Are Environmental Interventions Targeting Skin Cancer Prevention among Children and Adolescents Effective? A Systematic Review

Skin cancer, which is increasing exceedingly worldwide, is substantially preventable by reducing unprotected exposure to ultraviolet radiation (UVR). Several comprehensive interventions targeting sun protection behaviors among children and adolescents in various outdoor settings have been developed; however, there is a lack of insight on stand-alone effectiveness of environmental elements. To compose future skin cancer prevention interventions optimally, identification of effective environmental components is necessary. Hence, an extensive systematic literature search was conducted, using four scientific databases and one academic search engine. Seven relevant studies were evaluated based on stand-alone effects of various types of environmental sun safety interventions on socio-cognitive determinants, sun protection behaviors, UVR exposure, and incidence of sunburns and nevi. Free provision of sunscreen was most often the environmental component of interest, however showing inconsistent results in terms of effectiveness. Evidence regarding shade provision on shade-seeking behavior was most apparent. Even though more research is necessary to consolidate the findings, this review accentuates the promising role of environmental components in skin cancer prevention interventions and provides directions for future multi-component sun safety interventions targeted at children and adolescents in various outdoor settings.


Introduction
Melanoma and non-melanoma skin cancers (NMSC) are two of the most rapidly increasing cancer types among white populations [1]. Since the early 1980s, melanoma incidence rates have risen twofold in the United States to even threefold in Europe [2][3][4][5]. On a global level, more than 55,000 people died from melanoma in 2012, with the greatest burden in Europe, the United States, Australia, and New Zealand [6,7]. Continuation of the rising incidence rates of melanoma is predicted until 2022 at least in the United States and among European countries [3], also implying increased health care costs and need for a new skin cancer disease management strategy [8,9]. Even though exposure to ultraviolet radiation (UVR) is important for production of vitamin D [10], sun exposure and sunburns during early childhood are the most important risk factors for developing melanoma in later life [11][12][13][14][15] and should therefore be limited [16,17]. Objective data about the overall time children are exposed to UVR are inconsistent and vary per age group, latitude, and country of origin [18][19][20][21]. Even though insight in specific settings where children spend their outside time nowadays is scarce, children seem to often engage in outdoor activities at (pre)school [22,23], around the house and around the beach or swimming pools [24,25], when playing in outdoor playgrounds [26,27] or in public parks [28]. Compared to adolescents, younger children spend time outside after school and during the weekend to a greater extent [16,22].
The Surgeon General and World Health Organization (WHO) have documented five guidelines to enhance sun protection in a call to action addressing the rapidly increasing skin cancer incidence rates. These behaviors consist of wearing protective clothing, wearing sunglasses and a hat, seeking shade, avoiding peak sunlight hours, and applying sunscreen [29]. Over the years, several types of interventions have been developed to encourage various sun protection behaviors among parents and children. Educational interventions for example, can be individually directed and primarily focus on changing intentional decision-making processes [30], by increasing one's knowledge, improving socio-cognitive determinants (e.g., attitudes, self-efficacy) and learning skills to perform a certain behavior. Positive effects of such interventions have been shown for different sun protection behaviors [31,32] and on knowledge, attitudes, intentions, and behavior in various settings [31,33] for children and adolescents [34,35].
Although educational interventions have demonstrated positive effects, it is important to acknowledge that behavior can also be automatically triggered by environmental characteristics [36]. This is illustrated by dual process models, which state that behavior can both consciously and automatically be influenced by one's physical environment, such as by physical adaptations, policy or both [37]. One's environment can be characterized by both different levels of influence (e.g., micro/family setting or macro/community level) as by different types of the environment (e.g., political, economic, social, or physical). Hence, adapting the environment where children and adolescents are highly exposed to UVR can affect sun protection behavior [37].
Overall evidence for effectiveness of multi-component sun safety interventions, integrating both behavioral and environmental strategies, among children and adolescents in outdoor settings, is restricted [38]. Moreover, insight in effects of autonomous elements of these interventions is lacking and additional research is necessary. Since childhood is an important phase where consolidation of health behavior takes place [39] and life-long sun protection habits can be established [40], effectiveness of these interventions needs further examination.
In conclusion, a comprehensive approach in skin cancer prevention strategies among children, targeting both behavioral and environmental factors, is needed. Yet, in order to compose a mix of effective strategies targeting both types of factors and therefore design future interventions optimally, identifying the effects of separate components targeting these factors is necessary. Hence, the aim of this review was to systematically investigate available literature concerning stand-alone effectiveness of environmental interventions targeting sun protection behaviors among children and adolescents in various outdoor settings.

