You Are Here:
A Review of the Evidence for the U.S. Preventive Services Task Force
Release Date: February 2011
By Jennifer S. Lin, MD, MCR; Michelle Eder, PhD; and Sheila Weinmann, PhD, MPH
The information in this article is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This article is intended as a reference and not as a substitute for clinical judgment.
This article may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.
This article was first published in Annals of Internal Medicine on February 1, 2011 (Ann Intern Med 2011;154:190-201; http://www.annals.org).
Background: More than 2 million cases of skin cancer are diagnosed annually in the United States, and melanoma incidence is increasing.
Purpose: To assist the U.S. Preventive Services Task Force in updating its 2003 recommendation on behavioral counseling to prevent skin cancer.
Data Sources: Existing systematic reviews, database searches through February 2010, and outside experts.
Study Selection: English-language, primary care-relevant counseling trials to promote sun-protective behaviors and studies examining the association between sun-protective behaviors and skin cancer outcomes or potential adverse effects were included.
Data Extraction: Each study was appraised by using design-specific quality criteria. Important study details were abstracted into evidence tables.
Data Synthesis: 11 fair- or good-quality, randomized, controlled trials examined the counseling interventions' effect on sun-protective behaviors. In young women, appearance-focused behavioral interventions decrease indoor tanning and ultraviolet exposure. In young adolescents, computer support can decrease midday sun exposure and increase sunscreen use. Thirty-five mainly fair-quality observational studies examined the relationship between ultraviolet exposure or sunscreen use and skin cancer. Increasing intermittent sun exposure in childhood is associated with an increased risk for squamous cell carcinoma, basal cell carcinoma, and melanoma. Evidence suggests that regular or early use of indoor tanning may increase melanoma risk. On the basis of 1 fair-quality trial, regular sunscreen use can prevent squamous cell carcinoma, but it is yet unclear if it can prevent basal cell carcinoma or melanoma.
Limitations: There are limited rigorous counseling trials. Observational studies are limited by the complexity of measuring ultraviolet exposure and sunscreen use, and inadequate adjustment for important confounders.
Conclusion: Randomized, controlled trials suggest that primary care-relevant counseling can increase sun-protective behaviors and decrease indoor tanning.
Primary Funding Source: Agency for Healthcare Research and Quality.
In the United States, more than 2 million cases of nonmelanoma skin cancer are diagnosed each year. Of these cases, about two thirds are basal cell carcinoma and one quarter are squamous cell carcinoma (1). Although melanoma is considerably less common than basal cell or squamous cell carcinoma, it now accounts for about 75% of skin cancer deaths (1). Age-adjusted incidence rates for melanoma among white Americans have increased from approximately 8.7 per 100,000 in 1975 to 25.3 per 100,000 in 2007 (2). Several factors may contribute to this increase in incidence, including increased ultraviolet exposure, increased public awareness of the warning signs of melanoma, and increased screening by clinicians (3-5).
Skin cancer has well-known host and environmental risk factors. Several phenotypic characteristics are associated with skin cancer risk, including hair and eye color, freckles, and tendency to sunburn (6, 7). Exposure to solar ultraviolet radiation is the most important environmental risk factor for all types of skin cancer (8). Therefore, the primary strategies for preventing skin cancer include limiting ultraviolet exposure by avoiding midday sun, wearing protective clothing and broad-brimmed hats, applying sunscreen, and avoiding indoor tanning (7). However, sun-protective counseling in primary care varies in frequency and content (9-11), despite data suggesting that these behaviors need to be improved (12). Among adolescents in the United States, for example, about 83% reported at least 1 sunburn during the previous summer, only 34% reported sunscreen use, and nearly 10% of adolescents and 20% of young adults reported indoor tanning during the previous year (13, 14).
In 2003, the U.S. Preventive Services Task Force (USPSTF) concluded that evidence was insufficient to recommend for or against routine counseling by primary care clinicians to prevent skin cancer because of the uncertainty surrounding whether clinician counseling is effective in changing patient behaviors to reduce skin cancer, the uncertainty about potential harms of sun-protective behaviors, and availability of only fair-quality evidence linking sunscreen use or indoor tanning to skin cancer outcomes (15). Therefore, using the USPSTF methods (16), we developed an analytic framework with 5 key questions focusing on the evidence gaps identified in 2003 (Appendix Figure 1).
Key question 1: Is there direct evidence that counseling patients on sun-protective behaviors reduces sunburns, nevi, actinic keratoses, or skin cancer?
Key question 2: Do primary care-relevant counseling interventions change sun-protective behaviors?
