An Instrument to Measure Mental Health Professionals’ Beliefs and Attitudes towards Service Users’ Rights

We aimed at developing and validating a scale on the beliefs and attitudes of mental health professionals towards services users’ rights in order to provide a valid evaluation instrument for training activities with heterogeneous mental health professional groups. Items were extracted from a review of previous instruments, as well as from several focus groups which have been conducted with different mental health stakeholders, including mental health service users. The preliminary scale consisted of 44 items and was administered to 480 mental health professionals. After eliminating non-discriminant and low weighting items, a final scale of 25 items was obtained. Exploratory and confirmatory factor analyses produced a four-factor solution consisting of the following four dimensions; system criticism/justifying beliefs, freedom/coercion, empowerment/paternalism, and tolerance/discrimination. The scale shows high concordance with our theoretical model as well as adequate parameters of explained variance, model fit, and internal reliability. Additional work is required to assess the cultural equivalence and psychometrics of this tool in other settings and populations, including health students.


Introduction
The mental health sector has undergone two fundamental transformations in the last half century, namely Deinstitutionalization and Recovery [1].Both processes have involved an increase in service users' autonomy and freedom of choice.The Recovery movement has mostly been driven by service users themselves which has entailed a significant increase in participation, and consequently a reduction in paternalistic behaviours carried out by professionals.These processes have also improved the professionals' awareness of service users rights, have led to a reduction of coercive measures and a shift from symptom reduction to rehabilitative and recovery approaches [2].
Despite all these improvements, many service users still report stigmatizing attitudes, including professional paternalism and emotional estrangement [3,4].Therefore, receiving a mental health diagnosis is still considered as a predisposing factor that can lead to the experience of stigma from both the social environment [5], as well as mental health professionals [6].
Stereotypes depicting mental health service users as incompetent, weak, incurable, and violent, lead to social discrimination and coercive professional practices [7].Some examples are involuntary inpatient and outpatient treatments, forced medication, overmedication, electroconvulsive therapy under duress, mechanical restraints, seclusion, isolation and arbitrary legal incapacitations and guardianships.The underlying beliefs that influence the decisions that professionals make appear to be formed by a lack of awareness of one's own prejudices [4].In this sense, the perception by some professionals that the moral side of their decisions are not relevant to the recovery process, may contribute to the acceptance of coercion as a standard practice [8].
Given the extent of the consequences of stigma and coercion towards mental health service users, it is essential to raise awareness among mental health professionals in order to foster nonstigmatizing and empowering attitudes through frameworks such as Recovery [9] and Citizenship [10].In the context of planning and implementation of these training and awareness activities, there is a need to evaluate the impact this has on the beliefs and attitudes of professionals through standardised measures.

Previous measures
The first scale that included beliefs and attitudes towards mental health service users was developed by Gilbert and Levinson [11].This scale was used as a method to understand mental health practices on a continuum from custodial to humanistic [12].Another scale dealing with professionals opinions about mental illness was validated shortly afterwards [13].Alongside these early developments, Ervin Goffman resignified the word stigma to refer to a non-physical, invisible signal, making a person's social status undesirable [14].Relatedly, different mental health stigma measures have been developed and are usually applied to the general public [15][16][17][18][19][20][21][22], mental health service users [23], and some others to professional audiences [24][25][26][27][28][29].Another issue that has also generated assessment tools is the stigma that the medical profession in general have towards psychiatry in particular [30].
Over the last decades, in parallel to the rise of the Rehabilitation and Recovery movements, professionals have become more conscious of the need to offer a non-discriminant care based on users' rights.In this context, measures on recovery-based knowledge [31][32][33], attitudes [34], expectations [35] and practices [36][37][38][39][40][41][42] have been developed (see table 1).These instruments have enabled the evaluation of dozens of projects which have implemented the philosophy of recovery in health institutions.However, there are also limitations of these instruments, including the impossibility of administering the attitudes and knowledge measures at baseline with lay professionals, given that they assume a certain degree of knowledge of the principles of recovery.Furthermore, all current professional stigma instruments are designed for certain professional groups [24][25][26]29] or mental health conditions [27,28].Finally, scales measuring the recovery practice, assume that some level of implementation of these practices is being implemented.
So far, there is a lack of multidimensional measures that can be used to assess training and awareness activities attended by different mental health professional groups with heterogeneous levels of knowledge and awareness on the importance of respecting service users' rights.For this reason, the objective of this work is the development of a flexible instrument, in order to measure the beliefs and attitudes related to service users' rights among all types of mental health professionals.

