Participation preferences of health service users in health care decision‐making regarding rehabilitative care in Germany—A cross‐sectional study

Abstract Background Involving patients and citizens in health care decision‐making is considered increasingly important in Germany. Participatory structures have been implemented, especially in rehabilitative care. However, it is unknown whether and to what extent German patients and citizens want to participate in decisions that exceed their own medical treatment. Objective This study aimed to survey participation preferences and associated factors of health service users in decisions regarding rehabilitative care at micro, meso and macro levels. Methods A questionnaire was sent to 3872 former rehabilitants. We collected participation preferences using the Control Preference Scale or an adapted form. Possible influencing factors were examined using logistic regression models. Results The response rate was 5.7% (n = 217). At all decision‐making levels, joint decision‐making was preferred. At the macro level, preferences for actively participating were the highest. Preferences were significantly interrelated between decision‐making levels. At the micro level, an orthopaedic indication significantly decreased the desire for participation compared to psychosomatic indications (odds ratio = 0.44, p = .019). Discussion Participants wanted to be equally involved in decision‐making as experts. Higher preferences for active participation at the macro level might be due to dissatisfaction with the current health care organisation and lack of trust in politicians. Compared to the general public, our study sample was older (73.3% between 50 and 69 years) and more often chronically ill—factors associated with increased participation preferences in the literature. Conclusion Contrary to the identified preferences, participation opportunities in the German health care system are rare. Further research on participation preferences and structures that enable meaningful involvement are needed.

1 | BACKGROUND 1.1 | Relevance of public and patient involvement in health care decision-making Public and patient involvement (PPI) in health care decisions is increasingly important in Germany. [1][2][3][4][5][6] Not only is involvement in decisions at the micro level discussed but also involvement at the meso and macro levels. At the micro level, patients can be involved, for example, in decisions regarding their own treatment, the medical agenda or the place of treatment. 7,8 In our study, we focus on treatment decisions. Meso-level decisions concern particular geographical regions or health care facilities, whereas macro level decisions concern the whole health care system on a nation, state or province level (e.g., the financing and organisation of the overall service provision). 9,10 While micro level decisions are referred to as individual, meso and macro level decisions can be summarized under health policy decisions. 11 The relevant literature on PPI in health care decisions indicates that PPI leads to improved health care. At the micro level, patient participation leads to increased quality of the decision-making process, 12 improved patient knowledge 12,13 and higher patient satisfaction. [13][14][15] At the meso level, it can result in more patientcentred care as well as improved care processes and health outcomes of health care facilities. [16][17][18][19] At the macro level, PPI can ensure patient-oriented health policy, leading to more patient-friendly structures and improved service delivery. 17,18,[20][21][22] As the definition of PPI varies in the literature, we define PPI for the purpose of our study as the involvement of health service users in health care decision-making processes. Health service users include patients who are acute users of health services as well as citizens who are past and potential users of health services. [23][24][25]

| Previous research on participation preferences of citizens and patients
When thinking about increasing PPI in health care decisions, the question arises, whether and to what extent patients and citizens want to participate. While evidence on participation preferences at the micro level is increasing, 13,15,[26][27][28][29][30][31][32][33][34] participation preferences in health policy decisions have been less studied. The majority of studies surveying preferences at the micro level indicate that patients prefer a collaborative decision-making process, 26,28,[32][33][34] while two systematic reviews with a focus on oncological care found that patients prefer a passive decision-making process. 27,31 Studies focusing on health policy decision-making indicate varying preferences, but most conclude that the public prefers a consultative role. [35][36][37][38][39][40][41][42][43][44][45] The final decision is rather left to physicians, [37][38][39]41,[43][44][45] traditional decision-makers (e.g., elected officials, experts or politicians), 35-37 a multiprofessional group 40 or to patients and their families. 43 So far, participation preferences have been studied either only at the micro level or only at the health policy level, except for Fredriksson et al., 38 who emphasized that looking at this together will lead to a deeper understanding of the requirements for PPI in health care decision-making. However, how participation preferences between different decision-making levels are interrelated has not been studied yet.
Factors associated with increased desire for participation in health care decisions at all three decision-making levels include missing trust in the health care system or in physicians 36,38,46,47 and female sex. 13,32,36 Rising age is associated with an increased desire for involvement in health policy decisions initially. Only in very old age groups do participation preferences decrease. 36,38,46 In contrast, at the micro level, younger age is associated with increased preferences for participation. 13,26,27,32,33 The influence of education on participation preferences at the micro and macro levels is controversial-in some studies, higher education and in other studies lower education led to increased participation preferences. At the micro level, it is further suggested that participation preferences vary between indications and disease patterns. 13,34 For health policy decisions, it is emphasized that participation preferences can vary between countries and care settings due to different democratic understanding and culture or the organisation of health care.
Therefore, considering the context while assessing participation preferences is important. 38,41

