Background
Cognitive behaviour therapy (CBT), delivered over a minimum of 16 sessions, is the only individual psychological therapy recommended for the treatment of psychosis in a number of countries [
1‐
4]. However, access to CBT for psychosis (CBTp) is poor, with recent figures from the UK suggesting only 10% of people with a psychosis diagnosis are offered CBTp [
5]. The poor implementation of CBTp is not just limited to the UK, but is an international problem that is reported in the United States [
6‐
9], Canada [
10], and Australia [
11]. The CBTp access rates are not available for many countries, but it is likely that if these more affluent countries are not able to facilitate access, then countries with less economic resource would also experiencing implementation challenges.
One of the most commonly cited reasons for the poor access to CBTp is a lack of resources, including lack of protected time and staff shortages [
12]. One possible approach to increase access to CBTp is by developing interventions that can be delivered using comparatively less resources. A recent meta-analysis found briefer forms of CBTp (i.e. fewer than the recommended 16 therapy sessions) led to a significant reduction in psychosis symptoms compared to control conditions [
13] These brief CBTp interventions typically targeted a specific symptom associated with psychosis (e.g. delusions or voices). Consequently, there is potential for brief forms of symptom-specific CBTp to be offered in the first instance. Drawing on a stepped care approach [
14], more resource intensive forms of CBTp could then be delivered only to those still in need. However, based on the broader CBTp literature, we know demonstrating effectiveness does not necessarily lead to widespread implementation. For example, a recent audit of CBTp implementation within a NHS healthcare trust found that only 6.9% of people with psychosis were offered CBT, despite NICE [
2] recommending that everyone with psychosis should be offered CBTp [
15]. Therefore, in addition to investigating the effects of brief CBTp, we need to consider the potential challenges and facilitators to implementing this novel intervention.
This problem of implementation does not just apply to CBTp. It is common for healthcare services to experience delays in the process of implementing new treatments more broadly [
16]. The difficulties associated with implementation has led to the development of numerous theoretical models that aim to understand and simplify this process [
17]. A review by Tabak, Khoong, Chambers and Brownson [
18] identified 12 separate models of implementation; however only two of these (Conceptual Model of Implementation Research, [
19]; Normalisation Process Theory, [
20]) consider implementation at multiple levels, including the individual and system levels.
The Conceptual Model of Implementation Research [
19] synthesises previous theories of implementation to create a model that suggests different ways that implementation can be conceptualised (i.e. systems environment, organisational, learning, supervision, individual providers), and measured (i.e. feasibility, fidelity, penetration, acceptability, sustainability, uptake, and costs). The main purpose of this model is to explain the different outcomes that can be used to assess implementation, and the relationship between these outcomes [
21]. Although this model is useful, it is not appropriate to be used within the present study, as its purpose is not in line with our study aims. This study aims to explore the barriers and facilitators to implementing a brief CBTp intervention prospectively; whereas the Conceptual Model of Implementation Research considers implementation retrospectively, and does not include a framework for exploring what these barriers and facilitators could be.
Conversely, the flexibility of the Normalisation Process Theory (NPT; [
20]) means it can be appropriately applied to the present research study. NPT provides a theoretical framework to guide the implementation process. The theory specifies four factors that may enhance the likelihood of successfully implementing a new idea into an existing service: (1) coherence: the attitude of staff towards the new idea, (2) cognitive participation: the willingness of staff to be involved in implementation, (3) collective action: service level pragmatics involved in implementation, and (4) reflexivity: how the implementation process should be evaluated. This model can be used to consider the implementation of a brief CBTp intervention prospectively, and the NPT factors provide a theoretical basis from which barriers and facilitators can be explored.
NPT [
20] has been applied to many different healthcare interventions and contexts, including physical health, service infrastructure, and mental health [
22]. Looking specifically at the mental health related research, NPT [
20] has been used to explore the implementation of stepped care [
23], depression interventions [
24] and collaborative care [
25], primary mental health care [
26], bipolar treatment guidelines [
27], and problem-solving therapies [
28]. Some of these studies applied the NPT model [
20] retrospectively as a means of reviewing a previous implementation process e.g. [
26]; and some utilised NPT [
20] prospectively to develop an implementation plan e.g. [
25]. All of these studies used qualitative research methods to understand implementation within the NPT framework, and all concluded that NPT was a useful and comprehensive model to guide the implementation process in mental health service settings. There are currently no studies however that have tested the validity of the NPT model using a quantitative design in a mental health context.
