Decision aids to prepare patients for shared decision making: Two randomized controlled experiments on the impact of awareness of preference‐sensitivity and personal motives

Abstract Objective To participate in shared decision making (SDM), patients need to understand their options and develop trust in their own decision‐making abilities. Two experiments investigated the potential of decision aids (DAs) in preparing patients for SDM by raising awareness of preference‐sensitivity (Study 1) and showing possible personal motives for decision making (Study 2) in addition to providing information about the treatment options. Methods Participants (Study 1: N = 117; Study 2: N = 217) were put into two scenarios (Study 1: cruciate ligament rupture; Study 2: contraception), watched a consultation video and were randomized into one of three groups where they received additional information in the form of (a) narrative patient testimonials; (b) non‐narrative decision strategies; and (c) an unrelated text (control group). Results Participants who viewed the patient testimonials or decision strategies felt better prepared for a decision (Study 1: P < .001, ηP2 = 0.43; Study 2: P < .001, ηP2 = 0.57) and evaluated the decision‐making process more positively (Study 2: P < .001, ηP2 = 0.13) than participants in the control condition. Decision certainty (Study 1: P < .001, ηP2 = 0.05) and satisfaction (Study 1: P < .001, ηP2 = 0.11; Study 2: P = .003, d = 0.29) were higher across all conditions after watching the consultation video, and certainty and satisfaction were lower in the control condition (Study 2: P < .001, ηP2 = 0.05). Discussion Decision aids that explain preference‐sensitivity and personal motives can be beneficial for improving people's feelings of being prepared and their perception of the decision‐making process. To reach decision certainty and satisfaction, being well informed of one's options is particularly relevant. We discuss the implications of our findings for future research and the design of DAs.

about the treatment options.
Methods: Participants (Study 1: N = 117; Study 2: N = 217) were put into two scenarios (Study 1: cruciate ligament rupture; Study 2: contraception), watched a consultation video and were randomized into one of three groups where they received additional information in the form of (a) narrative patient testimonials; (b) non-narrative decision strategies; and (c) an unrelated text (control group).
Results: Participants who viewed the patient testimonials or decision strategies felt better prepared for a decision (Study 1: P < .001, 2 P = 0.43; Study 2: P < .001, 2 P = 0.57) and evaluated the decision-making process more positively (Study 2: P < .001, 2 P = 0.13) than participants in the control condition. Decision certainty (Study 1: P < .001, 2 P = 0.05) and satisfaction (Study 1: P < .001, 2 P = 0.11; Study 2: P = .003, d = 0.29) were higher across all conditions after watching the consultation video, and certainty and satisfaction were lower in the control condition (Study 2: P < .001, 2 P = 0.05). Discussion: Decision aids that explain preference-sensitivity and personal motives can be beneficial for improving people's feelings of being prepared and their perception of the decision-making process. To reach decision certainty and satisfaction, being well informed of one's options is particularly relevant. We discuss the implications of our findings for future research and the design of DAs.

K E Y W O R D S
decision aids, patient experiences, personal motives, preference-sensitivity, Shared decision making

