Clinical trials in neurosciences: Need for standards for innovation on clinical outcomes

Abstract

Advances in understanding neurobiological factors underlying neurologic and psychiatric conditions have yet to be adequately translated into successful treatments. In neuroscience therapeutic areas, a high percentage of clinical trials fail to demonstrate safety and efficacy, resulting in a lower innovation index compared to other fields. Developing new treatments requires innovative strategies for outcome decision-making, including tailoring existing instruments or creating novel measurement tools to address validity gaps and capture the clinical benefits of new therapeutic approaches. The interdisciplinary expert group forming the Network on Clinical Outcomes in Early-Phase Clinical Trials within the European College of Neuropsychopharmacology (ECNP) promotes harmonization of outcome research practices to increase the likelihood of identifying safe and effective treatments in neuroscience. Consensus-based recommendations on outcome selection standards were issued to guide researchers designing clinical trials aligned with regulatory agency guidelines (Zaragoza Domingo et al., 2024). This paper highlights two common concerns regarding outcomes in neuroscience clinical trials. First, the availability of reliable, sensitive, and fit-for-purpose (FFP) clinical outcome assessment (COA) tools. In fact, several well-established instruments would not meet current qualification requirements, despite their continued use. Second, instruments validated in one clinical context are being applied to a different condition without confirming validity, compromising content validity and measurement properties. A 7-step standard framework is presented to operationalize outcome decisions across early-to-late-phase pharmacological research, reinforcing content validity and psychometric rigour. A recent expert survey broadly endorsed this framework, and future plans include piloting the method across challenging therapeutic fields and engaging regulatory agencies.

Graphical abstract

Challenges and future steps in innovative Clinical Outcome Assessment (COA).

Highlights

• •Outcome research innovation in neurosciences is needed for more conclusive trials.
• •Legacy outcome assessment instruments may lack sensitivity to detect benefit from novel therapies.
• •Applying instruments across contexts without validity confirmation is a key risk.
• •A 7-step consensus framework standardizes outcomes selection in drug development.
• •Early regulatory engagement and framework adoption remain critical unmet needs.

1. Neurosciences - high percentage of negative trials

Advances in understanding the neurobiological factors underlying neurologic and psychiatric conditions have yet to be adequately translated into successful treatments. In neuroscience therapeutic areas, a high percentage of clinical trials fail to demonstrate safety and efficacy, resulting in a lower innovation index, in terms of registered drugs, compared to other therapeutic fields (van der Doef et al., 2018; Butlen-Ducuing et al., 2016). It is widely recognized that, consequently, neuroscience clinical trials consistently show high failure rates across various therapeutic areas. The field is rapidly innovating in novel therapeutics, but at the same time, it needs good standards in designing outcome strategies accordingly. Therefore, advances in drug discovery are not always driven by novel outcome measures that validly capture the benefit or safety of the interventions, increasing the risk of failure when testing promising therapies.

The Outcomes Research in Neurosciences group issued consensus-based recommendations to guide decisions in clinical outcome processes relevant to researchers when designing clinical trials for new therapeutics. The guidance document, which has been described in more detail elsewhere (Zaragoza Domingo et al., 2024a), is an attempt to highlight key activities needed to drive the decision process on outcomes strategy, the risks of shortening the process, to inform regulators and to increase the chances of achieving the regulatory agencies engagement; eventually to obtain more conclusive results from clinical trials data. The issued guidelines are updated regularly, covering novel publications, regulations, and case examples from the field. The graphical abstract summarizes the main challenges in COA innovation and future development plans; Fig. 1 offers a workflow for the process followed by the Network during the development of the framework. A briefing on the proposed standard methodology is presented in Table 1 to provide a more comprehensive overview for readers of the suggested activities.

Fig. 1.

The development of a playbook for selecting, modifying, and developing COAs for neuroscience CTs. An expert survey was conducted to collect opinions on current practices in outcome research among industry partners. The results of this survey were used to further inform the development of the outcome research guidance. Abbreviations: clinical trials, CT; clinical outcome assessments, COAs; central nervous system, CNS.

Table 1.

Briefing of the 7-step standard method.

Briefing Outcomes Research Guidance

Proposed 7-Step Standards for Clinical Outcomes Selection

(Extracted from COA Guidance V. 3).

Step 1 – Disease Model and Background Refers to identifying a disease model and fully describing the background of the research. This includes reviewing existing literature, working on the content validity through qualitative research, defining the unmet need to cover and describe the drug target.

