Authors of EWG’s 2012 sunscreen investigation are: David Andrews, Ph.D., Senior Scientist; Sean Gray, M.S., Senior Analyst; Jane Houlihan, M.S.C.E., Senior Vice President for Research; Nneka Leiba, M.P.H., Research Analyst; Sonya Lunder, M.P.H., Senior Analyst; Olga Naidenko, Ph.D., Senior Scientist; Paul Pestano, M.S., Research Analyst.

Summary of Methodology

EWG’s 6th annual analysis of sunscreens includes safety and effectiveness ratings for 1,800 SPF-rated products, including sunscreens and SPF-labeled lip balms, makeup and moisturizers. Our ratings are based on an in-house compilation of standard industry, government and academic data sources, models we constructed over the past eight years and a thorough review of the technical literature for sunscreens. We have incorporated sunscreen ratings from this investigation into our Skin Deep cosmetic safety database, an online consumer tool available at www.ewg.com/skindeep.

We based our analysis on sunscreen ingredient listings obtained primarily from online retailers. We rated products for overall safety and sun protection efficacy based on five factors:

• Health hazards associated with listed ingredients (based on a review of nearly 60 standard industry, academic and government regulatory and toxicity databases);
• UVB protection (using SPF rating as the indicator of effectiveness);
• UVA protection (using a standard industry absorbance model);
• The balance of UVA/UVB protection (using the ratio of UVA absorbance to SPF);
• Stability (how quickly a sunscreen ingredient breaks down in the sun, using an in-house stability database compiled from published findings of industry and peer-reviewed stability studies).

Our calculated overall rating for each product reflects a combination of the product’s health hazard rating and efficacy rating.

The methods and content of our analysis are based on our review of the technical sunscreen literature, including hundreds of industry and peer-reviewed studies. We compiled the results of our analyses in an online, interactive sunscreen guide. The details of our methodology are described below.

What’s in EWG’s 2012 Sunscreen Guide?

EWG’s 2012 Sunscreen Guide reflects the latest science on UV exposure and sun protection.

• UVA-UVB balance. EWG’s sunscreen efficacy ratings consider not only absolute protection from UVA and UVB radiation, but also UVA-UVB balance, reflecting a growing consensus that balanced protection across the sun’s UVA-UVB spectrum better protects sunscreen users (AAD 2009b, FDA 2007). Our balanced assessment relies on manufacturers’ listed SPF values and EWG-modeled UVA protection.
• Absorbance spectra. We include absorbance spectra for active ingredients drawn from our review of industry publications and peer-reviewed scientific literature. These inform our evaluations of the effectiveness of each sunscreen in protecting users from UV radiation.
• Health hazard ratings. As in past years, hazard ratings reflect hazard data compiled for all sunscreen ingredients, active and inactive. Hazard scores are weighted to account for the percent of active ingredients in each product, reflecting the potential for greater exposures as concentrations increase.
• Significant concerns for sunscreens. Hazard scores account for properties of particular concern for sunscreens, including products that contain oxybenzone or vitamin A, products in a spray or powder form that may pose a risk when inhaled, and products listing SPF values exceeding “SPF 50+,” the limit suggested by the FDA in its 2011 proposed regulation (FDA 2011a,b). For sunscreens with a single significant concern, we assign a rating no lower than 3 (moderate hazard), and for sunscreens with two or more significant concerns, we assign a rating of no lower than 7 to reflect a higher level of concern.

Sunscreen Rating System — Summary

Our sunscreen rating system is based on a product-by-product analysis of safety and effectiveness. In our analysis of a product’s effectiveness we weigh four factors: UVA and UVB protection, balance and the stability of active ingredient combinations. We combine these factors to derive a product effectiveness (sun hazard) rating. In our analysis of product safety we assess the potential for skin absorption and health hazards for all active ingredients (FDA-approved sunscreens) and inactive ingredients to derive a product safety (health hazard) rating. We combine these  effectiveness and safety ratings to derive an overall product rating using the following algorithm:

The sunscreen’s efficacy is calculated by taking 28% of the UVB protection + 28% of the UVA protection score + 28% of the UVA/UVB balance score + 16% of the sunscreen stability score. This result is then combined with the health hazard score using a variable ratio. For high hazard (health hazard score of 7 and higher), the efficacy score and health hazard score are combined at a 1:1 ratio. Using a linear regression, the efficacy score ratio is increased to 2:1 when the health hazard score reaches 0. This combined score is rounded to the nearest integer between 0 and 10.

