Sun protection for preventing basal cell and squamous cell skin cancers
Abstract
Background
'Keratinocyte cancer' is now the preferred term for the most commonly identified skin cancers basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC), which were previously commonly categorised as non‐melanoma skin cancers (NMSC). Keratinocyte cancer (KC) represents about 95% of malignant skin tumours. Lifestyle changes have led to increased exposure to the sun, which has, in turn, led to a significant increase of new cases of KC, with a worldwide annual incidence of between 3% and 8%. The successful use of preventive measures could mean a significant reduction in the resources used by health systems, compared with the high cost of the treatment of these conditions. At present, there is no information about the quality of the evidence for the use of these sun protection strategies with an assessment of their benefits and risks.
Objectives
To assess the effects of sun protection strategies (i.e. sunscreen and barrier methods) for preventing keratinocyte cancer (that is, basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) of the skin) in the general population.
Search methods
We searched the following databases up to May 2016: the Cochrane Skin Group Specialised Register, CENTRAL, MEDLINE, Embase, and LILACS. We also searched five trial registries and the bibliographies of included studies for further references to relevant trials.
Selection criteria
We included randomised controlled clinical trials (RCTs) of preventive strategies for keratinocyte cancer, such as physical barriers and sunscreens, in the general population (children and adults), which may provide information about benefits and adverse events related to the use of solar protection measures. We did not include trials focused on educational strategies to prevent KC or preventive strategies in high‐risk groups. Our prespecified primary outcomes were BCC or cSCC confirmed clinically or by histopathology at any follow‐up and adverse events.
Data collection and analysis
Two review authors independently selected studies for eligibility using Early Review Organizing Software (EROS). Similarly, two review authors independently used predesigned data collection forms to extract information from the original study reports about the participants, methods of randomisation, blinding, comparisons of interest, number of participants originally randomised by arm, follow‐up losses, and outcomes, and they assessed the risk of bias. We resolved any disagreement by consulting a third author and contacted trial investigators of identified trials to obtain additional information. We used standard methodological procedures expected by Cochrane.
Main results
We included one RCT (factorial design) that randomised 1621 participants.
This study compared the daily application of sunscreen compared with discretionary use of sunscreen, with or without beta‐carotene administration, in the general population. The study was undertaken in Australia; 55.2% of participants had fair skin, and they were monitored for 4.5 years for new cases of BCC or cSCC assessed by histopathology. We found this study to be at low risk of bias for domains such as allocation, blinding, and incomplete outcome data. However, we found multiple unclear risks related to other biases, including an unclear assessment of possible interactions between the effects of the different interventions evaluated (that is, sunscreen and beta‐carotene). We found no difference in terms of the number of participants developing BCC (n = 1621; risk ratio (RR) 1.03, 95% confidence interval (CI) 0.74 to 1.43) or cSCC (n = 1621; RR 0.88, 95% CI 0.50 to 1.54) when comparing daily application of sunscreen with discretionary use, even when analyses were restricted to groups without beta‐carotene supplementation. This evidence was of low quality, which means that there is some certainty that future studies may alter our confidence in this evidence.
We reported adverse events in a narrative way and included skin irritation or contact allergy.
We identified no studies that evaluated other sun protection measures, such as the use of sun‐protective clothing, sunglasses, or hats, or seeking the shade when outdoors.
Authors' conclusions
In this review, we assessed the effect of solar protection in preventing the occurrence of new cases of keratinocyte cancer. We only found one study that was suitable for inclusion. This was a study of sunscreens, so we were unable to assess any other forms of sun protection. The study addressed our prespecified primary outcomes, but not most of our secondary outcomes. We were unable to demonstrate from the available evidence whether sunscreen was effective for the prevention of basal cell carcinoma (BCC) or cutaneous squamous cell carcinoma (cSCC).
Our certainty in the evidence was low because there was a lack of histopathological confirmation of BCC or cSCC in a significant percentage of cases. Amongst other sources of bias, it was not clear whether the study authors had assessed any interaction effects between the sunscreen and beta‐carotene interventions. We think that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
PICOs
Plain language summary
Sun protection (including sunscreens) to prevent basal cell carcinoma and cutaneous squamous cell carcinoma of the skin
What is the aim of this review?
The aim of this Cochrane Review was to find out if using topical sunscreen and physical barrier methods (such as sun‐protective clothing, hats, sunglasses, and the active search for shade when outdoors) compared with no specific precautionary activity prevented the development of basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) of the skin in adults and children.
What was studied in this review?
Keratinocyte cancer (BCC and cSCC of the skin) is the most commonly identified type of skin cancer. The main risk is exposure to ultraviolet radiation, which is a component of sunlight. Prevention has become an important way to manage this cancer, so it is important to assess the effectiveness of methods used to prevent keratinocyte cancer in the general population. In this review, we assessed the effects of using topical sunscreen and physical barrier methods (such as sun‐protective clothing, hats, sunglasses, and the active search for shade when outdoors) compared with no specific precautionary interventions aimed at preventing the development of BCC and cSCC in adults and children.
We searched the medical literature up to May 2016 for randomised controlled trials that evaluated preventive strategies. We found only one study suitable for inclusion. This study compared the daily application of sunscreen (with or without beta‐carotene, which is a precursor of vitamin A) compared with the occasional use of sunscreen (with or without beta‐carotene) in the general population, without restriction by gender or age. The study was undertaken in Australia, where 1621 participants, 55% of them with fair skin, were monitored for 4.5 years for new cases of BCC or cSCC assessed by histopathology (which is a method used to detect cancerous cells under the microscope).
What are the main results of this review?
We found no difference between the number of people who developed BCC or cSCC in the two groups over the time period of the trial. So, there did not seem to be a difference in applying sunscreen daily compared with using it occasionally.
Key messages
Our one included study was a study of sunscreens, so we were unable to assess any other forms of sun protection.
