This section profiles the dataset in terms of the research methodologies used, and the distribution of articles over time, by discipline and by country.

The 34 articles in this category argue that trust (or credibility) can be defined as the end result of a series of judgments people apply during their online search processes. Such judgments are usually exercised on both the design features of websites and the content features of the information found. Tables 2 and 3 summarize the factors influencing trust formation. They are coded to show whether they have a positive or negative effect on trust formation, and whether the authors are using the terms “credibility” or “trust” in their research. In two instances [51,52], “Evaluation” (“E”) is used throughout, although the research is contextualized with reference to trust and credibility. In tables 2 and 3, factors are listed in order of decreasing number of citations.

There might be some overlap between the features identified as design and those as content, due to the personal interpretation that some authors provide of certain aspects. For example, discussion groups could be classed as a content feature, implying access to information created by other peers rather than professionals, and indeed, Sillence et al’s [44] discussion groups are classified as content features. However, in this review, the distinction is based on whether the website offers a link to a discussion forum or group to its users (design feature) or the material discussed in the forum or group is used for study purposes (content feature). In addition, personalization (here reported as design feature) could also encompass both design and content aspects as, for example, cookies could help remember preferences (design), and include opportunities to interact with other people on the site (content). However, in the articles analyzed here, only design aspects were discussed under the umbrella “personalization.”

When discussing design features, many authors agree that a clear layout of the website is a strong positive factor influencing trust formation; the presence of contact details and the authority of the owner of the website follow as the second and third most cited factors to consider when addressing trust or credibility issues. The availability of interactive features is another desirable trait of Web design, but it should be noted that the concept of “interactive” has been evolving with time. For example, Walther et al [53] defined a website as “dynamic” if it had changing features such as graphics and fonts. However, Diviani et al [52], more than 10 years later, considered “interactive features” the presence of chat-rooms and fora linked to the website. Interestingly, Fogg et al [43] were the only authors to take into consideration elements such as functionality, customer service, and affiliations.

Much less work has been conducted on negative aspects of design features (Table 2). There is consensus among authors that the presence of explicit advertising on a health website is the least desirable visual feature. Other than this, the majority of negative website design features discussed in the literature emerge from the work of Sillence et al conducted in early 2000s [6,44,48,54]. This is an interesting finding as, although the positive aspects can, in principle, be argued in opposite terms, very little explicit distinction is made between what is good and what is bad in WHI.

The most widely discussed positive aspect of content features in the literature (Table 3) is the authority of the author, representing the level of expertise of the person or persons writing the information; this is followed by the credibility or trustworthiness of the information (in articles where the main focus was either trust or credibility). For example, Freeman and Spyridakis [37] in their study on the measure of Web-based credibility, described credibility itself as being defined by two main components: trustworthiness and expertise, but Corritore et al [11] found that credibility was one of the direct predictors of trust. The objectivity of the information is also equally important and usually defined by authors in terms of how impartial and unbiased the source is. Ease of use, the fourth most common factor identified in the literature, is here seen as an aspect of content and how the information is made intelligible, and is based on a user’s perceptions [11].

Readability, familiarity, currency, triangulation, and usefulness are other factors that have fuelled researchers’ interest and which have been the subject of a number of studies. Less common positive content aspects, mentioned only marginally in literature, include the presence of quotations, statistics, and empathy.

The discussion around the negative aspects of the content characteristics of WHI has been much more limited and fragmented. Very few authors have dedicated time to assess what hinders trust (or credibility) in WHI sources. Some researchers have discussed the concept of risk being associated with trust, particularly from a philosophical perspective, stressing that every transaction that requires trust has a degree of associated risk [11,50,63]. The detrimental effects on trust formation deriving from the information being inappropriate or irrelevant, have only been discussed by Sillence and coauthors [6,44,54], with one work focusing on the complexity of the information [52] and another on the bias of the information [43].

Nine articles (Table 4) use trust as an independent variable in their research model and view it as an antecedent to (1) the evaluation of information “quality,” or (2) to the intention to “use” the information found. In addition, Escoffery et al [72] report on the ranking that college students give to a number of criteria when looking for WHI.

Table 5 lists the 24 papers analyzing demographic aspects in connection with the concept of trust in WHI.

