Qualitative Research to Inform Economic Modelling: Older People’s Views on Implementing the Nice Falls Prevention Guideline in English Community Context.

Background: High prevalence of falls among older persons makes falls prevention a public health priority. Yet community-based falls prevention face complexity in implementation and any commissioning strategy should be subject to rigorous economic evaluation to ensure cost-effective use of scarce healthcare resources. The study aims to capture the subjective views of older people in Sheeld on implementing the National Institute for Health and Care Excellence (NICE) guideline on community-based falls prevention and explore how the qualitative data can be used to inform commissioning strategies and the conceptual modelling of falls prevention economic evaluation. Methods: Focus group and interview participants (n=27) were recruited from Sheeld, England, and comprised falls prevention service users and eligible non-users of varying falls risks. Topics concerned key components of the NICE-recommended falls prevention pathway, including falls risk screening, multifactorial risk assessment and treatment uptake and adherence. Views on other topics concerning falls were also invited. Framework analysis was applied for data analysis, involving data familiarisation, identifying themes, indexing, charting and mapping and interpretation. The qualitative data were mapped to three frameworks: (1) facilitators and barriers to implementing the NICE-recommended pathway and contextual factors; (2) intervention-related causal mechanisms for formulating commissioning strategies spanning context, priority setting, need, supply and demand; and (3) methodological and evaluative challenges for public health economic modelling. Results: Two cross-component factors were identied: health motives of older persons; and professional competence. Participants highlighted the need for intersectoral approaches and prioritising the vulnerable groups. The local commissioning strategy should consider the socioeconomic, linguistic, geographical, legal and cultural contexts, priority setting challenges, supply-side mechanisms spanning provider, organisation, funding and policy (including intersectoral) and health and non-health demand motives. Methodological and evaluative challenges identied included: incorporating wider costs and effects; considering dynamic complexity in ageing process; considering social determinants of health; and conducting equity analyses. Conclusions: Holistic qualitative research can inform how commissioned falls prevention pathways can be feasible and effective. Qualitative data can inform commissioning strategies and conceptual modelling for economic evaluation of falls prevention and other geriatric interventions. This would improve the structural validity of quantitative models used to inform geriatric public

engage in healthy behaviour (41) -which can operate independently of supply. Hence, CICI could be supplemented by the health needs assessment (HNA) framework that incorporates demand, supply and need/eligibility as distinct yet overlapping domains (42). Such combined framework can be used to organise the themes identi ed from qualitative research with older persons into a format that guides commissioning strategies. Speci cally, the framework can highlight which implementation factors are modi able by commissioning and to what extent given the decision space (i.e., the scope for feasible action, determined by a range of factors including the stakeholders involved, decision time horizon and budget/capacity constraints) of the commissioner.
Secondly, the nature of falls being a public health problem faced by a broad spectrum of older populations -rather than a clinical problem faced by a well-de ned, narrow patient group -presents further complexity beyond intervention-related causal mechanisms (43). Though falls prevention clearly involves clinical practice informed by the NICE clinical guideline, its population-wide commissioning constitutes a much broader decision space as evidenced by the involvement of a wide range of allied health and nonclinical (e.g., Age UK) stakeholder organisations in formulating the Public Health England consensus statement on falls prevention (44). According to a systematic methodological review, the key methodological and evaluative challenges for public health economic evaluation include: incorporating wider costs and effects (e.g., intervention bene ts beyond improvements in narrow measures of health such as physical capacity and functioning); considering social determinants of health and assessing the intervention effect on social inequities of health; considering dynamic complexity in health determinants and intervention need; and considering theories and models of human behaviour based on psychology and sociology (45). Addressing such challenges is part of the INTEGRATE-HTA recommendations (see Chap. 3) (19), and is necessary for improving the structural validity of the decision model (43). The challenges speci c to the decision problem can be identi ed by consulting relevant stakeholders, including commissioners, professionals and service users; a conceptual model can then be developed that informs subsequent data gathering and model parameterisation (43). Hence, the framework of key methodological and evaluative challenges can be used to organise the relevant qualitative data obtained from older persons. This would then facilitate the development of the conceptual model of falls prevention economic evaluation.
In all, a de novo qualitative study of older people can holistically explore the facilitators and barriers for implementing the NICE-recommended falls prevention pathway. The resulting data can improve two types of understanding on complexity: one concerning the causal mechanisms of falls prevention in speci c local contexts; the other concerning falls as a public health problem. Both types inform the development of the conceptual model for economic evaluation of falls prevention and thereby improve the credibility of the nal economic model used for decision-making.

