The Use of Commercially Available Personal UV-meters Does Cause Less Safe Tanning Habits: A Randomized-controlled Trial

Authors


*e-mail: e.crocetti@cspo.it (Emanuele Crocetti)

Abstract

UV Index information is currently recommended as a vehicle to raise public awareness about the risk of sun-exposure. It remains unknown to what extent this information can change personal sun-protective behavior. The aim of the study was to analyze the effects of UV-Index (UV-I) information provided by low cost, commercially available UV-I sensors on major indicators of sun-tanning behavior. A randomized-controlled trial was carried out on 94 healthy volunteers aged 21–23 years. After the exclusion of subjects with photosensitive disorders (n = 3), 91 subjects were randomized in two arms after stratification based on phototype and sex. Both arms received a diary to be filled every day with a log of intentional sun-exposure during summer. Subjects in the intervention group also received a commercially available UV-I sensor. The UV-I sensors were switched on and the UV-value was recorded in 77% of days with sun-exposure. During days of sun-exposure, subjects randomized to the intervention group had longer average time of sun-exposure (227.7 vs 208.7 min per day, P = 0.003), also between noon and 4 pm (P < 0.001), and less frequently adopted sun protective measures than controls (hat [6.4%vs 10.2%, P = 0.007], sunglasses [23.9%vs 30.8%, P = 0.003], sunscreen [41.4%vs 47.2%, P = 0.02]) and they experienced more frequent sunburns (27.8%vs 21.5%, P = 0.004). The odd ratio of sunburns was 1.60 for subjects in the intervention group compared with controls (after adjustment for sex, sunscreen use and skin type). The mean UV-I value recorded by volunteers was lower (5.6 [SD ± 0.9]) than that (7.3 [SD ± 0.46]) recorded by a professional instrument in the same period at the same latitude. Poststudy laboratory tests showed that the sensor was able to detect only about 60% of the solar diffuse radiation. The use of UV-I sensors changed the sun protective behavior of sunbathers in the direction of less use of sun protective measures. One possible explanation is that the low cost UV-meters may have functioned incorrectly and under-reported UV exposure. This may have led to an underestimation of UV-I values, erroneously reassuring subjects and causing a less protective sunbathing behavior. Another hypothesis relies on a cognitive pitfall in the subjects’ dealing with intermediate UV-I values, as they may have been discouraged in the use of sunscreen as they did not feel that they had yet been exposed to very harmful UV radiation.

Introduction

An excess of intermittent sun-exposure, particularly if followed by sunburn, is the only preventable risk factor for cutaneous melanoma and basal cell carcinoma which has been identified (1).

Accordingly, the promotion of a safe exposure to the sun is the major task for primary prevention strategies against melanoma and nonmelanoma skin cancer. However, the most effective way of achieving change in sun-tanning behavior in sun-seeking subjects remains to be identified (2).

The UV-Index (UV-I) is a dimensionless measure of the UV irradiance effective in producing erythema reaching the surface on a horizontal plane. The index is computed by integrating the spectral irradiance over 280–400 nm weighted by the erythema action spectrum multiplied by a constant value of 40 W−1m2, and its value is independent of the individual skin sensitivity. The index was proposed in 2002 as a result of the INTERSUN program of the World Health Organization (3). The UV-I values varies from 0 to 12 (0–10 at a Mediterranean latitude), where each unit represents a multiple of the minimal dose able to affect biological tissue. The UV-I has been recommended as a vehicle to raise the public awareness of the potential detrimental health effects from solar UV exposure and to alert people of the need to adopt protective measures (4). A comprehensive list of scientific organizations that report the daily UV-I measurements across the world is available on the web (5–8).

However, little is known on the influence that information of the UV-I has on sun-exposure habits among those who sunbathe. Many newspapers, TV stations and radio bulletins across the world provide a UV-I forecast within the weather forecast, in particular during the summer. However, this method of distributing the information has several limitations since the value is a forecast, and as each UV-I value covers a wide geographical area (according to ZIP code in the USA) and refers to the expected value around solar noon, in many cases for clear sky conditions. Therefore, the estimate is only approximate due to factors which influence the UV-I such as clouds, altitude, surrounding surfaces, scattering, air pollution and the change of the UV-I throughout the day (9). A recent randomized trial addressed to Australian adults showed that the provision of solar-UV forecast in weather forecasts did not significantly changed their sun-protection practices (10).

