The influence of psychosocial factors on the intention to incorporate complementary and integrative medicine into psychiatric clinical practices Complementary Therapies in Clinical Practice

Complementary and integrative medicine (CIM) can be of great support to individuals suffering from psychiatric conditions; however, it is still rarely incorporated into clinical practice. Objective: To examine the influences of psychosocial and sociodemographic factors on health-care professionals ’ intention to use CIM in their psychiatric clinical practice. Method: One-hundred-and-five participants completed a questionnaire developed from an adapted version of Triandis ’ Theory of Interpersonal Behavior (TIB). Intentions to use CIM (yes or no) were analyzed using logistic regression models. Results: The multivariate model retained three main factors: affect, perceived social norms, and conditions facilitating CIM. These predicted health-care professionals ’ intention to use CIM with an AUC = 94.7%. Results: underlined that positive affective attitudes towards CIM, feeling that CIM was congruent with professional and institutional goals, and having sufficient skills in CIM were essential to ensuring that health-care professionals would integrate CIM into their clinical practice.


Introduction
Mental health disorders represent one of the largest parts of the global burden of disease [1][2][3]. One study showed that almost one in five adults (17.6%) had suffered from a common mental disorder in the past 12 months and that 29.2% had suffered from one in their lifetime [1]. Mental disorders cause acute suffering and significant loss of autonomy [4]. Indeed, they are the principal cause of disability globally, generating significant social and financial costs [2,5]. These illnesses also impact the social environment, with a risk of secondary psychological difficulties among family and friends [6,7], which may lead to increased feelings of loneliness [8,9]. Thus, mental health patients often suffer a significant reduction in their quality of life [4,6]. These multiple problems, in addition to the potential side effects of medication, such as weight gain and a decreased libido [10], can reduce medication adherence [11]. The complexity of the situations that patients with mental illnesses find themselves in, however, requires combined approaches to care [3,12]. Associating non-medicinal and pharmacological approaches is common [13].
Indeed, this may explain why it is relatively common for psychiatric patients in Western societies to use complementary and integrative medicine (CIM) to supplement conventional treatments [3]. As its use is rising worldwide [14][15][16], it is estimated that 16%-44% of people with mental illness also use CIM [17]. Indeed, 78.3% of people with a 12-month DSM-IV disorder and using complementary and alternative medicine reported being 'satisfied/very satisfied' with it [3]. CIM is recognized as improving individuals' physical and emotional well-being, with more than 96% of patients who use it perceiving benefits [18]. Integrative medicine is defined as the coordinated and selective incorporation of elements of complementary and alternative medicine into standard medical treatment plans derived from conventional diagnostic methods [19,20]. CIM approaches include diverse beliefs and practices, such as spiritual therapies, products derived from plants, animals, or minerals, manipulation-based techniques, and exercises, none of which are generally considered part of conventional medicine [16,20,21].
Providing psychosocial, person-centered care [22] is essential to avoid the dramatic consequences of mental illnesses in terms of stigmatization, social isolation, and reduced quality of life [12,[23][24][25]. Many health-care professionals perceive CIM as a useful tool for supporting psychiatric patients, one which can contribute to person-centered care [26][27][28]. Indeed, CIM has become an unavoidable part of psychiatric clinical practice [3]. This trend is in line with the WHO's Traditional Medicine Strategy, which aims to promote a broader vision of care, targeting improved health, person-centered care, and the development of patient autonomy through the high-quality, safe, and effective practice of CIM [16,21]. Despite all these benefits, CIM is not invariably incorporated into clinical practices; incorporation is very heterogeneous, depending on the hospital unit or medical specialty [29]. In the French-speaking part of Switzerland, 50% of hospitals have indicated that they offered at least one complementary and alternative medicine therapy [30]. One third of hospitals in Denmark offer CIM [31]. That rises to 64.4% in Norway, where no major differences were found between somatic and psychiatric hospitals [32].
Health-care professionals in somatic units face various barriers to integrating CIM, and their use of it depends on several types of factors [27,28,[33][34][35][36]. This also may explain why its use varies considerably from one health specialty or hospital unit to another [16,29]. Given patients' and health-care professionals' growing interest in CIM as part of routine psychiatric treatments, describing and understanding the multiple factors affecting whether health-care professionals are able to incorporate CIM into their clinical practice seems essential. To the best of our knowledge, no studies to date have examined the factors explaining health-care professionals' use of CIM in psychiatric hospital settings based on the psychosocial factors in the Triandis' Theory of Interpersonal Behavior (TIB). Furthermore, the role of sociodemographic factors remains little-studied even though it is known that they can influence the integration of CIM into professional health-care practices. Improved knowledge in these areas will provide a better understanding of what encourages professionals to integrate CIM into psychiatric health-care.