Key question 3: Do primary care-relevant counseling interventions have adverse effects?
Key question 4: Are certain behaviors (for example, changes in sun exposure, indoor tanning, or sunscreen use) associated with skin cancer outcomes?
Key question 5: Are sun-protective behaviors associated with adverse effects?
Data Sources and Searches
We initially searched for existing systematic reviews from 2001 to March 2008 and evaluated 15 relevant systematic reviews, in addition to the previous evidence report, for quality and their potential in answering questions or identifying primary research for each question (15, 17-31). We used 10 reviews to identify primary evidence and subsequently searched from the end dates of existing systematic reviews through February 2010 (Table 1) (15, 17, 18, 21, 23, 32-36). Details of the existing systematic reviews search are included in the full report (37). We identified 6132 abstracts through MEDLINE and the Cochrane Central Register of Controlled Trials and 165 articles from outside experts and reviewing bibliographies of other relevant articles and existing systematic reviews (Appendix Figure 2).
We reviewed all abstracts and articles for potential inclusion on the basis of a priori-determined inclusion criteria (Appendix Table 1). For key questions 1 to 3, we included randomized or controlled clinical trials evaluating behavioral interventions that were conducted in primary care settings, judged to be feasible for delivery in primary care (for example, mailed or electronic interventions) or widely available for referral from primary care. Outcomes for key question 2 included self-reported or directly observed measures of sun-protective behaviors (for example, limitation or avoidance of midday sun, use of sun-protective clothing, use of sunscreen, or limitation or avoidance of indoor tanning) at 3 months of follow-up or longer. For key questions 4 and 5, we included trials, cohort studies, and population-based case-control studies. We excluded cross-sectional studies that were ecological analyses and hospital-based case-control studies because hospital-based control participants are not generally representative of the community, and hospital-based cases can introduce considerable selection bias (38, 39). Outcomes for key question 5 included potentially clinically important harms (for example, paradoxical increase in sun exposure, reduced physical activity, dysphoric mood, vitamin D deficiency, and increased incidence of nonskin cancer).
Two investigators independently screened 6132 abstracts, 73 articles for key questions 1 to 3, and 309 articles for key questions 4 and 5.
Data Abstraction and Quality Assessment
Two investigators independently assessed study quality using the USPSTF's study design-specific quality criteria and the Newcastle-Ottawa Scale for assessing cohort and case-control studies (16, 40). All poor-quality studies were excluded. Listings of all excluded articles are included in the full evidence report (37).
We found no trials for key question 1, 13 articles (11 unique trials) for key questions 2 and 3, 60 articles (35 unique studies) for key question 4, and 19 articles (17 unique studies) for key question 5. One primary reviewer abstracted relevant information into standardized evidence tables for each included article. A second reviewer checked the abstracted data for accuracy and completeness.
Data Synthesis and Analysis
We were unable to conduct quantitative synthesis primarily because of the heterogeneity of the populations addressed and counseling intervention methods and measurement of exposures and outcomes. Instead, we qualitatively synthesized our results, stratified by population counseled (adults, young adults with a mean age of 18 to 21 years, and children) or type of exposure.
Role of the Funding Source
The Agency for Healthcare Research and Quality funded this research under a contract to support the USPSTF, provided project oversight, reviewed the draft evidence synthesis, and assisted in external review of the draft evidence synthesis. The Agency for Healthcare Research and Quality had no role in the study selection, quality assessment, or evidence synthesis.
Key Questions 1 and 2: Effectiveness of Counseling to Promote Sun-Protective Behaviors
We found no trials meeting our inclusion criteria that directly examined whether behavioral counseling interventions can reduce skin cancer. We included 11 fair-quality, randomized, controlled trials (RCTs) examining counseling interventions that included sun-protective behavior outcomes (Table 2). Quality considerations for these trials are summarized in Table 2.
In adults, 1 trial was conducted in the United Kingdom (41) and 4 trials were conducted in the United States (42-45). All of the trials used tailored risk feedback to promote sun-protective behaviors. Three of the counseling interventions conducted in the United States were coupled with in-office computer support on the basis of the transtheoretical model to generate printed stage-based tailored feedback (43-45). The trial conducted in the United Kingdom used a self-directed computer station in primary care practice to deliver the counseling intervention (41). Populations studied included predominantly middle-aged white men and women. Interventions ranged from a single 15-minute self-directed session to several sessions with in-person counseling, phone counseling, or written assessments followed by tailored written feedback. Overall, 4 of 5 trials (6949 participants) showed that primary care-relevant counseling with tailored feedback (with or without computer support) can modestly affect self-reported sun-protective behaviors, as measured by composite behavior scores (Table 2) (42-44). The differences in these scores, although statistically significant, were small, and it is unclear whether these differences translate into clinically meaningful behavior change. In the 1 trial (724 participants) that also reported individual types of behavior change, only the change in use of sunglasses was statistically significant (Table 2) (42). One trial conducted among siblings of patients with melanoma, which evaluated a similar counseling intervention, did not show any statistically significant changes in sun-protective behaviors (Table 2) (45). This trial, however, used different outcome measures than the other trials and had only 64% follow-up at 12 months.
