Predictors of the intention to use integrative medicine in psychiatric hospitals

Abstract

Background

The present study aimed to examine the influence of psychosocial and demographic factors on health professionals’ intention to integrate complementary and integrative medicine (CIM) into hospital-based psychiatric clinical practice. It also sought to validate the theoretical model derived from our previous exploratory study.

Method

A cross-sectional survey study using an online questionnaire sent to 4111 potential participants based on an adapted version of Triandis’ Theory of Interpersonal Behaviour (TIB).

Results

Participants reported high levels of positive attitudes towards CIM, with 61.6 % of the 1561 respondents reporting previous use in clinical practice and 37.8 % having received formal CIM training. Analysis of professionals’ intention to use CIM revealed four influential psychosocial factors - affect, perceived social norms, descriptive norms and past behaviour (p < 0.0001) - and three significant demographic factors - older age, a managerial position and fewer years of clinical practice (p < 0.05). With an area under the ROC curve of 95.53 %, the resulting model showed high discriminatory power and excellent fit.

Conclusions

These findings highlight the need for a supportive institutional environment to promote CIM. This requires the support of a wide range of professional leaders and the promotion of a shared CIM culture among healthcare professionals and interdisciplinary discussions.

1. Introduction

Reducing the prevalence of mental disorders is a public health priority.1, 2, 3 Conventional pharmacological treatments have their limitations in mental disorders.⁴ There is a growing trend towards integrative approaches that combine different modes of treatment, including complementary medicine.4, 5, 6 These approaches are usually classified into five categories: Mind–body interventions, such as relaxation techniques or hypnosis; Biologically based treatments, including herbal remedies, aromatherapy and homoeopathy; Manipulative and body-based methods, such as massage and osteopathic manipulation; Energy-based therapies, such as reiki and magnetic field therapy; and Whole medical systems, including traditional Chinese medicine and ayurveda.^7,8

However, the use of such practices remains unevenly distributed and varies across healthcare settings.10, 11, 12 We acknowledge that many healthcare professionals find the terminology unclear and that complementary medicine and integrative medicine are still frequently used interchangeably. For clarity and consistency, we use the term complementary and integrative medicine (CIM) throughout this work. It is also important to specify that the practices examined here are intended to complement, not replace, conventional care.

This manuscript uses the term ‘integrative medicine’ as defined internationally: a coordinated approach that aims to combine the best evidence from both conventional and complementary medicine.⁹

One recent study found that 62 % of 489 health units in Sweden used CIM to manage conditions such as anxiety, sleep disorders and depression.¹¹ The main motivations for integrating CIM into conventional treatments were symptom relief, meeting patient needs and reducing conventional medication use, and there were minimal reports of adverse effects. The most common reason for discontinuing the use of CIM in psychiatric hospital settings was a lack of trained practitioners.¹¹ Patient satisfaction with these approaches has been well-documented.^6,13 The quality of research on the effectiveness of CIM in psychiatry is increasing, as evidenced by advances in studies on acupuncture,¹⁴ brief mindfulness-based interventions,^15,16 yoga¹⁷ and aromatherapy.¹⁸

Given the importance of CIM in the management of mental health, there is an urgent need to understand the factors that influence their use in clinical practice by healthcare professionals in psychiatric hospitals.^10,12,19 A comprehensive understanding of these factors, as highlighted by Metri et al. (2022),⁶ Olsson et al. (2022)¹⁰ and Wemrell et al. (2020),¹¹ is crucial if healthcare leaders are to facilitate the adoption of innovative, evidence-based CIM in clinical practice. A qualitative study revealed scepticism among decision-makers, who prioritised evidence-based practice and national guidelines, as opposed to practitioners, who emphasised safety and CIM’s patient-centred approach.¹⁰ This scepticism may arise from a lack of familiarity with the latest evidence on CIM.²⁰

Several psychosocial factors influence the adoption of new professional behaviours.21, 22, 23 Using an adapted version of Triandis’ theory of interpersonal behaviour (TIB),²⁴ our previous exploratory study (N = 105) aimed to understand how those factors influenced health professionals’ intentions to integrate CIM into psychiatric clinical practice with adult patients (Fig. 1). It successfully discriminated between professionals who intended to use CIM and those who did not, with an accuracy rate of 94.74 %.¹²

Fig. 1.

