How narratives influence colorectal cancer screening decision making and uptake: A realist review

Abstract Background Although narratives have been found to affect decisions about preventive behaviours, including participation in cancer screening, the underlying mechanisms of narratives remain unclear. Objective The purpose of this study was to summarize and synthesize existing literature on narrative interventions in the context of colorectal cancer screening. Our main research question was as follows: How, when and for whom do narratives work context of decision making about colorectal cancer screening participation? Methods We undertook a realist review to collect evidence on narratives in the context of colorectal cancer screening. A search of the literature was performed in Embase, MEDLINE/PubMed, Cinahl and PsycINFO. We included empirical evaluations (qualitative or quantitative) of narrative interventions. In total, 15 studies met the inclusion criteria. A content‐based taxonomy of patient narrative types in decision aids formed the basis for our initial programme theory. Main result We identified four mechanisms: (a) process narratives that address perceived barriers towards screening lead to improved affective forecasting, (b) experience narratives that demonstrate the screening procedure lead to increased self‐efficacy, (c) experience narratives that depict experiences from similar others lead to more engagement and (d) outcome narratives that focus on outcomes of colorectal cancer (CRC) screening decision decrease or increase fear of colorectal cancer. The evidence was limited on which narrative type may facilitate or bias informed decision making in colorectal cancer screening. Discussion and conclusion The findings indicate the importance of more detailed descriptions of narrative interventions in order to understand how mechanisms may facilitate or bias informed decision making in colorectal cancer screening.

explicit purpose of cancer screening programmes in many European countries. 1 IDM assumes that individuals make a rational and autonomous choice that is based on relevant knowledge and is consistent with their attitude towards undergoing screening. 2 Previous studies showed that certain groups, including those with lower socio-economic status (SES), ethnic minority groups and those with lower health literacy levels, are less likely to participate in CRC screening. [3][4][5] Numerous reasons for lower screening participation in those groups have been suggested, such as lower engagement with cancer screening information, lack of time, financial resources and lower perceived self-efficacy. 6 Lately, several studies found that screening invitations, which are typically written materials, are often too difficult to understand and do not help to make informed decisions about health. 7,8 IDM requires more than just the ability to read and understand cancer screening information. It also requires the ability to appraise the potential benefits and harms of screening and apply the information to one's personal situation. 9 This combination of skills is referred to as health literacy, which is broadly defined as an individual's capacity to assess, understand and use information to make informed decisions in health care. 10,11 Poor health literacy is more common among patients who have lower educational level, older patients and racial and ethnic minorities. 12 In order to reduce inequalities in cancer screening, it is crucial to investigate new communication strategies that are culturally sensitive and are presented in an accessible and comprehensible format. 13 The inclusion of narratives in cancer screening information is increasingly being suggested as a valuable tool for greater engagement with screening information for ethnic minority groups and people with low SES. 14 Narratives are personal stories that convey information through others' health situations or experiences. 14 Previous research showed that cancer narratives may be especially useful to overcome resistance to cancer screening information and to facilitate the mental simulation of unknown or frightening procedures, including screening. 15 However, research on narratives interventions and decision making in the context of cancer screening is still in its early stages.
It has remained unclear how narratives affect the decision-making process and cancer screening uptake, and subsequently how they should be used in informed decision making. 16 Despite their potential to reduce ethnic and socio-economic inequalities in participation in cancer screening, 14 narratives can be considered as complex interventions (ie, interventions whose effects are crucially dependent on context and implementation) that may either facilitate or hinder informed decision making in health care.
Narrative interventions work only if they are targeted for participants in the appropriate circumstances and are implemented in the right way. They are also strongly related to the decisions and actions taken by the participants. From a realist perspective, a certain type of narrative may work well in a certain setting but poorly or not at all in another setting. 17 This knowledge is crucial in order to understand how, when and for whom narratives may work in informed decision making about CRC screening. 17 A realist approach is a theory-driven way of analysing complex interventions and is based on the idea that an intervention works (or does not work) because participants make certain decisions or act in a certain way in response to the intervention. 17 Our main research question was as follows: How, when and for whom do narratives work in the context of decision making about colorectal cancer screening participation? 2 | ME THODS

| Design
Following the realist review principles, our methods included: (a) formulating an initial programme theory about how narratives are meant to work and what impacts they are expected to have, (b) selecting and appraising studies and (c) testing the programme theory by extracting, analysing and synthesizing relevant data. 18 We used the RAMESES publication standards for realist reviews. 19

| Phase I: Initial programme theory
In the first phase, we formulated an initial programme about how a narrative intervention is expected to lead to its effects and in which context it should do so. 17 One way to develop an initial programme theory is to use concepts from another theory that informs current or comparable interventions. 20 In addition, countervailing mechanisms can be distinguished (ie, mechanisms that explain why an intervention does not work). We chose the taxonomy by Shaffer and Zikmund-Fisher 21 of narratives in decision aids as our initial programme theory. Shaffer and Zikmund-Fisher 21 argued that narratives are too often perceived as a homogenous entity and one single construct in existing research. The use of narratives to support decision making should therefore not be classified as "good" or "bad" but rather "whether certain narrative types are suited for certain purposes." They therefore designed a general taxonomy of narrative content types (ie, "process," "experience" and "outcome" narratives) in decision aids to better understand when and how narratives affect decisions about health care. These narrative types are hypothesized to depend on three narrative dimensions: purpose, content and overall tone. 21

