A Randomized Controlled Trial of an Appearance-focused Intervention to Prevent Skin Cancer

Abstract

BACKGROUND

Skin cancer represents a significant health threat with over 1.3 million diagnoses, 8000 melanoma deaths, and more than $1 billion spent annually for skin cancer healthcare in the US. Despite findings from laboratory, case-control, and prospective studies that indicate a link between youthful indoor tanning (IT) and skin cancer, IT is increasing among US youth. Appearance-focused interventions represent a promising method to counteract these trends.

METHODS

A total of 430 female indoor tanners were randomized into intervention or no intervention control conditions. Intervention participants received an appearance-focused booklet based on decision-theoretical models of health behavior. Outcome variables included self-reports of IT behavior and intentions, as well as measures of cognitive mediating variables.

RESULTS

Normative increases in springtime IT rates were significantly lower (ie, over 35%) at 6-month follow-up in intervention versus control participants with similar reductions in future intentions. Mediation analyses revealed 6 cognitive variables (IT attitudes, fashion attitudes, perceived susceptibility to skin cancer and skin damage, subjective norms, and image norms) that significantly mediated change in IT behavior.

CONCLUSIONS

The appearance-focused intervention demonstrated strong effects on IT behavior and intentions in young indoor tanners. Appearance-focused approaches to skin cancer prevention need to present alternative behaviors as well as alter IT attitudes. Mediational results provide guides for strengthening future appearance-focused interventions directed at behaviors that increase risk of skin cancer.

With more than 1.3 million diagnoses, over 10,000 deaths, and more than $1 billion spent annually for skin cancer treatment in the US, skin cancer represent a serious public health threat.^1,2 Despite laboratory, case-control, and prospective studies all pointing toward a positive relation between youthful indoor tanning (IT) and melanoma and squamous cell carcinoma,^3-5 IT is gaining popularity, particularly among youth. The IT industry has an estimated 30 million individual US users.⁶ Recent reviews report approximately 10% of US adolescents under age 15 have used IT in the past year, with prevalence among older adolescent females estimated at 25% to 40%.⁷

Studies indicate that appearance factors are most predictive of IT use,^8-16 followed by social factors,¹⁷ perceived susceptibility to skin damage,^18,19 and dependence on ultraviolet (UV) light exposure.^20,21 Recent interventions have emphasized the appearance damage associated with UV tanning^18,22-27 (see Table 1 for a summary). Although these efforts demonstrate good efficacy, there are limitations that need to be empirically addressed before wide-scale diffusion is feasible. For example, Hillhouse and Turrisi²³ used a booklet emphasizing appearance damage in young women. Although this intervention is a relatively easy and low-cost method, the study used a small, limited sample. Furthermore, they did not examine mediatory processes underlying intervention efficacy.

TABLE 1.

Summary of Appearance-focused Skin Cancer Prevention Interventions

Publication	Intervention Type	Outcome	Time Interval for Follow-up Measure
Gibbons 200522	Skin damage UV photos	Decrease in tanning bed usage	3 wk
Hillhouse & Turrisi 200223	Workbook highlighting appearance-damaging effect of UV exposure	Decreased intentions to indoor tan and amount of indoor tanning	2 mo
Jackson & Aiken 200624	Multicomponent intervention with focus on image norms	Increase in sun protection behavior and intentions; decrease in intentions to sunbathe and sunbathing	2 wk
Jones & Leary 199425	Compared a health-focused to a appearance focused intervention	Appearance intervention showed reduced intentions to tan	Immediately after intervention
Mahler 200326	Appearance-focused information intervention and/or a UV photograph	Reduced sunbathing when received both	1 mo
Pagoto 200327	Multicomponent intervention including skin damage UV photos	Increased sun protection, no difference in sun exposure	2 mo

UV indicates ultraviolet.

Similarly, Gibbons et al²² reported promising results getting young people to alter UV-risk behavior through the use of UV photographs. However, valid UV photography currently requires significant equipment expense and interpretative expertise,²⁸ and can only be administered to 1 individual at a time, limiting its feasibility for large-scale campaigns.