Aim And Objectives
The study aims to capture the subjective views of older people in She eld on implementing the NICE CG161 guideline on community-based falls prevention and explore how the qualitative data can be used to inform the conceptual model of falls prevention economic evaluation. She eld was chosen as a representative local setting to which CG161 is applicable. The research objectives are to: 1. Identify the facilitators and barriers for implementing key components of the CG161 communitybased falls prevention pathway -including falls risk screening and assessment, falls risk awareness, and uptake and adherence of treatments within multifactorial intervention -and contextual factors in uencing the pathway implementation.
2. Inform potential commissioning strategies on falls prevention by understanding the causal mechanisms in context, supply, need and demand that in uence implementation.
3. Identify the methodological and evaluative challenges associated with economic evaluation of falls prevention as a public health decision problem.
The target audience hence includes local commissioners of falls prevention and other geriatric public health interventions, and economic modellers who provide analytic support. The identi ed facilitators and barriers would also be of interest to professionals and patient groups seeking to improve local implementation of falls prevention.

Methods
The qualitative research involved focus groups and interviews with older persons living in the community. The ethics approval was obtained from the Research Ethics Committee at the School of Health and Related Research, University of She eld (ref. 025248). Written consent was obtained from willing participants.

Target population and sampling
The target population comprised persons aged 65 + in She eld, England, and persons aged 50-64 who are at high falls risk. The latter group was included to explore the rationale for earlier prevention as is currently recommended for inpatient settings by CG161 (2). Purposive sampling covered multiple categories of participant characteristics in terms of falls risk and service use as illustrated in Fig. 1.
According to CG161, those with history of fall(s) requiring medical attention or recurrent falls in the past year and/or mobility and balance problems were de ned as high-risk (2). Low-risk individuals were sampled because they are still eligible for falls risk screening and/or interested in early prevention.
Recruitment continued until all participant categories were covered and themes saturated. Speci cally, two focus groups were formed from two separate cohorts enrolled in Dance to Health, a falls prevention programme that combines evidence-based Otago and Falls Management Exercise components in dance routines (46, 47); these groups contained high and low risk service users. Two further groups were formed from a Patient and Public Involvement group meeting regularly at the Northern General Hospital and a social group meeting at Zest Community, a local social enterprise offering leisure, health and work support services to diverse age groups; these contained high and low risk service non-users. Two interview participants were recruited from Dance to Health and Zest Community.
Focus groups were held directly before/after the regular meetings. Community organisation staff con rmed before research commencement whether their members could give informed consent. One participant declared memory problems while another a recent diagnosis of Alzheimer's disease; but both were regular attendees of community groups and expressed con dence in participating. After obtaining written consents, questionnaires were administered to collect data on demographics, falls history and fear of falling, current physical activity, and contact with falls prevention services.

Discussion topics
The main discussion topics were structured around the sequential steps of the proactive prevention pathway recommended by CG161 (2), namely: (i) falls risk screening/assessment by professionals; (ii) participant suggestions on raising falls risk awareness in the community; (iii) initial uptake of different treatments; and (iv) long-term adherence to treatments. The pathway is proactive in that it is initiated by professional referral of high-risk individuals after falls risk screening. If mentioned by participants, two further pathways were discussed: the reactive pathway -where older persons are referred to falls prevention by professionals after medical attention for a fall, which is also recommended by CG161 (see recommendations 1.1.2.1, 1.1.3.2 and 1.1.6.1) (2); and the self-referred pathway -where older persons enrol in falls prevention without professional referral.
A simpli ed graphical summary of the proactive pathway, as shown in Fig. 2, was used to explain the main topics to participants. Four treatment types -exercise, HAM, medication change and vision improvement -were explained while emphasising that other types exist, such as chiropody. It was also highlighted that reactive pathway after a serious fall is commonly used, and that self-referred pathway is recommended by experts (48). Further contextual factors in uencing falls risk and prevention (e.g., safety of pedestrian walks in Winter) were actively explored as they emerged during discussion.

Data collection
Recorded audio data were transcribed and anonymised. The questionnaire data were similarly transferred to an Excel spreadsheet and anonymised. Both data were stored securely in the University designated folder.