A more suitable and reliable way of giving information on the UV-I is the use of a personal UV-meter worn by the subject. Other approaches can be conceived including an SMS message on personal mobile phones or a central UV-I measure provided by each bathing establishment.

Low cost UV-I meters are already commercially available on about 10 websites from about $15–70 USD each. They are generally small, easy to use and can be worn comfortably like a wristwatch. Therefore, following recommendations about the UV-I, anyone can purchase these instruments in order to monitor the UV-I and prevent excessive sun-exposure.

The aim of this study is to conduct a randomized-controlled trial to determine the effects of the use of a low cost, commercially available UV-I meter on some indicators of sun-tanning behavior such as time spent sunbathing per day, use of sun-protective measures such as sunscreen, clothes and sunglasses and frequency of sunburns.

Materials and methods

Study design.  In May 2004, 94 healthy volunteers aged 21–23 years were recruited for the study from students of the University of Florence medical school. After a brief explanation of the study protocol, an informed written consent approved by the Institutional Review Board was obtained from those that were willing to participate. Each subject was then interviewed for exclusionary criteria such as photosensitive diseases and diseases requiring phototherapy or sun-exposure for therapeutic purposes, including vitiligo, psoriasis, atopic dermatitis and chronic eczema. Three subjects were excluded because they reported photodermatitis, which was probably polymorphic light eruption according to a tentative diagnosis based on the anamnesis. The remaining 91 subjects were randomized into intervention (n = 46) and control groups (n = 45) (Fig. 1).

Figure 1.

 Randomization tree describing the study groups.

Before randomization, a dermatologic visit was performed to determine Fitzpatrick phototype and selected variables which could possibly modify the attitude and habits of sun-exposure, including total number of nevi, family history of melanoma in first- and second-degree relatives and the presence of atypical nevi.

A block randomization list (block size n = 10) was generated stratified according to sex (males and females) and phototype (Fitzpatrick II, III and IV), considered to be the main variables determining sun-tanning behavior. Due to the narrow age range, age was not considered to be a confounder.

The results of the randomization list, produced by E.C., were unknown to any of the other coauthors and were contained in a set of sealed envelopes, each bearing on the outside only the block identification (sex and phototype). After the preliminary interview and dermatologic visit, for each patient the appropriate numbered envelope was opened; the card inside told if the patient was to be enrolled in the intervention or control group.

The intervention group received a diary to be filled every day of sunbathing in the months of July and August 2004 and a UV-meter (see below) to be used to measure the UV-I in case of sun-exposure. Controls received the diary only.

 Personal UV-meter: technical characteristics.  After research on the web (April 2004), the Suncast UV monitor watch (http://www.safety-devices.com/suncast.htm) was chosen because of its easy use (continuous measure of UV-I on the display), small dimension (like a wristwatch) and intermediate cost (about $30 USD each) among similar instruments available.

The technical features of the instrument given on the manufacturer’s Website matched the minimal requirements of this study purpose. Once the instruments were received (n = 50), the technical characteristics of a sample group (n = 5) were reassessed.

The calibration test was performed in the laboratory using a collimated Xenon light lamp and a reference UV–VIS spectroradiometer (SpectrAIR, Flyby s.r.l., Livorno, Italy). The agreement was acceptable for the purposes of this experiment.

 Instructions to participants.  UV-meters were given to subjects allocated to the intervention arm along with a brief written explanation on their use. In order to avoid misreporting if the instruments were not correctly exposed to the sun, use on the wrist was discouraged and participants were asked to position the instrument parallel to the ground on their palm held at chest level for UV-I measurements. This measurement procedure was the most reliable in the preliminary tests performed by the Flyby Company. Participants were asked to use their UV-meter whenever they had intentional sun-exposure, repeating the measure every 2 h in cases of prolonged exposure. Both cases and controls received a short leaflet with statements about advice for safe sun-exposure in accordance with the UV-I value. The recommendations given are in accordance with international UV-I guidelines reported by the WHO (3). The same information was given both to cases and controls in order to minimize the risk of bias between the two study arms due to education about sun protection.

All the subjects (46/46) assigned to the intervention arm and 89% (40/45) of those assigned to control arm completed the study and were included in the analysis.

 Validity of randomization within the study. Tables 1 and 2 show the distribution of randomization variables (age and phototype); possible confounders which could affect the subjects’ attitudes toward suntanning including skin and hair color, number of nevi, family history of melanoma; and prestudy habits in the two study arms. No significant differences were found for any of the selected variables.