An adapted version of Triandis’ Theory of Interpersonal Behaviour (TIB).

Three primary factors were identified as influencing their intention to use CIM: perceived social norms emphasising institutional support; facilitating conditions, such as training policies and access to therapist; and the impact of affect on professionals’ comfort with using CIM. Factors such as consequences for staff and patients, descriptive norms, past behaviour and personal normative beliefs also played significant roles. Sociodemographic factors like personal experiences with CIM, previous use and institutional recognition also influenced intention, with greater interest among younger professionals. Despite shedding light on these influencing factors, that study had some limitations, including potential positive bias and a small sample size. This required further validation with a larger and more diverse sample.¹²

The present study focused on investigating psychosocial and demographic influences by validating a specific questionnaire, with the aim of helping psychiatric hospitals understand the factors influencing the integration of CIM, assessing the feasibility of implementing it and outlining requirements for future projects.

2. Method

2.1. Design

A cross-sectional survey study using an online questionnaire.

2.2. Specific aims

1. To determine the psychosocial and demographic factors influencing professionals’ intentions to integrate CIM into their clinical practice. 2. To use confirmatory factor analysis to validate the theoretical model derived from our preceding exploratory study.

2.3. Population

The study surveyed health professionals working in psychiatric units across French-speaking Switzerland. These included four general hospitals with psychiatric departments and three specialised public psychiatric clinics. According to Switzerland’s official hospital categorisation system,²⁵ all of these facilities are classified as psychiatric care sector institutions, regardless of their administrative classification. The system distinguishes general hospitals from specialised clinics based on the number of specialist wards and the ratio of general to specialist inpatient days in care. It also assigns institutions a level of service (level 1 or 2) according to their training capacity, infrastructure and case volume. This classification system is specific to Switzerland and may not correspond directly to international healthcare terminology. Despite organisational differences, healthcare professionals across these institutions undergo standardised training that ensures consistent national qualifications and care and fosters collaboration.

2.4. Participants and data collection

This present study reached out to 4111 health professionals (physicians, nurses, healthcare assistants and paramedical staff, including physiotherapists, occupational therapists, rehabilitation therapists, social workers, socio-educational professionals, and psychologists).

The study was conducted from March to November 2022 using a self-administered questionnaire accessed via a link sent by email. Mailing lists were provided by the directors or administrators of the participating mental health hospitals and clinics. The questionnaire was administered using the REDCap software electronic data collection tool and hosted on the servers of the University of Applied Sciences and Arts Western Switzerland.^26,27 Three to five reminders to complete the questionnaire were emailed to the professionals who had not yet answered. The questionnaire involved participants rating the extent to which they agreed with a series of statements about different factors related to the TIB and the independent variable of intention. Participation was purely voluntary, and there were no consequences for refusing to participate.

2.5. Ethics

An initial email was sent to potential survey participants explaining the aims and nature of the study before the link to the online questionnaire was sent in a second email. Respondents provided their implicit, voluntary consent to participate in the study by completing the questionnaire, as had been explicitly stated in the initial email. All data were coded to ensure anonymity and confidentiality. The study was approved by the Human Research Ethics Committee of the Canton of Vaud (No. AO_2021–00,046) in accordance with all the applicable rules for research involving human subjects.

2.6. Instrument

The questionnaire’s introduction defined CIM and the behaviour being investigated. CIM was characterised as a variety of diagnostic and therapeutic practices outside mainstream medical education, including the use of mind–body techniques, natural products, body manipulation practices, physical exercise and traditional medicine.^28,29 Certain validated, non-pharmacological interventions do not fall under the definition of CIM, such as dietary guidelines (diets, changes in eating behaviour, food hygiene practices), physical activity recommendations, psychological therapies (therapies inspired by analysis and psychoanalysis, cognitive behavioural therapies, systemic therapies) and physical therapies (rehabilitation techniques, physiotherapy, occupational therapy).

