abuse (CPA). The review included 17 studies featuring 15 campaigns conducted from 1989 to 2011 in ﬁve coun-tries. Seven studies used experimental designs, but most were quasi-experimental. CPA inci-dence was assessed in only three studies and decreased signiﬁcantly in two. Studies also found signiﬁcant reductions in relevant outcomes such as dysfunctional parenting, child problem behaviors and parental anger as well as increases in parental self-efﬁcacy and knowledge of concepts and actions relevant to preventing child abuse. The following risk factors were most frequently targeted in campaigns: lack of knowledge regarding positive parenting techniques, parental impulsivity, the stigma of asking for help, inadequate social support and inappropriate expectations for a child’s developmental stage. The evidence base for universal campaigns designed to prevent CPA remains inconclusive due to the limited availability of rigorous evaluations; however, Triple-P is a not-able exception. Given the potential for such interventions to shift population norms relevant to CPA and reduce rates of CPA, there is a need to further develop and rigorously evaluate such campaigns.


Introduction
Child maltreatment is a major public health problem that places children at an increased risk for acute injury as well as long-term negative mental, social, behavioral and physical health outcomes. For example, victims of child maltreatment report significantly higher rates of depression, post-traumatic stress disorder, anxiety, suicide, behavior problems, interpersonal challenges, decreased productivity, cognitive deficits and chronic diseases compared with non-victims [1][2][3][4][5][6][7][8][9][10]. Brain imaging and epigenetic studies speak to the damaging biological effects child maltreatment has on a child's developing brain and DNA [5,11]. The economic burden of child maltreatment is estimated to cost an average of $210 012 per victim in the United States [12]. Early toxic stress and adverse experiences in childhood can set trajectories for poor health into adulthood that are linked with pervasive health disparities and public health problems, such as smoking, obesity, heart disease and stroke [2,8,10,13,14]. In 2011 in the United States, there were over 1500 fatalities from child maltreatment and nearly 700 000 substantiated cases [15,16]. However, because so many cases go unreported, the true prevalence of child maltreatment is unknown.
To date, most child physical abuse and neglect prevention programs target select populations of high-risk individuals, outnumbering populationlevel or universal target programs by about two to one [17]. There is a strong evidence base for the effectiveness of such programs targeting selected populations, such as the Nurse Family Partnership home visitation program for first time Medicaid eligible mothers [18] and the Incredible Years Parenting Training Program for parents of children with conduct disorders [19], both of which have shown significant reductions in harsh parenting for program participants. However, the most effective of these programs are highly resource intensive and require high levels of participant involvement [20]. Less resource intensive population-level prevention strategies that target universal populations can serve as important complements to these programs.
According to a seminal article by Rose [21], population-level interventions have some advantages over strategies targeting selected or high-risk populations. First, they tend to focus on root causes of a problem that are highly prevalent in a population. Second, because behavior is contagious, population-level strategies help to enhance and ease the efforts of interventions targeting selected populations. Third, by focusing on highly prevalent risks in a population, they have the potential to have a large impact on population attributable risk. The usefulness of population-level educational strategies, such as media campaigns, for addressing and improving public health problems is well documented. They are frequently utilized for increasing awareness of important health issues as well as decreasing the prevalence of or preventing a behavior [22,23]. Physical activity campaigns have raised public awareness of the issue and increased levels of physical activity [22]. Smoking [24,25], drunk driving and alcohol-related crashes [26] have also declined as a result of media campaigns. Other examples include the VERB campaign to promote physical activity among 'tweens' [27], campaigns to impact eating habits [28] and the Truth campaign designed to reduce smoking among teens [29].
To be most effective, such prevention strategies should focus on reducing risk factors that are highly prevalent and modifiable. The Social Ecological Model (SEM) posits that such factors exist at multiple levels, ranging from the individual and interpersonal to broader community and societal levels [30]. Klevens and Whitaker [17] identified a number of risk factors addressed in primary prevention interventions for child physical abuse (CPA) and neglect and categorized them by SEM levels. At the individual level, they identified some key modifiable parent characteristics commonly addressed such as: emotional arousal/reactivity/impulsivity, attributional biases, inappropriate expectations, lack of empathy, substance abuse, poor parenting skills, lack of knowledge of parenting techniques and use of harsh discipline. Key modifiable family level factors