| Participation of citizens and patients in the German setting of rehabilitation
The opportunities for citizens and patients to participate in decision-making processes differ within the German health care system. In rehabilitative care, participatory structures are already further developed compared to other health care settings. [48][49][50][51][52][53] This can be seen, for example, in its unique legal anchoring: Rehabilitants should be involved in their own treatment but also in the organization and evaluation of rehabilitative services (German Social Code IX). Different approaches for PPI have been implemented as a result (e.g., patient involvement in quality assurance or development of therapy standards). 48 However, to implement participation opportunities that correspond to the patients' and citizens' preferences and are perceived accordingly by them, these preferences must be known. We identified only one study that assessed participation preferences in rehabilitative care at the macro level (by one selfdeveloped question). 6 Seventy percent of the included 50 patients from an inpatient cardiac rehabilitation site wanted to be involved in decisions concerning the financing of health services. Studies on participation preferences in overall health policy indicate that German citizens do not see their interests well represented and that they see a need for greater involvement of citizen and patient representatives. 58

| Study sample and data collection
To determine participation preferences, we conducted a crosssectional survey of health service users who previously received rehabilitative treatment in three inpatient rehabilitation centres of the German Pension Insurance Oldenburg-Bremen between August and December 2020. The study was approved by the responsible Ethics Board (number 2019-150). Persons were eligible for inclusion if they had completed psychosomatic or orthopaedic rehabilitation at one of the three facilities in 2019 (the discharge report was available).
Normally, patients stay in a rehabilitation centre for approximately 3 (orthopaedic) or 6 (psychosomatic) weeks. Due to their recent rehabilitation stay, study participants had individual experiences as patients, but could also take the broader public perspective as they had already completed their inpatient rehabilitation treatment.
The main diagnoses treated in the rehabilitation centres are depression, burnout, personality and behavioural disorders or anxiety disorder for psychosomatic rehabilitation and diseases of the musculoskeletal system, related chronic pain and psychosomatic comorbidities for orthopaedic rehabilitation. A survey questionnaire was sent out via post to 3872 former rehabilitants. The participants could decide whether they wanted to fill out the online survey or the paper-based survey.

| Survey
Our survey was embedded in a larger study on action and research need in rehabilitative care from the viewpoints of rehabilitants and people working in rehabilitative care. The questionnaire consisted of three questions regarding participation preferences in decisions on rehabilitative care at the micro, meso and macro levels (Table 1).
Additionally, questions about sociodemographic data, the type of indication and satisfaction with one's own rehabilitation were included.

| Survey of the main outcome variable
Participation preferences at the micro level were measured using the standardized and validated Control Preferences Scale (CPS) (Q1). 61 The CPS is considered a reliable tool to measure preferences in health care decisions and is frequently used in the literature for this purpose. 26 The adapted version of the CPS for macro level decisions is shown in Figure 1. To achieve a sufficiently large number of responses in each answer category, we used the three merged categories active, joint or passive for some statistical analyses (see Figure 1).

