Outsmart HPV was developed using a framework that included aspects of the Protection Motivation Theory [35], Information-Motivation-Behavioral Skills Model [36], and the Minority Stress Model [37]. The intervention for the Out-U group will consist of 2 components: (1) population-targeted, individually tailored Outsmart HPV content on the Web and (2) unidirectional monthly HPV vaccination reminders sent via text messages. The Web-based content will be presented in 4 sequential sections (Figure 2) that contain infographics, other visual formats, and tailored testimonials. The 4 sections are as follows:

The monthly vaccination reminders for the Out-U group will be unidirectional, meaning that participants will not have the option to respond to the reminders. The unidirectional reminders will be sent via an automated process and contain only text. The text will provide information about HPV and the HPV vaccine, steps for getting vaccinated (eg, making an appointment with a health care provider), and content addressing common barriers and concerns about the HPV vaccine. After completing their posttest survey, participants will have the ability to print or save a version of their Action Plan. Participants will also be able to log back into the project website and review the Web-based content throughout the study, except when completing surveys.

The intervention for the Out-I group will consist of 2 components: (1) population-targeted, individually tailored Outsmart HPV content on the Web and (2) interactive monthly HPV vaccination reminders sent via text messages. Thus, the interventions for the Out-I and Out-U groups will be identical except in the type of monthly vaccination reminders sent to participants: participants in the Out-I group will have the option to respond to and/or ask questions to obtain additional tailored information and resources. Research suggests that interactive reminders may be more preferable and effective than unidirectional reminders [41,42], which could be partly because of their ability to obtain additional information/resources and increased participant engagement. Communications for the interactive reminders will be sent through a combination of automated and manual processes, as described further below. To increase participant engagement, some of the interactive reminder text messages will also contain a meme or brief animation in Graphics Interchange Format (GIF). Similar to the approach taken in other technology-based interventions with adolescents and young adults [43], the memes and GIFs are intended to be funny or motivational and reinforce behavior. The text portion of the messages will focus on areas similar to those described above for the Out-U group.

For each month, the initial text message sent to participants will be automated and include a question that prompts a response (Figure 5). Including a prompted message can increase participant responsivity compared with texts with unprompted messages [44]. Depending on how a participant responds to this initial communication, additional automated communications with information and resources will be sent to participants. Participants will then have the opportunity to text further open-ended questions that they may have to the study team (participants can ask as many questions as needed). A study team member will review the received questions and manually send an appropriate response based on a library of example responses that has been developed for this study. Responses to unanticipated questions not included in the library will be discussed and agreed upon by members of the study team. Participants will be notified that it could take up to a few days for the study team to respond to questions.

Participants in the control group will receive standard information about the HPV vaccine that is closely modeled after the Vaccine Information Statement (VIS) for the HPV vaccine. The VIS is created by the Centers for Disease Control and Prevention to provide easy-to-understand information about the HPV vaccine that is publicly available [45]. We modeled the control group content after the VIS because health care providers are required to give a copy of the VIS to patients before vaccination [45]. We formatted the VIS content to match the color and font scheme of the project website. Similar to the Out-U and Out-I groups, participants in the control group will have the ability to print or save a summary of their viewed content after completing their posttest survey, and they will be able to log back into the project website and review their materials throughout the study (except when completing surveys).

Our primary outcomes will include both HPV vaccine initiation (ie, receipt of 1 or more doses of the HPV vaccine series) and completion (ie, receipt of all 3 doses recommended for this age range [13]) during the 9-month follow-up period. The 3- and 9-month surveys will assess how many HPV vaccine doses participants have received. To help maximize the quality of data on our primary outcomes, HPV vaccination status will be confirmed by and based on data from medical records for all participants who complete an MRR form. For those who do not complete an MRR form or who remain unvaccinated during the study, HPV vaccination status will be based on self-reported data from these surveys.

