The development process of the DA was based on the IPDAS guidelines for DA development [30]. In line with these guidelines, both end users’ (ie, smokers) and clinicians’ needs were assessed. In addition, we asked scientific smoking cessation experts for extra input as smoking cessation is often done without consulting clinicians — especially in the Netherlands [8]. In the end user needs assessment, 20 interviews were conducted to assess end users’ needs regarding potential cessation assistance’s characteristics that should be described in the DA and DA functions (eg, a knowledge quiz). The input of the experts was gathered through 2 Delphi studies (24 scientists and 38 clinicians completed the final round). The input of these 3 qualitative studies was used to inform the design and content of the DA, supplemented by the IPDAS background papers (eg, [31]) as well as other relevant literature in the field (eg, [23]) and in consultation with established experts in the field. The SDT served as the theoretical framework. After this, a first prototype was developed that was alpha-tested in September 2019 with potential end users (n=3) and Dutch smoking cessation experts (n=8). During the alpha test, participants were asked to focus on the content of the DA. The end users alpha-tested parts of the DA, while the smoking cessation experts tested the whole first version. Before the beta test, the DA was evaluated by 1 of the co-authors (JR) with regard to the comprehensibility of the text for people with limited health literacy. On the basis of that evaluation, some sections and sentences were rephrased or simplified.

In December 2019, after the alpha test, the DA was beta-tested among 15 experts and end users: 5 smoking cessation counselors, 5 eHealth experts (of whom, 2 had digital DA experience), and 5 potential end users (ie, people motivated to stop smoking). Respondents were given access to the DA to assess the DA’s usability using the heuristic evaluation method (smoking cessation counselors and eHealth experts) and the think-aloud method (potential end users) [32].

Using the heuristic evaluation method, smoking cessation counselors and eHealth experts were asked to evaluate the DA against a list of recognized usability principles: (1) use simple and natural dialogue, (2) speak the user’s language, (3) minimize memory load, (4) be consistent, (5) provide feedback, (6) provide clearly marked exits, (7) provide shortcuts, (8) provide good error messages, (9) prevent errors, and (10) provide help and documentation. The smoking cessation counselors and eHealth experts were asked to use predetermined scenarios (eg, whether cessation attempts have been undertaken before) during the evaluation.

With the think-aloud method, end users were asked to complete the DA while verbalizing their thoughts. The think-aloud method is considered particularly useful in understanding processes of cognition and is considered to be of high value in evaluating an intervention’s design on usability flaws [32].

Data gathered through these tests were compared and compiled into a summary describing the usability flaws of the DA. Based on these results, adjustments to the DA were identified, which can be seen in Textbox 1.

An interactive Dutch DA with an explicit VCM and tailored advice that facilitates the process of choosing an evidence-based cessation assistance was developed. The DA follows a stepwise approach and is designed to encourage a decision that is autonomous and consistent with participant’s values and smoking behavior. The DA consists of 8 sequential sections: (1) information section, (2) optional knowledge quiz, (3) brief smoking assessment, (4) intuitive decision, (5) intermediate advice, (6) explicit VCM, (7) tailored advice, and (8) access information.

In step 1, the information section explains the decision at hand, as well as all the available cessation assistance in the Netherlands. The following topics are addressed: (1) smoking cessation with and without evidence-based cessation assistance, (2) non-evidence–based cessation assistance, and (3) evidence-based cessation assistance.

For the first topic, differences in smoking cessation outcomes are described between people using and not using evidence-based cessation assistance using simple frequency formats (eg, “By using one or more of these stop methods, 15 out of 100 people are still smoke-free after one year.”).

For the second topic, a list of various non-evidence–based cessation assistance tools (ie, acupuncture, laser therapy, hypnotherapy, mindfulness, smartphone apps, self-help books, and e-cigarettes) are described, and a remark is made that the use of these tools is not advised.

For the third topic, evidence-based cessation assistance is described (Table 2), including behavioral support in general, in which users can choose to read more detailed information about all possible options (ie, face-to-face counseling, counseling over the phone, group coaching, and eHealth); NRT in general, in which users can choose to read more detailed information about all possible options (ie, nicotine patches, nicotine gum, nicotine lozenge, nicotine mouth spray, and nicotine inhaler); prescription medication in general, in which users can choose to read more detailed information about all possible options (ie, varenicline, bupropion, and nortriptyline); and cessation assistance during pregnancy and breastfeeding. Users can compare all cessation assistance options using an option grid and are informed about combinations of multiple cessation assistance options. In addition, costs and reimbursement are described, and sources used to inform the information section content (ie, reports, scientific manuscripts, websites) are provided.

