Study design

This study is qualitative and draws on deductive and inductive analysis. Cross-sectional online focus groups were designed and delivered using the Nominal Group Technique (NGT) method to identify and establish a consensus on the factors that influence HCPs’ choice of medication for patients with PD. The NGT method was chosen because it is a highly structured approach that offers a discursive and democratic method of collecting data and creates a collaborative balance among participants.15 In comparison to other qualitative research techniques (eg, in-depth interviews), NGT diminishes the potential for facilitator bias and encourages participants to occupy an active, democratically led role. Originally developed to assist in healthcare planning,16 the method lends itself particularly well to healthcare research and has been used to develop a framework for care coordination,17 explore stakeholder views on hypertension medication adherence,18 manage Alzheimer’s disease (using a modified version of the NGT)19 and identify research priorities for PD management.20 The benefit of the NGT approach is that it is possible to generate ideas, problem solve and establish a consensus that will identify key priorities for a given topic in a timely manner.20

Notably, the NGT allows a thematic structure to emerge organically within each group, without imposing any preconceived analytic framework by the facilitator.21 The approach taken aligns closely with a critical realist perspective.22 Central to critical realism are principles asserting the existence of a multilayered real world shaped by underlying causal mechanisms. These mechanisms generate phenomena, which are then experienced by individuals (eg, how do individual clinicians interpret guidelines to make prescription decisions?). However, as these mechanisms are not directly observable due to the complex nature of reality, they are inferred through exploring how people construct and attribute meaning to their experiences of the phenomena (eg, how clinicians use guidelines to make prescription decisions).23

In the context of prescribing medication for PD, understanding the underlying mechanisms that drive prescribing behaviours becomes crucial. While mechanisms possess the potential to produce phenomena, their causal efficiency is contingent on the contextual conditions within which they operate. This notion of an ‘open system’ acknowledges that the effectiveness of a mechanism can be influenced by other coexisting mechanisms within a given context.21

Employing a ‘group-by-group’ critical realist analytical approach allows for the exploration of variation within and between responses, ultimately leading to the identification of themes across the focus groups. The unique strength of this approach is that it can facilitate the between-group comparison of key responses. These themes can serve as valuable indicators of the similarities and differences in how HCPs construct their experiences of prescribing, shedding light on the nuanced factors shaping prescribing behaviours within the complex landscape of PD management.

Participants and recruitment

Participants were opportunistically recruited and initiated using an online poster disseminated via neurodegenerative disease societies and social network platforms. Experts in the PD field were individually identified using relevant conference/society programmes and websites and, where possible, contacted via email and invited to take part. Snowball sampling was also employed to support further participant recruitment. Participation in the study was open to HCPs from across Europe who spoke English, were currently employed as medical professionals (eg, consultant, specialist nurse, neurologist) and had prescribed medication for PD in the past 2 years (2021–2023). Prospective participants were excluded if they were retired or did not hold a licence to prescribe medication. The recruitment process ran from March 2023 and ended once 18 HCPs had enrolled (and seven groups had been recruited), which is consistent with the critical realist assumptions underpinning the study22 and also supported by previous research, which indicates that 80% of data saturation can be achieved with two to three groups.23 One participant who agreed to take part left the study before the discussion began, as they were no longer involved in prescribing medication. All participants provided digitally informed consent, thereby indicating that they understood their participation was voluntary and confidential and that they were free to withdraw at any point without having to provide a reason. To thank participants for their time, they were remunerated with a £/€20 online shopping voucher once the focus group meetings were complete. This study received full ethical approval from the Aston University College of Health and Life Sciences Research Ethics Committee, Birmingham, UK, in addition to the NHS Health Research Authority and Care Research Wales Research Ethics Committee (24/HRA/0792).

Data collection

To ensure the authenticity and depth of insights gathered from the participants, each focus group was facilitated by the lead author (EB), who had no prior contact with the participants, thus minimising the potential for social desirability artefacts to influence the data collection process. Participants had no prior knowledge of the research or of the researcher and only understood the aims of doing the research by reading the participant information sheet. Each of the seven focus groups lasted 1 hour and took place online using Microsoft Teams, with each participant attending only one meeting. Hosting the meetings virtually enabled wider data collection from participants taking part across Europe.24 The video of each meeting was digitally recorded for sense-checking purposes but was not transcribed because the focus group consensus of priority factors was the only relevant output data; no participant quotations were used. As well as EB, an assistant was present at the meetings to create field notes that would support later interpretation of the context of the factors discussed.

Nominal group technique (NGT)

Each focus group commenced with a scene setting (10 min—welcome, aims, purpose, research interest and procedural details) followed by the presentation of the nominal prompt, ‘What factors influence your choice of prescription treatment for Parkinson’s disease?’ to guide subsequent discussion (the prompt was not piloted, nor were participants shown it before taking part). A silent generation phase (5 min) allowed participants to independently scribe responses to the prompt. Each participant then shared their ideas during a round-robin discussion (10 min) facilitated by EB, who compiled a comprehensive list and shared it on screen for participants. During the clarification phase (10 min), participants were encouraged to refine, merge or eliminate duplicate ideas to streamline the list for further assessment. In the subsequent ranking stage (15 min), participants individually ranked their top five factors in order of importance (1=most important; 5=least important) before they were collectively discussed and grouped by the facilitator. Finally, the ranked factors were revealed, allowing participants to collectively agree on the group ranking and address any disparities through discussion or voting for amendments (10 min).