Data analysis
A framework analysis was employed for the analysis of obtained data (49,50). The approach involved ve stages: (a) familiarisation -which involves repeated listening to audio and reading of transcripts for immersion in the data; (b) identifying a thematic framework -which may be based on an a priori set of issues related to the research objectives and/or themes emerging from the data; (c) indexing -which systematically applies the thematic framework to the transcripts; (d) charting -which 'lifts' the data from the transcripts and rearranges them (e.g., in a tabular format) according to the thematic framework; and (e) mapping and interpretation -which seeks associations and develops policy-related strategies from the charted data based on a priori issues and/or emerging themes. Stages (a) to (c) were conducted independently by two authors (JK and YL). All authors contributed to stages (d) and (e).
From stage (b) onwards, three frameworks related to the research objectives were constructed using a priori concepts and themes emerging from the data: 1. Framework to understand the facilitators and barriers to components of the NICE CG161 falls prevention pathway and cross-component and contextual factors. 2. Framework to inform potential commissioning strategies by accounting for causal mechanisms in context, priority setting, need/eligibility, supply and demand. 3. Framework to understand the key methodological and evaluative challenges to public health economic model development.

Framework (I): Facilitators and barriers and crosscomponent and contextual factors
This framework closely follows the structure of the discussion topics and charts the main themes identi ed from the data. Themes are divided into pathway components -i.e., (i) falls risk screening/assessment by professionals; (ii) raising falls risk awareness; (iii) initial uptake of treatments; and (iv) long-term adherence to treatments -and then into facilitators and barriers. Cross-component factors are facilitators and barriers that in uence multiple components; these are highlighted as the main determinants of successful pathway implementation. Contextual factors in uence the feasibility and impact of the components and also inform the commissioning strategies as discussed below.

Framework (II): Potential commissioning strategies
This framework rearranges the main themes under Framework (I) into a format that guides commissioning strategies. The strategies may inform immediate commissioning decisions or the conceptualisation of a de novo decision model. A combined framework is constructed that incorporates the implementation context-mechanism distinction of the CICI framework (20) and the need-supplydemand distinction of the HNA framework (42), as shown in Fig. 3.
The CICI framework highlights eight domains of context, ranging from the immediate intervention setting to political in uences. The contextual factors already identi ed under Framework (I) are mapped to these domains. The implementation context also presents priority setting challenges to the decision-maker. The three main priority setting criteria highlighted by an international panel of experts and stakeholders are reducing social inequities of health, prioritising the frailest and reducing the non-health costs (35).
Context also in uences the implementation mechanisms of provider, organisation, funding and health system-wide policy, shown in the Supply circle of the HNA Venn diagram. Providers and organisations are micro-and meso-level entities delivering the commissioned interventions. Funding supports these entities as well as wider, auxiliary implementation strategies (e.g., community marketing to in uence demand). Policies concern macro-, system-level changes to facilitate implementation (e.g., changes to GP reimbursement structure to facilitate regular falls risk screening).
The key consideration for formulating commissioning strategies is the decision space that de nes which contextual factors and mechanisms are modi able and to what extent. The decision space is determined by the combination of context and priority setting challenges, range of stakeholders involved, decision time horizon, and any budget and capacity constraints. For example, improving professional competence requires the cooperation of professional training institutions and may not be feasible in the short run; conversely, changing housing regulations may be feasible if the local Council and housing associations are actively involved in decision-making. The decision space may be largely pre-established prior to the qualitative study; alternatively, the qualitative ndings may motivate changes to the decision space.
Intervention need/eligibility in the HNA Venn diagram is chie y determined by normative clinical and public health guidelines and intervention studies that have used rigorous research designs to demonstrate the ability to bene t from the interventions (42). Yet, eligibility criteria may fall within the decision space if there is exibility in how the criteria are applied in the local context. The CG161, for example, does not prescribe any speci c care pathway for cognitively impaired persons (2); hence, the local commissioners and professionals may design a locally speci c pathway. Framework (II) similarly seeks to identify major determinants of demand including personal factors underlying uptake/adherence decisions (e.g., health-related motives for healthy behaviour (41)) and external in uences on demand (e.g., community marketing, self-e cacy promotions (51,52)). The implications on commissioning are inferred from the types of demand-side factors and whether these fall within the decision space.