Four trials in young adults were conducted in university settings and used "appearance-based" behavioral interventions that emphasized the effects of photoaging effects of ultraviolet exposure and norms about tanning and appearing tan instead of a primarily "health-based" message about skin cancer prevention (46-49). Interventions ranged from a written self-guided booklet to a brief video and to a 30-minute 1:1 peer-counseling session. In 3 trials (897 participants), the appearance-focused counseling intervention successfully reduced indoor tanning among women who had the intention to tan indoors (Table 2) (46, 48, 49). Although the interventions decreased indoor tanning behavior by up to 35% (46), follow-up for these trials was only 3 to 6 months. In another RCT (133 participants), a brief video intervention with or without an ultraviolet facial photograph produced a moderate decrease in objectively measured skin pigmentation (using skin reflectance spectrophotometry) at 12 months (Table 2) (47). The change in pigmentation was judged "moderate" on the basis of the Cohen d statistic.
In children, we found only 2 trials (50, 51). Participants in both trials were predominantly white. In 1 trial (819 participants), young adolescents randomly assigned to brief counseling by their primary care providers, coupled with in-office computer support to generate printed tailored feedback, reported both higher composite sun-protection scores and a greater likelihood of avoiding or limiting midday sun exposure or using sunscreen on the face or sun-exposed areas at 24 months than the attention control group (Table 2). The other cluster RCT, conducted in a large managed care organization, integrated counseling into 4 sequential well-child visits at the discretion of the primary care provider (51). Parents of newborns (728 participants) in practices randomly assigned to receive the intervention reported higher composite sun-protection scores at 36-month follow-up than those in control practices (Table 2). The clinical significance of these higher scores, however, is unclear, given the very small numerical differences and the lack of statistically significant differences in 6 of 7 sun-protection questions that contribute to the composite score.
Key Question 4: Association Between Sun Exposure, Sunscreen Use, and Indoor Tanning and Skin Cancer
Sixty articles representing 35 unique fair- or good-quality studies evaluated the epidemiologic association between sun exposure, indoor tanning, or sunscreen use and skin cancer (Table 3 and Appendix Table 2). We found only 1 good-quality trial, the Nambour Skin Cancer and Actinic Eye Disease Prevention Trial (The Nambour Trial) (53, 85, 87-89); 6 fair- or good-quality cohort studies (52, 54-56, 64, 86); and 28 fair- or good-quality, population-based, case-control studies (31, 57-63, 65-84), 3 of which were nested case-control studies (57, 73, 75). Odds ratios (ORs) and risk ratios provide a general estimate of the magnitude of the association between the highest- and lowest-risk groups. The ORs and risk ratios, however, should not be compared between studies because the studies used very different measures of exposures and choice of reference groups. Although measures of sun exposure varied greatly among studies, they can be generally categorized as intermittent, which includes measures of recreational sun exposure; chronic, which includes occupational measures of sun exposure; or total, which are cumulative estimates of sun exposure. This section for key question 4 includes a higher-level synthesis of results (Table 3) and a summary of the major limitations of these results; interested readers may refer to Appendix Table 2 for individual study details with outcome data.
On the basis of 5 fair- or good-quality cohort studies and 7 fair- or good-quality case-control studies, increasing intermittent sun exposure in childhood and during one's lifetime is associated with an increased risk for both squamous cell carcinoma and basal cell carcinoma (range of ORs, 1.27 to 3.86) (Appendix Table 2) (52-63). The evidence is more consistent for intermittent sun exposure in childhood leading to an increased risk for squamous cell carcinoma and basal cell carcinoma than in adulthood (52, 58, 60, 62). Although few studies examined the association between total (or cumulative) and chronic (or occupational) sun exposure, most existing studies did not suggest a strong association between total or chronic sun exposure and squamous cell carcinoma or basal cell carcinoma (Appendix Table 2) (53, 54, 58, 59, 61, 62).