Table 1.   Demographic, constitutional variables and risk factors for melanoma in the intervention and control groups.
  1. *No subject with phototype I was found. †Statistical test = ‡Student’s t-test and §chi-squared test. Sex and phototype were used as randomization variables.

VariableIntervention group (n = 46)Control group (n = 40)Probability according to statistical tests†
Mean age (years)24.023.7P = 0.60
Sex
 Male1214 
 Female3426§P = 0.37
Phototype*
 II1916 
 III2218 
 IV56§P = 0.85
Skin color
 Fair2020 
 Intermediate2116 
 Olive54§P = 0.83
Eye color
 Black/brown2920 
 Green1014 
 Grey/blue76§P = 0.37
Hair color
 Black/dark brown3334 
 Light brown103 
 Blond33§P = 0.18
Family history of melanoma (first- and second-degree relatives)
 No4438 
 Yes22§P = 0.89
Number of melanocytic nevi
 ≤102720 
 11–301211 
 >3079§P = 0.63
Atypical nevi
 None3933 
 1–264 
 >213§P = 0.48
Table 2.   Pre-study attitudes and habits on sun-exposure in subjects allocated to the intervention and control groups.
VariableIntervention group (n = 46)Control group (n = 40)Probability according to the χ2 test
Sunscreen use when sun exposed
 Never37P = 0.12
 Seldom2012
 Frequently126
 Always1115
Sunscreen users: degree of protection commonly chosen
 None25P = 0.42
 High 105
 Intermediate 2421
 Low 108
Amount of sun-exposure: average number of weeks per year spent at the beach
 Up to 21719P = 0.71
 3–42014
 5–643
 >653
Do you love sunbathing?
 No1614P = 0.98
 Yes3026
Do you love sunbathing between noon and 4 in the afternoon?
 No1212P = 0.69
 Yes3428
History of sunburn in childhood (≤15)
 No1013P = 0.63
 1–31414
 4–595
 6–10116
 >1022
History of sunburn in adulthood (>15 years of age)
 no97P = 0.75
 1–31820
 4–594
 6–1087
 >1022

 Sunbathing locations.  Five types of possible sunbathing locations were considered: beach, country, mountain, lake and city (i.e. at a swimming pool or during recreational activity such as tennis or running).

 Sunbathing latitude.  According to information given by the volunteers about the exact location where sunbathing occurred, the latitude was classified as central Mediterranean (Tuscany and central Italy), south Mediterranean or tropical (South of Italy including Sicily and Sardinia Island, Greece, South of Spain).

 Data analysis.  In order to analyze differences in suntanning habits between subjects allocated to the intervention and control arms we computed “days with sunbathing,” then “amount of sunbathing sun-exposure during days with sunbathing” and then “average sunbathing time during days with sunbathing.” The same also for days with non-sunbathing sun-exposure (e.g. open-air sport activities).

Sun-protection habits are presented overall and according to the two types of sun-exposure. The intervention and control groups were compared using the Student’s t-test and chi-squared test.

A logistic model was used to evaluate the risk of sunburn according to the group (intervention/control), the days with sunscreen use, skin type and sex.

Results

UV-index measurements

According to the study diaries, the personal UV meter was switched on and the UV-value was recorded on 77.0% of the total study days with sun-exposure (579/752). At least one measurement per day was taken on 579 days (77%), two on 300 (39.9%), three on 112 (14.9%) and four on 36 days (4.8%).

The mean UV-I value recorded across all the diaries was 3.9 (SD ± 1.9) for the first reading, 3.8 (SD ± 1.5) for the second, 3.3 (SD ± 1.6) for the third and 2.6 (SD ± 1.2) for the fourth reading.

Sun-tanning behavior: duration of sun-exposure and use of protective tools

Table 3 shows the results in terms of duration and characteristics of sun-exposure for both days of sunbathing and no sunbathing sun-exposure. Subjects in the intervention group had on average a longer time of sun-exposure than those in the control groups (227.6 vs 208.7 min for day of sun-exposure, P = 0.003), also in the time interval 12 pm to 4 pm (129.2 vs 106.0 min for day of sun-exposure from noon to 4 pm, P < 0.001).

Table 3.   Results on sun-exposure and adoption of sun protective measures according to subject diaries during the study period in the intervention and control groups, overall and by type of sun-exposure (sunbathing and no sunbathing).
Variable Intervention group (n = 46) Control group (n = 40)Probability according to the statistical test*
  1. 95% CI = 95% confidence intervals. *Statistical test = Student’s t-test and chi-squared test.

