Introduction

Vitamin D is produced on exposure of the skin to solar ultraviolent B (UVB) radiation and solar UVB is the primary source of vitamin D for most persons1,2. Vitamin D receptors have been discovered in most cells in the body and enzymes capable of converting circulating 25-hydroxyvitamin D [25(OH)D] to the active 1,25 hydroxyvitamin D [1,25(OH)D] are now known to exist outside the kidneys, including the skin, prompting a plethora of new discoveries about its function3. In addition to its protective effect on bone fractures, rickets, osteomalacia and osteoporosis, vitamin D is now thought to decrease a spectrum of chronic illnesses including internal cancers, cardiovascular disease, autoimmune diseases, metabolic disorders and mental illness4,5. Epidemiological and preclinical studies have provided evidence that vitamin D has protective effects against the development of cancer6,7. Garland et al8. explored the associations between sunlight and the breast cancer incidence and mortality in the United States and found a strong, inverse association between sunlight exposure and breast cancer mortality (−0.80, P < 0.0001). Although sun exposure increased the risk of skin cancer in Chinese9 and other10 populations, the effects of avoidance of suboptimal vitamin D levels on cancer cell proliferation are likely to be beneficial to the melanoma patient11. Serum levels in the range 70–100 nmol/L might be a reasonable target for melanoma patients as much as for other members of the population11. Individual sunlight exposure can be measured with objective methods including observations, skin reflectance using colorimeters or spectrophotometers, personal dosimetry using polysulphone film, skin swabbing using spectrophotometer and inspections of moles10,12,13,14. Compared with other measurements, questionnaires remain the most cost-effective assessment of population sunlight exposure10,12,13,15,16. Existing sunlight exposure questionnaires were mostly applied to Caucasians or non-Asians and generally did not collect the exposure information over lifetime or have not been validated13,14,15,16,17. Therefore this study aimed to develop and validate a lifetime sunlight exposure questionnaire for use in the Chinese population15,16.

Results

Content validity

The CVI of the questionnaire itmes ranged from 0.67 to 1.0. Except for one itme, all had CVI values of 0.83 or above (Table 1). However, this item was still retained in the questionnaire based on the study by Fitzpatrick18. The other 61 items were retained in the questionnaire.

Table 1 Content Validity Index (CVI) Values for Chinese Lifetime Sunlight Exposure Questionnaire
Full size table

Construct validity

A principal components analysis was conducted to evaluate the construct validity of the sunlight exposure questionnaire (n = 650). The Scree Plot (Figure 1) indicated that a two-factor solution was optimal. Four factors had eigenvalues greater than one. Principal components analysis revealed that the Total Variance explained by the first two factors were 52.9% and 17.5% respectively (Table 2) and a corresponding Component Matrix showed the correlation coefficients between each question and the two factors were positive and good (Table 3). Although 4 principal component factors could be extracted, the first two could explain 70.4% of the total variance and included all items analyzed. The two factors were labeled: (1) frequency and duration worked in the sun in four respective seasons in life; (2) hours per day spent in the sun in summer and other 3 seasons in 4 life stages. This analysis indicated that no items (continuous variables) need to be excluded for the Chinese sunlight exposure questionnaire.

Table 2 Two-factor Solution by Principal Items Loading for the Chinese Version of the Sunlight Exposure Questionnaire (n = 650)
Full size table
Table 3 Two-factor Solution: Factor Loadings by Principal Components Analysis on Items of the Sunlight Exposure Questionnaire (n = 650)
Full size table
Figure 1
figure 1

Scree Plot by Principle Component Analysis of the Chinese Lifetime Sunlight Exposure Questionnaire (n = 650).

Full size image

Reliability

Table 4 shows that the reliability was excellent for the average hours spent in the sun during the 4 respective life stages (ICC: 0.750–0.925), moderate to good for lifetime duration worked in the sun in the respective four seasons (ICC: 0.586–0.744). The item-total correlations for most items were moderate to good (0.419–0.886). Table 5 shows that eight items (ever went to a summer climate in winter during the four life stages, the living places from age 35 y to present and whether ever walked in the sun from 35 y to present, whether ever used sunlamp or sunbed) were consistent between the first and second interviews. The agreements were excellent (Kappa ranged between 0.82 and 0.90) for place of residence, whether usually used sunscreen, umbrella from 35 y to present. The agreements for most items for sun protection during the 4 life stages and the skin reaction to the sun were moderate to good (Kappa ranged between 0.51 and 0.75).

Table 4 Medians (P25, P75) of the Time Spent in the Sun as Estimated by the Sunlight Exposure Administered Twice (n = 94)
Full size table
Table 5 Variable Reliability on Sunlight Exposure Information and the Agreement between two Repeated Interviews (n = 94)
Full size table

Discussion

This is the first sunlight exposure questionnaire developed and validated for use in the Chinese population. The questionnaire was designed to capture the usual time spent in the sun, sun protection used and outdoor activities in the sun during the different life stages among Hong Kong Chinese premenopausal women. The content validity, construct validity and reliability of the questionnaire were adequate.

