Data on the respondents’ DNA damages caused by sun exposure were obtained through analysis of urine samples. The most frequent DNA damages caused by UVR is the formation of thymidine dimers, which are believed to be a risk factor for all skin cancers [11, 12]. Thymidine dimers were previously measured using an isotope method, which is no longer available [13]. Here a new method using liquid chromatography–mass spectrometry (LC-MS) was used [3]. The method detects thymidine dimers only. The technical procedures of the urine analysis are described in an additional publication on another part of the study by Lerche and colleagues [14]. Alongside using this new method, we conducted this mixed methods study using qualitative interviews and quantitative risk scoring to explore risk perceptions related to skin cancer among the study’s participants. See Fig. 1 for an overview of the study framework.

Participants were recruited online. An advertisement invited 35–55-year-olds in the Copenhagen area travelling on beach holidays of minimum seven days during February/March 2023 to the study. All skin types were welcomed. The participants received DKK 1500 for their participation, given they provided urine samples before and after travelling on holiday and participated in an interview about risk perceptions related to skin cancer.

Recent sun bed users as well as individuals, who had been on a ski or beach holiday within the last four weeks were excluded, as recent UVR exposure immediately preceding the experiment would disrupt the results. Pregnancy, prior or current skin cancer, as well as using medicine, which could interfere with DNA repair processes or increase photosensitivity were also exclusion criteria.

Ninety people initially signed up to participate in the study. 27 met the criteria of the study and were enrolled. 20 participants proceeded to the qualitative interviews. Seven opted out of the qualitative interview, due to lack of DNA damage detected (n = 3), failure to complete the required tasks of the experiment (n = 1) or inability to schedule an interview within the timeframe (n = 3) (See Fig. 2).

Flow diagram of participant enrollment

The group consisted of 20 Danes aged 36-56 years with a mean age of 47 years, all living in the Greater Copenhagen area. 16 individuals identified as women and four as men. Most of our participants had a light skin tone, as it is typical to ethnic Danes. Using self-reported Fitzpatrick classification of skin types 45% (n = 9) had skin type II, 35% (n = 7) had skin type III, and 20% (n=4) had skin type IV [15]. The majority had long-term higher education (>4½ years) (n = 8) or medium-term higher education (2-4½ years) (n = 6). Most participants had children and travelled with their families (n = 17). The remaining participants (n = 3) travelled with friends or partners. The destinations were spread over four continents, and all had a UV-index above five during February/March 2023 with a holiday duration ranging from 7 to 23 days (see table 1). In terms of latitude, the destinations were located from 28° North (Grand Canaria, Spain) to 33° South (Cape Town, South Africa), but with the majority of the destinations around 15° from Equator.

The study was approved by the Danish Research Ethics Committee (H-20076172) and the Danish Knowledge Center for Data Reviews (P-2021-591). The study was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans, informed consent was obtained, and data anonymized.

We conducted semi-structured qualitative interviews with a duration of 30–60 minutes. The interviews were based on an interview guide covering: everyday life and social relations, the activities and weather during the beach holiday, behaviour in the sun, perceived benefits vs. barriers of protecting your skin from UV radiation, as well as the participants’ sense of self-efficacy in terms of using sun protection, and their perceived seriousness of skin cancer. Eventually their perception of personal risk of skin cancer before biofeedback information, and reactions to biofeedback information, and perception of risk of skin cancer after the given biofeedback information were assessed. All interviews were carried out by an experienced interviewer and an observer.

Baseline perception of personal risk of skin cancer before biofeedback information was firstly assessed through an open-ended questioning technique asking the participants to reflect qualitatively on their sense of susceptibility to skin cancer. Afterwards, we handed them a document with a simple scale illustrated as a horizontal line with numbers from 0 to 10, where 0 = low risk and 10 = high risk of developing skin cancer. The participants were asked to score and mark on the scale their own perceived risk (susceptibility) of developing skin cancer ranging from 0 to 10. The participants assessed their individual risk on the scale from a subjective perspective based on, e.g. skin type, family history of skin cancer, knowledge about skin cancer susceptibility, their history of sun protective behaviours, as well as individual aspects generally affecting perceptions of disease risk such as personal anxiety levels and general health history. In such, participants with a sensitive skin type, family history of skin cancer, history of sun burns, lack of use of sun protection, general cancer worries, etc. generally assessed their pre-biofeedback information risk of developing skin cancer as higher compared to participants with fewer of such experiences.