Widely discussed in literature are the two extremes of the age spectrum, the young and the old, and their alleged profound differences in selecting, evaluating, and trusting WHI. Old age has been associated with an overall low trust in Web resources [83,84,87,91], as people in this age group rely more on interpersonal relations with physicians, pharmacists, friends, and family [84]; in this respect, the better the quality of doctor-patient communications or other health care providers, the less people tend to go online to look for alternative health resources [86]. Medlock et al [93] found that, indeed, older people depend more on face-to-face interactions with doctors (first) and pharmacists (second), but the use of the Internet is their third chosen source on health information, and aging people who use the Internet more than their peers tend to use all other health information sources as well. McMillan and Macias [94] made the distinction between “health technologists,” who are younger seniors using online resources frequently and reporting higher trust in the information retrieved and “health traditionalists,” who are the older segment of seniors using the Internet seldom and, therefore, trusting its information much less. Distrust in Web-based resources is associated with difficulties in navigating through large amounts of often confusing information. Zulman et al [95] noted that health websites reporting clearer features that identify easily the source and authorship of the information would promote the use of the Internet among more senior people. This age group acknowledges the importance of selecting trustworthy, credible information but lacks the experience in identifying what trust indicators should be used, hence, simpler and clearer layouts would be easier to navigate. The use of the Internet is associated with an intrinsic trust in the information found and with the perception that searches are easy to carry out, but such perception and the associated feeling of trust decrease with age [63]. A comparative study conducted in Germany by Feufel and Stahl [97] on young and elderly people, emphasized how highly skilled Internet users (identified as young, with high levels of education and more Internet experience) are more confident about the quality of the information retrieved, achieve more focused results, and conduct searches to objectively inform themselves as opposed to low skilled people (identified as older, with lower education levels and patchy Internet experience), who perform less effective searches only to confirm their own preexisting opinions on a topic. Better Internet skills have also been correlated with less trust in the health information [37].

Not surprisingly, at the opposite end of the age scale, young adults, in particular teenage users, seem to experience the same lack of judgmental skills of their much older peers when evaluating health information on the Web. In studies conducted on UK and US adolescents, Gray et al [12,13] reported that young people have difficulties in evaluating online information, which are further exacerbated by low functional, critical, and interactive literacy skills. Very young adults base their assessments of WHI on aesthetics prompts, how familiar they are with a certain website [92], and on how easy it is to access the information [55]. The confidence in their own search strategies contributes to increasing the trust in the information [92]. However, it might be that such confidence is entirely subjective and unjustified, particularly if other older and more expert people are asked to review the same sources [12,13]. Johnson et al [55] found that assessment skills become deeper and more content-oriented with age, so that university students in their third year of study show more discerning judgments than their peers in the first year.

The intermediate age group, here generally defined as comprising people between 25-55 years, shows an online behavior that is more influenced by other demographic factors, particularly education and income levels. As reported in Table 5, several studies have confirmed that people with higher incomes [79-81,83,85,91] and a higher education level [79-81,83,85,86] trust and use digital health information more than people in lower socioeconomic groups. A study conducted in Australia by Dart [45] showed how people from high socioeconomic and university-based backgrounds used the Internet for health information more than disadvantaged people but, in spite of considerable differences in the Internet use, neither group particularly trusted online information. Only Ye [87] reports conflicting findings; according to this research, neither personal capital (ie, income, age, education, and health status) nor social capital (one’s network of social interactions) have an impact on trust judgments of health digital information. A study by Dutta-Bergman [99] described how, at the dawn of the Internet era, trust in WHI was segmented according to the sources accessed; for example, younger people with strong health beliefs would trust local doctors’ websites the most, whereas less educated people with weaker health beliefs would trust in hospitals more; people with higher income and education would prefer medical universities’ websites for their information needs. A more recent trend, due to the widespread use of social media, is the willingness to share health information with others online.

The perception of one’s own health status is another determining factor in health information evaluation; however, research has identified conflicting trends. Ye [89] reported that trust in WHI is associated with the ability to assess and look after one’s own health (self-efficacy) and with negative emotions due to perceived poor health. This finding is in agreement with Atkinson et al [80] and Hou and Shim [86], who have also demonstrated how people in good health tend to use offline information sources more. This contrasts with findings from Cotten and Gupta [79] and Soederberg et al [107] who assert that a good health status is indicative of more online activity. Other authors [85,87] have shown that there is no relationship between perceived health status and trust in health information.

A considerable number of the studies reviewed cover the issue of gender, but mostly as part of a larger set of demographic attributes (see Table 5); some authors agree that women use and trust the Internet more than men when it comes to health problems [45,67,79-82]. However, a number of studies showed no differences [66,83-85], leaving the gender debate open and in need of further research. Only two studies focus on gender differences in online health searches behavior: one from Korea [67]and one from the United States [82]. In contrast, the role of gender in influencing online trust judgments, in contexts other than health, has received more attention (eg, [100,101]).

