Almost 5 million Americans are treated for skin cancer annually, and its incidence is rising [1-5]. Additionally, keratinocyte carcinomas can be a chronic disease, requiring ongoing costly treatments and resulting in compromised quality of life similar to some other cancers [6]. Although largely preventable, skin cancers, particularly melanomas, can be deadly, debilitating, damaging to tissues and organs, and disfiguring. Risk factors for skin cancers include personal or family history of melanoma or keratinocyte carcinomas, certain phenotypic (eg, fair skin) and other physical characteristics (eg, numerous moles) [7-19], as well as excessive UV radiation (UVR) exposure from the sun or tanning devices [20-25]. Most skin cancers are preventable with sun protection such as minimizing UVR exposure and wearing protective clothing and sunscreen [26,27]. Early detection via skin examination by an individual, partner, or health care provider may also help improve treatment outcomes and reduce morbidity and mortality [28].

Young adulthood, defined here as ages 18 to 25 years, is an important window for skin cancer risk reduction interventions. Melanoma is one of the most common cancers in people younger than 30 years [29]. Adolescents in the United States have the lowest sun protection rates of all age groups [30] and engage in increased UVR exposure as they move into adulthood [31]. This time period represents a critical period in forming sun protection habits because only 25% of lifetime UVR is accumulated by age 18 years [32]. Skin cancer risk behaviors, including lack of sun protection, sunburns, and indoor tanning, peak around age 25 years [33,34]. Unfortunately, young adults tend to be resistant to preventive health recommendations because, as a group, they perceive themselves as having more immediate priorities than disease prevention and believe that the consequences of their current health behaviors are in the distant future [35-37]. In addition, young adults are more likely to engage in risky behaviors and can be more influenced by their peers than older adults [35-37]. Prior studies of behavioral interventions to increase sun protection or decrease UVR exposure among healthy young adults in the United States [38-45] have been limited in their reach or duration of assessment.

Social media has been almost universally adopted by young adults. In 2021, almost all 18-29 year olds used the internet [46], approximately 96% owned a smartphone [47], and approximately 70% used Facebook and Instagram [48]. Social media recruitment can be efficient [49-52], and digital behavioral interventions can be disseminated widely and be cost-effective to maintain [53,54]. Digital interventions designed to improve health behaviors (eg, exercise and weight loss) in various populations have been found to produce medium effect sizes and consistently outperform similar nondigital interventions [54], but longer-term outcomes have been minimally evaluated for skin cancer prevention digital interventions [55].

Given the potential of digital behavioral health approaches, we developed an individually tailored, interactive, multimedia, and theoretically grounded (Integrative Model of Behavioral Prediction [IM]) web-based intervention targeted to young adults (basic) [56,57]. We evaluated the intervention in a large sample of 18-25 year olds (n=964) at moderate to high risk of developing skin cancer, recruited from a consumer research panel hosted by a market research company, in a randomized controlled efficacy trial. Behavioral risk factors for melanoma were significantly improved. Intervention effect sizes (Cohen d) were 0.53 for sun protection and 0.43 for UVR exposure behaviors at the 3-month follow-up compared with an assessment-only condition [58]. Sunburns were also significantly reduced. This intervention is the basic intervention for the study described in this paper.

Although the results of the efficacy trial were promising, we proposed that the intervention and outcomes could be further improved. Our goal was to improve engagement with and impact of the Basic Intervention on sun protection, UVR exposure, and engagement in skin cancer examination. To improve effects of the basic intervention, we added several key interactive features and strategies suggested by prior participants, our data, and supported by the literature (ie, by creating a mobile version, adding incentives embedded in the intervention, an enhanced goal-oriented feature, ongoing email and SMS text messages related to previously identified mediators of behavior change, and ongoing news and video updates).

To assess the potential for future dissemination, we conducted a hybrid type II effectiveness-implementation trial [59], which has the goal of increasing efficiency of moving novel evidence-based interventions to the community through mass access and reach. The purpose of this study was to implement the enhanced intervention (enhanced) with young adults aged 18-25 years at moderate to high risk of developing skin cancer and evaluate the intervention’s effectiveness in a sample recruited via the web through national dissemination to a general population. We hypothesized that effectiveness and longer-term maintenance effects on UVR and sun protection outcomes would be best for enhanced, second for the original basic condition, and worst for an electronic pamphlet (e-pamphlet) control group. To our knowledge, these are the first entirely digital behavioral skin cancer prevention interventions targeted to healthy young adults recruited from the community that have been evaluated in a national longitudinal randomized controlled trial (RCT) in the United States. If found to be effective, these interventions could be feasibly scaled up for widespread delivery, with the potential to reduce skin cancer risk in a large population of US young adults.

This study assessed 2 digital skin cancer risk reduction programs compared with a control condition for young adults in the United States between the ages of 18 and 25 years (ClinicalTrials.gov NCT03313492). Recruitment efforts are described in detail elsewhere [60]. Briefly, recruitment of study participants was conducted through paid advertising on Facebook, Instagram, and Twitter and unpaid social media posts between September 2018 and April 2019. Audience demographic and behavioral characteristics including age range, location (United States), and interests such as outdoor activities and physical fitness were identified to efficiently target the intended audience. Advertisement content focused on potential heath, appearance, and financial benefits of participation (eg, “Healthy skin is beautiful skin”). The objective of the advertisements was to encourage potential participants to click the call-to-action buttons such as “Sign Up” or “Learn More” that directed them to a study-specific landing web page or sign-up web page. The landing page included relevant images and brief information about the study and reasons why individuals might want to participate, including brief testimonials from prior participants. Individuals were then instructed to create an account on the sign-up web page with a phone number, email, and password. Once an individual indicated their interest and created a study account, they were automatically directed to complete a brief eligibility screener. If eligible, individuals were invited to complete the online informed consent form. After consent, participants were directed to the 10-minute online baseline survey. If enrollees completed the baseline survey, they were automatically directed to their intervention condition and received a US $5 electronic gift card. A computerized program was created to randomize participants on a 2:2:1 basis in blocks of 10 to either basic, enhanced (described below), or a noninteractive skin cancer prevention educational web page modeled after an e-pamphlet from the American Cancer Society [61]. Participants were invited by email, SMS text message, and telephone if necessary to complete online 10- to 15-minute follow-up surveys at 4, 12, 24, and 52 weeks.

