Background
Knowledge translation (KT), the application of research evidence into clinical practice, has been characterised as a haphazard process [
1]. The KT process can be broken down into a series of behaviours performed by individuals to reach a goal (
i.e., goal-directed behaviours, or GDBs). When viewed as such, theories of human behaviour can be employed to identify factors that predict the behaviours involved in translating research evidence into practice [
2]. For example, clinical practice guidelines in the UK recommend that primary care health professionals provide all patients, and especially those at greater cardiovascular risk, with advice on engaging in regular physical activity (PA) for health promotion and disease prevention [
3,
4]. However, evidence suggests that provision of PA advice is less than optimal [
5,
6]. By acknowledging the provision of PA advice as a health professional behaviour, behavioural theory can be used to understand factors that account for variability in optimal PA advice provision.
Among theories of behaviour, the theory of planned behaviour (TPB) [
7] has been tested across a variety of populations, behaviours, and contexts [
8]. The TPB suggests that behaviour is a function of four constructs: intention, attitude (evaluation of the behaviour), subjective norm (perceived social pressures), and perceived behavioural control (PBC; ability). Intention, the key construct in the model, is a proximal predictor of behaviour as well as a mediator of the effect of attitude and subjective norm on behaviour and a partial mediator of the effect of PBC on behaviour. While the TPB is among the models with the best utility in predicting health professionals' GDBs [
9,
10], it is not without its limitations [
11]. Among them is the issue of behavioural segregation: the TPB focuses on a single GDB, isolated from other GDBs engaged in by health professionals. In contexts of multiple goal pursuit, such as clinical consultations, these other GDBs may have a helpful or hindering influence on a focal GDB. Competition for limited resources (
e.g., time, energy) may lead to goal conflict. However, engaging in some GDBs may be helpful and increase the likelihood that a particular GDB is performed, thereby representing goal facilitation. Goal conflict and goal facilitation may influence the extent to which a health professional engages in a given guideline-recommended behaviour. If so, the incorporation of these constructs into the behavioural pathway may supplement the explanatory power of the TPB and help to further understand KT processes. The current study aimed to explore whether goal conflict and goal facilitation are predictive of health professional behaviour beyond the proximal predictors of behaviour from the TPB.
The TPB has been frequently used to predict health professional behaviour. A systematic review of social cognition models applied to predict health professional behaviour identified 14 prospective studies testing the TPB with 1,882 health professionals [
10]. The identified studies explained a frequency-weighted mean of 35% of the variance in health professional behaviour, and intention and PBC were each consistent predictors of behaviour [
10]. Furthermore, when compared against other social cognition models within the same sample, the TPB is the most predictive model [
9]. The TPB posits that while additional background constructs might be relevant to understanding behaviour, their effect should be mediated through the model [
12]. Nevertheless, a number of other social cognitive constructs have been proposed to supplement the TPB. For example, Godin and colleagues [
10] hypothesised an augmented TPB that includes additional predictors of intention (role and identity, moral norm, and health professional characteristics) and behaviour (habit and past behaviour). Although these constructs may increase the predictive utility of the model, they do not address the TPB's focus on a single GDB segregated from other concurrently pursued GDBs.
Clinical practice often involves health professionals performing numerous GDBs, each competing for limited resources in patient consultations, in particular time-related resources [
13]. GDBs might conflict with (
i.e., hinder) pursuing a particular GDB while other GDBs might create opportunities and be perceived to facilitate (
i.e., help). Assessing perceptions about how conflicting and facilitating GDBs influence a focal GDB provides a way of accounting for the influence of the wider context of multiple goal pursuit which often characterises clinical practice. General medical practitioners perceive many of their GDBs as facilitating and conflicting with guideline-recommended GDBs such as prescribing to reduce blood pressure and providing PA-related advice. For example, GPs have reported that addressing the patient's agenda, treating acute illnesses, and prescribing to reduce cholesterol are among the GDBs perceived to conflict with giving PA advice [
13]. Furthermore, taking a patient's history, addressing alcohol consumption and smoking, checking body mass index, and addressing well-being and stress are perceived by GPs to facilitate giving PA advice [
13]. Thus, not only do health professionals engage in numerous behaviours, but many of these are also perceived as facilitating or conflicting.
It is not clear whether goal facilitation or goal conflict actually predict health professionals' behaviour beyond the predictive efficacy of leading social cognition models such as the TPB. However, evidence from other populations supports the potential of goal conflict and goal facilitation as predictors of health professional behaviour.
In other professional contexts, both goal facilitation and goal conflict have been shown to be associated with behaviour. In a management setting, goal conflict was negatively associated (medium effect size [
14]) with attainment of a novel self-set goal four months later [
15]. However, goal conflict was assessed on a bipolar scale ranging from instrumental (negative values) to conflicting (positive values) and the observed mean of 'goal conflict' was negative and within a range that would be considered as goal facilitation. The observed association may be more appropriately characterised as evidence of the relationship between goal facilitation and behaviour. In an academic context, university professors' conflict between teaching and research negatively predicted their research performance [
16]. In a context of medical equipment sales, goal conflict was negatively associated with commitment and self-efficacy (conceptually similar constructs to intention and PBC in the TPB), and performance [
17].