The specific behaviour considered was using CIM in the routine clinical care of psychiatric patients, which could involve direct administration to the patients or referral to practitioners trained in CIM as a complementary approach to standard psychiatric care.

After the introductory definitions, participants provided their sociodemographic and professional details. Using classic 5-point and 7-point Likert scales and a 5-point semantic scale, participants responded to items on 7 factors related to the TIB: perceived social norms, 5 items (Cronbach’s alpha (α) = 0.89); conditions facilitating CIM, 3 items (α = 0.89); perceived consequences for patients, 8 items (α = 0.89); perceived consequences for staff, 4 items (α = 0.89); descriptive norms, 3 items (alpha = 0.79); structural facilitating conditions, 7 items (α = 0.79); affect, 5 items (α = 0.92); and personal normative beliefs (extra factor), 2 items. These alpha coefficients were calculated based on our preceding exploratory study.

The variable of intention to use CIM was measured using 3 items (α = 0.79), and past behaviour was assessed using a single item with four response options. Detailed information on these measures can be found in our preceding article on validating the scale¹² (Table 1).

Table 1.

Sociodemographic characteristics of the 1561 respondents.

Variable	All, n = 1561 (100 %)		Physicians, n = 323 (20.7 %)		Nurses, n = 796 (51.0 %)		Healthcare assistants, n = 53 (3.4 %)		Paramedical staff, n = 61 (3.9 %)		Psychologists, n = 215 (13.8 %)		Others, n = 113 (7.2 %)
	n	%	n	%	n	%	n	%	n	%	n	%	n	%
Sex
-Women	1082	69.3	183	56.7	541	68.0	41	77.4	52	85.3	185	86.1	80	70.8
-Men	479	30.7	140	43.3	255	32.0	12	22.6	9	14.7	30	13.9	33	29.2
Age (mean (s.d.))	41.3	(10.7)	40.6	(9.7)	41.7	(10.9)	36.8	(11.5)	39.5	(10.8)	40.0	(10.1)	46.4	(10.6)
Years of clinical practice (mean (s.d.))	14.1	(10.3)	11.5	(9.6)	15.9	(10.5)	11.4	(9.3)	13.5	(9.9)	11.9	(9.5)	14.7	(9.9)
Works in
-Outpatient care	675	43.3	171	52.9	275	34.5	6	11.3	17	27.9	166	77.2	40	35.4
-Inpatient care	639	40.9	81	25.1	425	53.4	46	86.8	32	52.4	12	5.6	43	38.1
-Both	247	15.8	71	22.0	96	12.1	1	1.9	12	19.7	37	17.2	30	26.5
Managerial position
-Yes	425	27.2	171	52.9	171	21.5	1	1.9	1	1.6	65	30.2	16	14.2
-No	1136	72.8	152	47.1	625	78.5	52	98.1	60	98.4	150	69.8	97	85.8
Undergone recognised CIM training
-Yes	386	24.7	46	14.2	239	30.0	9	17.0	18	29.5	47	21.9	27	23.9
-No	1175	75.3	277	85.8	557	70.0	44	83.0	43	70.5	168	78.1	86	76.1
Used CIM in clinical practice
-Yes	961	61.6	191	59.1	536	67.3	32	60.4	34	55.7	116	53.9	52	46.0
-No	600	38.4	132	40.9	260	32.7	21	39.6	27	44.3	99	46.1	61	54.0
Used CIM personally
-Yes	1143	73.2	184	57.0	603	75.8	39	73.6	55	90.2	170	79.1	92	81.4
-No	418	26.8	139	43.0	193	24.2	14	26.4	6	9.8	45	20.9	21	18.6

2.7. Statistical method

All statistical analyses were performed using Stata 17 software (StataCorp. 2021).