| INTRODUC TI ON
The question as to how shared decision making (SDM) between clinicians and patients can be improved is frequently asked in patient-education research. SDM is an approach to reaching medical decisions 'where clinicians and patients make decisions together using the best available evidence, where patients are encouraged to consider available screening, treatment, or management options and the likely benefits and harms of each'. 1 Including patients in medical decisions is particularly important in preference-sensitive situations, where there is no scientific evidence for the superiority of one treatment option. 2 In these situations, the decision should depend on individual circumstances, values and preferences of the patient. 3,4 Most patients wish to be included in medical decisions and prefer SDM to paternalistic or purely autonomous approaches. 5,6 Using SDM results more often in decisions that fit patients' preferences and has also been found to have positive effects on several outcomes, like satisfaction, 7-9 patient-physician relationship, 10 treatment co-operation and compliance. 10,11 However, there are reasons why SDM is often challenging for both physicians and patients which consequently discourage implementation in everyday clinical practice.
A major difficulty for successful SDM is the fact that most patients have little medical background knowledge. Gaining trustworthy knowledge about their diagnosis and treatment options for making an informed decision 12,13 thus becomes a big challenge, and many patients remember little about the treatment options after their consultation with a physician. 14 A popular approach to deal with this problem is the application of decision aids. [15][16][17][18] Decision aids (DAs) are evidence-based tools that are developed to support patients in making choices among health-care options, 19 complementary to personal consultations. They have been found to have positive effects on knowledge gain, [19][20][21] satisfaction with the decision process 20,21 as well as on active participation in decision making and awareness of personal values 19 in different medical fields.
Moreover, providing information alone is not enough to enable successful participation in medical decision making. [22][23][24] Many patients tend to underestimate the importance of their personal preferences in decision making. Joseph-Williams et al 22 concluded that research should therefore examine methods that enable patients to recognize this importance and prepare them for SDM. In their theory, Waldron et al 25 proposed that when patients are 'able to express their preferences and values through the implementation of SDM, then they experience higher confidence in their ability to participate in SDM, resulting in higher levels of SDM engagement' (p. 12). While most DAs aim to impart medical knowledge, they should also include additional material for decision support. 26 The studies presented here address this research gap.

| Decision support material
Two experimental studies investigated the potential of DAs that aim to prepare patients for and support SDM, in addition to being thoroughly informed about the treatment options. As many patients find it hard to grasp that their personal preferences matter in a medical decision, one way to prepare them for SDM is to explain the concept of preference-sensitivity better. But even if patients have understood that they themselves and the physician involved should be more sensitive to their personal preferences when medical decisions are made, it is still challenging for them to figure out what is particularly important to them. 27,28 So, for patients who have already understood the preference-sensitivity of the situation, the next step is to make them aware of potential personal motives in a decision.
This approach aims at giving them a clear idea of what their decision might be based on. Obviously, in preference-sensitive decision situations, this should be done in a way that does not manipulate patients towards one option.

| Narrative and non-narrative formats
There are different ways to explain the concept of preference-sensitivity and make people aware of different motives. In the studies presented here, we compared two different kinds of formats: narrative patient testimonials and non-narrative decision strategies.
Many DAs include narrative patient testimonials, 29-31 as patients perceive personal experiences of others combined with factual information to be very helpful for decision making. 32 Narrative formats have the advantage that they are vivid, easy to understand, and not abstract, making it easier for patients to comprehend and remember the information they contain. 33,34 They arouse interest and can support patients both in understanding their role in the decision-making process and in clarifying their personal preferences.
When using narratives in DAs, one needs to be careful, however, because reading about the experiences of others can bias decision making by triggering heuristic thinking. [34][35][36][37] According to Shaffer et al, 29,38 patient testimonials should focus on the decision process rather than on the outcome of this process.
Information regarding preference-sensitivity and motives can also be presented in a non-narrative format. Strategies for recognizing preference-sensitivity or for becoming aware of personal motives may be just listed without any personal context. While a non-narrative format appears to be less vivid and more abstract than narrative testimonials, it remains unclear which format is more helpful for people faced with a medical decision. In both studies presented here, we investigated the impact of narrative patient testimonials compared to non-narrative decision strategies.

