Introduction

Neurodegenerative diseases (NDs), such as Parkinson’s disease (PD), are a leading cause of disability and death worldwide [1]. Year on year, the incidence and prevalence rates for NDs have increased, and there is growing concern that the burden of these diseases will double across the globe by 2043 [2–4]. In the absence of a cure, various medications are used to manage the symptoms of NDs [5]; however, the variability of patient symptoms, health priorities and drug effectiveness are challenging obstacles that healthcare professionals (HCPs) often face when determining the most appropriate medication to prescribe [6,7]. Hence, there is a need to understand the role of the contextual influences at play when HCPs make medication choices, so that efforts to optimise prescribing can be established.

Much is already understood about factors that influence prescribing choices. For example, a recent systematic review identified patient age and type of prescriber (e.g., general practitioner or neurologist) as key determinants for prescribing medication to patients with PD [8]. More broadly, a qualitative study reported that factors such as the severity of patients’ health concerns, physicians’ preferred approaches, and availability of resources influenced clinical decision making for patients with complex multimorbidity [9]. The psychology of decision making has also been explored, with some evidence indicating that various heuristic strategies lead to cognitive biases (e.g., availability or outcome bias) that influence HCP judgement [10]. The factors reported in these studies add valuable knowledge about prescribing; however, little is known about which factors are most important or how they can be used to optimise prescribing behaviour.

Enhancing treatment choice requires an in-depth understanding of prescribing behaviour to identify what needs to change and how to effectively bring about the desired change. The behaviour change wheel (BCW) is a leading theoretical framework that guides users through the identification of barriers and determinants that underpin an individual’s Capability, Opportunity and Motivation to perform a Behaviour (COM-B) [11]. This enables users to subsequently identify the most effective approaches (e.g., education or training) and behaviour change technique (BCT; e.g., goal setting or use of a creditable source) to change the targeted determinant(s) [12]. The assessment of behavioural determinants can also be explored by using the Theoretical Domains Framework (TDF), which comprises 14 behavioural determinants that map onto behaviour change components of the COM-B model [13]. The advantage of using both the TDF and COM-B model together is that they provide a greater level of understanding of the theoretical constructs underpinning a behaviour. For example, a review investigating interventions that optimise prescribing reported that 1) knowledge and 2) memory, attention and decision making were two of the most common TDF behavioural determinants identified [14]. Identifying these TDF determinants allowed the authors of the review to use the BCW manual and its associated BCT taxonomy to retrospectively identify which BCTs (e.g., prompts and defaults) could be used to influence a change in behaviour.

There is a clear research gap to understand more about the decision-making drivers behind prescribing. This is a fundamental element of medical practice, and an area where HCPs report difficulty and discomfort when making clinical decisions [15,16]. Exploring and better understanding the factors that influence prescribing for patients with NDs offers an opportunity to improve the prescribing pathway by supporting HCPs to optimise their treatment choices. Pharmacotherapy is the dominant treatment regimen for NDs [17]. For patients with PD specifically, the use of the dopaminergic precursor levodopa as first-line monotherapy is most prescribed [18]. Some studies have indicated that levodopa therapy is prescribed in up to 70% of cases in the USA [19]. The treatment of Alzheimer’s disease is not dominated to the same degree as that of PD but cholinesterase inhibitors, such as donepezil, still account for 60% of the first-line treatments in the USA [20].

This study aimed to identify the factors that predict HCP prescribing decisions in order to demonstrate the potential to optimise treatment choice. It is hypothesised that the factors which influence prescribing choices, and the behavioural determinants that they map to, will vary depending on the clinical presentation of a patient.

Methods

Prescribing scenarios

Two situational prescribing scenarios that describe patients with newly diagnosed and advanced stage ND were used to capture outcome measures that reflect a range of potential real-world prescribing practices (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Prescribing scenarios 1 and 2.