Search Strategy
The PRISMA guidelines for systematic reviews were followed to enhance reproducibility of this study [41]. The formulated research question included study characteristics according to the PICOS tool, affirmed by the Cochrane Collaboration [42]. Development of the search strategies was accomplished with help of a scientific information specialist (author S.J.). Since the setting where an intervention took place was specifically important, 'setting' characteristics replaced the standardized 'comparison' element out of the standardized PICOS in the search strings. An extensive literature search regarding environmental sun safety interventions was conducted using four databases (i.e., PubMed, PsycInfo, Cochrane, Web of Science) and one academic search engine (i.e., Google Scholar) applying systematic formulated search strings. The search strings contained five index terms: (1) population, (2) intervention, (3) setting, (4) outcomes, and (5) study design. These databases together covered a broad range of health, behavioral, and social science subjects as well as the ability to search for scholarly literature. Examples of two search strings are depicted in Appendix A.

Eligibility Criteria
Prior formulated inclusion criteria were for studies looking at: (1) stand-alone effects on socio-cognitive determinants, sun safety behaviors, UVR exposure, and sunburn and nevi incidence; (2) physical environmental, policy, economic, and/or socio-cultural interventions [37]; (3) outdoor and school settings; (4) among children aged between 0 and 18 years; (5) including intervention designs with at least one comparison group, were eligible for inclusion. Stand-alone effectiveness of environmental interventions was interpreted as such that the sun safety intervention, whether it represented a single or multi-component intervention, should consist of one separate intervention arm in which the environmental component was exclusively tested. Inclusion and exclusion criteria are depicted in Table 1. Moreover, studies that were conducted before 1990 and were written in non-English were omitted. No filters were used in the databases and the academic search engine since the search strings were optimally and carefully designed and inclusion and exclusion criteria were clear.

Study Selection
The systematic search took place from the 17 August to the 4 October 2018. An updated search to ensure inclusion of all available recent studies was performed on the 17 September 2019. After eliminating duplicate studies according to the method described by Bramer and colleagues [43], title selection of the studies took place by the first (KT) and last (FS) author in the first round. Consensus between KT and FS about selected studies based on titles needed to be reached before continuing to the next stage. In the second round, eligible abstracts were selected. In the third round, full-text articles were selected. When discrepancies between the two researchers were observed based on the title or abstract, the paper was taken to the second or third round in the study selection process. Both KT and FS determined upon final selection of relevant articles that were used for data abstraction and disagreements were discussed. Furthermore, when no consensus about eligibility could be reached, a third researcher (author L.O.) was consulted.

Data Abstraction
Study characteristics and study outcomes were extracted from the selected studies and quality assessments to estimate the risk of bias were performed. Descriptive information about detailed study characteristics, study outcomes and data regarding quality of studies were reviewed independently before abstracting relevant data.

Study Characteristics
A standardized data abstraction form [44] was critically examined and altered regarding specific characteristics of the studies that were selected, in consensus with the study objectives. Characteristics of the selected studies that were abstracted were predominantly formulated based on the PICOS framework and can be found in Table 2. After entirely reading the first included study [45], study characteristics were further specified according to elaborate data that was present in this study.

Study Outcomes
Study outcomes were regarded important for extraction based on previous systematic reviews [32,38] and were related to socio-cognitive determinants, sun safety behaviors, UVR exposure, and health-related outcomes (i.e., reported sunburns and/or melanocytic nevi (i.e., moles)). The behavioral outcomes were based on recommended sun safety guidelines [29,46]. Since occurrence of sunburn and melanocytic nevi are both objectifications of (over)exposure to UVR, these were included as outcome measures [32]. Finally, information about statistical analyses that were conducted, statistical results, reported stand-alone effects of the intervention and if applicable, reported effect sizes were also abstracted in order to gain an overview of relevant study data.