Theoretical framework
According to the literature, adopting a new professional behavior depends on a wide range of psychosocial factors [37,38]. Triandis' original TIB [39] explored cultural, social, and moral elements to create a better understanding of behaviors involving personal or ethical responsibilities during the development of new clinical practices in psychiatry, such as CIM [40], which was not yet a part of standard treatment and practice [35].
In the TIB, the adoption of a specific human behavior is facilitated by the three direct antecedents of habit, intention, and conditions. Habit and intention vary according to the novelty of the behavior being studied. The strength of intention is critical when a behavior is new, whereas the strength of habit increases with the degree of automatism that may result from the behavior's repeated performance. We considered past behavior instead of habit because incorporating CIM into clinical psychiatric practice is not yet the standard professional clinical practice expected within institutions. Habit corresponds to the frequency with which a behavior has been adopted in the past, and some studies have indicated that this might explain 7.2% of the variance in intention [41]. Facilitating conditions are elements that might aid or impede a behavior. Previous studies using the TIB found that the factors of habit and facilitating conditions were directly related to intention and thus were important predictors of intention [42], even though the original TIB model conceptualized them as the direct antecedents of behavior [39]. The intention to adopt a new behavior depends on several factors: perceived consequences, affect, perceived social norms (including both normative beliefs and perceived social roles), and personal normative beliefs. Perceived consequences are the result of a subjective, personal cognitive analysis of consequences as the advantages or disadvantages which might result from a specific behavior [40]. Affect represents an individual's emotional responses to the adoption of a new behavior. These responses result from the person's previous experiences; they involve long-term memory and generate positive or negative feelings that will lead to psychological states that the individual will seek to reinforce if positive or stop if negative [40]. Perceived social norms are composed of two sub-components. The first sub-component corresponds to normative beliefs, as defined by Fishbein and Ajzen [40,43]. These result from a personal subjective analysis of the opinions of individuals or groups that the individual considers important with regard to the behavior being studied. The second sub-component corresponds to a perceived social role representing a personal assessment of the relevance of adopting or not adopting a new behavior in relation to one's reference group. These norms differ according to the prevailing norms of societies and social groups. Personal normative beliefs result from a personal assessment of the relevance of adopting a behavior according to one's values and principles and, therefore, from the obligation that a person feels to adopt this behavior [40]. The person no longer refers to the opinions of others, as these would be social norms.
Some authors have added other factors to the original TIB to study health-care professionals' behaviors because the specificities of contexts and situations may influence the adoption of a new professional behavior [44,45]. One systematic review reported that descriptive norms explained an extra 5% of the variance in intention [46] and seemed to constitute a statistically significant determinant of behavior. Descriptive norms are respondents' perceptions about the adoption of a particular behavior by other individuals. The factor of self-identity-which is not in the original TIB model-had a negative strength in the prediction of intention (β = − 0.33) in Gagnon's study [44]. Self-identity is the degree of similarity between an individual's perception of themself and the characteristics that they associate with a particular behavior. Indeed, health-care professionals could perceive the use of CIM in clinical practice to be either positive or negative, especially with regard to energy medicines, which they are frequently skeptical about [47]. Sociodemographic variables were also included, such as type of health-care profession, sex, and the number of years of professional experience. These, too, are considered to influence professionals' intentions to use CIM [35,44].

Study objectives
Based on an adapted TIB model ( Fig. 1), the present study aimed to determine the influences of psychosocial factors (perceived social norms, facilitating conditions, affect, perceived consequences, descriptive norms, past behavior, personal normative beliefs, self-identity) and sociodemographic factors on health-care professionals' intention to use CIM in their psychiatric clinical practice with adults.

Study population
This study was conducted in the French-speaking part of Switzerland among general practitioners (GPs), nurses, and paramedical staff (physiotherapists, occupational therapists, psychologists, and social workers) working in Lausanne University Hospital's northern Vaud psychiatric care units. The focus group and the interviews took place from April to May 2018. The survey took place from February to March 2019 through a self-administered questionnaire sent by email. Email addresses were provided by northern Vaud's psychiatric care units.

