Clinical research in dementia: A perspective on implementing innovation

The increasing global prevalence of dementia demands concrete actions that are aimed strategically at optimizing processes that drive clinical innovation. The first step in this direction requires outlining hurdles in the transition from research to practice. The different parties needed to support translational processes have communication mismatches; methodological gaps hamper evidence‐based decision‐making; and data are insufficient to provide reliable estimates of long‐term health benefits and costs in decisional models. Pilot projects are tackling some of these gaps, but appropriate methods often still need to be devised or adapted to the dementia field. A consistent implementation perspective along the whole translational continuum, explicitly defined and shared among the relevant stakeholders, should overcome the “research‐versus‐adoption” dichotomy, and tackle the implementation cliff early on. Concrete next steps may consist of providing tools that support the effective participation of heterogeneous stakeholders and agreeing on a definition of clinical significance that facilitates the selection of proper outcome measures.

"research-versus-adoption" dichotomy, and tackle the implementation cliff early on.
Concrete next steps may consist of providing tools that support the effective participation of heterogeneous stakeholders and agreeing on a definition of clinical significance that facilitates the selection of proper outcome measures.

K E Y W O R D S
Alzheimer's disease, clinical innovation, dementia, implementation, methodology, neurocognitive disorders, translational research

BACKGROUND
Medical research aims at providing new diagnostics, treatment, and care by testing new knowledge gained in basic research through clinical studies. However, achieving clinical innovation requires specific efforts that bring this new knowledge into practice, an urgently needed step in the field of dementia due to increased life expectancy and the consequent global increase of its prevalence. 1,2 The discrepancy between scientific knowledge production and its adoption in practice is evident from enduring gaps in diagnosis, treatment, and care. The 2020 Alzheimer Europe report on national responses to dementia shows that in half of the European countries, the actions taken to support dementia health care still cover only 40% to 63% of expressed needs (see Figure 17 in 3 ). The dementia detection rate itself is estimated to lag well below 70%, even in high-income countries, with imprecise diagnoses despite the availability of etiological biomarkers. 4 Routine clinical use of such biomarkers would increase early and accurate diagnosis. Implementing non-pharmacological treatment would improve the quality of care and of life of patients and caregivers (Table 1). Such actions would contribute to tackling the global priority of dementia concretely, but their implementation depends on proper completion of relevant intermediate steps.
Slow translation of scientific findings into practice is common in many medical fields. The definition of evidence-based clinical procedures requires strict validation studies, quality assessment of produced evidence, and demonstration of impact on relevant outcomes.
On the other hand, translation to practice entails obtaining regulatory approval, access and reimbursement, and adoption by clinicians, steps that are hampered by a variety of hurdles rooted in logistical procedures and cultural traditions. 5 We propose that the "implementation cliff" 6 cannot be overcome as long as these two aspects are not considered as the two faces of a single coin.
The World Health Organization (WHO) global action plan 7 inspires local policies to achieve concrete aims, incorporated in the national dementia strategies developed so far. 8 Formal monitoring based on specific indicators 9 helps to assess whether such aims are achieved.
However, hurdles between planned aims and their achievement need to be identified to mitigate them and proceed effectively. In this article, we identify some barriers and how they may be tackled to pull scientific advancements beyond publication in scientific journals and reach patients, carers, institutions, and communities. Identifying and over-coming such gaps is a pivotal step that complements and supports the WHO-inspired national dementia strategies.

CHALLENGES BETWEEN RESEARCH AND ADOPTION
The development (eg, definition of procedures, creation of tools, data collection) required to bring a valid and meaningful diagnostic tool or therapy to practice requires collaboration between professionals and researchers from different fields, as well as different entities, such as scientific societies, regulators, and decisionmakers. Consequently, gaps range from concrete methodological faults in the produced studies to the more elusive, but no less relevant, issue of communication among such parties.

Challenge 1: Communication
Given their diverse backgrounds and experience, scientists, clinicians, and related professionals, as well as stakeholders, may approach the same topic using different conceptual representations. Taking for granted that others understand exactly what we mean from our words leads to misunderstanding and ineffective collaboration. 10

Within research
Even among researchers, the concept of translational research itself is heterogeneous, covering different steps and aims along the translational continuum (Figure 1), where a unidirectional flow of information from basic research, to clinical studies, to adoption is usually implied. The heterogeneity of terms and concepts is even wider between the research and implementation fields, still considered as separate. 6 Figure 2) to play an active role. 12,13,5 Achieving such communication flow is challenging and still limited in dementia research.

Between research and clinics
Although less critical than lack of approval and reimbursement, often  [19][20][21] but their level of empowerment in the field of dementia still seems insufficient for active participation. The additional time required to engage, build capacity among both researchers and citizens, and structure cooperation is rarely funded, which affects the quality and efficiency of clinical research.
As an example, 22 outcomes of clinical utility of amyloid positron emission tomography are mostly defined in terms of physicians' diagnostic confidence, a relevant but limited scope of the utility that diagnostic biomarkers should demonstrate. 23,24 Aspects like clinical significance, quality of life, or patient-relevant outcomes require such complex collaboration to select, operationalize, and assess meaningful and measurable variables. Consensus and dialogue are necessary to "upgrade" relevant outcomes, that is, outcomes that matter within a patientcentered approach, to the well-acknowledged status of current precision medicine. Collaboration with institutions and research funders should aim at incentivizing the assessment of such outcomes and the collection of the required evidence enabling to reach patients: this may possibly start a virtuous circle of increasing investment for research able to provide greater return to society. Structuring such collaboration is thus necessary for scientific innovations to be developed, 25,26 approved, 27 and refunded. 28

Challenge 2: Methodology
Sound methodology is required to publish in peer-reviewed journals.
However, methodology is not sound per se, but rather context-specific:  31 The adoption of proper methodology requires that researchers, research funders, and publishers be aware of the kind of data necessary for decision-making.
To improve the methodology of diagnostic biomarker validation, a European consortium defined a systematic methodology, 23,26,32 known as the "Strategic Biomarker Roadmap" (SBR) (Figure 1

Challenge 3: Models for decision-making
Evidence is important to support regulatory or clinical decisions on new treatments or diagnostics (Figure 1

Methodology and validity
Recent systematic reviews on the methodology of decision-analytic models in general 36  In general, limited evidence has been reported on the external validation of decision models.