A confirmatory factor analysis (CFA) was conducted using the structural equation model (SEM) package in Stata to assess the measurement and structural invariance of our 7-factor model, as per the recommendations by Brown.³⁰ Maximum likelihood estimation with missing values (MLMV) was used to account for missing data.

To assess the 7-factor model’s equivalence for men and women (configural invariance), a same-form model was fitted as the baseline model. This model assumes the same form and loading of the indicators on each group’s latent variables, without imposing equality constraints on those latent variables’ loadings, error variances, variances or covariances.

Subsequent tests of measurement invariance involved fitting more restrictive models: Model-2 used equal loadings for men and women; Model-3 used equal loadings and equal error variances; Model-4 used equal loadings, equal error variances and equal covariances; and Model-5 used equal loadings and equal intercepts.

Goodness of fit and structural invariance were assessed using a comparative fit index (CFI) with a recommended cut-off of approximately 0.95 and a root mean squared error of approximation (RMSEA) using an ideal threshold of 0.05 or less for a good fit. The models were also compared. More restrictive models were considered invariant from less restrictive models if their CFI and RMSEA differences were < 0.01 and < 0.015, respectively.^31,32 Confirmation of the measurement and structural invariance between groups enabled the comparison of means between men and women.

To identify factors associated with healthcare professionals’ intentions to use CIM, an intention score was calculated by summing the item scores within each factor (range [3–15]). Respondents were then divided into two groups based on the median intention score of 12, thus distinguishing those with a greater intention (GI) to use CIM (intention score ≥ 12, coded as 1) from those with a lesser intention (LI) (intention score < 12, coded as 0). In the absence of any validated cut-off available in the literature, the median was chosen as a pragmatic threshold, allowing for a balanced distribution of participants and facilitating comparison between groups with higher and lower intentions to use CIM.

Univariate logistic regression models were used to examine associations, quantified using odds ratios (ORs) and p-values, between each independent variable and the intention outcome. Variables with p < 0.2 were used in a backwards selection procedure to fit a multivariate logistic regression model.

A fractional polynomial model was used to assess the linearity of the relationships between continuous variables and the intention to use CIM. Potential interactions were explored, and model diagnostics were performed to evaluate residuals, identify influential observations, and assess model calibration using the Hosmer–Lemeshow goodness-of-fit test.

The model’s power to discriminate between the two groups was determined by the area under the ROC curve (AUC).

All of our statistical analyses were performed by Dr Mohamed Faouzi, a biostatistics expert who also supervised data management.

3. Results

3.1. Respondents’ characteristics

The survey was distributed to 4111 health professionals in general and psychiatric hospitals and state-run psychiatric clinics, resulting in 1561 respondents (a 38 % response rate) (Table 1).

Table 1 shows the respondents’ sociodemographic characteristics. Mean age was 41.3 years old (± 10.7), and the majority were women (69.3 %). The responding healthcare professionals included 796 (51 %) nurses, 323 (20.7 %) physicians, 215 (13.8 %) psychologists, 61 (3.9 %) paramedical staff, including physiotherapists and occupational therapists, and 53 (3.4 %) healthcare assistants. There were 113 (7.2 %) other participants, including social workers, socio-educational professionals and specialists in specific therapies such as art therapy. On average, respondents had 14.1 years (± 10.3) of clinical experience.

The response rates by profession are as follows: 66 % occupational therapists, 58 % social workers:, 57 % psychomotor therapist, 49 % psychologists, 43 % nurses, 39 % physiotherapists, 36 % physicians, 32 % other participants, 32 % healthcare assistants.

More than half (61.6 %) had used CIM in their clinical practice, and 24.7 % had received accredited CIM training, with 73.2 % also using CIM for personal or family care.

3.2. Confirmatory factor analysis of the theoretical model

We assessed the internal consistency of our 7-factor model, which showed good reliability (Cronbach’s alpha ranged from 0.76 to 0.95). The initial 7-factor model (with no correlation between errors) showed a poor fit to the data. We refined the model by iteratively introducing correlated errors between specific items, resulting in our baseline model (same-form model), which showed improved fit statistics (lower χ2/dl, lower RMSEA and higher CFI).

