In the first stage, an expert team of dermatologists and epidemiologists drew up the CHACES questionnaire by adapting a previous questionnaire created and validated by the same group in 2009 (“Cuestionario a pie de playa” [“The Beach Questionnaire”]).7 The adaptation involved an analysis of each item on the questionnaire by evaluating its content validity, with nuanced drafting of several questions in order to adapt it to schoolchildren and the addition of new items that would be considered of interest in this setting. The questionnaire approved by the committee was subsequently piloted by applying it to 10 health care professionals from Hospital Costa del Sol, Marbella, Spain. This approach aimed to identify ambiguous questions, possible errors, and difficulties in comprehension.

The second stage, or validation stage, involved an evaluation of the psychometric properties of validity and reliability. This stage consisted of a cross-sectional study in 2 phases:

• none-

  Phase I, analysis of the construct validity and reliability-consistency of the questionnaire.

• none-

  Phase II, analysis of the reliability-stability of the questionnaire.

The study was performed in the south of Spain in the province of Málaga, in the western part of the Costa del Sol, which includes the area between the towns of Benalmádena and Torremolinos. This is a tourist area with a Mediterranean climate and more than 300 days of sunshine per year.8,9 The study population comprised adolescents (aged 11–17 years) and adults (≥18 years). The sample was a convenience sample, chosen based on a single inclusion criterion, namely, complete fluency in spoken and written Spanish in order to be able to complete the questionnaire. The questionnaires considered valid for the study were those completed by individuals within the age range (older than 11 years) who provided responses for the variables of age, sex, and sunburn during the previous summer.

The minimum statistical power necessary to evaluate the validity and reliability of the items on the questionnaire was set at 200 individuals per group (40 items multiplied by 5).10

No information other than that specified in the questionnaire was collected. However, all of the study data were recorded anonymously, with strict adherence to current national legislation and regulations on data protection. Consent to participate in the study was obtained before participants completed the questionnaire. The project was approved by the Clinical Research Ethics Committee of Hospital Costa del Sol, since it was performed within a public health framework.

Two versions of the CHACES questionnaire were created, namely, an adolescent version (11–17 years) and an adult version (≥18 years) (See Appendix B Supplementary Material). The only differences between the 2 versions were the way the respondent was addressed (i.e., using the polite form “usted” in the adult questionnaire and the second person singular familiar pronoun “tú” in the adolescent version) and the inclusion of 4 additional items in the adult questionnaire (demographic data and sun exposure habits in the workplace).

The 2 versions of the questionnaire (adolescent and adult) are self-completed. The first has 39 items and the second 43. The questionnaires comprise the following sections: 1) Request to participate, 2) Personal and demographic data, and 3) Information requested (in 3 large blocks: habits, attitudes, and knowledge with respect to sun exposure). These are used to explore the following fields:

• 1

  Demographic data (5 common items in both questionnaires and 3 more in the adult version): sex, age, country of birth, town of residence, and educational level. The adult questionnaire also included profession, marital status, and whether or not the respondent had children aged under 12 years.

• 2

  Skin color (1 item in both questionnaires): color of skin not exposed to sunlight (5 response categories).

• 3

  Fitzpatrick skin type (1 item in both questionnaires with 4 response categories).

• 4

  Sun exposure habits (4 similar items in both questionnaires and 2 more in the adult version): Sun exposure habits are evaluated in 2 different scenarios, namely, sunbathing at the beach or pool and sports or leisure activities in the open air. We explored both the number of days per year and the number of hours per day of sun exposure (5 response categories). The adult questionnaire also explored the number of days per year of sun exposure in the workplace (5 response categories).

• 5

  Sunburn in the previous year (1 item in both questionnaires, with 5 response categories). Sunburn is defined as the presence of reddening and pain after exposure to sunlight.

• 6

  Sun protection behaviors when participating in activities in the open air (7 items in both questionnaires): Six sun protection behaviors recommended by the World Health Organization are explored, as follows11: staying in the shade, using sunglasses, using a hat or cap, wearing long sleeves or trousers, avoiding sunlight during the middle of the day (11:00 to 17:00), and using high-factor sunscreen (sun protection factor). Likert-type responses (5 categories) are used. As for sun protection habits, the sun protection factor of the usual sunscreen is recorded (5 categories).