Study Quality and Risk of Bias
The quality of included studies was assessed by using the Quality Assessment Tool for Quantitative Studies from the Effective Public Health Practice Project (EPHPP) [47], which is a validated tool to use in the assessment of quality of studies [48,49]. With regard to assessing the quality of studies included in the review, methodological characteristics were rated as 'strong', 'moderate', or 'weak'.

Results
In total, 1085 articles were found in the five databases and were screened for eligibility. After the de-duplication process, 753 titles remained. After screening of titles and abstracts, 45 articles were eligible for full-text screening. In the full-text phase, three authors were approached to retrieve more information upon deciding whether these articles were eligible for data abstraction, since follow-up data was not present [50][51][52]. The first authors of these studies were contacted to investigate whether data were available. In one case, a study protocol described the methods for conducting a study in which effects of shaded areas in public parks was investigated from 2013 to 2016 [51]. The first author of this article mentioned that the data would be submitted on a short notice and that follow-up results were presented at two conferences. The follow-up data of this study was therefore included in the analysis as a conference abstract. Another study [52] only described baseline data. After contacting the first author, it was mentioned that follow-up data was retrieved and presented at a conference in 2010, however publication was not expected to be established anytime soon. Therefore, the data from this study were included as a conference abstract too [53]. Finally, the follow-up data of one study [50] was eventually excluded in the analysis after contacting the authors, since stand-alone effects could not be distinguished for children or adolescents specifically. Ultimately, from the 45 articles, eight studies that described the stand-alone effects of environmental interventions were qualified for data-abstraction (see Appendix B for the flowchart). After excluding the last article [50], this resulted in a total of seven studies that were used in the data-extraction process. A full overview of study outcomes and results is provided in Table 3.

Types of Environmental Components
In most of the studies, an economic intervention [54] was included in an intervention-arm. More specifically, proactive provision of free sunscreen was the most applied environmental component. In three studies free sunscreen was the exclusive addition in the environmental intervention-arm, while in one study sunscreen provision was one of multiple components. In two studies, bottles of sunscreen were provided to parents at school near the end of the school year [55] or daycare center before spring [56]. In one study, sunscreen was handed out to children themselves, at the end of the school year [57]. Lastly, free sunscreen was provided among other factors (portable shade tents, posters, and policy consultation), in which sunscreen dispensers were placed in city parks, community centers and outdoor recreation sites [45]. In another economic intervention, provision of protective clothing was used [53]. The entire intervention was based on arrangement of clothing, hats, as well as swim shirts for children attending daycare centers.
Subsequently, the second most used intervention type was an adaptation in the physical environment, consisting of shade provision in three studies [54]. The effectiveness of shade sails at secondary school sites [58] and purpose built shade and trees covering public park areas [59] were investigated. Moreover, effects of portable tents, besides sunscreen dispensers, were examined in one study [45].

Effects of Environmental Interventions on Socio-Cognitive Determinants
One study described socio-cognitive determinants among other outcomes [57]. In this case, the environmental intervention consisted of free provision of sunscreen. Children showed a significantly greater reduction in their desire to have a tan in comparison with children from the other intervention and control group. No effects were found on knowledge and awareness.

Effects of Environmental Interventions on Sun Safety Behaviors and UVR Exposure
Six studies had one or more sun safety behaviors as outcome(s); UVR exposure was part of the outcome measures in two studies [55,56]. Two studies solely assessed shade-seeking behavior [58,59] and one study focused on sunscreen use [55]. In the remaining studies more than one sun safety behavior was measured [45,56,57]. Parents estimated their children's execution of sun safety behaviors in most of the studies [45,[55][56][57]. In one study, parents and children were both assessed [57]. The effects of different types of environmental interventions on sun safety behaviors were variable. For instance, economic interventions consisting of free sunscreen or sunscreen dispensers did not account for significant improvements in parental or children's sun safety behaviors or UVR exposure [55][56][57]. Physical changes in the environment, consisting of shaded areas and purposely planted trees, showed positive effects on shade-seeking behavior of children and adolescents in two studies [58,59]. However, in a study where sunscreen dispensers were combined with shaded areas, behavioral effects were absent [45].

