Stroke secondary prevention, a non-surgical and non-pharmacological consensus definition: results of a Delphi study

Objective Evidence supporting lifestyle modification in vascular risk reduction is limited, drawn largely from primary prevention studies. To advance the evidence base for non-pharmacological and non-surgical stroke secondary prevention (SSP), empirical research is needed, informed by a consensus-derived definition of SSP. To date, no such definition has been published. We used Delphi methods to generate an evidence-based definition of non-pharmacological and non-surgical SSP. Results The 16 participants were members of INSsPiRE (International Network of Stroke Secondary Prevention Researchers), a multidisciplinary group of trialists, academics and clinicians. The Elicitation stage identified 49 key elements, grouped into 3 overarching domains: Risk factors, Education, and Theory before being subjected to iterative stages of elicitation, ranking, discussion, and anonymous voting. In the Action stage, following an experience-based engagement with key stakeholders, a consensus-derived definition, complementing current pharmacological and surgical SSP pathways, was finalised: Non-pharmacological and non-surgical stroke secondary prevention supports and improves long-term health and well-being in everyday life and reduces the risk of another stroke, by drawing from a spectrum of theoretically informed interventions and educational strategies. Interventions to self-manage modifiable lifestyle risk factors are contextualized and individualized to the capacities, needs, and personally meaningful priorities of individuals with stroke and their families.


Introduction
International best practice guidelines for stroke secondary prevention (SSP), while aetiology dependent, generally include medication prescription (anti-hypertensive, lipid lowering, anti-platelet/coagulant); high level evidence supports this recommendation [1,2]. Conversely, recommendations for lifestyle modifications have lower levels of evidence, largely drawn from primary prevention studies, and as a results some population-attributable stroke risk factors (e.g. psychosocial stress) [3] are inadequately addressed [1,2].
Conclusive evidence is lacking on how best to support stroke survivors to engage in risk reducing behaviours. In recent systematic reviews of complex interventions in SSP, meta-analysis was possible for limited outcomes due to primary study heterogeneity across key definitions, population and intervention characteristics, outcomes and associated measures [4,5]. To advance the evidence-base, empirical research is needed, informed Open Access BMC Research Notes *Correspondence: maggie.lawrence@gcu.ac.uk 1 Department of Nursing and Community Health, School of Health and Life Sciences, Glasgow Caledonian University, Glasgow G4 0BA, UK Full list of author information is available at the end of the article by a consensus-derived definition of non-pharmacological and non-surgical SSP and an agreed core set of outcomes. No published consensus on these foundational tools exists.
In 2016, INSsPiRE (International Network of SSP Researchers), comprising secondary prevention trialists identified by ML's reviews [4,6] agreed a programme of work, focusing initially on a consensus-derived definition of secondary prevention, beyond pharmacological and surgical interventions, to inform research standards, facilitate data synthesis, guideline development and service delivery.

Methods
Delphi technique: a structured, iterative process that pools knowledge and understanding from a range of experts to arrive at an agreed standpoint on an issue [7]. Data can be gathered using electronic, internet-mediated tools; ideal where geographical barriers exist. Data were collected and synthesized in seven stages [7], using both online and in-person modes (Fig. 1).
Stages 1 facilitator assignment and 2 participant identification: ML facilitated stages 1-4. Stage 6 was facilitated by BD, and stage 7 by AHP; independent, experienced researchers. Participants were identified as the contemporaneous members of INSsPiRE i.e. published researchers, academics, and clinicians working in the specialty of non-pharmacological and non-surgical SSP. Stage 3: Problem definition: The problem was defined as the lack of an evidence-based, consensual definition of non-pharmacological and non-surgical SSP. Stage 4: Elicitation of definitions and key elements of SSP: Participants were sent an elicitation survey by URL link. Comprising two parts, the survey asked participants to:

Provide examples of published definitions of non-
pharmacological and non-surgical SSP from systematic reviews, research studies, clinical guidelines, or third sector websites, including source reference(s). 2. List key intervention elements and/or underpinning concepts and theories considered essential to non-pharmacological and non-surgical SSP.
Stage 5: Ranking: Participants received a link to a survey where they ranked, without consultation with other participants, each key element and concept previously identified, on a 10-point Likert scale: Not important (1-3), Important but not critical (4-6), Critical (7-9); Unable to score (10). Stage 6: Ranking revision: This iterative stage used online and in-person modes. Participants accessed the Stage 6 survey, developed by augmenting the stage 5 survey with the item-by-item response data (i.e. voting frequency). Participants reviewed overall response rates and considered whether to revise their original ranking. Another round of anonymous ranking followed group-based discussions. Consensus was defined as ≥ 70% ranking agreement [8]. Stage 7: Action: In small groups participants worked (phone/Skype/in-person) to draft definitions. At a subsequent in-person meeting, following experiencebased stakeholder engagement, participants reviewed Stage 6 results and draft definitions, and discussed and agreed a definition.
Ethical approval was received from Glasgow Caledonian University's (GCU) School of Health and Life Sciences Ethics Committee (HLS/NCH/16/020). Table 1 identifies INSsPiRE members who participated in ≥ 1 Delphi process stage.

Results
In Stage 4 (Elicitation), 14(88%) participants identified 26 unique definitions of SSP and 85 unique 'key elements' . XW collated the definitions, removed duplicates, and shared the resultant list with participants. ML collated 'key elements' , removed duplicates, and categorised elements into eight domains: Modifiable lifestyle risk factors, Modifiable physiological risk factors, Education about stroke, Education about modifiable risk factors, Education about managing other lifestyle issues, Education about managing psychosocial factors, Skills education/training, and Underpinning theories and approaches. The domains were collapsed into three overarching domains: Risk factors, Education, and Theory, and used to structure the Stage 5 survey. In Stage 5 14(88%) participants ranked each element using the Likert scale (above).
In Stage 6 (Ranking revision) online consensus (n = 14, 88%) was achieved to include 24 of the 49 elements. At a subsequent, independently facilitated in-person meeting (n = 14, 88%) at GCU in June 2017, discussion and debate was followed by anonymous voting on the remaining 25 elements. Three elements were merged with others, one was removed, and consensus to include 14 and exclude three further elements achieved. Four outstanding elements remained; after two further online-rounds (n = 15, 94%; n = 10, 67%), consensus was achieved to include all four. By Stage 6 conclusion, 42 key elements were agreed ( Table 2).

Discussion
This consensus-driven definition moves the concept of non-pharmacological/non-surgical SSP forward. Previous ambiguity around SSP meant inherent difficulty in formulating appropriate research questions, standardising outcome measures, and synthesising evidence.
Non-pharmacological/non-surgical SSP, as defined, is not intended to stand-alone as a preventive strategy. Rather, it raises awareness of additional avenues for focus to maximise reduction in recurrent events. Gains from modest lifestyle changes in addition to pharmacological interventions have an estimated cumulative relative risk reduction for recurrent events of 80% (Numbers Needed to Treat: 5) [9]; notably pharmacological adherence is a health behaviour core to our definition [10].
Reaching consensus by electronic voting alone proved challenging for a number of 'key elements' including unsafe alcohol consumption, and psychosocial stress, both of which receive little attention in SSP RCTs [4,5,11], and addressing psychosocial stress is not an SSP guideline recommendation [1,2]. Similarly, insufficient evidence exists to recommend any one behaviour change and/or self-management theory in SSP. When achieving consensus became protracted, face-to-face meetings allowed effective open debate prior to anonymous voting.
Future work must include agreement on core outcomes for non-pharmacological, non-surgical SSP. Informed by this definition, a planned overview review will determine the quantity and quality of evidence from theoreticallyinformed studies employing behavioural and/or selfmanagement strategies on mortality, cardiovascular end points, and risk-reducing behaviours [12]. The consensus definition presented here is an important first step in building an impactful, evidence-based field in SSP.