• 7

  Attitudes toward sun exposure (10 items in both questionnaires): Attitude is defined as a general permanent positive or negative feeling. Attitudes to sun exposure and protection are investigated based on 3 dimensions (health, cosmetic value, and pleasure/leisure). The questions are answered on a Likert-type scale (5 categories).

• 8

  Knowledge with respect to sun exposure (10 items in both questionnaires): Ten basic concepts related to sun exposure and skin cancer are explored using dichotomous responses (true or false).

In the first phase, an expert panel comprising dermatologists and epidemiologists evaluated the content validity of the questionnaire.

In the second phase, a validation study of the 2 questionnaires was performed; both were performed in parallel using the same statistical analysis plan. For each of the 2 phases, a descriptive analysis was performed of the demographic characteristics, attitudes toward sun exposure, and knowledge of sun exposure in the population samples included. We studied the frequency distribution (endorsement frequency) for all the sections.

We analyzed both the construct validity (ability of the items to measure what they purport to measure) and the internal consistency (homogeneity of the items that measure the same concept) for the items corresponding to the concepts of sun protection behaviors, attitudes toward sun exposure, and knowledge with respect to sun exposure. The statistical analysis enabled us to eliminate items with the poorest yield and redefine the corresponding dimensions.

The concept validity of habits of and attitudes to sun exposure was measured using an exploratory factor analysis of the main components based on transformation of the solutions using a varimax rotation for the 6 items with the same response scale in the sun protection behaviors. The adequacy of the factor analysis was verified using the Kaiser–Meyer–Olkin test and the Bartlett test of sphericity. We considered as valid those items with communalities greater than 0.3 and factor saturations greater than 0.4. The internal consistency of the items for each of the resulting factors was explored using the Cronbach α. As for the area of knowledge, concept validity was studied using the facility index, together with the discrimination index, and the Cronbach α value for global evaluation of the scale. Similarly, we studied the mean (SD) of the number of correct answers for the whole scale.

The stability of the questionnaire was studied using the test-retest technique. We evaluated concordance between the pre-post evaluation of each of the items (15 days between surveys) using the absolute agreement percentage and the κ value with Cicchetti weights including the respective 95% CI. In accordance with the criteria of Landis and Koch,12 we considered as the minimum acceptable value per item a κ value greater than 0.20 (median degree of agreement); values greater than 0.6 were considered to indicate substantial agreement.

These properties were analyzed using a first population sample of 1482 adults and adolescents obtained from the health care setting. The participants were recruited with the help of health care professionals working in public health centers in the western part of the Costa del Sol between April and September 2014. A total of 15 health centers and 109 professionals participated in collection of the questionnaires.

A total of 1482 questionnaires were completed (78.4% adults [≥18 years] and 21.6% adolescents [11–17 years]). The mean (SD) age in the 2 groups was 39 (8.6) and 13.9 (1.7) years, respectively, with a mean overall age of 33.7 (12.8) years. There was a female predominance (68.9% women and 31.1% men), and 90.5% of the population were Spanish. The educational level was medium-high in 85%; only 2% had no formal education (Table 1).

Skin color was mainly pale or intermediate (32.8% and 38.5%, respectively), with a homogeneous distribution of skin types i to iv, although values were slightly higher for skin type iii (34.4%) (Table 1). More than 50% of the total population had been sunburnt the previous year; this was higher in adolescents (67.3% vs. 46.5% in adults). Sun exposure habits are shown in Table 1.

The frequency of use of different sun protection techniques and attitudes to sun exposure in each of the population groups surveyed is shown in Figs. 1 and 2.

Figure 1.

Sun protection behaviors: frequency distribution (% of the total).

(0.14MB).

Figure 2.

Attitudes toward sun exposure. Frequency distribution.

(0.14MB).

The construct validity and internal consistency-reliability of the questionnaire was assessed by analyzing its 3 components independently: sun protection habits, attitudes, and knowledge.

The frequency distribution and item-to-scale correlation for the items on the questionnaire associated with sun protection behaviors is shown in Table 2. None of the items had a response category greater than 50%, thus indicating their consistency. The score on the Kaiser–Meyer–Olkin test was 0.703, with a P value of < .001 in the Bartlett test of sphericity. Therefore, the fit of the 6 items analyzed was correct.

Factor analysis did not determine the loss of items on the questionnaire and showed that they all clustered in 2 components: component 1, which included sun protection measures not related to clothing (shade-sun umbrella, sunglasses, avoiding the sun in the middle of the day, and sunscreens); and component 2, which covers all measures associated with clothing (hat/cap and t-shirt/long trousers).