Framework (III): Methodological and evaluative challenges for decision modelling
The categories of key methodological challenges for public health economic model development are taken from the systematic methodological review of Squires and colleagues (45), namely: (a) incorporating wider costs outside public healthcare sector and wider non-health effects of interventions (e.g., high social participation, productivity) not captured by generic health utility measures, such as EQ-5D, that are recommended for use in health technology assessment (34,53,54); (b) considering social determinants of health; (c) considering dynamic complexity in falls risk and intervention need; and (d) considering psychological and sociological theories of human behaviour.
The evaluative challenges are inherent in all economic evaluations and include the perspective, type of analysis and time horizon of the evaluation and the assessment of equity and other priority setting criteria (35,54,55). They also include the choice of methods (e.g., costing) and scenarios (i.e., base case and alternative scenarios) to evaluate the range of intervention-related causal mechanisms identi ed under Framework (II).

Participant characteristics
Twenty-seven persons participated in research across four focus groups (FG1-4) and two interviews (INT1-2) between October 2019 and January 2020. Table 1 summarises their characteristics.  3 The list of services was taken from Cochrane systematic review of falls prevention trials (9). However, the questionnaire did not explicitly label these services as falls prevention interventions in order to invite responses from participants who may have received a multi-purpose service (e.g., physiotherapy or vitamin D supplementation) without awareness of its falls prevention property. Overall, 21 participants (78%) indicated use of one or more service.
Regarding current access to falls prevention, 11 reported having spoken to a professional about falls risk. Nevertheless, 21 reported recent use of services with some falls prevention properties (9), suggesting that the main falls prevention pathway under current practice is self-referral by older persons. Of the 21 users, 13 reported accessing multiple interventions. The most widely accessed services were physiotherapy and falls education.  Falls risk screening and assessment by professionals [1] (A) Professional competence

