Keywords

FormalPara Take Home Points for Reader

  • Traditionally, health education programs were directed toward either general audiences or more targeted audiences.

  • Tailoring is a newer practice in health communication, where messages are designed for an individual.

  • Tailoring was made possible by computer technology, which allowed for complex algorithms that could select particular messages, from messages libraries, based upon an assessment of the individual.

  • Tailoring has now been applied across a wide range of communication channels, including Internet and mobile channels.

  • Tailoring results in messages that are more relevant to the individual, and as a result are more likely to be attended to and processed by the individual.

  • Tailored health programs have been found to be efficacious across a wide range of health behaviors.

  • Further work is needed to better understand the mechanisms underlying effective tailoring and to understand how to best disseminate tailored programs so that they can reach more people.

Traditionally, health education and health communication efforts have been directed toward either a general audience or a particular target audience. While this approach is efficient with regard to reaching audiences, it may often not be efficient with regard to public health impact. That is, if impact is defined as population reach x efficacy (Abrams et al. 1996), interventions must not only reach large audiences but also must be efficacious in changing those audiences’ beliefs and behaviors. Several decades of research have taught us that general audience approaches to health communication are unlikely to be efficacious, while the efficacy of targeted health communication approaches varies (Noar et al. 2009; Snyder et al. 2004; Strecher 2009).

A newer practice in health communication , and one that has shown much promise over the past two decades, is tailored communication. Tailored communication has been defined as “any combination of strategies and information intended to reach one specific person, based on characteristics that are unique to that person, related to the outcome of interest, and derived from an individual assessment” (Kreuter et al. 1999, p. 277). Tailored communications are uniquely individualized to each person, whereas targeted messages are developed to be effective with an entire segment of the population. Targeting and tailoring can be thought of on a continuum of message design that ranges from the most generic mass audience communications, which are meant for everyone, to the more customized targeted communications, which are meant for a specific group of people, and finally to the most customized tailored communications, which are meant for individuals. To achieve their customization, tailored communications require an assessment of the individual (see Fig. 18.1).

Fig. 18.1
figure 1

Level of customization of health content for general audience, group targeted, and individually tailored interventions. Note This figure was adapted from Kreuter et al. (1999) and Hawkins et al. (2008)

Full size image

Message tailoring operates on the premise that although targeting (at the group level) may enhance the perceived relevance of a health message for members of a group, there will still be a “mismatch” between a message designed for an entire group and some members of that group. Interestingly, while most advertising uses group-targeting practices, advertisers are increasingly using new technologies to provide tailored communications. For example, websites such as Amazon.com assess a person’s browsing and buying tendencies and make tailored recommendations for what that person may be interested in buying. Netflix and Tivo use similar tailoring practices, making personalized suggestions for what movies one should rent or television programs to record and watch. Even supermarket scanning technology assesses the kinds of items shoppers typically purchase and provides customized coupons at the checkout that reflect buying tendencies. While tailored health communication and this type of advertising tailoring uses transparent means, it is worth noting that advertisers are increasingly applying this type of tailoring in a more covert manner. That is, there have been several recent reports indicating that customized advertising is greatly increasing on the Internet, using location data and data that are accessed from one’s personal computer (e.g., recent web pages viewed, search queries conducted, items bought, etc.) (Turow 2012). This kind of information is sometimes accessed without a person’s knowledge, raising ethical and privacy-related concerns.

In the health promotion area, a perfect storm that ultimately led to the development of tailored communication interventions occurred: first, recognition of the poor outcomes of many health promotion programs using print materials (e.g., self-help manuals); second, development of stage-based theories such as the transtheoretical model (Prochaska et al. 1992), which suggested that because people are often at differing stages in the change process, they require different, more individualized messages; and third, technological developments in computer technology that made tailoring on a large scale basis possible. The first studies in the message tailoring area compared print materials tailored on determinants from health behavior theories (e.g., transtheoretical model, health belief model, social cognitive theory, theory of reasoned action) (Ajzen and Fishbein 1980; Bandura 1986; Prochaska et al. 1992; Rosenthal 1974) to materials that were more generic or targeted in nature and found that tailored materials led to more behavior change in areas such as smoking cessation and dietary practices (Skinner et al. 1999). The success of these early studies ultimately led to a burgeoning literature on tailored health communication . Tailoring has since been applied across many other communication channels, to more than 20 health behaviors, and to a number of populations (Noar et al. 2009). More recently tailoring applications in the area of Internet-based interventions (Lustria et al. 2009) and text messaging interventions (Fjeldsoe et al. 2012) have been rapidly growing.