1.1 Review of existing literature Identification of potential outcomes from existing work as information provided to inform the consensus process for next steps.

1.2 Review of existing qualitative studies or disease models Identification of qualitative research work either published or unpublished with patients suffering from the disorder or relatives or caregivers either formal or informal.

1.3 Defining the unmet need (or gaps to fill) Description of the rationale for the development of a COS in the context of what is already known with the focus on the contribution to actual clinical application.

1.4 Describing “drug targets” and “drug actions”: actual/desired Clarification about what is known, the mechanism of action of the new drug under research and the pathophysiological pathway by which the therapeutic effect is thought to be achieved.

Step 2 – Defining the Scope of Use for the Clinical Outcomes Set Refers to defining the Scope of Use for the Outcome Set. This includes defining the condition of interest, the intervention, and the setting or context of use.

2.1 Health condition, condition stage, target patient population Definition of the purpose of the COS, including therapeutic indication and intended use, i.e., to validate a pharmacological intervention. Ultimately, this consists of defining the scope of the COS and its applicability.

2.2 Defining the intervention Specification of how the new drug under research is expected to produce its effect.

2.3 Defining setting Specification of the therapeutic settings in which the new intervention will be conducted as much as possible. An international scope should also be taken into account, considering different health systems.

2.4 COS application in a “nutshell” To summarize the application of the COS within the context of intervention in early-stage RCTs.

Step 3 Stakeholder involvement The goal is to be explicit on what categories of stakeholders should be considered (patients, public, practitioner, policy maker, program manager, professor, and payer) and on the minimum requirements for consensus on the specific COS project.

3.1 Key stakeholders To identify stakeholders that must be included in a consensus COS definition.

3.2 Balancing stakeholders To clarify the composition of the panel groups regarding the variety of stakeholders and the number of participants for each stakeholder and whether a specific balance is needed.

Step 4 Determining “What to Measure” as Dimensions Refers to determining “what to measure” as dimensions, including identifying dimensions, building a disease impact model, setting candidate domains and selecting the relevant ones using a Delphi technique for determining the final COS.

4.1 Defining the method for identifying dimensions Revision and list all key areas that need to be represented as domains/concepts in the core domain set of the condition.

4.2 Organization and structure of domains - Disease Impact Model Organization of the preselected domains/concepts of interest in a taxonomy or theoretical model useful to confirming the coverage of relevant aspects of the disorder. Models including symptoms & signs and impact in daily life.

4.3 Setting a list of candidate domains To build an initial (long) list of outcomes covering each core area, as the starting point for consensus.

4.4 Defining the method for consensus Stablishing the method for achieving consensus among the panel of participants regarding the final core domains (e.g. Delphi panels, etc.).

4.5 Determining final COS/COMs Agreement on the final COS/COM based on consensus as a final vote from the entire group of panelists.

Step 5 - Determining “How to Measure” Outcomes Refers to determining “how to measure” outcomes. This includes defining the concepts of interest, identifying existing instruments, analyzing the known psychometric properties, and summarizing how the instruments cover the dimensions or existing gaps.

5.1 Conceptual definitions of dimensions under consensus Developing an accurate definition of the dimensions that will facilitate the process of finding suitable measurement instruments.

5.2 Identifying existing instruments Identification of existing validated instruments as candidates to measure the dimensions.

5.1.1 Defining the method to select instruments

5.1.2 Quality assessment of psychometric properties of key instruments including content validity thorough cognitive debriefing with patients “content validity confirmation with key stakeholders. Content validity may include patient, caregivers, treating physicians and payers as appropriate'.

5.3 Summary coverage of OMIs for dimension measurement Mapping the dimensions that are covered/uncovered by the existing and selected instruments.

5.4 Considering alternative methods for endpoint measurement Considering the utility of DHTs, Adaptive, Goal Attainment Scaling, etc.,.

5.5 Additional information from post-hoc analysis or RWE databases Use of existing data sets to fill the existing gaps related to the psychometric features of the instruments or dimensions.

Step 6 - Final generic recommendations for COS measurement Refers to the final generic recommendations for COS measurement including suggested clinical outcome measures (COMs) for its use in clinical trials, as primary and secondary outcomes, with mention of special factors to consider and efficiency of the measurement method from recent trials learning from recent applications in research.