Four articles focused on information sources and trust or credibility. LaValley et al [102] reported that almost 3/4 of Americans use commercially sponsored websites to satisfy their health information requirements, but stressed how different website types have different reasons for sponsoring health information, which may affect the website’s content and design. Hu and Sundar [103] showed that websites are preferred to bulletin boards, home pages, and blogs, and credibility was strongly associated with users’ perceptions of the relevance of the message associated to a certain source. Stoerger [47] found that websites with lower credibility levels were associated with a lot of interactive features and advertising. In an earlier article by Rains [70], based on the Health Information National Trends Survey (HINTS), the author demonstrated a link between a person’s trust in mass media and one’s health care provider and an increased use in WHI resources.

Two articles discussed trust from a theoretical perspective. Sillence and Briggs [104] explained how ubiquitous computing [105] has long-term and still unknown implications for the health care sector because it produces a shift of people’s trust from physicians to artificial agents (ie, computers). Singal and Shruti [106] claimed that there is no standard that exemplifies how to make trust decisions in health; however the authors envisaged to develop a technique to rerank search results using trust as a determining factor so that the more trustworthy a website, the higher its position in a result list.

This article reports on a systematic literature review of the peer reviewed literature exploring the concept of trust and credibility formation in WHI seeking. The review demonstrates that there is still no consensus of the relationship between the terms “trust” and “credibility”. This review also demonstrates that trust and credibility have been investigated both as the dependent variable, representing the end product of a series of cues and factors influencing the process of information seeking and as independent variable, alongside other variables associated with the quality or use of information. In addition, other research has examined the impact of demographic variables such as gender and age, on trust and credibility judgments in WHI seeking. Nevertheless, while there is a growing body of research in this area, given the importance of the trustworthiness of WHI, there is considerable scope for future research and theoretical development in this area. This includes:

Conceptual or theoretical: It would be of considerable benefit to be able to arrive at a consensus on the relationships between the various variables associated with research into WHI seeking. For example, there needs to be further consideration as to whether trust and credibility or information quality are the most important outcome variables. We propose that consideration of the context may be important in differentiating between trust and credibility, with, for example, trust being the appropriate term to use when information is used to inform a decision or action. This stance would also necessitate the development of an improved understanding of the relationship between trust and risk in digital health information seeking. Having established, or at least, further explored the relationships between trust and credibility, it will be important to increase understanding of the key influencing factors, and the extent to which context might impact on these.

Methods and measurement: Most research on trust and credibility in WHI seeking has adopted a quantitative approach. The purpose of quantitative studies is typically to test theory. However, in the absence of a consensus on definitions of variables and the dominant relationships between them, such studies are unlikely to lead to an integrated and coherent body of knowledge. This is further undermined by the considerable variability in the measurement scales used for trust and credibility and related variables. This needs to be addressed by a much greater number of qualitative studies that offer deeper insights into the context, processes, and judgments associated with WHI seeking and the relationship between these.

Topics: Most studies have investigated the factors that influence trust judgment in relation to WHI in general. Hence, there is scope for more studies that take into consideration judgments on specific health information sources, including specific websites and social media platforms, and the role of the owner and community associated with these platforms in influencing trust judgments. This review has not embraced research on trust regarding health information received in social support groups; this would also be an important agenda for future research. In addition, most prior research has privileged factors that have a positive effect on trust judgments, with few reporting on those factors (such as advertising) that might undermine trust. Finally, the dynamic between trust and risk deserves greater attention, particularly with regard to patients’ perceptions of the seriousness of their complaint.

Impact of demographic variables: There is evidence that various demographic variables (eg, age, income, and gender) may influence WHI-seeking behaviors, but the evidence that this also impacts on their trust judgments is scant. Further research is needed in this area. In particular, the research on disadvantaged groups has focused on identifying their needs, but little work has been done on how these needs and the ability of members of these groups to discriminate between trustworthy and untrustworthy information can be enhanced. In addition, whereas some research has been conducted on the role of gender on trust formation in information seeking, which makes links to the role of the psycho-social context, there is considerable scope for further research into the role of this context in trust formation in health information seeking. Finally, the current research base focuses on health information seekers in the United States, United Kingdom, and Australia. There is a need for research in countries where not only technological differences can play a role in information seeking, but also culture and, more specifically, trust formation and relationships with health organizations and professionals are likely to differ from non-western countries.