Ethics
The study's nature and purposes were explained in an email sent to both its focus group participants and the survey participants before the administration of the online questionnaire. By participating in a focus group or filling out the questionnaire, respondents gave implicit voluntary consent to their study participation. All the data were coded to protect anonymity and confidentiality. The Human Research Ethics Committee of the Canton of Vaud validated the study (no. 2018-00518) with respect to all the rules applicable to research on human beings.

Instrument development process
The instrument was developed following a mixed qualitative and quantitative approach [48].

Item generation
Questionnaire items were generated using a qualitative approach [48]. We led two 1-h focus groups and six individual interviews with participants representative of the target population. The focus groups and interviewees were asked identical open questions, with the aim of drawing out as many of their beliefs about CIM as possible, covering each of the factors in the adapted TIB model (Fig. 1.) Audio recordings of each discussion and interview were transcribed verbatim. Two researchers then independently made a qualitative analysis of the transcriptions based on the factors in the adapted TIB model. They agreed on the labeling and classification of the themes extracted from the transcripts and summarized them to establish the content of the phase 2 questionnaire. Only topics addressed by at least two participants were retained.

Content validity
To ensure content validity, the questionnaire's preliminary version was individually presented to eight experts from the target population and two experts on questionnaires; minor modifications were made to standardize and clarify item wording. The TIB questionnaire validation process was conducted using a sample of 22 health-care professionals with an almost identical profile to that of the target population (GPs, nurses, and paramedical staff working in Lausanne University Hospital's psychiatric care units, but not in northern Vaud). Participants were asked to take the questionnaire twice, with a two-week interval. The questionnaire's test-retest reliability was assessed statistically using Cohen's kappa. Results indicated good construct-reliability, with kappas averaging 0.75 and ranging from 0.4 to 1, indicating moderate to almost perfect agreement.

Instrument
In its introduction, the questionnaire defined CIM and the behavior being studied. CIM was defined as "A set of diagnostic and therapeutic methods (including natural health products) which are neither readily accepted by the current medical system nor taught in recognized medical schools. They can be classified into three broad categories of treatments: 1. Natural biological products (phytotherapy, essential oils, dietary supplements, homoeopathy, aromatherapy, etc.); 2. Mind-body interventions (acupuncture, reflexology, auriculotherapy, hypnosis, mindfulness, relaxing massages, sophrology, cardiac coherence, Eye Movement Desensitization and Reprocessing (EMDR), art therapies such as music therapy); 3. Other health-care approaches from traditional medicine (e.g., traditional Chinese medicine). The following validated, non-medicinal treatments do not fit into the definition of CIM: hygiene and dietary rules (dietary regime, physical activities and sports, modifications to eating habits, hygiene rules); usual psychological treatments (psychoanalysis, cognitive behavioral therapy, systemic therapy); physical therapeutics (rehabilitation techniques, physiotherapy, occupational therapy)." The studied behavior was described as "The behavior (incorporating CIM into my usual clinical practice with psychiatric patients) refers to the act of administering CIM to patients yourself or referring them to somebody trained in CIM as a complement to the usual psychiatric care provided." The questionnaire's introductory section was then followed by sociodemographic and professional questions.
Sociodemographic and professional characteristics were assessed using eleven questions inquiring about the respondents' age, sex, profession (GP, nurse, etc.), whether they were parents, whether they practiced in inpatient or outpatient care settings, whether they held a management position, whether they worked full-time or part-time, how many years of professional experience they had, whether they had used CIM in their clinical practice in the past, whether they had undergone recognized training in CIM by Lausanne University Hospital, and whether they used CIM in their personal life.
Participants were asked to indicate the extent to which they agreed with the item statements related to the factors of the TIB and the independent variable of intention based on the initial TIB model (see Fig. 1). Respondents then answered questions using either a 5-point Likert scale, a 7-point Likert scale, or a 5-point semantic scale related to the adapted factors of the TIB and the independent variable of intention. The factor of past behavior used 1 item with 4 response options (all factors and intention are detailed in Table 1). All the factor score calculations were the sums of their items.