Content validity is the determination of the content representativeness or content relevance of the elements/items of an instrument. The content validity of questionnaires focusing on various exposure assessments fields has been assessed in a few studies. A newly developed Korean Acupuncture Sensation Questionnaire validated by expert panel judgment has sufficient content validity for de qi (CVI > 0.80)19. Thrush et al20 developed and established the content validity of a 43-item fixed-response instrument designed to measure the organizational climate for research integrity in academic health centers and established that the instrument has an excellent content validity (CVI = 0.90). The items included in our lifetime sunlight exposure questionnaire were established with sufficient content validity.

Our Chinese lifetime sunlight exposure questionnaire has also been found to have good construct validity. The Scree plot indicated a two-factor construct for the continuous variables. Principal Components Analysis revealed a satisfactory percentage (70.4%) of the Total Variance was explained by the two factors with eigenvalues greater than one. This analysis indicated that no items (continuous variables) needed to be excluded for the Chinese sunlight exposure questionnaire. Consistently, a Spanish questionnaire evaluated habits, attitudes and understanding of exposure to sunlight and factorial analysis of the principal components confirmed the construct validity with commonalities and factor saturations > 0.5021. Therefore, the construct validity of our questionnaire can be supported.

The existing sunlight exposure questionnaires have mostly been applied to the Caucasian or non-Asian populations; and generally have not collected exposure information over lifetime or have not been validated13,14,15,16,17. Knight et al used a sunlight exposure questionnaire to acquire the exposure data from Caucasians for three life periods: 10 to 19, 20 to 29 and 45 to 54 years. Chen et al collected information on sun exposure at different lifetime periods and tested skin reaction after 2-h sun exposure in the Taiwanese population9. However, the questionnaires used in these studies have not been validated. The objective measures of sunlight exposure, such as personal UV dosimetry, have been used to validate the questionnaire on recent sunlight exposure; but correlations between questionnaires and objective measures are usually not strong13. Our questionnaire was found to have sufficient construct validity to measure lifetime sunlight exposure among Chinese women. It could thus be applied to assess the association between vitamin D from sunlight exposure and health outcomes.

Consistent with other previous studies, the reproducibility for lifetime sunlight exposure questionnaires in other studies was relatively good. The Australian case-control study found that the test-retest k statistic of self-reported sun exposure ranged from 0.43 to 0.7422. Another case-control study on skin cancer conducted in Southern Europe showed good reproducibility between answers given on two different occasions to a sunlight exposure questionnaire over several different life stages (ICC: 0.68–0.79)23.

Our study has a few limitations. These included the difficulties in obtaining the detailed information of time spent in the sun for each activity and sun protection methods used during the 4 life stages13,17. There were no experts in the field of skin cancer or dermatology for our content validity assessment, but the diverse backgrounds of the expert panel and the consistent results from the experts support the good content validity of the questionnaire. Our study population was selected from an ongoing population-based cohort study of premenopausal women recruited from stratified-cluster sampling of housing estates in Shatin (a density populated district in Hong Kong). Therefore, the generalizability of this study to other age groups or gender is limited.

Our study also did not evaluate the concurrent validity of the questionnaire, but correlations between questionnaires and objective measures are usually not strong13. A study investigated the self-reported versus observed sun habits in beachgoers in Honolulu found correlations of 0.54 to 0.72 between self-reported use of sunscreen and objective measurement of sunscreen use; and correlations of 0.11 to 0.79 between self-reported and observed use of clothing24. Attempts have also been made to validate lifetime sunlight exposure questionnaires with measures on sun damage to the skin for studies of multiple sclerosis22 or skin cancer23. However, the correlation values between questionnaires and objective measures from these studies were also not high probably due to measurement errors or small numbers of cases. In contrast, our study has sufficient validity and good reliability to measure lifetime sunlight exposure.

Accuracy of recall is a concern. The reproducibility of sun exposure-related questions has been examined in a number of studies over time periods ranging from a few weeks to several years22,23,25. In general, these studies have found evidence that sun exposure and outdoor activities, whether in childhood and adolescence, or recent years, could be reasonably recalled. In spite of these limitations, to our knowledge, this is the first study to assess the content validity, construct validity and reproducibility of the Chinese version of lifetime sunlight exposure questionnaire.

Future studies might explore the criterion validity of the current questionnaire using UV monitors. More importantly, research into the relationship between diseases such as breast cancer and sun exposure creates new opportunities in diseases prevention. Using our questionnaire could evaluate or determine which life stage(s) of sun exposure is more closely related to the health or disease outcomes. There is much concern among the public with respect to the risks and benefits of sun exposure. Evidence-based recommendations could be generated on the achievable and beneficial sunlight exposure rather than complete avoidance15,26,27.