Failure to focus on viable, healthy alternative behaviors is another limitation.^24,25 Literature from other health domains (eg, teen pregnancy, alcohol abuse, etc)^29,30 have demonstrated the need to simultaneously improve attitudes toward behavioral alternatives for long-term sustainability of behavior change. The dearth of information concerning healthy, appearance-enhancing alternative choices currently limits potential for long-term stability of effects.

Thus, although promising, there remain gaps in the literature examining sustainable interventions with a strong evidence base, which can be delivered to large audiences easily and cost effectively. The present research extends prior efforts to address these limitations. It evaluates an appearance-focused intervention designed to reduce IT tendencies using a randomized controlled trial. The study used a larger, more diverse sample. It examined mediational processes related to intervention effectiveness, and was guided by strong theoretical models that focus on providing healthy alternatives.

The intervention was a 24-page booklet (available upon request from the authors) based on decision-theoretical frameworks in general,^31,32 and the Jaccard behavioral alternative model³³ in particular. The Jaccard model focuses on the decision to use IT as well as healthy alternatives to this. For example, young people often use IT before important events like a prom. A typical young woman preparing for a prom is faced with the decision of whether to use IT. This decision is influenced, in part, by the way she construes IT, and in part by the alternatives she perceives to be available to her. An effective analysis involves consideration of IT, along with other viable alternatives, such as tanning using sunless tanning products, choosing a dress that does not require a complimentary tan, etc (see Fig. 1). Effective interventions reduce unhealthy behavioral motivations by 1) reducing positive attitudes toward the behavior, 2) decreasing positive subjective norms (ie, the perception that the behavior is approved by others), and 3) increasing perceived susceptibility to significant negative effects, while simultaneously increasing motivation for healthy alternative behaviors by improving attitudes and norms toward the healthy behavior. Although cessation of deliberate tanning is the ultimate goal, research indicates that harm reduction strategies can mitigate harm for frequent users.³⁴

FIGURE 1.

A theoretical model is shown of appearance intervention influence on indoor tanning outcomes. AF indicates appearance-focused.

Understanding mediational processes is critical for prevention campaigns. According to the model, IT behaviors are directly affected by IT attitudes and attitudes toward viable alternative behaviors such as fashion choices and sunless tanning. Attitudes toward alternatives are proximal mediators. Attitudinal constructs, in turn, are impacted by core variables identified in previous empirical work¹³: 1) perception that tans are attractive, 2) perception that IT is relaxing, 3) belief that IT increases skin damage susceptibility, 4) belief that IT increases skin cancer susceptibility, 5) IT subjective norms, 6) descriptive norms for peers (ie, perception of how much peers use IT), 7) image norms (ie, norms for an attractive body image) for IT relating to peers, and 8) image norms for IT relating to media stars. The present study evaluated a brief appearance-focused intervention based on a theoretical model with mediational analyses designed to assess whether observed program effects are a result of changes in targeted individual level variables.

MATERIALS AND METHODS

Sample

After institutional review board approval, 430 female tanners (median age, 18.6 years; standard deviation, 0.78 years; range, 17-21 years) signed informed consents, were randomized into intervention (n = 200) or no intervention control conditions (n = 230), and completed initial assessments. Intervention and control participants were treated identically, except that intervention participants received the booklet before the post-test.

Recruitment

A computer generated a random list of 1690 female university students from 2 universities. (Although it is possible that a very small number of participants could have been enrolled in an author's class during the study, the finding that recruitment lists were drawn from the general university female population [N > 23,000], along with the composition of the authors’ classes [ie, graduate and senior level] make this highly unlikely.) The students were sent an invitation email with a screening survey assessing past IT behavior and future intentions. Email receipt was confirmed for 853 individuals. Inclusion criteria were past year IT or a score of 5 or above on a 7-point scale that measured intentions to tan in the next year. Eligibility requirements were met in 454 of the 853 (53.0%), and 430 agreed and completed baseline measures (95% participation rate). Participants were offered monetary compensation for completing assessments and evaluating the intervention ($20 for each assessment). A small number who completed baseline measures dropped out (total n = 18; 5 intervention and 13 control). Group assignment was individually randomized across the conditions. The final groups consisted of 195 intervention participants and 217 controls. We found no differences between intervention and control group participants on background characteristics (see Table 2).