https://doi.org/10.1371/journal.pone.0353479.t001

Results

Regression results for the long-form 93-item questionnaire

For scenario 1 (patient with newly diagnosed ND), the stepwise regression generated a statistically significant model comprising three predictors. Intentions and behavioural regulation both positively predicted the likelihood of an HCP prescribing new, novel or different to usual medication (β = 0.32, p=<0.001, 95% confidence interval (CI) 4.75, 10.70; β = 0.19, p = 0.005, 95% CI 1.57, 8.92). This indicates that HCPs with greater intention (i.e., a conscious readiness to prescribe) and greater behavioural regulation (i.e., an ability to monitor and control one’s prescribing behaviour) are more likely to prescribe new, novel or different to usual medication. Conversely, social/professional role and identity negatively predicted the likelihood of prescribing new, novel or different to usual medication (β = ‑0.184, p = 0.002, 95% CI −11.50, −2.54). This indicates that HCPs with greater role responsibility and self-identity (i.e., perceiving prescribing to be an important part of their role and identifying as a prescribing HCP) are less likely to prescribe new, novel or different to usual medication (see Fig 1). Overall, this model explained 18.4% of the variance in how often participants would prescribe new, novel or different to usual medication to a newly diagnosed patient with a ND.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Main predictors of prescribing behaviour for neurodegenerative disease.

Horizontal bars show the standardised direction and magnitude of the main predictors identified in the regression models. Positive values indicate predictors associated with greater likelihood of prescribing a new, novel, or different medication, whereas negative values indicate predictors associated with lower likelihood of prescribing.

https://doi.org/10.1371/journal.pone.0353479.g001

For scenario 2 (patient with advanced stage ND), the stepwise regression model generated a statistically significant model comprising two predictor variables. Intention positively predicted the likelihood of prescribing new, novel or different to usual medication (β = 0.23, p < 0.001, 95% CI 2.73, 8.11); thus, as above, HCPs with greater intention are more likely to prescribe new, novel or different to usual medication. Conversely, belief about consequences negatively predicted likelihood of prescribing new, novel or different to usual medication (β = −0.145, p = 0.014, 95% CI −6.79, −0.75). This indicates that HCPs who are more concerned about the consequences of prescribing out of habit (i.e., who believe there is more potential risk than benefit) are less likely to prescribe a new, novel or different to usual medication. Overall, this model explained 6.4% of the variance in how often participants would prescribe new, novel or different to usual medication to a patient with advanced stage ND.

Further data from the stepwise regression models are presented in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Regression results for the long-form 93-item questionnaire.

https://doi.org/10.1371/journal.pone.0353479.t002

Discussion

The purpose of this study was to explore the likelihood of HCPs prescribing a new, novel or different medication for NDs and to map the strongest predictors of prescribing to behaviour change determinants aligned to the TDF and COM-B theoretical frameworks. Overall, when positioned with a hypothetical choice of prescribing a medication different to routine choice, HCPs were more likely to consider prescribing a new, novel or different medication to patients with advanced stage ND (scenario 2) than to patients with newly diagnosed ND (scenario 1). This finding complements the wider literature, which highlights that patient stage of disease is a key factor that influences prescribing choice, [32–34] and that HCPs are more likely to adhere to guideline recommendations on using first-line therapy [35] (i.e., less likely to deviate from routine prescribing for newly diagnosed patients). Ensuring that HCPs are aware of up-to-date guidelines and have access to them is clearly an important gateway to influencing future prescribing changes.

Further, the regression model identified that the behavioural determinants most likely to predict a change in prescribing differ depending on whether HCPs treat newly diagnosed patients or those with advanced stage ND, which supports the proposed hypothesis (see Table 3). Specifically, the data indicated that the behavioural determinant ‘intentions (to prescribe)’ may be a more generic target for changing prescribing behaviour as it was a significant predictor for both new and advance stage patient scenarios, whereas behavioural regulation, social/professional role and identity, and belief about consequences may represent more selective targets relevant in context-dependent scenarios. Collectively, these predictors are hypothetically lucrative targets for an intervention that aims to optimise medication choice in NDs; however, further study is required to test their impact on change.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Mean values for each of the domains and outcome measures.

https://doi.org/10.1371/journal.pone.0353479.t003

Perceiving the significant predictive factors of this study through a behaviour change lens indicated that HCPs’ psychological capability (led by their behavioural regulation, e.g., how they self-monitor or objectively measure their actions) and reflective motivation (led by their intention, e.g., making conscious decisions) are important positive influences of prescribing. At the same time, other aspects of reflective motivation (led by HCPs’ social/professional role, e.g., how they perceive themselves and their role, and their belief about consequences, e.g., outcome expectancies) may also be important predictors; however, these should be carefully considered as they may negatively predict HCPs’ likelihood of prescribing outside of habit.

