Melanoma Risk in Relation to Use of Sunscreen or Other Sun Protection Methods

Longstanding recommendations to reduce the harmful effects of sun exposure to the skin include sun avoidance, especially during peak day time hours, skin coverage via clothing and wide brimmed hats, use of sunscreen with a high sun protection factor (SPF), and seeking shade. Of these, sunscreen use is the action that the lay public most associates with skin cancer prevention—in one recent report, 72% of respondents volunteered sunscreen compared with 63% offering sun avoidance and 46% suggesting protective clothing or hats to prevent skin cancer (1). Despite public perception, the scientific evidence that sunscreen prevents skin cancer is controversial (2–5). Randomized controlled trials have shown reductions with sunscreen use in actinic keratoses in adults (6, 7) and nevi in children (8) precursors of squamous cell carcinoma and melanoma, respectively. One trial, involving 1,621 participants, found squamous, but not basal cell, skin cancers to be lower among individuals assigned to daily sunscreen use relative to those assigned to continue their usual sunscreen practices (9, 10). An extension of that trial found 11 melanomas diagnosed in the sunscreen compared with 22 melanomas in the comparison group (11). This small trial is the only experimental study to examine melanoma incidence in relation to sunscreen use.

Owing to melanoma's rarity, additional randomized controlled trials testing sunscreen's efficacy to prevent melanoma are unlikely. For this reason, case–control studies are useful because they allow for greater precision to estimate the association between sunscreen use and melanoma if the sample size is sufficiently large, and for more detailed assessment of lifetime sunscreen use. However, case–control studies to date have been inconsistent, with sunscreen found to increase, decrease or to be unrelated to melanoma risk, resulting in a summary OR near the null in meta-analyses (12, 13). These inconsistencies may be due to study design (hospital- vs. population based), small sample size, variation in sunscreen exposure measurement (ever, frequency, or years used), and inability to control potential confounding, such as sunburn, skin sensitivity to the sun, or sun exposure. In addition, almost all case–control studies were conducted prior to availability of sunscreen products with higher SPF, and they lacked information about adequacy of sunscreen practices.

We recently completed a case–control study of risk factors for melanoma (14), with cases diagnosed between 2004 and 2007, in which we collected detailed information about sunscreen use. Our study differs from other published case–control studies in several ways. First, we expanded the assessment of frequency of sunscreen use to ascertain sunscreen use during various year-round outdoor activities throughout the lifetime. Second, we collected lifetime sunscreen use patterns, such as use of sunscreen with an SPF of 15 or greater, amount applied, amount of skin covered, reapplication, and routine use when not planning to be in the sun. Third, we included a measure of use of other sun protection methods, such as clothing or staying in the shade. Finally, we collected information on established melanoma risk factors, including a detailed assessment of sun exposure, and recruited a sample (1,167 cases; 1,101 controls) that was far larger than any previous case–control study.

Methods for the Skin Health Study have been previously published (14). Briefly, we conducted a population-based case–control study in Minnesota among persons diagnosed with invasive cutaneous melanoma (cases) between 2004 and 2007 at ages 25 to 59, ascertained by the state cancer registry. Controls were persons without melanoma, randomly selected from the state drivers' license list (including persons with state identification cards) and frequency matched to cases on age and gender. We were unable to determine eligibility for 557 (27.5%) of 2,026 total cases due to physician refusal (n = 124; 6.1%), inability to locate a phone number necessary for data collection or to reach them by phone (n = 235; 11.6%), patient refusal (n = 79; 3.9%), death (n = 23; 1.1%) or other (n = 96; 4.7%). A total of 89 (4.4%) cases were not eligible due to having prior melanoma, noncutaneous melanoma, an ineligible diagnosis, or a language barrier. Among 3,095 controls, eligibility status was unknown for 1,354 (43.7%) for the following reasons: no phone number located (n = 598; 19.3%), no response to our telephone calls (n = 273; 8.8%), subject refusal (n = 468; 15.1%), or death (n = 15; 0.5%). A total of 151 controls (4.9%) were found to be ineligible, due to residing outside of Minnesota at the time of contact (n = 63), language barriers (n = 74), or a prior melanoma diagnosis (n = 14). Altogether, 1,380 of 2,026 (68.1%) cases and 1,590 of 3,095 (51.4%) controls were screened and eligible; among these potential participants, 1,167 cases and 1,101 controls (84.6% and 69.2% of eligible, respectively) completed a self-administered questionnaire and a telephone interview.