The baseline model (Model-1) was then used to assess configural and measurement invariance between the sexes. Model-2 (Table 2), using equal loadings, did not differ from Model-1, which was not using equal loadings. The changes in CFI (0.002 < 0.01) and RMSEA (0.00 < 0.015) did not exceed the threshold for rejecting invariance between the sexes.^31,32 This result supports the conclusion that the 7-factor model is invariant between the sexes.

Table 2.

Comparison between models.

Indices	Model 1a	Model 2b	Model 3c	Model 4d	Model 5e
RMSEAf	0.050	0.050	0.050	0.051	0.051
CFIg	0.942	0.940	0.936	0.934	0.936
Comparison		2 vs 1	3 vs 2	4 vs 5	5 vs 2
RMSEA-Diff	Not applicable	0.00	0.00	0.00	0.001
CFI-Diff		0.002	0.004	0.002	0.004

^aSame-form model.

^bEqual loadings model.

^cEqual loadings and errors.

^dEqual loadings, errors, variances and covariances.

^eEqual loadings and intercepts.

^fRoot Mean Square Error of Approximation.

^gComparative Fit Index.

3.3. Factors related to the TIB and demographic factors associated with the intention to use CIM

A summary of the factors associated with the intention to use CIM is presented in Table 3.

Table 3.

Comparison between the greater intention (GI) and lesser intention (LI) to use CIM groups using univariate and multivariate logistic regression analyses.

	Descriptive statistics				Univariate		Multivariate
	Greater Intention (GI) to use CIM		Lesser Intention (LI) to use CIM		Greater Intention (GI) to use CIM		Greater Intention (GI) to use CIM
	%	(n)	%	(n)
Total number of participants (n = 1561)	53.4	(834)	46.6	(727)	–		–
Factors related to the TIB	mean	(s.d.)	mean	(s.d.)	OR	p-value	OR	p-value
Perceived social norms (5 items)	30.3	(4.3)	22.3	(4.9)	1.42	< 0.0001*	1.21	< 0.0001*
Conditions facilitating CIM (3 items)	18.6	(2.7)	15.8	(4.1)	1.27	< 0.0001*	–
Affect (5 items)	31.7	(3.1)	25.7	(6.4)	1.33	< 0.0001*	1.30	< 0.0001*
Perceived consequences for patients (8 items)1	47.0	(5.8)	40.2	(7.8)	1.16	< 0.0001*	–
Perceived consequences for staff (3 items)	21.9	(3.6)	19.1	(4.0)	1.22	< 0.0001*	–
Descriptive norms (3 items)	17.2	(3.4)	12.4	(4.2)	1.37	< 0.0001*	1.19	< 0.0001*
Structural facilitating conditions (7 items)	39.4	(6.4)	36.2	(7.3)	1.07	<0.0001*	–
Past behaviour (preceding month)	%	(n)	%	(n)
-Never (ref)	15.7	(60)	77.1	(280)	–	–	–
-Once	11.0	(42)	6.1	(22)	8.91	< 0.0001*	7.54	< 0.0001*
-2–5 times	37.7	(144)	13.2	(48)	14	< 0.0001*	9.26	< 0.0001*
-More than 5 times	35.6	(136)	3.6	(13)	48.82	< 0.0001*	21.14	< 0.0001*
Extra factor	mean	(sd)	mean	(sd)	OR	p-value	OR	p-value
Personal normative beliefs (2 items)	13.2	(1.3)	10.7	(3.0)	1.81	< 0.0001*	–
Sociodemographic factors	%	n	%	n	OR	p-value	OR	p-value
Sex (woman)	73.3	(611)	64.8	(471)	1.49	< 0.0001*	–
Age (mean (s.d.))	41.9	(10.8)	40.7	(10.6)	1.01	0.032*	1.06	0.012*
Outpatient or inpatient practice
-Outpatient (ref.)	43.1	(359)	43.5	(316)	–	–	–
-Inpatient	38.2	(319)	44.0	(320)	0.88	0.237	–
-Outpatient and inpatient	18.7	(156)	12.5	(91)	1.51	0.007*	–
Has a managerial position (Yes)	29.4	(245)	24.8	(180)	1.26	0.041*	2.41	0.005*
Years of clinical practice (mean (s.d.))	14.8	(10.5)	13.3	(10.0)	1.01	0.006*	0.94	0.017*
Job
-Physicians	19.8	(165)	21.7	(158)	0.82	0.123	–
-Nurses (ref.)	53.6	(447)	48.0	(349)	–	–	–
-Healthcare assistants	2.8	(23)	4.1	(30)	0.60	0.073	–
-Paramedical staff	3.9	(33)	3.9	(28)	0.92	0.755	–
-Psychologists	14.3	(119)	13.2	(96)	0.97	0.833	–
-Other	5.6	(47)	9.1	(66)	0.56	0.004*	–
Has used CIM in their psychiatric clinical practice (Yes)	83.5	(696)	36.5	(265)	8.79	< 0.0001*	–
Has undergone recognised training in CIM (Yes)	37.8	(315)	9.8	(71)	5.61	< 0.0001*	–
Used CIM in personal life (Yes)	83.6	(697)	61.4	(446)	3.21	< 0.0001*	–

