Summary of Findings

In 2003, the USPSTF concluded that there was insufficient evidence to recommend for or against routine counseling by primary care clinicians to prevent skin cancer. At the time of this recommendation, only a single trial that evaluated primary care skin cancer counseling was available, and this was part of a larger community-based intervention. Therefore, the contribution of office-based counseling could not be isolated. In addition, there was uncertainty about potential harms of counseling to encourage sun-protective behaviors, and limited evidence examining the effect of sunscreen or indoor tanning use on skin cancer risk. We summarize our findings according to the evidence gaps identified by the 2003 USPSTF recommendation (Table 7).

Table 7. Summary of Evidence By Key Question.

Table 7

Summary of Evidence By Key Question.

Effectiveness of Counseling to Promote Sun-Protective Behaviors

Although we did not find any studies examining whether counseling interventions could reduce skin cancer or intermediate outcomes (e.g., sunburns, nevi, or actinic keratoses), we found 10 fair-or good-quality RCTs that examined the impact of primary care relevant skin cancer counseling interventions on sun-protective behaviors. In two of these trials, however, the counseling delivered through primary care was a small part of a much larger coordinated multimodal, community-based intervention, and thus will not be discussed further.

In adults (n=6,225), primary care relevant counseling with computer support can increase composite scores measuring sun-protective behaviors at 6 to 24 months. In young adults (n=563), brief appearance-focused behavioral interventions can decrease normative indoor tanning behaviors at 6 months, and decrease UV exposure, as objectively measured by skin pigmentation, at 12 months. In young adolescents (n=819), primary care counseling with computer support, similar to that used in adults, can decrease midday sun exposure and increase sunscreen use at 12 and 24 months. In parents of newborns (n=728), primary care counseling integrated into sequential well-child care visits can increase composite scores measuring sun-protective behaviors at 36 months. It is important to note that the trials in adults only reported composite sun protection scores and not changes in individual behavior. It is unclear if the small, but statistically significant, differences in composite scores of self-reported sun-protective behaviors translate into clinically meaningful behavior change that will prevent skin cancer. All but one trial used self-reported behavioral outcomes and therefore could be affected by social desirability bias. However, it appears that self-reported skin cancer risk behaviors are subject to minimal social desirability bias.111

Most of these trials were conducted in exclusively or predominantly white populations. Only one trial in young adolescents and one trial in college students reported inclusion of a sizeable nonwhite population. However, this restriction is reasonable, given the much higher incidence of skin cancer in white persons. In addition, one trial in the United Kingdom included only persons with “high-risk” skin characteristics (e.g., red hair, multiple nevi, history of sunburn as a child, freckling, family history of melanoma, or fair sun-sensitive skin),46 and one trial included only young women who expressed intention to indoor tan.49 Three trials used a low-intensity intervention: a single 15-minute self-directed session on a computer workstation “prescribed” by their primary care provider, an appearance-focused booklet aimed at decreasing indoor tanning, or an appearance-focused video on the effects of photoaging on skin. The remaining trials examined high-intensity counseling interventions with about four sessions (either in person or by telephone), many of which included computerized support. Although all trials reported the theoretical underpinnings of the intervention, they are described in varying detail. As such, it is often unclear if the intervention was guided by theory only or empirical data as well (see Appendix B Table 1 for details on intervention theory).

Harms of Counseling and Practicing Sun-Protective Behaviors

Overall, we found little evidence that sun-protective counseling or practicing sun-protective behaviors causes significant harms. Of the 10 trials examining the effectiveness of counseling interventions to improve sun-protective behaviors, we found no evidence that primary care relevant counseling interventions or community-based interventions involving primary care counseling paradoxically decrease sun-protective behavior. In addition, based on two trials, there is no evidence to suggest that sun-protective behavioral counseling in children or adolescents negatively impacts physical activity or BMI.54,94

Based on limited but good-quality trial evidence, it appears that higher SPF sunscreen use can increase intentional sun exposure in young adults on sunbathing vacations compared with lower SPF sunscreen use (but not sunburns). However, other fair-to good-quality trial evidence suggests that sunscreen use in general does not appear to increase sun exposure in adults or children. Two case-control studies suggest an increased risk for melanoma with sunscreen use, although other studies found no association or a protective effect. However, these studies have major methodological limitations, including the use of very crude measures of sunscreen use and lack of adequate adjustment for confounding by indication.