| Survey of study participants' characteristics
In addition to participation preferences, we obtained information on Before further analyses, we imputed missing data using the fully conditional specification method (MICE). MICE is recommended for data sets containing variables of different types and allowed us to take the uncertainty about the imputed value into account by imputing multiple times. Therefore, parameter estimations are less biased. 62,63 We assumed that our missing data are missing at random.
Overall, we created 40 imputed data sets as recommended by Azur et al. 62 For imputation, we used the "mice"-package in R. 64 To assess the possible influencing factors on participation preferences, we ran an ordered logistic regression model on each of the imputed data sets. As independent variables, we considered age, gender, education and the indication, as these are mentioned as important predictors in the literature (see Section 1). Furthermore, we considered satisfaction with one's own rehabilitation as we assumed that this is related to trust in the physician and could therefore be an important predictor. We set the significance level for the regression analyses to a two-sided p value of less than .05. We pooled the results of the regression models to one outcome set and calculated odds ratios for the participation preferences depending on the variation of the independent variables.
To assess differences in the distribution of preferences between decision-making levels, we conducted a Friedman test, followed by a Nemenyi post-hoc test for pairwise comparisons. To check whether the preferences for an active, joint or passive decision-making process are correlated between different levels, we conducted a χ 2 test of independence. When there was a significant correlation, we calculated the Spearman's rank correlation coefficient to assess the strength of the correlation. The significance level was set to a twosided p value of less than .05.
For the statistical analyses, we had to exclude one case, where the answer for gender was diverse, as we had only one person in this category. A statistical analyses was, therefore, not reasonable. We tested, however, whether the assignment of this person to the group female or male would lead to a significant difference in the results.

| RESULTS
Of the 3872 former rehabilitants contacted, 90 could not be reached.
A total of 217 persons participated in our study (response rate 5.7%).
Slightly more than half of the participants were male (52.1%).
The majority were between 50 and 59 years old (53.5%), had a secondary school diploma (70.5%) and had participated in orthopaedic rehabilitation (57.6%). An overall overview of the characteristics of the study participants is shown in sample with the overall study population for these variables, our sample differs only slightly from the overall study population.

| Descriptive results
The participation preferences at all three decision-making levels are shown in Figure 2.

| Differences and interrelationships in participation preferences between decision-making levels
The variation in the distribution of participation preferences between decision-making levels is slightly significant (χ 2 = 7.30, p = .026).
In post-hoc pairwise comparisons, this significant difference could not be verified.
Between the participation preferences at the different decisionmaking levels, we identified significant correlations (p < .001 for all combinations of decision-making levels, see Table 3). We found a large positive correlation between participation preferences at the micro and meso levels (r s = .55, p < .001), a medium positive correlation between preferences at the meso and macro levels (r s = .34, p < .001) and a small positive correlation between preferences at the micro and macro levels (r s = .21, p < .01). 65

| Influencing factors on participation preferences
At the micro level, we identified that participants with an orthopaedic indication are significantly less likely to want to be involved in individual treatment decisions compared to those with a psychosomatic indication (p = .019). At the meso and macro levels, we did not observe this correlation. We also did not find any influence on participation preferences regarding gender, age, education or satisfaction with one's own rehabilitation. The results of the regression analyses are shown in Table 4.

| DISCUSSION
In this study, we identified the participation preferences of health service users in decisions regarding rehabilitative care at the micro, meso and macro levels as well as associated factors with these preferences. Our findings indicate that study participants prefer to be equally involved as experts in decision-making. At the very least, F I G U R E 2 Preference for joint decisionmaking at all three levels of health care decisionmaking (missing data not shown in the graphic) they want their interests to be heard and considered, even when preferring to not actively make the final decision themselves.
In the following, we discuss participation preferences at the individual decision-making levels, existing differences and interrelationships between these levels and influencing factors on participation preferences. Finally, we discuss the practical implications of our findings.