Raising awareness of falls risk
Participants generally recognised that falls risk awareness is a matter of understanding the ageing process, not only from a certain senior age but from earlier adult life stages. For example, one participant expressed the di culty of staying aware of falls risks at home during the gradual ageing process: (FG1) "Well, it happens so gradually, doesn't it… when it is part of ageing and degenerative thing, it's not like they go over night from being perfect to being in a wheelchair. It's such a gradual thing. And you get used to stuff. You get used to the fact that the rug was curled up at the end." The role of informal caregivers in maintaining awareness of falls risk, particularly in the living environment shared with older persons, was also highlighted.
Initial uptake of falls prevention treatments Important intervention characteristics included cost, enjoyability, suitable di culty, safety, location, timing, support facilities (e.g., lack of handrail at venue entrance), and transport issues (availability and cost). Individuals considered whether the speci c combination of these characteristics suited their preference and ability to pay. For example, one participant perceived modest private cost as an acceptable trade-off to enjoyability, while another perceived transport costs as a key main barrier: Participants acknowledged the in uential role of professionals in determining their treatment uptake, more in uential than their peers according to theme [3][4][5][6][7][8][9][10][11][12]. The key steps were professional awareness of falls prevention initiatives in the community, followed by proactive recommendations or referrals made in a respectful and person-centred manner: (FG1) "One person when we had a meeting found out that so many doctors were handing out too many drugs instead of an alternative. There was an alternative. [My doctor at surgery] said, 'I'd want you to go and do an aquarobics' and that helped me, that helped me so much that I didn't need the drugs." Long-term adherence to falls prevention treatments (FG2) "I've got loads of medication variation problems. For me, I don't really expect GPs to improve things, but they never told me 'Oh we could change this into that'. He [the GP] just expects me to just keep preordering the medications. So I leave it that way." There was a close overlap in factors determining treatment uptake and adherence. As for factor differences, experience of falling was mentioned as a facilitator for uptake but not for adherence. Socioeconomic and linguistic barriers were mentioned only for uptake, likely because they are su cient to discourage both uptake and adherence for the marginalised subgroups. Funding constraints impeded both uptake and adherence, though in different ways: adherence was predictably curtailed by the funding cut at the end of the pilot period (theme [4][5][6][7][8][9][10][11][12][13][14][15][16][17]); while uptake was impeded by deliberate policy to concentrate funding in deprived areas despite higher demand in well-off areas: (FG3) "Now, to be honest, this [well-off] area doesn't usually have anything. You know, I mean, all the money and the grant has been put into only deprived areas." Contextual factors in uencing the falls prevention pathway Table 3 summarises the contextual factors that in uenced the pathway implementation. They are grouped under two categories: (i) intersectoral factors; and (ii) prioritising the vulnerable groups. Table B in Supplementary Material shows the direct transcript quotes. • Health-promoting local public spaces [5 − 2] • Home ownership and modi cation [5 − 3] • Communitarian approaches [5 − 4] • Persons with complex comorbidities [6 − 1] • Persons experiencing cognitive decline [6 − 2] • Socially isolated persons [6 − 3] Intersectoral factors Intersectoral factors concerned matters typically addressed outside the healthcare system, including the safety and health-promoting features of local public spaces, the relationship between home ownership and ability to implement home modi cations, and potential communitarian approaches that mobilise the community to meet common goals. Older participants mentioned how in the past the local community would handle the challenges that lie outside the local/central government's responsibility; the decline in communal responsibility was perceived to explain the increase in local health hazards: (FG1) "I don't think neighbours are neighbours anymore, either. When we were younger, I remember when snow came here, all the men of each family would come and make a path. And they don't do that now." Prioritising the vulnerable groups Another set of themes concerned the need to prioritise the most vulnerable individuals at risk of a serious fall or loss of independence. Three groups were identi ed: persons with complex comorbidities; persons experiencing cognitive decline; and socially isolated persons. The reported experience of the diabetic participant who was below age 65 (hence below the eligibility age for the proactive pathway) illustrated how vulnerable individuals concurrently face multiple risk factors for serious falls: (FG1) "If I had a bad day with my high sugar levels. I've had my bad day with blurriness. And I come down a lot of stairs and I fell X times coming down from attic and obviously coming out of my building which is a high old building. And then you've got to come down some more which is always full of leaves." Despite this, public support for home assessment and modi cation was denied due to her ability to walk 100 meters without problem, and support from other care professionals was similarly lacking.
Framework (II): Potential commissioning strategies Table 4 rearranges the identi ed themes according to the CICI-HNA framework. This could potentially be addressed by new housing regulations that incentivise relevant action by landlords. The culture of communal responsibility could be enhanced to some extent by supporting community organisations and civic initiatives. If chosen as a priority criterion, bridging the socioeconomic and linguistic gaps in intervention access and outcomes would require cooperation of representatives of the marginalised groups to devise appropriate, tailored strategies.
The commissioner may also decide to change the eligibility criteria for falls prevention according to local priorities. Currently, CG161 recommends community-based falls risk screening for those aged 65 and over, followed by referral to multifactorial intervention for those at high falls risk de ned by falls history and abnormal gait/balance. The screening protocol can be expanded to include those with complex comorbidities who are aged less than 65; the risk factors examined for referral can similarly be expanded to cover frailty and non-health factors such as social isolation. A separate pathway may be designed for cognitively impaired persons who require tailored support from dedicated organisations: (INT2) "But with these walks which are organised by the Alzheimer's Society is that there are quali ed people leading the walks." Older participants identi ed a broad range of supply-side issues and solutions at provider/organisation, funding/policy and intersectoral levels as shown in the second column. The commissioner should determine which solutions lie within the decision space which would depend on factors such as stakeholder involvement, time horizon and budget/capacity constraint: e.g., professional attributes such as commandeering attitude may take time to remove. Signi cant investments -e.g., a new Falls Clinics, changes to GP reimbursement schedule for risk screening -would similarly take time and be constrained by the budget. Certain intervention characteristics -e.g., private costs and sta ng level -may be solved by commissioning; but other characteristics (e.g., timing because venue is shared by other sessions) may be unmodi able.
The last column arranges the demand-side themes by three types: health and fall-related motives of older persons; non-health and social motives; and external in uences on demand. Importantly, the external in uences are modi able by using auxiliary implementation strategies (e.g., community marketing). Older persons are also receptive of professional recommendations; hence, this in uence can be maximised by improving professional attributes such as awareness of community initiatives:  Table 5 summarises the methodological and evaluative challenges for falls prevention economic model identi ed from the qualitative data.  [3-20, 3-23, 4-11].
• Model should capture wider health bene ts of interventions beyond falls prevention [4 − 3].
Perspective, type of analysis and time horizon • Societal perspective is likely necessary to capture private intervention costs [3-20, 3-23, 4-11] and private care expenditures due to falls.
• CCA can be used if non-monetary outcomes without clear cost-effectiveness thresholds (e.g., number of intervention volunteers and participants as a measure of civic strength [5 − 4]) are tracked.
• Time horizon should be long enough to capture the full impact of falls and falls prevention and dynamic trajectories of ageing and falls risk.
• Model should incorporate seasonal changes in falls risk due to environmental risk factors [5 − 1].