Sunscreen use was collected via telephone interview, an approach we developed after in-person interviews with 11 sunscreen users and piloted with 32 individuals. First, we asked about time spent in 12 outdoor activities in which the participant had engaged for at least 4 days per year in the decade years, for example, at age 10, 20, 30, 40, and 50, depending on a person's reference age. The outdoor activities included time spent at the beach or pool, sunbathing, boating or waterskiing, fishing, playing or coaching outdoor team sports, walking, hiking or jogging, biking, roller skating or rollerblading, golfing, playing tennis, playing outside, gardening or performing outdoor chores, and winter sports (e.g., snow skiing, ice skating, sledding, hockey). For each reported outdoor activity, participants were asked how often they used sunscreen lotions or creams to prevent sunburn (0 for never to 5 for almost always).

If sunscreen use was reported for any activity during a decade year, we asked additional questions about sunscreen use patterns in that decade year; questions about frequency of use were based on the same 6-point scale described above. The questions included the frequency of sunscreen use with a SPF of 15 or greater (SPF15+), the thickness of sunscreen application (from 1 to 3 for thin, medium, thick), the amount of exposed skin that was covered (from 1 to 3 for none/very little, some, to most/all), the frequency of sunscreen reapplication after being in the sun for longer than 2 hours, and the frequency of sunscreen use when not specifically doing a sun-related activity (hereafter referred to as “routine” sunscreen use). After the sunscreen questions, all participants were also asked a single question in each decade year: “how often did you do other things to protect your skin from the sun, like wearing a hat, long-sleeved shirt or staying in the shade?”. Response categories were 0 for never to 5 for almost always.

To obtain lifetime use of sunscreen, we averaged a participant's reported frequency score of sunscreen use across all reported activities within a decade and across all decade years. For the specific sunscreen use patterns, we averaged values for each decade-specific pattern across each decade year among persons who reported any outdoor activity-related sunscreen use. We further classified individuals into 3 categories to represent non-, inconsistent-, or optimal-users based on scores from the 2 most recent decades. We chose this period because (i) all participants had a minimum of 2 decades of potential exposure regardless of reference age, (ii) this period encompassed public health recommendations for sun protection and availability of higher SPF sunscreens, and (iii) the highest mean values for each measure were observed in the most recent 2 decades. For all sunscreen measures, individuals who reported no sunscreen use in both decades were classified as nonusers. Then, depending on the sunscreen measure, we classified optimal users as those individuals who had scores that exhibited the most desired sunscreen pattern in both decades, as follows: (i) mean score in the highest tertile (after excluding nonusers or score of zero) for sunscreen use during outdoor activities, (ii) use of SPF15+ sunscreen more than half the time, (iii) reapplication of sunscreen more than half the time, (iv) routine sunscreen use more than half the time, (v) thick sunscreen application, or (vi) coverage of all or most of the skin. All other participants fell into the inconsistent sunscreen user category. For other sun protection methods, individuals who reported rarely or never using these methods in both decades were classified as nonusers, those who reported using these methods more than half the time in both decades were considered to be optimal users, with the remainder designated as inconsistent users.