Due to the inclusion criteria for this review, we only included two studies examining sun-protective behaviors and possible harms of vitamin D deficiency. From one fair-quality trial, it appears that regular sunscreen use does not lead to vitamin D deficiency.98 One cohort study suggests that vitamin D levels are greatly influenced by sun exposure and that women living at high latitudes may be at risk for vitamin D deficiency during the winter and spring months.99 However, the study investigators state that “most of the women with low serum 25-hydroxyvitamin D3 were only deficient for part of the year.” A recent full report on vitamin D and cancer from the World Health Organization International Agency for Research on Cancer includes a detailed discussion of the complex relationship between serum 25-hydroxyvitamin D levels and sun exposure, and the multiple variables that potentially affect endogenous vitamin D production.112 Cutaneous vitamin D synthesis, however, varies significantly among individuals. In general, this synthesis happens relatively quickly, and prolonged sun exposure does not result in continuous increases in vitamin D synthesis, so that maximum vitamin D synthesis occurs at suberthemogenic UV doses.112 According to best estimates, during sunny summer days at approximately 40 degrees latitude, a fair-skinned person could achieve maximum cutaneous vitamin D synthesis with 5 to 10 minutes midday sun exposure to the face and forearms a few times a week. Longer exposure, approximately 30 minutes, is needed for darker-skinned persons or with less-intense sun exposure (e.g., cloudy days).112 In addition, this report recognizes the importance of exogenous vitamin D found in diet, and that fortified foods and supplements are important sources of vitamin D in the winter when skin synthesis of vitamin D is insufficient.112 Finally, it has been hypothesized that vitamin D production may be protective against certain types of cancer through vitamin D receptor-dependent or independent mechanisms. The few case-control studies published on this topic suggest that sun exposure and intermittent sun exposure in lighter pigmented persons may be inversely related to risk for advanced breast cancer, prostate cancer, and nonHodgkin lymphoma. However, this literature is very sparse, and the case-control studies have important methodological limitations, including the adequate measurement of sun exposure and lack of adjustment for vitamin D intake. Furthermore, none of the studies directly assessed vitamin D status, and the relationship between sun exposure and vitamin D status is not direct. Given the limited number of published studies, it is likely that this body of literature is significantly affected by publication bias.112

Association Between Sun Exposure, Sunscreen Use, or Indoor Tanning and Skin Cancer

We did not find any studies meeting our inclusion criteria that examined whether a change in sun exposure (e.g., due to protective clothing or avoidance of midday sun) resulted in a decrease in skin cancer outcomes. We found mainly fair-quality cohort and case-control studies examining the relationship between sun exposure and skin cancer (11 studies for squamous cell and basal cell carcinoma, 18 studies for melanoma). We found that increasing intermittent (or recreational) sun exposure is associated with an increased risk for squamous cell and basal cell carcinoma (range OR, 1.27 to 3.86). Case-control studies examining the risk for melanoma with intermittent sun exposure are inconsistent, but some studies suggest that increasing recreational sun exposure increases the risk for melanoma (range OR, 1.3 to 5.0). However, the evidence is more consistent for intermittent sun exposure in childhood leading to an increased risk for both melanoma and squamous cell and basal cell carcinoma. Fewer studies examined the association of total or chronic (or occupational) sun exposure. These studies do not suggest a strong association between total or chronic sun exposure and skin cancer. However, some evidence suggests that total sun exposure in childhood is associated with an increased risk for melanoma and occupational sun exposure may be associated with a decreased risk for melanoma. Our findings are consistent with a fair-quality systematic review by Gandini and colleagues that found a positive association for intermittent sun exposure and an inverse association for high levels of occupational or chronic sun exposure.17 Unlike our review, the meta-analysis by Gandini and colleagues included both population-based and nonpopulation-based case-control studies.