6.1 Considerations regarding drug labeling Ascertain whether the final COS will be accepted by regulators to support labeling claims and identify the risks and next steps

Step 7- Conclusions of gap analysis and future steps Refers to describing the conclusions of gap analysis and specifying future actions in view of future work for clinical trials design.

7.1 Selecting the most suitable format for existing OMIs Identify measurement formats suitable for the context and for the clinical trial (electronic –eCOA-, centralized/remote, etc.).

COS Core Outcomes Set, COA Clinical Outcomes Assessment, COM, Clinical Outcome Measure, COU Context of Use, DHT Digital Health Technology, eCOA Electronic Clinical Outcome Assessment, RCT Randomized Clinical Trials, OMI Outcome Measurement Instrument.

This outcomes research Network is composed of independent and interdisciplinary experts and was recognized in 2023 by the European College of Neuropsychopharmacology (ECNP) as a Thematic Working Group and, more recently in 2026, upgraded to a full Network integrated into the existing networks. The ECNP networks are multi-disciplinary collaborative platforms that bring together researchers across Europe to share ideas, discoveries and best practices in translational neuroscience. Networks are designed to facilitate the collection of essential biological, clinical and therapeutic data in a robust and replicable way, covering the range of both disease-oriented and transnosological research in translational neuroscience. The overarching objectives of the network on Clinical Outcomes in Early-phase Clinical Trials are as follows: (i) setting the standardization of outcomes research practices and promotion of its application in clinical trials in neuroscience in the continuum, from early to late phases of drug development; (ii) facilitating the communication between different stakeholders including drug developers, regulatory agencies and patient's advocacy groups; (iii) disseminating by offering educational programs tailored to different users/stakeholders involved in drug development with a special focus to next the generation of scientists; and (iv) providing methodological support to the other condition-specific working groups across different scientific societies.

In a multidisciplinary context for drug development, there was agreement within the group on the potential impact of harmonizing outcome research practices to increase the likelihood of successfully identifying safe and effective treatments in psychiatry, neurology and allied fields. Among others, one major challenge in neuroscience clinical trials is the limited availability of reliable, sensitive, and fit-for-purpose (FFP) clinical outcome assessment (COA) tools. The FDA defines an FFP COA as a measure that has proven to be appropriate to support claims in a specific clinical context, meaning that the measure meets the qualifications for supporting claims outlined in the corresponding regulatory guidance, specific to a study population and protocol, and to the claim sought/hypothesis tested.

In this context, it has been noted that some well-established instruments, commonly regarded as gold standards in clinical trials, may not fully meet the mentioned FFP criteria. However, legacy instruments remain widely used because they are familiar to clinicians and broadly accepted by regulatory authorities. Despite being endorsed, these instruments have limited utility for demonstrating efficacy in certain contexts of use. Moreover, some of the legacy outcome measurement tools exhibit uncertain sensitivity to capture change in specific disease dimensions or concepts, particularly for pharmacological treatments with novel mechanisms of action that are different to those marketed drugs, with added expected benefits in other dimensions of health (Schooler et al., 2023). Therefore, the development of new treatments requires innovative outcome strategies to tailor existing instruments or design novel measurement tools that address validity gaps, thereby capturing clinical benefits from novel therapies.

Experts in COA science agree that fostering awareness of good research practices may increase the likelihood that a COA will successfully demonstrate the efficacy and safety of new interventions, or at least help mitigate potential concerns about measurement methods in the event of negative or inconclusive trials. In recent years, the science of COAs related to clinical trials has been increasingly recognized as a key component in the decision-making processes of research and development (R&D) departments within the pharmaceutical and biotech industries. Therefore, the main goal of our Network is to address and to provide methodological guidance on the decisions around measurement strategy, which is currently lacking in the neuroscience research field, and framed in a way that is in line with current regulatory patient-centered recommendations placing patients and caregivers at the centre of strategies for informing regulatory decisions on new drug applications (NDAs), i.e. FDA guidelines for Patient-Focused Drug Development (FDA-PFDD) (U.S. Food and Drug Administration, 2025) and the EMA initiative on patient experience data (PED).

A couple of case examples of the application of PFDD principles integrating patient-experience data into the NDA can be found in the development of esketamine as an antidepressant agent and in the development of a medical product for spinal muscular atrophy (SMA). In both cases, novel instruments included in the clinical trial design were developed with clear efforts to integrate patients and caregivers, reflecting the needs of the patient community.