We also flag sunscreen-specific hazards. Recent data suggest that sunscreens should not: contain oxybenzone or vitamin A; be aerosolized or in powder form; list SPF values above “50+”. Products with one of these hazards cannot have a final score less than 3, products with 2 of more of these concerns cannot score lower than 7. Powdered sunscreens cannot be scored lower than 7 due to intense concerns about inhalation toxicity.

We also include a score to account for UVA-UVB balance. There is growing awareness that ideal sunscreens offer reasonably similar protection across the UV spectrum. The growth of ultra-high SPF sunscreens further exacerbates this problem, as UVA protection currently maxes out at about 20 in U.S. sunscreens (BASF 2011).

Sunscreen Efficacy (Sun Hazard)

Overview of Sunscreen Efficacy Evaluation

In our analysis of a product’s effectiveness, we weigh the four contributing factors (Shaath 2005): UVA protection; UVB protection; UVA/UVB balance; stability of active ingredient combinations, considering both the potential for active ingredient molecules to break down in sunlight, react with other ingredients or otherwise transform into compounds less effective at filtering UV radiation. We assign a score for each of these factors based on an evaluation of  the label SPF, EWG-modeled UV protection and the technical literature.

We derive an overall rating for product effectiveness (sun hazard) as the sum of these four factors, weighted by our judgment of their relative importance. In this calculation we assign a weight of 0.28 to each of UVA and UVB protection and balance, and a weight of 0.16 to stability. The procedures we used in our analysis of sunscreen efficacy are described below.

Absorbance Spectra for Active Ingredients

We based our analysis of sunscreen effectiveness in part on the absorbance spectrum of each active ingredient.

Example: modeled absorption spectrum of octyl methoxycinnamate

Figure 1: Source (Herzog 2002)

EWG uses the reported SPF and our calculated absorbance spectra to determine the UV blocking strength for the ingredients listed in this report. We gathered the absorbance data for each active ingredient from a variety of published scientific sources, listed below.

With these absorbance spectra we calculate the amount of UV radiation expected to be blocked (i.e., absorbed or scattered) along the UVA and UVB wavelengths. We use these calculations to aid in assessing the effectiveness of products.

Table. Sources for absorbance spectra used in EWG sunscreen analysis

¹ For inorganic active ingredients – titanium dioxide and zinc oxide – the “absorbance spectra” also takes into account the chemical’s ability to scatter UV radiation in the UVA range (Schlossman and Shao 2005). Two different zinc oxide spectra are used to reflect the large variation in efficacy with changing particle size. The default categorization for zinc oxide ingredients is particle size size >100nm (Lewicka 2011).

Absorbance spectra are represented in most of these sources either in graphic or tabular format as a function of wavelength. To use these absorbance spectra in our computations of sunscreen effectiveness, we developed an equation to represent each measured spectrum. When necessary, we digitized the graphical absorbance spectra from the sources listed above. We used the graphing and statistical analysis software package xmGrace (Turner, Team et al. 2004) to determine the best-fit polynomial expression for each absorbance spectrum. The maximum error between the digitized data and final fitted values was 1%, and for any given point was less than 0.05% in most cases.

Monochromatic Protection Factor (MPF) and Transmission Spectra for Ingredients and Products

Both SPF and MPF are unitless factors that provide a measure of the amount of UV radiation blocked by sunscreen. SPF is a single value, while MPF varies based on wavelength. In the U.S., SPF is derived from sunburn experiments on human volunteers, while MPF is derived from lab measurements of UV transmission (Herzog 2005). SPF can also be computed by combining the MPF spectrum with the effective action spectrum (EA) for sunburn (a measure of how much damage a particular wavelength of light will cause) (McKinlay and Diffey 1987).