TABLE 2.

Demographic Variables by Group (N = 412 Female Indoor Tanners)

Background Characteristics	Control n=217	Intervention n=195
Fitzpatrick skin type, χ2 (chi-square [4, N=412]=4.32; P=.37)
I	7.8	7.0
II	19.8	27.6
III	41.5	40.0
IV	27.6	23.8
V	3.2	1.6
Family socioeconomic status, χ2 (chi-square [4, N=412]=3.58; P=.47)
Much lower than most families	1.4	0.0
Moderately lower	9.7	11.9
About average	54.4	56.8
Moderately higher	30.9	28.6
Much higher than most families	3.7	2.7
Year in school, χ2 (chi-square [3, N = 412]=4.91; P=.18)
Freshman	58.1	63.2
Sophomore	31.3	31.9
Junior	10.1	4.9
Senior	0.5	0.0
Mean age, y (SD) (t[400]=1.59; P=.11)	18.69 (.82)	18.56 (.74)
Mean age began indoor tanning (SD) (t[344]=–1.00; P=.32)	15.7 (1.58)	15.8 (1.43)

SD indicates standard deviation.

Materials and Procedures

Participants completed the baseline assessment in the fall. Intervention participants were then given the intervention, and asked to provide feedback 1 month later. All participants were reassessed at 1-month follow-up for mediational analyses. All post-tests took place in April at 6-month follow-up to intervention distribution. This post-test time was selected to assess IT during the period of heaviest use³⁵ (see Fig. 2 for a study design flow chart). Participants also completed biweekly diaries of IT behavior as a validity check of the global self-reports.

FIGURE 2.

The study design is represented in a flow chart.

Contents of intervention

The prototype booklet was written at an eighth grade reading level, based on empirical results from the theoretical model.^13,14 This prototype was modified using focus group feedback from current indoor tanners (n = 30) and administered to another 147 indoor tanners.²³ Follow-up interviews (n = 25) with participants led to further modifications, resulting in the final booklet containing 3 appearance-enhancement alternatives: exercise, clothing, and sunless tanning. The booklet used in this study was produced by a professional commercial art firm using the content from the process described above. A brief summary of each section is provided below. Section 1 introduced the history of tanning, providing a context for current tanning norms. Section 2 provided an analysis of current tanning norms, as well as examining media and peer image norms. Section 3 discussed the effects of UV radiation on skin with the intent of increasing perceived susceptibility to skin damage from UV exposure. Section 4 discussed the specific effects of IT, again with the intent to increase perceived susceptibility. Section 5 provided IT guidelines emphasizing tanning abstinence as well as harm reduction recommendations. The last section provided information on healthier, appearance-enhancing alternatives to tanning, including exercise, choosing fashion that does not require a complimentary tan, and sunless tanning products, along with links to interesting and informative websites on these alternatives. Both the booklet content and the links related to sunless tanning emphasized that sunless tanning does not protect against UV exposure.

Intervention integrity

Participants were told the research was designed to: 1) provide educational material on IT, and 2) obtain feedback on the readability and interest value of the intervention. They were asked to write short statements summarizing each section and to complete rating scales evaluating specific topics. These efforts provide a formal evaluation of whether the materials were attended to and comprehended. A random subset of intervention participants returned the booklets with written summaries. Analysis of these summaries clearly reflected knowledge and comprehension of section content. Comparisons of this subset with other participants revealed no differences in efficacy or mediational outcomes. Thus, there is good evidence that intervention participants read and attended to the materials.

Evaluation of booklet content

One month after distribution, intervention participants evaluated the booklet for interest, readability, comprehension, and usefulness on 10-point scales (not at all to extremely). The means and standard deviations for the booklet evaluations, which are highly positive, are presented in Table 3. These data are consistent with the suggestion that the participants did process the contents of the booklet and were positive in their evaluative assessments.

TABLE 3.

