The extent to which cancer patients trust in cancer-related online information: a systematic review

Introduction

Information needs are among the most prevalent unmet supportive care needs of cancer patients throughout their cancer journey (Harrison et al., 2009). The majority of cancer patients want to have all available information concerning their illness and treatment (Davies et al., 2008; Jenkins, Fallowfield & Saul, 2001; Newnham et al., 2006). The internet allows cancer patients to fulfill their needs for information regarding their diagnosis, prognosis or likelihood of cure, disease stage, and treatment options or the side effects of the treatment in question (Castleton et al., 2011; Maddock et al., 2011; Tariman et al., 2014). Internet utilization is widespread in advanced economies: 89% of the US population and 80% of the European population reported using the internet at least occasionally (Poushter, 2016). Compared with other information sources, the internet has the unique advantage of convenience. Cancer patients can anonymously access the internet anytime from almost anywhere (Maddock et al., 2011; Ziebland et al., 2004).

The prevalence of cancer patients who used the internet to look for cancer-related information in a Dutch sample, two American samples and a Swedish sample ranged from 60% to 75% (Castleton et al., 2011; Mattsson et al., 2017; Mayer et al., 2007; Van de Poll-Franse & Van Eenbergen, 2008), and the prevalence appears to be increasing (Finney Rutten et al., 2016).

There are various motivational reasons for cancer patients to search for cancer information on the internet. Patients reported going online because they wanted to develop questions to discuss with their physician, verify information given by their physician, or seek alternative treatments (Castleton et al., 2011) and because they felt that the amount of information they received from their physician was insufficient (Chen & Siu, 2001).

Information obtained from the internet can impact clinical care in different ways. Cancer-related online information can change patients’ choice of treatment, their choice of physician, and their decisions regarding enrollment in a clinical trial (Castleton et al., 2011). Most cancer patients do not believe that online information searching negatively affects the doctor-patient relationship (Newnham et al., 2006). However, some cancer patients are careful about discussing online information with their physicians. These patients worry that their online searches might have a negative influence on their relationship and might cause physicians to treat them as a problematic patient (Broom, 2005; Chiu, 2011). Additionally, some oncologists admit to having some difficulty discussing internet-based information with their patients. These oncologists are more likely to report that information obtained from the internet confuses patients (Helft, Hlubocky & Daugherty, 2003).

Cancer patients who search the internet for cancer-related information tend to be younger and more highly educated than those who do not (Castleton et al., 2011; Mattsson et al., 2017; Shahrokni, Mahmoudzadeh & Lu, 2014), and they are more likely to have a partner (Mattsson et al., 2017). Both age and education are also associated with higher unmet information needs among cancer patients (Sondergaard et al., 2013). One study found that female gender was a factor associated with seeking cancer-related information (Mayer et al., 2007), while another study did not (Castleton et al., 2011). A higher likelihood of internet use was further associated with better self-reported health among cancer survivors (Chou et al., 2011).

The quality of the cancer-related information that can be accessed on the internet is highly variable. Several studies used the DISCERN scale, a validated instrument developed to help consumers evaluate the quality of health-related information on treatment choices for a specific health problem (Charnock et al., 1999), to evaluate the quality of websites that provide cancer-related information. The websites were generated by typing cancer-related search terms (i.e., ‘cancer’, ‘cancer therapy’, ‘breast cancer’ or ‘colon cancer’) in popular search engines, such as Google or Bing and focusing mainly on the first search results (Borgmann et al., 2016; Bruce et al., 2015; Hargrave, Hargrave & Bouffet, 2006; Liebl et al., 2015; Nghiem, Mahmoud & Som, 2016; Ni Riordain & McCreary, 2009; Wasserman et al., 2014), as most users will not proceed any further (Eysenbach & Köhler, 2002). The evaluated websites’ information was often incomplete and did not provide all of the details necessary to allow cancer patients to make well-informed decisions (Al-Bahrani & Plusa, 2004; Borgmann et al., 2016; Bruce et al., 2015; Hargrave, Hargrave & Bouffet, 2006; Liebl et al., 2015; Nghiem, Mahmoud & Som, 2016; Ni Riordain & McCreary, 2009; Wasserman et al., 2014). However, the results of two studies indicate a difference between the information quality of different website types. The quality of websites from nonprofit organizations or the government was higher than the quality of websites from the medical practices or commercial health information websites (Liebl et al., 2015; Nghiem, Mahmoud & Som, 2016).