We identified no studies that evaluated other sun protection measures, such as the use of sun‐protective clothing, sunglasses, or hats, or seeking the shade when outdoors.
We did not find evidence for the effectiveness of daily sunscreen for preventing BCC or cSCC compared with the occasional use of sunscreen. The certainty of the evidence was low, which means that future studies may alter this result.
Side effects from the sunscreen used with or without the addition of beta‐carotene included a low percentage of cases of contact allergy and skin irritation.
How up to date is this review?
This review included studies identified up to May 2016.
Authors' conclusions
Summary of findings
| Daily application of sunscreen compared with discretionary use (including beta‐carotene use) for preventing basal cell and cutaneous squamous cell skin cancers | ||||||
| Patient or population: general participants | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | Number of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Discretionary use (including beta‐carotene use) | Daily application of sunscreen (including beta‐carotene use) | |||||
| Basal cell carcinoma | 78 per 1000 | 80 per 1000 (58 to 111) | RR 1.03 (95% CI 0.74 to 1.43) | 1621 | ⊕⊕⊝⊝ | The study authors reported narratively that the incidence was similar, even when analysis was restricted to cases of BCC diagnosed histologically. Also, the study authors reported a similar risk when groups receiving beta‐carotene were excluded (IDR 0.96, 95% CI 0.59 to 1.57). |
| Cutaneous squamous cell carcinoma | 31 per 1000 | 27 per 1000 (15 to 48) | RR 0.88 (95% CI 0.50 to 1.54) | 1621 | ⊕⊕⊝⊝ | The study authors reported an IDR of 0.74 (95% CI 0.39 to 1.38) when the analyses were restricted to histologically diagnosed cSCC. Also, the study authors reported a similar risk when groups receiving beta‐carotene were excluded (IDR 0.74, 95% CI 0.31 to 1.77). |
| Adverse events | See comment | See comment | ‐ | 1621 | ⊕⊕⊝⊝ | This was reported narratively for the daily sunscreen use group only. The most frequent adverse events were contact allergy or skin irritation (25 participants), skin greasiness (10 participants), and interference with perspiration or stinging eyes after facial perspiration (6 participants). |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| ¹Downgraded by one level due to other potential sources of bias, including an unclear assessment of interaction between the effects of the different interventions evaluated (that is, sunscreen and beta‐carotene). | ||||||
Background
Description of the condition
Basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) were previously commonly categorised as non‐melanoma skin cancers (NMSC), but now the preferred term 'keratinocyte cancer' is used (Albert 2003). Throughout the review, we have used this term when we refer to BCC and cSCC collectively.
Keratinocyte cancer (KC) is a term that includes BCC and cSCC of the skin. Together, KC represents about 95% of all skin cancers (Dubas 2013). Since 1960, changes in lifestyle have led to increased exposure to the sun, which has in turn led to a significant increase of new cases of KC, with a worldwide annual incidence of between 3% and 8% (Glass 1989; Ricotti 2009). Around 1.2 million keratinocyte cancers are diagnosed per year in the USA (Ricotti 2009). In Australia, the estimated incidence is 400 cases per 100,000 inhabitants (Johnson 1992; Stern 2010). In Colombia, an equatorial country with a predominant population of phototype III (skin type), a significant increase in KC figures has been documented, rising from 23 cases per 100,000 inhabitants in 2003 to 41 cases per 100,000 in 2007 (Sánchez 2011). In Europe, there is an increasing trend in the incidence of keratinocyte cancer, with exceptional cases such as Switzerland where the highest rate of cSCC has been documented in the whole continent, rising from 14.2 cases per 100,000 inhabitants in 1978 to 28.9 cases per 100,000 inhabitants in 1997 (Lomas 2012).
The action of ultraviolet radiation on the keratinocyte layer of cells in the epidermis is the common site of the development of cSCC and BCC lesions (Preston 1992). The mechanism involves damage to DNA and its repair system, specifically the formation of pyrimidine dimers (the bases that are part of the structure of DNA), as well as mutations of p53 tumour suppressor genes (Preston 1992). Basal cell carcinoma is characterised by a slow rate of growth and extremely low probability of metastasis; however, this condition can compromise wide areas of tissue, cartilage, and bone and produce local damage (Rubin 2005). By contrast, the potential for distant metastasis of cSCC is greater, with up to 5% of lesions metastasising within five years (Miller 2010).
The diagnosis of KC is based on clinical examination of the lesion and confirmation by histopathology (Motley 2003; Telfer 2008), and is classified into two large groups: basal cell carcinomas and cutaneous squamous cell carcinomas (Gloster 1996). The American Cancer Society reports that around 50% to 65% of KCs are classified as BCC, and 20% to 25%, as cSCC (Kornek 2013). Basal cell carcinoma is a tumour of epithelial (outer layer of the skin) origin, which rarely metastasises (ranging from an incidence of 0.0028% to 0.1%) (Glass 1989), but has significant disease burden and considerable costs (Green 2010; Sánchez 2011; Stern 2010). The main risk factor for developing BCC is exposure to ultraviolet radiation (UV) (Corona 2001). Other factors, such as sun exposure during work, skin cancer family history, personal history of sunburn, presence of actinic keratoses, and having skin phototypes I to III, have also been associated with its onset (Lear 1997; Sánchez 2012). By contrast, cSCC is a locally invasive malignant tumour with a great potential to spread to distant parts of the body (Lansbury 2010). Presence of cSCC has been associated with different factors, such as chronic exposure to UV radiation, history of burns and occupational exposure, inherited skin conditions (albinism or xeroderma pigmentosum), and human papilloma virus infection, among other factors (Green 2010; Sánchez 2013).
Description of the intervention
Prevention is an important component in the management of KC and includes measures to avoid and reduce the consequences of exposure to the sun (Kornek 2013). These measures are broadly divided into sunscreens and physical barriers.