How Tailoring Operates

How is a tailored intervention typically implemented? The process is depicted in Fig. 18.2 (also see Dijkstra and De Vries 1999; Kreuter et al. 2000). First, an individual is assessed using standardized measures on a variety of characteristics that are relevant to the behavior under study (e.g., demographic, behavioral, cultural, psychosocial characteristics). As the tailoring literature has been greatly driven by the transtheoretical (or stages of change) model and other psychosocial theories, such assessments often measure variables from one or more behavioral theories. Assessments can be made in a variety of ways—for example, via telephone, mail, local computer/kiosk, Internet, or in person. These data are then processed and, when the intervention relies on computer technology (which represents the vast majority of studies), decision rules (computer algorithms) select the particular messages that are most appropriate for the individual. These messages are derived from a message library, which consists of hundreds or even thousands of messages created for the program. In some cases, additional libraries may exist, such as an image library that contains a variety of images to be matched to participant characteristics (e.g., gender, race, age). When delivery of tailored messages is done by a counselor or health educator, the same process of matching assessment to feedback is conducted, but typically without the use of computerized algorithms.

Fig. 18.2
figure 2

Overview of the tailoring process

Full size image

Delivery of tailored content varies depending on what channel is used. In the case of tailored print materials, a feedback report is compiled (again by the computer program), printed out, and presented to the participant in person or through the mail. Tailored computer programs that run on local computers, kiosks, or over the Internet operate in a similar fashion, except that the feedback messages are presented onscreen (immediately after assessment); in some cases, the feedback also may be printed. Computer-based programs have the opportunity to use interactivity and multimedia and thus may have additional content libraries (e.g., video library). Finally, counselors or health educators can deliver tailored messages over the telephone or in person. While most counseling interventions would not be considered tailored because of lack of a standardized assessment and subsequent matching of that assessment to tailored feedback, a small number of studies that used standardized measures for assessment and counselors/health educators to deliver feedback based on assessment can be considered tailored interventions (Brinberg and Axelson 1990; Brinberg et al. 2000). It is worth noting that the cost-effectiveness of tailoring may be reduced when human counselors are used for delivery of intervention feedback rather than print materials or the Internet. Finally, automated voice programs can also tailor content that is delivered in audio form over the telephone.

Tailoring Channels

The current discussion highlights the many assessment and delivery channels that tailored interventions have used to date. Table 18.1 provides a list of channels and channel attributes; an X indicates positive attributes of the channels, whereas a missing X indicates a negative attribute. While this list is not exhaustive, it provides a comparison of basic advantages and disadvantages of the major tailoring delivery channels at present. The list also provides generalizations; therefore, there may be exceptions. For example, print and telephone are generally viewed as lower cost options than computer and Internet; however, there are circumstances in which this may not be the case, such as when the number of participants in a program becomes very large and an Internet-based program may become more cost effective than a print-based program. Similarly, as time passes and programming and technical support become less costly, the Internet may overtake print as a more cost effective channel , regardless of the number of participants in a program.

Table 18.1 Comparison of six delivery channels for tailored health communication programs
Full size table

A logical question that has been raised in this area is which delivery channel is “best.” For the most part, the limited evidence to date does not suggest the clear superiority of one channel over another. The few trials that have conducted head-to-head tests comparing print, Internet, and telephone programs have found all of them to demonstrate short-term efficacy (Kroeze et al. 2008a; Marcus et al. 2007), and two meta-analyses support this conclusion (Krebs et al. 2010; Sohl and Moyer 2007). There is some evidence that print interventions may lead to longer maintenance of behavioral changes than telephone (Marcus et al. 2007) or web delivery (Kroeze et al. 2008a). However, a study comparing an online tailored web site with motivational interviewing (interpersonal tailoring) found both interventions to be equally effective (Prochaska et al. 2008).

The broader answer to this question may be that differing channels will suit differing circumstances (which seems quite apropos given the concept of tailoring). For instance, for program applications that demand interactivity and/or multimedia, local computer, mobile device, or Internet interventions may be best. However, these applications are not without their disadvantages, including requirements for ongoing technical assistance for the program, significant development costs, and a level of skill on the part of the user. Alternatively, channels such as mail (print materials) and telephone are lower cost options that are capable of reaching large numbers of individuals who are likely to have access to these media. On the downside, however, is the fact that these kinds of programs are not capable of approaching the kind of sophistication of media that is easily achieved with computer delivery.