| Process, experience and outcome narratives
In the context of CRC screening, "process narratives" describe how an individual made the CRC screening decision. "Experience narratives" describe experiential aspects of CRC or CRC screening and essentially provide information about what it is like to have CRC or undergo CRC screening. "Outcome narratives" describe the psychological (eg, patient's quality of life or regret) or physical (eg, CRC patient's survival) health outcomes associated with the CRC screening decision. 21 The taxonomy hypothesizes that these three narratives types have five different, yet overlapping purposes: (a) to inform, (b) to engage, (c) to model behaviour, (d) to persuade and (e) to comfort.
The essential difference between these three narrative types is to the purpose to inform (ie, "process" and "experience" narratives) and the purpose to persuade (ie, "outcome" narratives). Accordingly, Shaffer and Zikmund-Fisher 21 suggest not to use "outcome" narratives as patient decision aids. In contrast, they hypothesize that "experience" narratives as well as "process" narratives might be a helpful component in patient decision aids.
In line with the initial programme theory, we therefore hypothesize that "process" and "experience" narratives may facilitate IDM in CRC screening, whereas "outcome" narratives may be more persuasive and should not be used as decision aids for CRC screening.  Table 1 for the characteristics of the included studies and the narrative interventions).

| Phase III: Testing the programme theory
To test the programme theory, the included studies were qualitatively analysed, using framework analysis. 22 Framework analysis is well-suited for a realist approach, because it can be applied to studies using a mixed-method design. 22 Following the framework analysis' five steps, we first familiarized with the data, using a deductive approach to categorize the included narratives into a narrative content type (ie, "process," "experience" and "outcome") according to the initial programme theory. Initially, the authors (AW and JS) reviewed ten articles. In the second step, we used an inductive approach, identifying a thematic framework of mechanisms. The third step (indexing) involved coding the contexts, mechanisms and outcomes in all included studies, for which we used the qualitative data software MAXQDA. 23 These contexts, mechanisms and outcomes were then compared with the initial programme theory, which we modified using the evaluation findings.

| Assessing the quality
Two authors (AW and JS) independently appraised the evidence and generally agreed on the quality of the included articles. Following the realist principles, the assessment of the quality of included studies was guided by Pawsons' 18 stages of relevance and rigour (see Appendices S1 and S2). Relevance entails whether the included study can contribute to theory building and rigour entails whether the method used to generate the data is credible and trustworthy.
For assessing relevance, we focused especially on the theory building in the studies, which can be examined by the use of "thick description" in an article. Lincoln and Guba 24 describe "thick description" as a way to achieve external validity. Thick description is contrasted with thin description which can be seen as a superficial account (see Box 2). For example, when studies described theoretical concepts in sufficient depth to be relevant to our research questions, we evaluated these as being relevant to our study (see Appendix S1).
For assessing rigour, we examined the study design, the data collection and the analysis (see Appendix S2). Disagreements, for example about whether a narrative intervention should be categorized as "outcome" or "experience," or both, were discussed by both authors (AW and JS) until agreement was reached.

| Overview of included studies
Of the 15 included studies, one study used a qualitative design, 25 two studies used a mixed methods design, 26,27 and twelve studies used a quantitative design. Of these twelve studies, nine used a randomized (controlled) trial design. In these randomized (controlled) trial studies, the main intervention was narrative information with the control groups receiving: a culturally targeted presentation, 28 general information about CRC, 30,32 a stock (ie, no tailoring, no narrative), tailored narrative or tailored educational message, 33 a numeric risk tool, 34 risk information, 35 an educational video about car safety 36 and a promotora and video intervention. 37 Narrative interventions were targeted at various ethnic groups. Three studies were set in the UK, and all other twelve studies were set in the United States (see Table 1).

| Narrative content type
By analysing the content of the narratives, we identified only two "process narratives" that informed participants about the decisionmaking process of CRC screening. 32, 38 We identified six "experience" narratives that focused primarily on experiences with CRC or with CRC screening. We identified six "outcome" narratives, described by seven studies 30,33,35,36,39 that focused primarily on the outcomes of the CRC screening decision (see Table 1).

| Mechanisms of narrative content types
From the framework analysis, we identified four mechanisms and eight associated countervailing mechanisms (see Figure 2 for the final programme theory).

| Mechanism 1 Process narratives that address perceived barriers towards screening lead to improved affective forecasting
The process narratives 32,38 that we identified described an experience from a similar other who had recently made the screening decision. In these narratives, the character first felt uncertain about screening and knew little about screening tests. The character then talked about the decision process, including encountered barriers towards screening (eg, inconvenience of the screening procedure including having to take the laxative). Dillard and Fagerlin 32 described how the narrative intervention was tailored to participants' overestimated barriers towards CRC screening, so-called "affective forecasts." For instance, the narrative character stated: "When trying

