Main findings
This study used TPB-based constructs supplemented by a multiple goals approach to investigate control beliefs and the facilitating and interfering goal-directed behaviours that GPs perceived as affecting their performance of two evidence-based behaviours in a diabetes consultation. Results showed that indeed GPs perceived other goal-directed behaviours as interfering with and facilitating performing the focal evidence-based behaviours, though to a different extent between behaviours. The majority of perceived goal facilitation and interference was elicited beyond the standard control belief elicitation. Results were in line with quantitative research conducted with other populations that found that the interfering [18–20, 43] and facilitating (Presseau J, Sniehotta FF, Francis JJ, Gebhardt WA: With a little help from my goals: Integrating intergoal facilitation with the theory of planned behaviour to predict physical activity, Submitted) [17] effect of other goal pursuits were related to the performance of a particular behaviour. This study contributes to this research by providing qualitative evidence that GPs perceive that goals they pursue when managing diabetes interfere with and facilitate their performance of evidence-based behaviours. This study adds to the literature by considering how both the content and duration of this perceived interference and facilitation may affect performance. In doing so, this study suggests promising lines of development of behavioural theory to reflect physicians' perceived competing demands in clinical practice. Behavioural approaches to implementation research may benefit from further investigation of the perceived influences of pursuing multiple goals over and above intentions and PBC.
Content of perceived goal interference and facilitation between focal behaviours
While similar types of goal-directed behaviours were perceived to interfere with both focal behaviours (though more frequently for PA advice), BP prescribing was consistently described as strongly intended whereas intention to provide PA advice varied between GPs. This suggests an underlying (and perhaps not surprising) potential difference in relative priority between the two focal behaviours for some GPs. The implication is that when goals compete, the less prioritised goal-directed behaviours may be subject to a greater influence by other interfering goal-directed behaviours.
As opposed to goal interference, as many participants described goal-directed behaviours that facilitate giving PA advice as prescribing to reduce BP. Though some goal-directed behaviours were perceived to facilitate both focal behaviours (including each other), a key content-related difference distinguishes the two: one-half described 'consultation' goal-directed behaviours as facilitating BP prescribing (compared to one GP for PA advice), whereas eight described other 'lifestyle' goal-directed behaviours as facilitating giving PA advice (compared to three GPs for BP prescribing). Performing 'consultation' goal-directed behaviours may effectively provide a supportive context for performing the highly intended behaviour. Conversely, the behaviour with more variable levels of intention was not described as being facilitated by such consultation goal-directed behaviours, but rather by the cluster of other similar lifestyle goal-directed behaviours. These differences between focal behaviours again suggest an underlying difference in relative priority. When time is limited, we question whether facilitating similar (e.g., other lifestyle) goal-directed behaviours would increase the likelihood of a focal behaviour being performed, because that facilitating effect would depend on those similar behaviours also being performed. However, facilitating goal-directed behaviours at the consultation level may provide a context that favours the facilitated focal behaviour despite time limitations. Certain types of goal-directed behaviours may therefore be more useful for promoting the performance of a focal evidence-based behaviour.
Goal facilitation and interference along a temporal dimension
Despite the interviews focusing on perceived intergoal relations within a single consultation, the longitudinal and chronic nature of diabetes care was often reflected in GPs' responses when discussing facilitating goal-directed behaviours. This suggests that goal facilitation may operate beyond the single consultation and that pursuing such goals over a series of consultations eventually facilitates performing the focal behaviour (i.e., prospective facilitation). While this lead-up prospective facilitation is reminiscent of Bandura's 'proximal subgoals' [44] and Bagozzi's 'instrumental acts' [45], the latter concepts are framed within a perspective that is explicitly focused on a single behaviour. Conversely, the concept of prospective goal facilitation takes a systems-based perspective. The system can be considered as made up of multiple goal-directed and valued behaviours that are performed in and of themselves, rather than expressly to facilitate a particular behaviour. This temporal perspective of prospective goal facilitation may help to account for the longitudinal aspects of general practice often recognised as a main advantage, such as continuity of care [46]. It also presents with the possibility of developing strategies for promoting facilitation based on planning (e.g., facilitation planning) that extend over many consultations.
While an equivalent temporal dimension for goal interference was not overtly described by GPs, the perceived interfering relationship between goal-directed behaviours can nevertheless be considered along a temporal continuum. For instance, some identified interfering goal-directed behaviours can be considered as one-offs, representing a more transient form of interference confined to a single consultation (e.g., treating an acute illness, dealing with pressing issues). Other goal-directed behaviours presented a more enduring interference because they are potentially performed frequently and recurrently over time (e.g., fitting in the patient agenda, capturing other information for the GP contract). The advantage of distinguishing this temporal dimension lies in the possibility that separate strategies may exist for dealing with such perceived interfering goals. Transient interference can be dealt with using deferral strategies [47], whereas enduring interference is by definition longitudinal in nature and thus continuous deferral would likely be detrimental. Enduringly interfering goal pursuits may also be an indication of the relative priority of a goal-directed behaviour; if many goals interfere over a long period of time with performing a particular behaviour, the latter may not be seen as important or useful. Enduring interference may be particularly problematic for optimal performance of evidence-based behaviours, and future research could specifically identify whether duration of perceived interference affects performance of particular focal clinical behaviours. That said, identifying and promoting facilitating goal-directed behaviours may circumvent these more enduring perceived interfering goal-directed behaviours, as could re-evaluating, modifying or disengaging from a particularly interfering goal [47].