Applications of Theory

To date, numerous behavioral theories have been used to inform tailored interventions. Constructs from such theories are often used in the tailoring process itself, a strategy referred to as behavioral construct tailoring (Kreuter et al. 2000). Theory can also be used in other ways, however, such as providing detailed message design guidance in intervention development (Harrington and Noar 2012).

Reviews of the tailoring literature indicate that the transtheoretical model and stages of change may be the most dominant theoretical perspective in this literature (Noar et al. 2007, 2011; Richards et al. 2007). This is not surprising, given the influential role that this model played early in the tailoring literature (Noar et al. 2007). In fact, many of the first tailored interventions applied stages of change (Campbell et al. 1994; Skinner et al. 1994) or the full transtheoretical model (Prochaska et al. 1993). Many other widely used theories have also been applied in this literature, however, including social cognitive theory , the health belief model, and the theory of planned behavior (Lustria et al. 2009; Noar et al. 2007, 2011; Richards et al. 2007; Sohl and Moyer 2007).

It is important to note that tailored interventions often apply multiple theories to inform a single tailored intervention (Noar et al. 2009; Richards et al. 2007). Also, it cannot be assumed that because a particular intervention is “based on” a theory that the intervention tailors messages on the basis of all of that theory’s core constructs (Noar et al. 2007) (also see Painter et al. 2008). Instead, theory has been used in a very eclectic and utilitarian manner in this literature, with select constructs from a variety of theories often informing one intervention. While theoretical purists may object to this manner of using theory , others defend the use of theoretical constructs for interventions on the basis of their empirical usefulness (Bartholomew et al. 2006; Kreuter et al. 2000) or suggest that the use of multiple theories that make complementary contributions can strengthen interventions (Glanz and Bishop 2010).

Evidence of Efficacy

Two decades of literature provides a strong empirical basis upon which to judge the efficacy of computer tailored interventions (CTIs). Early in the literature, a number of seminal studies demonstrated efficacy and thus provided reasons for optimism (Campbell et al. 1994; Prochaska et al. 1993; Skinner et al. 1994). Subsequently, narrative reviews of the literature concluded that CTIs generally were successful in affecting health behavior change in diverse areas (Skinner et al. 1999), including smoking cessation (Strecher 1999), nutrition (Brug et al. 1999), and cancer prevention (Rimer and Glassman 1999). There have also been more recent systematic reviews that have concluded that CTIs are generally efficacious (Kroeze et al. 2006; Neville et al. 2009; Richards et al. 2007). Perhaps most importantly, four meta-analytic projects have examined the literature and provided a more fine-grained analysis of CTI efficacy. We describe these meta-analyses in greater detail next (also see Table 18.2).

Table 18.2 Summary of meta-analyses of tailored interventions
Full size table

Noar et al. (2007) conducted a meta-analysis of 57 studies that tested the ability of tailored print materials to affect health behavior change . The studies primarily consisted of smoking cessation (26 %), diet (23 %), and mammography screening (21 %) interventions. The overall effect size in this study was r = .074 for tailored interventions compared with no-treatment control and alternative interventions, which converts to d = .15. Perhaps more importantly, a subsequent analysis that excluded studies containing only no-treatment control comparison conditions revealed that tailored interventions still outperformed generic or targeted interventions (r = .058 or d = .12).

Across behaviors, smoking and diet had the largest effect sizes, followed by mammography screening and then exercise. Also, a variety of intervention characteristics moderated intervention efficacy. For example, studies that generated tailored reports in the form of pamphlets/leaflets and newsletters/magazines had significantly larger effect sizes than those generating letters or manuals. In addition, those interventions with more than one contact with participants had significantly larger effects than those with just one contact. Finally, studies tailoring on particular theoretical (i.e., attitudes, self-efficacy, stage of change) and other (i.e., demographic) factors had larger effect sizes than those not tailoring on these variables (Noar et al. 2007).

That same year, Sohl and Moyer (2007) published a meta-analysis of 28 tailored intervention studies focused on increasing mammography screening within print, telephone, and in-person interventions. Results indicated a statistically significant overall effect of the tailored interventions compared with no-treatment control and alternative interventions, with an odds ratio (OR) equal to 1.42 (which converts to d = .21). Telephone, in person, and print tailored interventions all had similar impact. In addition, studies that tailored on variables from the health belief model had significantly greater impact than those that did not, and interventions were significantly more effective when a physician recommendation was part of the intervention. Also, studies that measured recent mammography as opposed to regular mammography tended to have larger effect sizes.