Data analysis

The NGT approach enables data analysis to take place during the focus group. Data were interpreted qualitatively by allowing a subordinate list of themes to emerge at the round-robin stage of each group and later refined and merged where possible in the clarification stage.25 Using these lists, each group agreed on the superordinate ranked factors by voting on their importance. The superordinate ranked factors that emerged from group consensus were then deductively coded to the behavioural domains of the TDF and Capability, Opportunity, Motivation–Behaviour (COM-B) model to identify the behavioural determinants underpinning these factors. This deductive stage involved an iterative process of agreement between authors EB, JT and CS, who coded a sample of the NGT output until the superordinate themes could not be reduced further on to the behavioural domains of the TDF and COM-B models. After meeting regularly to discuss mapping priorities, EB mapped the remaining output into the TDF domains, a process that was iteratively revised with regular discussions with CS. This multistage deductive approach to mapping the behaviours from the focus groups to the specific TDF domains followed the existing work in the field.26 A codebook defining each of the TDF domains and how each factor is mapped onto them is provided in online supplemental file 1.

Research team and reflexivity

Dr Emma Begley PhD is a qualified female researcher with 11 years of experience in public health and behavioural change. At the time of the study, EB was employed as a Knowledge Transfer Partnership Research Associate at Aston University, Birmingham, UK. Dr Jason Thomas and Dr Carl Senior are behavioural scientists with 30 years of experience between them in applying behavioural change techniques in areas such as healthy eating. Any prior knowledge about PD was established during a literature review for EB, JT and CS. William Hind is the Founder and CEO of Alpharmaxim.

Patient and public involvement

No patients or members of the public were involved in designing or carrying out this research.

Participant characteristics

Across the seven focus groups, a total of 18 participants between 25 and ≥65 years of age took part (female n=10; male n=8). Each focus group comprised two or three participants. Participants’ experience of working in PD ranged from two to ≥20 years as neurologists, consultants, specialist nurses or specialist practitioners (n=5, n=3, n=8 and n=1, respectively; data unavailable for one participant). Most participants were recruited from the UK (n=11); other participants were recruited from Austria, France, Ireland, Italy, Netherlands, Sweden and Switzerland (n=1, from each county).

Silent generation of factors and round robin (NGT stages 2 and 3)

The raw data that emerged from the silent generation and round-robin stage yielded 60 unique factors that participants thought influenced their prescribing decisions; individual focus groups reported a range of 22–41 factors. There was a good consensus of factors that emerged across the focus groups, and data saturation was reached by the seventh focus group (FG7).

All seven focus groups consistently discussed patient age, HCP and/or patient experience of using a medication, symptoms (including severity and burden) and suitability of a medication (eg, ease of use of medication, route of administration) as important factors. When making prescribing decisions, participants also mentioned how they considered reducing pill burden and simplifying a drug regimen for patients, thereby making it more suitable and acceptable. Other factors commonly reported included consideration of patient cognition (eg, level of education and literacy), complications of comorbidities (eg, hypotension or high blood pressure), side effect profiles, availability of drugs on practice formularies (although participants explained this was not an issue across Austria, Germany or Switzerland), patient quality of life (QoL; eg, daily activity level and ability to still work) and support at home to administer and manage medication.

Drug cost was discussed by all groups, but only five included it in their ranked priorities. There was, however, some discrepancy between participants about whether they considered the cost to be a swaying factor; one participant each from the UK (FG3), France (FG4) and Italy (FG7) expressed that cost was not an influencing factor as efficacy was valued more. There were also inconsistencies regarding the perceived importance of guidelines. Although five focus groups ranked guidelines as being important, either as a factor on its own or as part of a broader theme (eg, external factors, practicalities or drug factors), a few participants (one from FG3 and two from FG5) recalled that they only use them to some extent and preferred to refer to published evidence. Other factors occasionally reported related to polypharmacy, patient and HCP treatment expectations and desired outcomes, patient preference, HCP authority to prescribe, drug efficacy, multidisciplinary team support, psychological health, trials and evidence, capacity to monitor treatment, patient frailty, risk of impulse control disorders and type of PD.

Clarification and ranking (NGT stages 4 and 5)

Participants ranked between 3 and 8 priority factors; differences in the number of priorities between focus groups occurred due to participant preferences to specify individual factors rather than collapse them into broader themes. As such, the order of rankings is broadly covered (see table 1 for a full overview):

• Medical or symptom concern included factors such as the stage of disease, symptom concern or severity and was discussed by six groups, and although four of these groups categorised this factor into a broader theme (disease prognosis, patient characteristics or clinical factors), it was generally ranked a first priority.