The communalities were between 0.5 and 0.64, with the exception of item 2 (“Use of sunglasses”, 0.369), and the factor saturations of the 2 components exceeded the value of 0.5 and thus accounted for 55.34% of the variance.

The analysis of internal consistency showed a Cronbach α of 0.63 for the first component (shade/sun umbrella, sunglasses, avoiding the sun in the middle of the day, and sunscreens) and 0.45 for the second (hat/cap and t-shirt/long trousers) (Table 3).

Table 2 shows the frequency distribution and the item-to-scale correlation for the items associated with attitudes toward sun exposure. Even though a slight ceiling effect was observed for some items, none had a response category greater than 80%, thus demonstrating their consistency.

The exploratory factor analysis initially demonstrated the poor fit of 2 items: A4 (“I don’t like using sunscreen”; communality, 0.22) and A10 (“It’s easy to protect yourself from the sun by wearing a hat and clothes that cover your skin”; communality, 0.17). Given the poor fit of these 2 items, we decided to modify item A4 by making it more positive (“I like using sunscreen”). The subsequent factor analysis of the components with A4 rotated enabled the item to be fitted (Table 3).

The Kaiser–Mayer–Olkin value was 0.778 (P < .0001 in the Bartlett sphericity test); therefore, the fit for the items analyzed was correct. The factor saturations of the 2 components exceeded the value of 0.4 and thus explained 54.4% of the variance.

Therefore, all of the items explored in the attitudes toward sun exposure were maintained and grouped into 2 components: component 1 covered attitudes to sun protection measures and the harmful effects of the sun (use of creams, use of hats/clothing, shade, and effects such as burns, spots/wrinkles, and skin cancer); and component 2, which included general attitudes on liking the sun (desire to sunbathe and get a suntan and/or considering the sun as a source of health) (Table 3).

The internal consistency analysis revealed a Cronbach α of 0.729 for the first component (attitudes to sun protection measures and the harmful effects of the sun) and 0.861 for the second component (general attitudes on liking the sun). The 10 items were maintained, with none excluded, since the Cronbach α value decreased with respect to that of its component in all items when it was deleted, except for items A4 and A10, in which it increased slightly (Table 3).

A total of 1333 persons completed the items on knowledge. The mean number of right answers was 6.86 (1.33), with no significant differences between the 3 groups analyzed (P<.440). However, the adolescent group had the lowest level of knowledge (6.5 vs. 7.4 for teachers and 7.1 for parents).

The facility index was calculated for the 10 items and showed all the values to be extreme (none were between 0.4 and 0.6). The discrimination index in all cases was below 0.25. The Cronbach α was 0.335, thus necessitating deletion of 5 items to obtain a modest Cronbach α (greater than 0.5) (Table 4). Nevertheless, we decided to maintain these 5 items, given that they were very stable, in order to replicate them in future populations.

The stability of the questionnaire was analyzed using a test-retest approach. A group of 39 adolescents/adults were surveyed at 2 time points separated by an interval of 2 weeks. The respondents were recruited by the research team. We analyzed the stability of the items associated with skin type, exposure habits, sun protection behaviors, and attitudes toward and knowledge of sun exposure by studying the absolute percentage and κ value with Cicchetti weights for each of these items (Table 5).

With respect to skin type, exposure habits, and sun protection behaviors, all values for the intraclass correlation coefficient were above 0.55 in all of the items except 3: “Habits: days per year - Sunbathing” (0.46), “Sun protection behaviors - Hat/cap” (0.49), and “Sun protection behaviors - Avoiding the sun in the middle of the day” (0.52), although the absolute percentage of agreement was in fact higher for these 3 items (61.5%, 51.4%, and 60.5%, respectively). The percentage of absolute agreement is higher than 60% for all the items, except “Sun protection behaviors - hat/cap” (51.4%).

As for attitudes toward and knowledge of sun exposure, the intraclass correlation coefficient values were mainly low (0.15–0.78). Subsequent analysis of the percentage of absolute agreement revealed a marked increase in these values for all of the items. The percentage of absolute agreement was higher than 60% for all the items, except 3: “Attitudes - I like being tan” (53.8%), “Attitudes - I prefer to be in the shade than in the sun in the middle of the day” (55.3%), and “Attitudes - It’s easy to protect yourself from the sun by wearing a hat and clothes that cover your skin” (56.4%).