When attitudes and beliefs get in the way of shared decision‐making: A mediation analysis of participation preference

Abstract Introduction Certain sociodemographic characteristics (e.g., older age) have previously been identified as barriers to patients' participation preference in shared decision‐making (SDM). We aim to demonstrate that this relationship is mediated by the perceived power imbalance that manifests itself in patients' negative attitudes and beliefs about their role in decision‐making. Methods We recruited a large sample (N = 434) of outpatients with a range of urological diagnoses (42.2% urooncological). Before the medical consultation at a university hospital, patients completed the Patients' Attitudes and Beliefs Scale and the Autonomy Preference Index. We evaluated attitudes as a mediator between sociodemographic factors and participation preference in a path model. Results We replicated associations between relevant sociodemographic factors and participation preference. Importantly, attitudes and beliefs about one's own role as a patient mediated this relationship. The mediation path model explained a substantial proportion of the variance in participation preference (27.8%). Participation preferences and attitudes did not differ for oncological and nononcological patients. Conclusion Patients' attitudes and beliefs about their role determine whether they are willing to participate in medical decision‐making. Thus, inviting patients to participate in SDM should encompass an assessment of their attitudes and beliefs. Importantly, negative attitudes may be accessible to change. Unlike stable sociodemographic characteristics, such values are promising targets for interventions to foster more active participation in SDM. Patient or Public Contribution This study was part of a larger project on implementing SDM in urological practice. Several stakeholders were involved in the design, planning and conduction of this study, for example, three authors are practising urologists, and three are psychologists with experience in patient care. In addition, the survey was piloted with patients, and their feedback was integrated into the questionnaire. The data presented in this study is based on patients' responses. Results may help to empower our patients.


| Relevance of participation preference
SDM has quantifiable benefits; the involvement of patients in treatment decisions is, for example, associated with better treatment adherence, [12][13][14] more accurate knowledge about their condition and recommended treatments 15 and less decisional conflict. 16 Ultimately, this leads to higher satisfaction among patients 16,17 and physicians 7 alike. Some studies even report a higher quality of life. 18,19 Beyond these sizable effects, SDM is also considered the ethical imperative for treatment decisions as it fosters patients' right to selfdetermination. 9,20 Consequently, most modern healthcare systems have implemented legislation to facilitate SDM. 8 Despite these benefits, not all patients seek participation in decision-making to the same extent. 15,[21][22][23] Although a general trend towards a stronger preference to participate has been observed over the last decades, 24 about a quarter of patients (Md = 27% in a review of 31 studies 23 ) express a low desire for autonomy. Research has recently started to identify reasons for this.
Existing research reported lower participation preference in patients with certain sociodemographic characteristics. In particular, older patients, less educated individuals and men 18,[25][26][27] express lower interest in active participation. However, a comprehensive review indicates that the exact nature and mechanisms involved in such sociodemographic variables' on participation preference remain to be specified in cancer patients. 28 Thus, one goal of this study was to investigate the association of sociodemographic variables on the participation preference of urooncological patients.

| Power imbalance as a potential barrier
It has recently been suggested that a seemingly low desire for patient autonomy may be due to a perceived lack of personal ability to participate rather than a genuine lack of actual desire. 29 In particular, the power imbalance (i.e., the perceived permission to participate) in the physician-patient relationship has been hypothesized to be a critical determinant of patients' ability to participate in medical decision-making. Patients often underestimate the value of their contributions to the consultation, that is, their ability to understand medical information and the importance of their personal preferences, 30,31 which contributes to an increased feeling of a power imbalance. Consequently, these individuals tend to delegate decisions to medical experts. The asymmetry in the physician-patient relationship is deepened if patients feel dependent on their physician's benevolence. 32 For example, patients may fear being labelled a difficult patient and the negative impact that such a label could have on their BÜDENBENDER ET AL. | 741 care. [33][34][35] Thus, they refrain from asking too many questions or asserting their interests, such as actively participating in decisionmaking. All too often, the normative belief prevails that a 'good' patient is characterized by conformity and passivity, which may overshadow patients' desire for autonomy. 36 Several studies report that even when patients witness a potentially harmful and preventable event during their care (e.g., a lack of hygiene measures), a majority of (cancer) patients opt not to speak up or assert their interest. [37][38][39] This phenomenon has previously been labelled as white-coat silence, 40 and it is closely related to patients' attitudes regarding their role in decision-making and beliefs about what constitutes a 'good' patient. This effect is particularly pronounced when the stakes are high, as is the case for life-threatening illnesses due to the increased dependence on the physician. In extreme cases, patients may even describe themselves as being hostages to the situation. 41 Considering patients' attitudes and beliefs about their role thus appears essential in optimized SDM. However, these patient variables are not routinely assessed even when SDM is carefully implemented. Therefore, it is crucial to investigate the potential barriers that attitudes and beliefs impose and may curtail patients' potential to participate in medical consultation actively. Thus, we seek to analyse the relationship between patients' attitudes and beliefs about their role and their participation preference.