We collected skin, hair, and eye color, presence and pattern of moles, and ever use of indoor tanning via self-administered questionnaire. Education, income, all sun exposure measures, history and number of painful sunburns, tanning ability, and family history of melanoma were collected during the telephone interview. Lifetime routine sun exposure was obtained by multiplying the number of days by the number of hours typically spent outside on weekdays and weekends during winter and summer months and summing across decades (15). For outdoor activity-related sun exposure, the total number of days spent in each outdoor activity was multiplied by the number of hours for each activity, and summed across activities and decades. On the basis of hair and eye color, and tanning ability, we derived a phenotype risk score, ranging from 1 (lowest) to 5 (highest) for risk of melanoma. For hair color, a score of 1 was assigned for dark brown or black, 2 for blond or light brown, and 3 for red; having eye colors of grey, blue, green or hazel contributed 1 additional point to the phenotype score, as did having no ability to tan (16).

Using general linear regression models, we first compared mean lifetime scores between cases and controls for sunscreen use associated with outdoor activities, each specific sunscreen pattern, and use of other sun protection methods. For the categorical measures of exposure, we used logistic regression to calculate OR and 95% CI to compare optimal- and inconsistent- to nonusers among cases and controls. We repeated this analysis for melanoma site (head and neck, trunk, upper, or lower limbs) and Breslow's depth of the tumor (< 1 mm, 1–1.99 mm, ≥ 2.0 mm). Mean differences and P values between cases and controls or ORs and CIs were adjusted for gender, age (continuous), income (≤ $60,000; > $60,000, missing), education (completed college, did not complete college), phenotype risk score (continuous), moles (none, few, some, many, missing), family history (yes, no, missing), total lifetime painful sunburns lasting more than one day (continuous), ever use of indoor tanning (yes, no), routine sun exposure (continuous) and sun exposure from outdoor activities (continuous).

To address concerns about confounding by indication for sunscreen use and questions related to the compensation hypothesis (i.e., sunscreen users compensate for high SPF sunscreen by spending more time in the sun, thereby increasing their risk of melanoma), we conducted stratified analyses for the association between SPF15+ or other sun protection methods and melanoma risk by phenotypic risk or sun exposure. We restricted these analyses to cases and controls, depending on the analysis, who were either (i) optimal users of SPF15+ sunscreen or nonusers of sunscreen, or (ii) optimal- or nonusers of other sun protection methods. Crude and adjusted ORs and CI were calculated within strata of phenotypic risk (scores of 1 or 2, low; 3, intermediate; 4 or 5, high), routine sun exposure (in tertiles), sun exposure during outdoor activities (in tertiles), sun exposure during water-related activities (in tertiles), sun exposure from sunbathing (none, low, high), and sun exposure during outdoor work (none, low, high). When stratifying by phenotypic risk score, we adjusted for all other characteristics as previously described except for the stratification variable. When stratifying by sun exposure, we adjusted for all other characteristics except those related to sun exposure.

Information about sunscreen use was available for about 97% of all cases and controls. Completeness of data for use of specific sunscreen patterns or other sun protection methods, along with all adjustment variables, ranged from 979 (83.9%) to 1106 (94.8%) of cases, and from 888 (80.7%) to 1034 (93.9%) of controls, depending on the measure.

As previously published (14), cases were significantly more likely than controls to have blue or green eyes, red or blond hair, fair or very fair skin, many moles or freckles (any amount), and higher income. A history of 3 or more sunburns or use of indoor tanning was associated with an increased risk of melanoma. Positive associations between melanoma and routine sun exposure, sun exposure from outdoor activities or jobs, or family history, were not observed.

The mean frequency of outdoor activity-related sunscreen use reported by controls in each decade year according to their age at the reference date was low (Fig. 1). Within each age group, controls reported greater use of sunscreen with each passing decade, except for the youngest, whose use leveled off between age 10 and age 20, marking the transition from child- to adulthood. In any given decade year, the youngest controls to reach that decade had the highest while the oldest had the lowest mean sunscreen scores. With some minor variation, these trends by age within and across decades were observed among controls for each specific sunscreen use pattern and for use of other sun protection methods (data not shown).