Univariate analyses revealed several significant psychosocial predictors of intention to use CIM. All the psychosocial factors—perceived social norms, conditions facilitating CIM, affect, perceived consequences for patients and staff, descriptive norms, structural facilitating conditions, past behaviour and personal normative beliefs—were highly significant (p < 0.0001). Seven sociodemographic variables, including age, sex, outpatient and inpatient practice, managerial position, clinical experience, professional and personal use of CIM, and formal CIM training, also significantly predicted intention to use CIM (p < 0.05).

Multivariate analysis identified four key factors related to the TIB: perceived social norms, affect, descriptive norms and frequency of past behaviour (p < 0.0001). In addition, three sociodemographic variables—age, managerial position and years of clinical practice—contributed significantly to the model (p < 0.05). The final model showed strong discriminatory power and an excellent fit, with an area under the ROC curve of 95.53 %.

4. Discussion

In the present study, 1561 health professionals from psychiatric hospitals and state-run clinics shared their perceptions of how CIM is practised in mental healthcare in French-speaking regions of Switzerland. Their attitudes towards CIM were largely positive, with 61.6 % reporting having used it previously in their clinical practice and 37.8 % having received formal training in CIM. With possible intention to use CIM scores ranging from 0 to 15, the median score was 12, indicating a strong intention to use. Our validation of the theoretical model of the psychosocial determinants influencing intention to use CIM was successful. With a ROC curve of 95.53 %, our model demonstrated robust predictive ability, particularly in discriminating between GI and LI groups in psychiatric clinical practice.

Both this study and the preceding pilot study¹² identified two significant factors influencing the intention to integrate CIM into practice. The first factor involved perceived social norms and incorporated two sub-factors. The first, normative norms, reflects mental healthcare professionals’ perceptions of the attitudes of colleagues, their healthcare team, the institution and direct supervisors to integrating CIM in mental healthcare. Positive support from these groups tended to increase the intention to use CIM, while perceived unsupportiveness may have hindered it.33, 34, 35 The second sub-factor was professional role, which represented how the integration of CIM aligned with the responsibilities and functions of psychiatric patient care. This aspect strongly predicted intentions to adopt new healthcare behaviours.²³ If CIM is compatible with one’s professional role, it increases one’s intention to use it; conversely, perceived incompatibility may decrease that intention, may be due to concerns about patient safety, lack of training in complementary medicine or ethical considerations.