| Research questions and hypotheses
In order to understand the potential of elements in DAs that prepare patients for SDM and are given in addition to detailed and balanced information about the options, we conducted two experimental studies. We addressed two aspects that may support patients in SDM: explaining the concept of preference-sensitivity (Study 1) and presenting personal motives (Study 2). Understanding preference-sensitivity is necessary to comprehend the concept of SDM, whereas becoming aware of possible motives underlying a decision implies that patients have already understood the concept of SDM and have started to consider their treatment options more closely.
While both approaches may be helpful to prepare patients for SDM, they address different steps in the decision-making process.
In both studies, participants received DAs that consisted of a video about a consultation with a physician and additional material.
The additional material differed between the conditions and was either (a) narrative patient testimonials, (b) decision strategies or (c) a text about an unrelated topic (control condition). Both studies used hypothetical scenarios, where participants were faced with a choice regarding a medical decision. In Study 1 (additional material about preference-sensitivity), we chose the scenario of suffering from a cruciate ligament rupture, a sports injury that can be treated either surgically or with intense physiotherapy. Neither treatment has been found to be clearly superior, 39,40 so patients in this situation were faced with an individual, preference-sensitive decision. In Study 2 (additional material about personal motives), we used the scenario of thinking about switching from an oral contraceptive pill to an intrauterine device (IUD) with copper. A decision about a contraceptive method is preference-sensitive, as there are many equally effective options, and women are aware of that. 41,42 Nevertheless, many women seek consultation when making this decision. 43 This scenario allowed us to examine the potential benefits of being presented possible motives underlying a decision.
For the selection of our outcome variables, we built on Elwyn and Miron-Shatz 4 who suggested that the evaluation of SDM should focus on patients' personal perceptions of being well informed, certain about their choice, and satisfied with the decision-making process, rather than on outcomes of the decision. In both studies, we measured participants' preparation for decision making, their decisional conflict and their evaluation of the decision process. In Study 1, we also measured participants' control preferences (ie the amount of control one wants to assume in the decision in terms of partly handing over decision-making powers to a physician). In Study 2, the control preferences were not included, because women demand to have personal control of their decision in the choice of a contraceptive method. 44 We hypothesized a positive impact of reading an additional element (narrative patient testimonial or decision strategies) compared to the control condition. Previous research suggests that encouragement to participate in the decision-making process as well as support regarding the formation of personal preferences may be helpful in addition to factual information. 22,45,46 In particular, we expected that participants who viewed a DA with such additional material would feel better prepared for the decision (H1a), show a stronger increase in decision certainty (confidence that the decision is right for them) and satisfaction regarding the decision (H2a), and evaluate the decision process more positively (H3a) than participants in the control condition. In Study 1, we also expected them to prefer a more active role in the decision-making process (control preferences; H4a).
We also hypothesized that the narrative testimonials would be more effective than the non-narrative decision strategies. Former research has shown that narratives in DAs can support patients in understanding their role in the decision-making process and in clarifying their personal preferences. 29,32,38,47 Consequently, we expected that participants in the narrative condition would feel better prepared for the decision (H1b), show a stronger increase in decision certainty and satisfaction (H2b), and evaluate the decision process more positively (H3b) than participants in the non-narrative condition. In Study 1, we also expected them to prefer a more active role in decision making (H4b).
In Study 2, as an exploratory research question, we measured participants' individual motives to examine whether the DAs resulted in changes of what they perceived as personally important and whether there were differences between the conditions.

| Participants
Participants for Study 1 were recruited from the participant database

| Procedure
In both studies, all instructions and questionnaires were presented to the participants on a computer screen. After reading the study description, participants answered demographic questions, read a general information text, and indicated their prior experiences with cruciate ligament injuries or contraception, respectively, in order to examine potential prior differences (see Measures for details).
Participants then read the fictional situation (Appendix S1) and responded to questionnaires (point of measurement 1; POM 1).
Afterwards, they watched a video showing a consultation with a physician/a gynaecologist. Only in Study 1 was there a measurement following the video (POM 2). This decision was made to investigate explicitly the impact of the different components of the DAs. In Study 2, we did not use POM 2 due to the timing constraints of the online study. Subsequently, participants were randomly assigned to one of three conditions and either read (a) two narrative patient testimonials, (b) non-narrative decision strategies or (c) an unrelated text (control condition). Then, participants responded once again to questionnaires ( Figure 3). To ensure that participants actually watched the video and read the texts, we included timers in the survey so that they could only continue to the next page after an appropriate time.
To avoid missing data, participants were forced to respond to all of the questions before they could proceed with the survey.