Mean frequency of lifetime sunscreen use during outdoor activities was similarly low among both cases and controls (Table 1). Among sunscreen users, however, mean lifetime use of sunscreen with SPF of at least 15 was significantly higher among controls than cases (adjusted mean, 3.55 vs. 3.40, P value = 0.03). For the remaining sunscreen patterns (amount applied, reapplication, amount of skin covered and routine use), no differences in adjusted mean lifetime scores between cases and controls were observed. In contrast, the adjusted mean score for lifetime use of other sun protection methods was statistically significantly higher among controls than cases (adjusted mean, 2.36 vs. 2.20, P value = 0.0061).

The proportion of controls categorized as optimal sunscreen users in both decades was less than 15% for almost all measures, including outdoor activity-related sunscreen use (13.6%), application thickness (4.8%), reapplication frequency (9.6%), and routine sunscreen use (3.5%; Table 2). Only use of SPF15+ sunscreen and covering most exposed skin were consistently reported in both decades by a substantial proportion of controls (48.3% and 52.8%, respectively). Optimal use of sunscreen during outdoor activities in the most recent 2 decades was not associated with risk of melanoma, but a suggestion of a trend toward decreased risk of melanoma with increasing use of SPF15+ sunscreen was observed (P_trend = 0.03). Frequent routine sunscreen use in both decades was strongly associated with lower likelihood of melanoma (adjusted OR 0.44, 95% CI: 0.23–0.86), observed in both men and women (data not shown). Adjusted ORs for optimal or inconsistent versus nonuse for the remaining sunscreen patterns and melanoma risk were largely null. However, individuals who reported use of other types of sun protection methods in both decades compared with those who rarely or never used these methods were at low risk of developing melanoma, even if their use was inconsistent across time. Relative to never or rare users of other sun protection methods, the multivariate-adjusted OR for the likelihood of melanoma among inconsistent users was 0.80 (95% CI: 0.63–1.01), and for optimal users of other sun protection methods, the multivariate-adjusted OR was 0.59 (95% CI: 0.44–0.78). This trend was highly statistically significant (P = 0.0003).

For the sun protection strategies with any significant finding related to reduced melanoma risk, specifically SPF15+, routine sunscreen use, and other sun protection methods, analyses were restricted to optimal versus nonusers (SPF15+ sunscreen, routine sunscreen), or the least frequent users (other sun protection methods) and repeated according to Breslow's depth or tumor site of melanoma diagnosed in the cases (Table 3). The previously observed statistically significant inverse association for optimal users relative to infrequent users of other sun protection methods remained regardless of Breslow's depth or tumor site. Except for melanomas arising on the head or neck (adjusted OR 1.33, 95% CI: 0.25–7.20), inverse associations were similarly observed for the optimal use of sunscreen on a routine basis across Breslow's depth categories and other tumor sites. However, the proportion of participants reporting optimal use was very low, resulting in wide CIs that tended to include the null value. Among optimal users of SPF15+ sunscreen relative to nonusers, the likelihood of decreased melanoma risk was observed only for the thinnest tumors (adjusted OR 0.65, 95% CI: 0.45–0.94).

Within both the highest and lowest levels of phenotypic risk, crude ORs comparing optimal SPF15+ sunscreen users to nonusers of sunscreen showed either a very weak reduction or null association with melanoma (Table 4). These associations became more strongly consistent with a decreased risk of melanoma, after accounting for demographic and other risk factors, routine and outdoor activity sun exposure; however, CIs included the null value. When stratified by sun exposure levels, crude ORs for melanoma and optimal use of SPF15+ sunscreen were elevated among individuals in the highest sun exposure category, regardless of whether time spent in the sun accrued from routine daily living, outdoor activities, outdoor water activities, sunbathing or working outdoors; no clear pattern was observed within low or intermediate levels of sun exposure. After adjustment, individuals in the low or intermediate levels of sun exposure who were optimal users of SPF15+ were consistently less likely to be cases than controls compared with nonusers of sunscreen. In the highest level of sun exposure, adjustment resulted in attenuation of the elevated crude ORs toward the null, but did not reverse the direction of these associations. The P value for interaction for these observed patterns between use of SPF15+ sunscreen and melanoma according to amount of sun exposure was statistically significant for outdoor activity-related sun exposure (P = 0.0334).