Sunscreen agents include all the products designed to reduce contact between the skin and UV radiation (Mulliken 2012). The ability of sunscreens to prevent the damage related to UV radiation is measured by the sun protection factor (SPF) (Mulliken 2012). Briefly, the SPF is defined as the time needed to produce sunburn when the sunscreen is applied to the skin, divided by the time needed to cause sunburn when nothing is applied to the skin (Schalka 2011). The sun protection factor is accepted as the worldwide standard for the assessment of protection against the erythemogenic (reddening) effects of UVB and UVA radiation (Schalka 2011). The effectiveness of a sunscreen is dependent on such characteristics as specific ingredients, general formulation, water‐resistance, time over which the solar filter has been exposed to the sun, and the quantity of sunscreen applied (Saki 2012). Similarly, numerous studies have confirmed that instances of sunburn throughout life increase the risk of developing skin cancer (Sánchez 2012; Sánchez 2013; Zanetti 1996). UVB generates direct changes in DNA (Ravanat 2001). So, sunscreens that block UVB radiation would potentially increase the time of sun exposure without sunburn and, in theory, reduce the risk of developing skin cancer. The level of protection against UVB can be quantified by measuring the MDE (minimum dose of radiation that can produce erythema, which is reddening of the skin) and the SPF. The SPF is the result of the relationship between the minimal erythema dose (MED) with and without protection (MDE with sunscreen/MDE without sunscreen) (Jansen 2013a). Then, the question is to define the SPF that is able to decrease the risk of DNA damage resulting from sunburn and thus prevent the occurrence of new cases of KC.
Physical barriers can also help in preventing and minimising the harmful effects of UV radiation (Gasparro 1998; Latha 2013). Some authors have proposed that such measures could be more effective than sunscreen to prevent skin cancer (Linos 2011). Physical barriers include photoprotection with special clothes made of different materials, which provide protection against both UVB and UVA radiation (Diaz 2013). Polyester is the material with the highest ultraviolet light absorption capacity, while cotton has the lowest capacity (Diaz 2013). Hats can provide protection to the head and neck depending on their size and shape and the materials from which they are made (Klostermann 2013). Sunglasses, depending on their size, shape, and ability to block UV radiation, are effective in protecting the periorbital region, usually on the lower eyelid and inner edge, which are both areas highly exposed to solar radiation (Klostermann 2013). Finally, the active search for shade may be from a physical barrier, such as from a roof or the use of an umbrella when outdoors. These could provide significant protection against UV radiation without modifying exposure to visible light (Burnett 2012; Cooley 2013).
How the intervention might work
Sun protection strategies, such as those previously mentioned, act by blocking or diminishing the contact of UVA and UVB radiation with the skin, thus, avoiding DNA damage and the development of keratinocyte cancer (Sambandan 2011). Although there is a larger proportion of UVA (wavelength = 320 nm to 400 nm) radiation in the solar spectrum, UVB radiation (290 nm to 320 nm) causes most of the acute and chronic biological damage to the skin (Cole 2001). Any strategy suitable for wide implementation should protect against both types of radiation (Latha 2013). An intervention, such as beta‐carotene supplementation, has been evaluated in its role as a protective agent of the skin, for example, in the prevention of sunburn (Kopcke 2008).
Why it is important to do this review
Ultraviolet radiation is a risk factor in the development of KC (Lucas 2006), which has an effect on the quality of life of those with the condition, as well as having serious cost consequences for countries' health systems. The use of preventive measures could mean a significant reduction of the resources used by health systems compared with the high cost that the treatment of KCs and their aftermath entails. At present, there is no information about the evidence for the use of these sun protection strategies with an assessment of their benefits and harms.
Our review aims to systematically assess the effectiveness and related adverse events of these protective elements, providing information on the possible routes for controlling the increasing incidence of keratinocyte cancer in the general population.
The plans for this review were published as a protocol 'Sun protection for preventing basal cell and squamous cell skin cancers' (Sánchez 2014).
Objectives
To assess the effects of sun protection strategies (i.e. sunscreen and barrier methods) for preventing keratinocyte cancer (that is, basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC) of the skin) in the general population.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled clinical trials that may provide information about benefits and adverse events related to the use of solar protection elements, such as physical barriers and sunscreens, as preventive strategies for keratinocyte cancer.
We did not include trials focused on educational strategies to prevent KC (Langbecker 2014), nor preventive strategies in high‐risk groups (for example, people with actinic keratoses, organ transplant recipients, etc.), because there are other Cochrane Reviews published or in development that focus on these specific issues (Bath‐Hextall 2007; Morales‐Sánchez 2016).
Types of participants
We included trials focused on the general population (children and adults) without restriction by gender or age. We did not include trials focusing on special populations (for example, people with actinic keratoses, organ transplant recipients, etc.).
Types of interventions
Studies included one or more of the following interventions versus no intervention, placebo, or other interventions.
-
Sunscreens with any sun protection factor (SPF).
-
Wearing hats, sunglasses, and special clothing outdoors.
-
Staying out of the sun (e.g. use of shady places, roofs, and umbrellas outdoors, etc.).
Types of outcome measures
Primary outcomes
-
Basal cell carcinoma (BCC) confirmed clinically or histopathologically at any follow‐up.
-
Cutaneous squamous cell carcinoma (cSCC) confirmed clinically or histopathologically at any follow‐up.
-
Adverse events (e.g. dermatitis from sunscreens, acne secondary to the use of sunscreens, dermatitis from the use of hats and clothes, vitamin D deficiency from lack of exposure to the sun, etc.) reported by a number of participants or individually.
We have explained our reason for the addition of 'clinical' to our two primary outcomes in the Differences between protocol and review section.
Secondary outcomes
-
Number of self‐reported sunburns or skin lesions, defined by each study, at the end of follow‐up.
-
Actinic or solar keratoses at any follow‐up.
-
Total hours of ultraviolet radiation exposure at the end of follow‐up.
-
Total hours outdoors in peak exposure times at the end of follow‐up.
-
Minimal erythema dose (MED) at the end of follow‐up.