reviewed included family conflict, stress and isolation. Neighborhood factors identified mostly tied in with social isolation perceived at the individual and family level including social disorganization, low social cohesion and lack of access to needed services and support. Just one risk factor was covered at the societal level: social tolerance or acceptance of abuse-or use of corporal punishment (CP) for purposes of child discipline. However, Klevens and Whitaker considered that this factor '. . . may be the most prevalent risk factor for child abuse in the United States (p. 371)'. Indeed, CP is a strong risk factor for CPA [31][32][33] as well as other poor outcomes in children [34,35]. Further, a majority of children have experienced CP [36,37] and most adults believe it is a necessary form of discipline [38]. Aligned with Klevens and Whitaker's review, and written nearly four decades before, Garbarino identified two necessary conditions for child maltreatment to occur: parental isolation from support systems and cultural support for the use of physical force and discipline with children [39]. Based on these reviews as well as the unique ability of universal campaigns to address societal level risk factors, we included in our review those campaigns that address the use of CP as well as CPA. With these prior works as a guide, we sought to identify key, modifiable and prevalent risk factors that were addressed by the campaigns in our review.
Within the field of child maltreatment prevention, many of these individual and family level risk factors have been addressed successfully within programs focused on selective or high-risk populations [e.g. 18,40,41], but less is known about the ability of programs focused on universal populations Review of child physical abuse prevention campaigns to address these factors. Further, the majority of media campaigns within child abuse prevention have focused on child sexual abuse [42,43]. Yet, population-level strategies to prevent CPA can help to overcome some barriers often associated with selective parenting programs such as transportation issues, time conflicts, costs and limited capacity to reach all parents in need. The use of media in particular could provide help to normalize parenting challenges, to model appropriate behaviors and for parents to learn anonymously [23]. Hence, there is a need to better understand what risk factors are targeted and how successfully, specifically in universal campaigns for CPA prevention.
This review will be the first to focus solely on the prevention of CPA via population-level interventions (i.e. focused on universal targets) with a media component. Reviews by Klevens and Whitaker [17], Mikton and Butchart [44] and MacLeod and Nelson [20], all differed from the current review in their inclusion of selective and/or indicated interventions and by addressing multiple types of abuse. Although maltreatment types are certainly correlated, etiologies and prevention strategies can vary substantially and hence are worth examining separately for each type of abuse. Similarly, although interventions targeting universal and selective or indicated populations might address some overlapping risk factors, approaches and strategies for these intervention types are likely to be quite different.
Our systematic review of universal CPA prevention interventions with a media component has two primary aims. The first aim is to examine existing evidence from outcome evaluations to assess the effectiveness of these interventions. As we anticipate a paucity of rigorous evaluation studies [e.g. randomized control trials (RCTs)], we expect a systematic review to be more appropriate than a metaanalysis. The second aim is to examine and assess key risk factors addressed, campaign messages used and formative and process data to provide formative research that can be used to inform and improve future universal CPA prevention campaigns. As the application of evidence-informed, universal interventions to CPA prevention is a relatively young and underdeveloped field, we anticipate that this review will yield preliminary but modest evidence of program effectiveness (aim 1). At this stage of field development, we expect our formative research to be of particular use to public health practitioners and researchers engaged in the development and testing of such universal interventions for CPA prevention (aim 2).
Pubmed, Science Direct (Elsevier) and Web of Science (ISI). Second, a similar approach was used in Google rather than a journal database. Third, literature reviews of general child abuse campaigns were examined for additional campaigns or evaluations. Fourth, health or public health organizations were entered into Google search engine and organization websites were scanned for relevant campaigns, including Centers for Disease Control and Prevention (CDC), National Institute of Health (NIH) and World Health Organization (WHO). Finally, we contacted 11 authors and/or sponsoring organizations for which we identified campaigns but no evaluations via our search methods. Three of the 11 responded and two new evaluations were identified. However, neither of them met our inclusion criteria. Primary study authors and/or sponsoring organizations of the campaigns also were contacted by phone and email to obtain campaign materials and additional information regarding the campaigns. Internet searches through Google and YouTube were conducted to locate campaign materials.

