Sun protection education for adolescents: a feasibility study of a wait-list controlled trial of an intervention involving a presentation, action planning, and SMS messages and using objective measurement of sun exposure

Sun exposure is important for health; it is involved in vitamin D synthesis and may induce feelings of well-being [1]. Sun overexposure is detrimental to health; 86% of melanoma cases in the UK are caused by overexposure to ultraviolet radiation (UVR) [2]. Overexposure leads to both DNA damage and immunosuppression, which mediate carcinogenesis [3]. In the last decade, melanoma incidence rates have increased by 50% in the UK and are projected to rise by 7% between 2014 and 2035, to 32 cases per 100,000 people by 2035 (https://​www.​cancerresearchuk​.​org/​health-professional/​cancer-statistics/​statistics-by-cancer-type/​melanoma-skin-cancer#heading-Zero). The estimated cost to the National Health Service (NHS) due to skin cancer will amount to over £180 million per annum in 2020 [4]. Evidence from meta-analyses show that melanoma risk is more closely linked with intermittent exposure to high-intensity sunlight than to chronic sunlight exposure [5]. Another meta-analysis found that an increased risk of melanoma was seen with increasing number of sunburns for all time-periods (childhood, adolescence, adulthood, and lifetime) [6]. Melanoma risk is increased regardless of whether sunburn occurs in childhood or adulthood [5‐7]. However, adolescence is a key period for increasing melanoma risk; there is a greater propensity for sunburn during adolescence than childhood [8‐11] or adulthood [12]. Several reports indicate that there is a steady decline in sun protection behaviours from childhood to adolescence [13, 14]. UK studies show that 51% of adolescents experience sunburn in the summer [15] and 44% of adolescents do not use sunscreen [16]. Health in adulthood has antecedents in childhood [17‐19]; for example, behaviours (e.g. sunbathing) and attitudes (e.g. pro-tanning) associated with skin cancer emerge in adolescence and track into adulthood [20, 21]. Adolescence therefore provides a critical window of opportunity for the primary prevention of skin cancer caused by sunburn across the life-course.

Theories inform what cognitions and emotions need to be addressed in behaviour change interventions. The Common-Sense Model of illness representation and self-regulation (CSM) [22] and Health Action Process Approach (HAPA) [23] informed the intervention being tested in this feasibility study. The CSM suggests that an ‘illness representation’ (e.g., skin cancer), has four dimensions: the cause dimension represents beliefs regarding the factors that are responsible for causing the illness (e.g., sunburn causes skin cancer), the consequence dimension refers to beliefs regarding the impact of an illness on overall quality of life (e.g., I will die if I get skin cancer), illness identity refers to beliefs about the illness label and knowledge about its symptoms (e.g., a mole that changes shape is a sign of skin cancer), the timeline dimension refers to beliefs about the time-scale or course of the illness (e.g., if I detect skin cancer early I will receive treatment and be cured) [24]. Expanded versions of the CSM include additional dimensions such as risk perception (e.g., people my age are not at risk of skin cancer) [25] and controllability (e.g., I can reduce my risk of skin cancer by using sunscreen) [26]. CSM is different to other health and risk behaviour models by explicitly recognising that people make simultaneous cognitive and emotional representations of an illness [24]. Thus, interventions using CSM ought to incorporate content that evoke an emotional response. According to HAPA, behavioural intentions are more likely to be translated into action when people generate specific plans [27]. Hence, sun safe interventions using HAPA will include planning to protect oneself from sun overexposure. Our previous study suggests that CSM and HAPA constructs are associated with sun protection behaviours [28] and other studies have also shown that illness representations [28, 29], risk perceptions [28, 30‐33], and action planning [28, 30] are associated with sun protection behaviours in adolescence. A systematic review examining the efficacy of appearance-based interventions concluded that they generally produce positive effects of sun protection behaviours [34], which suggests that appearance beliefs is another important dimension to address in sun protection interventions. However, it is not known which psychosocial constructs and risk behaviour models have most explanatory power for sun protection behaviours in adolescence.

Education interventions involve imparting knowledge and developing sun protection skills [35]. Some forms of delivery of an education intervention are likely to be more acceptable to the target audience than others and may influence intervention adherence and thereby effectiveness [36]. Smartphone technology is a form of delivery that offers opportunities to deliver sun protection information using text messaging (also called short messaging service (SMS)) because of high levels (83%) of ownership in the target age group [37]. Moreover, this form of delivering sun protection information may represent a practical and cost-effective approach relative to interventions that are delivered in-person. A systematic review of eight studies concluded that the use of SMS and similar electronic technology improves sun protection behaviours [38]. However, only one included study involved adolescents, which was a pilot study with no control group of 113 adolescents (11–14 years) who received 36 text messages [39]. The study reported significant increases in self-reported wearing of sunscreen, hats and sunglasses [39].