In this comprehensive epidemiologic assessment of use of sunscreen or other sun protection methods in relation to melanoma, optimal use of sunscreen during outdoor activities across 2 decades prior to reference date was uncommon among cases and controls, and the amount of sunscreen used and the frequency with which it was reapplied was low. This may explain why sunscreen use or most usage patterns were not associated with risk of melanoma. In contrast, individuals who reported high frequency of sunscreen use when not planning to be in the sun or use of other sun protection methods, like hats, long-sleeved shirts or staying in the shade, experienced lower risk of melanoma than individuals who did not adhere to these practices. Although more diligent adherence to proper sunscreen practices during outdoor activities may yet be important for melanoma prevention, our data suggest that routine sunscreen use or use of sun barrier and sun avoidance methods have the potential to be important strategies to reduce melanoma incidence.

Although the proportion of cases and controls reporting optimal use of sunscreen on a routine basis was very low, our observation of an inverse association with melanoma risk is consistent with findings from the sole experimental study to examine sunscreen use and melanoma, the Nambour Skin Cancer Prevention Trial (11). Study participants assigned to the intervention were instructed to apply sunscreen with an SPF of 16 daily to head, neck, arms and hands, and to reapply, if necessary, throughout the day; control participants followed usual sunscreen practices. Among intervention participants, compliance with daily sunscreen use was 56%, mean quantity of sunscreen applied was about half of the currently recommended amount (1 mg/cm²; ref. 17), and use waned after conclusion of the trial, although it remained higher 5 years later compared with control participants (18). Despite these limitations, incidence of in situ and invasive melanoma in the intervention group was half that of the comparison group 10 years following the trial's conclusion. One other case–control study assessed “regular” sunscreen use, and found it to be associated with an increased risk of melanoma (19), but the definition of regular sunscreen use was not provided and likely differed from our measure, and adjustment factors were less extensive than in our study. We observed this inverse association among women, who would be expected to routinely use sunscreens in moisturizers and cosmetics. These products, if they are formulated to reflect or absorb both UVB and UVA, have been shown to decrease oxidative stress and reduce biological markers of photoaging associated with suberythemal sun exposure (20), so our results are plausible. Surprisingly, we did not observe an association between routine sunscreen use and head and neck melanomas, where these products are most commonly applied, perhaps due to residual confounding from higher sun exposure to the body site that is most likely to be regularly exposed to the sun. Alternatively, our measure may be a marker for healthy behavior, albeit employed by an exceedingly small group, rather than representing actual routine sunscreen use.

Although sunscreen use was the focus of our study, our most robust finding was that use of other sun protection methods was consistently associated with lower risk of melanoma, observed among persons regardless of the tumor thickness or location, their melanoma risk profile or their amount or source of sun exposure. Intuitively, this latter observation is logical considering that individuals often apply sunscreen to spend more time in the sun, whereas clothing and use of shade are ways to block the skin from the harmful effects of the sun. Although experimental studies show the ability of clothing to block UVB and UVA (21, 22), and surveys find severe sunburns, which increase risk of melanoma, are less commonly reported by persons who regularly choose clothing, shade or mid-day sun avoidance compared with persons who do not (23), epidemiologic evidence is limited for the effect of these strategies on melanoma occurrence. In a case–control study conducted in Australia, melanoma was more common among persons whose melanoma site was exposed during summer outdoor work compared with persons who were usually covered (24). Similarly, sun sensitive women who wore bikinis were more likely than women who wore full swimsuits to develop melanoma on their trunks in a nested case–control study (25). A major limitation of our measure is that we lumped together alternative methods to sunscreen, including use of hats (without specifying that they be broad brimmed), long-sleeved shirts (but not long pants) and seeking shade (but not avoidance of sun during peak hours). Nevertheless, this nonspecific measure is likely to represent at least one of these actions, thereby reflecting the experience of individuals who take particular precautions to avoid exposing their skin to the sun. Clearly, additional work is needed to tease out the contribution of the various forms of sun avoidance strategies to a reduced risk of melanoma.