Relative priority between goal-directed behaviours
The relative priority between the focal behaviours was an underlying finding in this study. Despite more barriers expressed for prescribing to reduce BP, it was also consistently described as strongly intended whereas intention to give PA advice was variable. Differences in relative priority are not surprising because PA advice can often also be provided by other primary care staff (e.g., practice nurse), whereas prescribing to reduce BP is primarily the GP's role (though increasing dosage can be nurse-led). While some GPs may indeed prioritise diagnosing and treating diabetes, the variation in described strength of intention to give PA advice suggests that this is not true of all GPs. Future research should investigate whether perceptions about professional role influence the priority of a particular evidence-based clinical behaviour relative to other goal-directed behaviours performed in a consultation.
In a null-sum situation of limited time something must give way, and this is likely determined by the perceived priority of each goal-directed behaviour. However, applications of single behaviour models to health professional behaviour [5, 8] inherently do not consider this. A GP may intend to address cholesterol and BP with a patient, and defer addressing BP to the next consultation in order to be able to pursue both. However, this still raises the question of which behaviour should take precedence and which should be deferred. This may be less of an issue when follow-up consultations or extra time slots [48] are readily available. However, the follow-up consultation also presents with another set of goal-directed behaviours themselves potentially interfering with the now deferred behaviour. Whether or not the deferred behaviour's priority has changed may again be a function of what other goal-directed behaviours the GP performs in the follow-up consultation. The effectiveness of strategies for dealing with interference and promoting facilitation may also ultimately depend on which goal-directed behaviours are prioritised at any given time. Given that BP prescribing for people with diabetes is currently related to a GP contract-remunerated target in the UK, while PA advice is not seems a likely reason for differences in relative priority. Indeed, relative priority is likely to be influenced by a number of behavioural, normative, and control beliefs, and future research focusing on influences of priority seems justified.
Comparing control beliefs and perceived intergoal relationships
Control beliefs and perceived intergoal relationships have similarities; indeed both reflected similar themes in this study. In theory, one would expect intergoal conflict and facilitation to be reflected in perceptions of perceived control. Regardless of whether they represent a more detailed facet of control beliefs or are independent constructs, questions and prompts of goal facilitation and interference elicit content that might otherwise be missed in standard belief elicitation studies. Indeed, while some of the coded perceived intergoal relationships emerged following control belief elicitation, the vast majority of coded perceived goal facilitation and interference (71% to 92% of codes) was elicited using questions and prompts for these constructs or when discussing intention. In itself, this argues that it may be important to further consider the context within which focal clinical behaviours are performed, including competing goal-directed behaviours.
Further conceptual and empirical factors can also attest to their distinctiveness. Conceptually, control beliefs 'deal with the presence or absence of requisite resources and opportunities' [4]. Conversely, goal-directed behaviours compete for those resources and opportunities, are performed independently for their own sake, and are determined by their own set of beliefs, perceptions, and intentions. Perceived intergoal facilitation and interference are constructs that partly represent sources of resource competition, and thus may influence control beliefs about a particular goal-directed behaviour. For instance, 'focusing on GP contract goals' was described as a goal-directed behaviour that interfered with giving PA advice. Pursuing these perceived interfering contract goals may then lead the GP to perceive a time constraint (i.e., a control belief). Perceived intergoal relationships might also influence other control-related beliefs. For instance, 'engaging the patient' and 'negotiating with the patient' were goal-directed behaviours described as facilitating prescribing to reduce BP, and their pursuit may influence control beliefs described as making it easier to prescribe, such as 'knowing the patient'. These examples suggest that perceived intergoal relationships may contribute towards control beliefs about a particular goal-directed behaviour, but are conceptually separate.
That said, despite our focus on control beliefs, perceived intergoal relationships may also inform other types of beliefs. For instance, the perceived facilitating effect of 'talking about weight' might affect a behavioural belief that it is good practice to talk about exercise, and the perceived interference of 'pursuing other GP contract targets' might affect normative beliefs about whether colleagues think the GP should prescribe. Furthermore, these perceived intergoal relationships may influence a behaviour without necessarily informing specific beliefs about the behaviour, leading to a independent influence on behaviour. While these effects require quantitative substantiation in a clinical sample, perceived intergoal facilitation has been shown to be partially mediated by the TPB and also additionally independently predict behaviour in a non-clinical population [Presseau J, Sniehotta FF, Francis JJ, Gebhardt WA: With a little help from my goals: Integrating intergoal facilitation with the theory of planned behaviour to predict physical activity, Submitted]. This further attests to the distinction between control factors and perceived intergoal relationships.