In step 2, users can choose to complete the knowledge quiz or skip the knowledge quiz and proceed to the following step. If they complete the knowledge quiz, they will receive the results, see the correct answers to the questions that they answered incorrectly or could not answer (seeing the correct answers can also be skipped even if a respondent opts to do the knowledge quiz), and proceed to the following step.

The brief smoking assessment in step 3 consists of 2 closed (yes/no) questions: “To ensure that our advice suits you best, we would like to ask you if you smoke more than 10 cigarettes (or other tobacco products) on a normal day?” and “To ensure that our advice suits you best, we would like to ask you if you have ever made one or more smoking cessation attempts in the past that were unsuccessful?”

In step 4, an intuitive decision occurs between different clusters of cessation assistance tools: behavioral support; NRT; combination of behavioral support, NRT, and prescription medication; combination of behavioral support and NRT; combination of behavioral support and prescription medication; other, non-evidence–based cessation assistance; and no cessation assistance at all.

In step 5, intermediate advice is provided for users to reevaluate their intuitive decision in step 4 when they chose behavioral support only or NRT only, while also affirming at least 1 of the 2 questions in step 3 (indicating that they smoke more than 10 cigarettes on a normal day or have made one or more smoking cessation attempts in the past): They are advised to consider using a combination of behavioral and pharmacological cessation assistance tools. Intermediate advice is also given to users who chose non-evidence–based cessation assistance or no cessation assistance at all regardless of their answers in step 3: They are advised to consider using one or more evidence-based cessation assistance tools.

The explicit VCM in step 6 is provided for users that chose evidence-based cessation assistance tools in steps 4 or 5; users are asked to rate certain statements regarding cessation assistance characteristics (eg, “I prefer a stop method that works better, even if that means that I have to leave the house.”). These statements are also described in Table 2. Users only rate statements for options that belong to the cluster of cessation assistance options they selected in the previous step.

Step 7 is tailored advice based on the explicit VCM and an optional ranking of all options. The advice is given only when it is possible to give clear advice (ie, if users’ scores do not suggest that more than 2 cessation assistance tools are suitable based on their indicated values).

Step 8 involves accessing information on how to obtain the chosen cessation assistance (eg, nicotine patches).

Above and beyond these 8 steps, framing throughout the DA is positive and autonomy-supportive [33] (in line with the SDT as theoretical framework) to support users’ need for autonomy (eg, we refrained from using fear appeals [34] and controlling language [12,33]). We also clearly communicate that users can make their own choice at the end and included “cues” to support this process (again in order to support users’ need for autonomy; eg, by stating that the advice is only a recommendation and not a command and that it might be a good idea to discuss this with either the social environment, a health care provider, or both). A flow chart of the DA can be seen in Figure 1. The digital DA is hosted by the Dutch company OverNite Software Europe BV, and we make use of their product “TailorBuilder” [35]. Translated screenshots of different sections can be seen in Figure 2 (example of the information section) and Figure 3 (example of the explicit VCM).

An information section was predominantly included as relevant knowledge is needed to engage in high-quality decision making [36] and as it is regarded as an active ingredient by the IPDAS [37,38]. Given this fact, and as potential end users indicated that they would like a knowledge quiz within the DA, a knowledge quiz was included at the end of the information section. Users are thus able to check for themselves if they have the necessary knowledge to engage in the decision process and can obtain additional information if desired. The content of the information section was mainly based on Dutch guidelines (eg, [39]), supplemented by various Cochrane reviews (eg, [40]), online information from the Dutch National Health Care Institute [41], and the Dutch Medicines Evaluation Board [42].

An intuitive decision to limit the choice set for the explicit VCM was included to both prevent choice overload [43] and to include intuitive decision-making processes in the DA. Choice overload describes a scenario in which the difficulty of the decision to be made is greater than the cognitive means of the person faced with the decision, due to too many options. Many DAs focus on a rather small set of available options [44] (eg, deciding between intensive or less intensive aftercare after breast cancer treatment [45]); yet, a cessation assistance DA has to include more options to present all available possibilities. Considering that the comparison of too many characteristics of too many options can lead to choice overload, which can in turn lead to worse decisional outcomes [43], in step 4 of the DA we decided to have users intuitively choose to go on with a limited choice set (eg, only behavioral support). Another benefit of this approach is that users are encouraged to make use of both deliberative and intuitive decision-making processes. While many DAs focus on facilitating deliberation, de Vries et al [36] have argued that combining those 2 strategies can lead to improved decisional outcomes. Importantly, however, users receive information about all possible options before they intuitively make a first, broad decision as knowledge is regarded as a necessary prerequisite for high-quality intuitive decision making [36].