We found very limited evidence (a limited number of studies using crude measures of indoor tanning exposure) that exposure to indoor tanning devices may increase the risk for squamous cell and basal cell carcinoma, after adjusting for all important confounders. Results generally suggest no association. However, a slightly larger body of higher quality evidence suggests that “regular” or “early” use of indoor tanning may increase the risk for developing melanoma (range OR, 1.55 to 2.3). Most of these studies used crude measures of indoor tanning exposure. The one study that examined sunlamp (earlier technology) and tanning bed (more recent technology) exposure separately found a statistically significant trend (p=0.02) for frequent sunlamp use (≥6 times) and melanoma risk (OR, 1.54 [CI, 0.93–2.57]), but not for frequent tanning bed use (≥10 times) and melanoma risk (OR, 1.25 [CI, 0.79–1.98]).92 However, the study investigators state that although no association with tanning bed use was found, sufficient lag time may not have elapsed to assess a potential effect, given the more recent use of tanning beds. Our findings are consistent with a fair-quality systematic review and meta-analysis by the International Agency for Research on Cancer on artificial UV light and skin cancer that found evidence to suggest that first use of indoor tanning equipment before age 35 years increases risk for melanoma.30 This review estimated the risk for melanoma at 1.15 (CI, 1.00–1.31) based on “ever use” in 19 studies, and at 1.75 (CI, 1.35–2.26) based on first exposure during youth in 7 studies. The risk for squamous cell carcinoma, based on “ever use” in three studies, was 2.25 (CI, 1.08–4.70), and not significant for basal cell carcinoma. Unlike our review, the meta-analysis included both population-based and nonpopulation-based case-control studies.

Based on one fair-quality trial, regular sunscreen use may prevent squamous cell carcinoma (RR, 0.65 [CI, 0.45–0.94])but not basal cell carcinoma. Case-control studies that suggest sunscreen use reduces the risk for basal cell carcinoma have major limitations. Based on one fair-quality cohort (n=178,155) and four fair-quality case-control studies, there is no clear protective or harmful effect of sunscreen use on the risk for melanoma. This finding is consistent with a fair-quality systematic review and meta-analysis by Dennis and colleagues that found no association between melanoma and sunscreen use.31 This meta-analysis, however, did not report any sensitivity analyses. The primary research, nonrandomized studies examining sunscreen use, included in both our report and the meta-analysis by Dennis and colleagues, have major methodological limitations, including the use of very crude measures of sunscreen use and lack of adequate adjustment for confounding by indication.

Limitations

Given the purview of the USPSTF and the scope of our evidence report, we did not review community-based behavioral interventions to promote sun-protective behaviors (e.g., those conducted in schools, recreational, or occupational settings or media campaigns), as these were not considered feasible to implement in primary care or referable from primary care. However, interested readers can refer to the Task Force on Community Preventive Service’s recommendations and evidence report on interventions to prevent skin cancer.24,113,114

There are two major limitations in the body of evidence evaluating the effectiveness of primary care relevant counseling to prevent skin cancer. The first limitation is the generalizability of the interventions to current primary care practice. Based on rigorous trial evidence, many of the effective counseling interventions to promote sun-protective behaviors incorporated computerized support providing tailored patient education. This type of computerized support is not necessarily widely available, and the implementation of this type of support would require additional effort and cost. It is also unclear if this type of support is essential to the effectiveness of the interventions. Only one trial specifically evaluated counseling to reduce indoor tanning, and none of the trials using composite behavior scores included indoor tanning. Both trials in young adults used “appearance-focused” behavioral interventions. It is possible that different counseling messages will be effective in differently aged populations. Second, many of the counseling trials used composite sun behavior scores. It is unclear if these small changes in scores represent meaningful changes in sun-protection behavior that would reduce skin cancer or even prevent a number of sunburns. In addition, only two trials addressed skin cancer prevention counseling in children and adolescents, which based on the epidemiological evidence, is the ideal time to intervene on sun-protective behaviors. Although most of the counseling trials were conducted in predominantly white populations, this is not really a limitation for this body of literature, as white persons represent a higher risk population. One counseling trial in young adolescents and one trial in college students included a sizeable proportion of nonwhite participants. However, given that practically all of the epidemiologic studies included exclusively or predominantly white individuals, it is unclear if sun-protective behavior counseling will have similar benefits in nonwhite populations, especially given the lower incidence of skin cancer in nonwhite populations.