Means and Standard Deviations (SD) for Evaluation of the Intervention Booklet

	Interesting Mean (SD)	Readable Mean (SD)	Understandable Mean (SD)	Useful Mean (SD)
Introduction	6.74 (1.95)	8.15 (1.69)	8.74 (1.48)	6.95 (2.07)
Current fashion trends	7.55 (1.85)	8.21 (1.62)	8.70 (1.41)	7.57 (1.89)
Effects of UVR on skin	7.71 (2.01)	8.29 (1.74)	8.59 (1.61)	8.30 (1.71)
Effects of indoor tanning on skin	7.28 (2.01)	8.25 (1.57)	8.74 (1.49)	7.77 (1.90)
Indoor tanning guidelines	7.62 (2.03)	8.47 (1.55)	8.88 (1.44)	7.74 (2.12)
Healthy alternatives	7.72 (1.80)	8.51 (1.52)	8.86 (1.31)	8.21 (1.78)

The scale for all of the evaluation items are anchored by 0 (not at all) and 10 (extremely) with 3 (slightly) and 7 (moderately) as reference points.

UVR indicates ultraviolet radiation.

Measures

Indoor tanning behavior

IT frequency

At baseline, IT frequency was measured for the past 3, 6, and 12 months. At 6-month follow-up in April, the 3-month frequency was reassessed to measure peak months of IT use. (Although it would have been optimal to have identical assessments during the 3 months of spring before the intervention, this would have required following participants for 1 year to have the same measurement metric [ie, a global assessment of spring tanning taken immediately at the end of spring], which was beyond the project scope.) These measures demonstrate strong correlations with diary measures of behavior over the same time frames (r = .77-.86; P < .001) in previous work.³⁶ In the current study, biweekly diaries provided a validity check. Aggregates of the diary measures were strongly related to the global assessment of IT frequency (r = .87; P < .001).

Indoor tanning intentions

For IT intentions, participants were asked how strongly they intended to engage in 2 IT-related future behaviors on 7-point Likert-type scales. The measure exhibited excellent internal reliability (Cronbach α = .89) as well as a significant relation with future IT behavior (r = .37-.42; P < .001).

Proximal mediators

Attitudes toward IT

To assess IT attitudes, participants indicated agreement or disagreement with 5 statements describing IT advantages or disadvantages on 5-point Likert-type scales. This IT attitude scale has a long history of use in the literature,^23,37 where it is consistently reliable (r > .90) and strongly related to IT intentions.

Attitudes toward alternatives

Attitudes toward IT alternatives were assessed with scales that demonstrate excellent internal reliabilities (Cronbach α = .77-.90), as well as test-retest reliabilities (4 weeks: r = .60-.63). The sunless tanning attitudes scale consisted of 2 items, and the fashion attitudes scale also had 2 items measured on 5-point, Likert-type scales. Because our earlier studies have not found significant effects for exercise attitudes on IT outcome variables, they were not assessed in this study.

Distal mediators

Indoor tanning beliefs (eg, belief that tanning is attractive, perception that IT is a good way to relax and relieve stress)

Respondents indicated the degree to which they believe tanning improves attractiveness using 3 items (Cronbach α = .73). The perception that IT is a good way to relax and relieve stress was assessed with 3 items (Cronbach a 5 .85).

Perceived susceptibility

The conditional perceived susceptibility to damaging physical appearance was assessed using 1 item, whereas skin cancer susceptibility was assessed using 2 items (Cronbach α = .84).

Norms

We used methods suggested by Ajzen and Madden³⁸ to assess subjective norms. Respondents indicated whether friends and typical and popular college students think they should (or should not) “indoor tan.” Respondents indicated how motivated they are to comply with referent's IT opinions. Subjective norm scores are calculated by multiplying normative perceptions and motivations to comply and summing across the referents. This scale evidenced good internal consistency (Cronbach α = .87). In addition, we assessed descriptive norms of friends and other college students with 2 items (Cronbach α = .65), perceived image norms for peers with 1 item, and perceived image norms of media stars with 2 items (Cronbach α = .88).