The individual biofeedback information based on the participants’ results of their urine samples showing DNA damage (cyclobutane pyrimidine dimers, with thymidine dimers being the most common) was given during the interview immediately following assessment of the baseline risk perception. The biofeedback information was given by the researcher responsible for the urine analysis. The participants were just told that skin DNA damage had been identified in their urine samples after their holiday, and that this was DNA damage deriving from UVR exposure on their holiday, as their initial urine sample immediately before their holiday was free from or very low in DNA-repair products. Unfortunately, at the time of the interviews it was not possible to provide our participants with a meaningful quantified measurement or level of the detected DNA damage, because we had no background average as a basis for comparison, as background data is still sparse given the novelty of the method. The urine analysis and interviews were conducted continuously as the participants returned home from their vacations during a period of 4.5 weeks, and therefore, a comparison between the participants’ levels of DNA damage was unfortunately not possible until after completion of the experiment.

During the interview and the disclosure of biofeedback information participants had the opportunity to ask questions and get more detailed explanations about the detected DNA damage. After disclosure of the biofeedback information, the researcher left the interview room, and the interviewer proceeded with the interview, recording the participants’ reactions to the biofeedback information, and assessing the post-biofeedback perceptions of risk of skin cancer. Again, we used the scale from 0 to 10, where the participants once more were asked to quantify their own subjectively perceived risk of skin cancer after being informed about their UVR-induced DNA damage. Lastly, we explored the participants’ motivation for intended behaviour change considering the disclosed biofeedback information.

The qualitative data analysis followed the procedures described by Steinar Kvale [16] and corresponded with the key criteria in the standards for reporting qualitative research (SRQR) [17]. The 20 interviews were recorded and transcribed. A coding system was developed based on the themes from the interview guide. After reading through the transcribed interviews, a total of 21 categories and subcategories were developed. The transcribed interviews were coded using the software NVivo12. This allowed for a systematic overview of data, enabling comparisons to be made. Subsequent analysis of the coded data was based on identifying patterns in the material and relating such patterns to the theoretical framework of The Health Belief Model [10].

When comparing the aggregated summary of 20 participants’ risk scores before and after disclosed biofeedback information means (95 C.I.) are provided. Differences in means were tested with Χ2. For all tests, p values <0.05 were considered statistically significant.

The Health Belief Model (HBM) was used as analytical framework. The HBM was developed in the early 1950s by social scientists with the aim of better understanding why people fail to adopt disease prevention strategies. The HBM suggests that a person's perception of threat of an illness or disease together with a person's belief in the benefits, barriers, and self-efficacy of a recommended health behavior will predict the likelihood of the person adopting the behavior. Also, the HBM suggests that individuals are more likely to follow health recommendations if they are exposed to cues to action, i.e., events that trigger health-promoting behaviour, as in this case personal biofeedback information. Furthermore, the HBM also includes modifying variables such as age, sex, ethnicity etc., perceived seriousness, and perceived susceptibility of the illness or disease in the model as primary to people’s likelihood of engaging in health-promoting behaviour [18]. In this experiment personal biofeedback information is inserted in the model as a cue to action, while the health-promoting behaviour that we wish to investigate the likelihood of is intended behaviour change in terms of sun protection. In the interviews, the participants’ pre and post biofeedback information risk scores can be inserted in the HBM as their perceived susceptibility to skin cancer. And the biofeedback information as a cue to action shows us how the participants’ perceived sense of threat is affected. Aligned with the model, the qualitative interviews also covered the participants’ perceived seriousness of skin cancer, their perceived benefits vs. barriers of protecting their skin from UV radiation, and their sense of self-efficacy in terms of using sun protection. See Fig. 3.

Components of the health belief model

Figure 3 the Health Belief Model is applied to the experiment’s analytical framework such that personal biofeedback information is inserted in the model as a cue to action. The experiment investigates how the participants’ perceived sense of threat and perceived susceptibility of skin cancer (i.e. here their risk scores), and likelihood of engaging in health-promoting behaviour (i.e. here intended behaviour change in terms of sun protection), is affected by personal biofeedback information.