-
Participant's compliance with preventive strategies at the end of the trial.
Search methods for identification of studies
We aimed to identify all relevant randomised controlled trials (RCTs) regardless of language or publication status (published, unpublished, in press, or in progress).
Electronic searches
We searched the following databases up to 10 May 2016:
-
the Cochrane Skin Group Specialised Register using the search strategy in Appendix 1;
-
the Cochrane Central Register of Controlled Trials (CENTRAL) 2016, Issue 5, in the Cochrane Library using the strategy in Appendix 2;
-
MEDLINE via Ovid (from 1946) using the strategy in Appendix 3;
-
Embase via Ovid (from 1974) using the strategy in Appendix 4; and
-
LILACS (Latin American and Caribbean Health Science Information database, from 1982) using the strategy in Appendix 5.
Trial registries
We searched the following trials registries and portals up to 10 May 2016:
-
the metaRegister of Controlled Trials (www.controlled‐trials.com);
-
the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov);
-
the Australian New Zealand Clinical Trials Registry (www.anzctr.org.au);
-
the World Health Organization International Clinical Trials Registry Platform (www.who.int/trialsearch); and
-
the EU Clinical Trials Register (www.clinicaltrialsregister.eu).
We used the terms "non melanoma skin cancer", "basal cell carcinoma", "squamous cell carcinoma", "bowen disease", and "actinic keratosis".
Searching other resources
References from included studies
We checked the bibliographies of included studies for further references to relevant trials.
Adverse effects
We did not perform a separate search for adverse effects of the target intervention, but we examined data on adverse effects from the included study that we identified.
Data collection and analysis
Due to the lack of studies, we were unable to carry out analyses that we had planned in our protocol. We have explained this in the Differences between protocol and review section.
Selection of studies
Two review authors (MO and GS) independently selected studies for eligibility with Early Review Organizing Software (EROS) (Ciapponi 2011; Glujovsky 2010; Glujovsky 2011). We checked the titles and abstracts of all of the retrieved studies to determine if they fulfilled the inclusion criteria previously proposed. We assessed the full texts of identified studies to confirm their final inclusion. We resolved all disagreements by involving a third author (IA‐R). We were not blinded to characteristics such as the authors' names, institutions, or the journal of publication at any stage of the review. We recorded the reasons for our exclusion of potential studies in the 'Characteristics of excluded studies' tables (Higgins 2011).
Data extraction and management
Two review authors (CS and JG) used predesigned data collection forms to retrieve information, such as randomisation methods, blinding of participants and personnel, comparisons, number of participants by arm, and follow‐up losses, from the original study in an independent way (Higgins 2011). We tested this format prior to extended use. We resolved any disagreement by discussion with a third review author (JN). We entered extracted data into Review Manager 5 for further analyses (Review Manager 2012).
Assessment of risk of bias in included studies
Two authors (ARH and IA‐R) used the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions to independently assess the risk of bias of the included trial (Higgins 2011). We took six domains into consideration: random sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessment; incomplete outcome data; selective reporting; and other biases. We resolved any disagreement by discussion with a third author (GS). We assessed the likely magnitude and direction of the bias and whether we considered it likely to impact on the findings (Higgins 2011). We summarised the information in a 'Risk of bias' table, which is an extension of the 'Characteristics of included studies' table. We planned to explore the impact of the level of bias through undertaking sensitivity analyses (see the Sensitivity analysis section), but we did not perform this analysis because of the number of studies included.
Measures of treatment effect
We presented results of dichotomous outcomes (such as BCC and cSCC) as summary risk ratios (RR) with 95% confidence intervals (CIs), as well as the number needed to treat for an additional harmful outcome (NNTH) as an absolute measure of harm, and NNTH as the reciprocal of risk differences (RD) (McQuay 1998). We also planned that for continuous outcomes (such as the number of sunburns and total hours of ultraviolet radiation exposure), we would report the mean difference (MD) with its corresponding 95% CI, but these outcomes were not reported (see Differences between protocol and review section).
Unit of analysis issues
We did not expect to find any unit of analysis issues as we did not expect to find cross‐over studies. In the case of within‐participant trials (e.g. split face) related to the effectiveness of the sunscreen, we planned to report the results in a narrative way and not include this information in planned meta‐analyses. However, we found that information about these kinds of tumours can be expressed as number of participants affected, as well as total number of tumours by groups.
Dealing with missing data
We planned to retrieve levels of attrition of information if possible. We planned to explore, by using sensitivity analyses, the impact of trials with high levels of attrition in the assessment of treatment effects. If possible, we carried out analyses, as far as possible on an intention‐to‐treat (ITT) basis (i.e. we attempted to include in the analyses all randomised participants in the denominator of the assessed groups).
Assessment of heterogeneity
We planned to investigate heterogeneity in the first instance through visual examination of measures of treatment effect forest plots. Main sources of heterogeneity could include skin phototype, duration of follow‐up, and gender and age groups. We planned to evaluate statistical heterogeneity by means of the I² statistic. This statistic estimates the percentage of total variation across included trials due to heterogeneity rather than sampling error (Higgins 2003; Higgins 2011). We planned to explore heterogeneity if the I² statistic was greater than 30%, and in cases where the I² statistic was more than 80%, we did not plan to present pooled results.
Assessment of reporting biases
We planned to create funnel plots of the primary outcomes to provide a visual assessment of reporting bias if at least 10 trials were available (Higgins 2011; Sterne 2011). Also, we planned to use two tests to assess asymmetry in the corresponding funnel plot: the regression asymmetry test (Egger 1997) and the adjusted rank correlation test (Begg 1994). However, we did not perform this analysis because of the number of studies included.