The guidance in place, based on 7-step processes, was based on a consensus-building process, which included 22 experts in clinical trial methodologies in neurosciences, either members of the International Society for CNS Trial Methodology (ISCTM) (The International Society for CNS Clinical Trials and Methodology and ISCTM, 2025), and/or of the ECNP Network (European College of Neuropsychopharmacology, 2023). Now, this paper aims to raise researchers’ awareness of the importance of harmonizing outcome science research in clinical trials, and for the first time, to present the diversity of current research practices in COA methodologies across companies gathered through a recent survey of field experts. This paper also aims to disseminate the identified risks associated with streamlining the process, increasing the risk of misleading outcome measurement and jeopardizing the final regulatory acceptance. Overall, the reader will see this paper as an attempt to contribute to achieving more efficient, in the sense of more conclusive, clinical trials in the neurosciences and allied therapeutic areas.

2. Expected impact of promoting good practices in outcomes research

In this new era of patient-focused drug development, outcomes sciences hold a fundamental role in demonstrating the safety and efficacy of novel interventions. The science and recent guidance documents focused on clinical outcome assessments in patient-focused drug development (U.S. Food and Drug Administration, 2025; U.S. Food and Drug Administrationa; Silva et al., 2024) is not only relevant to private pharmaceutical and biotechnological companies, but also to clinical researchers aiming to test novel therapeutic strategies, repurpose drugs, or demonstrate additional benefits of existing treatments.

It is worth noting that the concept of success related to novel therapies is multiple. Not only does it refer to demonstrating efficacy/safety in clinical trials in agreement with regulatory agencies, but also having the capacity to obtain final approval/authorization, and subsequent acceptance of the treatment by coverage and payment post-regulatory endorsement/approval by payers as health technology assessment agencies (HTAs).

Activities needed to adhere to the proposed good practices in outcomes research are to be embedded within the integrated plan for evidence generation (IEP). The IEP is a strategic approach specific to a novel drug, that optimizes drug development and market access by ensuring evidence generation aligns with regulatory, clinical, and market needs; when done in a comprehensive way, allows to identify relevant evidences that are missing or not completely known, therefore helps to prioritize studies ahead, and draw a plan to satisfy diverse stakeholders (regulators, clinicians, payers). The IEP is usually set at the early stages of drug development, when trials offer a unique context for innovation (phase 2), whether conducted by the pharmaceutical industry or investigator-sponsored (e.g., trials initiated by researchers or clinicians). Nevertheless, early-phase trials can also carry significant risk, as there is typically only one opportunity to demonstrate the benefits of the new intervention, which can compel continued investment in its development following a positive outcome.

Good practices for evidence generation in outcomes research start with a comprehensive description of the disease, encompassing not only the clinician's perspective but also that of the patient and their environment (e.g., carers, in-patient clinics, rehabilitation and home settings staff). Based on qualitative research, one can reach an adequate disease definition and build the conceptual disease model that identifies and describes symptoms and signs (proximal and distal to the core condition) and their impact on the patient's life. This analytical process may unveil new facets of the disease that, for their nature, are more relevant to patients and, therefore, require a deep review or even the inclusion of additional concepts other than those already known and planned to measure. In this context, qualitative research is often used to collect patients' data that support the identification of salient concepts in health that are both highly prevalent and scored as highly bothersome by patients. In line with the official guidance from regulators, the current imperative in COA science is to listen to patients within their own environments to learn more about the disorder and how patients feel, function, and survive (U.S. Food and Drug Administration, 2025). The same principle of concept identification, applies to digital health technologies (DHT) based measures, to qualified for its use as relevant endpoints in clinical trials; in this way, it is important to underscore that for existing endpoints or for those new technological endpoints based on sensors a validity plan starts as well with the corresponding condition model, i.e. from health concepts of interest to specific measures as defined later as study endpoints (Weinfurt, 2022).

The resulting conceptual disease models are not always published, potentially due to a range of factors related to the type of research of qualitative nature involving and often with limited number of participants. Nevertheless, informative examples are available in neurology, such as those developed in Parkinson's Disease (Muro-de-la-Herran et al., 2014; Staunton et al., 2022). In psychiatry, the relative scarcity of published conceptual disease models may reflect challenges related to limited sample sizes and concerns regarding representativeness. Despite these constraints, such models remain valuable as foundational frameworks. Importantly, they highlight meaningful opportunities for clinicians to advance future qualitative research initiatives, engaging patients and caregivers across psychiatric and related disorders.