Administration of the instrument to participants
The survey was distributed to all the 197 health-care professionals working in northern Vaud's psychiatric care units-66 GPs (33%), 82 nurses (40%), and 49 paramedical staff (27%)-of whom 122 began the questionnaire and 105 fully completed it (53%). All the data were coded to ensure confidentiality in REDCap software.
The survey was distributed electronically using REDCap software's electronic data capture tools and was hosted on servers at the University of Applied Sciences and Arts Western Switzerland [49,50]. Three to five reminder emails were sent to participants who had not yet completed the questionnaire. Participants were asked to indicate the extent to which they agreed with the item statements related to the TIB's different factors and the independent variable of intention (see Fig. 1).

The principal component analysis process
A principal component analysis (PCA) was performed to assess how appropriate each of the observed variables was with regards to the latent constructs they were supposed to measure, and to try to reduce the number of items, if possible. The first step assessed the data's  factorability using the Kaiser-Meyer-Olkin measure of sampling adequacy and Bartlett's test of sphericity [51]. The eigenvalues criterion was then used to determine how many factors to extract, and the direct oblimin rotation method was performed because the factors were likely to be correlated [51].
Candidates for potential deletion were selected using the following criteria [52]: an item's factor loading had to be > 0.4; otherwise it was a candidate for deletion; if an item was cross-loaded on several components, its highest factor loading had to be ≥ 0.5 and its difference from the second-highest factor loading had to be ≥ 0.3; otherwise it was a candidate for deletion. Thus, an item whose highest factor loading was 0.55 and whose second-highest loading was 0.3 was a candidate for deletion. Past behavior, which was measured using a single question, was not included in the PCA.
The authors then evaluated items selected as potential candidates for deletion to determine their impacts on the model. If a candidate was deemed essential to the model, it was retained; otherwise, it was deleted. Items were deleted one at a time and then the PCA was run again, following the same steps (KMO, eigenvalues, candidate screening, evaluation, deletion) until no more potential candidates for deletion emerged.
The results indicated that most of the factors taken from the adapted TIB were selected by the PCA. Fig. 2 shows the final TIB model tested following the PCA analyses. The factors of perceived consequences and facilitating conditions were shared between two different factor groups of each original factor, which refined and specified their content: perceived consequences for patients and perceived consequences for staff, and structural facilitating conditions and conditions facilitating CIM. The factors of self-identity and personal normative beliefs did not pass the PCA test. Nevertheless, because the literature often describes somatic unit healthcare professionals' attitudes towards CIM, we decided to retain personal normative beliefs and test the strength of their effects anyway. The final TIB model tested was thus composed of seven factors retained by the PCA plus the extra non-retained factor of personal normative beliefs. Table 1 describes the factors and the items retained by the PCA.

Statistical analysis of the intention to use CIM
To identify the factors that played a significant part in predicting the health-care professionals' intentions to use CIM, respondents were classified into two groups according to the median value of the intention score (median = 12; range = [3][4][5][6][7][8][9][10][11][12][13][14][15]): the group with the intention to use CIM (intention score ≥ 12, coded 1) and the group without that intention (intention score < 12, coded 0). The scores of the items composing each factor were summed to give an overall total factor score. Data (sociodemographic variables and calculated scores by latent construct) were then summarized by group as numbers and percentages for categorical variables and as means and standard deviations (SD) for continuous variables. Associations between each independent variable and the outcomes were assessed using a univariate logistic regression model, and the strengths of these associations were measured using the odds ratio (OR) and the p-value.
Those variables associated with the intention to use CIM that had pvalues < 20% were then used in a logistic regression backward procedure to fit the multivariate model. A fractional polynomial model was used to check the linearity of the relationship between continuous variables and the intention to use CIM. Potential interactions were also tested, model diagnostics were performed to check for residuals, and influential observations and model calibration were tested using the Hosmer-Lemeshow goodness-of-fit test. The power of discrimination between the two groups was calculated using the area under the ROC curve (AUC). To get an accurate estimate of our model's power of discrimination, we performed an internal validation using the bootstrap method described by Harrell et al. [53]. This method enables the calculation of the optimism of our model's predictive discrimination. The  Note. a PCA split the initial factor of facilitating conditions into two separate factors: conditions facilitating CIM and structural facilitating conditions. b PCA split the initial factor of perceived consequences into two separate factors: perceived consequence for patients and perceived consequence for staff. c PCA did not retain the factor of personal normative beliefs, but it was nevertheless tested on intention because of its repeated appearance in the literature. This requires further research. d PCA deleted 2 of the initial 7 items of perceived social norms-1 item from perceived social roles and 1 item from normative beliefs. e PCA deleted 1 of the initial 11 items of facilitating conditions. f PCA deleted 1 of the initial 13 items of perceived consequences.