Data synthesis
We planned to summarise the findings using random‐effects models with the DerSimonian and Laird method and carry out statistical analyses using Review Manager 5 (Review Manager 2012). We planned not to present a pooled result if we identified substantial heterogeneity (I² statistic was greater than 80%). Also, we planned to conduct a trial sequential analysis (TSA), which combines an information size calculation (cumulated sample sizes of included trials) for meta‐analysis with the threshold of statistical significance. We planned to conduct a TSA using the TSA software on binary outcomes (Brok 2009; Pogue 1997; Pogue 1998; Thorlund 2009; Wetterslev 2008) and apply trial sequential monitoring boundaries according to a heterogeneity‐adjusted required information size, based on an a priori 10% relative risk reduction (RRR) employing alpha = 0.05 and beta = 0.20 (CTU 2011; Thorlund 2011). However, we did not perform these analyses due to the number of studies included, and we present the results in a narrative way, with figures to illustrate the main information.
Subgroup analysis and investigation of heterogeneity
We planned to undertake subgroup analysis and perform interaction tests to check for subgroup differences where this assessment would be meaningful. For the primary outcomes, we planned to consider subgroup analyses for the following factors, as appropriate:
-
gender;
-
age groups (i.e. less than 18 years, 18 to 40 years, 41 to 60 years, greater than 60 years);
-
skin phototype (i.e. I, II, III, IV, V, VI);
-
hair colour phenotype (i.e. redhead, blonde, black, etc.);
-
eye colour phenotype (i.e. blue, green, black, brown, etc.);
-
duration at follow‐up (less than one year, between one to two years, between two to five years, more than five years); and
-
history of skin cancer or precancerous lesions (personal and family).
However, we did not perform these analyses because of the number of studies included (see Differences between protocol and review section).
Sensitivity analysis
We planned to perform a sensitivity analysis using those trials classified as having a low risk of bias (Higgins 2011) in three core domains: allocation concealment, incomplete outcome data, and blinding of outcome assessment. However, we did not perform this analysis because of the number of studies included (see Differences between protocol and review section).
'Summary of findings' tables
We used the guidelines of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) Working Group (Guyatt 2008) to assess the quality of the evidence related to primary outcomes, and developed a 'Summary of findings' table with the GRADE profiler software. The GRADE system assesses the quality of evidence based on the extent to which users can be confident that an association reflects the item being evaluated (Guyatt 2008). Assessment of the quality of evidence included consideration of the risk of bias, heterogeneity, directness of the evidence, risk of publication bias, and precision of effect estimates (Guyatt 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g).
Results
Description of studies
See Characteristics of included studies and Characteristics of excluded studies.
Results of the search
We identified 482 records from the database searches after removal of duplicates. We excluded 452 records based on titles and abstracts and identified 30 records for further evaluation in full text. We excluded a further 21 records (See Characteristics of excluded studies). We included one trial, Green 1999, with data reported in nine different references (Figure 1). We did not find additional references from trial registers or other sources of information.
Study flow diagram
Included studies
We included one study, which compared daily application of sunscreen versus discretionary use (mostly for recreational use). We did not find studies of children or studies comparing sunscreen at any sun protection factor (SPF) with no intervention or placebo. Also, we did not identify studies about interventions such as wearing hats, sunglasses, or sun‐protective clothing outdoors, or staying out of the sun (use of shady places, roofs, umbrellas, etc.).
Design
We included one study that randomly assigned 1621 people and included 1383 with full skin examination in the analysis (see the Characteristics of included studies section). This study was a randomised 2x2 factorial trial developed in Australia, which evaluated the following four interventions: daily sunscreen and beta‐carotene supplementation (30 mg beta‐carotene tablet each day); daily sunscreen and placebo; discretionary sunscreen and beta‐carotene supplementation (30 mg beta‐carotene tablet each day); and discretionary sunscreen and placebo. The use of placebo cream alone was considered "unethical"; thus, the daily application of sunscreen was compared with the discretionary use of sunscreen. However, the study authors effectively reduced the comparisons to two by combining the groups (see below).
Participants
Adult participants were selected from a population‐based prevalence survey of skin cancer conducted in 1986, with 43.7% being men; the average age was 48.77 years. Fifty‐two per cent of the participants were fair skinned; 18.7% had an occupation that they carried out mainly outdoors; and 27% had a previous diagnosis of skin cancer.
Sample size and setting
One thousand two hundred and one eligible residents of Nambour were randomised to the different groups of this trial. The study authors presented analysis based on 1383 trial participants who had full skin examination during the follow‐up period. The study authors estimated that a minimum of 1600 participants needed to be enrolled to detect a 36% reduction of basal cell carcinoma (BCC) and a 59% reduction of cutaneous squamous cell carcinoma (cSCC).
Interventions
Four groups were evaluated.
-
Participants in group one received instructions for daily use of SPF 16 sunscreen and a 30 mg beta‐carotene tablet to be taken each day.
-
Participants in group two received instructions for daily use of SPF 16 sunscreen and a 30 mg placebo tablet each day.
The study authors added groups one and two together to obtain results for all participants with daily administration of SPF 16 sunscreen (daily sunscreen group: number of participants randomised = 812).
-
Participants in group three received a 30 mg beta‐carotene tablet each day and received instructions to use sunscreen on a discretionary basis.
-
Participants in group four received a 30 mg placebo tablet each day and received instructions to use sunscreen on a discretionary basis.
The study authors added groups three and four together to obtain results for all participants with discretionary administration of SPF 16 sunscreen (non‐daily sunscreen group: number of participants randomised = 809).
One thousand three hundred and eighty‐three participants completed the full follow‐up period of 4.5 years (85.3%). We included the 1621 participants originally allocated to our analysis (intention‐to‐treat analysis). Losses by group ranged from 12.3% to 17.1%. The administration of sunscreen on a daily basis involved the application of a layer of sunscreen to all exposed sites of the head, neck, arms, and hands every morning.