In conclusion, this study suggests that lifetime sunlight exposure questionnaire developed in the present study has sufficient content validity, construct validity and good reliability to measure lifetime sunlight exposure among Chinese women in Hong Kong. This questionnaire could be applied to assess the association between vitamin D from sunlight exposure and health outcomes.

Methods

Development of sunlight exposure questionnaire

The sunlight exposure questionnaire (Supplementary Information) was developed based on literature on sunlight exposure assessment13,14,17 and also on existing sunlight exposure questionnaires that have been used in several previous epidemiological studies of cancer15,16. A total of 62 items (questions) were included in the questionnaire covering the skin reaction to sunlight exposure (always/easily burns; burns rarely/never; always/easily tans; tans rarely/never); frequency and duration spent outdoor in the sun over lifetime (6–12 y, 13–19 y and 20–34 y and from 35 y to present); and personal protection used. Sunlight exposure questions covered for each age group included the following items: place of residence (country, province and city); outdoor activities in the sun and the average hours per day spent in the summer and the other 3 seasons; usual (≥50% of the time) sun protection used in each season; trips to summer climate in winter and the frequency (none, once every 3–4 years, once every 2 years, every year); or under shade when outside in summer and specific outdoor activities from ages 35 yr to present; outdoor jobs in the sun and information on frequency (hours per day) and duration (number of weeks and months per season and total number of years); sunlamp or sunbed use (with age when first used, age at last used and total number of sessions over lifetime).

Content validity

To assess the suitability of questionnaire items for use in a Chinese population, the content validity of the sunlight exposure questionnaire was evaluated by a panel of 6 experts with discipline in women's health, lifestyle and health, environmental health and physical education. The experts were academicians/professionals with relevant experiences between 2 and 25 years (mean [standard deviation; SD] 8.3, 16.1) in research or work on sunlight exposure, physical activity or nutrition. The range of experiences provide a wide and relevant perspective on the appropriateness and validity of the items to be included.

Procedures of the content validity assessment

The experts were provided with a delineation of the full content domain of the questionnaires, with specific questions pertaining to the content relevance of each item, was derived based on the ratings of the content relevance of the questionnaire items. For the level of each question's validity, a 4-point ordinal rating scale was used, where 4 point means “very relevant”; 3, “somewhat relevant”; 2, “hardly relevant” and 1, “totally irrelevant”.

The index of content validity (CVI) was derived based on the ratings of the content relevance of the questionnaire items. The CVI is the proportion of items (questions) in the questionnaire that received a rating of 3 or 4 by the experts28. Only items with a CVI of 0.83 or above were retained in the questionnaire28, meaning items regarded as valid by more than 80%of the experts, were selected as significant28. Other items were eliminated or revised according to the literature and suggestions from the expert panel.

The experts were also asked to identify any areas that might have been omitted and to suggest any areas requiring improvement or modification.

Test of construct validity and reliability

Two pretests (n = 5 and 22 respectively) were carried out among women aged 27 to 51 years to test the flow and comprehensibility of the questionnaire. The pre-tested questionnaire was adopted for use in the telephone survey conducted from January to April 2010. Participants were invited from an ongoing population-based cohort study of premenopausal women previously recruited through stratified-cluster sampling from different housing types in Shatin, Hong Kong. A letter was first mailed to the study participants explaining the study aim and then, followed by a telephone call for arranging the telephone interview. Among the 676 potential participants, 5 were excluded due to invalid telephone numbers (0.74%). Of the remaining 671 women, 650 participated in the telephone survey, with a response rate of 96.9%. To test the questionnaire reliability, the sunlight exposure questionnaire was re-administered after 2 weeks among a 15% random sample of the study participants (n = 94).

The study was approved by Survey and Behavioural Research Ethics Committee of Chinese University of Hong Kong. Informed consent was obtained from all subjects.

Statistical analysis

The construct validity was estimated based on the questionnaire interview among the 650 women who answered the sunlight exposure questionnaire. Principal components analysis was used to analyze the factorial structure of the continuous variables from the sunlight exposure questionnaire. To eliminate the effect of different measurement units on the results, the variable (x) was standardized (z) using the theoretical (population) mean and standard deviation: , where μ = E(x) is the mean and σ = the standard deviation of the probability distribution of x29, z is the standardized value of x.

Reliability of the continuous variables in the questionnaire between the first and second interview (n = 94) was assessed by calculating intraclass correlation coefficient (ICC)30. Agreement rates by dichotomous and ordinal variables for the questionnaire were estimated by Cohen's Kappa (k)31. The values of k and ICCs less than 0.40, 0.40–0.75 and greater than 0.75 were considered to indicate poor, moderate to good and excellent agreement, respectively32.

Statistical significance was defined as two-sided P < 0.05. All statistical analyses were done using SPSS version 16.0 for windows.