Vaccination is widely recognized as one of the most effective public health interventions ever [1]. However, despite the well-established safety and effectiveness of vaccines, a growing number of parents are choosing to delay or forgo them for their children because of questions about vaccines’ necessity and safety (also sometimes because of firmly held religious or political beliefs). This phenomenon is referred to as vaccine hesitancy [2,3]. Vaccine hesitancy often results in undervaccination among children and has led to increasing numbers of vaccine-preventable disease outbreaks in the United States over the last two decades [4,5]. For example, measles, which was considered eradicated from the United States in 2000, caused infections among more than 2000 people between 2014 and 2018 alone, with the majority of cases occurring in individuals under- or unvaccinated against the virus [6]. Vaccine-preventable disease outbreaks are associated with significant cost and morbidity, and in some cases, even death [7]. Therefore, it is a public health priority to find interventions to mitigate this trend of parental vaccine hesitancy and childhood undervaccination [8]. This paper reports on the development of a novel, tailored, Web-based intervention to promote timely vaccination by addressing these issues. A more detailed description of the study design and protocol is available elsewhere [9].

Most interventions to increase vaccination developed thus far have focused primarily on correcting knowledge deficits, with the hypothesis that correcting these deficits will lead to improved attitudes and behaviors (ie, parents become less vaccine hesitant and thus are more likely to get their children vaccinated). Unfortunately, the majority of interventions based on this concept have not been effective at increasing vaccination rates [10,11]. This is because, as research has elucidated, parents’ vaccination decisions are multiply determined—based not just on their knowledge about the risk and severity of infectious diseases and the benefits and risks of vaccines but also on trust, emotion, values, past experience, access to health care, and peer influences [12,13].

Given the multiple determinants of vaccination decisions, novel intervention strategies that account for factors beyond knowledge deficits—such as personal values and emotions associated with parents’ individual concerns and barriers to vaccination—are needed [14]. One such approach to address these multiple issues is message tailoring. Message tailoring involves providing customized vaccine-promotion messages based on an individual’s unique beliefs, experiences, knowledge, and barriers to action [15]. Research on tailored messaging in multiple domains shows that by increasing the personal relevance of the information, people are more receptive to new information that may challenge their beliefs. Regardless of whether their knowledge of a given situation is altered by message tailoring, the technique is thought to work by lowering psychological resistance to information or suggested action that may counter what an individual initially thinks or believes [15]. Tailored messaging interventions have been shown to be effective for a number of health behaviors but have not been extensively tested for vaccine promotion [16,17].

In this paper, we describe the development of a Web-based, tailored messaging intervention used in a randomized controlled trial that is ongoing (ClinicalTrials.gov protocol number NCT02665013). This intervention, targeted to new and expectant mothers, was designed to promote vaccination by either reinforcing provaccination decisions among parents who are not hesitant to vaccinate or by decreasing vaccine hesitancy and thus increasing vaccination intention among parents who are vaccine hesitant. To do this, we developed our intervention based on evidence-based health behavior theories and included tailoring in both standard (ie, gender, primary vaccination beliefs, and concerns) and novel (personal values) domains. Although many additional factors could be novel targets for tailoring (trust, emotion, access to health care, etc), values were chosen as the potential new tailoring target for this intervention because, unlike beliefs, experiences, and barriers, which frequently change over time or by situation, values are believed to remain stable over the life course and across contexts [18-21]. Moreover, the behavior of vaccine hesitancy appears rooted in values as well as knowledge, skills, and self-efficacy [13,22-24]. Although there has been minimal research on how to use values to effectively promote vaccination, many researchers in the field recognize its potential importance [23,25-28]. Thus, a secondary goal in developing this values-tailored intervention was to begin to address the knowledge gap regarding the role values may play in creating messages effective for increasing maternal acceptance of vaccines. In addition, although fathers can be an important influence in childhood vaccination decisions, they were not included as the target for this intervention based on our previous research suggesting that mothers were the main vaccine decision makers in the study population of interest [29,30]. In this paper, we describe the overall intervention development process, with particular attention to how the novel values tailoring was developed.

The Web-based, tailored vaccine promotion messaging intervention was developed using a multiphase, iterative, user-centered design process. It was informed by behavioral theory, empirical data from maternal surveys and interviews, and expert and end-user input. When possible, end users’ opinions about intervention design were prioritized over those of the research team. For the intervention trial, we planned that mothers would be recruited to the website between the last trimester of their pregnancy and when their child was less than or equal to 2 months of age (the primary time for vaccination decisions to be solidified [30]). Mothers would then receive additional exposures to the website (with retailored information based on updated beliefs) at three additional time points between enrollment and infant age of 15 months. Vaccination timeliness would be assessed at age 200 days (ie, approximately 6 months) and 489 days (ie, approximately 15 months). Two versions of the website were developed for the randomized trial: a version with messages tailored to the individual participant based on personal characteristics (age, baby’s gender, and pregnancy status) and mothers’ vaccination beliefs and concerns, vaccination values, and intention to vaccinate (described in this paper), and an untailored version that was identical in appearance and content to the tailored version except for the tailored components (described elsewhere) [31].