| Attitudes' and beliefs' influence on behaviour
The relevance of attitudes and beliefs for behaviour preference (i.e., participation preference) is well-founded in psychological theories such as the theory of planned behaviour. 42 From this perspective, a particular behaviour (e.g., participation in decision-making) is preceded by an intention (or behavioural preference). 43 In the SDM context, this implies that the act of participating requires a certain participation preference.
The theory of planned behaviour postulates three important determinants for such a specific behavioural preference: attitudes towards the behaviour, control beliefs and normative beliefs. We, therefore, examine whether the relationship between patient characteristics and participation preference is mediated by attitudes towards the behaviour (participating in decision-making), control beliefs (presence of factors that facilitate or impede participation, e.g., lack of knowledge), as well as normative beliefs (what defines a 'good' patient).
Importantly, prior work has identified the potential of targeting attitudes and beliefs regarding SDM in healthcare professionals to foster change in behaviour intentions and behaviour itself. 44 However, there are no investigations on the role of attitudes and beliefs in patients.
1.5 | Current study: SDM in urooncology SDM has been used and evaluated in a diverse range of medical subdisciplines. A recent review by the American Urological Association (AUA) concluded that it is woefully underutilized 45 in the urological domain. However, SDM is increasingly requested in treatment guidelines and has proven to positively impact urological patients' knowledge, decisional conflict and quality of life. 45 Thus, in the context of a larger project to foster SDM in general urological practice, we set out to study barriers that impede urological patients' desire and ability to participate. Integration of SDM and engagement of patients is especially relevant in urology because patients in urology and urooncology often face very difficult and highly preference-sensitive decisions, 46,47 in which patient participation can have favourable effects. For instance, earlystage prostate cancer treatment demands patients and physicians to collaboratively decide among many treatment choices in equipoise, such as active surveillance, surgery and various forms of radiation treatments. 47,48 Regarding the participation preference of patients in urology, there have been a few studies on prostate cancer. In this patient group, there is a relatively high preference for participation compared to other tumour entities 22,23 ; however, around 10%-15% of patients report a low participation preference. 15,23,49 While previous research has typically addressed the influence of stable sociodemographic variables 18,25-28 on participation preference, we also aimed to gather information on potential mediators. We hypothesize that the previously reported association between certain patient characteristics and a low preference for participation 18,26,50 is mediated by a perceived power imbalance, which we expect to be reflected in patients' attitudes towards active participation, control beliefs and normative beliefs about the patient role.
With this research, we aim to identify potential targets for interventions that may help to empower patients who are reluctant to engage in SDM. Unlike stable sociodemographic characteristics, attitudes and beliefs are more accessible to change 43 and, therefore, may be ideal targets for structured interventions to help foster patient engagement and increase patient participation in the SDM process. For example, rather than assuming that older cohorts will most likely not be as open to active participation in decision-making, knowing more about the mechanisms in this process may encourage clinicians to pay close attention to their attitudes and beliefs.

| Participants
We recruited a sample of N = 468 consecutive patients who had a scheduled appointment or presented unscheduled with urgent symptoms at the outpatient clinic of the Department of Urology and Urosurgery at University Medical Center Mannheim, Germany.
We specified the following inclusion criteria: at least 18 years old, adequate proficiency in German and the ability to give informed consent. We excluded 3 patients with obviously invalid responses (i.e., only marked the middle response category on all questionnaires) and 15 patients (3.2%) with more than 50% missing data. In addition, 16 patients (3.4%) dropped out after giving informed consent.
Excluded patients (n = 31, 6.6%) were compared to those in the final sample (N = 434) by Pearson χ 2 -tests. When necessary (a cell with an expected cell count below one), we calculated Fisher's exact test. 51 There were no differences between excluded patients and the final sample regarding the type of appointment (appointment or unscheduled), diagnosis and occupational status (all p's > .117).