Case–control studies that have found sunscreen to be associated with an increased risk of melanoma have been criticized due to possible confounding by indication, namely, individuals at highest risk of melanoma are more likely to use sunscreen. Another criticism is that individuals compensate for use of high SPF sunscreen by spending more time in the sun, leading to a positive association between sunscreen use and melanoma. We specifically addressed these criticisms by performing stratified analyses according to phenotypic risk for melanoma and the amount of time spent in the sun in either a chronic or intermittent manner. Because we limited these analyses to non- and optimal users and further divided the sample size across 3 categories of phenotypic risk or sun exposure, the sample size within strata was small and CIs tended to include the null value even if the observed ORs were strongly consistent with a decreased risk of melanoma. Although the proportion of individuals who were optimal users of SPF15+ sunscreen was indeed highest for those in the highest category of risk based on phenotype, a reduced risk of melanoma was apparent among persons in the low and high-risk phenotype groups after accounting for potential confounding. Our results may differ from other case–control studies that found the opposite, in part, because we utilized a phenotype index rather than examining risk separately by hair, eye or skin color, as others have done (19, 26–28), or because we accounted more thoroughly for confounding, especially from sun exposure and sunburns.

To assess the compensation hypothesis as the explanation for higher risk of melanoma among sunscreen users relative to nonusers, we considered various opportunities when sun exposure occurs, from that associated with routine or outdoor activities to sunbathing and outdoor work, each of which may have different implications for sunscreen's effectiveness to prevent melanoma. Three previous case–control studies attempted to examine this issue, but they were limited to stratifying cases and controls according to number of holiday weeks per year in sunny resorts (19), frequency or duration (hours/occasion) of sunbathing (27), or duration of sunny vacations (weeks/year) (28), limiting direct comparison with our results. Our findings suggest a complex interrelationship between sunscreen practices and sun exposure for their effect on melanoma risk. Among persons with the highest amount of sun exposure regardless of source, optimal use of SPF15+ increased the risk of melanoma consistent with the compensation hypothesis. Although this association was largely explained after adjustment for phenotypic risk and other confounders, sunscreen users with high sun exposure did not experience a reduction in melanoma risk. On the other hand, the likelihood of melanoma was clearly reduced among persons with the lowest amount of sun exposure from outdoor activities and who were also optimal users of SPF15+ sunscreen, consistent with more stringent sun protection practices that combine sun avoidance with sunscreen use.

We acknowledge that our response rates were low, differentially so among controls than cases. Selection bias could be problematic for our key findings if cases who used sunscreen on a routine basis or other sun protection methods, or if controls who did not, were less likely to participate. Although we did not have information on these measures among nonparticipants, we found that past indoor tanning use was similar between participating and a subset of nonparticipating cases and controls (14). Recall bias is also a concern, however, if present, we would anticipate that controls would overreport all skin cancer prevention practices relative to cases, which we did not observe. In addition, our exposure measures were unique to our study and have not been validated. To date, studies of the accuracy of self-report of sunscreen use have primarily assessed reliability (29–32), or obtained a measure of validity by comparing self-report to a skin swab for presence or absence of sunscreen (33, 34), a technique that is not feasible for recall of past exposures. A few studies have also compared recent recall of wearing hats, long sleeves, or pants to direct observation, in beachgoers, parents visiting pools, or postal workers (34–36). Kappa coefficients reported in these studies have indicated good to excellent agreement for reliability or validity. Except for one study that reported kappa coefficients of 0.54 to 0.74 for sunscreen use and 0.59 to 0.71 for the SPF by occasion or location 10 years ago (32), none of these studies provides data for sunscreen or clothing use across the lifetime. The trends we observed for sunscreen use by decade and age are reassuring about the validity of our measures as they follow what we would expect with adoption of sunscreen recommendations over time.