Eligibility criteria
First, the intervention had to focus on reducing or preventing CPA or CP. Second, the intervention had to be a population-level strategy such as a mass media campaign or other educational efforts focused on universal targets. Interventions included various forms of mass communication (e.g., TV, radio, billboards, posters, report cards) or were delivered via community services with broad population access (e.g. hospitals, pediatric offices or schools). Studies of Triple P were eligible only if they included an examination of the Level 1 media component of Triple P designed for universal targets. Finally, the study had to report outcome evaluation results.

Exclusion criteria
Interventions dedicated solely to sexual, verbal or emotional abuse and/or neglect were excluded as were all parent training and home visitation programs without a 'campaign' (i.e. media or mass communications) component or focus on universal targets (i.e. interventions targeting selected or indicated populations were excluded).

Article selection
Method 1 identified 24 articles and 16 campaigns. Method 2 identified four more articles and 10 other campaigns. Method 3 identified three additional articles and three more campaigns. Method 4 yielded no additional reports or campaigns. One more article was obtained using Method 5. In all, 29 different campaigns were identified and fully reviewed. After review, 17 articles that reported on evaluations of 15 campaigns, met our study criteria and were included in our final analysis [46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61][62]. As this is a formative review and not a formal meta-analysis, multiple evaluations of the same intervention were permitted and counted as separate articles.

Data extraction
Information relevant to this review was extracted from campaign materials and evaluation articles and entered into a 'data extraction table' (e.g. see [22]). The following information was retrieved for each campaign: name, purpose and/or mission, description, program theory, specific campaign messages, format, materials used, links to materials and any additional information. For each evaluation study, the following information was extracted: author, year, title, name of sponsoring organization, organization/author contact information, source with corresponding link, evaluation design, sample size, sample characteristics, relevant outcomes to CPA or CP, evaluation tools, findings and any additional information.
All studies were analyzed to identify risk factors that were targeted to meet campaign goals. In a systematic review of primary prevention interventions for CPA and neglect, 30 risk factors were identified that have been targeted in previous CPA interventions [17]. This framework was used in our review of risk factors. However, new constructs emerged from a thorough investigation of program theories in these studies. The intervention was considered to have addressed a risk factor if it was explicitly stated or addressed in the evaluation article, in another Review of child physical abuse prevention campaigns article that described the campaign (e.g. [45,63]), or in the actual campaign materials.

Campaign descriptions
In total, 17 evaluation articles, featuring 15 separate campaigns, were analyzed and reviewed ( Table I). The evaluated interventions were carried out in five countries, including the United States (n ¼ 9), Australia (n ¼ 2), United Kingdom (n ¼ 2), New Zealand (n ¼ 2), Canada (n ¼ 1) and Japan (n ¼ 1). Target audiences ranged from the general public to adults over the age of 18 years, parents, parents and caregivers with children of a certain age and caregivers. Implementation of the campaigns spanned a period of 12 years with the first campaign beginning in 1989 and the most recent campaign ending in 2011. Campaign durations ranged from 6 weeks to 4 months (n ¼ 3), 1-2 years (n ¼ 11) and 3-9 years (n ¼ 3). Although all campaigns focused on reducing or preventing CPA or CP, some were dedicated to specific issues such as SBS or positive parenting. Five campaigns included at least the universal, Level 1 media component of Triple P [45].
A wide range of media was used in the campaigns including print, broadcast and online sources. Print media was widely used through billboards, newsletters, publications, posters, work books, compliance statements, pamphlets, resource guides, balloons, press releases, newspapers, handouts, postcards, street posters or bus advertisements (n ¼ 14). Most campaigns used broadcast media such as public service announcements (PSAs) on TV, TV segments/episodes, videos, PSA videos, radio segments or PSAs on the radio (n ¼ 14). Online media such as email, websites, web videos, online support or access to online resources was less common (n ¼ 3). Several campaigns provided telephone support and toll-free helplines for fast access to counseling services and parent tips and to report cases of CPA (n ¼ 3). Both pre-existing and novel materials were utilized by the campaigns. For example, two SBS campaigns used 'The Period of PURPLE Crying' materials that had already been implemented in healthcare settings [51,62]. Other campaigns developed novel materials such as the 'STOP. LOOK. LISTEN' card that was attached to children's report cards [52].

Formative evaluation
Six studies (35.3%) reported formative evaluation procedures (Table II). Reported formative strategies included pre-testing of campaign messages with focus groups [46,52,62], meetings/discussions with key informants [46,52,61], review of literature and resources relevant to the intervention and message development [50,54,61] and tailoring materials to the sample population such as via matching literacy levels and pre-campaign surveys of attitudes [46,52].