Assumptions and input data
Decision-model predictions are based on extrapolating short-term trial outcomes to a lifetime horizon. Indeed, the largest health benefit and care savings are obtained by postponing moderate-to-severe dementia stages, which have the highest impact on health and care use, and mortality. 41 Estimating the impact of early diagnosis or preclinical interventions requires additional assumptions on translating surrogate outcomes, like physicians' diagnostic confidence, amyloid load, or cognitive scores into patient-or caregiver-relevant outcomes. Finally, models are also affected by the limitations of input data, deriving from the mentioned low-dementia detection, 1,4 the unbalanced representation of ethnicity or socioeconomic status of study participants, the TA B L E 2 Defining a common language for inter-stakeholder participation. AD, Alzheimer's disease; BPSD, behavioral and psychological symptoms in dementia; EtD, evidence to decision procedures; 18F FDG-PET, 18F-fluorodeoxyglucose positron-emission tomography; PSA, prostate specific antigen

Examples of recommendations Gaps
Talking about dementia Avoid pessimistic or disempowering terms "A form of dementia" rather than "dementing illness" "Dementia" does not include conditions like mild cognitive impairment (MCI) or subjective cognitive decline (SCD), affecting the community and object of medical treatment and research Talking about people with dementia Avoid defining persons or life based on the condition "A person with dementia" rather than "dements," "sufferers," etc.
Need to define people attending memory clinics without a diagnosis of dementia. "Patients," "consumers," "end-users" are all not considered suitable by either the community or physicians/researchers Talking about carers Avoid assumptions: carers have different experiences "A person living alongside someone who has dementia" rather than "someone carrying the burden of caring" Research may need very short and direct terms. This needs to be respectful, but directedness may not be considered as assuming or judgmental in specific contexts Talking about symptoms and behavioral disturbances Try to describe it objectively and avoid negatively connoted terms "Behavioral and psychiatric symptoms" rather than "challenging behaviors" In research Address the condition or people as participants "Person living with dementia" rather than "subject" Need to include other conditions than dementia (MCI, SCD). Terms used in research contexts should be considered as elements in mathematical formulas; they need being short and precise without conveying judgmental assumptions. Agreement may need to be found specific to this context In medical practice Address the condition avoiding apparently degrading terms "Person living with dementia" rather than "case," or rather than abbreviations like "PWD" A language guideline has been proposed by the Australian Alzheimer's Association (https://www.dementia.org.au/files/resources/dementia-languageguidelines.pdf) and translated into different languages. This can be used as the basis for an update, like incorporating conditions like mild cognitive impairment (MCI) and adaptation to specific contexts of use. For example, research needs precise and direct terms that should be chosen among those not perceived as diminishing, if possible. However, understanding the context of use should also enable access to direct terms used in research as not necessarily assuming or judgmental. This requires communication and agreement among different parties.
scanty inclusion of the real range of comorbidities and clinical diversity of patients. Moreover, collecting such data focusing on contexts of specific resources and needs, rather than relative to dementia as an illness, would provide implementation-relevant information.

POSSIBLE WAYS FORWARD
What strategies can address the requirements of implementation and bridge research, clinics, and routine care providers in the dementia field? Useful methods can be adapted from theory-driven approaches, 42

Methodology
Increased awareness about the methodological requirements for

Modeling
Many of the limitations related to decision-analytic models originate from a lack of high-quality data. Nevertheless, some models have Innovative technology can of course contribute coordination platforms and computational power to interpolate missing information and process big data sources. Still, they cannot compensate per se for the outlined methodological gaps.

DISCUSSION
In this article we have outlined some hurdles that hamper the transfer of new research findings to clinical practice in the field of dementia, and we have proposed possible ways forward. Resolute efforts to improve our ability to bring clinical innovation are necessary to tackle the global increase of dementia prevalence. The investment in dementia research is lower than in other fields, like oncology. However, more effective implementation of interventions that are able to improve patients' well-being would provide a sizeable return of such investment to society, possibly starting a virtuous circle that increases investment.
Although this proposed approach is not new per se, 12,13,5 in this article we tried to identify concrete hurdles and resources to integrate the research and implementation fields and boost processes that bring clinical innovation for dementia.
The challenges identified in this article can be roughly grouped into two main categories, addressing how do we make progress (1)  Such implementation-oriented approach is challenging, requiring a structured effort to bridge the heterogeneous aspects and stakeholders within a consistent perspective. Specific methodology is needed to promote accessibility, empowerment, and synergies. This requires no less than a turn in culture, which is, however, timely given the increas- Typically, they simulate a simplification of the natural disease progression, the impact of the new technology on that progression, and the consequences in terms of their change in health and change in use of care resources for a specific population. In the field of AD and related disorders, such policy-decision models often extrapolate short-term trial results to long-term impact on health and care use. Open access funding enabled and organized by Projekt DEAL. The other authors report no conflicts of interest.