The MPF is a measure of the amount of UV radiation blocked (i.e., absorbed or scattered) at a particular wavelength. It is a key component in our evaluation of sunscreen effectiveness. We developed UV transmission spectra for individual active ingredients and for all combinations of active ingredients in products that we assessed. We use the MPF transmission spectrum in our sunscreen report both to graphically represent the effectiveness of sunscreen products and ingredients across the UV spectrum, and to calculate the effectiveness of products in the UVA range. (We use SPF instead of MPF as the measure of product effectiveness in the UVB range.)

We computed the MPF transmission spectra following the method detailed by Herzog and implemented by the BASF sunscreen simulator (formerly the Ciba sunscreen simulator) (Herzog 2002; Herzog 2006; BASF 2011). This model accounts for the effect of uneven skin surfaces – a series of ridges and valleys instead of a smooth surface. The model represents sunscreen on the skin as a thin film distributed unevenly. The sunscreen thickness is modeled using a continuous height distribution that matches a gamma distribution function (Ferrero 2003). The use of the gamma method provides a significant improvement in calculated correlation with measured SPF in relation to the previously used two-step model (O’Neill 1984).

MPF is given by:

where T is the percent transmission of light; ε(λ) is the average molecular absorption coefficient (defined in Herzog), c is the average molar concentration of the active ingredients in moles/liter, d is the path length (20 micrometers is the assumed thickness of sunscreen based upon the recommended applied dose of 2 mg/cm2), and g and f are parameters fitted by Herzog (Herzog 2002; Herzog 2006) to match experimental data on European sunscreens and equaling 0.269 and 0.935, respectively. Once the transmission spectrum is obtained, it can be transformed into an absorbance spectrum and monochromatic protection factors (MPF).

We used the Herzog method (Herzog 2002; Herzog 2006) described above to compute the UV transmission spectra both for individual ingredients and for all variations of active ingredients in the products we assessed. The method requires as input the concentrations of active ingredients. In computations of MPF spectra for individual ingredients, we used the average concentration of that ingredient found in products we assessed. In computations of MPF spectra for products, we used the concentrations of active ingredients specified on the product label. For some products in our database, the concentrations of active ingredients were not available from our data sources. In those cases, we used the following hierarchy to establish assumed concentrations of active ingredients used in our MPF analysis:

1. Average concentration of active ingredient for products with the same SPF and active ingredient combination.
2. Average concentration of active ingredient for products with the same SPF and active ingredient in a different combination.
3. Average concentration of active ingredient for products with the same ingredient combination over all available SPFs.
4. Average concentration of active ingredient for all products containing that active ingredient.

We evaluated sunscreen effectiveness for a product based in part on our computation of the transmission spectrum for the product’s combination of active ingredients. We integrate over the combined effective absorbance spectrum as described by Herzog (Herzog 2002; Herzog 2006) over 1 nm wavelength intervals to obtain overall sunscreen product spectra based on the individual ingredient spectra described above. The spectral information is presented in this report over 10 nm wavelength intervals.

A sunscreen product must generally contain multiple active ingredients to achieve a high SPF rating due to FDA-imposed concentration limits and constraints on product formulation (Chatelain and Gabard 2001). In graphic representations of “UV blocking” in this report, we present the MPFs in 10 nm intervals for each active ingredient.

Evaluating Products’ UVA Protection

Calculating the UVA protection score

In evaluating the overall UVA effectiveness, a second UVA metric is calculated – the percentage of UVA light blocked or absorbed, calculated from our modeled spectrum of the product.  This is useful because it captures the intensity of UVA protection. This value is calculated by integrating the MPF between 320-400nm.

Using this method, a score was assigned to each product:

Calculating the number of sunscreens meeting the EU COLIPA UVA standard

The European standard for UVA protection in sunscreens set by Cosmetics Europe–The Personal Care Association requires sunscreens to have both a critical wavelength of 370 and a ratio of UVA protection factor/SPF greater than 1/3.  The UVA protection factor is calculated by weighting the absorbance spectrum between 320 to 400nm with the persistent pigment-darkening action spectra (Cosmetics Europe 2011).