On the basis of 1 fair-quality cohort study and 13 fair-quality case-control studies, it seems that increasing intermittent sun exposure is generally associated with an increased risk for melanoma (Appendix Table 2). A large, fair-quality cohort study from Norway and Sweden showed a statistically significant trend between frequency of sunbathing vacations (childhood and adulthood) and the risk for melanoma (64). Of the 8 case-control studies that examined lifetime recreational sun exposure (31, 57, 65, 66, 69, 70, 72, 76), 5 studies showed that increasing total recreational sun exposure was associated with melanoma risk (range of ORs, 1.3 to 5.0) (57, 65, 66, 69, 70). Three of 4 case-control studies that examined recreational sun exposure during childhood suggest that increasing sun-bathing behavior in childhood is associated with an increased risk for melanoma (range of ORs, 1.7 to 3.5) (70, 71, 73, 75). On the basis of fair-quality case-control studies, it seems that both total and chronic sun exposure are not as strongly associated with melanoma. Six case-control studies included some measure of total sun exposure, either during childhood, during the recent past, or over the lifetime (Appendix Table 2) (65, 67, 69, 79-81). These studies showed mixed results: two studies found a statistically significant association between total lifetime sun exposure and melanoma (65, 81) and 4 did not (67, 69, 79, 80). All 3 studies that examined total sun exposure during childhood, however, showed a statistically significant association between increasing sun exposure and melanoma (range of ORs, 1.81 to 4.4) (67, 79, 81). Nine case-control studies included some measure of chronic or occupational sun exposure (Appendix Table 2) (65, 66, 68, 69, 71, 77-80). Three of these studies suggest that occupational sun exposure is associated with an increased risk for melanoma. These studies, however, used crude measures of occupational sun exposure (66, 77, 78), and 1 study showed an increased risk only with the highest level of occupational exposure (>20 years' exposure) (78). In contrast, 5 of the remaining 6 studies suggest that occupational sun exposure is inversely associated with melanoma risk (65, 68, 69, 79, 80).
Five fair-quality case-control studies examined the association between indoor tanning and the risk for squamous cell carcinoma or basal cell carcinoma (Appendix Table 2) (57, 59, 61, 62, 82). Four of 5 studies used only a crude dichotomous measure of indoor tanning, and none of these studies found a statistically significant association between ever and never use (57, 59, 61, 62). Three studies adjusted for both skin phenotype and sun exposure (57, 61, 62). One fair-quality case-control study that was larger and had a slightly higher proportion of exposed persons showed a statistically significant association between indoor tanning and risk for squamous cell carcinoma and basal cell carcinoma, with greater risk for persons who reported early first use (before age 20 years). This study, however, did not adjust for sun exposure (82).
We found 1 fair-quality cohort study and 11 fair-quality case-control studies that examined the association between indoor tanning and melanoma (Appendix Table 2) (31, 57, 64, 66, 68, 72-74, 76, 83, 84, 90). Most studies used crude measures of indoor tanning exposure. The Norwegian-Swedish Women's Lifestyle and Health Cohort Study found that women who reported regular solarium use (≥1 time per month over 2 or 3 decades) from age 10 to 39 years had an increased risk for melanoma (risk ratio, 2.37 [95% CI, 1.37 to 4.08]) after adjustment for important confounders, including skin phenotype and intermittent sun exposure (64). Six of 11 case-control studies did not find a statistically significant association between ever or never use of indoor tanning and melanoma (Appendix Table 2) (66, 68, 72, 73, 84, 90). Only 1 of 6 negative studies adjusted for both skin phenotype and some measure of sun exposure (90). Of the 4 studies that found a statistically significant association between indoor tanning exposure and melanoma, 2 adjusted for both skin phenotype and some measure of sun exposure (57, 76) and 1 adjusted only for skin phenotype (74). These studies suggest that regular or higher frequency of indoor tanning or use at a younger age may increase risk for melanoma. Only 1 study examined sun lamp (older technology) and tanning bed (newer technology) exposure separately. Although only frequent sun lamp use was associated with increased melanoma risk, study investigators caution that sufficient lag time may not have elapsed to assess a potential effect, given the more recent use of tanning beds (83).
We found 1 RCT (1621 participants) examining whether regular sunscreen use can prevent squamous cell carcinoma or basal cell carcinoma (85, 87, 88). After 8 years of follow-up, persons randomly assigned to regular sunscreen use had a decreased risk for squamous cell carcinoma (risk ratio, 0.65 [CI, 0.45 to 0.94]) but not basal cell carcinoma (risk ratio, 1.02 [CI 0.78 to 1.35]). Two fair-quality cohort studies from the Nurses' Health Study did not show a decrease in squamous cell carcinoma or basal cell carcinoma risk with sunscreen use after adjusting for skin phenotype and sun exposure (Appendix Table 2) (55, 56). Both of these studies, however, used only a crude dichotomous measure of sunscreen use. Although 2 fair-quality case-control studies suggest a protective effect of sunscreen for basal cell carcinoma, both used crude measures of sunscreen use and neither adjusted for sun exposure (Appendix Table 2) (58, 59).