A NPT questionnaire has recently been developed (NoMAD; [
29]). The psychometric properties of this measure are currently under assessment [
30]. Similar to the Conceptual Model of Implementation Research [
19], the items are phrased to look at implementation retrospectively. Furthermore, while the NoMAD measure is suitably vague to enable its use across multiple settings, this does limit its practical use in certain contexts. For example, one item on the NoMAD asks whether ‘sufficient resources are available to support the intervention’. In the context of our brief CBTp intervention, resources can be taken to mean the number of clinicians, clinician’s time, training, or information [
12]. Consequently, we have developed our own questionnaire measure based on the NPT model [
20] that has been specifically developed to investigate the prospective implementation of a brief CBTp intervention.
We plan to implement a brief CBTp intervention for distressing voices (CBTv) into National Health Service (NHS) mental health services in the UK [
31]. The intervention is designed for adults who are distressed by hearing voices and who are currently receiving mental health care in an NHS service. Services would most typically be either secondary care community teams or early intervention for psychosis services. Accredited therapists generally have a positive attitude towards, and frequently use, self-help materials in their clinical work [
32]. In contrast, mental health nurses have reported feeling sceptical about the value, and even the appropriateness, of talking to people about their voice hearing experiences [
33]. These differing attitudes towards these aspects of guided self-help CBTv, suggests that mental health practitioners as a workforce may not be a homogeneous population. This has implications for our study as the findings from our NPT questionnaire could be moderated by the sample characteristics.
In light of this research, our study aims to: (1) test the validity of the four factor NPT model within our questionnaire using factor analysis; and based on the established factor structure, (2) identify mental health practitioner views on the implementation of guided self-help CBTv, and what sample characteristics may moderate these views. To meet our second study aim we will explore the following research questions: (a) Do attitudes differ between those who do and do not have accreditation to deliver therapy? (b) Do attitudes differ depending on the participants’ level of experience working with people who hear voices?
Discussion
The first aim of our study was to test the proposed factor structure of Normalisation Process Theory (NPT) [
20] using exploratory factor analysis. We found partial support for the NPT model as the three factors extracted were akin to three of the NPT factors: (1) coherence, (2) cognitive participation, and (3) reflexive monitoring. This study seems to be the first test of the NPT model using factor analysis, although the findings of the NoMAD factor analysis are imminent [
30]. Our findings suggest that coherence, cognitive participation, and reflexive monitoring are important facets of implementation that should be considered prior to the dissemination of brief CBTp interventions. This result is particularly compelling as the use of EFA meant that any factor structure could have emerged. It is possible that these three factors would also emerge as important when understanding the implementation of healthcare interventions more generally. However, as our study asked participants about a specific intervention (guided self-help CBTv) further research using factor analysis is required to determine the generalizability of the NPT model.
We failed to find support for the fourth NPT factor, collective action, which speaks to the feasibility of implementing the new intervention into the existing service. This study recruited clinicians currently working in mental health services, rather than staff in more senior positions – such as service managers. Clinicians generally do not have the power or responsibility to make service-level decisions. As the collective action factor describes a facet of implementation that occurs at the service-level, this factor could be argued as inconsequential to our sample, and therefore explain why this factor did not emerge in our analysis. If participants in these more senior positions had been recruited to the study, it is possible that we may have found support for the collective action factor.
A kin to our study, most NPT studies in mental health settings recruited practitioners [
24,
25,
28]. However the NPT studies by Gask et al. [
26] and Franx et al. [
23] did include service leads and managers to investigate the implementation of mental health care, and stepped care, into primary care services respectively. Both of these studies found qualitative support for the Collective Action factor, as having coherent and consistent leadership across the services was associated with successful implementation. However, neither study explored the moderating effect of profession, nor did they validate the NPT model quantitatively. Consequently we suggest that future tests of the NPT model should include participants of varying levels of seniority. These studies would benefit from the use of quantitative, moderation analysis to explore the effects of seniority, and address this limitation of our study and previous research.