| Data collection and measures
Patients were approached before their consultation with a urologist by a nurse or a research assistant; they were informed about the protocol. After they signed informed consent, patients provided sociodemographic characteristics and filled in a set of self-report questionnaires, including the Autonomy Preference Index (API) 53,54 to assess participation preference and the Patients' Attitudes and Beliefs Scale (PABS). 55 Additional questionnaire data, for example, on patients' anxiety and depression, are reported elsewhere. 56,57 Medical diagnoses were retrieved from patients' electronic health records.

| Demographic and medical variables
We collected standard sociodemographic information about patients' age, gender, level of education, marital status and living arrangement.
We dichotomized the following variables: education (0 = without and 1 = with a university degree), living arrangement (0 = living alone or 1 = with others) and occupation (0 = currently unemployed/retired or 1 = currently either employed/in training). In addition, medical diagnoses were categorized as (0) nononcological versus (1) oncological. For consistency, we applied dichotomization cut-offs established in previous studies. 56,57

| Participation preference and intention to participate
The primary outcome of our study is patients' behaviour preference for participating in medical decision-making. Patients' preferences for involvement were measured with the German version of the API. 53,54 The measure consists of two subscales: decision-making (API-dm) and information-seeking (API-is). The German API-dm subscale has four items, which assess generic participation preference, and the API-is contains seven items, which assess information seeking. On both subscales, a linear transformation (0-100) is applied. 58 Higher scores indicate a stronger desire for autonomy. We report descriptive statistics for the API-is subscale. However, because there were high ceiling effects and little variance, the subscale was not used in further analyses (compare, e.g., Benbassat and colleagues 27,59,60 ). The primary outcome of our study, patients' participation preference, was thus assessed with the API-dm subscale. The API-dm is often used on its own to assess participation preference, and it has been validated in various languages (e.g., German) and settings. 24,61 Internal consistency for the scale was good α = . 85. 58 In addition, we generated an item to ask patients about their intention to participate: 'Regarding your upcoming consultation: Do you intend to participate in decision-making?'. The item was rated on a 6-point Likert Scale ranging from ('Absolutely not' to 'Absolutely yes').

| Attitudes and beliefs
We assessed patients' attitudes and beliefs about their role in shared treatment decisions using the PABS. 55 The PABS was translated to

| Statistical analyses
A drop-out analysis was conducted to assess systematic differences between completers and noncompleters (see Section 2.1). We compared patient characteristics for patients with urooncological versus nonurooncological diagnoses with χ 2 -tests. Furthermore, patient subgroups were compared (independent sample t-test) regarding participation preference, attitudes and beliefs. Univariate descriptive statistics are reported for participation preference, attitudes and beliefs and intention to participate, along with the Pearson correlation coefficients for the bivariate associations between these variables. The association between nominal scaled sample characteristics (e.g., gender) and the dependent variable participation preference were analysed with the correlation ratio η.
The expected mediation (i.e., sociodemographic variablesattitudes and beliefs-participation preference) was assessed with a path model (based on structural equation modelling). We checked the multivariate normality assumption between all included endogenous variables with the MVN R-package. 62 The path analysis was calculated with the lavaan R-package 63 using R. 64 The complete list of the R-packages is listed in Supporting Information: Appendix B. All other analyses were conducted in IBM SPSS, Version 27.0. Significance for all tests was set at α = .05. Where applicable, we report effect sizes; interpretation is based on Cohen's taxonomy. 52 3 | RESULTS  Table 1 for further comparison of both subgroups. However, the two diagnostic groups did not differ in their attitudes and beliefs, intention to participate or the primary outcome participation preference (all p's ≥ .104).