The present study confirmed affect’s significant influence on professional decision making, acting as the second key factor influencing intention to use CIM.¹² Both positive and negative affect play central roles in shaping professionals’ propensity to integrate CIM into clinical practice. Positive affect, such as feelings of confidence, comfort and satisfaction derived from integrating CIM, can significantly enhance usage intentions by modulating cognitive control and supporting motivation to continue using CIM.^36,37 Conversely, negative affect, including worry, discomfort, dissatisfaction, discouragement and irritation, can create barriers.³⁸ These emotions may indicate professionals’ confidence or concerns about the efficacy, safety and impact of CIM on patients. They may also indicate competence and familiarity with CIM, or even job dissatisfaction, especially if perceived as an additional workload. Understanding the institutional and organisational context is essential to deciphering variations in positive and negative affect.¹² In mental healthcare, where professionals need to manage their own emotions in order to respond effectively to those of patients, using CIM may provide a unique opportunity to foster therapeutic relationships and increase self-awareness.^39,40 Future research could explore CIM’s potential to promote self-regulation and emotional well-being in both patients and carers, potentially leading to more rewarding therapeutic interactions and increased job satisfaction.^12,41,42

Two other factors related to the TIB complement the main model: descriptive norms (examining the frequency with which CIM is integrated into practice by colleagues and institutions, which influences health professionals’ decisions) and past behaviour (a measure of use of CIM in the last month). The factor of the conditions facilitating CIM, however, was not significant in our multivariate analyses, contrary to the exploratory study.¹²

Descriptive norms’ and past behaviour’s influences on the intention to use CIM have some important and notable implications. Descriptive norms provide insight into the influence of social and organisational dynamics. This suggests that healthcare professionals are strongly influenced by the prevailing practices within their professional and institutional environment. Frequent use of CIM by colleagues and within the institution serve as strong social signals that positively influence an individual’s intention to conform.^43,44 This highlights the importance of fostering a supportive institutional culture, one openly favourable towards CIM. We suggest that messages concerning descriptive norms that convey patterns of CIM use in healthcare practice should be communicated to healthcare professionals more frequently.⁴⁴ Indeed, interventions using descriptive norm messages aim to encourage individuals to adopt or maintain positive behaviours. They can do this by providing information highlighting unusual undesirable behaviours or by emphasising the prevalence of desirable behaviours within a relevant social group.⁴⁴ Regarding the factor of past behaviour, we found that previous use of CIM in clinical practice strongly predicted intention to use it in the future. This is consistent with the literature, particularly for novel behaviours. Past behaviour is a valuable source of information that reinforces the intention to repeat that behaviour.⁴⁵ An in-depth examination of the demographic characteristics of these respondents would provide new insights and a better understanding of this factor, as recommended by the literature.⁴⁵

Our results showed that some demographic factors also significantly influenced the intention to use CIM, particularly being older and having fewer years of clinical experience. However, conflicting studies provide a nuanced perspective on these conclusions. Some researchers have suggested that older professionals may be more inclined to integrate novel methods into their practice due to their broader experience and the greater flexibility in their approaches. In a study comparing younger general practitioners (GPs) with their more experienced peers, some authors found that younger GPs were more critical of CIM.⁴⁶ They frequently expressed doubts about the evidence base for CIM and its specific efficacy beyond the placebo effect. In contrast, another study showed that despite having limited knowledge of CIM, university students from both pharmacy and non-pharmacy backgrounds demonstrated positive attitudes towards it.⁴⁷

Our study also showed that holding a managerial position significantly influenced the intention to use CIM. This finding was consistent with the key psychosocial factor of social determinants. Individuals in managerial positions may be more inclined to explore new approaches to improving care, as they are responsible for developing healthcare practices and, perhaps, reducing costs.48, 49, 50

These differences highlight the complexity of the factors that influence professional attitudes towards CIM. They suggest that the influence of age, experience and managerial position may vary depending on the medical speciality and professional context. Discrepancies related to age and experience may reflect generational differences in training and perceptions of complementary practices, all shaped by systemic and cultural factors within healthcare systems. For example, one recent study showed that GPs in Germany perceived CIM as a more integral part of primary care than their counterparts in countries where a stronger adherence to evidence-based medicine was noted, such as The Netherlands, Norway and the UK.⁵¹ These differences may also be linked to varying patient expectations and reimbursement systems across countries. These findings suggest that variations in the perception and use of CIM may arise not only from generational training differences but also from the interaction between age and experience and broader professional norms. Accordingly, age-specific analyses embedded within broader sociocultural and institutional contexts would be a valuable focus for future research.