Development
An initial set of 44 items were developed and reviewed by experts on stigma awareness and community mental health.Half of these items were derived from a systematic review of previous measures (see table 1).We reviewed scales on stigmatizing attitudes and beliefs as well as recoverybased knowledge and practices.We found items related to three cognitive levels: attitudes, awareness and knowledge, and four thematic domains: empowerment/paternalism, recovery, stigma and rights.
Additionally, we conducted 11 focus groups with mental health professionals.The first six groups were used for the adaptation to the current context of items based on the literature and the creation of a pool of 22 completely new items.In the last five, the scale was presented at the beginning of the session, leaving time for participants to respond.During the discussion, professionals could comment on the content and contextualize how they had answered the items.
Regarding the anchor points of the scale, we chose a four-point Likert scale (I fully disagree, I disagree, I agree, I fully agree).This mode helps to minimise middle response bias, which is frequent in attitude research [43].

Sample and procedure
The sample used for the psychometric validation of the Beliefs and Attitudes towards Mental Health service users' rights Scale (BAMHS) was comprised of a total of 480 Spanish-speaking mental health professionals.Among these, there were psychologists (29%), nurses (15%), social educators (15%), psychiatrists (12%), social workers (7%), occupational therapists (4%), and other allied professions (15%).The average age was 40.13, ranging from 23 to 65 years of age.Approximately 77% of the sample were women.
Participants in the study were gathered from a pool of professionals who had participated in discussions, training sessions and awareness activities on mental health service user's rights.The scale was designed to be used as a baseline and follow-up measure.
The study received ethical clearance from the University of Barcelona institutional review board (IRB00003099).All participants gave informed consent and the questionnaires were completed anonymously.

Analysis
Before analysing the data, we carried out a search for outliers by calculating the mean of all the responses for each participant.We excluded a total of 7 questionnaires from the analysis as they were indicative of extreme values (most answers corresponding to one of the Likert scale anchor points).
We calculated frequencies, asymmetry and kurtosis parameters, as well as item-total correlations for each item, in order to decide upon its inclusion in exploratory and confirmatory factor analyses (EFAs and CFAs), as well as univariate and multivariate Item Response Theory (IRT) exploratory and confirmatory analyses.EFAs were combined with CFAs through the identification of stable and theoretically congruent dimensions appearing in consecutive principal components analyses, whose fit could be tested through structural equation modelling.In parallel, discriminative capacity of each item was tested using uni-and multivariate IRT analyses.Through the analysis of item-total correlations, EFA and CFA factor loading valences, items that indicated that they should be reversed were recoded, with higher scores indicating larger presence of negative attitudes or beliefs (see italicized items in supplementary table 1).Finally, reliability was tested using Cronbach's alpha.The psych [44], lavaan [45], ltm [46] and mirt [47] packages for the R software [48] were used to compute all the statistical analyses.

Results
Frequencies, asymmetry and kurtosis parameters for each item can be seen in supplementary table 1. Due to their low discriminative capacity, we decided to remove items with an asymmetry and kurtosis greater than 1 or less than -1 (8 items) and/or 90% of the cases included in one of the two halves of the Likert scale (9 additional items).We then calculated item-total correlations as well as a unidimensional IRT unconstrained latent variable model with the remaining 27 items.An item on professional pessimism (9), was removed because of nil (r=-.034)correlation with the rest of items and low discrimination parameter (-0.057).All the remaining items had discrimination parameters above 0.5 within a unidimensional IRT.*Items marked with an asterisk were the only four that weighted higher within a different dimension when performing an EFA (Exploratory Factor Analysis) with the final item set.IRT: Item Response Theory.
Consecutive exploratory factor analyses using Varimax and Oblimin rotations as well as exploratory IRT models were conducted with the 26 remaining items, using the eigenvalue higher than one criterion in CFA and forcing the structure to 2, 3 and 4 factors in CFA and exploratory IRT.This procedure was repeated, temporarily excluding items with low and distributed loadings and low multivariate discriminant parameters.Once we identified a coherent item group (those that tended to remain under the same dimension with high factorial weights and discriminant parameters in different CFA and IRT analyses), we calculated its unidimensionality through Cronbach's alpha and confirmatory factorial analyses as well as the discrimination parameters of the items.Each group was subsequently removed from the total pool of items and the whole process was repeated with the remaining items, until we obtained a congruent model ( 2).Discrimination parameters within a multidimensional IRT (MIRT) were also satisfactory (0.80 -2.62).Additional items were added one by one, based on theoretical coherence, factor loadings and discriminant parameters in all EFAs and confirmatory IRT models which were incorporated during the previous process (see supplementary table 2).We tested again the EFA weights (table 2), reliability (table 3) and unidimensionality (supplementary table 2) for the whole set of 25 items and for each dimension.We also tested the fit of the whole model having added only a specific item (see table 3 and supplementary table 2).All items added to each subscale (11, 13, 14, 37, and 39) improved its internal reliability and the fit of the whole model considered as unidimensional, without substantially affecting the fit of the four-dimensional model.As it can be seen in supplementary table 2, adding item 21 worsened all unidimensionality parameters.Additionally, we considered that it could be included as part of the tolerance/discrimination subscale.However, it did not improve the reliability of that subscale, nor any of the rest, and hence, it was removed.
Figure 1 shows the CFA path diagram of the final model.Discrimination parameters within the final MIRT were also satisfactory (0.70 -2.04, see table 2).