Few studies have investigated which types of websites cancer patients visit to satisfy their information needs. Nonetheless, cancer patients consider health information websites to be a more valuable source of health information than forums or blogs (Mattsson et al., 2017). These patients report preferring to obtain reliable information regarding cancer from websites of their oncologist, hospital, or cancer society and are less likely to access websites with a profit interest. However, the same patients mostly accessed websites that were financed and created by pharmaceutic industries (Van de Poll-Franse & Van Eenbergen, 2008), which might promote their own interests and can be of lower quality than nonprofit websites (Liebl et al., 2015). Additionally, studies have revealed that health seekers do not consistently check the source and date of the health information they found online (Eysenbach & Köhler, 2002; Fox, 2006).

The varying quality of the cancer-related information available online presents cancer patients with significant challenges in evaluating and selecting reliable online information sources and, more specifically, in assessing the credibility and trustworthiness of these sources. Trust is an important factor associated with the intention to use information found on a website (Dutton & Shepherd, 2006; Lemire et al., 2008). People who trust in online health information become motivated to participate in various online health-related activities that meet their informational and emotional needs (Fisher et al., 2008).

The decision whether to trust a cancer-related online information can be a complex one as information-searching skills, prior experiences with the source and medium of the information, and characteristics of the source may influence the cancer patient’s decision. There are different levels (individual, interpersonal, relational and societal) of trust that have been studied in the literature. The interpersonal level appears to be the appropriate level for determining trust in online information, as information is provided by an author (trustee) and communicated over a certain channel (the internet) to a receiver (trustor) (Kelton, Fleischmann & Wallace, 2008). A definition of trust that is often used at this level is that: “The willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party” (Mayer, Davis & Schoorman, 1995).

The terms trust and credibility are often used interchangeably. However, credibility can be described as perceived information quality, or the evaluation of information quality by a user (Fogg & Tseng, 1999). After evaluating the credibility of information, a reader may decide whether to trust or not trust it. The credibility evaluation process in online environments can be explained by a dual processing model (Metzger, 2007) or the 3S-model (where the three “Ss” stand for semantics, surface, and source features of information) (Lucassen & Schraagen, 2011). The dual processing model states that the decision whether a heuristic (peripheral) or systematic (central) evaluation is performed is decided by the users’ motivation and ability. The users’ motivation results from the consequence of receiving inferior, unreliable or inaccurate information online (Metzger, 2007). We prefer the 3S-model, because we believe that due to existential concerns and their need for hope, cancer patients are a vulnerable population (Davey, Butow & Armstrong, 2003) and therefore would be highly motivated. The 3S-model asserts that the most direct strategy for evaluating credibility is to search for semantic cues (factual accuracy, neutrality or completeness of the information) in the information (Lucassen & Schraagen, 2011). However, cancer patients usually search for information that is new to them and thus do not always have the necessary expertise to evaluate the semantics of the information and consequently revert to surface cues (writing style, text length or number of references) (Lucassen & Schraagen, 2011). Systematic and heuristic processing are thus both used within a single search process. Additionally, trust in online information is influenced by trust in its source (website), which in turn is influenced by trust in the medium (internet) of this source and a general propensity to trust (Lucassen et al., 2013). Users with low trust in the source (website) cannot distinguish between high quality and inferior information (Lucassen & Schraagen, 2012). Trust in the internet is largely affected by prior experience with this medium (Dutton & Shepherd, 2006). In this study, we are interested in cancer patients’ trust in the cancer-related information online, their trust in certain cancer information websites (source) as well as their trust in the internet as a medium of cancer information.