The relationship between goal facilitation and conflict and behaviour has also been investigated to further understand preventive health behaviour, such as participation in PA. Prospective studies predicting engagement in PA have demonstrated that goal facilitation, but not goal conflict, predicts PA beyond TPB constructs [
18‐
20].
Goal conflict may be more readily perceived and predictive of behaviour when the conflicting GDBs under consideration are pursued within the same context as a focal GDB. Focusing on goal conflict perceived within a resource-constrained clinical setting may be a more appropriate test of the predictive utility of this goal construct. As such, the present study was interested in conflict and facilitation between a health professional's GDBs. We aimed to explore the predictive utility of goal facilitation and goal conflict in a health professional context. We hypothesised that goal facilitation and goal conflict would predict health professional behaviour over and above intention and PBC.
Methods
Participants
To our knowledge, the present study was the first to test goal conflict and goal facilitation as predictors of health professional behaviour in primary care. There was little existing evidence upon which to estimate the effect sizes for a formal power calculation, and thus this study was considered to be exploratory. We sent questionnaires to a random sample of health professionals from all 84 GP practices in NHS Grampian and all 69 practices in NHS Tayside, Scotland at baseline, targeting a final sample size of at least 157 health professionals. We estimated a 40% response at baseline and a 65% response at follow-up. Baseline questionnaires were sent to 606 health professionals (453 general practitioners, or GPs, and 153 nurses).
Measures and data collection procedures
The focal goal-directed behaviour of interest in the current study concerned providing PA advice, a guideline-recommended behaviour [
3]. Patients with hypertension have an elevated risk of cardiovascular disease, and increased PA is associated with a reduction in blood pressure [
21]. The focal behaviour was specified as giving patients with an existing diagnosis of uncomplicated hypertension lifestyle advice for increasing their PA.
At baseline in March 2009, participants were sent a four-page postal questionnaire along with an invitation letter, an information sheet, an informed consent sheet, and a freepost return envelope. An identical follow-up questionnaire was sent to baseline respondents six months later, in October 2009 along with an invitation letter and follow-up reminders to non-respondents. This length of follow-up is consistent with previous research testing goal conflict and goal facilitation in other settings [
18,
19] and tests of the TPB in this population [
22,
23].
Theory of planned behaviour
TPB constructs were measured at baseline using single items (to maximise response rates) in a single block prefaced with 'Please rate the following statements based on the following action: In the next two weeks, personally giving lifestyle advice for increasing physical activity to your patients with an existing diagnosis of uncomplicated hypertension.' Intention was measured with one item: 'I intend to do this' (1-strongly disagree to 7-strongly agree). PBC was measured with one item using a semantic differential scale: 'For me, doing this is...' (1-very difficult to 7- very easy). Attitude was also measured on a single semantic differential scale: 'For me to do this is...' (1-very bad practice to 7- very good practice). Subjective norm was assessed using one item: 'People whose opinion I value expect me to do this' (1-strongly disagree to 7-strongly agree).
Goal facilitation and goal conflict
Measures for goal facilitation and goal conflict were adapted from existing scales [
18,
19] into two single items (to maximise response rates) and assessed at baseline. For goal facilitation, participants were asked to rate their agreement with the statement 'During these consultations, other things I do helpfully lead me to give lifestyle advice for increasing physical activity' on a scale ranging from 1-strongly disagree to 7-strongly agree. To measure goal conflict, participants were asked to indicate their agreement with the statement 'During these consultations, other things I do lead me to spend less time giving lifestyle advice for increasing physical activity' on a scale ranging from 1-strongly disagree to 7-strongly agree. Factor analytic and predictive evidence has shown that goal conflict and goal facilitation are best considered as independent constructs, and were therefore measured separately [
18,
19].
Demographics
Participants were asked a series of demographic questions to assess their age, sex, graduation year, employment status (full-time or part-time) and role (GP or practice nurse).
Behaviour
The behavioural outcome measure was administered at follow-up and consisted of two items. The first item asked participants 'How many patients with an existing diagnosis of uncomplicated hypertension have you personally seen in the past two weeks?' The second item asked 'and of those, for how many did you give lifestyle advice for increasing physical activity?' The outcome measure was computed as the proportion of patients to whom advice was provided, out of the patients with existing uncomplicated hypertension seen in the past two weeks.
Ethics approval
Ethical approval for the current study was obtained from the North of Scotland Research Ethics Committee (REC No. 09/S0801/4).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
This study was conceived by JP, JJF, NCC, and FFS. The study was run by JP. Data handling and analyses were conducted by JP. JP led the writing of this paper and all authors commented on drafts and approved the final version.