• Patient characteristics included factors such as patient age, QoL or frailty and was a preferred theme across the groups (five out of seven groups) and was often ranked as a first or second priority, although FG5 ranked it fourth. The two groups that did not include patient characteristics as a superordinate theme decided either to prioritise and individually rank patient age, lifestyle and comorbidity (FG4) or included patient characteristics within a highly ranked ‘clinical factors’ theme (FG7). Despite this variance, the majority of focus groups agreed that factors related to patient characteristics are of high importance when making prescribing decisions.

• Side effect profile was discussed by six groups and ranked either as a category by itself or within a broader theme such as response to treatment, practicalities or clinical factors. Three groups ranked this theme as being the third most important. In FG6, it was mentioned that certain medications may be prescribed to manage side effects, but they did not include a side effect profile as one of their ranked themes.

• Access to treatment, including availability of treatment, was selected as a priority by four groups, but there was no consistency in its ranking. For example, FG1 ranked this least important (sixth), while FG2, 3 and 6 categorised it into broader themes that were ranked at first, third and fourth priority, respectively. Although the positioning of importance is indifferent between groups, access to treatment was still an important consideration when making medication choices.

• Clinical guidelines, including knowledge and practice of, was discussed by six groups, but there was no consensus on ranking between groups. The groups that prioritised guidelines (or included it in broader themes) ranked them first (FG1 and FG2), third (FG3 and FG7) or fourth (FG6). FG5 discussed guidelines but did not prioritise them (see online supplemental file 2), while FG4 did not discuss guidelines at all.

• Social support, including carer support at home to manage treatment, was a priority for five groups; three of these groups categorised it into broader themes (eg, patient characteristics [FG1 and FG2], external factors [FG2]). There was no consensus on its ranking, but participants did discuss the importance of having support at home to manage certain treatment.

• Patient preference was a priority for five groups: two groups ranked it as a fifth priority, while the other three groups included it within broader themes that were ranked first (FG3) or second (FG1 and 7). Generally, participants discussed the relevance of understanding what patients want and listening to their requests when they have done their own research about the available treatment.

Other factors that were prioritised and ranked but could not be easily grouped into the above themes included interpersonal relationships, such as multidisciplinary support (ranked fifth by FG1); efficacy and safety of a drug (ranked second by FG5, while three other focus groups included it within broader themes); treatment needs, such as patient expectation and symptom treatment (ranked fourth by FG7); and cost (ranked fifth by FG7, and included within broader themes for four other groups; however, some participants did not perceive cost as a swaying factor).

Theoretical mapping of priorities

Mapping priorities to the TDF provides a structured method to uncover the various behavioural determinants that may influence prescribing outcomes. This approach can help tailor evidence-based strategies to the unique context of that behaviour, increasing the likelihood of successful and sustainable change. To achieve this, the factors discussed and prioritised by participants aligned with seven of the 14 TDF domains, which subsequently mapped onto four of the COM-B subdomains (see figure 1 below for a mapping overview). Prioritised factors from the focus groups aligned with the following two TDF domains: knowledge and memory, attention and decision processes, which themselves map onto the COM-B capability domain (specifically, the psychological subdomain). Most of the factors mapped onto these domains were linked to HCP decision-making processes and were informed by patient and clinical characteristics: factors that participants ranked as being the most important. In terms of knowledge, participants discussed that their understanding of the efficacy and safety of a drug, including the interaction of other medication, influenced their prescribing (see description of factors in online supplemental file 2).

Another suite of prioritised factors aligned with the following three TDF domains: belief about consequences, goals and social/professional role and identity, which map onto the COM-B motivation domain (specifically, the reflective subdomain). Participants often shared concerns about the consequences of a medication causing side effects, such as reduced impulsive control and the potential of medication to negatively impact patient QoL, or reduce or impair treatment outcomes. As such, participants discussed how it was important for them to consider patient preferences and identify treatment needs, subsequently informing treatment goals. Participants also felt that HCP experience and familiarity with prescribing a medication would influence future prescribing decisions (see online supplemental file 2); this was particularly evident among specialist nurses who explained how they often discuss prescribing options with senior consultants or neurologists.

Finally, the remaining prioritised factors from the focus groups aligned with the following two TDF domains: environment context and resources, and social influences, which map onto the COM-B opportunity domain (the physical and social subdomains, respectively). Factors within these domains reflected influences arising from the participants’ physical environment (eg, medication cost or availability) and resources (eg, guidelines or access to medication) that somewhat determined prescribing decisions. Additionally, a range of social influences, such as the interpersonal relationships that arise from shared decision-making with patients or HCP peers and support at home to help patients manage medication and a general prescribing culture, was reported by participants.

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Aston University College of Health and Life Sciences Research Ethics Committee, Birmingham, UK, in addition to the NHS Health Research Authority and Care Research Wales Research Ethics Committee (24/HRA/0792). Participants gave informed consent to participate in the study before taking part.