More recently, Krebs et al. (2010) conducted a meta-analysis of 88 studies testing tailored interventions delivered using print (75 %), computer (22 %), and automated telephone (3 %) channels . Behaviors examined in this review were diet, smoking, physical activity, and mammography screening, and some studies intervened on more than one behavior at a time. Overall, there was a statistically significant effect of tailored interventions on health behavior change (d = .17), and this effect did not vary significantly across tailoring channels . Effects were similar across behaviors, with dietary fat reduction being most efficacious (d = .22). In addition, interventions that focused on multiple behaviors did not have smaller effects than those focusing on a single behavior.

Krebs et al.’s (2010) meta-analysis also examined tailoring effects over time. Results revealed that effects tended to peak between 4–12 months and then gradually decline over time. Interestingly, there was also evidence that dynamically tailored interventions (those reassessing individuals before providing new tailored feedback) had significantly larger effects at most timepoints (including 13–24 month follow-up) than statically tailored interventions (those providing new tailored feedback based on the same baseline assessment). Indeed, only dynamically tailored interventions demonstrated statically significant effects at long-term follow-up.

Most recently, Lustria and colleagues conducted a meta-analysis of web-delivered, tailored health behavior change interventions (Lustria et al. 2013). The 40 studies primarily consisted of physical activity (42 %), diet (25 %), and smoking (18 %) interventions. The overall weighted mean effect size in this study was d = .14. Studies that contained a longer term follow-up timepoint (53 % of studies) also exhibited a significant effect at the follow-up timepoint, d = .16. This bodes well for maintenance of intervention effects using tailored interventions, as it suggests no decay of intervention effects over the course of the study. Another key finding is the fact that multiple behavior interventions had similar effects as single behavior interventions, suggesting that intervening on multiple behaviors may not undermine behavior change .

Overall, these meta-analyses suggest that tailored interventions have often been successful in stimulating behavior change , and they each contribute to our knowledge about what may make efficacious interventions. The evidence to date suggests that messages that are more customized to an individual are more successful in influencing health behavior change (Noar et al. 2007) and that carefully constructed interventions can maintain changes over the longer term (Krebs et al. 2010; Lustria et al. 2013). Effect sizes across all of these meta-analyses were similar (ranging from d = .14–.21), giving us some indication of what the “typical” effect of a tailored intervention may be. Other important findings from these meta-analyses indicate that tailoring channel does not in and of itself appear to make a difference, but how tailoring is carried out (e.g., choice of theoretical constructs, dynamic vs. static tailoring, design of print materials) does appear to have a measurable impact on the efficacy of a tailored intervention. Finally, tailoring may be a promising strategy for those who wish to intervene on multiple behaviors (Krebs et al. 2010; Lustria et al. 2013), although a meta-analysis focused only on multiple behavior interventions would better enable us to understand the effects of such interventions.

Future Directions

Cumulative Science of Tailoring

New research on tailored health communication is critical to advance the field. The field is currently shifting from research that addresses whether tailoring works to research that addresses under what conditions it works. Many of the meta-analyses discussed in this chapter have conducted analyses to explore this question. In addition, recent conceptual work has distinguished among the various types of personalization and feedback strategies that tailored interventions can deliver (Dijkstra 2008; Hawkins et al. 2008). This “new language” will help tailoring researchers better disentangle the various components of tailored messages in order to advance an understanding of how tailoring operates most effectively (see Table 18.3).

Table 18.3 Contemporary terminology and definitions in tailoring research
Full size table

In addition, an important observation about tailored interventions is the following: To date, tailoring has almost entirely been conceived of as a way to customize intervention content to individuals. As a result, nearly all tailoring has focused on what scholars believe to be the behavioral determinants of tailoring , which come from the theories of behavior and behavior change described earlier in this chapter (also see Noar and Zimmerman 2005). Many other factors affect how health content will be received, however, and a number of these communication-oriented factors could also be tailored on. Table 18.4 lists four domains that can be considered in tailoring (Noar et al. 2009; Rimer and Kreuter 2006). To date, the literature has focused almost entirely on the first row (content) and often neglected the other three domains in tailoring. The potential here is enormous: Not only could CTIs ultimately deliver the right content to the individual but also they could deliver it in a way that best ensures that content is attended to, cognitively processed, and perceived as personally relevant.