Overall days of sun-exposure
N   752   772 
Time (min)171139161135 
Average time of sun-exposure (min)   227.6   208.70.003
Average time (min) of sun-exposure between noon and 4 pm   129.2   106.0<0.001
Use of sunscreen
 No   441   408 
 Yes   311 (41.4%)   364 (47.2%)0.02
Use of T-shirt
 No   562   587 
 Yes   190 (25.3%)   185 (24.0%)0.56
Use of sunglasses
 No   572   534 
 Yes   180 (23.9%)   238 (30.8%)0.003
Use of hat
 No   704   693 
 Yes    48 (6.4%)    79 (10.2%)0.007
Days with sunburns
 No   543   606 
 Yes   209 (27.8%)   166 (21.5%) 0.004
Days of sunbathing sun-exposure
Average time
 N   518   512 
 Time (min)121430112662 
Average daily time of sun-exposure (min)   234.4   220.00.04
Average time (min) of sun-exposure between noon and 4 pm   127.7   107.3<0.001
Use of sunscreen
 No   230   208 
 Yes   288 (55.6%)   304 (59.4%)0.22
Use of T-shirt
 No   494   483 
 Yes    24 (4.6%)    29 (5.7%)0.45
Use of sunglasses
 No   404   361 
 Yes   114 (22.0%)   151 (29.5%)0.006
Use of hat
 No   498   467 
 Yes    20 (3.9%)    45 (8.8%)0.001
Days with sunburns
 No   348   372 
 Yes   170 (32.8%)   140 (27.3%) 0.055
Days of no sunbathing sun-exposure
 N   234   260 
 Time (min) 49709 48473 
Average daily time of sun-exposure (min)   212.4   186.40.06
Average time (min) of sun-exposure between noon and 4 pm   133.5   102.6<0.001
Use of sunscreen
 No   211   200 
 Yes    23 (9.8%)    60 (23.1%)0.0001
Use of T-shirt
 No    68   104 
 Yes   166 (70.9%)   156 (60.0%)0.01
Use of sunglasses
 No   168   173 
 Yes    66 (28.2%)    87 (33.5%)0.21
Use of hat
 No   206   226 
 Yes    28 (12.0%)    34 (13.1%)0.71
Days with sunburns
 No   195   234 
 Yes    39 (16.7%)    26 (10.0%)0.03

Regarding the adoption of sun protective tools, a statistically significant difference was found between the intervention and control groups in the frequency of the use of sunscreen (41.4%vs 47.2%, P = 0.02), sunglasses (23.9%vs 30.8%, P = 0.003) and hat (6.4%vs 10.2%, P = 0.007) with less use in the intervention group than in the control arm. On the contrary, clothes (T-shirts) were worn at the same percentage (P = 0.56) during the exposed days in both the intervention (190/752, 25.3%) and control groups (185/772, 24.0%).

The frequency of days with sunburns was significantly higher (P = 0.004) in the intervention (209/752, 27.8%) than in the control group (166/722, 21.5%).

Almost similar results were present during days of sunbathing and no sunbathing sun-exposure.

In particular, among the former, the intervention group had longer average time of exposure, also from noon to 4 pm, lower use of sunglasses and hat and higher frequency of sunburns (P = 0.06); during the no-sunbathing sun-exposure days the longer exposure mean time was confirmed as the lower use of T-shirt and sunscreen as the higher proportion of sunburns.

As regards the number of overall days of sun-exposure with sunburns, there were 3 subjects in the intervention group and 4 controls with 0 day, 4 and 2, respectively, with 1 day, 6 in both groups with 2 days, 7 and 9 with 3 days, 10 and 2 with 4 days, 2 and 4 with 5 days, 6 and 4 with 6 days, 1 and 3 with 7 days, 2 and 3 with 8 days, only 2 controls with 9 days, 2 and 1 with 10 days and 1 subject in the intervention with 11, 12 and 20 days of sun-exposure with sunburns.

With regards to the place of sun-exposure, 1039/1524 sunbathing days (68.2 %) occurred at the beach (72.1% cases, 64.4% controls), 242 (15.9%) in the city (12.2% cases, 19.4% controls), 167 (11.0%) in the countryside (9.3% cases, 12.6% controls), 51 (3.4%) in the mountains (3.9% cases, 2.9% controls) and 25 (1.6%) on the lake (2.5% cases, 0.8% controls). Overall, the locations of sun-exposure were different between the groups (χ2 [4 df] = 27.8; P < 0.001).