Methods used to establish the tailored website’s architecture, content, and tailoring included the following:

The theoretical framework for this intervention (Figure 1) was a hybrid of the Theory of Planned Behavior (TPB) [33] and the Values-Attitudes-Behavior (VAB) model [34]. According to the TPB, behavior (in this case, childhood vaccination) is directly influenced by intentions (intention to vaccinate), which are based on one’s attitudes, perceived norms, and perceived behavioral control. In the TPB, perceived behavioral control is a function of control beliefs (eg, beliefs that the decision to vaccinate is in the parents’ power to control) and attitudes are a function of behavioral beliefs (eg, beliefs that vaccinating one’s child will prevent infectious disease and will not harm the child). The VAB theory goes beyond the TPB to posit that personal values are factors influencing attitudes—this is the basis for our hypothesis that values-tailored messaging would be an effective behavior change target in the intervention. We also used select principles of motivational interviewing (MI) [35] and persuasive messaging [36,37] to inform intervention design, aiming to increase the personal relevance of the materials while minimizing reactance. Specifically for MI, the idea of rolling with resistance (not directly counteracting a person’s antivaccination attitudes for example) is believed to decrease a person’s reactance to receiving information that may be counter to their current beliefs [35,38]. Persuasive messaging in this context refers to making the information more personally relevant by tailoring the information to the person’s needs and more trustworthy by providing information that is perceived as accurate and balanced (two-sided messaging) with regard to potential risks and harms related to vaccination.

The setting for this project was Kaiser Permanente Colorado (KPCO). KPCO is a managed care organization in the Denver metropolitan area that maintains an electronic health record with demographic, medical encounter, and vaccination data on all members. Between January 2015 and October 2015, various convenience samples (described below and in Figure 2) of mothers of young children currently enrolled at KPCO were recruited for the design and tailored message testing interviews, usability testing, and surveys.

Informational content for the website was developed by editing content adapted from an intervention used in a previous trial to reflect the most current vaccine information and recommendations [32], and developing new content based on emerging vaccination issues identified by vaccine experts and from vaccine questions received from parents in a previous trial [39]. When possible, content was contextualized to the local Colorado environment and health system (ie, reflected vaccines in the KPCO vaccine formulary) and was written at an eighth-grade reading level. All information presented was evidence based and included updated references from peer-reviewed journal articles and materials from the Centers for Disease Control and Prevention.

To create a personalized Web-based experience for each participant, the intervention was designed such that a preintervention survey would assess individuals’ values, beliefs, and vaccine hesitancy. The Web-based tailoring engine would then use this information to present corresponding tailored messages incorporated with evidence-based informational content about vaccination to each user. The process by which we developed and evaluated the values tailoring and message framing strategies is described in the following sections.

In total, there were 17 general topic areas that the intervention addressed, divided into 41 specific content areas (Table 2). These 41 content areas were further grouped into 6 broad categories corresponding to the 6 standard tiles (ie, not the Just for You tiles) presented on the intervention’s home page (Figure 3). Furthermore, 4 of the 41 content areas were newly developed for the intervention. The remaining 37 were adapted from the previous intervention [39]. In addition, three newly developed, interactive infographics were also included on the following topics: herd immunity (included under the Community Benefits topic), antigen counts in the past and current vaccines (included under the Vaccine Ingredient Types topic), and disease rates before and after vaccines (included under the Risk of Diseases topic). Content areas included in the 6 home page tiles were identical in the tailored and untailored websites.

In this paper, we described in detail how we developed a Web-based, tailored messaging intervention to address maternal vaccine hesitancy using an iterative development process and a mixed method approach. This intervention, which was developed using validated health behavior theories, is expected to be effective, engaging, and easy to use based on end-user feedback and pilot testing. Ultimately, the intervention included common tailoring elements such as demographics and beliefs as well as a novel tailoring target, personal values. In addition to examining tailoring targets, we also used user feedback to assess the potential impact of different message framing strategies in combination with the tailored elements. This was necessary as there is ongoing debate about the optimal messages framing strategy with regard to vaccine hesitancy and immunization [46-51].

