Analysis System for Self-Efficacy Training (ASSET)

Abstract

Objective

The paper presents the development of a coding tool for self-efficacy orientated interventions in diabetes self-management programmes (Analysis System for Self-Efficacy Training, ASSET) and explores its construct validity and clinical utility.

Methods

Based on four sources of self-efficacy (i.e., mastery experience, role modelling, verbal persuasion and physiological and affective states), published self-efficacy based interventions for diabetes care were analysed in order to identify specific verbal behavioural techniques. Video-recorded facilitating behaviours were evaluated using ASSET.

Results

The reliability between four coders was high (K = 0.71). ASSET enabled assessment of both self-efficacy based techniques and participants’ response to those techniques. Individual patterns of delivery and shifts over time across facilitators were found. In the presented intervention we observed that self-efficacy utterances were followed by longer patient verbal responses than non-self-efficacy utterances.

Conclusion

These detailed analyses with ASSET provide rich data and give the researcher an insight into the underlying mechanism of the intervention process.

Practice implications

By providing a detailed description of self-efficacy strategies ASSET can be used by health care professionals to guide reflective practice and support training programmes.

Introduction

Over the last decade, research on intervention studies has led to the development of standards on how to run, evaluate and present interventions [1], [2], [3], [4]. Despite the fact that systematic approaches are broadly adopted, evidence on the fidelity of interventions is still scarce. Interventions are evaluated on the basis of outcome measures as it is assumed that the underlying theory has been properly implemented [5]. However, detailed analysis of intervention delivery could provide insights into behaviour change mechanisms (e.g., intention—behaviour gap), enable theory testing (e.g., hierarchy versus complementary relationship of four sources of self-efficacy) and control for potential mediators of intervention delivery process (e.g., Hawthorne effect).

Even when applying a specific systematic approach, there has been no evidence that interventionists adhere to the training protocol and act in line with the underlying theory. Lack of an active control group and of process evaluation in their studies raises the question of whether the results are based on active components of the intervention (i.e., theory-driven techniques) or are caused by non-specific factors such as therapist time or attention, e.g. [6], [7]. Studies showed that nurses trained to deliver specific strategies tended to skip them when pressed for time [8]. Furthermore, even when health professionals were convinced that they delivered specific interventions, according to external observers they did not do so. For example, Koopman-van der Berg and van der Bijl [9] evaluated self-efficacy based interventions delivered by nurses facilitating diabetes self-management programmes. Nurses reported utilization of the strategy of setting goals but did not deliver it according to the independent raters.

This raises the question as to how can we know whether an intervention has implemented a theory as intended? Michie et al. have proposed a common taxonomy of behaviour change techniques [10] so researchers can use pre-defined techniques to describe their interventions. This taxonomy provides a useful common language to describe but not evaluate an intervention. There are a few coding tools in the area of health psychology designed to evaluate interventions [11], [12], [13], [14]. However, none of these tools can be implemented to test treatment fidelity, as they were not designed to test a specific theory. Rather they provide a means to describe certain features of the interaction between a health professional and a patient.

Therefore, it is clear that there is a need for coding tools that can be used to evaluate whether theories have been implemented as designed. However, the problem with theory-based tools is that they would largely only be relevant to researchers using specific theories, and would thus have restricted utility. Social cognitive theory [15] and its focus on the development of self-efficacy is a notable exception to this general rule. Self-efficacy and self-efficacy like constructs are now incorporated into a large number of health behaviour theories [16], [17]. Thus, developing a tool that can be used to assess the implementation of self-efficacy in health interventions would have high applicability for theory validation. The coding of the use of different techniques in this way can also be used to provide feedback to health professionals, supporting ongoing reflective practice and to inform training programmes to develop self-management support skills. Furthermore, by examining how patients respond to the different strategies used by the health care professional, this tool could potentially help develop clearer guidance for health care professionals in how to respond to patient questions. Also, by looking at the implementation of techniques and patients’ response to them the interaction between patients and health professionals can be explored. For example, one could examine to what extent nurses’ way of talking encourages patients to talk. This has the potential to enhance understanding of the most efficient way for professionals to deliver self-management education to patients with chronic illness.

Self-efficacy, the belief in the personal ability to master adversities, is a crucial predictor of successful self-management and the most often targeted component in health behaviour change interventions [18], [19], [20]. Social cognitive theory incorporates not only the description, but also operationalisation of self-efficacy [5], [15]. Bandura introduced the four sources of self-efficacy in a hierarchical model with ‘mastery experience’ being the most powerful source of self-efficacy [21]. We introduce a complementary model of the self-efficacy based strategies. The four sources of self-efficacy can be considered on two broad dimensions: actor of and behaviour involved in pursuing the technique [22]:

• Who talks (i.e., who is the source of self-efficacy).

• What happens (i.e., action or action related talk versus appraisal of an event).

Whilst ‘mastery experience’ relates to gaining self-efficacy belief from one's own experience (i.e., the facilitator creates the opportunity for an individual to be in action), ‘role modelling’ addresses the opportunity to gain self-efficacy belief from successful others (i.e., facilitator creates the opportunity to observe others in action). On the other hand, ‘verbal persuasion’ is based not on action but on an appraisal made by others (i.e., the facilitator appraises an individual's skilfulness). Physiological and affective states in turn relate to appraisal of symptoms experienced by an individual (i.e., the facilitator creates the opportunity for an individual to attribute physiological and affective symptoms). Fig. 1 represents the conceptual framework of the underlying categories.

This provides a framework for coding self-efficacy promoting attempts of health professionals. However, these broad strategies need to be operationalised into specific verbal techniques. Therefore, the aim of the studies reported here were to develop a coding tool for the implementation of social cognitive theory (the Analysis System for Self-Efficacy Training, ASSET), to demonstrate its inter-rater reliability and to explore its construct validity and clinical utility. In order to identify the techniques used in self-management interventions to enhance peoples’ self-efficacy we focused on diabetes care. Self-efficacy has been widely targeted in self-management programmes [36], [37] and widely investigated in diabetes research [9], [38], [39]. Self-efficacy appears to have an important role in diabetes in terms of improvement of psychological [33] and physiological health [34] as well as maintenance of health behaviours [35].

We expected that carrying out data analysis using ASSET would provide us with detailed descriptions of the intervention delivery process and enable us to identify individual patterns of delivery among facilitators and across time. We expected that self-efficacy driven techniques would be significantly shorter in time than non-self-efficacy orientated speech, as the purpose of the intervention was to activate rather than solve. We also expected that participants would develop more diabetes-related knowledge and skills as the programme progressed, resulting in more personal experiences to reflect on. Thus, we expected the amount of self-efficacy driven techniques to increase over the duration of the intervention.