Outcomes
The primary outcome of our sole included study was the number of new cases of BCC and cSCC (Green 1999). Secondary outcomes were change in the prevalence of solar keratosis (as full body counts and per sunscreen site counts), adverse events (main reported complaints), and the degree of photoageing. The researchers assessed the participants' compliance with sunscreen use every three months by comparing the weight of the sunscreen provided with the average rate of consumption for a group of non‐participants. Likewise, the use of sunscreen in the control group was measured through a standard questionnaire delivered annually. Two follow‐up clinics in 1994 (two years after the beginning of the study) and 1996 (at the end of the study) assessed the participants, and all lesions identified at those times were evaluated by histopathology. Local doctors documented the rest of the lesions on prespecified cards.
Excluded studies
We excluded 21 studies for reasons reported in the 'Characteristics of excluded studies' tables after assessment of the full text of the report and most frequently because they did not evaluate the effectiveness of sun protection strategies in the prevention of keratinocyte cancer.
Risk of bias in included studies
We assessed the one included study for risk of bias and reported the judgments for the individual domains in the 'Risk of bias' table (see Figure 2).
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
Allocation
Regarding randomisation issues, Green 1999 declared, "A customized randomization computer program was used to assign all attending the baseline study clinics to one of four treatment groups." Likewise, regarding allocation concealment, they added, "The treatment code was known only to the investigator who generated it and the two people who packaged the tablets for distribution. None of these people had contact with participants." We therefore judged these two domains to be at low risk of bias.
Blinding
Regarding blinding of participants, personnel, and assessment of outcomes, the authors of Green 1999 reported that the treatment code was known only to the investigator who generated it and to the two people who packaged the tablets for distribution. None of these people had contact with participants. Likewise, a single dermatopathologist histologically examined all clinically diagnosed skin cancers during these follow‐up surveys. Dermatologists involved in the study survey were unaware of the treatment allocation. We judged these domains at low risk of bias.
Incomplete outcome data
Regarding attrition bias for our primary outcomes, Green 1999 declared: "At the end of the study in year 5, 238 (15%) participants had withdrawn without a complete skin examination by a dermatologist in the follow‐up period." The percentage of losses to follow‐up was similar for all groups, and it was the same for all primary outcomes (group one: 12.37%; group two: 17.15%; group three: 14.1%; group four: 15.01%). We judged this domain at low risk of bias.
By contrast, the number of participants lost to follow up was bigger for actinic keratosis (ranging from 28% to 36%), and the risk was considered high for this outcome. One thousand three hundred and eighty‐three participants completed the full follow‐up period of 4.5 years, but we included the 1621 participants originally randomised under intention‐to‐treat (ITT) analysis.
Selective reporting
In a secondary reference, Green 1999 reported the association between possible predictors of sunscreen use and frequency of sunscreen application, including general information on variables such as time outdoors on weekdays in summer and burns during the trial, but this information was not reported in a sufficient way to enable analysis. So, we judged this domain to be at unclear risk of bias.
Other potential sources of bias
We identified other potential sources of bias in the study, including an unclear assessment of the interaction between the effects of the different interventions evaluated (that is, sunscreen and beta‐carotene), an unclear impact of multiple posthoc analyses not planned a priori (repeated significance testing), and an unclear impact of clinical versus histological diagnosis of keratinocyte cancer. We did not consider funding as a source of bias as the Public Health Research and Development Committee of the National Health and Medical Research Council of Australia provided funding.
Effects of interventions
We only found information from one study comparing the daily application of sunscreen (in which participants applied a layer of sunscreen themselves every morning to all uncovered areas on the head, neck, arms, and hands, reapplying after substantial sweating, bathing, or extended sun exposure) versus discretionary use (mostly for recreational use).
For the main analyses, Green 1999 included all participants in the trial, including those who received beta‐carotene tablets. They combined groups one and two to obtain results for all participants with daily administration of SPF 16 sunscreen (i.e. those participants who received instructions for daily use of SPF 16 sunscreen and a 30 mg beta‐carotene tablet to be taken each day, plus those participants who received instructions for daily use of SPF 16 sunscreen and a 30 mg placebo tablet each day). These were compared to all participants (groups three and four) with discretionary administration of SPF sunscreen (i.e. those participants who received a 30 mg beta‐carotene tablet each day and received instructions to use sunscreen on a discretionary basis, plus those participants who received a 30 mg placebo tablet each day and received instructions to use sunscreen on a discretionary basis).
For BCC and cSCC, the trial authors provided information for cases confirmed histopathologically in terms of incidence density ratios (IDR); these estimates are provided below as supplementary information. For actinic keratosis (AK), authors combined all lesions on the whole body at each examination (full body counts) and also added all lesions on the sites to which sunscreen was applied (sunscreen site counts). Also, Green 1999 provided supplementary analyses for BCC and cSCC excluding participants assigned to beta‐carotene tablets (i.e. groups one and three). However, there was not enough information about the number of cases per arm to estimate risk ratios (RR). Information about IDR (which was the effect measure reported by the study authors) is provided as supplementary information.
We present the quality of the evidence in summary of findings Table for the main comparison.
Primary outcomes
Basal cell carcinoma (BCC) confirmed clinically or histopathologically at any follow‐up
The incidence of new BCC was similar in the daily‐application group (812 participants randomly assigned) compared with the discretionary‐use group (809 participants randomly assigned) (RR 1.03, 95% confidence interval (CI) 0.74 to 1.43; Analysis 1.1). This evidence was of low quality, which means that there is some certainty that future studies may alter our confidence in this evidence.
It was reported in a narrative way that the incidence was similar, even when authors restricted analysis to cases of BCC "that had been diagnosed histologically". The study authors also provided information in terms of IDR when they excluded groups receiving beta‐carotene (that is, the authors excluded groups one and three), but they did not find differences between groups (IDR 0.96, 95% CI 0.59 to 1.56).
Cutaneous squamous cell carcinoma (cSCC) confirmed clinically or histopathologically at any follow‐up
The incidence of new cSCC was similar in the daily‐application group (812 participants randomly assigned) compared with the discretionary‐use group (809 participants randomly assigned) (RR 0.88, 95% CI 0.50 to 1.54; Analysis 1.2). This evidence was of low quality, which means that there is some certainty that future studies may alter our confidence in this evidence.