Consumers’ trust in health information websites can be influenced by various factors. Two systematic reviews (Kim, 2016; Sbaffi & Rowley, 2017) assessed these factors, which can be organized into three categories as follows. (1) Individual consumer characteristics: consumers who are younger, are mostly highly educated, are female, have a higher level of agreeableness, have a higher income, reported being in good health condition and have a higher level of health literacy appear to be more trusting of health information websites. (2) Website-related factors: websites that are complete, understandable, unbiased, modern, useful, and easy to navigate; have a clear and professional layout; are easy to access; are run by medical universities or the federal government; and contain high-quality information are more likely to be trusted. (3) Consumer-to-website interaction-related factors: experience in using the internet and familiarity with the website are likely to influence consumer trust in the health information. Experienced users are more confident in the internet and less concerned over the risks entailed in its use, thereby increasing the likelihood of trusting health-related websites. Additionally, patients prefer health information written by people experiencing similar health issues (Kim, 2016; Sbaffi & Rowley, 2017).

In summary, it can be stated that cancer patients can only benefit from online cancer information if they can trust the information or the internet as a medium of this information. To date, no systematic review that analyzed cancer patients’ trust in online health information has been published. The primary goal of this systematic review is to identify the extent to which cancer patients trust cancer-related online information, internet websites as a source of cancer-related information or the internet as a medium of cancer-related information. As a secondary goal, the review seeks to determine whether trust in cancer-related online information differs across patients of different ages, genders, health statuses, education levels or cancer types.

Methods

A systematic literature review was conducted to explore cancer patients’ trust in cancer-related online information, internet websites as a source of cancer-related information or in the internet as a medium of cancer information. The protocol for the systematic review was registered in the international prospective register of reviews (PROSPERO) with the registration code CRD42017070190 (File S1). Additionally, the reporting of this review followed the recommendations of the PRISMA statement (Moher et al., 2009), an evidence-based minimum set of items for reporting in systematic reviews and meta-analyses (Table S1).

Results

Study selection

The search of the databases (Medline, CINHAL, Web of Science, PsychINFO and PSYINDEX) resulted in 7,314 citations (Fig. 1). All citations are available at Figshare (https://doi.org/10.6084/m9.figshare.7701014.v1). After the removal of duplicate articles, 6132 titles and abstracts were scanned for eligibility. Of these, 54 studies fulfilled the eligibility criteria of the first selection phase. Four additional studies were added after the reference lists of the 54 potentially relevant studies were manually searched. The full text of two articles could not be retrieved despite contacting the authors. Of the remaining 51 articles, seven (Crutzen et al., 2014; Losken et al., 2005; Lussiez et al., 2017; Mayer et al., 2007; Pereira et al., 2000; Roach et al., 2009; Shea-Budgell et al., 2014) met the eligibility criteria of the second phase and were therefore included in the review.

Discussion

This review includes seven studies that describe cancer patients’ trust in cancer-related online information. Overall, approximately half of the cancer patients in the included studies reported that they trusted cancer-related online information, internet websites as a source of cancer-related information or the internet as a medium of cancer-related information.