A general criticism of evidence reporting the effect of sun protection interventions is reliance on self-report. People may present a favourable image of themselves on questionnaires, which is called socially desirable responding [40]. As several systematic reviews have highlighted, a major limitation of previous sun protection intervention studies is lack of objective measurement of sun protection behaviour and clinically-related proximal targets such as, sunburn occurrence [38, 41].

The feasibility study of an educational sun protection intervention reported in this manuscript was designed to address limitations of our previous education intervention study [28]. The previous intervention was delivered in-person and sun protection behaviours were self-reported. Hence, the main purpose of this study was to assess the feasibility and acceptability of a sun protection education intervention with an additional intervention component – Short Messaging System (SMS). A further aim was to evaluate trial parameters such as, recruitment, use of a wait-list controlled trial design, and objective measurement of skin colour (erythema and melanin). The feasibility study was not powered to measure effect; rather, the purpose was to observe changes in potential outcomes of interest, and in particular, if outcomes changed in the intended direction. We do not report effectiveness data because it is generally recommended that feasibility and pilot studies descriptively evaluate a trial’s feasibility, acceptability and safety rather than test the effectiveness hypotheses of the planned main large-scale trial [2, 16‐18, 42‐45]. This is because the small amount of effect data available in feasibility and pilot studies means the degree of uncertainty is such that the chance of reaching inaccurate conclusions about intervention effect is high. Hence, robust and rigorous assessment of an intervention’s therapeutic implications must await adequately sized definitive pivotal trials [42]. Hence, recruitment and data completion rates are reported for intervention and control schools but outcomes (e.g., objective measure of skin colour) are only reported for the intervention school.

We conducted a feasibility study of a wait-list controlled trial, with five schools allocated by the research team to an intervention group and one school to a wait-list control group. The last school to be recruited was allocated to the control group. We did not use randomisation to reduce bias in this feasibility study because we were not aiming to measure effect; rather, the purpose was to estimate important parameters that are needed to design the main trial so that the future main trial is internally and externally valid. The study was conducted in the United Kingdom between June and September in 2018 to deliberately coincide with summertime when risk of overexposure to the sun is greatest. University of the Highlands and Islands Research and Ethical Committee approved the study (REF: OLETHSHE1004).

The criteria for inclusion in the study were males and females aged between 13 and 15 years. Head teachers of 132 state secondary schools (i.e. schools with students in our age category) in 6 local education authorities in different parts of Scotland were contacted by email about the study. Schools were followed up by telephone calls if the head teacher (also called head master, Principal etc. in other countries) expressed interest in participating in the study. Recruitment stopped once six schools consented to participate.

For each participating school, study information booklets were distributed to all parents/carers of all eligible students on the school roll, including a form that could be returned if the parent/carer did not assent to their child’s participation in the study. Only students who are opted out of the study by a parent/carer were excluded. Students who have insufficient English language were not excluded because the Education (Additional Support for Learning) (Scotland) Act 2004 entitles those students to additional support for learning. Contact details for the research team were included in the booklet, allowing parents/carers the opportunity to contact the research team if they wished to discuss the study. Students whose parents did not opt them out of the study were provided with a verbal overview of the study by a researcher, a study information booklet and consent form. For those students who were opted out of the study by parents/carers and those that did not consent to participate, alternative educational opportunities were provided by the school whilst their classmates participated in the study.

The intervention being tested was a refinement of our previous intervention [28]; two intervention components were the same (Components 1 and 2) and our previous study shows that these components improve sun safe intentions [28]; the third additional component was novel and had not been previously tested. The intervention was designed in accordance with two theoretical models - CSM and HAPA. The intervention was developed to address social cognitions by changing beliefs about skin cancer, evoke an emotional response to skin cancer, and shift sun protection intentions to actual behaviour by including action and coping planning.

Component 1: Information delivered during a presentation was designed to address key CSM dimensions (cause, consequence, identity, risk perception, controllability) and included information about personal experiences of skin cancer, incidence patterns, risk factors (e.g. when during the day risk of overexposure is greatest in the UK), associations between disease staging and survival, and benefits of SSE. The presentation also briefly touched on appearance e.g., concepts of beauty in relation to tanned and pale skin and the importance of some sun exposure for vitamin D. A skin cancer nurse specialist delivered the 50-min presentation on one occasion during the school day in a classroom or hall. After playing a 5-min film ‘Dear 16-year-old me’ (http://​dcmf.​ca), the nurse delivered the presentation with the aid of Microsoft PowerPoint slides. A Manual of the presentation that was developed by the research team was used as a guide for delivering the same presentation each time. A young adult skin cancer survivor (in this study the individual was male and hence, male pronoun is used) gave a brief 5-min talk after the nurse-delivered presentation. The talk was about his personal experience of melanoma diagnosis at 16 years old, impacts on his life and his views on sunscreen use and SSE. The film and the young person’s talk aimed to evoke an emotional response to skin cancer.