Unlike other case–control studies, our sample size was larger, and we examined longer duration of sunscreen use and the effect of its proper use on melanoma risk, while accounting for potential confounding by phenotype or ultraviolet radiation exposure from artificial or solar sources. Although metaanalyses find moderately strong associations with intermittent sun exposure (usually defined as sunbathing, time spent during sunny vacations, or outdoor recreational activities; ref. 37), we found that cases were just as likely as controls to report sun exposure during intermittent outdoor activities. This could be due to differences in the way we assessed intermittent exposure, or measurement error, which could be problematic for interpretation of our results. Finally, we were unable to directly address hypotheses, first proposed in 1993 (38), that sunscreen increases melanoma risk by increasing exposure to UVA because SPF only protects against UVB, by absorption of chemicals in sunscreen, by decreasing vitamin D synthesis, or by interfering with the skin's innate defense mechanisms. Because we found no relationship between melanoma and sunscreen use or related usage patterns, and decreased risk of melanoma risk for use of sunscreen on a routine basis or other sun protection methods, these hypothesized mechanisms are unlikely to be operating in our study.

The public health message for skin cancer prevention is complex, offering multiple recommendations that vary with regard to acceptability, convenience, and cost, and conflict with the public's positive attitudes that tanned skin is healthy and attractive. The message is further complicated by recent recognition of vitamin D insufficiency in the population, and of the need to be more physically active to prevent obesity (39). Furthermore, popular media reinforces sunscreen use, but not other sun protection methods, as it simultaneously promotes tanned skin as a marker of health and beauty (40). Our data suggest ranking other sun protection methods, such as clothing or sun avoidance, higher than sunscreen for reducing melanoma risk. Although many health agencies list sun avoidance and sun blocking mechanisms over or to be used in conjunction with sunscreen, examples exist where this is not the case, and this inconsistency may contribute to public perception that sunscreen is the first action for skin cancer prevention (1).

If sunscreen continues to be the public's preferred choice for skin cancer prevention, more emphasis needs to be placed on its correct use to maximize its potential efficacy, as our documentation of sunscreen patterns shows considerable room for improvement. Results from the 2005 National Health Interview Survey underline the challenges in conveying clear skin cancer prevention messages to the public–only 4.2% to 24.4% of respondents reported practicing all skin cancer prevention recommendations, with the lowest prevalence (4.2%) found among individuals 18 to 29 years old and the highest prevalence (24.4%) among persons age 65 or older (41). Of the 5 recommended behaviors, use of sun-protective clothing was the least commonly reported strategy employed. And adoption of these behaviors is not improving—between 1992 and 2010, the average annual percent change decreased for use of clothing and did not change for sunscreen or staying in the shade (42). To reverse these trends and reduce melanoma incidence, identification of effective strategies will need to carefully consider the target group, the intervention medium, and the context (43). For example, a recent trial found that 3 tailored skin cancer prevention messages compared with a single general message delivered to adults via mail increased hat use, but not use of shirts or sunscreen, nor sun avoidance (44). Another trial reported that adult beachgoers who received education and a UV photograph of their damaged skin compared with education alone increased their hat and sunscreen use but did not reduce their time spent in the sun (45). These 2 examples highlight the limitations of a “one-size-fits-all” approach to adoption of skin cancer prevention behavior.

In summary, our study provides some new insights about sunscreen patterns and alternative sun protection methods in relation to risk of melanoma. The infrequent and incorrect use of sunscreen, however, continues to limit our ability to draw firm conclusions about the efficacy of sunscreen to prevent melanoma. In the short term, our data offer some support for improving public health messages about the relative order and proper use of sun protection strategies to reduce the harmful effects of the sun on skin health, and in the long term, call for continued research on the extent to which these strategies are effective against melanoma.

No potential conflicts of interest were disclosed.

Funding for this research was provided by the American Cancer Society (RSGPB-04-083-01-CCE) and the National Cancer Institute (5R01CA106807).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.