Process evaluation
Nearly all studies (94.1%) reported at least one form of process evaluation, such as the distribution of campaign materials, procedures such as trainings of professionals and measures for monitoring reach. Two of the most commonly assessed process evaluation methods were awareness of the campaign and recall of receiving intervention materials. Over 40% assessed awareness or recall of the campaigns at mid-or post-intervention. Rates of campaign awareness or recall ranged from a high of 85% [54] to a low of 17.1% [56]. Nearly all participants (92-98%) in two of the SBS studies [48,49] confirmed that they received information about SBS.

Risk factors addressed in campaigns
Among the 17 studies, eight key risk factors were identified and definitions were established for each (Table III). Sample campaign messages also were selected to exemplify how campaigns targeted each risk factor. The most frequent risk factors addressed by the campaigns were lack of knowledge or skills regarding positive parenting techniques (82.4%), parental impulsivity (76.5%), lack of knowledge regarding child development or inappropriate expectations for a child's developmental stage (64.7%), inadequate social support (64.7%) and the stigma of asking for help (64.7%). Less commonly targeted M. K. Poole et al. Review of child physical abuse prevention campaigns

Press releases
Other Triple P Components: .

Consultations with parent
training (20 min Review of child physical abuse prevention campaigns Review of child physical abuse prevention campaigns

Radio segments with medical professionals (10 min)
.

Posters in medical and education centers
Simple pre-and postintervention

OE
from pre-to post-intervention Review of child physical abuse prevention campaigns Review of child physical abuse prevention campaigns   'Crying is not a reflection on your skills as a parent or caregiver. Crying can't be controlled'.
. 'Think about the 2-2-2 theory. Babies begin to cry as early as 2 weeks, crying peaks at 2 months and a baby can cry up to 2 hours a day'.
. Infant soothing techniques include the following: 'holding, rocking, singing, playing soft music, and/or feeding'.
. 'Identify a family member or friend who can support mom or dad if they need a break anytime during the day or night. Write their phone number next to the phone and call them'.
. Self-coping strategies includes the following: 'exercising, calling a friend or family member, taking a time-out, meditating or deep breathing, and/or listening to music'. (continued) Review of child physical abuse prevention campaigns 'We need to understand that crying is a way babies communicate. It might mean the baby is hungry. It might mean that the baby is tired. It might mean that the baby is sick or that the baby just wants to be held. But it doesn't mean the baby is angry. It doesn't mean that baby is out to get his parent. Or that the parent isn't doing a good job'.
. 'When a baby is fussy or is crying, you might try feeding the baby slowly, offering a pacifier, taking the baby for a walk or a ride in the car or simply holding the baby. Maybe the baby is too cold or too warm or isn't feeling well. If you think the baby might be sick, call a healthcare professional for advice'. e    . 'Being a parent is hard, but there is help. Call Florida's FREE and CONFIDENTIAL parent helpline'.
. 'Distract and divert is a way to take your child's attention off one thing and focus it onto something else. When your child is doing something you don't like or that may be dangerous, give your child something else to do'. 'I was worried at first. Her mama was so young. I told her mama I would watch the baby so she could go to the grocery store and X X X X X X (continued) Review of child physical abuse prevention campaigns Table III. just take a few minutes to herself. I still do it today. It doesn't take much of my time and her mama said it was such a big help. It's your turn to raise the leaders of tomorrow. Find out how at ounce.org'.

Alter damaging behaviors in families
. 'It's not OK to be cruel to your boy just because it never did me any harm, or scream abuse at your children just because you're unhappy. It's not OK to control your family with threats, to bully them, to intimidate them or  McLaren [53] .

Alter damaging behaviors in families
.
Reduce social norms that tolerate family violence (Same as above) X X X X Norton et al. [55] .  Review of child physical abuse prevention campaigns . 'Some babies cry a lot when they are hungry, wet, tired, or just want company. Some infants cry at certain times of the day or night (usually when you want to sleep or eat). Feeding and changing them may help, but sometimes even that doesn't work'.

Raise public's awareness of SBS
. 'If your baby cries a lot, try the following: feed the baby slowly; burp the baby often. Offer the baby a pacifier. Hold the baby against your chest and walk or rock him/her. Take the baby for a ride in a stroller or car or put him/her in a baby swing. If you breastfeed, avoid eating onions or beans or drinking coffee, tea, or cola'.