Intervention scenarios evaluated
• Usual care scenario should re ect limited access rates due to barriers. 4 • Main intervention scenario should incorporate: local eligibility criteria tailored to changing falls risk pro le; multiple non-mutually exclusive intervention pathways; external evidence on interventions which have similar characteristics as those preferred by local older persons. 5 • Intervention costing should incorporate: cost of risk identi cation; cost of auxiliary implementation strategies; xed/sunk costs for major system changes; cost of additional resources to achieve full set of positive intervention characteristics; cost of professional training to obtain positive attributes; and funding to sustain intervention over su ciently long period. 6 Methodological challenges [Theme #] 1 Evaluative challenges [Theme #] • Additional scenarios conducting value of implementation analyses to evaluate auxiliary implementation strategies [2-2, 3-4, 3-5, 3-6]. The data identi ed costs incurred outside the public sector (e.g., private intervention and transport costs, costs of venues donated by local church) and non-health intervention effects (e.g., high social participation) which were important facilitators and barriers. Accordingly, evaluation should consider taking the societal perspective and using a broader wellbeing measure. Private intervention costs should be matched by private care expenditures incurred from falls. Participants also highlighted wider health bene ts of exercise beyond falls prevention, including improved mobility and mental health: Social determinants of health identi ed from the data included socioeconomic and ethnic/linguistic barriers to intervention access and social isolation that left individuals without close support in case of serious fall or functional loss. The lower intervention access may mean that the intervention is less costeffective for the socioeconomically deprived and ethnic minority subgroups. A strategy that prioritises access for these groups to reduce the social inequities of health (e.g., concentrating funding in deprived areas ) would introduce an equity-e ciency trade-off. The nal model should parameterise the causal mechanisms to quantify the trade-off; the strategy would be accepted if stakeholders nd the trade-off to be reasonable. A similar process of equity-e ciency evaluation can be applied to other vulnerable subgroups identi ed, i.e., those with complex comorbidities and cognitive impairment.
The dynamic processes of ageing and falls risk progression, starting before the age of 65, were mentioned by some participants as motivating factors for intervention uptake/adherence; yet others perceived the emerging illnesses as major barriers: (FG4) "Well, I used to go swimming a lot every week. But then, since a long period of illness, I stopped going." Either way, the model should seek to capture the dynamic trajectories of physical capacity, functional status and health perception as key determinants of intervention demand. Moreover, the dynamic progression means that persons at different stages of the falls risk progression have different intervention needs; the model can quantify the added bene ts of an intervention strategy that tailors treatments to progression stages relative to a strategy that does not. An example of the latter was perceived by older participants: (INT2) "I think [the professionals] ought to check things like stairs and back steps. And not expect the elderly people to report it, because they are probably so used to these things when they've lived in the house all the time and are not necessarily aware of how less well coordinated they are from before." Another key outcome of dynamic complexity is heterogeneity; hence, the model should capture the changes in the composition of vulnerable groups and conduct appropriate equity-e ciency evaluations for different strategies as discussed above.
The data, especially when organised under Framework (II), can inform the range of intervention scenarios evaluated under base case analysis and alternative scenarios analyses. All three prevention pathwaysproactive, self-referred and reactive -were mentioned in the data (see theme [1][2][3][4][5] for participant discussion of a reactive HAM receipt), and hence should be considered in the base case analysis. The main intervention scenario (compared to usual care under base case analysis) should incorporate interventions that have some or all of the positive characteristics (see Table 4) such as allowing individually tailored di culty. Where external studies are used as data sources (e.g., RCT for e cacy), they should evaluate interventions with similar characteristics as the model scenario.
Intervention costing should incorporate not only the cost of intervention delivery but also the cost of auxiliary implementation strategies used to generate the given uptake and adherence; for the proactive pathway, the cost of professional risk screening and referral should be included. Major system-level changes (e.g., integrated data system for risk screening) would incur xed/sunk costs which may be incorporated as annuitized overheads. Costs would be incurred if additional professional training (resources) is required to obtain positive professional attributes (intervention characteristics).
Key psychological and sociological factors identi ed from the qualitative data (e.g., health motives in uencing demand) can be parameterised in the nal model based on relevant theories and external quantitative data. An alternative, heuristic method is to conduct value of implementation analyses as alternative intervention scenarios (57). Additional monetary value of hypothetical improvements in intervention uptake/adherence can be estimated without knowing what psychological or sociological factors contributed to the improvements. The additional value is the maximum amount that can be invested in auxiliary implementation strategies that produce the given improvements.