In relation to the existing literature, use of behaviour change theory or frameworks to improve prescribing practice is growing. The methods used vary and most target antibiotic or inappropriate prescribing [14,36,37], thus making it difficult to compare and identify causal relationships between the determinants and behaviour change in NDs. That said, behavioural regulation was found to be important (and, therefore, a target behaviour) for improving polypharmacy prescribing in older adults [38] and partly explained a positive relationship in antimicrobial stewardship behaviour [30]. Similarly, behavioural regulation and intention were also identified as important determinants in influencing prescribing nurses to prescribe more medication in community care settings [39]. Conversely, behavioural regulation and intention were not reported by a recent systematic review of interventions designed to optimise prescribing [14]. Rather, the review by Talat et al. 2022 [14] identified two interventions that targeted social/professional role and identity, and belief about consequences: determinants this current study also found as negative predictors of prescribing. Nevertheless, there is other evidence that HCP belief about consequences is often an important target of changing prescribing [38–40]. Differences in study findings may be due to the often-missing information about how interventions identify and report prescribing determinants, thus requiring a level of interpretation from the review authors [14]. Although this report found differences in the predictive direction of the determinants identified (two negative and two positive predictors), they still comprise part of COM-B’s reflective motivation, which is theorised to play a key part in habit formation [41], the role of which has been shown to have a positive effect on HCP behaviour [42]. The inconsistencies in the targets of change and their importance highlights the need to test the effectiveness of changing prescribing behaviour for the behavioural determinants reported here.

Understanding prescribing practices in this way fulfils the first stage of the BCW framework [12] and helps to identify relevant theoretically driven intervention content and the circumstances in which there may be a greater likelihood of change [43]. The potential of this approach is highlighted in a recent systematic review, which identified 16 interventions that effectively optimised medication prescribing and, using the TDF (which is compatible with the BCW’s COM-B model), identified relevant behavioural determinants and BCTs that the interventions used to trigger change [14]. Unlike the interventions described in the review by Talat et al. 2022 [14], this current study clearly highlights how the behavioural determinants influencing prescribing for NDs were identified. Hence, this work contributes new knowledge specific to the ND field and, by identifying relevant intervention content using the subsequent stages outlined in the BCW framework [12], provides an opportunity to explore the mechanisms of action [44] potentially linked to changing prescribing practice.

This study has several strengths, one of which is that a significantly powered sample of participants was recruited from across Europe, thereby providing a wider perspective on prescribing for NDs. In addition, the data captured enabled the statistical quantification of predictors. Other strengths include the development of a robust survey, which was informed by existing literature, a primary (related) research study, and underpinned by a behaviour change theoretical framework. However, using a newly developed survey poses certain limitations. For example, the survey requires further validation to ensure its robustness. Designing a situational scenario is also another limitation as it does not capture actual prescribing practice and is not sufficient to indicate causality. In addition, the design of the present study ensured that core prescribing competencies were addressed comprehensively and without redundancy across the two situational scenarios. Future research in this area should examine a broader range of prescribing competencies through the inclusion of additional scenarios.

That said, the results are enough to guide the development of future experimental studies to explore the change in prescribing patterns when certain behavioural determinants are targeted. While the aim of the current study was to examine the facilitators of broad behaviour change, future research should investigate whether the findings reported here can also be observed when focusing on specific classes of drugs or even a comparison of different disease states.

Future research should also consider replicating this study longitudinally (for example, pre-/post-launch of a new medication) to explore whether the behavioural determinants identified here are malleable over time and result in prescribing change, a sub group analysis across different countries or job seniority, or if other factors such as sensitivity of missing data also need addressing (see, e.g., [45]).

In conclusion, this study identified a set of generic and more selective behavioural predictors that may be useful targets to help optimise prescribing for NDs in the future. This insight is valuable to those developing behaviour change interventions as it can be used to identify relevant intervention content and offers an opportunity to examine whether there is a causal relationship between the behavioural determinants and outcome changes. The findings of the current study have clear implications for the clinical management of NDs. A better understanding of the factors that influence prescription decisions can help identify more effective pathways to pharmacotherapy. This, in turn, is likely to reduce the healthcare burden and enhance patients’ quality of life.

Supporting information

Acknowledgments

Dr Catrina Milgate from Alpharmaxim contributed to reviewing and editing draft versions of the manuscript; she also oversaw the supervision of resource support from professional Medical Writers and Editorial Assistants who conducted data checks and formatted the manuscript.