Inclusion criteria
Empirical studies examining the effects of narratives in the context of CRC screening.
Studies reporting on cancer diagnosis or treatment (and not screening).
Studies describing narratives that solely promote cancer preventive behaviours, such as smoking, nutrition and physical activity, rather than screening.
Studies reporting on the feasibility of narrative interventions in cancer screening. That is, studies that simply examined whether it was possible to use narrative interventions, rather than examine the effects of narratives themselves. Description of factors or mechanisms mentioning "model," "process," "functions," "investigates," "describes," "explains," "experiences" etc

| Mechanism 2 Experience narratives that demonstrate the screening procedure lead to increased self-efficacy
Several "experience" narratives demonstrated the screening pro-

| Mechanism 3 Experience narratives that depict experiences from similar others lead to more engagement
Several "experience" narratives presented CRC or CRC screening experiences from others who were depicted as "just ordinary people" from the community, 25 as "real" people, 31 similar others, 32 but also as respected role models in the community. 28 For instance, one participant in the study by Bennett and von Wagner 25 stated: "saying what they do and that they're married and they have children make you identify a little bit more with them." These "experience" narratives showed that if narratives present a range of experiences from similar others, this makes narrative information more reassuring, more credible, more vivid, more engaging, 32 helps to reduce fears people have about the test 38 and legitimizes the quotes and stories. 30 This leads to normalization of cancer screening, 25  terms, availability in different languages, recruiting actors from the community and the setting of a typical home or clinic, can increase perceived similarity with the narrative character. 26,27,34 Countervailing mechanism 3.4: When solely presenting narrative information, this mechanism might not be effective for those who also regard factual information as essential to making a decision about CRC screening. 25

| Mechanism 4 Outcome narratives that focus on outcomes of the CRC screening decision increase or decrease anxiety
There was a wide variety in the framing of the psychological and physical outcomes of CRC screening within the identified "outcome"

| Mode of information
Narrative information might be effective in influencing decision making, regardless of the mode of information (paper-based, video, promotora-only). 37 In addition, Pignone and Harris 36 found that narrative information, in combination with an educational brochure that is tailored to an individual's decision-making stage, increased screening intention and uptake. Studies focusing on diverse groups showed that as long as the presenter of the narrative information is culturally sensitive, the narrative can be successfully presented or delivered by any individual, regardless of ethnic background. 28,37 However, the narrative information must be easily accessible to the participants. Hwang

| D ISCUSS I ON
The aim of this review was to understand how, when and for whom narratives work in the context of CRC screening. Using the taxonomy by Shaffer and Zikmund-Fisher 21 as our initial programme theory, we categorized the included narrative interventions into three narrative types: "process," "experience" and "outcome." We identified only two "process" narrative interventions that targeted the decision-making process, six "experience" narratives and six "outcome" narratives. We further specified and refined four mechanisms, which provide an exploratory account of how these CRC screening programmes must ensure that screening invitees receive accurate information based on the most recent available evidence and information about the potential benefits and harms of CRC screening. 41 Accordingly, decision aids should inform about these benefits and harms of screening, and should present these in terms of absolute risk, not relative risk. Moreover, the information must not be directive and must facilitate an autonomous choice. 42

| Limitations
This synthesis of literature on narrative interventions in the context of cancer screening has some important limitations. First, the narrative interventions varied in terms of narrative length, intervention (eg, video, leaflet, and testimonial), purpose, content, overall tone and participants, making it difficult to compare between the studies and categorize the narrative interventions. Second, few narratives have been experimentally evaluated and the majority of quantitative studies used a cross-sectional design, precluding causal inference.
Third, none of the studies assessed decisional certainty, deliberation or informed decision making. Therefore, we are not certain whether the narrative interventions facilitated decision making that was congruent with participants' values and preferences. Fourth, it might be possible that we missed studies on CRC screening narratives as we focused on studies evaluating narrative effects. Studies that solely described the development of narrative, for example, were excluded.
Fifth, most narrative interventions in the studies were carried out in the United States, which means that the development process of narrative interventions reflects the access to health insurance and care provided in public and private systems. In addition, the focus in the United States is more on screening uptake rather than on informed decision making, through actively promoting the message "the best test is the one that gets done." 43 Hence, our synthesis provided more empirical evidence on the positive impact of narrative interventions in promoting screening uptake than on its influence on informed decision making.

| CON CLUS ION
An important finding of this review is that narratives that solely describe the decision-making process are hugely underrepresented in the literature on CRC screening narratives. Future studies must report on the purpose and the content of narrative interventions more clearly in order to understand which and how narrative types lead to which outcomes and for whom they work (and do not work) in the context of decision making about CRC screening participation.

ACK N OWLED G M ENTS
The authors thank Prof. Dr. Marie-Louise Essink-Bot and Prof. Dr.
Ellen Smets for their valuable contributions to this article. We also acknowledge Faridi van Etten for helping in our search for relevant studies. This study was co-funded by European Integration Fund (EIF) (Grant no. 2013EIF036).