RESULTS

Statistical Analyses

We conducted 2 (condition) × 2 (time) mixed measures analyses of variance to examine the effects of intervention on IT behavioral outcomes. The joint significance test was used to assess mediation.³⁹ Missing data (<3%) were imputed using the EM algorithm in SPSS 15.0 (SPSS Inc., Chicago, Ill). The analyses were intent-to-treat, including all randomized cases, whether they read the intervention or not.

Assessment of the effects of the intervention

Efficacy of the intervention on IT intentions and behavior

The baseline 3-month assessment covered August through October, a period with some of the lowest expected IT use.⁴⁰ In contrast, the 3-month assessment at long-term follow-up covered February through April, which has the highest rates. Consequently, we expected overall IT frequency to increase across these assessments. At baseline, the groups were equivalent on all estimates of past IT frequency (3-, 6-, and 12-month Student t < 0.5; P > .60) and averaged over 26 IT sessions in the previous year. As expected, there was a significant effect for time (F_1,400 5 49.51; P < .001). Of more interest, however, are the significant interaction effects suggesting that the intervention differentially reduced normative increases in IT during the heaviest IT months by over 35% (see Table 4). Similar changes were noted for future IT intentions. The results suggest the intervention was effective at reducing intentions to indoor tan at the long-term follow-up.

TABLE 4.

Comparisons of Means (Standard Errors) Between Intervention Groups Across Indoor Tanning Tendencies

	Baseline*		Long-term Follow-up†		Significance F(df=1400)
	Control (SE)	Intervention (SE)	Control (SE)	Intervention (SE)
Past 3-mo IT sessions	4.48 (0.55)	4.67 (0.60)	10.90 (0.93)	6.80 (0.93)	12.42‡
Intention to use indoor tanning	10.01 (0.26)	9.55 (0.29)	10.51 (0.28)	8.65 (0.30)	15.64‡

SE indicates standard error; IT, indoor tanning.

^*Baseline represents indoor tanning for August through October.

^†Long-term follow-up represents indoor tanning for February through April.

^‡P < .001, 2-tailed.

Examination of Mediating Processes

Mediation analyses results are presented in Table 5. All mediators were measured at 1-month follow-up. We tested for the significance of the mediated effect using the methods of MacKinnon et al.³⁹ First, a regression analysis tests for significant effects of intervention condition on the hypothesized mediator measured at 1-month follow-up (the α path). A second regression tests for the effect of the mediator on the outcome at long-term follow-up while controlling for intervention condition (the β path). If both paths show statistical significance, there is evidence for a significant mediating relation (eg, the intervention affects the outcome variable through changes in the mediator variables). The product of the α and β paths is the mediated effect (αβ) and provides an estimate of the strength of the mediation.

TABLE 5.

Baseline and 1-Month Follow-up Means Between Intervention Groups (Standard Deviations), Intervention Effects on Mediators, Mediator Effects on Indoor Tanning, and Mediated Effects

Mediator	Baseline Control Means	Baseline Intervention Means	Follow-up Control Means	Follow-up Intervention Means	α Intervention Effect on Mediator	β Mediator Effect on Outcome	αβ Mediated Effect
Tanning alternatives
Attitude toward indoor tanning	16.3 (.33)	15.9 (.35)	17.2 (.34)	14.0 (.36)	–3.18*	0.47*	–1.49†
Attitude toward sunless tanning	6.4 (.16)	6.0 (.17)	6.4 (.15)	6.9 (.17)	0.46‡	–0.24	–0.11
Attitude toward fashion	8.0 (.10)	8.2 (.11)	8.1 (.10)	8.6 (.11)	0.48*	0.88‡	0.42‡
Cognitive mediators
Tan is attractive	12.6 (.14)	13.0 (.15)	12.8 (.15)	12.3 (.16)	–0.47‡	1.24*	–0.58‡
Indoor tanning is relaxing	10.9 (.21)	11.0 (.22)	11.2 (.21)	10.3 (.23)	–0.88†	0.80*	–0.71†
Perceived susceptibility to skin damage	4.4 (.06)	4.4 (.06)	4.3 (.06)	4.5 (.06)	0.14§	–1.22§	–0.18§
Perceived susceptibility to skin cancer	8.0 (.12)	8.1 (.13)	8.0 (.12)	8.3 (.13)	0.27	–0.96†	–0.26§
Subjective norms	11.0 (.96)	9.7 (1.04)	11.2 (1.05)	5.7 (1.13)	–5.51*	0.17*	–0.92*
Descriptive norms, peers	7.5 (.13)	7.3 (.14)	7.5 (.12)	7.4 (.13)	–0.13	1.44*	–0.19
Image norms, peers	4.2 (.06)	4.1 (.06)	4.3 (.05)	4.2 (.05)	–0.11	1.96‡	–0.22§
Image norms, media	6.8 (.13)	7.3 (.13)	7.0 (.14)	6.1 (.14)	–1.20*	0.09	–0.11