Discussion
This study explored older people's views on facilitators and barriers for implementing the communitybased falls prevention pathway recommended by NICE as well as broader themes on raising falls risk awareness, intersectoral initiatives and prioritisation of vulnerable groups. Participants included service users and non-users and those at high and low risks of falling. The study also explored how the identi ed themes can be mapped on to frameworks that can inform commissioning decisions via a de novo falls prevention economic model. It was thereby shown that the framework analysis approach (49) can exibly accommodate diverse frameworks according to research aims.
The methods and results of this study contribute to the growing eld of research exploring how qualitative evidence can be effectively used to inform health technology assessment (HTA) (40). The recent NICE Decision Support Unit (DSU) report, for example, critiques the limited consideration of qualitative evidence in the current NICE HTA guideline (process and methods guideline 9; PMG9) (53) and sees the use of established, purpose-speci c frameworks -including the CICI framework -as a tool for accelerated and standardised incorporation of qualitative evidence in the HTA decision-making process (40). This study showed that the CICI framework, despite its focus on supply-side conditions, can be applied to service users and eligible non-users. Previous qualitative studies have indeed shown that older people are sensitive to supply-side issues including cultural-linguistic context of intervention, professional attributes and intervention characteristics (50,(58)(59)(60), making their views highly relevant to commissioning decisions that must consider not only the supply-side conditions but also how these are perceived and accepted by service users. This study facilitated attention on users' perception and demand by supplementing the CICI framework with the HNA framework that conceptualises intervention access as an outcome of interactions between demand, supply and normative need. Such exible adaptation and application of the CICI framework is encouraged by the framework developers (20).
Moreover, both the CICI framework developers and the DSU report focus on the application of CICI to qualitative and mixed-methods systematic reviews and not to primary qualitative research (20,40). By applying the framework to primary research -which is arguably more relevant for local decision-making -this study demonstrates the wider potential reach of the framework. Indeed, wider primary application would facilitate secondary syntheses.
This study also showed that the primary qualitative research on service users can identify the key methodological and evaluative challenges to public health economic evaluation and thus function as a vital step within the conceptual modelling process (43). Having identi ed the key causal mechanisms, the qualitative data can also identify the necessary group of stakeholders to modify them, and those not already involved in the project can subsequently be recruited. These are ex-ante, or prospective, applications of the qualitative evidence to inform the de novo model development. Yet ex-post application may be equally valuable: in England and Wales, local clinical commissioning groups (CCGs) and local authorities are required to implement an intervention approved by NICE HTA within three months of the approval unless major local barriers to implementation can be identi ed (recommendation 1.5.1) (53).
The local qualitative evidence can identify such barriers and/or anticipate any major differences in the local cost-effectiveness and population-level outcomes relative to those predicted by the HTA. Moreover, the decision model underlying the HTA approval can be critiqued based on the methodological and evaluative challenges identi ed by the local qualitative evidence. If the model performs poorly in addressing the challenges, then a de novo model can be commissioned; the qualitative data would then be applied ex-ante. As mentioned, the ex-ante approach is more relevant for community-based falls prevention since no HTA has been conducted, and existing models (11, 39) do not adequately address the methodological challenges. The 2019 surveillance for the update to NICE CG161 (not yet published at the time of writing, May 2021) also mentions no plan for economic evaluation nor indeed for primary/secondary qualitative research with older persons (61).
The holistic approach to exploring the falls prevention facilitators/barriers identi ed two crosscomponent factors: health motives of older persons; and professional competence. The role of health motives in in uencing older persons' health behaviour has been debated in the literature. One study in Scotland found that older people are unlikely to participate in exercise for health reasons but rather for the social rewards (62); while another found that health motives (e.g., maintaining functional independence) help translate intentions into actual change in health behaviour (41). This study found that health motives operate alongside the social rewards of interventions which corroborates the ndings of a previous qualitative systematic review of older persons' views (58). CG161 similarly recognises both factors and recommends that care professionals provide information on the physical bene ts of modifying falls risk to older persons and caregivers (recommendation 1.1.10.2), while also promoting the social values of interventions (1.1.9.2) (2). The absolute and relative strengths of health and non-health motives thus impact the nal combination of intervention characteristics and auxiliary implementation strategies: for example, strengthening the health motives would require well-framed health messaging (51) and health literacy/awareness promotion in earlier life course (63); while addressing the nonhealth/social motives is a matter of better intervention design (e.g., a more sociable group environment).