Outcome evaluation
Seven different evaluation designs were used, with some studies using more than one type: randomized control trial (RCT) (n ¼ 7); Single survey at mid-or post-intervention (n ¼ 4); Pre-and mid-or postintervention, one group (n ¼ 3); Simple time series (n ¼ 2); Time series with comparison group (n ¼ 1); Non-equivalent groups, pre-and post-intervention (n ¼ 1) and survey with timing ambiguous in relation to the campaign (n ¼ 1). Participants were recruited through a variety of probability and nonprobability sampling methods. Only three studies examined child abuse outcomes specifically and found a reduction in the incidence of abusive head injuries [49], child maltreatment injuries [56] and child maltreatment cases [52,56]; though one was not statistically significant [52]. However, some measure of behavior change was assessed in more than half (58.8%) of the studies [46,47,[49][50][51][52][53][56][57][58][59]. Significant decreases were seen in child behavior problems [47,[57][58][59] and in dysfunctional or coercive parenting behaviors [47,51,57,59]. Two studies reported increases in calls to helplines to report child abuse cases [46,52]. Increases also were seen in the number of callers wanting to seek assistance from the helpline or report parental alcohol and drug abuse [46]. Two studies reported statistically significant increases in the number of attempts by parents and/or community members to prevent child abuse through strategies promoted in the campaign, such as assisting parents by watching their children [50] and sharing information about excessive infant crying with other caregivers [51].
Attitudes were assessed in seven of the studies [46, 50, 53-55, 61, 62]. However, only one study reported a significant improvement in positive attitudes toward preventing child abuse from pre-to post-intervention and across two groups [50]. Knowledge was assessed in six studies [48,50,51,55,61,62]. Significant increases were seen in knowledge of SBS [48,62], child development, community resources [50] and the causes and patterns of infant crying [51,62].
Additional indicators that were frequently assessed included parental self-efficacy and parental anger. Significant increases in parenting self-efficacy or competence were reported in three [47,57,58] of four studies, while significantly decreased parental anger or frustration was reported in two [47,57] of three studies. One study assessed and found improvements in parents' intentions to use appropriate and positive child discipline strategies [60].
Of the 17 studies we reviewed, six were focused on the prevention of SBS. Two utilized 'The PERIOD of PURPLE Crying' materials [51,62] and the remaining used a variety of multimedia materials [48,49,55,61]. Significant improvements were seen in variables such as knowledge of SBS [48,51,62], knowledge of infant crying patterns [51,62] and ways for parents to cope with excessive baby crying, such as walking away and self-talk strategies [51]. One study found a significant decline in the number of abusive head injuries among infants at follow-up [49].
An overview of study designs and interventions is provided in Table IV.