The new FDA standard for broad spectrum protection

Starting in mid-December 2012, sunscreens sold and labeled as providing broad spectrum protection will be required to have a critical wavelength of 370nm (FDA 2011).  The new FDA standards do little to differentiate mediocre from excellent products. The standards have been criticized for providing no incentive for improvement (Diffey 2012).

Evaluating UVA/UVB Protection Balance

The need for UVA protection to scale with SPF protection has been recognized by the American Association of Dermatology, the FDA and the sunscreen-regulating agencies of the EU, the UK and Australia.

In assessing the balance of UVA-UVB protection of a product, we shifted from using the spectral uniformity index, our procedure in 2009, to the ratio of UVA-PF to labeled SPF. This improved method better accounts for imbalance, particularly in the high-SPF range, and relies directly on manufacturers’ measured SPF values, rather than modeled values.

We calculated a balance factor for each sunscreen as the ratio of the UVA Protection Factor (UVA-PF) for persistent pigment darkening to the SPF value listed on product labels.

Using this method, a UVA/UVB balance score was assigned to each product:

Evaluating Effectiveness of UVB Protection

We based our evaluation of UVB protection on each product’s SPF (Sun Protection Factor) rating, which is the accepted metric. We scaled the SPF factor to create a UVB rating for each product that ranged from 0 (effective) to 10 (ineffective). This calculation involved the following:

We set a linear relationship between SPF and a product’s UVB rating using two pre-established points, assigning a UVB rating of 1 (effective) to SPF 30 products and a UVB rating of 6.4 to SPF 15 products (moderately effective). These points were set to correspond to the three standard score ranges established in EWG’s personal care product rating systems, used in this sunscreen analysis as well as in our Skin Deep personal care product assessment guide.

Our standard three-tiered scoring system maps integer scores from 0 to 10 into the following three categories: 0-2 (low hazard or effective); 3-6 (moderate concern or moderate effectiveness); and 7-10 (higher concern or ineffective). A value of 5 was selected for SPF 15 products to correspond to the moderate score range. This effectively sets SPF 15 at the middle of the “moderate” score range, with SPF values lower than 15 getting higher hazard scores because they are generally not recommended by health authorities (e.g., AAD 2009).

With this linear relationship established between the UVB rating and SPF, we then calculated scores for the full range of SPF values on products using the following procedure:

• We calculated UVB ratings for products with SPF values from 0 to 40 and by interpolating or extrapolating along the line described above.
• We set the UVB rating at zero (effective) for products with SPF values of “40+” or with listed SPF values exceeding 40.

Ingredient Stability

Absorption of UV light causes many sunscreen active ingredients to undergo chemical reactions or structural changes on the skin. In most cases, these ingredients quickly return to their original form to absorb more energy. However, ingredients can also degrade and may lose their UV protectiveness. In fact, a study by Shaath found that seven of 14 common sunscreens in Europe photo-degraded significantly after exposure to UV radiation, specifically UVA (Shaath, Fares et al. 1990).

In certain cases, the degradation may also produce other chemicals that are toxic to skin and body cells, especially if the sunscreen has been absorbed into the skin (Gasparro 1997) or the reactions can speed up (catalyze) the degradation of other ingredients in the sunscreen mixture (Bonda 2005).

Ideally, we’d like to have laboratory results of photo-degradation for each active ingredient in every sunscreen product. Since this information is not publicly available and such testing is not required of manufacturers, a large number of studies from different sources were analyzed. In quantifying these studies, it is difficult to compare results of different studies because different experimental conditions were used (solvent versus sunscreen formulation; measurement of light energy; sample preparation). Additionally, the degradation rate of an ingredient in a dilute laboratory solvent (such as water or ethanol) may or may not be representative of the results during consumer use. Even results in one sunscreen formulation may not be representative of the results in another because of the different ways that active ingredients behave in different environments.

EWG performed linear regression analysis of percent degradation versus minimal erythemal dose (MED) exposures in solvent and sunscreen formulations. The regression equations for solvent and sunscreen systems were then weighted equally and classified into three categories:

We weighted solvent and formulation results equally because of the wide variation in test conditions and the possibility that a single sunscreen formulation may not be representative of other formulations.