Discussion
According to our results, the BAMHS may be a useful tool to assess the impact of awareness and training activities on professionals' beliefs and attitudes towards service users' rights.This new scale offers flexibility and assumes no prior awareness or knowledge, making it especially suitable for its use in areas where user-led and progressive professional movements are carrying out activities with professionals without previous Recovery knowledge or awareness of user rights violations.
The results illustrate four final dimensions, namely: system criticism/justifying beliefs, freedom/coercion, empowerment/paternalism and tolerance/discrimination.The scale can be scored conveniently in any direction, with higher scores signifying higher respect or a higher violation of rights.We simply advise potential users to make it clear in the methodology of their research report.The final structure of the BAMHS showed an adequate fit according to CFA parameters, good reliability and good discrimination parameters.Adding six items to the core model did not substantively affect the overall fit of the model, nor of each of the modified dimensions or the discrimination capacity of each of the items.Additionally, none of the items that were included worsened the reliability of each dimension or the whole model.
The first dimension of the BAMHS materializes the professional beliefs that health-related professionals have which justify the status quo.Claiming that mental disorders are diseases like any other, that their aggressiveness is due to their mental disorders, that it is not possible to recover without the intervention of a professional, and even that some patients will never recover are statements that reinforce the need for mental health staff and their interventions.Regarding the former topic, some authors have stressed the role of biological and genetic attributions in the process of stigmatisation, including the belief that most mental disorders are chronic conditions [54,55].In some way, understanding that mental disorders are unrecoverable biological conditions, might tip the moral balance towards the justification of coercion [56].Accordingly, professionals scoring high on this subscale might also think that they only use these measures when necessary.In this context, declaring someone incapacitated, might be considered an adequate way of care.Finding a justification for the use of extraordinary measures in the very nature of mental disorders, might facilitate the concealment between the use of such measures and stating that 'mental health service users now have the same rights as other people' and that 'mental health professionals, in general, work collaboratively with patients'.These types of assertions are related to the complacency usually found among some mental health professionals despite the continuous use of coercion [57][58][59].
The freedom/coercion dimension addresses recurrent topics with mental health professionals when discussing service users' rights.The subscale includes questions on involuntary hospitalization, mechanical restraints, and, inversely, respect for service users' autonomy.We would like to highlight that more than half of our sample believed that mechanical restraints are sometimes necessary and that one should be involuntarily hospitalised even if they do not pose a threat to others.This is in contrast to the evidence that shows a worse prognosis [60], iatrogenia [61,62], and even death [63] caused to people subjected to such coercive measures.Conversely, restraint reduction has been shown to be feasible [64] and to reduce the risk of injury and medical leave among nursing staff [65].
The next subscale, empowerment/paternalism, represents a series of beliefs related to the supposed inability of people diagnosed with mental disorders to take charge of their lives including having children, making decisions regarding their treatment or prioritizing treatment over dignity [66,67].This justifies paternalism in the form of guidelines and constant support, emotionally distant and value-free practices [68].
Finally, a fourth subscale tolerance/discrimination materializes widespread prejudices towards mental health service users.Discrimination occurs in different contexts, for instance this can include employment discrimination (as many would not feel comfortable with a diagnosed teacher) which is evidenced through low occupational rates [69].Likewise, social distance reflects the main reason for the stigma that people with mental health problems experience [70,71].Other discriminatory practices include access to healthcare [72], and those included in legislation, such as the prohibition to vote [73].The main limitation of this validation study is the use of a convenience sample formed by professionals willing to participate in awareness activities.This may have caused biases, such as social desirability, due to the profile of the participants in the activities in which this validation is contextualized.However, this scale is designed to evaluate changes in professionals willing to participate in activities where patients' rights are discussed.Therefore, we believe that it can be a useful tool to evaluate awareness activities in the mental health field.

Conclusions
In conclusion, we believe that our instrument brings a new perspective to the measure of beliefs and attitudes of mental health professionals in the context of the new era opened by the Convention on the Rights of Persons with Disabilities [74].
Supplementary Materials: Table S1: Descriptive data of the initial 44 item pool; Table S2

Figure 1 .
Figure 1.Structural equations diagram of the scale.

Table 1 .
Previous measures of mental health professionals' beliefs and attitudes towards service users' rights.

Table 2 .
Factorial weights and discrimination data of the final item pool.

Table 3 .
Reliability of total scale and 4 final dimensions.