Discussion
Our extensive review of evaluations of populationlevel CPA prevention programs with a media Review of child physical abuse prevention campaigns  component revealed only 17 articles that met our eligibility criteria. This paucity points to a gap in existing CPA prevention efforts focused on universal target populations. In addition, the shortage of evaluations retrieved for this review underscores the importance of evaluating CPA campaigns. Findings from our review indicate that the evidence regarding this method's ability to reduce CPA is insufficient, primarily due to weak evaluation methodologies. However, the evidence for the Triple P program in particular is promising. Lending to the strength of this review, nine of the evaluation studies used comparison (two) group designs, seven of which were RCTs. Three studies actually assessed CPA rates and detected significant reductions. But only two of these used strong evaluation designs including an RCT of an intervention that addressed seven out of eight identified risk factors [56] and a time series with a comparison group that addressed all eight risk factors [49]. The third study used a simple time series design and only addressed half of the identified risk factors (parental impulsivity, lack of knowledge or skills regarding positive parenting techniques, lack of social support and stigma of asking for help) [52]. Further, many of the reviewed interventions produced significant parent and child behavioral effects [47,51,[57][58][59] as well as improvements in parents' knowledge of CPA, CP, SBS and/or neglect [48,50,51,62]. The impact of these interventions on attitudes, beliefs and change in awareness is less conclusive because these outcomes were measured less frequently or the evaluation designs were less strong. We found that eight key risk factors were frequently targeted for CPA prevention. These risk factors were identified a total of 85 times across the examined articles. Campaigns that targeted the most common risk factor addressed, lack of knowledge or skills regarding positive parenting techniques, aimed to promote positive parenting strategies and to empower parents to use non-punitive discipline techniques. Campaigns that targeted the next most common risk, parental impulsivity, aimed to teach parents how to control angry outbursts, which can lead to CPA or CP, through positive parenting strategies such as walking away or taking a break when parenting becomes too frustrating. The next most common risk was inappropriate expectations that parents can have for their children. Campaigns that addressed this aimed to improve parents' knowledge of children's developmental milestones and normal behavior for certain ages so that parents were less likely to over-or underestimate their child's capabilities [47-49, 56-59, 61]. Such lack of knowledge can lead to parental frustration, anger and use of punishment in the face of 'normal' child behavior. Also targeted just as frequently as inappropriate expectations were lack of social support and stigma of asking for help. To improve social support, campaigns encouraged community members to be aware of signs of child abuse, to offer a helping hand to other parents and to increase awareness of parenting resources. To reduce the stigma of asking for help, campaigns encouraged parents to seek help when they became overwhelmed or frustrated and also normalized the need for help with parenting. This approach aims to discourage parents from feeling inadequate in their own parenting skills and to normalize parenting challenges.
The three least commonly targeted risk factorslack of knowledge about the consequences of physical punishment or infant shaking, parents' negative attribution biases and parents' lack of self-carewere all still targeted in about half of the studies. Many campaigns sought to increase the knowledge that use of physical punishment even that which is not classified as abuse, can harm children's mental and physical health. To address negative attribution biases, campaigns aimed to help parents understand that although children can be difficult, they rarely do things deliberately to anger parents. For example, parents may misinterpret normal child exploration that results in making a big mess or pulling items off shelves as a sign of disrespect or trying to anger the parent. And finally, many campaigns reminded parents not to neglect their own mental and physical health.
Of the two interventions that demonstrated statistically significant decreases in CPA, one targeted reductions in SBS and addressed all eight identified risk factors [49] and the other used the full Triple P Review of child physical abuse prevention campaigns program and addressed seven of the identified risk factors [56]. Many of the studies that revealed significant outcomes were those that targeted SBS or that used Triple P intervention materials. The Triple P-based interventions are also noteworthy because they consistently demonstrated significant improvements in dysfunctional parenting, parental self-efficacy and child behavior problems. Triple P appears to focus primarily on promoting positive parenting strategies. In contrast, interventions targeting SBS tended to include explicit details about the harmful effects of physical punishment or shaking infants in addition to promoting positive alternatives. The consistency in these effective interventions suggests that the risk factors they target should continue to be integrated into future campaigns to prevent CPA. Other programs might learn a lot from these successful interventions.
Although we had hoped to learn more about each campaign's success in addressing specific risk factors, our ability to do this was highly limited. First, most studies did not assess changes in all of the key risk factors that they targeted. For example, although most interventions addressed inappropriate expectations for child developmental stage and also the stigma of asking for help, very few studies actually measured and assessed changes in these risk factors. Second, many studies that did measure their targeted risk factors did not apply a rigorous enough methodology to draw strong conclusions from their findings. We would encourage future program evaluators to address both of these issues whenever possible: that is, to explicitly assess key targeted risk factors and, of course, to use the strongest methodologies possible to ensure greater confidence in results. Such efforts will help to advance program theory in this field by deepening our understanding of how and why certain interventions 'work' while others do not. Such knowledge is essential to developing future successful campaign interventions.
Beyond the key risk factors that we tracked, program effectiveness might well be tied to other key intervention adaptations and sensitivities that were not well documented, such as attention to issues of cultural relevance and stages of child development.
On the former issue, while many of the reviewed campaigns reached broad audiences, not all of the campaigns were designed to be culturally sensitive or tailored to different ethnic groups. Three interventions offered campaign materials in multiple languages [49,60,62]. Another intervention used program materials that were thoroughly translated from English to Japanese and then back translated to ensure that the campaign messages were not lost in translation in Japan [51]. Other interventions included diverse characters in the campaigns to make them relatable to the target population [50] with one specifically matching campaign materials to the mother's race [61]. Other interventions were conscious of local cultures and national trends. Mandell [52] wrote the campaign materials on a second or third grade reading level to account for the high levels of illiteracy in the target population, whereas several Triple P Level 1 interventions were based on the nationwide popularity of 'infotainment' [47,57,58]. Finally, several articles cited previous studies or presented findings that the campaign materials were designed to be appropriate for [52,53] or were effective across multiple cultures [56,59,60]. In short, some issues of cultural sensitivity and adaptation were addressed by close to half of the campaigns, but very few demonstrated the extent, success and effectiveness of these adaptations.
Extent of intervention tailoring to relevant stages of child development might also be an important factor in program success, particularly given that appropriate parenting instructions and CPA risk varies by child age. The SBS interventions, which specifically target parents of infants, are strong examples of tailoring to child developmental stage. All of the SBS interventions reviewed aimed to educate parents about the dangers of shaking during the critical newborn period [48,49,51,55,61,62]. Other interventions based their campaigns on stages of child development. For example, noting an increase in CPA during report card periods, Mandell [52] attached resources for CPA prevention to schoolaged children's report cards. Evans et al. [50], who found a significant increase in knowledge of child development but not in knowledge of M. K. Poole et al.
age-appropriate discipline strategies, distributed parenting resource guides with tips for managing common child behavior problems at each developmental stage. Sanders et al. [59] developed their intervention for parents of children making the transition to school, citing that parents might be more willing to engage in programs during an important time for the parents and child. Several articles mentioned Triple P's ability to assist parents with children of all ages [47,[56][57][58] and one emphasized the benefits of Play Nicely in teaching parents age-appropriate discipline strategies in the early years of the child's life [60]. Future reviews would benefit from more careful documentation and tracking of program adaptations and attention to issues such as child development and cultural sensitivity.