There is insufficient information in the literature on the subject of photo-stability to reliably guide a sunscreen formulator, let alone the consumer. Our classifications are presented here:

In order to account for a situation where an individual ingredient may photo-degrade but the sunscreen itself continues to provide significant protection relative to its original level due to the presence of other active ingredients, we assume that the UV blocking effectiveness of an active ingredient experiencing major degradation is reduced by 50%, and the UV blocking effectiveness of an active ingredient with minor degradation is reduced by 25%.

We then re-integrate over the entire spectrum and compare the degraded spectrum to the original. UVA and UVB protection are weighted equally. Based on the amount of relative degradation, the following scores are applied separately to the UVA and UVB portions:

Menthyl Anthranilate and Padimate O fluoresce when exposed to sunlight, meaning they absorb in energy in the UVB range, and re-emit it in the UVA range. If an active ingredient fluoresces, we increase the stability score by 1 point.

The scores for UVA, UVB and fluorescence were added together for the overall stability score, which ranges from 0 to 9, and were then scaled to a range of 0 to 10.

Several inactive ingredients help prevent sun damage through mechanisms other than blocking UV rays. For example, a variety of anti-oxidants scavenge free radicals in cells (Klein and Palefsky 2005). In some cases, claims made for these ingredients are unregulated (Klein and Palefsky 2005), while in others, the SPF itself can no longer be predicted by the sunblocking ability of the actives alone (Stanfield 2005). In the later case, consumers are misled into believing they are receiving more protection than they actually are. For these ingredients, we attenuate the UVA and UVB scores as follows:

Particle Size Assumptions for Mineral Sunscreens

Small mineral particles with sizes in the nanometer range are used in sunscreens because they provide strong UV attenuation and because they are transparent when applied to the skin. Concerns about toxic effects increase as size decreases due to the potential for these small particles to be absorbed through the skin and bypass the body’s defense mechanisms.

Companies are not required to provide particle size information on package labels, leaving regulators and the public with little information to determine the prevalence of micro- and nano-scale materials in sunscreens (FOE 2009, FDA 2007). We reviewed available information about particle size and surface coatings for zinc oxide and titanium dioxide sunscreens on company websites and product labels, as well as the results of a manufacturers’ survey conducted by Friends of the Earth.

All mineral UV blockers in sunscreens involve a distribution of sizes, and the information we have seen indicates that a significant portion of the primary particles sizes are < 100nm. The largest advertised primary particle size of any manufacturer is BASF’s zinc oxide, which gives a mean particle size of 140nm for 20% of the particles < 100nm (BASF 2011).

We assume the following manufacturers use zinc oxide particles with a mean size > 100nm: Alba Botanica, Allergan, Avalon Organics, Black Opal, Blistex/M.D. Forte, Bullfrog, Tatoo Goo, Lavera Schwarzkopf & Henkel, Soleo Organics, Badger, EcoLani, Gaia, Kabana, Lotus Moon, Mexitan and Miessence — based on claims made on their website or claims made to Friends of the Earth (FOE 2009).

Titanium Dioxide

When the shortest dimension of the primary particles of titanium dioxide is 15 nm, it appears transparent on the skin, but at 35-60nm it appears opaque (Schlossman 2005). All information we have amassed about 10 different titanium dioxide suppliers indicated primary particles sizes of 10-35nm. Without clear regulatory guidelines, manufacturers and product formulators can claim they are using or not using nanoparticles without providing information to back up that claim. We assume that all UV attenuation-grade titanium dioxide sunscreens that apply clear on the skin use titanium dioxide with a mean primary particle size of 15-35 nm in the shortest dimension.

Zinc Oxide

Scientists estimate zinc oxide provides maximum UVB protection with particles sized 20 to 30nm, and the typical sizes of zinc particles in sunscreen are 30 to 200 nm (BASF 2011, Cross 2007, Nohynek 2007, Stamatakis 1990). Compared to larger particles, smaller particles provide greater UVB protection but less UVA protection (Schlossman 2005). Particles larger than about 200-300 nm tint the skin white and are unacceptable to most consumers (BASF 2000).