The epidemiological evidence examining skin cancer risk with sun exposure, indoor tanning, and sunscreen use has numerous limitations. The internal validity of the observational literature is threatened by the complex and variable nature of measuring sun exposure and sunscreen use, omission of adjustment for important confounders in many studies, and problems with recall bias for determining true exposure in case-control studies. The literature as a whole may also be influenced by publication bias. The generalizability of the observational literature addressing indoor tanning and sunscreen use is limited by the inclusion of outdated indoor tanning devices and sunscreen formulations.

Even though we limited our included studies to fair-or good-quality cohort and population-based case-control studies, this body of evidence had some consistent limitations in internal validity. Most of the cohort studies (and all of the large cohort studies) included were derived from larger cohort studies that were not primarily designed to address skin cancer-related behaviors, and therefore use fairly crude measures of exposure. In addition, a few of these cohort studies were not true inception cohort studies, meaning the cohorts were defined by those persons who answered relevant questions and did not have missing data. Even narrowing our inclusion criteria to population-based case-control studies, it is possible that cases were not necessarily representative of the whole spectrum of the examined disease in the general population. For example, many case-control studies were interview studies that excluded cases of death. This means that cases with the most aggressive cancer or advanced disease were likely underrepresented. Case-control studies often reported different ways of calculating participation rates; therefore, comparison of participation rates across studies is difficult, although we excluded studies reporting extremely poor participation rates. Also, some case-control studies excluded different types of melanoma, again making comparisons across studies difficult. In addition, melanoma research is now beginning to distinguish among types of melanoma by somatic gene mutations, and is finding differences in risk factors for the different types of melanoma. Evidence to suggest that melanomas at different body sites are associated with distinct patterns of sun exposure support this hypothesis.115 However, we did not examine cross-sectional studies or studies without true controls, and we did not include selected studies that may elucidate this association. One included case-control study presented site-specific melanoma outcomes; however, associations between different measures of sun exposure and site-specific melanoma outcomes did not seem to differ in this study (Appendix C Table 1).73

Perhaps the biggest limitation in interpreting this body of evidence is the complexity and variability in the measurement of sun exposure and important confounders, particularly for sun exposure and sunscreen use. Sun exposure is extremely complex to measure, even when broken down into total, intermittent, and chronic sun exposure. There was a large amount of heterogeneity in the actual measurement of sun exposure between studies, the categorization of levels of exposure, and the choice of reference groups. Sun-exposure measurement was defined differently, was assessed differently (e.g., objectively measured pigmentation, interview, questionnaire), and was often used in different periods of life. Complexity of measurement ranged from sun exposure indexes accounting for some aspect of measured ambient UV exposure, to cumulative hours to very broad categories (e.g., mainly indoors, indoors and outdoors, or mainly outdoors). In general, the measurement of indoor tanning or sunscreen use was crude. Measurement of sunscreen use rarely included important details, such as SPF, amount, frequency and duration, and year, as sunscreen formulations have changed over time. Likewise, measurement of indoor tanning use rarely included important details, such as rationale or motivation of use, frequency and duration, and year, as indoor tanning devices have changed over time as well.