There appears to be differences in the results of the seven included studies. The lowest trust scores were reported in the three studies with the largest samples of cancer patients with various types of cancer (Mayer et al., 2007; Roach et al., 2009; Shea-Budgell et al., 2014). In these studies, the participants were asked to appraise their trust in the internet as a medium of cancer-related information. In three of the other four studies (Losken et al., 2005; Lussiez et al., 2017; Pereira et al., 2000), the participants were asked how much they trusted the cancer- or health-specific information they found online. The seventh study measured patients’ trust in one specific cancer information website (Crutzen et al., 2014). Possible explanations for the differences in trust reported by the groups of participants in these studies might be the formulation of the items that measured patients’ trust or the gender of the study participants. When patients are asked to report how much they trust the information they find online, they might be more likely to interpret the question as a rating of their information literacy (Shenton, 2009) or eHealth literacy (Norman & Skinner, 2006), which are defined as the skills needed to find, retrieve and analyze information in general (information literacy) or health information online (eHealth literacy) and use it appropriately. The potential bias that occurs when patients rate their own information literacy or eHealth literacy is that people are likely to overestimate their own abilities (Merritt, Smith & Di Renzo, 2005; Mohmood, 2016; Van der Vaart et al., 2011). Furthermore, two of the studies with higher reported trust scores (Losken et al., 2005; Pereira et al., 2000) mainly focused on breast cancer patients. These studies reported mean trust levels of 4.3 and 3.5, which is in line with prior research that stated that female consumers are more likely to trust online information (Kim, 2016; Sbaffi & Rowley, 2017). Nevertheless, a between-group comparison within the studies showed no difference between female and male patients.

The internet appears to be the second-most-trusted cancer information medium behind health care professionals (Mayer et al., 2007; Roach et al., 2009; Shea-Budgell et al., 2014). Patients are more likely to trust online information than information from newspapers, magazines, the radio, family or friends or the television. A possible explanation might be that the internet offers information that has greater relevance to the consumer and therefore is considered more trustworthy (Song & Zahedi, 2007). Magazines, radio and television present information that might target cancer patients or patients with a certain cancer type, whereas the internet allows cancer patients to search for information that is relevant to their individual situation or question.

The number of breast cancer patients in one of the included studies (Pereira et al., 2000) that reported that they were undecided regarding whether they should trust the medical information they found online was higher than expected. A possible explanation might be the varying quality of online information, which makes it difficult for cancer patients to make well-informed medical decisions (Al-Bahrani & Plusa, 2004; Borgmann et al., 2016; Bruce et al., 2015; Hargrave, Hargrave & Bouffet, 2006; Liebl et al., 2015; Nghiem, Mahmoud & Som, 2016; Ni Riordain & McCreary, 2009; Wasserman et al., 2014). Physicians and oncologists should adopt an intermediary role when they discuss internet information with their patients. They should be able recommend reliable online information sources to their patients and help them understand and discuss the information found there (Halwas, Griebel & Huebner, 2017). Patients who search for information should not be viewed as a threat (Helft, Hlubocky & Daugherty, 2003) but as an opportunity to increase communication and shared decision-making ability (Kehl et al., 2015).

It was not possible to answer the secondary research question: Does trust in online health information differ between cancer patients with different ages, genders, health status or education levels? Most of the included studies did not examine how patients’ characteristics influence the amount of reported trust in online health information as measuring trust was not their main research goal. The results of the included studies that did examine within-group differences did not always confirm the results of recent systematic reviews that identified consumer characteristics that might influence the reported trust in health information websites (Kim, 2016; Sbaffi & Rowley, 2017): In one study, as expected, female patients reported higher levels of trust (Crutzen et al., 2014). The effect size for the MD between females and males was quite small. The results of the comparisons between patients with higher and lower education levels within the studies were mixed. One study (Lussiez et al., 2017) reported the expected difference, while the other study (Crutzen et al., 2014) stated that patients with lower education levels were more likely to trust online information. There appears to be no explanation for the higher mean trust scores of the low-education patient groups compared with the high-education patient groups. However, the study sample of cancer patients with low education levels only included five individuals (Crutzen et al., 2014) and therefore may not represent the entire population of cancer patients with low education levels. A within-study comparison indicated that cancer type appeared to have no influence on the reported trust in online cancer information (Shea-Budgell et al., 2014). The absence of a difference among different cancer types did not contradict any study findings. Cancer type was added to the list of possible confounding variables because a difference in reported trust was expected due to differences in attitudes towards eHealth among patients with different types of cancer (Jansen et al., 2015). The MD in reported trust between patients who had personally sought cancer information and those who had not had a large effect size. This effect may be explained by differences in research findings. Patients who are searching for cancer information are more likely to have completed a higher level of education (Ramanadhan & Viswanath, 2006), which is again associated with a higher tendency to trust online information (Kim, 2016; Sbaffi & Rowley, 2017). Additionally, experienced users are more confident in the internet and less concerned over the risks entailed in its use, both of which have a positive influence on trust (Kim, 2016).