Likewise, the incidence rate of histologically diagnosed cSCC, based on 40 participants with pathological confirmation, was similar in both groups (IDR 0.74, 95% CI 0.39 to 1.38). The study authors also provided information in terms of IDR when they excluded groups receiving beta‐carotene (that is, the authors excluded groups one and three), but they found no differences between groups (IDR 0.74, 95% CI 0.31 to 1.77).
Adverse events
In a narrative report, Green 1999 stated that the main complaints made by the daily sunscreen use group were related to skin irritation or contact allergy (25 participants out of 812) and skin oiliness (10 participants out of 812). This evidence was of low quality, which means that there is some certainty that future studies may alter our confidence in this evidence. Adverse events related to the discretionary‐use group were not reported.
Secondary outcomes
We did not find information about the number of self‐reported sunburns or skin lesions, total hours of ultraviolet radiation exposure, total hours outdoors in peak exposure times, or minimal erythema dose (MED). Our sole included study only reported the following secondary outcomes.
Actinic or solar keratoses at any follow‐up
The rate of change in the total number (full body count) of prevalent actinic keratoses between 1994 and 1996 was similar in the daily‐application group (n = 812 participants randomly assigned) and in the discretionary‐use group (n = 809 participants randomly assigned; 559 participants analysed) (RR 0.95, 95% CI 0.75 to 1.20; Analysis 1.3).
Participant's compliance with preventive strategies at the end of the trial
In a narrative report, Green 1999 stated, "75% of participants assigned to daily sunscreen use were applying sunscreen to their neck, arms and hands at least 3 to 4 days a week and those people not assigned to the sunscreen group were applying sunscreen to head, neck and arms not at all or no more than 1 or 2 days a week." For the daily use group, the median daily weight of sunscreen applied on average throughout the trial was 1.5 g/d (range 0 to 7.4 g/d), but the median decreased as the trial progressed (1992 = 1.67 g and 1996 = 1.22 g).
Discussion
We have summarised the evidence in summary of findings Table for the main comparison.
Summary of main results
In this review, we only identified one study, which evaluated whether regular daily use of sunscreen in comparison with discretionary or occasional use prevented new cases of basal cell carcinoma (BCC), cutaneous squamous cell carcinoma (cSCC), and our secondary outcome actinic keratoses at follow‐up (Green 1999). The study was undertaken in Nambour in South East Queensland (Australia); the researchers also studied adverse events and adherence related to the use of sunscreen. They found that sun protection factor (SPF) 16 sunscreen applied on exposed sites of the body, such as the head, neck, and upper limbs, every morning and repeated after heavy sweating or bathing made no difference in terms of reducing the occurrence of new cases of BCC and cSCC confirmed or not by histopathology at 4.5 years, in comparison with discretionary use (low‐quality evidence; see summary of findings Table for the main comparison). We did not identify studies comparing the use of sunscreen against no use of sunscreen.
Adverse events associated with the use of sunscreen in the daily sunscreen use group, such as occasional local reactions, like contact dermatitis; the feeling of oily skin; and interference with perspiration, were documented.
We identified no studies that evaluated other sun protection measures or regimens, such as wearing sun‐protective clothing, glasses, or hats, or seeking the shade when outdoors.
Overall completeness and applicability of evidence
Although we searched for randomised controlled trials (RCTs) of sun protection strategies aimed at preventing the development of cutaneous squamous cell carcinoma, basal cell carcinoma, and actinic keratoses, we found only one study, which was on the use of sunscreen. Limitations of this study were that it was restricted to only one SPF (SPF 16); it was in a population at higher risk of keratinocyte cancer due to the majority of participants having fair skin (55.2% were fair) and therefore a higher risk of sun damage and developing keratinocyte cancer (KC); and it did not address several of the prespecified secondary outcomes planned in our review.
Our review did not find differences between two sunscreen regimens (daily use versus discretionary use) in terms of the number of participants with new cases of BCC or cSCC, so to date, available evidence regarding the effectiveness and safety of preventive strategies for cSCC and BCC is scarce. This result could be attributed to a short period of exposure to sunscreen (4.5 years) and a limited tracking time. Given the above, it is possible that the clinical assessment did not last long enough to demonstrate that a protective effect can reduce the occurrence of BCC.
Cutaneous squamous cell carcinoma lesions are strongly related to sun exposure. It is estimated that 90% of these occur in body sites that are usually exposed to the sun (Miller 2010). Preventing early stages of carcinogenesis would indicate the benefit of performing preventive procedures from childhood with a longer follow‐up period (Zelen 1988). The pattern of BCC‐causing sun exposure seems to be different from that of cSCC, because although it is accepted that ultraviolet exposure (UV) is the main causal factor, the accurate relationship between the amount, timing, and pattern of exposure is unknown (i.e. exposure age, number of years, intensity of exposure, etc.) (Roewert‐Huber 2007). Although cSCC lesions show a higher metastatic profile in comparison with basal cell carcinoma lesions, there are BCC histologic patterns that might be locally aggressive without correlation with its metastatic risk (Roewert‐Huber 2007). However, reduced sunlight exposure can be deleterious for some populations, producing long‐term harms such as vitamin D deficiency and depression, which is a reason for not recommending reduced exposure to sunlight as a general measure (Maslin 2014).
The results of this study highlight that there are peculiarities in populations that could change the estimate of the effect, among which, phototype, type of radiation, geographic location (altitude above sea level, ozone layer, and latitude), and the number of hours spent outdoors are worth noting (Holman 1984). Thus, the benefits of using sunscreen may vary from one population to another. Finally, we must emphasise that Green 1999 conducted an evaluation of effectiveness with only one sun protection factor (SPF 16) and a once a day application scheme, which today would be considered inadequate (Pissavini 2013).