3. Need for harmonization and standardization of outcomes research practices

Considering the importance of selecting outcomes in clinical trials, standard frameworks for COA/COS selection have been elaborated by other groups, including OMERACT, COMET-COS-STAD, and COSMIN. Besides these valuable sources for researchers, given the uniqueness of neuroscience therapeutic areas, our proposed approach represents the first attempt to provide guidance specific to this field, based on existing frameworks, with a special focus on early-phase pharmacological clinical trials and maintaining consistency across all development phases. The methodology proposed by the Network is presented as a playbook to mitigate methodological and regulatory issues that remain in existing initiatives, which were originally conceived for clinical research to ensure standardization of key outcomes relevant to patients, clinicians, and policymakers, improving evidence quality for healthcare decisions closer to clinical practice. The 7-step approach emphasizes the integration with the specific needs of drug development, which in early phases are more oriented to signal detection as proof of concept linking clinical outcomes to the compound mechanisms of action while maintaining alignment with regulatory requirements. This means that, rather than repackaging already established principles, the recommendations specified by this Network serve as a tool to operationalize them, reinforcing content validity and psychometric rigour when selecting/developing measurement tools and, as said, providing practical guidance to implement outcome measure strategies consistent with current regulatory review processes in place.

Two main applications of this framework are clearly foreseen as worth distinguishing, based on the need either to: (i) guide the selection of COAs for specific study protocols for clinical trial study design, or (ii) guide initiatives aimed at comprehensively identifying and defining COS.

In this context, COS is understood as the minimal set of outcomes and recommended COA instruments to be consistently included in all clinical trials for the same condition using a consensus-based method, thereby facilitating the comparability of results across studies and serving as a benchmark for future research in that condition.

Moreover, potential situations where the 7-step method is applicable are listed here, situations where the decision-driven process can impact the final outcomes strategy:

• a)For conditions where there is a need to find substitutes for legacy instruments, which might or might not be considered gold standard, but, as legacy, the COA tools are broadly accepted without adhering to all regulatory requirements (fit-for-purpose, not completely validated).
• b)For conditions facing challenges when testing novel therapies, defined as, for instance, a different clinical context than the usual. For example, this may include disease-modifying therapies that alter the condition's trajectory, thus slowing down disease progression. In summary, where the benefit of treatment is expected not to be well captured with existing endpoints or tools for any reason.
• c)For novel biomarkers, either identified in specific conditions or as part of precision therapy, that are thought to be useful for different purposes (e.g., diagnostic value, staging disease severity, prognostic for evolution) but require validation for their clinical utility. Examples of such cases are in Multiple Sclerosis, Parkinson's Disease GBA1 mutation or Huntington's Disease, to mention some.
• d)In the transdiagnostic approach, when targeting neurobehavioural constructs that are shared across categorical diagnoses (e.g., anhedonia in depression, anxiety, schizophrenia, PTSD), which creates the need to provide evidence of benefit across different conditions.
• e)For clinical validation of DHTs, collecting novel outcomes and digital phenotyping through electronic devices (e.g., vocal markers for depression or digital phenotyping for negative symptoms in schizophrenia, among others in development).

Besides its relevance for the pharma industry and biotechnological companies, this method is also addressed to researchers in general, including clinicians or healthcare practitioners, aiming to independently define outcome measures when designing a research trial to explore the efficacy and safety of novel interventions (including drug repurposing) or when planning to develop a novel measurement instrument based on the clinical experience (See Supplementary F1 for a template protocol for outcomes research).

4. Convergence of psychiatry and neurology: can common processes apply?

Clinical trials in the neurosciences and allied fields cover a variety of conditions that are quite unique in nature and require addressing issues not found in other therapeutic fields, such as the subjectivity of clinical measures or the lack of biomarkers for diagnostic or monitoring purposes.

In general terms, the neurosciences encompass neurology and psychiatry, and extend to related areas such as immunology and neuroinflammation. However, one could argue that these fields include conditions that are completely different, with some commonalities. For example, these conditions share signs and symptoms and similar measurement methods characterized by a certain level of subjectivity, i.e. from those based on the clinician's evaluations (ClinROs), those based on patient-reported measures (PROs), or reported by others like observers or proxies (ObsROs/Proxies).