In our efficacy calculations we utilize two different zinc oxide particle sizes to reflect the larger range of sizes used in sunscreens. By default, EWG has assumed that zinc oxide used as an active ingredient in sunscreens has 40-60nm mean primary particle size, and that when the manufacturers indicate they use Z-Cote or non-nano minerals, we assume a mean primary particle size of 140 nm.

Safety Assessment (Hazard Score)

EWG’s health hazard scores were based upon the ingredient health hazard scoring system of our Skin Deep database (www.ewg.org/skindeep). This core database of chemical hazards, regulatory status and study availability pools data from nearly 60 databases and and from government agencies, industry panels, academic institutions or other credible bodies. The information in Skin Deep is used to create hazard ratings and data gap ratings for personal care products, as well as for individual ingredients.

The absorption potential of mineral ingredients is an important factor in our sunscreen scoring system. The hazard score for mineral ingredients is adjusted to account for exposure potential calculated from evidence regarding skin penetration or absorption, as described in greater detail on our Skin Deep “About” page.

We have given additional weight in our calculated hazard scores to properties of particular concern for sunscreens, including products that contain oxybenzone or vitamin A, products in a spray or powder form that may pose a risk when inhaled, and products listing SPF values exceeding “SPF 50+, the limit suggested by the FDA in its 2011 proposed regulation (FDA 2011a,b). For sunscreens with a single significant concern, we assigned a rating of no lower than 3 (moderate hazard), and for sunscreens with two or more significant concerns, we assigned a rating of no lower than 7 to reflect a higher level of concern.

Health hazard scores in our sunscreen evaluations reflect hazards specific to sunscreens as well as beneficial or potentially harmful effects of specific combinations of active ingredients. We assessed hazards identified by government, industry, and academic sources, and did not evaluate specific claims made by individual manufacturers.

This report includes a closer look at the 17 chemicals permitted by FDA for use as active ingredients in sunscreens (including the various sizes of inorganic sunscreens), and the 52 chemicals used in other countries to prevent UV exposure that are added to U.S. sunscreens for other purposes. We compiled relevant information from sources that included published reports in the peer-reviewed literature and risk assessments from the European Union, Japan and Australia, countries with robust sunscreen regulations.

Assessing Known or Suspected Chemical Hazards

Sunscreens sold in the U.S. are considered over-the-counter (OTC) drugs. They contain active ingredients that must undergo safety and effectiveness testing and inactive ingredients that, as in virtually all other personal care products, are not required to be tested for safety. We used different approaches to evaluate active and inactive ingredients.

Active ingredients, as well as assessments of specific active ingredient combinations, were evaluated by conducting an extensive review of the scientific literature. The review included peer-reviewed literature, filed and approved patents and reviews by government and industry panels, as well as cross-checks with the existing Skin Deep database. Certain inactive ingredients, such as those that are approved as active ingredients outside the U.S., are also treated as active ingredients in the health and sun hazard reviews.

Inactive ingredient assessments were conducted using the existing Skin Deep system cited above (EWG 2012). Skin Deep identifies chemicals that pose health hazards, including known and probable carcinogens, reproductive and developmental toxicants, neurotoxic and immunotoxic chemicals, chemicals flagged for persistence, bioaccumulation and toxicity in the environment, as well as chemicals banned or restricted in other countries. Skin Deep assessments also highlight the extensive data gaps for the majority of ingredients used in cosmetics and personal care products.

Briefly, EWG’s hazard ratings are a synthesis of known and suspected hazards associated with ingredients and products. Hazard ratings within Skin Deep are categorized as raising low, moderate or higher concern, with numeric rankings that range from 0 (low concern) to 10 (higher concern). Data gap ratings describe the extent to which ingredients or products have been definitively assessed for safety. Data gap ratings are represented in Skin Deep by a numeric percentage ranging from 100% (complete absence of safety data) to 0% (comprehensive safety data).

Further details concerning this methodology may be found on the Skin Deep website (www.ewg.org/skindeep).