| Participation preferences at the micro level
The desire for a joint decision-making process was strongest at the micro level, and simultaneously, we found the strongest rejection of a

| Participation preferences at the meso level
At the meso level, our study findings indicate a stronger desire to be involved in the decision-making process compared to previous studies, which conclude that a consultative role is preferred. [39][40][41][42]45 This difference may be due to the setting of rehabilitation. As mentioned in the Section 1, the principle of participation is already widly implemented in some parts of this care setting. 48,49 Furthermore, rehabilitants must request their rehabilitation stay themselves, so they already need to be actively concerned with their own health care.
Study participants may therefore be more familiar with participating in health care decisions. Our study participants were also older and mostly chronically ill, which is typical for rehabilitants and associated with an increased desire for involvement in health policy decisions. 36,38,46,66 Fredriksson et al. 38 assumed that this is because of the more frequent contact with the health care system and personal concern. Therefore, compared to the general public, participation preferences for rehabilitants might be higher.

| Participation preferences at the macro level
As for the meso level, we identified higher participation preferences at the macro level compared to other study findings. [35][36][37][38][39][40]44 The setting and the characteristics of the study participants can serve as possible explanatory factors here as well. Surprisingly, we found the highest preferences not only for an active but also for a passive form of participation at the macro level. Dissatisfaction with the current organization of rehabilitative care might also explain the higher preference of actively participating in macro-level decisions. 38 As mentioned in Section 1, the majority of the German population feels that their interests are not well represented in health policy, indicating a need for greater PPI from their viewpoint. 58,59,68 The lack of direct personal concern at this level, on

| Differences and interrelationships in participation preferences between decision-making levels
The results indicate that participation preferences are not equally distributed across the decision-making levels. As the Friedman test was just significant (p = .026) and we could not identify any significant difference for pairwise comparisons in the more con-

| Practical implications
As PPI in health care decision-making is related to improved health systems and patient-oriented care [12][13][14][15][16][17][18][19][20][21][22]  Our results may also have been influenced by the quantitative questionnaire format, which could lead to different participation preferences than qualitative methods, where the option for discussion and clarification of questions exists. 39,41,43 We used closedended questions as we expected a higher response rate, 78 to include a larger sample size and because of the complexity of the topic. We believe that close-ended questions are easier to understand and answer for participants. A qualitative research approach to gain insight into the reasons for different participation preferences would be interesting for further research.
We also recognized that the CPS is criticized in the literature for only measuring one aspect of decision-making at the micro level when focusing exclusively on treatment decisions. 7,8 More options of decision-making exist at the micro level and ignoring them might lead to inaccurate assumptions on participation preferences. This problem is also inherent in other comparable scales. Since the CPS proved to be a reliable and easy-to-understand instrument, was easily adaptable for the meso and macro levels due to its single-item characteristic and scale values can be used for ordinal regression analyses, we decided that the CPS was a suitable instrument for our purpose. As we used the adapted versions of the CPS for the first time, further validation would be necessary.

| CONCLUSION
The majority of the study participants wanted to be equally involved like experts at all health care decision-making levels (micro: 65.9%, meso: 55.8%, macro: 39.6%), regardless of age, gender, education or satisfaction with the received rehabilitative treatment. At the micro level, the patients' indication influenced their preferences.
Contrary to the identified preferences, PPI in health policy decisions in Germany is in its infancy. 4,49,54,72,73 Health care providers and financers are still the central decision-makers. The successful implementation of PPI depends on the willingness of policy-makers to redistribute decision-making power and on the motivation of service providers to meaningfully involve patients and citizens. It also depends on the motivation of patients and citizens to become involved. As participation preferences between decision-making levels were significantly correlated, patients can already be motivated and empowered to participate in health care decision-making at the micro level. This might be a good place to start fostering PPI also in health policy decisions.
To implement appropriate methods, research on the participation preferences of the general German population is needed. Further, it needs to be investigated how citizens and patients would like to become involved and what they need to be able to participate. It would also be interesting to assess why people want to or do not want to become involved. A qualitative research approach could provide valuable information here. Based on this evidence, the implementation of PPI and increased patient-centredness of the German health care system can be pushed further ahead.