The importance of the second cross-component factor, professional competence, is a rmed by CG161 which recommends that all healthcare professionals regularly dealing with older persons "develop and maintain basic professional competence in falls assessment and prevention" (1.1.10.1) (2). Yet older participants perceived external constraints placed even on competent professionals, including time constraints. This corroborates the ndings from a previous survey of English GPs which speci ed insu cient consultation time and lack of allied health professionals in the community as the most prominent barriers to implementing CG161 (25). Therefore, commissioning should comprehensively account for care system bottlenecks and carefully cost the solutions for their removal. One economic model, for example, incorporated the cost of a citywide falls risk screening that was assumed to operate like a cancer screening programme (66). Costs that are xed/sunk would interact with uptake rate to produce worse cost-effectiveness if uptake is inadequate (67) and economies of scale if uptake is increased (66). Hence, models should accurately portray the cost structure ( xed vs. variable) to characterise the impact of implementation quality on cost-effectiveness. Aggregate population-level health and/or economic impact is another outcome largely determined by implementation; the NICE PMG9, for example, stresses the need to account for such impact in HTA decisions (see recommendations 5.12.3 to 5.12.7) (53). Yet cost-per-unit ratios (e.g., incremental cost-effectiveness ratio) are often interpreted in isolation when using economic evidence for decision-making (68-70). The nal model informed by the qualitative evidence should present both ratio and aggregate outcomes so that the full impact of implementation quality could be quanti ed (71).
Less emphasised in CG161 but visible in the qualitative data (e.g., theme [4][5][6][7][8][9][10][11][12][13][14][15][16]) is the role of nonclinical professionals and volunteers who can substantially in uence both supply and demand given their proximity to older persons in the community (72): a pilot falls prevention scheme in She eld, for example, found that falls risk screening conducted at local community groups and lunch clubs signi cantly increased uptake (73). It is hence critical to value the nonclinical and volunteer contributions; and value of implementation analysis offers a heuristic method to that end (57). For example, one falls prevention model set in a Massachusetts community of population size 44,000 estimated that increasing falls prevention uptake from 50-75% would yield an additional $2.79 million which is the maximum amount that can be invested in community organisations to generate such uptake increase (74). Such monetary value can be combined with qualitative data on demand-side in uences to devise a costeffective implementation strategy.
The methods used in this study are applicable to other geriatric health areas. One care strategy attracting policy attention is integrated care, designed to create "connectivity, alignment and collaboration within and between the cure and care sectors at the funding, administrative and/or provider levels" (75). Since 2014 in England, the Better Care Fund obliges CCGs and local authorities to create a shared budget for health and social care and other public services, and also invests its own capital (£6.4 billion in  to facilitate integration (76). Such a strategy brings problems of implementation as diverse service components and teams are combined (77); the empirical results for integrated care schemes are accordingly mixed (78, 79). The holistic, cross-component qualitative investigation of the facilitators and barriers is likely critical for the schemes' implementation. The contextual factors are similarly critical as the decline in physical capacity during ageing raises the in uence of the wider environment in determining the older person's holistic capability (80-83). The key methodological and evaluative challenges must likewise be addressed by any economic model of geriatric public health interventions: for example, the social heterogeneity in health status is a prominent feature of geriatric population and raises equity issues (84, 85).

Strengths and limitations
The simultaneous coverage of three frameworks -cross-component factors, intervention-related causal mechanisms and public health modelling challenges -is a key strength of this study. Qualitative research is often conducted and interpreted separately from economic evaluation, even when both designs are conducted in the same project (86, 87). By contrast, this study explores how qualitative data can directly inform model-based economic evaluation. Another strength is its sampling of participants of varying service use history and falls risk re ecting the heterogeneity in older populations.
The study nevertheless has limitations. The purposive sampling could have accounted for social categories such as area-level deprivation, particularly given the importance of social determinants of falls prevention access. The sampling was concentrated around older persons living near the She eld city centre, meaning that persons living in more rural suburbs -and having different sets of preferences and constraints -were underrepresented. More views could have been elicited from informal caregivers given their central role in facilitating falls prevention.

Conclusion
Better understanding of older persons' health motives and higher professional competence can improve the implementation of the NICE-recommended falls prevention pathway. Older persons are sensitive to implementation causal mechanisms, meaning that their views can inform contextual and supply-side changes to promote falls prevention and wider health promotion. They are also important stakeholders who can inform the development of a complex public health economic model.   Graphical summary of the recommended falls prevention guideline used to introduce the discussion topics to focus group and interview participants.