Effects of Environmental Interventions on Melanocytic Nevi and Sunburns
One study assessed nevi incidence as exclusive outcome [53], two studies evaluated incidence of nevi among other outcomes [55,56], and one study investigated sunburn incidence [57]. Findings regarding the effects of economic interventions on nevi and sunburn incidence, such as free provision of sunscreen, were contradicting. One study found that children who received free sunscreen had developed less nevi at the end of the intervention period [55], whereas another study found no effects of free sunscreen provision on nevi incidence [56]. Lastly, in the only study that investigated the effects of free sunscreen provision on sunburn incidence, no evidence for effects was found [57]. In the only study that investigated free provision of UV-protective clothing, hats, and swim shirts [53], findings show that children who did not receive the intervention developed higher incidence of moles on their bodies than children in the intervention group. Specific information about the usage of these garments by child care staff or parents was however absent.

Quality Assessment of Studies
Weak coding was most often due to absence of reported controlling for confounding variables in study designs or statistical analyses. Furthermore, data collection methods were coded as weak in two studies, due to the lack of validated questionnaires [45,57]. Drop-out rates were above 40% in two studies, resulting in a weak coding for this sub-item [45,57]. All studies were randomized or cluster controlled trials, except for one non-randomized trial with pre-and post-tests [59]. Two studies guaranteed blinding in which the outcome assessors were not aware of the intervention status of participants [55,56] and in two studies, blinding of assessors was not possible due to observational methods [58,59]. In most studies, respondents were not aware of the research question. In one study, the blinding procedure was not explained [53]. Notably, none of the included studies reported effect sizes between intervention and control groups. Since the included studies were heterogeneous in terms of intervention type, outcomes, and statistical results and specific statistical information was absent in three studies, a pooled effect size was not calculated. See Table 4 for the quality rating per study.

Discussion
This systematic literature study is the first to examine stand-alone effects of environmental sun safety interventions among children and adolescents on socio-cognitive determinants, sun safety behaviors, UVR exposure, development of nevi, and sunburn incidence. Seven studies were included, showing that free provision of sunscreen (four times), shade supply (three times) and provision of UV-protective clothing and accessories (one time) were the environmental types of interventions implemented. Five studies showed significant effects of environmental components, assessed after one year on average (yet ranging from four months to three years). Positive effects were especially visible on shade-seeking behavior and incidence of nevi; effects on socio-cognitive determinants, other sun safety behaviors, UVR exposure, and sunburn incidence were not evident. Overall, shade provision seemed to show the most encouraging results. Recent reviews about the effects of shade provision on sun safety behaviors among adults show encouraging results as well [60][61][62].
First, in respect of environmental interventions implemented in the studies included, five used an economic component such as free provision of sunscreen [55][56][57], free sunscreen dispensers [45], or free provision of UV-protective clothing and garments [53]. Physical changes in the environment such as supplying shaded areas, portable shade tents, or planted trees were used in three studies [45,58,59]. Furthermore, sunscreen provision was the most frequently mentioned intervention, even though sunscreen use alone is not sufficient for UVR protection [63,64] and is currently considered an additional recommendation besides other sun safety practices [29,65]. Despite the fact that provision of sunscreen is relatively accessible and low-cost, economic interventions such as provision of clothing, hats, and sunglasses warrant further exploration as well [66,67], since these methods seem to be more effective than sunscreen application [63,68].
Second, when looking at the outcomes of the included studies, results showed that most outcomes were based on internationally recommended sun safety behaviors [29,46], in which sunscreen use was regularly the primary behavioral outcome [45,[55][56][57], followed by shade-seeking [58,59] and wearing UV-protective clothing [45,56]. The preference of measuring sunscreen use prior to shade-seeking and clothing behavior in the studies is in accordance with a general popularity of sunscreen application [68]. Even though it is encouraging that environmental sun safety interventions assess a variety of behavioral outcomes, results of the included studies show that actual effects of environmental interventions on behavioral change are scarce and were only found for shade-seeking behavior [58,59]. These results accentuate the importance of enhancing other sun protection behaviors besides sunscreen use [29]. Furthermore, effects of environmental interventions on other outcomes such as incidence of nevi were variable. Two studies showed positive results considering incidence of nevi on the body [53,55]. However, in these studies, no differences in UVR exposure between intervention and control groups were reported while UVR exposure is a great predictor for nevi development [23]. Consistent conclusions about effects of environmental interventions on sunscreen use and incidence of nevi can therefore not be drawn.
Thirdly, with regard to the setting in which interventions were implemented, the school setting (meso level) was most often used. This preference for the school setting is in conformity with the increased attention for school health promotion in general [69]. Although the recognized Healthy Schools Approach is emerging internationally, attention for sun protection behavior is often lacking [70][71][72]. The promising results of comprehensive health promoting school programs emphasize the importance of sun safety in school-settings, with an added value of integrated environmental components [73]. However, other settings can also be of considerable importance with regard to enhancing sun safety. As ecological systems theory imply, the macro level is also of great importance when targeting children's health behavior [74]. The minority of interventions directed at recreational venues (macro level) is therefore noticeable. Particularly, since the amount of time people spent at recreational settings is increasing among Western populations [75,76]. The amount of UVR exposure at these venues is often high [21] while sun protection is regularly lacking during outdoor activities [77][78][79] and no prevention policies are at place. Moreover, children specifically are at high risk of receiving large amounts of UVR at playgrounds due to unavailability of shaded areas, as revealed by a recent study conducted in Germany [80]. These findings accentuate the importance of intervening in recreational settings. Nevertheless, more knowledge about specific locations where children receive the largest amounts of UVR is necessary before important settings for interventions can be defined, since insight in harmful sun exposure patterns is still lacking [81] and adequate measurement is challenging [82].
Although this review showed that economic types of interventions are most often implemented, it would also be advantageous to analyse the effect of other types of interventions as well. For example, policy intervention types were absent, while these strategies seem to gain positive effects on other health behaviors such as food intake and physical activity [83,84]. Examples of these types of interventions could be scheduling outdoor activities outside of UV peak hours [32], regulation of wearing hats and playing indoors or in the shade [85] or increasing the availability of shaded areas [86].