Quality of the evidence
The results of this single study suggest that using sunscreen on a daily basis, compared with discretionary use, does not reduce the onset of new cases of basal cell or cutaneous squamous cell carcinoma. This evidence was of low quality, which means that there is some certainty that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. We downgraded the risk of bias due to a number of potential risks that had an unclear impact on the results, including an unclear assessment of interaction between the effects of the different interventions evaluated (that is, sunscreen and beta‐carotene), an unclear impact of multiple posthoc analyses not planned a priori (repeated significance testing), and an unclear impact of clinical versus histological diagnosis of keratinocyte cancer. We deemed other considerations such as inconsistency, imprecision, and publication bias as not serious, but readers should consider these issues when considering the main results of this trial.
Potential biases in the review process
This review was comprehensive in aiming to identify clinical trials addressing the issue of the effectiveness and safety of sun protection measures for the prevention of cSCC and BCC. In general, we followed most of the strategies recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) regarding identification of relevant studies. However, the number of references identified was low, demonstrating the poor level of research at trial level in this area. This is understandable given the difficulty of long‐term population experiments, which are needed to adequately identify incident cases of KC. Also, the effectiveness of sun protection measures is often studied in conjunction with educational strategies to maximise adherence to preventive strategies, but this review did not address the educational aspect.
In this review, we made some changes from the protocol, which may be sources of bias. We expanded the criteria for confirmation of BCC and cSCC lesions, because histopathological confirmation of keratinocyte cancer could be a difficulty in population‐based studies with large sample sizes, and then, clinical confirmation is more likely to be used. Despite expanding our criteria, we considered that there was a lack of applicability of our results to a wider population, so we downgraded for indirectness. This change could be considered a source of bias in this review.
Agreements and disagreements with other studies or reviews
van der Pols 2006 contained the results of a follow‐up study of the cohort of people participating in the Nambour study eight years after the cessation of the trial. Including the time of the trial, a 12‐year follow‐up was completed. Comparing the group of participants originally assigned to daily sunscreen use with the group not assigned to daily sunscreen use, the author found that sunscreen significantly reduced the incidence of cSCC in both the number of people affected (RR 0.65, 95% CI 0.45 to 0.94) and the number of tumours diagnosed (RR 0.59, 95% CI 0.38 to 0.90). However, it was not possible to demonstrate, even during this monitoring period, that the regular application of sunscreen could reduce the occurrence of new cases of BCC, which may take many years to develop (van der Pols 2006). Although our plan was to assess the development of BCC or cSCC at any follow‐up period, it was clear to us (because of the participants' compliance with the preventive strategies) that the information reported in this study was a follow‐up of a cohort of participants who did not necessarily use sunscreen as planned in the trial.
Bimczok 2007 found that the minimum dose of sunscreen to be effective was SPF 15, and the minimum amount should be 2 mg/cm². However, subsequent studies have reported that the general population applies close to 1 mg/cm², which would affect the effectiveness of the intervention (Jansen 2013a; Jansen 2013b). In 2011, the U.S. Food and Drug Administration (FDA) recommended the use of sunscreen with a SPF higher than 15, and that same year, the National Institute for Health and Care Excellence (NICE) in the UK published a guide for the prevention of skin cancer in which it too recommended the use of sunscreen with a SPF higher than 15 (NICE 2011). Pissavini and collaborators, who after studying the population's patterns of sunscreen use and ultraviolet (UV) exposure, recommended the use of SPF 30, have recently supplemented these results, given that the effectiveness of this measure is related to frequency of use and the dose used (Pissavini 2013).
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
Comparison 1 Daily application of sunscreen versus discretionary use (including beta‐carotene use), Outcome 1 BCC confirmed clinically or histopathologically at any follow‐up.
Comparison 1 Daily application of sunscreen versus discretionary use (including beta‐carotene use), Outcome 2 cSCC confirmed clinically or histopathologically at any follow‐up.
Comparison 1 Daily application of sunscreen versus discretionary use (including beta‐carotene use), Outcome 3 Actinic or solar keratoses at any follow‐up.
| Daily application of sunscreen compared with discretionary use (including beta‐carotene use) for preventing basal cell and cutaneous squamous cell skin cancers | ||||||
| Patient or population: general participants | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | Number of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Discretionary use (including beta‐carotene use) | Daily application of sunscreen (including beta‐carotene use) | |||||
| Basal cell carcinoma | 78 per 1000 | 80 per 1000 (58 to 111) | RR 1.03 (95% CI 0.74 to 1.43) | 1621 | ⊕⊕⊝⊝ | The study authors reported narratively that the incidence was similar, even when analysis was restricted to cases of BCC diagnosed histologically. Also, the study authors reported a similar risk when groups receiving beta‐carotene were excluded (IDR 0.96, 95% CI 0.59 to 1.57). |
| Cutaneous squamous cell carcinoma | 31 per 1000 | 27 per 1000 (15 to 48) | RR 0.88 (95% CI 0.50 to 1.54) | 1621 | ⊕⊕⊝⊝ | The study authors reported an IDR of 0.74 (95% CI 0.39 to 1.38) when the analyses were restricted to histologically diagnosed cSCC. Also, the study authors reported a similar risk when groups receiving beta‐carotene were excluded (IDR 0.74, 95% CI 0.31 to 1.77). |
| Adverse events | See comment | See comment | ‐ | 1621 | ⊕⊕⊝⊝ | This was reported narratively for the daily sunscreen use group only. The most frequent adverse events were contact allergy or skin irritation (25 participants), skin greasiness (10 participants), and interference with perspiration or stinging eyes after facial perspiration (6 participants). |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| ¹Downgraded by one level due to other potential sources of bias, including an unclear assessment of interaction between the effects of the different interventions evaluated (that is, sunscreen and beta‐carotene). | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 BCC confirmed clinically or histopathologically at any follow‐up Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 cSCC confirmed clinically or histopathologically at any follow‐up Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 3 Actinic or solar keratoses at any follow‐up Show forest plot | 1 | Risk Ratio (Random, 95% CI) | Totals not selected | |