^*P < .001.

^†P < .01.

^‡P < .05.

^§P < .10,2-tailed.

Examination of the α paths reveals that the intervention significantly reduced favorable attitudes toward IT, the perception that tanning is attractive, the perception that IT is relaxing, subjective normative beliefs, and image normative beliefs of fashion and media stars. The intervention significantly increased favorable attitudes toward sunless tanning and toward using clothing as an alternative for appearance enhancement, and marginally influenced perceived susceptibility to skin damage.

Examination of the β path reveals that attitudes toward IT, attitudes toward using fashion, the perception that tanning is attractive, the perception that tanning is relaxing, subjective normative beliefs, descriptive norms of friends and other college students, and image norms of peers led to significant increases in the 6-month follow-up IT measure. Measures of perceived susceptibility to damaged physical appearance and perceived susceptibility to skin cancer were significantly related to fewer IT sessions at long-term follow-up.

Mediated Effects

Statistically significant mediated effects were found for attitudes toward IT, attitudes toward using fashion, perceptions that tanning enhances attractiveness, perceptions that IT is relaxing, and subjective normative beliefs. The mediated effect for perceived susceptibility to skin damage, perceived susceptibility to skin cancer, and the image norms related to peers were marginally significant. Thus, the observed intervention effects can be explained, in part, by changes in these core cognitive and behavioral variables. We also tested a mediation chain model with the variables that were found to be significant based upon the mediation analyses presented here. The results of this model reveal that all paths were significant (P < .05), lending support to the mediation chain of the theoretical model. Further evidence for the theoretical plausibility of the mediation chain model is that a model that includes proximal mediators predictive at the same level as distal mediators has a significantly worse fit than the mediation chain model.

DISCUSSION

This study examined the efficacy of an appearance-focused intervention grounded in the Behavioral Alternative Model to reduce IT use in young adult females. Building on earlier work,²³ the intervention demonstrated strong effects on IT behavioral frequency and intentions (ie, reducing normative increases in springtime frequency by over 35% vs controls). These results support a growing literature that points to the success of appearance-focused approaches to skin cancer prevention when grounded in solid psychosocial health theories.

Of interest is that a simple message delivery method, a booklet, was able to achieve clinically significant reductions in UV exposure behavior by using a theoretically based, empirically tested message. Although communication methods are important for intervention appeal, dissemination costs, and integrity, this study demonstrates that messages guided by strong sociopsychological theories can have good effects without technologically innovative delivery systems. This result replicates previous work that has found comparable changes in tanning behavior and attitudes using a similar booklet,²³ as well as a more recent study that achieved change through the administration of a simple handout providing tanners normative feedback on their behavior.⁴⁰ The current study did impose conditions designed to ensure reading and processing the material presented in the intervention (ie, writing summaries and providing evaluations) that would not apply in real world settings. Therefore, further research will be needed to ascertain the optimal message delivery method (eg, booklet, Internet, cell phones, etc), by empirically evaluating the optimal match of delivery system to subgroup and individual preferences.