The second aim of our study was to examine clinicians’ attitudes towards guided self-help CBTv, and whether these differed as a function of therapy training (therapist versus non-therapist) and experience working with clients who hear voices. We found that clinicians’ attitudes were favourable across all three factors (all
Ms. < 3; see Table
2). With respect to each of the factors extracted, these ratings can be interpreted to mean clinicians are, on average, supportive of the concept of guided self-help CBTv (coherence), are willing to be involved in the implementation (cognitive participation), and agree with the proposed means of evaluating the implementation (reflexive monitoring). These are encouraging findings as they suggest that most clinicians working in NHS mental health services in the UK have a positive attitude about guided self-help CBTv and would be willing to support its implementation and evaluation. This suggests that clinicians’ attitudes and willingness to be involved would not be barriers to implementation of guided self-help CBTv in the NHS.
Only therapist training significantly moderated the clinicians’ attitude, with qualified therapists reporting significantly less favourable attitudes on the coherence subscale compared to non-therapists. In practical terms however, this difference may be negligible as the mean difference between these groups was 0.42 on the seven point Likert scale. In addition, in both cases, therapists and non-therapists had mean ratings that were in the favourable range (<3) suggesting that whilst therapists were somewhat more sceptical, they were still, on average, positive in their attitudes towards to the intervention. However, there is some evidence to suggest attitudes towards psychological therapy can vary as a function of the clinicians’ profession and training. The majority of the literature suggests therapists view psychosocial interventions for psychosis with a greater optimism compared to other mental health professionals [
42]. However there is some evidence that concurs with our findings, as therapists seem to be more pessimistic than mental health nurses about their ability to ‘treat’ psychosis [
43]. Therapists also report that delivering brief CBTp interventions can be problematic owing to the limited number of sessions involved, and the complex nature of many patients’ presenting problems [
44]. Whether the therapists’ reservations are realised in practice requires further research.
Overall, it seems clinicians show support for implementing guided self-help CBTv, which is encouraging. This finding contrasts with previous research that suggests mental health clinicians may not be supportive of interventions that invite people to talk about the voices they experience [
33]. Perhaps the recent growth of emancipatory approaches to voices, such as the Hearing Voices Movement, has helped to demonstrate the therapeutic value of openly discussing voices [
45]. Whether clinicians’ positive pre-implementation attitudes will aid the actual implementation process remains to be seen. There is evidence to suggest that negative clinician attitudes are associated with poorer intervention outcomes [
46]. The favourable attitudes of clinicians in the present study are therefore welcome as this will help to create the optimal service environment, from which we can explore the effectiveness of this intervention.
Limitations
It is possible that the reason we did not find support for the fourth NPT factor is because the items developed to target this factor were poor representations of the collective action factor. That is to say, our items may not have sufficiently examined implementation at the service-level. Before accepting this as a study limitation, it is important to first explore whether the sample characteristics may have contributed to the factor structure extracted. As mentioned previously, future studies should aim to include participants in more senior-level positions to see whether this causes the emergence of the collective action factor.
Our questionnaire was designed to investigate implementation prospectively. Our findings are therefore indicative of mental health practitioners anticipated barrier and facilitators to implementation – our study cannot determine whether this will translate into practice. This limitation highlights the benefits of longitudinal implementation research. Using this research design will help to determine whether positive pre-implementation attitudes translate into a successful initial implementation, and the sustained employment of the intervention. To our knowledge, there are currently no quantitative studies that have looked at the NPT’s model ability to predict implementation outcomes; however this is one of the study aims of the NoMAD psychometric assessment [
30] which is currently underway.
The measure we have developed appears to have factorial validity. However, we did not assess other forms of validity such as construct and divergent validity. This will be the focus of future research evaluating the psychometric properties of the measure. Furthermore, factors were extracted using a combination of Kaiser’s [
39] criteria and the Scree Plot. Although this method of factor extraction is arguably the most widely used and well-established in scale construction, other methods such as parallel analysis are gaining support [
47].
Research implications
Our study has identified a number of areas for future research. For example, future studies assessing the validity of the NPT model [
20] should aim to recruit participants of varying levels of seniority, and determine whether pre-implementation attitudes correspond to the subsequent ease of implementation. As our study seems to be the first to use factor analysis to test the NPT model, more factor analysis studies are needed to see whether the NPT factor structure can be generalised to the implementation of other interventions in both mental health and physical health contexts.