Table 18.4 Domains in which tailoring can be achieved and associated theories and variables
Full size table

That specific studies are beginning to examine the kinds of questions raised by Hawkins et al. (2008), Noar et al. (2009), Rimer and Kreuter (2006), and others (Dijkstra 2008; Strecher 2009) is promising. For example, newer studies are examining novel constructs on which to tailor—especially constructs in the areas of cultural tailoring (Kreuter et al. 2005; Resnicow et al. 2008; van der Veen et al. 2010), environmental tailoring (van Stralen et al. 2009, 2010), and message framing (Latimer et al. 2008; Ruiter et al. 2010). Research is also examining the effects of different types of tailoring personalization and feedback on intervention efficacy (de Vet et al. 2008; Dijkstra 2005; Kroeze et al. 2008b; Strecher et al. 2008), mediators and moderators of effective tailoring (Campbell et al. 2008; Ko et al. 2010, 2011; Strecher et al. 2006), different ways of delivering multiple behavior change interventions (Vandelanotte et al. 2008), and even brain responses to tailoring (Chua et al. 2009). Additional studies are needed to help build a cumulative science of best practices in tailoring . As this work is conducted, we strongly encourage researchers to follow recommended guidelines for reporting studies of tailored interventions (Harrington and Noar 2012).

Dissemination of Interventions

A second key area for future research on tailored interventions is the area of dissemination. While many tailored programs have shown efficacy in research trials, few have been disseminated into practice. Why is this the case? Key barriers include the following: (1) many platforms for dissemination (worksites, clinics, etc.) want an intervention that covers many behaviors, while most tailored interventions focus on a single behavior; (2) many of the computer platforms upon which tailored interventions were developed are now out of date; (3) ongoing technical assistance is needed for such interventions, but resources are not always available for such support; (4) intellectual property issues arise in the case of some tailored interventions; and (5) it is not clear whether a tailored intervention developed in one context or region of the country would work as effectively in another context or region of the country. Some of these barriers have resulted from the fact that many tailored interventions were not developed with dissemination in mind, and in that manner future studies will benefit from a more careful consideration of dissemination issues early in the development process (National Cancer Institute 2008; Vinson et al. 2011). Also, dissemination and implementation research specifically on the topic of tailored interventions will also be necessary (Rabin and Glasgow 2012).

Many of these barriers are surmountable, of course, and progress is being made toward potential solutions. Approaches include (1) increasing development of multiple behavior interventions that address many health behaviors in the context of a single program; (2) developing open source software that can be shared among developers of tailored interventions to provide a common platform for such interventions; (3) conducting studies to research the best manner in which to disseminate tailored interventions; (4) partnering with nonprofit and private sector agencies that have an interest in disseminating such interventions; and (5) increasing the dialogue among technical developers about the best ways to create sustainable interventions and provide technical support over the long term. The ultimate impact of tailored interventions on lifestyle behaviors could be substantial, but this impact is only possible if such programs are exported from the confines of funded research to the field setting. Although we are not there yet, efforts are underway to move us closer to the goal of bringing more of these efficacious programs into the public domain.

While reaching many of the above goals will take time, there has been some substantive movement toward dissemination. Dr. Victor Strecher and his colleagues in the Center for Health Communications Research at the University of Michigan have developed open source software for the creation of CTIs (Center for Health Communications Research 2011). The software, called the Michigan Tailoring System , is free to use and is open source, making modifications to it possible by any technical developer. Also, Dr. Marci Campbell and her colleagues in the Communications for Health Applications and Interventions (CHAI) core at the University of North Carolina at Chapel Hill have developed Tailortool, an open source software toolkit that enables the creation and delivery of CTIs over the Web (Communication for Health Applications and Interventions 2011). Tailortool allows a user to develop CTIs that are delivered in a newsletter format on the computer (in PDF format), which can then be printed or saved electronically for later reading. Also in existence are open source interactive voice response programs that are available for use in developing and delivering CTIs (Vinson et al. 2011). Moreover, while public sector dissemination has so far been a challenge to achieve, companies such as Health Media (founded in 1998; http://www.healthmedia.com/) and Pro-Change Behavior Systems (founded in 1997; http://www.prochange.com/) have been successfully developing and disseminating CTIs in the private sector.

Conclusion

In this chapter, we have provided an overview of tailored health communication research, providing a brief history of its genesis, highlighting its theoretical foundations, and summarizing what we know about tailoring effects. We have described the basic process of how tailoring operates and the channels through which tailored interventions may be delivered. We also have identified important directions for future research, including the importance of building a cumulative science of tailoring and translating our knowledge to applied settings.

Tailored health communication interventions represent a cutting edge approach to integrating persuasive message design with theoretical and technological developments. Ongoing research in this area offers an exceptional opportunity to extend health decision science by increasing our understanding of how to design and deliver tailored messages to individuals in order to help them make good health decisions. We look forward to continued advancements in this vibrant area of research.