Adjustment for important confounders and stratification to examine effect modification also varied across studies; however, studies examining sun exposure generally adjusted for age, sex, and some measure of skin phenotype or sun sensitivity. Although some studies did not adjust for sun sensitivity (i.e., skin type, ability to tan, or susceptibility to burn), most adjusted for some measure of skin phenotype in general(e.g., skin color, hair/eye color). Only four studies presented results stratified by skin phenotype, and these studies suggest an interaction between skin phenotype and skin cancer.36,71,87,116 Therefore, simply adjusting for skin type as a confounder in logistic regression may not be adequate to understand the effect of sun exposure in at-risk (e.g., poor tanners) populations. Lack of adequate adjustment and lack of stratification for skin phenotype or sun sensitivity may be an explanation for the lack of association or the inverse association reported with occupational sun exposure, as persons at low risk for skin cancer due to skin pigmentation are over-represented in outdoor workers. In addition, though most studies examining indoor tanning and sunscreen use adjusted for age, sex, and skin phenotype, not all adjusted for sun exposure. For sunscreen use, confounding by indication is extremely important and was generally not well adjusted for. Some studies also may have over-adjusted for confounding, such as adjusting for nevi, freckling, or sunburn history, as these are likely intermediate steps in carcinogenesis or surrogates for sun exposure.

The retrospective assessment of sun exposure, and in some cases important confounders, is subject to significant recall bias. This recall bias may have been less of a problem in earlier studies, such as in the 1980s when there was less public knowledge about the potential harms of UV exposure. As a corollary, assessment of past exposure, especially in childhood or the distant past, are subject to imprecision. Therefore, since most of this evidence is case-control studies, it is subject to these limitations.

Given these numerous limitations, we caution against lending much confidence to quantitative risk estimates. Given the extreme heterogeneity in the measurement of exposure, confounders, and, in some instances, outcomes (i.e., types of skin cancer), we did not attempt quantitative synthesis of risk estimates. It is also important to consider that, even though the epidemiological literature may show a statistically significant association trend in risk (e.g., from the lowest to highest percentile), people might not change their behavior to this degree (e.g., from that of the highest to that of the lowest percentile), so the estimates presented in the epidemiological literature are intended to describe primarily the strength of an observed association and dose response, important criteria for causality. These limitations also apply to the case-control studies examining the association between sun exposure and cancer other than skin cancer.

While it is also likely that this body of literature is subject to publication bias, the direction of bias may not be consistent. For example, studies showing a positive association between sun exposure and melanoma might be more likely to be published, while studies showing a positive association between sunscreen use and melanoma might be less likely to be published.

In addition to the limitations in the internal validity of this body of evidence, there are also important limitations in the generalizability of associations observed for indoor tanning and sunscreen use, both of which have changed in the recent past. Indoor tanning devices before 1980 had higher UVB content, and after 1980 had higher UVA content.92 Furthermore, modern tanning beds have undergone technological advances to enrich UVB that allow shorter duration of exposure. However, in practice, the proportion of UVB output of indoor tanning devices varies.30 Therefore, the potential harm of indoor tanning has changed during the period of all of the included studies, and adolescent or early adulthood sunbed exposure in observational studies may not be generalizable to the current exposure from indoor tanning devices. Likewise, sunscreen formulations have also changed drastically over time. SPF was introduced in 1978 and protection for UVA was not added until 1989, and UV sun exposure is approximately 5% UVB and 95% UVA.30 In addition, sunscreen formulations have also improved over time, offering higher level SPF and water resistance.

Due to the scope of this report, we did not examine the evidence between sun exposure, indoor tanning, or sunscreen use and outcomes other than skin cancer (e.g., nevi, premalignant lesions, or evidence of photoaging of the skin). We acknowledge that, therefore, we may have missed other potentially informative bodies of literature for skin cancer prevention. Due to the scope of our report and our inclusion criteria, our report does not discuss key bodies of literature on the relationship of sun exposure and vitamin D, and vitamin D and cancer risk. Interested readers should refer to the recent report by the International Agency for Research on Cancer.112