| General information text
In Study 1, the general information text consisted of 504 words and contained five pictures. It explained the location and function of the cruciate ligament, consequences of an injury, and treatment options (surgery or intense physiotherapy), all in a neutral way. It stated that in medical research neither treatment had been found clearly better than the other (Appendix S2).
In Study 2, the general information text provided basic information about the copper IUD and how it functions (47 words) including a picture of the location of the copper IUD in the uterus. On the next page, information about the security of different contraceptive methods was given (103 words; Appendix S3).

| Consultation videos
In Study 1, the consultation video took 07:30 minutes and showed a physician (portrayed by a white male actor in his late 30s) head-on.
The video had already been used in another study 28   In Study 2, the consultation video took 4:53 minutes and showed an interaction between a young female patient asking for information about the copper IUD and a female gynaecologist. This video had also already been used in a prior study. 41 The gynaecologist was shown head-on, allowing participants to put themselves in the patient's position, and was portrayed by a white actress in her 40s (Appendix S5).

| Patient testimonials and decision strategies
In both studies, the patient testimonials and the decision strategies were presented as content in an online portal, where patients could get information about a cruciate ligament rupture/about contraceptive methods and read about others' experiences. We created the material after an intensive literature search of patients' needs and experiences 50 as well as suggestions from previous research. 22,29 The material standardized the given information across the conditions and kept the content in the narrative condition and in the non-narrative condition parallel: The decision strategies and motives were the same as the ones mentioned in the narratives, but without any personal experiential information. The texts in the control condition were unrelated to the medical decision and dealt with the topic of migrating birds (Study 1) or potatoes (Study 2).
Following Shaffer et al, 29,38 the testimonials focused on the decision process and did not mention the outcome of the decision to avoid persuasive effects. To ensure that there was no persuasive effect, we asked participants which treatment they preferred and found no sig-  In addition to the dependent variables, we examined potential prior differences among the experimental conditions. Participants in Study 1 were asked for age, gender, treatment preference, frequency of physical activity and prior experiences with knee injuries.

| Measures
In Study 2, they were asked for age, preference, duration of pill use, desire to have children and prior experiences with the copper chain.
As a control question, in Study 2 participants were asked at the end of the survey whether they had read a patient testimonial, a list with decision strategies, or a text about potatoes.

TA B L E 1 Similarities and differences between Study 1 and Study 2
The similarities and differences between Study 1 and Study 2 are presented in Table 1.

| Prior analyses
In Study 1, no tested variable violated the assumption of normality or variance homogeneity (all P ≥ .069), and the repeated measurements did not violate the assumption of homogeneity of covariance matrices (P = .679). There were no group differences in Study 1 regarding age, gender distribution, treatment preferences, prior experiences with knee injuries or physical activity (all P ≥ .071). In Study 2, there were also no violations of the homogeneity of variances (all P ≥ .094), and there was no violation of the assumption of homogeneity of covariance matrices (P = .145) among the repeated measurements. For the preparation for decision making, the quantile plot revealed a non-normal distribution, but as previous literature has shown, the ANOVA is still robust even in the case of non-normal distributions. 62,63 The groups in Study 2 also did not differ in terms of age, preference, duration of pill use, desire to have children and prior experiences with the copper chain (all P ≥ .181). Table 2 shows the means and standard deviations for the dependent variables in Study 1, Table 3 those for the dependent variables in Study 2.

| Preparation for decision making
In Study 1, an ANOVA showed significant group differences in the feel- Contrary to Hypothesis 1b, there was no significant difference between the narrative and the non-narrative group, P = .670.  There were also group differences in decision certainty/ satisfaction, P < .001, 2 P = 0.09/ P < .001, 2 P = 0.13. Participants in the control condition reported significantly lower decision certainty/ satisfaction with the decision than participants in the narrative group, P < .001, d = 0.75/P < .001, d = 0.86, and participants in the non-narrative group, P = .002, d = 0.51/P < .001, d = 0.71. There were no significant differences in decision certainty/ satisfaction between the participants who had read the narratives or the decision strategies, P = .191/ P = .400.

