JOURNAL OF COSMETIC SCIENCE 192 affected. The most straightforward way to prevent skin cancer is to avoid intense sun exposure of the skin. Measures include staying indoors in the middle part of the day when ambient UV levels peak or seeking shade and wearing protective clothing, such as hats and long sleeves, when outdoors (3). The application of sunscreen that shields the skin by refl ecting or fi ltering UV radiation is another popular measure. This article reviews the available evidence regarding the use of sunscreens for skin cancer prevention. EVIDENCE FROM OBSERVATIONAL STUDIES One of the most extensive reviews of relevant observational studies was a recent system- atic review of the evidence in humans (available to February 2018) regarding the use of sunscreens for melanoma prevention (4). The authors identifi ed 23 case–control, one eco- logical, and three cohort studies, and their evidence was deemed weak and heterogeneous because of the challenges of controlling for “confounding by indication” (4). This con- founding occurs because factors such as sunburn susceptibility, high occupational or rec- reational sun exposure, or previous skin cancer determine both sunscreen use and skin cancer risk (5), so a priori people who use sunscreen are naturally the people most likely to develop skin cancer, and this cannot be adjusted for in analysis of observational studies. Randomized controlled trials (RCTs) are the only kind of study that avoid this confound- ing through the random allocation of sunscreen use irrespective of the risk of skin cancer and, therefore, the only kind of study that can properly evaluate the long-term effective- ness of sunscreen in preventing skin cancer (5). EVIDENCE FROM RCTS There has been only one RCT that has assessed the effects of using sunscreen on skin cancer occurrence. It was conducted in Nambour, a township in subtropical Queensland, Australia (6). The participants were initially selected at random from a community reg- ister (the electoral roll: voting is compulsory in Australia) in 1986 when they were aged 20–69 years, for a study of skin cancer prevalence followed by incidence studies (7). Of the 2,095 people in the original prevalence study, 1,621 consented to take part in the fi eld trial that evaluated daily application of sunscreen (and daily oral supplements of beta-carotene) to prevent skin cancer, from 1992 to 1996 (6). At baseline in 1992, they completed self-administered questionnaires recording personal characteristics such as education, smoking, medical conditions, and medications, and skin cancer risk factors, namely, skin color, sunburn tendency and past sunburns, and occupational (weekday) and recreational (weekend and holidays) sun exposure (6). All received full-body skin exami- nations for skin cancer and other signs of sun damage by dermatologists at trial baseline. Using a computer-generated random assignment sequence, trial participants were allo- cated to one of four intervention groups: daily application of a standard sunscreen (details in the following paragraph) and a 30-mg beta-carotene tablet each day, daily sunscreen and placebo tablet, beta-carotene only, and placebo only (8). (The beta-carotene interven- tion had no effect on skin cancer and is not discussed further.) Participants not assigned to daily application of sunscreen were not assigned placebo sunscreen for ethical reasons and were asked to continue the use of sunscreen at their discretion, which was mostly recreational use or no use.

SUNSCREENS AND SKIN CANCER PREVENTION 193 Those allocated to daily sunscreen received unlimited free supplies of the study sunscreen, a water-resistant, broad-spectrum, sun protection factor (SPF) 16 sunscreen (Ross Cosmet- ics, Melbourne, Australia). They were asked to apply it every morning to all exposed skin on the head, neck, arms, and hands, with reapplication after heavy sweating, bathing, or prolonged sun exposure. To estimate compliance with trial protocol, participants answered questionnaires in 1994 and at the end of the trial about frequency of sunscreen use on average each week and attended study clinics quarterly when they returned sunscreen bottles for weighing. Incident skin cancers were monitored in several ways to ensure complete capture. At the quarterly study clinics, participants reported new skin cancers they carried wallet-sized treatment cards which doctors completed if a skin cancer was treated, and they had full skin examinations by dermatologists in 1994 and 1996. Histological confi rmation was sought for all clinically diagnosed skin cancers. The Nambour Trial ceased in 1996, and the participants were followed up for another decade with regular questionnaires about habits of sun exposure and protection and about treatment of new skin cancers, all of which were confi rmed by review of medical records. All major regional and state pathology laboratories provided pathology reports for any skin cancers diagnosed among trial participants resulting in virtually complete ascertain- ment of all skin cancers confi rmed histologically (9,10). A priori, all new cancers diag- nosed in the fi rst year of the trial were excluded from analyses of the preventive effect of sunscreen (8). At the end of the trial, 75% of those allocated to daily sunscreen use were regular users (defi ned as applying sunscreen 3–4 d per week), and 74% of those allocated to discre- tionary sunscreen were either not using it at all or applying it at most 1–2 d per week (8). Sun exposure among those in the daily and discretionary sunscreen groups remained similar throughout the trial, as shown by measured UVB radiation exposure in a sub- sample, and by 79% and 77%, respectively, reporting that in the previous summer, they had spent 50% of their time outdoors in the sun on weekends at the trial’s end propor- tions of hat-wearing and shade-seeking people were also similar in each treatment group (11). After trial cessation, a large proportion of those allocated to daily sunscreen use continued to apply sunscreen to their skin regularly: 35% of pretrial regular users and 20% of those who were irregular or never users before the trial (12). B CC The incidence of BCC was not reduced by daily sunscreen use during the trial period [rate ratio (RR) = 1.03] (8) or at the end of the follow-up in 2004 (RR = 1.02) compared with controls (9). However, there was evidence that sunscreen application delayed the appear- ance of subsequent BCCs in those who developed multiple BCCs during follow-up (13). S CC At the end of the trial, new SCC tumors diagnosed clinically or histologically were re- duced by 40% in the intervention group (RR = 0.61 95% confi dence interval (CI) 0.46– 0.81) and by more than 50% based on only histologically confi rmed SCCs (RR = 0.48 95% CI 0.35–0.64) (8). By 2004, incidence rates of SCC were signifi cantly reduced in