| Patients' participation preference and associated variables
Overall, most patients wanted to participate to some degree in  Table 2. Additionally, we included bivariate correlations in Table 2   The associations of patients' gender and type of diagnosis with participation preference were nonsignificant (p ≥ .14) and, thus, excluded from the multivariate mediation analysis. Furthermore, we excluded occupation from the mediation analysis due to substantial multicollinearity with age (η = 0.7, p ≤ .001).
Based on the results of the bivariate analyses described above, the following variables with a significant association with participation preference were included in the mediation model: age, education and living arrangement as exogenous predictors; positive and negative attitudes as endogenous mediators of the effect and participation preference as the endogenous outcome variable.  We provide an overview of all effects in the path analysis in Table 3. All three exogenous predictor variables (patients' age, education and living arrangements) significantly affected participation preference (Table 3) With reference to earlier studies, we replicated the association of participation preference and sociodemographic characteristics, 26  Overall, the predictors in our mediation path model explained around 28% of the variance in patients' participation preferences, which corresponds to the strong explanatory power of our model. 52 In previous research, multiple regressions were mainly used to analyse factors influencing participation preferences and explained between 3% and 21% of the variance. [66][67][68] While our mediation path model explained a substantial amount of variance and, more importantly, uncovered the mediating effects of attitudes and beliefs, T A B L E 3 Effects in the mediation analysis between sociodemographic variables and participation preference Patients are thought to be especially prone to experiencing power imbalance in the relationship with their physicians when the stakes are high, for example, when they face a life-threatening disease. 41 However, we did not find a difference between the oncological and the nononcological subsamples. A possible explanation is that the nononcological patients in our sample presented with complex and long-lasting medical conditions and consequently faced substantial burdens as well (see Köther et al. 57 ).

| Mediating effects of patients' attitudes and beliefs
In line with this, it is possible that patients' attitudes, beliefs and participation preferences are not directly associated with illness severity but rather with the perceived burden and impairment associated with it.
In addition, there were significant differences between the oncological and the nononcological subsamples, for example, the urooncological sample contained more male patients, they were on average older and more were retired. Thus, the high homogeneity in this sample and lack of variance could have limited the possibility of finding differences in attitudes and beliefs. This might limit the generalizability to the bigger concept of the influence of illness severity (i.e., life-threatening or not). However, it does not speak against the generalizability to urological patients, as the described differences in the distribution of gender, age and occupation are typical in urology versus urooncology.
Regarding the information preference of the patients, we found considerable ceiling effects in the API-is, similar to previous reports. 27,59,60,70 Unfortunately, the lack of variance obtained by the measurement prevented us from including it in further analysis.
However, it is important to recognize that urological and urooncological patients generally have a genuinely high preference for being informed, even when reporting a lower preference for participation.
The finding of a universally high preference for information is consistent with studies on other patient groups, for example, patients at the general practitioner 70 or patients with end-stage renal disease 71 or a study with over 5000 older adults who all have a very high desire for information. 72 Thus, to treat uro(onco)logical patients according to their preferences, they should readily receive information on diagnostics, treatment options and side effects regardless of their preference for participation in the decision-making process.

| Limitations
One limitation of our study is the utilization of the attitudes and beliefs scale as a proxy for a power imbalance. Even though attitudes and beliefs, as measured by the PABS, 55

| Conclusion and future directions
With increasing international efforts to better implement SDM, 8,73 it is important to identify barriers that limit patients' ability to participate actively. We found low participation preference in older  79 and thus, they may miss an opportunity to pave the way for SDM. Also, participating in online self-help groups was shown to reduce negative attitudes in (cancer) patients and should be encouraged by clinicians. 80 Implementation of SDM in urology and urooncology is especially important, 45 given that patients, in the course of their treatment, face many preferencesensitive decisions. 47,48,81 Research on SDM has documented the positive impact patient participation can have in urology. [82][83][84] At its core, SDM requires a fundamental shift in power, 85 which can only be achieved by the active empowerment of patients in urology. Future studies should examine if targeting attitudes and beliefs, for example, in decision aids, holds its promise and helps empower patients in the process of SDM.
It is important to consider that patients' negative attitudes and beliefs and the accompanying fear of being perceived as a 'bad patient' may arise from experience. A patient may become quickly disheartened if he/she is willing to participate but experiences a lack of time or responsiveness from the clinicians. In line with the title of a recent SDM intervention study, 'Changing patients but not physicians is not enough', 77 it could be a valuable measure for SDM adaption to also prepare physicians to embrace their patients' willingness to participate in decision-making.
Finally, our study highlights the importance of patients' attitudes and beliefs regarding the perceived power asymmetry before the consultation for their active participation. Building on that, an interesting path for future studies would be the investigation of the experienced power asymmetry in the medical encounter and the influence on the perceived participation after the encounter. resources. All authors approved the final version of the report.