| Decision evaluation
Contradictory to Hypotheses 3a and 3b, there were no group differences in the evaluation of the decision in Study 1, F(2, 107)

| Control preferences
Contradictory to Hypothesis 4, Kruskal-Wallis tests showed no group differences regarding control preferences at any POM in Study 1, all P ≥ .145.

| Exploratory analysis of the effect on motives
In Study 2, the motives for making decisions about contraceptives the participants rated as most important were the effectiveness of the method, well-being while using it, knowing the method,

| D ISCUSS I ON
In both studies, we found that participants who received additional material in the DA felt that the material prepared them better for the decision than participants in the control group. This effect is in line with the findings of Osaka and Nakayama 64 and the assumptions of Elwyn et al. 26 Decision certainty (in Study 1) and satisfaction (in both studies) were higher across all conditions after watching the consultation video. In Study 2, participants in the control condition reported lower decision certainty and lower satisfaction with the decision than participants in the narrative and non-narrative groups.
In Study 2, we also found that participants who had read either the narrative patient testimonials or the decision strategies evaluated their decision more positively than participants who had read the control text.

| Limitations
There are limitations to our studies that need to be taken into consideration when interpreting the results. One point is that our decision situations were hypothetical and therefore maybe not personally relevant for the participants. Actual patients who suffer from a cruciate ligament rupture, for example, may think and feel differently from our participants and be more intensely involved. Also, we did not ask for prior experiences with medical treatments or the health system in general, and such prior experience could certainly influ- While the two studies were similar in many aspects, they also had important differences which make it difficult to compare their results. Study 1 took place in the laboratory, had three POMs, and the decision situation was purely hypothetical for the participants. Study 2 took place online, had only two POMs, and the participants were faced with a decision that was closer to their real-life situation, because they had all made a decision about a contraceptive method at some point in the past. Conducting Study 2 online had the disadvantages that many participants did not finish the survey and we had less control of how much attention participants paid to the video and the texts. In order to reduce this problem, we included a minimum processing time and a control question, but we cannot know to what extent participants focused on the survey.

| Implications
The participants benefitted consistently from the DAs used in our studies. The DAs supported participants' understanding of treatment options, including risks and benefits, in a comprehensible and insightful way. Moreover, participants benefitted from additional support regarding their own preferences or motives. For the design of DAs, it seems highly recommendable to have a mixture of wellbalanced treatment information as well as support for patients in reflecting on their own preferences. It could be beneficial for physicians to apply such DAs as additional support for patients in the decision-making process, or at least as a help to patients in becoming aware of their own preferences and motives during medical consultations.

ACK N OWLED G EM ENTS
Open Access funding enabled and organized by ProjektDEAL.

CO N FLI C T S O F I NTE R E S T
No conflicts of interest are declared.

AUTH O R S' CO NTR I B UTI O N
All authors contributed substantially to the conception and design of this work. SK was involved in the acquisition of data. SK, ME and MB analysed and interpreted the data; all the authors contributed significantly to the discussion. JK, MB and ME drafted the manuscript; UC and SK commented on it and critically revised it.
All of the authors approved to the final version to be published; all of the authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work would be appropriately investigated and resolved.

E TH I C A L A PPROVA L
The research reported here was performed in accordance with the Declaration of Helsinki and had full approval by the local ethics committee (approval number: LEK 2019/016).

TR I A L R EG I S TRY
The studies were pre-registered on the pre-registration platform AsPredicted (aspredicted.org) prior to data collection (Study

PATI E NT O F PU B LI C CO NTR I B UTI O N
There were no patients, service-users or care-givers involved in our study. The participants in our study were mostly university students who took part in an online experiment.