Implications for implementation science
Implementation science is concerned with understanding and promoting the application of research into practice, which involves the behaviour of health professionals. Theory-based models of behaviour allow us to build a cumulative science to understand the factors that are perceived to relate to performing according to the standards set by current evidence. Investigations of extensions to such models of behaviour allow us to maintain their foundations while attempting to address identified shortcomings. This qualitative study contributes hypothesis-generating results towards the further development of behavioural theory to better understand such variations in evidence-based health professional behaviour. This study suggests that what GPs do and pursue during a consultation are perceived to influence each other in a helpful or hindering way. Rather than solely focusing on a single investigator-identified behaviour, busy time-constrained consultations may be more appropriately conceptualised by also explicitly considering the perceived influence of GPs' other goal-directed behaviours. Gaps between research evidence and the performance of a particular clinical behaviour might be addressed by focusing attention upon what else the GP wants to do and does during the consultation, and how they relate to the focal behaviour. In some instances, many of the other goal-directed behaviours in the consultation are perceived to interfere with its performance. For others, the extent of interference is lesser (perhaps due to a higher relative priority), though behaviour may still be marred by a number of identified control beliefs. The value of a multiple goal-directed behaviour approach to implementation science may be as a means of: assessing how higher-level policy driven goals such as 'provide patient centred care' and 'provide evidence-based care' are pursued (i.e., goal-directed behaviours) and how these pursuits may facilitate or interfere with one another; identifying and promoting sustainable clinical goal pursuits that facilitate particular evidence-based behaviours; and identifying and addressing competing goal pursuits that interfere with these evidence-based behaviours.
For instance, eliciting the multiple goal-directed behaviours that professionals perform and assessing their perceived interfering and facilitating influence on a focal behaviour may raise the awareness and salience of otherwise habitually performed behaviours. This could provide the opportunity to target interfering goal relations (that may or may not be related to control belief-related barriers). Once this interference is identified, and if appropriate, strategies can be adopted to minimise its effects. In this study GPs reported that respecting patient choice interfered with prescribing to reduce BP (Table 1), and that whether the patient 'understands and is informed' made it easier to prescribe (Additional File 2). They also perceived that performing the goal-directed behaviour of 'educating patients' facilitated prescribing to reduce BP (Table 2). Thus, a strategy of educating patients may both facilitate performance of the target behaviour and promote the factors seen as making it easier to prescribe to reduce BP, minimising the potential influence of the interfering goal. Promoting such facilitating sequences of goal-directed behaviours uses the existing structure of goal pursuit, rather than necessarily introducing new goal-directed behaviours. This could involve prospective facilitation whereby facilitating goal-directed behaviours can be identified and prospectively planned to be performed over time, which may provide a theoretically-informed operationalisation of continuity of care.
Strengths and limitations
This study used an explicit and a priori-specified theory-based methodology as a foundation for thematic analysis. This approach is a strength of this study because it allowed us to integrate knowledge and evidence from existing theories to extend current ones, rather than (re)inventing a new theory [49]. While further quantitative evidence is needed to substantiate the qualitative findings in this study, by moving beyond single behaviours studied in isolation, this study attempted to bring some clarity to the complexity of clinical practice. The theory-based methods support the results in contributing to building a cumulative evidence base of the implementation of health professional behaviour. Methodologically, the double coding and inter-rater reliability assessment are also a strength. While this study is limited by a small sample size, this is mitigated by the purposive heterogeneity sampling strategy used to explore the breadth of responses. It became evident in the later interviews that the research questions had been sufficiently answered, i.e., that GPs did perceive their goal-directed behaviours as facilitating and influencing performing the two focal behaviours. Though the study was not designed to necessarily achieve data saturation, evidence from the literature suggesting that a sample size of 12 can provide as much information as a much larger sample in qualitative studies [37].
Unanswered Questions
While the study design precludes us from drawing conclusions about whether perceived intergoal relationships might augment the TPB, this study nevertheless allows us to generate hypotheses, particularly when also considering research in non-clinical populations. Future investigations could test hypotheses regarding whether perceived intergoal relationships build independently on TPB constructs, or moderate the relationship between clinicians' intentions and their behaviour. Whether promoting facilitating goal pursuits and reducing the effect of interfering goals might affect performance of a focal behaviour also remains an open question. Another unanswered question involves GPs' reports of high intention to prescribe to reduce BP, but expressing conditions related to the situational demands of the consultation that affect that high intention; future research should consider the implications of these 'conditionalities'. Finally, it seems plausible to have a strong intention towards many behaviours while still prioritising some over others, as priority implies urgency. Future investigations distinguishing 'priority' from alternative comparative measures, such as intention-choice [11] or relative intention (e.g., rank or difference between intention to perform multiple goal-directed behaviours) may contribute to understanding the effect that multiple goal-directed behaviours have on performing a focal behaviour.