Respondents' characteristics
The survey was distributed to all the 197 health-care professionals working in northern Vaud's psychiatric care units-66 GPs (33%), 82 nurses (40%), and 49 paramedical staff (27%)-of whom 122 began the questionnaire and 105 fully completed it (53%). Table 2 shows the sociodemographic characteristics of the 105 respondents, whose average age was 40.63 years old (±10.3). More females (79%) than males (21%) responded, but this fitted the sample population (77% female, 23% male). GPs accounted for 27.6% of respondents, paramedical staff represented 47.6%, and psychosocial nursing staff made up 24.8%. Paramedical staff were thus slightly over-represented among respondents as they composed 39.6% of the sample population. This was at the expense of GPs (33.5% of the sample population) and psychosocial nursing staff (26.9%). More than half of the respondents (54.3%) had used CIM in their clinical practice at least once in the past, even though only 24.8% of them had undergone approved training in CIM. Respondents had an average of 13.69 years (±9.88) of experience in clinical practice, and 75.2% used CIM personally (for themselves or close relatives).

Factors regarding the TIB and sociodemographic characteristics by groups' intentions to use CIM
Summary data for all factors, differentiated by the groups' intentions to use CIM or not (Yes or No), are presented in Table 3. Means (SD) for principal factors of the TIB were: perceived social norms, 28
In addition, four sociodemographic variables significantly predicted the intention to use CIM (Table 3) Variables predicting the intention to use CIM, with a p-value < 20%, were then used, via a backwards logistic regression procedure, to fit a multivariate model that would best explain that intention. The following variables were deemed to form the best model (Table 3): perceived social norms (OR = 1.53, p < 0.0001), conditions facilitating CIM (OR = 1.33, p = 0.041), and affect (OR = 1.24, p = 0.032). These three variables significantly predicted (p model < 0.0001) intention. As shown by the  [53], estimated an optimism of 1%. The model's corrected performance was therefore estimated at 93.74% (94.74%-1%). This value constitutes an unbiased estimate of the power of external predictive discrimination. The latter result suggests that priority actions to promote the incorporation of CIM into the clinical practice of psychiatric health-care professionals should focus on these three factors.