Strenghts and Limitations of the Included Studies
A few strengths of the included studies are worth mentioning. Almost all studies used methodologically strong designs, mostly (cluster) RCT designs, which strengthens the validation of evidence [87]. Furthermore, all studies used a relatively large sample size and drop-out rates were considered low. Most studies used multiple measurements which generated long-term data. Lastly, in the included studies, children of all age groups were represented. Also, a few limitations should be mentioned. Quality components such as selection bias, handling confounders, and handling withdrawals and drop-out were often rated as weak to moderate [47]. Furthermore, most weak ratings were given for controlling for confounders, which was often caused by an abstinence of information reported in the studies. Moreover, in some studies lack of details disabled the possibility to rate the study quality sufficiently, which may have resulted in inaccurate weak ratings. For example, data on statistical analyses performed were missing and contact with the author did not provide sufficient information [57]. These methodological shortcomings together may have affected the validity of the results. Furthermore, no effect sizes were reported which made comparison of statistical impact between studies difficult. Due to heterogeneity, it was not applicable nor eligible to calculate comparative effect sizes between studies [88]. Follow-up period was mixed in the studies reporting significant effects (ranging from three months to three years), thus making it difficult to assess the impact of various interventions. Lastly, even though it is regarded beneficial to include grey literature in a systematic review [89,90], the results extracted from two conference abstracts [53,59] should be interpreted with caution.

Strengths and Limitations of This Review
First, this review was conducted in accordance with common guidelines and with use of the PRISMA statement for performing and reporting systematic reviews [41,42], which includes relevant topics and concepts and enhances reporting of systematic reviews [91]. Furthermore, five databases were used in the literature search which is a considerable amount of sources [88], which has accounted for avoiding missing relevant titles [92]. Moreover, for assessing risk of bias of the included studies, the validated EPHPP tool was used [47,49], which has proven adequate inter-rater agreement [93]. The current systematic review has also a few limitations. First, the amount of eligible studies that specifically met our criteria turned out limited, which complicates comparison between studies. However, since sampling was done systematically in this review with high sensitivity by formulating specific search strings for five databases, the sample size of available studies seems adequate to draw conclusions [92]. Moreover, there is no golden standard stating a minimum of eligible studies to be included in the synthesis of systematic reviews. Lastly, years of publication of the included studies showed a shortage in recent studies conducted, which demonstrates a possible decrease in interest in environmental components in skin cancer prevention interventions.