Emerging Issues and Future Research

More primary care relevant counseling trials to promote sun-protective behaviors, including those that address indoor tanning, are needed, especially in children, adolescents, and young adults. Trials of low-intensity interventions, such as the 15-minute self-administered computer session or an appearance-focused video or booklet, should be replicated in other populations. In addition to using self-reported measures of avoidance of midday sun, use of protective clothing, and use of sunscreen, trials should also consistently include measures of indoor tanning and sunburns. Trials using composite behavioral scores would be strengthened if they also provided the proportion of individuals whose behavior changed as recommended. In 2005, the National Cancer Institute and the Emory Prevention Research Center convened a workshop for skin cancer prevention investigators in the United States to develop a consensus-based set of core measures to assess UV exposure, sun-protective behaviors, and nonsolar tanning behavior. 117,118 These measures should be used consistently so as to understand their validity and reliability across different settings and populations. In addition to using these consensus-based self-reported measures, objective measures, such as dosimeters or visual observation of participant behavior, would also strengthen this body of literature.

More studies, and better designed studies, that examine the potential effect of sunscreen use and decreased sun exposure on vitamin D and other diseases hypothesized to be affected by vitamin D (e.g., cancer, autoimmune disease, bone-related disease) are needed. Trial evidence suggests that sunscreen is effective in reducing risk for squamous cell carcinoma, but it is unclear if regular sunscreen use prevents basal cell carcinoma or melanoma. However, nonrandomized studies examining sunscreen use have serious methodological limitations. It is therefore important to determine if the increase in recreational sun exposure, even if it does not increase risk for sunburns, has clinically important sequelae. Currently, the epidemiologic literature supporting an association between sun exposure and breast and prostate cancer and nonHodgkin lymphoma is sparse and has serious methodological limitations. Therefore, more studies are needed that account for the measurement of sun exposure, adjustment for important confounders, and direct assessment of vitamin D levels, if possible. Currently, there is no evidence to suggest that sun-protective behavior messages aimed at reducing prolonged or intense sun exposure and sunburns cause significant harm, such as vitamin D deficiency or increasing risk for cancer. In addition, more studies with more detailed assessment of sunscreen and indoor tanning use are needed. It is important that these studies consistently adjust for both important host and environmental factors. Survey instruments to assess these types of exposure need to be reliable and validated. The body of evidence would be strengthened if studies used the same or comparable measurements to facilitate comparison across studies. It will likely take decades to see a potential protective effective of regular sunscreen use on melanoma risk or the potential harms of current tanning beds. Therefore, studies evaluating current sunscreen formulations will continue to be necessary over time.

Conclusions

A limited number of RCTs suggest that primary care relevant behavioral counseling can minimally increase composite scores measuring self-reported sun-protective behaviors in adults and their newborns, decrease self-reported indoor tanning use and objectively measured pigmentation in college students, and decrease self-reported midday sun exposure and increase sunscreen use in young adolescents. The clinical significance of small changes in sun protection composite scores is unclear. Many of the counseling interventions incorporated computerized support that could generate tailored feedback.

Primary care counseling to prevent skin cancer and the practice of sun-protective behaviors to limit intense or prolonged sun exposure do not appear to have significant harms, but methodologically rigorous studies examining the potential harms of vitamin D deficiency are lacking. There is evidence, mostly from case-control studies, to suggest that intermittent sun exposure, especially in childhood or adolescence, may increase risk for all types of skin cancer. Regular sunscreen use can decrease the incidence of squamous cell carcinoma, but it is unclear if it can prevent basal cell carcinoma or melanoma. Based on a limited number of studies, it appears that regular and early use of indoor tanning may increase the risk for melanoma. These risks, however, may not apply to current devices, since tanning devices have changed significantly over the past 20 to 30 years. Therefore, behavioral counseling to promote skin cancer prevention should focus on improving multiple behaviors to reduce UV exposure and not improving sunscreen use alone.

One counseling trial in young adolescents and one trial in college students included a sizeable proportion of nonwhite participants. However, given that practically all of the epidemiologic studies included exclusively or predominantly white persons, it is unclear if sun-protective behavior counseling will have similar benefits in nonwhite populations, especially given the lower incidence of skin cancer in nonwhite populations.