On the basis of 1 fair-quality cohort and 4 fair-quality case-control studies, sunscreen use has no clear protective or harmful effect on the risk for melanoma (Appendix Table 2) (31, 66, 66, 76, 86). One cohort and 1 case-control study found no significant association between a crude dichotomous measure of sunscreen use and risk for melanoma (66, 86). One study found a protective effect for women who reported always using sunscreen compared with those who reported sometimes or never using sunscreen. This study adjusted for skin phenotype and sunburn, but not sun exposure (68). Two studies conducted in Sweden found a statistically significant harmful effect of sunscreen, such that persons who reported always or almost always using sunscreen were at increased risk for melanoma, after adjustment for both skin phenotype and sun exposure (31, 76).
Study Heterogeneity and Methodological Limitations
This body of epidemiologic evidence examining sun exposure, indoor tanning, and sunscreen use has several important limitations. There was great heterogeneity in the actual measurement of sun exposure among studies, the categorization of levels of exposure, and in choice of reference groups. Sun-exposure measurements used different definitions and assessment methods and often covered different periods of a person's life. Measurement of sunscreen rarely included important details, such as sun protection factor, amount, frequency and duration, and years because sunscreens have changed over time. Likewise, measurement of indoor tanning rarely included important details, such as rationale or motivation of use, frequency and duration, and years because indoor tanning devices have also changed over time. Adjustment for important confounders and stratification to examine effect modification also varied across studies. Studies examining sun exposure generally adjusted for age, sex, and some measure of skin phenotype or sun sensitivity. Several studies examining indoor tanning and sunscreen use did not adjust for sun exposure. Some studies also may have overadjusted for confounding, such as adjustment for nevi, freckling, or sunburn history, because these are probably intermediate steps in carcinogenesis or surrogates for sun exposure. Finally, only 4 studies presented results stratified by skin phenotype; these studies suggest an interaction between skin phenotype and skin cancer (25, 57, 75, 91). Therefore, simply adjusting for skin type as a confounder in logistic regression may be insufficient to illuminate the effect of sun exposure in at-risk populations (for example, poor tanners). Lack of adequate adjustment and lack of stratification for skin phenotype may explain the lack of association seen in some studies or inverse association reported with occupational sun exposure.
Key Questions 3 and 5: Potential Harms of Sun-Protective Behaviors
On the basis of the trials included in key questions 1 and 2, we found no evidence for harms of counseling to prevent skin cancer. In addition, we found 17 fair- or good-quality studies that directly examined the potential harms of sun-protective behaviors (Table 4) (92-107). Overall validity concerns are summarized in Table 4. One fair-quality trial that examined whether adherence to sun-protective behaviors in children reduces physical activity found no difference in body mass index or self-reported time spent outdoors at long-term follow-up between children receiving sun-protection curricula versus standard health-education curricula in schools (92). This finding is consistent with 1 of the included counseling trials that found no difference in self-reported measures of physical activity (50, 106, 107). Six fair- or good-quality trials examined whether sunscreen use leads to increased sun exposure (93-95, 108-110). These RCTs suggest that sunscreen with a higher sun protection factor may increase intentional sun exposure in healthy student volunteers on vacation. Sunscreen use in general, however, does not promote increased sun exposure. Three fair-quality studies examined the effect of sun exposure or sunscreen use on vitamin D levels (96, 97, 111). One small, fair-quality trial showed that sunscreen use during the summer did not significantly decrease vitamin D levels or cause vitamin D deficiency (96). Two fair-quality cohort studies demonstrated that vitamin D levels were influenced by sun exposure, such that post- or perimenopausal women living at high altitudes were at risk for transient vitamin D deficiency during winter months (97, 111).
It is hypothesized that sun exposure may be protective against some types of cancer through vitamin D production. Seven fair- or good-quality studies examined the relationship between sun exposure and risk for nonskin cancer (98, 99, 101-105). On the basis of a sparse body of fair- or good-quality cohort and case-control studies, it seems that sun exposure in lighter pigmented persons may be inversely related to risk for advanced breast and prostate cancer after adjustment for well-established risk factors and that intermittent sun exposure may be inversely related to risk for non-Hodgkin lymphoma (98, 100-103, 105). None of these studies, however, directly measured vitamin D status.