Component 3: Automated text messages were delivered on two days of the week for seven weeks after the 50-min presentation. All messages were sent during the summer holiday period relating to the schools in the study. This is because the summer holidays coincide with UK warm weather when there is the greatest risk of overexposure to the harmful effects of the sun. It is also when students may travel abroad to countries with hot climates. Messages were developed by the study investigators to apply to key theoretical CSM dimensions and address appearance. A total of 14 messages were developed. These messages were tailored to the target audience following feedback from a focus group of students (n = 13) who attended one of the participating schools. Participants were shown the messages that the study investigators developed and were asked to provide feedback, including how to make the text messages more likely to motivate themselves to protect their skin. Messages that the participants indicated that they did not like were removed from the list or were revised based on specific suggestions (Additional file 1). Messages contained information around sun safety behaviours and information about the effects of excessive exposure to the sun (Table 1). Messages were scheduled in the morning on a Monday and Friday.

Outcome variables were measured before the school summer holidays in June (baseline) and after the summer holidays in September (follow-up). This is because the summer holidays coincide with UK warm weather when there is the greatest risk of overexposure to the harmful effects of the sun. Student responses were paired between the two timepoints via use of unique identifier after all questionnaires had been anonymised. Objective measures of sunburn and tanning were collected by two researchers in the classroom. A self-completed pen and paper questionnaire was completed by students in the classroom. Items for the self-report questionnaire were recommended by an international working group to measure sunburn and sun protection behaviours [46] and/or used in our previous study [28].

As this was a feasibility study, the quantitative data were analysed using descriptive statistics. Baseline measurements were reported as n (%) for categorical data and mean (standard deviation) for continuous variables. The melanin and erythema indices were summarized as the arithmetic mean of the three readings taken on the forearm, and the arithmetic mean of the three readings taken behind the ear. Changes in the outcome measures were analysed within individuals (paired analysis) and reported in cross-tabulations (pre- and post- intervention) for categorical variables and as mean (standard deviation) of within-individual changes for continuous variables.

The control group was included in the study solely to determine the feasibility of recruitment of such a group. Thus, the more detailed results reported in this manuscript only include participants in the intervention group, to describe the outcomes pre- and post-intervention and the participants’ subjective views of the intervention impact. Only complete data (i.e., individually paired baseline and follow up sampled data) are included.

Audio-recorded qualitative data from focus groups were transcribed verbatim and analysed thematically using the Framework approach [49]. Qualitative findings provided contextual and explanatory understandings of adolescents’ experiences of the intervention.

Initially eight out of 130 schools in Scotland indicated an interest in participating in the study. Three schools had difficulty in facilitating the study within the timetable. There were four schools allocated to the intervention group and one school allocated to the control group. As previously pointed out, randomisation as a method for reducing bias was not used because we were not measuring effect in this feasibility study.

The school registers across all sites (intervention and control) indicate 724 students were eligible. No students declined to participate but some were either absent on the days the research was conducted or in a class that was not included in the study (a teacher could decide if their class was to be included in the study without giving any reason) and therefore 487 students consented to the study at baseline (67%), of which 385 (79%) were in the intervention group and 102 (21%) in the control group. Four parents/carers opted their child out of the study.

The characteristics of the study group are shown in Table 2. On average the intervention group were statistically significantly older than the control group (average age = 14.3 compared to 13.4) and had slightly more male participants (although this difference was not statistically significant). The distribution of ethnic groups was similar in the intervention and control groups, with the majority of the participants from the White ethnic group.

Objective skin measurements were available for 255 (66%) of the intervention group at baseline and 237 (61%) of the group at follow up. In one intervention school, the researchers ran out of time and could not collect measures from all students. Complete self-report data in schools were available for 247 (64%) adolescents in the intervention schools (i.e. we could pair baseline and follow-up for analysis of change in self-report outcome measures). In the control school, objective skin measurements were available for 101 (99%) of the control group at baseline and 79 (77%) at follow up. Complete self-report data in the control group was 57 (56%).

The study flowchart (Fig. 1) shows the number of students screened and assessed for eligibility, excluded, allocated to intervention or control group and the number of students in the intervention schools receiving the intervention and assessed at follow up. Given this is a feasibility study that is not designed to measure effect, only complete data for intervention schools is included in the results below.