This research adhered to appropriate ethical review and approvals as per institutional guidelines (Institutional Review Board #Pro2018001543). Participants provided informed consent electronically and were informed of their ability to opt out. Identifiable data were stored in password-protected files accessible only to approved research personnel. Data were deidentified for analysis. Participants were informed that the total study incentives would be up to US $120 for completing all 5 online surveys over the course of 12 months, plus periodic gift card raffles.

A free online American Cancer Society pamphlet on skin cancer prevention and early detection [61] was used as the control condition and was distributed on a noninteractive web page.

Online eligibility surveys were completed by 4518 individuals (Figure 1). Of these, 2473 (54.7%) were deemed eligible. The main reason for being ineligible was a BRAT skin cancer risk score denoting lower risk (1056/4518, 23.4%, of individuals screened not meeting criteria), followed by age other than 18-25 years (767/4518, 17%, not meeting criteria) or self-reporting “always” engaging in sun protection behaviors outside (279/4518, 6.2%, not meeting criteria). Of these, 80.5% (1990/2473) submitted online baseline surveys. The study team dropped 13.7% (621/4518) of potentially eligible participants for attempting to enroll more than once or for poor data quality. This figure is lower than our prior study (22%) [58] but higher than the goal of 5% being careless responders. DQS ranged from 0 to 17 (worse quality), with a median (IQR) score of 3.00 (1.00-5.00) and a mean (SD) score of 3.63 (3.01). Of the 1369 who were randomized to 1 of the 3 intervention conditions, the proportion of completed surveys at follow-up was 1158 (84.6%) at 4 weeks, 1182 (86.3%) at 12 weeks, 1177 (86.0%) at 26 weeks, and 1144 (83.6%) at 52 weeks. These completion rates were similar across the 3 intervention conditions.

At baseline, participants’ average age was 22.3 (SD 2.31) years (Table 2). Approximately 69% (949/1369) of the sample was female, 83% (1141/1369) White non-Hispanic, 73% (999/1369) heterosexual, 34.3% (470/1369) had at least a college education, 47% (649/1369) earned ≤US $15,000 in the last year, 75% (1021/1369) lived in a state with a northern climate (consistent with classifications of states by climate), and over 87% (1185/1369) lived with someone part or all of the last year. There were no significant differences in these demographic variables across the intervention conditions (P values ≥.13), suggesting successful randomization.

Mean scores at baseline on primary outcomes were low, with an average score of 1.72 (SD 0.49) for overall sun protection, an average score of 1.48 (SD 0.84) for habitual sun protection on scales from 0 to 4, and an average of 0.72 (SD 1.35) for overall UVR exposure on a scale from 0 to 30 in the last month (Table 3). Baseline values on secondary outcomes were also low, with 11.2% (n=154) reporting at least 1 skin self-exam in the last 3 months, 10.4% (n=143) reporting a physician skin exam in the last 12 months, 31.6% (n=416) reporting wearing sunscreen on their body at least once in the last month, 6.9% (n=94) reporting indoor tanning at least once in the last month, yet 23.4% (n=321) reporting at least 1 sunburn in the last month (Table 4).

Figure 2 shows the error bar plots of the changes from Baseline of the primary outcomes. As the interactions of intervention and time were not significant for any primary outcome (P≥.23), they were not included in the final models. Overall sun protection in the e-pamphlet group was statistically significantly lower than that in the basic or the enhanced group (P=.007 and <.001 with Cohen d of 0.209 and 0.376, respectively). In addition, sun protection in the enhanced group was statistically significantly higher than that in the basic group (P=.01, d=0.149). Habitual sun protection in the e-pamphlet group was statistically significantly lower than that in the basic or the enhanced group (P=.02 and .004 with d of 0.186 and 0.224, respectively); there was no difference between the basic and the enhanced groups (P=.46). All groups reported increased UVR exposure during the summer, and UVR exposure in the e-pamphlet group was greater than that in the basic or the enhanced group, but these differences did not reach statistical significance (P=.07 and .09, respectively) In addition, there was no significant difference in UVR between the basic and enhanced groups (P=.89).

Figure 3 shows the error bar plots of the adjusted proportions for secondary outcomes. For the adjusted proportions of skin self-exam in the past 3 months, the basic group had a significantly higher proportion of individuals who reported doing at least 1 skin self-exam than those in the e-pamphlet group at week 52 (P=.04). The proportion for the enhanced group was also significantly higher than that for the e-pamphlet group at week 52 (P=.004). For the adjusted proportion of individuals who use sunscreen on their body, which increased for all groups during the summer, the proportions for both the basic and enhanced groups were significantly higher than that for the e-pamphlet group at week 12 (P=.03 and .01, respectively). For the adjusted percent of indoor tanning in the past month, the basic group had a significantly higher proportion of individuals who reported indoor tanning than the e-pamphlet group at week 52 (P=.048). All other group comparisons from week 4 to week 52 were not statistically significant.