Discussion
Based on Triandis' Theory of Interpersonal Behavior, this study aimed to determine how psychosocial and sociodemographic factors influenced health-care professionals' (N = 105) intentions to use CIM in their psychiatric clinical practice with adults. The study indicated that 55.2% of participants intended to use CIM in their clinical practice with adults with mental illness in the next three months. Our multivariate model identified three main factors (perceived social norms, conditions facilitating CIM, and affect) associated with the intention to practice CIM in psychiatry. The model allowed us to discriminate between people who intended to use CIM and those who did not, with an AUC of 94.74%. The adapted factors of the TIB that we used therefore formed an excellent model for explaining health-care professionals' intention to use CIM. This result suggests the need to prioritize actions supporting these three factors in order to encourage psychiatric health-care professionals to use CIM in their clinical practice.
Some studies [38,44] using the TIB model to study other health-care professionals' intentions to adopt a new professional behavior also revealed the significant influence of one or both of the two sub-components of perceived social norms (i.e., normative beliefs and perceived social roles). A study of physicians' intentions to use telemedicine identified the significant effect of perceived social norms [44]; a systematic review identified that the factor of role beliefs was a substantial determinant of intention [38]. We can therefore consider perceived social norms to be a key factor in predicting health-care professionals' intentions to adopt a new professional behavior. This means that professionals may be more willing to use CIM if they identify it as an integral part of their professional role, involving a person-centered approach and taking into account the patients' preferences, values, and beliefs [22,26]. The use of CIM also requires institutional, hierarchical, and peer acceptance. These findings are in line with other studies showing that physicians [34,35,47] and nurses [28,36,54] need clear direction from the greater medical community, their peer group, and their institution. The literature indicates physicians' central role in facilitating and supporting the implementation of CIM among their peers [55][56][57]. According to some authors, the institutional processes required for incorporating CIM into practice are often heterogeneous and disparate, and introducing CIM could be facilitated if physicians promoted such projects to their hospital management boards [29,30].
Concerning the factor of conditions facilitating CIM, the literature is fairly consistent in indicating that health-care professionals' lack of knowledge about CIM limits the incorporation of these approaches into their clinical care and affects the quality of the information and counselling that they can give to patients about CIM [27,28,33,34,36,47,54]. We suggest that this factor, which describes the conditions for the safe use of CIM (having knowledge of CIM, being trained in CIM by one's institution, or having CIM practitioners recognized by the institution), represents a necessary condition for the professional being able to integrate CIM effectively and safely and make appropriate use of it with patients. Training health-care professionals in CIM should be a priority action for institutions wishing to introduce it or, at the very least, professionals should have access to experts in CIM recognized by their institution. This would help them to better inform psychiatric patients, support them with the most appropriate CIM, and identify possible Table 3 Comparison of the group intending to use CIM (Yes) vs. the group not intending to use CIM (No), using univariate and multivariate logistic regression analyses. adverse effects or the risks of interactions with pharmaceutical treatments.
The scientific literature has yet to directly describe affect's importance in the use of CIM in psychiatric clinical practice. In studies using the TIB model, affect was not a significant predictive factor of intention among non-health-care staff [38,44,45]. Affect is the result of previous experiences and requires the use of long-term memory [40]. If positive affect is felt at the thought of adopting a given behavior, that psychological state will result in the body trying to maximize that feeling, both in terms of frequency and intensity, which will reinforce the intention to repeat the behavior [40,44]. Because psychiatric care requires interpersonal interaction involving personal involvement [58,59], we suggest that caregivers are particularly attentive not only to the patients' emotions but also their personal affective state [58]. We suggest that using CIM can constitute a non-conventional approach to promoting therapeutic relationships, with the implication that health-care professionals develop a different type of self-awareness about their feelings. This might help them to improve their capacity for self-regulation and emotional well-being, which will have implications on maintaining fruitful and satisfying therapeutic relationships with patients [60] and which may increase job satisfaction [28]. This factor will require further exploration to understand and better identify the origin of these emotions.
In addition to considering the three main factors, other psychosocial factors have a significant influence on intention. Following the PCA, our study differentiated perceived consequences for patients from perceived consequences for staff, and both had a very significant influence on intention. Our results concerning perceived consequences for patients were consistent with the literature, which describes improvements to mood disorders and anxiety [3,28,36], stress [28,36], sleep [28,36], and a more general contribution to improved health [28,35]. CIM also seems to encourage patient empowerment [28] and increase their involvement in their treatment [35]. Our findings on perceived consequences for staff indicated a great interest in how CIM could enrich clinical practice. We noted that health-care professionals perceived that incorporating CIM had multidimensional positive effects on psychiatric care. The literature also indicates that using CIM encourages robust therapeutic relationships [35] and facilitates patient-nurse communication, patient-centered care, and a more humanistic way of practicing that care [26,28]. This is consistent with the professional psychiatric values of person-centered care [22]. Despite their potential for indirect positive effects on psychiatric patients' conditions, there is little evidence of research in this field in the quantitative literature.
Descriptive norms also very significantly influenced our respondent's intentions to use CIM. This factor also remains relatively unexplored in the literature, and it is not clearly distinguishable from peer approval. The factor of structural facilitating conditions came close to having a significant effect on intention, and this result needs to be confirmed in further studies using larger populations. Several studies have described how such structural barriers as the lack of scientific evidence [36], the lack of reimbursement programs [29,36], and the lack of time [27] represent barriers to the use of CIM in practice. The past behavior of having used CIM in clinical practice in the preceding month was a very strong predictor of the intention not to use it in the future. Having used CIM 1-5 times or more in the preceding month led to a high statistical probability of the intention to use it again in the future. Previous studies had not examined past behavior in this way. The extra factor of personal normative beliefs, which did not pass the PCA, also had a high significant effect on intention. This extra factor demonstrated the importance of participants' opinions and personal values with regards to incorporating CIM into their practice-a result similar to Godin's indication of the importance of moral norms [61]. According to the literature on moral norms [61], participants whose intentions are most consistent with their own moral standards are more likely to adopt a new behavior. These authors suggested that internalized norms and personal expectations are important factors in the development of a person's motivation to engage in a given behavior [61].
The present study also noted the significant effects that the sociodemographic factors of having used CIM in one's personal life, having already incorporated CIM into one's clinical practice, and being trained in CIM approaches recognized by one's institution had on the intention to use CIM. However, we also found that the number of years of clinical practice had a negative statistical influence on the intention to use CIM, with older professionals perhaps having less interest in CIM than younger ones. A systematic review [28] noted seven studies that had reported that a previous positive personal experience of CIM could lead to it being used on or recommend to patients, even by professionals previously skeptical about the value of such therapies. Another study showed that nurses' and midwives' positive opinions of CIM were more closely associated with their personal experiences of it than were physicians' opinions [33]. Personal experience was the main factor explaining the positive attitudes and beliefs towards CIM among nursing students and chiropractic students [62]. Comparisons showed that health-care professionals trained in CIM were more influenced by their positive personal experiences of CIM and literature-based factors than were professionals not trained in CIM [33]. In another study, respondents trained in one or more CIM approaches reported a greater intention of using them than those who had never had any training [35]. Concerning professional seniority, Godin et al. [35] noted a slightly significant weaker intention to use CIM among 60-75-year-olds than among their younger colleagues. However, one systematic review [47] indicated that associations between physicians' sex or age and their attitudes towards CIM varied across studies. Those authors suggested that the lack of consistent correlations could be due to different definitions and categorizations of CIM.