The results of our pilot test suggest that several theoretically informed constructs may mediate the intervention’s effects on vaccination outcomes [32]. Guided by these previous findings, we will examine several constructs as potential mediators, including perceived vulnerability to HPV-related disease (3 items with a 4-point response scale ranging from no chance to high chance), perceived severity of HPV-related disease (3 items with a 4-point response scale ranging from not at all to very), and response efficacy (eg, perceived effectiveness of the HPV vaccine; 2 items with a 5-point response scale ranging from strongly disagree to strongly agree). Additional constructs will include rewards of the maladaptive response (eg, negative social norms of getting the HPV vaccine; 2 items with a 5-point response scale ranging from strongly disagree to strongly agree), self-efficacy to get the HPV vaccine and to talk with a health care provider about the HPV vaccine (2 items with a 5-point response scale ranging from strongly disagree to strongly agree), response costs of getting the HPV vaccine (ie, perceived barriers; 2 items with a 5-point response scale ranging from strongly disagree to strongly agree), and intention to get the HPV vaccine (1 item with a 5-point response scale ranging from strongly disagree to strongly agree). We will also assess knowledge about HPV and the HPV vaccine (7 items with response options of yes, no, and I don’t know), worry about getting HPV-related disease (3 items with a 4-point response scale ranging from not at all to a lot), and stigma around HPV (3 items with a 5-point response scale ranging from strongly disagree to strongly agree). These constructs will be assessed at each survey time point with existing measures and those used in our past studies [17,28,32,46,47].

The baseline survey will assess demographic and health-related characteristics that we will examine as potential moderators of intervention effects. Demographic characteristics will include participants’ age, race/ethnicity, socioeconomic status, relationship status, sexual orientation (eg, sexual identity, behavior, and attraction), and urbanicity. Health-related characteristics and experiences will include health insurance status, health care utilization, perceived experience of discrimination in receiving health care, disclosure and concealment of sexual orientation to their health care provider, sexual history (eg, number of sexual partners and age at first sexual intercourse), history of HPV-related disease, and HIV status (as some HIV-positive participants may not be aware they can receive the HPV vaccine [13]). The baseline survey will also assess participants’ attitudes about vaccines in general [48] and electronic health literacy [49] (which may affect participants’ understanding and perceptions of the study materials).

The posttest survey will assess acceptability of the Web-based content about the HPV vaccine across the 3 study groups [50]. Survey items will ask about participants’ perceptions of the information viewed (eg, understandability and relevance) and quality of the website (eg, appearance and usability). The 3-month survey will ask participants in the Out-U and Out-I groups about their acceptability of the Action Plan, and the 9-month survey will ask these participants about the acceptability of the monthly vaccination reminders (eg, understandability and helpfulness). Study surveys will also examine participants’ communication and information seeking about the HPV vaccine.

We will measure engagement using administrative data from the project website about study processes and participants’ interaction with study materials (ie, paradata). This information will assist in examining whether dosage influences intervention efficacy, which in turn will help inform future dissemination efforts [51]. Paradata will include counts of website log-ins, duration of each log-in, parts of the website visited, duration on each part of the website visited, and functions utilized. For the Out-I group, these data will also include the number and nature of communications that occur during the interactive vaccination reminder process.

We will examine descriptive statistics for participants’ baseline characteristics and make comparisons across study groups. Using an intention-to-treat approach, we will then use logistic regression models to determine intervention efficacy, with separate models for HPV vaccine initiation and completion. For each outcome, we will perform all pairwise comparisons across study groups. Logistic regression models will produce odds ratios (ORs) and 95% CIs. Sensitivity analyses will include adjusting for possible residual confounders and a tipping point analysis and multiple imputation to examine the impact of any missing outcome data.

We will examine theoretical constructs as potential mediators using structural equation modeling. We will examine HPV vaccine initiation and completion separately and use a probit link to transform each binary outcome to a standard normal scale [52]. For each outcome, we will fit 2 models using a maximum likelihood approach: a full model and a more parsimonious final model. The full model will include the study group and all theoretical constructs examined. We will assess the overall model fit of the full model through the root mean square error of approximation (RMSEA) and other metrics (eg, chi-square test) [53,54]. We will consider an RMSEA of less than 0.07 to indicate an acceptable fit [54]. To create the final model, we will remove pathways (ie, associations) from the full model based on effect size and the chi-square test for nested models. This process will continue until removing pathways no longer improves the model. The resulting model will be the final model. Structural equation models will produce standardized path coefficients that estimate the direct and indirect effects of study group and the theoretical constructs.

To examine moderation, we will use logistic regression models that include an interaction term between each potential moderator and study group. These logistic regression models will produce ORs and 95% CIs. We will consider moderation to be present if an interaction term has P<.05.