The information on the latitude of sunbathing was available only for 221/752 (29.4%) days for the cases and for 300/772 (38.9%) days of sun-exposure for the controls. A sunbathing day in a south Mediterranean or tropical latitude occurred in 44/221 (19.9%) days in cases and in 40/300 (13.3%) days in controls (χ2 [1 df] = 4.01; P = 0.044).

Risk of sunburns during sun-exposure

The risk of sunburns during sun-exposure days has been evaluated with a logistic model. The crude odds ratio (OR) showed a 40% increased risk (OR = 1.40, 95% confidence intervals [CI] 1.11–1.78) for subjects in the intervention group in comparison with controls. The risk for subjects in the intervention group was confirmed also including in the model sex, skin type and sunscreen use (OR = 1.60, 95% CI 1.23–2.00).

Discussion

A survey conducted in the United States in 1994 and in 1995 showed that the majority of television weather forecasters and newspapers reported the UV-I and that most of the public was aware of the UV-I, and some respondents declared that they were disposed to change their sun-protection practices (11). A more recent survey conducted in New Zealand (summer 2000–2001) showed that 66% of radio stations, both TV channels and 48% of newspapers reported UVR information in summertime weather reports (12). However, the impact of UV-I information on changes in sun protective behavior remains to be shown. Branstrom et al. (13) recently compared four different UV information packages directed to young adults and their effects on sun-related behaviors. Two out of the four study arms received a UV indicator that gave a rough indication of the UV intensity after a few seconds' exposure to sunlight with color changes, and these two groups were given different educational brochures. Only 42% of participants used the UV indicator, and 12% preferred to use the UV-I as presented in the news media. The study analysis was based on a questionnaire submitted before and after the intervention instead of on a diary as in this study. According to Branstrom’s study, sun-related behavior and belief changed in all groups, but information about UV-I or a personal UV intensity indicator did not decrease sunbathing or sunburn more than general written information (13). More recently also a randomized trial carried out in Australia did not show evidence of differences in sun-protection practices among adults randomized in three different groups of forecasts, two of which included UV forecasts and sun-protection messages (10).

This paper is the first randomized study aimed at investigating in healthy volunteers the effect on sun tanning habits of the UV-I information provided by a personal sensor. The study is based on a diary where the time of exposure, frequency of sunburns and use of sun protective measures were recorded every day by the subjects. These data demonstrate that the use of a low-cost personal UV-I sensor changes some sunbathing habits, in particular those regarding the adoption of sun protective measures. Unfortunately, this change was in the opposite direction as expected, as some sun protective tools such as hats, sunglasses and sunscreen were adopted less frequently by subjects using UV-I sensors than by controls. Probably as a consequence of the lower rate of sunscreen use, subjects randomized to the intervention arm had more frequent sunburns.

These results were almost the same both during sunbathing and no sunbathing sun-exposure days.

This study has several points of strength: a dermatologic examination allowed for the exclusion of subjects with photosensitive diseases or with diseases requiring sun-exposure for treatment. This made the sample of subjects that entered the study representative of the sun-exposure habits among healthy subjects. The randomization was perfectly balanced for a number of additional variables including phenotypic traits (fair skin, eye and hair color), risk factors for melanoma including family history, number of nevi and presence of atypical nevi as assessed in the dermatologic examination. A high participation rate was obtained, with 100% of subjects randomized to the intervention arm completing the study. Another point of strength concerns the UV-I meter used, which is already commercially available on the web at a cost accessible to a large portion of the public. This meant that the study focused on what happens using instruments that are likely to be chosen by laymen given price and availability concerns. On the contrary, the use of a cumbersome, expensive, high technology prototype would have given results not applicable to practice.

One possible weakness is the unbalanced distribution between the randomization arms in terms of location (beaches, mountains, country, lake and city) and of latitude (Mediterranean vs south Mediterranean) of days with sun-exposure. Concerning the latter point, the exact geographical location of each day was not included in the questionnaire items, but volunteers were encouraged to provide this information in a supplementary record. This information was only provided for 29.4% of days with sun-exposure among cases and 38.8% among controls. According to these limited data, the majority of days with sun-exposure occurred in the middle region of Italy, particularly on the beaches of Tuscany. Only a minority occurred in southern Italy, Greece or other sunnier locations. A slightly higher percentage of days with sun-exposure at lower latitudes was found among cases (19.9%) than among controls (13.3%). On a clear day in mid-July on the Tuscany beach, the UV-I can reach a maximum value of about 8.0, while in Heraclion (Cretan island) it can reach a value of 9.0, and the corresponding daily UV doses weighted for the Erythema can reach values of 4.5 or 4.9 KJm−2, respectively (G. Zipoli, LAMMA Weather Laboratory, personal communication). Therefore, a higher number of cases could have been exposed to a slightly higher UV irradiation. This could have increased the occurrence of sunburns, but it is very unlikely that it is involved in the lower adoption of sun protective tools among cases. It is noteworthy, however, that no statistically significant differences were seen between the mean UV measures among UV users in different latitudes (data not shown).