Strengths and limitations
This review has several limitations. Only one independent reviewer conducted the search for relevant literature using Method 1 (journal database search), therefore, some eligible articles may have been overlooked in the initial search. However, two research assistants exhausted Method 2 (search engine). Methods 3, 4 and 5 were used to further reduce the risk of missing any eligible studies. The linking of interventions with risk factors was based solely on what was written in the text description and available campaign materials and was not verified with the article author or campaign program staff. It is possible that some campaigns addressed more risk factors than we identified based on what was written in the text or campaign materials. However, as we did not have access to all campaign materials, this was the least biased method that we could choose. Moreover, the availability of many Triple P evaluations and formative articles (e.g. [45]) may have placed the program at an advantage over the other interventions in our review. Given the complexity of the multilevel intervention, it was difficult to dissect the program goals and risk factors that were targeted at each level. Therefore, it is likely that the interventions with only Level 1 components did not target as many risk factors as the interventions with Levels 1-5 of Triple P.
Additionally, the specific campaign messages in Table 3 were retrieved from available campaign materials located in the evaluation articles or online. One research assistant included the messages that she deemed most important and relevant to the targeted risk factors. The lack of accessible campaign materials for multiple interventions may have limited our ability to draw comprehensive conclusions about targeted risk factors, the most effective messages, links to program outcomes and overall recommendations.

Conclusions
Our formative research from this review suggests that CPA prevention program developers might consider integrating components into their programs that address the most commonly targeted risk factors highlighted in this review that were linked with significant outcomes. We recommend, at minimum, incorporating the most promising risk factors into CPA prevention programming, including reducing parental impulsivity, reducing the stigma associated with asking for parenting help, increasing social support for parents, increasing knowledge and use of positive parenting techniques and increasing knowledge of appropriate expectations for a child's developmental stage. In particular, Triple P and effective SBS program materials should undergo further rigorous evaluation to confirm their effectiveness in reducing CPA. The use of helplines also appears promising and should be integrated into future interventions and further evaluated.
This review provides a preliminary assessment of the impact of population-level interventions with a media component for CPA prevention and suggests that, although the evidence for Triple P is promising, the current level of evidence for other examined interventions is insufficient to draw solid conclusions and additional studies are needed. Overall, more rigorous evaluations should be conducted in order to broaden the evidence base for these types of interventions. Future evaluation studies would benefit from the inclusion of clear program theory Review of child physical abuse prevention campaigns descriptions along with a clear review of targeted risk factors and their linkages with program messages and components. Further development and testing of universal CPA prevention campaigns is important given their potential for communitylevel impact. The very high economic [12] and public health [8] costs of child maltreatment to society mean that even modest prevention program effects on universal targets could have a major positive impact on society.