Study limitations
The present study's target-group response rate was moderately satisfactory and comparable to Godin's study dealing with a closely related subject but using another theory of behavior [35]. Our study of 105 psychiatric-sector professionals should not be generalized to all health-care professionals in this field, and the questionnaire should be tested on larger populations. Although most respondents held positive views of CIM, it is quite possible that non-respondents had no interest in the subject or held negative views about it; thus, there is a risk of a positive bias towards CIM. Moreover, the study was conducted in Switzerland, where the general population is particularly interested in the use of complementary treatment approaches, which may also explain the interest of health professionals. Switzerland's obligatory basic health insurance reimburses four CIM approaches (anthroposophic medicine, homoeopathy, phytotherapy, and traditional Chinese medicine) if they are carried out by qualified physicians, and this may also encourage professionals to use them. Nevertheless, hospital billing processes for CIM treatments are complex, and these may limit their use in hospital settings [30].
The reliability of our adapted version of the factors of the TIB was good enough to study health-care professionals' intentions to use CIM in their clinical practice in psychiatry. The factors of past behavior and personal normative beliefs were considered despite them being rejected by the PCA. Testing them showed that both had a high statistical association with health-care professionals' intentions to incorporate CIM into their practice, but further investigation would be needed on a larger population. Finally, the English translation of our questionnaire needs to be validated.

Recommendations for practice and future research
Although the present study's results should be confirmed in a larger population, they can provide some guidance to health-care institutions wishing to support the incorporation of CIM into clinical practice. Health-care professionals should identify with CIM, promote it as part of a person-centered approach, and see this as a part of their professional role. We thus suggest that psychiatric institutions disseminate a clear position in favor of incorporating CIM into clinical care and that these approaches should always take place with the patient's full agreement. Furthermore, institutions should establish policies for training healthcare professionals in CIM or for ensuring access to CIM therapists, recognized by the institution, to whom their patients can confidently be referred. The only sociodemographic factors in our study that significantly predicted the intention to adopt CIM were related to personal or professional experiences with those approaches, which is consistent with the need to develop health-care professionals' skills and knowledge of them. Given the growing interest in CIM among patients worldwide, health-care professionals' initial training should also develop their knowledge in this domain so that they can support and advise their patients on appropriate choices. Our study also identified the importance of considering the emotional processes affecting psychiatric health-care professionals' intentions to use CIM. This factor requires further research.
Finally, we observed that psychiatric health-care professionals perceived multiple positive consequences when incorporating CIM into their clinical practice. These results provide new directions for research into how CIM improves therapeutic relationships or creates greater patient commitment to their own treatment. Although these elements could have indirect positive effects on psychiatric patients' conditions, they have yet to be significantly explored with quantitative studies.
The present study, using an expanded number of factors of the TIB, was well able to identify the psychosocial factors that best predicted psychiatric health-care professionals' intentions to incorporate CIM into their practice. Its innovative questionnaire, completed by respondents working in an environment rather favorable to CIM, requires a confirmatory study and external validation. Behavioral theories should be used more widely in research to better understand the multiple factors influencing professional behaviors.