Objective:

We conducted a systematic evidence review to support the U.S. Preventive Services Task Force (USPSTF) in updating their recommendations on behavioral counseling for skin cancer primary prevention and on secondary prevention with skin self-exam. Our review addresses the following Key Questions (KQs):

1. Does counseling patients in skin cancer prevention improve a) intermediate outcomes (sunburn or precursor lesions) or b) skin cancer outcomes (melanoma, squamous cell, or basal cell carcinoma incidence, morbidity, or mortality)?
2. Do primary care-relevant counseling interventions improve skin cancer prevention behaviors?
3. What are the harms of counseling interventions for skin cancer prevention?
4. What is the association between skin self-examination and skin cancer outcomes (melanoma, squamous cell, or basal cell carcinoma incidence, morbidity, or mortality)?
5. What are the harms of skin self-examination?

Data Sources:

We searched Cochrane Central Register of Controlled Trials, Medline, and PubMed to locate relevant studies for all KQs. For counseling on sun protection behaviors we searched for articles published from 2009 to March 31, 2016. For skin self-exam we searched for articles published from August 2005 to March 31, 2016. We supplemented our database searches by reviewing reference lists from recent and relevant systematic reviews. We also searched ClinicalTrials.gov and WHO International Clinical Trials Registry Platform (ICTRP), for relevant ongoing behavioral intervention trials. We updated our search on June 7, 2017.

Study Selection:

We reviewed 2,100 abstracts and 355 articles against specified inclusion criteria. Eligible studies included those written in English and conducted in people of any age or in caregivers of younger children, conducted in settings affiliated with primary care. Intermediate outcomes of interest were sunburn, nevi, and actinic keratosis; health outcomes included melanoma, basal cell or squamous cell carcinoma incidence, morbidity or mortality. Behavioral outcomes of interest were sun protection behaviors (e.g., composite scores, use of protective clothing, sun avoidance, use of sunscreen), skin self-exam, or indoor tanning use. Any harm of behavioral counseling intervention was of interest. We conducted dual, independent critical appraisal of all provisionally included studies and abstracted all important study details and results from fair- and good-quality studies.

Data Analysis:

Data were independently abstracted by one reviewer and confirmed by another. We synthesized the results for health outcomes and adverse events for pediatric and adult populations separately. The data did not allow for quantitative pooling due to the limited number of contributing studies and the variability of the outcomes measured. For sun protection and skin self-exam behavioral outcomes, we present forest plots showing the standardized mean differences in change between groups (using the Cohen's d statistic) to illustrate the range of effects seen across studies, but we do not provide pooled estimates. We summarized the overall strength of evidence for each KQ for child/adolescent populations and adult/young adult populations.

Results:

We included 21 trials that reported the impact of primary care-relevant behavioral interventions on skin cancer outcomes, sunburn, and sun protection behaviors. Six trials were conducted among child or adolescent populations (n=4252); 16 trials reported data in adult populations (n=16,309), and three of those were conducted exclusively in young adults (n=1528).

Intermediate and health outcomes. None of the six trials among children and adolescents reported skin cancer outcomes (KQ1). Three trials assessing parent-reported sunburn outcomes in children ages 3–10 generally found no intervention effect. A trial among six-year-olds (n=867) found a small intervention effect on nonsevere sunburn (effect size, -0.25 [95%, CI -0.47 to -0.04], p=0.02); but no effect on severe, blistering sunburn at 3 years. This same trial found no difference between the mean number of small or large nevi between intervention and control group children at 3-year followup.

One trial of six in adult populations found an intervention effect for sunburn outcomes. In a trial of online education for young adults (n=965, 86% fair skin), the proportion of participants reporting red/painful sunburn in the past month decreased more markedly from baseline to 3 months in the intervention group compared to two other groups (54.5% to 26.3% in the intervention group; 51.5% to 38.2% in the public website group; 56.3% to 41.2% in the assessment-only control group, p=0.014 for intervention-assessment only comparison). One trial (n=1356) assessed skin cancer outcomes at 12 months after a skin self-exam intervention, and found no difference in numbers of cancers and atypical nevi detected in intervention and control groups.

Behavioral outcomes. All six trials among children and adolescents reported the effect of interventions on composite sun protection behaviors; five of the six trials found a statistically significant benefit on parent-reported composite sun protection scores compared with controls at 3-month to 3-year followup. Standardized mean differences (Cohen's d) effect sizes ranged from 0 to 0.96, with the three larger trials suggesting small to moderate effect ranging from 0.16 to 0.50 (average around 0.32). Effects on sunscreen use and other individual sun protection behaviors were generally consistent within each trial, and there were no apparent trends in the effectiveness of the interventions according to intervention or population characteristics.

In 12 trials reporting sun protection behaviors among adults, evidence was mixed. One trial among young adults and five trials among adult populations found increases in sun protection composite measures compared to control groups. Standardized effect sizes ranged from -0.46 (favoring control group) to 0.57 (favoring the intervention group), and between 0.10 and 0.20 for most studies. Sunscreen use was the most commonly reported individual behavior. Only one in three trials found a significant change in self-reported indoor tanning behavior, a trial of an appearance-focused intervention among young adult female indoor tanners found an attenuated increase in mean number of indoor tanning sessions from baseline to 6 months in the intervention group (mean 4.67 to 6.8 sessions in previous 3 months) compared to a larger increase (mean 4.48 to 10.9 sessions) in the control group (p<0.001). We found no consistent patterns of intervention effectiveness by age or by intervention component, though trials of longer duration or more contacts with participants tended to find intervention effects. Evidence for skin self-exam was more consistent, with 9 of 11 trials finding significant increases in self-reported skin self-exam compared to control conditions. Odds ratios for skin exam in intervention groups compared to control groups ranged from 1.16 to 2.64.

Harms (adults only). No harms were assessed in trials of children or adolescents. Only two trials reported harms in adult trials. One trial focused on skin self-exam (n=1356) found that more intervention group participants reported a skin procedure compared to the attention-control group between 0 and 6 months (8.0% vs. 3.6%, p=0.0005). However, between 6 and 12 months, the proportions were similar between groups: (3.9% and 3.3%, not significant [NS]).

In one study of single-session primary care provider counseling with risk assessment and feedback compared to no intervention (n=217), a slightly higher proportion of adults in the intervention group versus control group reported worrying about developing melanoma, but this difference was not significant (28.9% vs. 18.4%, p=0.16).

No trials met our inclusion criteria for KQ4, on the association between skin self-exam and skin cancer outcomes, or for KQ5, on the harms of skin self-exam.

Limitations:

Trials of behavioral interventions used self-reported outcomes, which are subject to bias. The clinical relevance of incremental changes in composite measures of sun protection behaviors is difficult to assess. There were no new studies among children aged 0-3 or adolescents, and few studies among young adults. Skin cancer outcomes were reported only in a single study focused on skin self-exam.

Conclusions:

The body of evidence on the impact of behavioral interventions has increased substantially since the previous review, and generally reaffirms its findings, adding new but limited evidence on intermediate and health outcomes and for behavioral outcomes in children aged 3–10. The current fair-to-good evidence base suggests that behavioral interventions can increase sun protection behavior with few harms in both pediatric and, less consistently, in adult populations; but there is no consistent evidence that interventions are associated with improved sunburn frequency in children or adults. Interventions can increase skin self-exam in adults relative to control conditions, but may also lead to increased skin procedures without detecting additional atypical nevi or skin cancers.