Finally, this study only aimed to provide a general overview of how psychiatric care settings influenced the use of CIM. Differences between hospital and clinic settings were not analysed in detail, although they may play a role in the use of CIM. Further studies are warranted to examine the impacts of these settings more specifically.

4.1. Limitations

The present study had some limitations that may restrict its generalisability. Only 38 % of those invited to complete the questionnaire responded, which could have led to a participation bias affecting the sample’s representativeness. The 61.6 % of respondents who reported using CIM in the previous month may also represent a biased sample. Individuals already interested in CIM may have been more inclined to participate in our survey, potentially biasing the results towards more favourable attitudes and greater intentions to use CIM. In addition, this study was not designed to examine differences between specific professional groups, such as prescribers and non-prescribers. Future research focusing on physician-centred perspectives could provide more nuanced insights, particularly in relation to referral patterns, decision-making authority and the clinical use of CIM.

4.2. Conclusion

The present study provides a basis for future research along the lines suggested by Metri et al.⁶ It provided new insights into the psychosocial and demographic factors that predict the intention to use CIM in psychiatric hospitals. Future studies should seek to identify relevant organisational barriers in these treatment settings.

In order to successfully integrate CIM into the practice of psychiatric hospitals and clinics, it is crucial to foster a supportive institutional culture that embraces the CIM approach. This can be achieved by highlighting existing CIM practices within healthcare teams, establishing links between professional roles and CIM, and emphasising the positive impact of CIM on both patients and healthcare professionals.

Gaining the support of leaders in different professions is also essential to uniting an institution’s staff and cultivating a positive institutional culture for using evidence-based CIM.

To improve these strategies, it is essential to gain a deeper insight into how the organisational contexts within institutions can continue to nurture the use of CIM or open themselves to it. In addition, it would be worthwhile to understand which sociodemographic factors most influenced respondents’ recent use of CIM in clinical practice. Exploring these issues would provide valuable insights into how to promote the effective integration of evidence-based CIM practices.

Author contributions

Corinne Schaub and Mohamed Faouzi shared the position of first authors.

Corinne Schaub: Conceptualization; Project administration; Methodology; Investigation; Validation; Funding acquisiton; Writing the original draft.

Mohamed Faouzi: Methodology; Formal analysis; Validation; Writing the original draft.

Julien Vonlanthen: Investigation, Data Curation, Participation in formal analysis, Visualization, Writing the original draft.

Michaël Cordey: Validation; Writing - review & editing.

Pauline Marchand: Validation; Writing - review & editing.

Alexia Stantzos: Validation; Resources; Funding acquisiton; Writing - Review & Editing.

Chantal Berna: Validation; Resources; Funding acquisiton; Writing - Review & Editing.

Ketia Alexandre: Conceptualization; Methodology; Validation; Funding acquisiton; Writing the original draft.

Funding

This work was supported by the Swiss National Science Foundation (SNSF) [grant number 10001C_201107].

Ethics statement

An initial email was sent to potential survey participants explaining the aims and nature of the study before the link to the online questionnaire was sent in a second email. Respondents provided their implicit, voluntary consent to participate in the study by completing the questionnaire, as had been explicitly stated in the initial email. All data were coded to ensure anonymity and confidentiality. The study was approved by the Human Research Ethics Committee of the Canton of Vaud (No. AO_2021–00,046) in accordance with all the applicable rules for research involving human subjects.

Data availability

The data will be made available upon request

Declaration of competing interest

None.