The analyses further revealed that the intervention caused significant reductions in favorable attitudes toward IT, and accompanying improvements in attitudes toward using sunless tanning and fashion to enhance appearance in intervention participants. Although IT attitude changes appear relatively modest (1.9 of 25 total points), the mean scale scores did move from generally favorable (above the midpoint) to generally unfavorable (below the midpoint). Furthermore, these seemingly small changes fall into the range of change (1 standard deviation) that our theoretical model predicts can lead to relatively large changes in IT behavior odds ratios (OR, 7.6+) when combined with similar changes in other beliefs and attitudes. The intervention also decreased the perception that tans are attractive, the perception that IT is relaxing and stress relieving, and the belief that media stars indoor tan. However, we saw no changes in perceived susceptibility to skin damage or to skin cancer from IT, likely because of ceiling effects with these measures.

The mediational analyses revealed significant mediated effects for IT attitudes, fashion attitudes, perceptions tans are attractive, perceptions IT is relaxing, perceived susceptibility to skin damage and to skin cancer, subjective norms, and image norms of peers. Of interest is the finding that there was no mediational effect for sunless tanning attitudes despite the finding that there was a significant effect of the intervention on sunless tanning attitudes, as reflected in the a path in the mediational analysis. Despite the lack of a mediational effect in this study, we believe that sunless tanning represents a tanning alternative that should be further studied.

Some authors have expressed concern about recommending sunless tanning as a UV tanning alternative because of findings that sunless tanning users are more likely to report sunburn and IT use in the past year.^41,42 Furthermore, some health professionals think that promoting sunless tanning sends the wrong message. This concern is contrary to empirical results from studies on many health issues (eg, alcohol, tobacco, and illicit drug use, safe sex behaviors, etc) that find no evidence for increased negative health behaviors in individuals exposed to harm-reduction messages similar to sunless tanning promotion.^34,43 Furthermore, such a shift toward attitudes in favor of tanning, in our study, should be reflected in 1) increased positive attitudes toward tanning, 2) increased subjective tanning norms, and 3) a significant relation between sunless tanning attitudes and subjective norms among the intervention participants exposed to the sunless tanning information in this intervention. However, tanning attitudes and subjective norms decreased significantly after exposure to the intervention, and the relation between sunless tanning attitudes and subjective norms was small and nonsignificant at follow-up (r = .05; P = .51).

It is clear from the mediated effect sizes that IT attitudes had the strongest effect, indicating its importance as an intervention target. The perception that IT is a good way to relax and relieve stress had the next strongest effect size. This latter result supports the work of Danoff-Burg and Mosher,¹¹ who believe that future work in this area needs to broaden its focus to include the socialization and relaxation aspects of the tanning experience. The intervention also changed perceptions that a tan is attractive. This has proven difficult to accomplish in past research. This study demonstrates that interventions based on sound theory and empirical data can impact even relatively entrenched beliefs and perceptions.

Subjective norms, that is, the perception that peers and friends approve of IT, were significantly changed by the intervention and proved a relatively strong mediator of intervention effectiveness. Jackson and Aiken²⁴ predicted that changing descriptive peer norms for tanning would be difficult, because a tan is easily observable. This prediction was confirmed in this study. However, our intervention demonstrated that you can alter the perception about what peers think about tanning and IT. Furthermore, the changes in subjective norms proved important for changing long-term IT behavior.

Limitations of this study include the finding that we only evaluated the effects of the intervention at 6 months after intervention. Future work needs to explore long-term effects of this type of intervention to determine whether these results can be sustained. Because outdoor tanning habits were not assessed, it is not possible to know whether some participants switched to outdoor tanning because of perceptions that it is safer than tanning indoors. However, given that the intervention discouraged all UV exposure, indoors and out, we believe that this would be relatively unlikely. Future studies should consider this possibility and assess all forms of intentional UV exposure.

Despite the limitations, this study further supports the growing literature affirming the efficacy of appearance-focused interventions. The theoretical model guiding the intervention accounts for viable alternative behaviors to UV tanning for improving appearance, thus increasing the likelihood effects can be sustained. Furthermore, the mediational analyses affirm the usefulness of this model, and point to variables (eg, relaxation/socialization aspects of tanning) that might be fruitful focuses of future interventions. This randomized controlled trial supports the use of theoretically and empirically based intervention messages to change young persons’ UV risky behaviors and ultimately reduce skin cancer morbidity and mortality.