The University of Twente gave ethical clearance (registration 210899). All respondents gave online informed consent before starting the survey.

An online cross-sectional survey targeted individuals aged 50–75 years without a lung cancer diagnosis who consider themselves current or former smokers. This cohort follows less strict inclusion criteria than screening trials. Ineligible respondents were identified through introductory questions (age and smoking history) and excluded from the survey. The survey targeted individuals from France, Germany, Italy and the Netherlands and was translated into the respective languages. The full survey is in the Supplementary Material.

The descriptive framework followed an iterative process, starting with the preference elicitation method, followed by the decision criteria. Preferences were elicited using analytic hierarchy process (AHP). AHP, a widely used multi-criteria decision analysis (MCDA) technique, aids preference-sensitive decisions involving multiple criteria. It supports individual or group decision-makers, revealing their own and stakeholders’ preferences [29]. Decision criteria are defined, and pairwise comparisons made, resulting in preference weights [30].

To elicit the willingness of respondents to participate in screening and answer the first research question, direct questions were used. Participants were asked what the likelihood is that they would participate in specific screening programs.

All respondents answered general questions on age, sex, educational level and smoking status, as well as family history for each of the Big-3 diseases. Following smoking status, questions were asked to estimate the pack-years smoked and, if applicable, how many years they have quit smoking. Other general risk-related questions included chest complaints for which they have not seen a physician, perceived 5-year risk of lung cancer, likelihood of smoking cessation within one year and likelihood of smoking cessation if diagnosed with one of the diseases.

Literature-sourced criteria driving screening participation were identified. Discussing key criteria led to two new criteria, resulting in a final list of eight. These criteria are relevant, non-redundant, non-overlapping and independent. The two added criteria were Diseases screened for and immediate feedback. The first is added due to this study’s focus on multi-disease screening. The second is adapted from mammography screening research, where results within 48 h are considered important [31]. Table 1 describes and provides ranges of the final eight criteria.

The preference elicitation section started with a straightforward ranking of the eight criteria, from most to least important. Then, respondents were asked to complete pairwise comparisons using a Likert scale to compare the relative importance of the criteria. Figure 1 shows an example of one of these questions. Only the top 5 criteria from the ranking were used in the pairwise comparisons, which reduced the number of comparisons from 28 to 10. The sub-criteria of diseases screened for (different combinations of Big-3 diseases) were also compared using pairwise comparisons. For the elicitation of preferences, only answers with a consistency ratio smaller than 0.3 (indicating the most consistent answers) were included. Due to the reduction in pairwise comparisons, the applied consistency ratio is at the upper limit of the typically accepted ratios of 0.1 to 0.3.

Example of a preference question and a stated willingness to participate question following the ranking of criteria in the English version of the survey. The full survey is available in the Supplementary Material

To elicit respondents’ willingness to participate in screening, respondents were given more information on the benefits and risks of screening for each of the Big-3 diseases and asked what their likelihood of participation would be for screening for each of the combinations of diseases used as sub-criteria, as shown in Fig. 1. The respondent’s own answer to the likelihood of participating in LCS was provided in the following questions as a reference.

The translated surveys were used in think-aloud interviews to make improvements. Researchers from each country identified 5 eligible respondents to interview, using convenience sampling. A delay in ethical approval in Germany resulted in the survey being filled in by 5 researchers who were asked to imagine that they are part of the target group. The interviewees had to explain their thoughts when answering the survey to ensure that concepts were well understood. The guidelines used by interviewers can be found in the Supplementary Material.

After these interviews, translations of concepts were improved, and the option of indicating that a person smokes “occasionally” was added to be able to more accurately calculate the pack-years.

The online survey was set up in Qualtrics [32] and sent out by Dynata, a global online market research firm (https://www.dynata.com) between 1 and 28 December 2021. Dynata used a multi-sourcing panel recruitment model with a variety of contact methods including loyalty partnerships, apps and emails amongst others, which is a form of convenience sampling. In each country, a target of 330 respondents was set and recruitment was stopped after the quota was reached, which means that some participants were already recruited and could continue filling in the survey. Quotas were implemented to match population demographics per country.

The statistical analysis was conducted in R version 4.2.2 [33]. Aggregation of individual priorities (AIP) with geometric mean was used to report the overall opinion of respondents [34]. The statistical significance (p<0.05) of differences in willingness to participate between subgroups was tested using the Kruskal-Wallis test. Furthermore, pairwise comparisons using the Wilcoxon rank sum test, with continuity correction data, were used to determine the statistical significance between groups.

The analysis could have confounding factors, influencing the factors with significant impact on the willingness to participate in screening. Therefore, we used a multivariate regression model to identify the factors that have a significant influence (p<0.05) on the willingness to participate while also correcting for the influence of other factors. This was used to confirm the importance of different factors on the willingness to participate in screening. The factors included in the multivariate regression to investigate their effect were age, smoking status, presence of already diagnosed COPD or CVD, sex, whether the respondent already had chest complaints, whether the respondent would plan to quit smoking within 1 year (if diagnosed with one of the Big-3), country, perceived 5-year lung cancer risk, education, pack-years and family history of each of the Big-3. Additionally, the interaction effects between the following factors were included: smoking status and the presence of diagnosed COPD or CVD, smoking status and the perceived lung cancer risk, smoking status and education level as well as pack-years and age or educational level.