As the Dutch smoking cessation guideline indicates that a combination of both behavioral and pharmacological support should be considered if smokers smoke >10 cigarettes per day or if many of their past cessation attempts failed [39], users that indicated in step 3 that this applies to them are provided with information on the improved effectiveness of a behavioral and pharmacological combination. Of course, even then, they can autonomously decide to continue with another cessation assistance set — in line with their preference and the SDT. In case smokers choose to use either no cessation assistance or no evidence-based cessation assistance, they are also reminded that evidence-based cessation assistance will probably lead to more successful cessation outcomes, but if they choose not to change their mind, they are directed to the last step (ie, access information) if they decided to use no cessation assistance or directly to the end if they decided to use no evidence-based cessation assistance.

We decided to use an explicit VCM in the intervention group as VCMs are regarded as active ingredients by the IPDAS [20,37], and while conflicting evidence exists [46], an overwhelming amount of research has shown that interactive methods to clarify people’s values seem to result in better decisional outcomes [22], especially in the long run [21] and if people are supported in understanding the implications of their clarified values [23,24]. This is why we decided to integrate individual tailored advice that is supplemented with information on how to obtain the different cessation assistance options to remove barriers for the users.

Before conducting any analyses, we will assess the scale quality of all measurement instruments where appropriate, following the 2 steps proposed by Crutzen and Peters [62]: (1) conformation of intended structure of the measurement instrument by means of exploratory factor analyses and (2) calculation of omega [63] as a less biased alternative to (Cronbach) α. Compared to α, omega has more realistic assumptions regarding variances of and covariance between items [64]. Omega_hierarchical is based upon the sum of the squared loadings of items on the general factor and reduces the risks of misjudging the internal consistency of scientific scales [65]. Values will be calculated with R [66], making use of the integrated development environment RStudio [67].

To assess the effects of the DA on dichotomous and continuous (primary and secondary) outcome measures, logistic and linear regression analyses will be conducted, respectively. Mediation analyses will be conducted to determine whether the effects of the DA on smoking abstinence are mediated through cessation assistance use. All analyses will include covariates that were selected a priori (as recommended by Gruijters [68] and De Boer et al [69]), if these are also associated with the 3 outcome measures as described in our hypotheses (ie, smoking abstinence, evidence-based cessation assistance use, and decisional conflict) within our sample. Demographic factors (ie, age, gender, and education) were selected for all 3 outcome measures. The Revised Fagerström Test for Nicotine Dependence was also selected for the smoking-related outcome measures, whereas stage of decision making was selected for decisional conflict. More information on the rationale for selecting those covariates can be found in Multimedia Appendix 1. Results will be reported from both the fully adjusted as well as crude analyses [68]. These analyses as well will be done with R [66], making use of the integrated development environment RStudio [67]. To test the robustness of the results, all analyses will be conducted according to 3 different approaches (if applicable): analyses based on (1) worst case scenario (dropout respondents are considered not to have changed), (2) multiple imputations, and (3) complete cases only. We will also test whether selective dropout has occurred in order to subsequently minimize the bias that this can cause.

An economic evaluation will be performed from a societal perspective with a time horizon of 12 months. Both a cost-effectiveness analysis and a cost-utility analysis will be conducted. For the cost-effectiveness analysis, 2 incremental cost-effectiveness ratios will be calculated, based on the cost per abstinent respondent and the cost per individual who uses evidence-based cessation assistance. For the cost-utility analysis, an incremental cost-utility ratio will be calculated, based on the ICEpop CAPability measure for Adults, as proposed by Smit et al [70]. Uncertainty will be accounted for by bootstrapping and several univariate and multivariate sensitivity analyses. Cost-effectiveness acceptability curves will be constructed, showing for varying willingness-to-pay thresholds the probability that the DA with an explicit VCM and tailored advice is cost-effective compared to the DA without an explicit VCM and tailored advice.