An online survey including a DCE was administered, between April and May 2023, to 16- to 23-year-olds (n = 300) and parents of 16- to 18-year-olds (n = 300) in the US.

Parents must have at least one 16- to 18-year-old for whom they had arranged at least 75% of visits to primary care providers in the preceding 3 years, and they must have attended at least 50% of visits with the 16- to 18-year-old. Both 16- to 23-year-olds and parents must be able to read and understand English to provide informed consent. Resource parents (i.e., foster or adoptive parents) were excluded.

Potential participants were identified via a participant recruitment database, SAGO, as well as healthcare provider (HCP) referrals, patient associations, and social media advertising (as needed) (see Supplemental file S1). All eligible participants were invited to complete an eligibility screener. Eligible participants who completed the screener and provided informed consent were directed to the online preference survey link. Eligible 16- and 17-year-olds provided their assent in addition to their parents’ consent. The survey was estimated to take approximately 40–50 min to complete.

Good research practice guidelines were applied to the DCE experimental design and the pilot [22, 23] and the analysis methods [24]. The DCE consisted of questions quantifying the relative importance that individuals place on vaccination features and estimated trade-offs. The survey also captured demographic characteristics, and information on meningococcal disease knowledge and vaccination.

Vaccination attributes and their associated attribute levels (Fig. 1) were determined following a targeted literature review [25] and interviews (focus groups) with 16- to 23-year-olds and parents [26]. Descriptions were provided for each attribute in lay terms (Table S1).

Attributes and attribute levels used in the DCE

A pilot phase was conducted with 20 respondents from each group to confirm the feasibility of completing the survey. Learnings from the pilot were incorporated into the final survey. The pilot survey respondents were all included in the final survey samples.

A DCE was used to elicit and quantify relative preferences for attribute levels of treatments.

Trade-offs between hypothetical meningococcal vaccinations were presented independently to allow for estimation of the preference weights for each attribute level (i.e., trade-offs were uncorrelated). The trade-offs were evenly presented across respondents to avoid biasing the preference weight estimates (i.e., trade-offs needed to be balanced). Infeasible combinations of attribute levels were excluded, e.g., the number of injections could only be equal to or less than the number of visits to an HCP.

The experimental design implemented a commonly used D-optimal algorithm to construct a fractional factorial experimental design (statistically chosen subset of choice sets) [27]. This design maximizes information gathered about preferences by optimizing the number of discrete choice pairings required to compare all attribute levels.

To avoid presenting more choice tasks than respondents could feasibly answer, the experimental design was split into three blocks, each with ten trade-off combinations. Choices were not repeated across blocks. A block design was allocated to minimize selection bias, which could help minimize extreme views when choosing different choice options in the DCE task. Each respondent was randomly assigned to complete only one block in the DCE survey.

During the DCE, respondents were presented with an illustrative example of a pairwise choice set consisting of two hypothetical meningococcal vaccinations (Fig. 2). Respondents were presented with different pairwise choices including different combinations of attribute levels from five attributes. Respondents were then asked to choose their preferred vaccination profile (forced choice).

Example of a DCE choice task. DCE discrete choice experiment

Summary descriptive statistics were used to describe the respondents invited to participate in the survey, participant characteristics, and responses to questions on meningococcal disease knowledge and vaccination experience.

For analysis of the primary data, the choice data were analyzed using a random parameter logit model to estimate preference parameters (see Supplemental file 2 for details). The responses to each pair of forced-choice trade-offs were modeled. Data were then analyzed to estimate predicted choice probabilities, i.e., the likelihood that an average respondent would select a specific vaccination profile. This approach allowed for preference weights for all attribute levels included in the survey to be estimated. A higher preference weight describes a more appealing attribute level, and the magnitude of difference between attribute levels represents the relative impact of changes in an attribute.

It was hypothesized a priori that factors such as out-of-pocket costs, lack of meningococcal vaccine awareness, or lack of trust in HCPs may reduce the preference for a meningococcal vaccine, while factors such as CDC recommendations or having an at-risk child may increase the preferences for a meningococcal vaccine. Subgroup analysis was used to determine whether average preferences varied among respondents in 13 mutually exclusive subgroups identified in post hoc analysis (e.g., by education level, household income, risk factors, awareness of meningococcal vaccines, or belief that vaccination interferes with natural protection; see full list in Tables S2 and S3). Preferences were explored to test for systematic differences in attribute preferences among 16- to 23-year-olds and parents (see Supplemental file 2 for details).

Institutional review board approval was received on 11/27/2022 from WCG IRB (IRB Tracking number 20226172). All survey respondents provided informed consent to participate in the study.