In terms of the methodological quality of the included studies, it should be noted that the assumptions of this review are exclusively based on the results of cross-sectional studies, which are likely to have different biases (Viswanathan & Berkman, 2012). Overall, the assessment of the included studies with the RTI item bank indicated that estimations of selective outcome reporting, confounding and overall assessment gave a positive impression of the methodological quality of the included studies. However, four studies failed to report whether the inclusion criteria varied across participants. Due to the lack of information on inclusion criteria, a selection bias cannot be ruled out which may limit the generalizability of the study results (Hernán, Hernández-Díaz & Robins, 2004). However, not reporting the inclusion criteria only allow conclusions to be drawn about the reporting quality of the included studies, but not about the quality of the studies (Margulis et al., 2014). Additionally, only three of the studies used valid and reliable instruments, which is an indicator for detection bias (Viswanathan et al., 2013). Additional research needs to be conducted using a validated instrument to measure patients’ trust in cancer-related information websites.

Six of the studies only used one item to measure the primary outcome of trust, creating a possible source of bias. Latent variables are usually complex and not easily measured. The use of multiple items helps to average out errors that are inherent in single items and therefore have higher reliability and criterion validity than a single item (Sarstedt & Wilczynski, 2009). Single items have practical advantages, such as parsimony and ease of administration (Bergkvist & Rossiter, 2009), and they usually promote higher response rates (Bergkvist & Rossiter, 2007). However, they only perform as well as multi-item scales under very specific conditions (Diamantopoulos et al., 2012). Under these conditions, the construct should be unidimensional and unambiguous to the respondent (Wanous, Reichers & Hudy, 1997). Examples of these types of constructs are job satisfaction (Wanous, Reichers & Hudy, 1997) and attitude towards advertisement and brand in marketing (Bergkvist & Rossiter, 2007; Bergkvist & Rossiter, 2009). Because assessing trust in information always will contain a certain degree of heuristics (Lucassen et al., 2013), it appears as if one item might provide a sufficient indication of whether cancer patient trust online information or the internet as an information source. Nevertheless, there appears to be a need for validated questionnaires that measure consumers’ trust in online health information.

Conclusions

This systematic review included seven cross-sectional studies out of 7,314 citations obtained from a search. The results of the included studies indicate that approximately half of cancer patients appear to trust information found on the internet, internet websites as a source of cancer-related information or trust the internet itself as a medium of cancer information. However, the small number of included studies, high heterogeneity of participants, methods and outcomes, and the diverse quality of the included studies call for further systematic research.

Further research on cancer patients needs to be conducted using a validated instrument to measure patients’ perceived trust and the credibility of health information websites, especially for groups of patients such as older adults and those with a low socioeconomic status, who appear to have lower online information searching skills and tend to be less likely to trust cancer information found online.

Furthermore, it is important to understand that cancer patients’ decision making is influenced by online information and that even if the physician remains the most trusted medium of advice, patients do and will increasingly use cancer-related websites to search information concerning their disease and its treatment. Therefore, physicians, nurses and other health care providers should provide more support and advice to patients seeking health information. Additionally, patients should be encouraged to ask their doctors questions and to discuss the results of their online information searches with them to ensure that false information is not included.

Supplemental Information