Of the 4 health behavior theories and techniques used to inform the development of the intervention, the TPB has the most evidence for its applicability to vaccine decision making. Owing to this, we planned a priori to incorporate elements from this theory into our intervention and therefore did not focus any of our data collection described in this paper on these elements. Specifically, the intervention was tailored based on TPB constructs of attitudes, and when possible, elements of social norms were incorporated into the messages. The other remaining theories and techniques—the VAB theory and elements of persuasive messaging and MI—did not have a strong evidence base with regard to vaccine decision making at the time the intervention was being developed. However, the data collected in this study suggest that each is relevant to the vaccination decision. Specifically, results from our quantitative survey of values and from the qualitative interviews on message framing and values tailoring both suggest that certain values are more important than others with regard to vaccine decision making, and that some values are better incorporated with certain beliefs than others. A subsequent study done by our group examining the role that values play in mothers’ vaccine decision making further supports this notion [24]. Support for incorporating techniques from persuasive messaging into the intervention was also derived from the qualitative message framing and values tailoring interviews. The respondents clearly indicated that vaccine-hesitant parents strongly preferred messages that were perceived as balanced, where both positive and negative information (ie, two-sided messages) about vaccine safety or benefits was presented as these messages were deemed more trustworthy than messages that only conveyed one side of these issues. This point supports the concept in persuasive messaging that messages deemed as trustworthy are more likely to be reflected upon and more persuasive. Also supporting the importance of persuasive messaging was the finding from the message framing and values tailoring interviews that showed that messages tailored to the user were deemed more personally relevant. The design feedback interviews showing that mothers uniformly liked intervention architecture that clearly allowed a user to choose the specific information to view also support this concept. Evidence from our study for incorporating techniques from MI into the intervention is somewhat indirect in that mothers participating in the message framing and values interviews who were following an alternative vaccination schedule did not like black-and-white messages that argued against this practice. Such messages could be considered to counter the MI tenet of rolling with resistance. Another important MI tenet is that of using intrinsic motivation to effect behavior change. Evidence in support of this concept also comes from the message framing and values tailoring interviews showing that mothers generally preferred values-tailored messages to those not tailored to values. Further work by our group and others that occurred after our intervention was developed lends additional support to the important role that MI likely plays in motivating parents to vaccinate [52,53].

The next step in our study is to assess the efficacy of this intervention in a randomized controlled trial of KPCO expectant and new mothers. This trial (ClinicalTrials.gov protocol number NCT02665013) will examine the relative effect of the tailored versus untailored websites for their effect on timely infant vaccine utilization during the first 15 months of life. In addition, a number of secondary outcomes will also be assessed including whether the intervention modified maternal vaccination beliefs and concerns or vaccine hesitancy, and how any changes in these outcomes relate to vaccination values. This will allow for a more thorough investigation of the VAB model used as a basis for this study.

Tailoring on values represents a novel, and potentially important, innovation incorporated into this intervention. A large body of literature in the social sciences provides compelling evidence that aligning educational messages with personal values can make information more salient and actionable [54-57]. The somewhat more limited number of studies examining values tailoring in the realm of health behavior change specifically further support this concept [58-61]. Recently, researchers interested in improving immunization delivery have begun to recognize the potential role of values tailoring in promoting vaccination [23,27,62]. Although there has been limited prior research examining the impact of values tailoring on vaccine acceptance, our own study [24] and that of others [63], which were completed after the study presented in this paper, have shown that differences in personal values are associated with variation in the acceptance of recommended vaccines. Taken together, values appear to be a potentially untapped tailoring resource that warrants further exploration. We believe results from the randomized controlled trial that will evaluate this intervention will add important and novel information to this growing body of research.

This paper should be considered in the context of some important limitations. First, when developing the various aspects of the intervention, we generally received input from only a small number of mothers, most of whom were white and non-Hispanic. Maternal input was not designed to be comprehensive. Rather, we opted to obtain in-depth information from a handful of mothers to provide more nuanced insight for optimizing the intervention and making it more relevant to potential end users, and we generally solicited mothers’ input until thematic saturation was reached (although we did not do a formal qualitative analysis). In addition, we focused much of the intervention development on the incorporation of values as a tailoring element, an approach to tailoring that is in need of further study. Moreover, although the study population providing input into the interventions’ design was diverse with regard to demographic characteristic such as race, ethnicity, education, and income, the population was relatively homogeneous with regard to insurance status (all had access to care) and primary language (we only gathered data from English-speaking mothers) and did not include fathers. As such, any impact of the intervention found in the upcoming clinical trial may not be generalizable to other populations that differ in these respects, and the intervention may need further refinement based on these characteristics. In addition, there are several items that previous research has shown as potentially important variables in the vaccination decision that could represent potential tailoring variables (education level, exposure to scientific controversy, degree of social networking, etc) [64,65] and were not included in our intervention. This was a purposeful decision to be able to isolate any potential impacts of values tailoring on vaccine uptake. However, these variables may need to be considered as tailoring targets in future iterations of the intervention. A notable strength of the project was the use of multiple methods to collect data (qualitative and quantitative), which may increase the validity of our findings.

We used both qualitative and quantitative approaches and significant end-user input to develop a Web-based, theory-driven, tailored messaging intervention designed to address maternal vaccine hesitancy and subsequently improve uptake of infant vaccines. Our results suggest that elements from each of the 4 models and techniques incorporated into our conceptual model for the intervention were important. Specifically, results of qualitative user design and message framing interviews supported using several techniques from persuasive messaging and MI in the intervention. A quantitative survey on parents’ vaccination values, combined with qualitative data from the message framing interviews, supported the importance of the VAB model as a foundation of the intervention and supported the idea of using values as a novel tailoring variable. Elements from the TPB were not assessed directly in this study as there was already an evidence base supporting the importance of this theory in parents’ vaccine decision making. By incorporating elements from these four theories and techniques into the intervention, we believe it will be highly effective in changing mothers’ vaccination attitudes and behaviors. This hypothesis will be tested in a subsequent randomized controlled trial.