Recommendations
Considering the results of this systematic review, several recommendations can be made. The large body of evidence for presence of health promoting sun safety interventions in general is positive, since health promoting adaptations in the environment show promising results on various other health behaviors [94][95][96][97], and the call for sun protection encouragement specifically is reported [29]. Moreover, within skin cancer prevention specifically, a recent study revealed that three decennia of dissemination of the multi-component SunSmart program led to significant improvements in various sun protection behaviors [98]. However, to gain knowledge on effectiveness of stand-alone effects of environmental interventions, it is recommended to investigate the effects of isolated components specifically in future studies. Albeit this review showed availability of a large amount of interventions that were initially eligible for inclusion, the absence of exclusively reported results of the environmental component(s) in most cases, restricted inclusion of those studies and therefore the ability to report extensively on the effectiveness of these components.
Second, sun protection behaviors of children themselves, in various settings, should be considered. Since children and adolescents were the target groups for this review, it is convenient that the responsibility of sun safety for children was expected to be among parents and teachers primarily. However, in a recent study [99], we found that children between approximately 11 and 14 years old, increasingly execute sun safety measures themselves, while parental protection towards their children declines. Hence, examination of children's own sun protection behaviors should certainly be included in future effect studies.
Third, the included studies in this review showed most studies were conducted in countries outside of Europe-inhabiting Caucasian populations, such as Oceania and Northern America, where high doses of ambient UVR exposure are present [6,100]. Since the incidence rates and melanoma risk have started risen earlier in these countries located at a lower latitude [101], it is understandable that skin cancer prevention strategies are already developed to a greater extent. Moreover, societal norms regarding UVR exposure might differ from those in European countries. Since latitudinal differences or seasonal variation can account for differences in need for sun protection strategies [102], more extensive research in which latitudinal differences are taken into account and, ideally performed in countries where skin cancer prevention is not yet normalized, is needed to translate research into practice. Specifically, since the need for skin cancer prevention due to rapidly growing incidence of melanoma in European countries is crucial [9,103].
Fourth, in all studies, subjective measures such as self-report questionnaires and/or observations were used. To further increase validity, application of objective measures of UVR exposure, such as handheld meters and time-stamped dosimetry [104] and wrist worn dosimetry devices [20], is recommended. Especially, the combination of self-reported and personal dosimeter measurements is promising to consider in future studies [105].
In conclusion, this review demonstrated overall positive results of environmental interventions in five of the seven included studies. Among those, shade provision was the most promising and consistent in increasing shade-seeking behavior. However, more research is necessary to investigate the perpetuation of these findings. As supplying shade provides intervention opportunities in various settings, in both schools and public areas, integrating shade provision in sun safety interventions for children is highly recommended. Moreover, future environmental interventions should focus more specifically on micro and macro levels of influence in children's social environments, such as the homeand recreational setting. Funding: No funding has been received for conducting this study nor for preparation of this manuscript.

Conflicts of Interest:
The authors declare that they have no conflicting interests.

Appendix A.2 Google Scholar
Infants OR children OR adolescents AND Protection OR prevention OR "prevention and control" OR intervention OR "health promotion" OR "environment design" OR "environment and public health" OR "built environment" OR shade OR "environmental exposure" OR "purpose built" OR "shade provision" OR "urban design" OR "physical environment" OR "urban architecture" AND Sunburn OR sunlight OR sunshine OR sun OR sunbathing OR "sun exposure" OR "UV radiation" OR UVR OR "overexposure" OR "skin cancer" OR melanoma OR "sun safe" AND Outdoor OR outside OR "swimming pool" OR "recreational area" OR recreation OR "recreational place" OR "open air" OR school OR playground OR "child day care center" OR park OR vacationing OR play OR plaything AND RCT OR "randomized controlled trial" OR "observational study" OR "cluster randomized trial" OR "experimental study" OR "quasi experiment" OR "experimental design" OR "group comparison" OR "control group"