How can this less safe sun-exposure behavior among the intervention subjects be explained?

One hypothesis is that the UV-sensor functioned incorrectly in the hands of volunteers during the study. To investigate this possibility, the UV-I values read by sensors and recorded by volunteers in the diary were compared with those obtained by professional instruments over the same period by the weather laboratory LAMMA (14) operating in the Florence area in Tuscany in the center of Italy in the month of July 2004. The average of the highest UV-I values in each day among those recorded by the volunteers was 5.6 (SD ± 0.9), lower than 7.3 (SD ± 0.46) recorded by LAMMA. Similar values were found during the month of August (data not shown). According to WHO recommendations reported in the leaflet given to participants, a UV-I from 3 to 5 means a moderate risk of damage from unprotected sun-exposure. The advice in this situation is to take precautions such as covering up if you will be outside and to stay in the shade near midday when the sun is strongest. The leaflet does not mention the use of sunscreen, which may have encouraged intervention subjects towards a less protected sun-exposure with less sunscreen use.

In order to investigate the possible reasons of the lower than expected values recorded by the UV-I sensor when used by volunteers, we performed additional laboratory investigations on the instrument performance after the end of the study. The angular response of the instrument was compared with the cosine (ideal) response. This reveals that the sensor is able to detect only about 60% of the solar diffuse radiation. Transferring this laboratory information to operative conditions when the UV-I was recorded by volunteers, this would mean that when the sensor is directed towards the sun in clear sky conditions at midday, the measurement of the direct normal radiation is underestimated by about 20%. This underestimation could increase up to about 40% in the case of higher solar zenith angles and with a cloudy or hazy sky, when the diffuse radiation dominates over direct radiation. Moreover, the narrow field of view of the sensor makes the orientation of the sensor toward the sun disk critical during the volunteer measurement, so orientation errors could also contribute to the underestimation of the direct component of the solar radiation. It is plausible that these limitations found for the UV-meter adopted in this study equally apply to all low-cost UV-I sensors currently available, as the technology used among them is quite similar.

Another hypothesis that could explain why the knowledge of the UV-I is associated with less safe sun-exposure is a cognitive one. As the possible UV-I values range from 1 to 10, knowledge of being exposed to an intermediate UV-I (i.e. about 5–6 as occurred in this study) could cause an erroneous reassurance leading to subjects omitting sunscreen use despite recommendations, as they feel that they are still far from a very harmful exposure.

If this hypothesis was true, the UV-I information would enhance protective behavior only when very high values are reached. Accordingly, the UV-I information may not be useful in locations with intermediate irradiation such as those at the Mediterranean latitude.

In conclusion, the personal UV-I meter was switched on by sunbathers on 77.0% of days with sun-exposure with 1–4 readings taken over the course of the day. A significant impact was found on the adoption of sun protective measures, but the impact was in the opposite direction than expected. The lower frequency of the use of a hat, sunglasses and sunscreen in the intervention group are in the direction of more harmful sun-exposure. The higher frequency of sunburns confirms the clinical relevance of the behavioral changes associated with UV-I sensor use. One possible explanation of this result is an incorrect functioning of low cost UV-meters under real conditions, leading to an underestimation of the UV-I value. Another explanation is a cognitive pitfall that could discourage the use of sun protective measures in cases of intermediate UV-I values as the subjects consider themselves to not be exposed to harmful irradiation.

Concerning the possible incorrect functioning of the sensor, one possible alternative is new real-time UV-I services which have recently appeared on the web and are based on SMS text information provided to each sunbather on their mobile phone (15). This approach has the advantage that no sensors are used by the subjects so that no errors in the measurement can occur due to misuse of the radiometer, radiometer defects or lack of calibration. However, specifically designed studies must investigate the capability of this new approach to determine whether it has a favorable effect to enhance personal sun-protection practices.

Ancillary