In neurosciences therapeutic fields where the central nervous system (CNS) is involved, but also in the conditions where a biomarker is available (e.g. in Multiple Sclerosis (MS), but in order to reach acceptance of a new drug application, regulators also require developers to demonstrate benefit (i.e., how the patient feels, functions and survives) based on patient-centric measures. This poses a special methodological challenge, offering a variety of situations: conditions with well-known biomarkers (e.g., neuroimaging in MS), those without a biomarker, or those with potential biomarkers that have not yet been validated. As for subjectivity, conditions such as mood disorders or pain-related conditions, what the patient reports is actually considered feasible and reliable. However, one can find other conditions where patient awareness of severity and impact might be compromised, thus PROs not being a reliable measure of drug effects, as can be the case of psychosis, dementia or even, to some extent, stroke.

Alternatively, neuroscience conditions can also be classified based on the availability of objective markers for diagnosis and severity assessment, distinguishing between conditions that depend on patient-reported outcomes (PRO) and those that are not. In this way, PRO diseases are defined as those for which the approval of drugs traditionally relies on the benefit measured by PROs as primary or secondary endpoints. Otherwise, approval of novel treatments for diseases that traditionally rely on survival, biomarkers, or COAs other than PROs for evaluating treatment benefit can be classified as non-PRO-dependent.

5. Standard good practices adoption by regulatory agencies in the USA and Europe

Regulatory agencies in the USA and Europe are focused on promoting the main methodology for collecting patients' experience data (PED). Examples of such efforts are the progressively published versions of the FDA's PFDD guidance (4 issued), including substantial details on the overarching principles and methods for data collection and reporting (U.S. Food and Drug Administration, 2025). This set of guidance documents helps to ensure that COAs used during medical product development measure what really matters to patients, clearly define what is being measured, appropriately evaluate the effectiveness, tolerability, and safety of treatments, and ultimately avoid misleading marketing claims. The last PFDD Guidance series #4 was finalized in October 2025, and focuses on Selecting, Developing, or Modifying Fit-for-Purpose Clinical Outcome Assessments (U.S. Food and Drug Administrationb). In line with the FDA, similar recommendations have been issued in Europe, with a recent reflection paper on PED being published by the EMA/CHMO Committee of Human Medicinal Products (European Medicines Agency (EMA)). The release of these guidelines is timely, coinciding with the field's pressing need for innovative approaches to evaluating clinical benefit, with a particular emphasis on the patient's experience.

A couple of controversial situations can be mentioned, showing how current practices in neuroscience outcomes can fail regulatory expectations. A first controversy concerns legacy instruments commented on earlier, which predominantly rely on clinician evaluations (i.e., ClinROs type of COA) to screen for symptoms and rate their severity. Since outcomes are reported by clinical practitioners, these instruments lack patient focus and are therefore far from regulatory recommendations.

A second common practice in outcomes research is the reuse or repurposing of instruments, when the decision is to modify existing instruments or to transfer assessment tools already validated in one specific condition or study population to others. This practice poses validity risks, mainly related to face or content validity of the COA used in a context other than the original validation, exemplifying one of the many issues stemming from current practices when designing clinical trials. During internal Network meetings, other pitfalls encountered in decision-making regarding outcomes were discussed. These included the assumption that a given measurement instrument maintains the same psychometric properties when used in different clinical contexts, even within the same condition (e.g., acute versus chronic), or when used in different conditions sharing the same symptoms. In such cases, it is necessary to conduct validation programs to confirm patients' accurate understanding of the items (cognitive debriefing studies) and to assess the instrument's metrics in every novel clinical context. This second concern is relevant given the current trend toward a more patient-centric approach, in which PROs for measuring efficacy and safety are increasingly being elevated to primary endpoints in clinical trials, with statistical tests ordered accordingly.

In parallel to the relevance of PROs to collect patients’ feedback, there is an increased interest in collecting more objective data in more ecological and continuous formats to minimize the influence of the recall period between study visits. Examples of such are, by including DHTs devices, improving accuracy and time-resolution of the measurements, either for digitalizing qualified instruments or including sensors to collect passive data for novel endpoints. Since DHTs can demonstrate both the clinical meaningfulness of what is measured and the sensitivity to detect benefits in interventional trials, these novel methodologies are likely to gradually replace subjective reporting from all sources of information. For all described scenarios, the proposed 7-step standard process applies to all types of COAs, including DHTs, and outlines the fundamental steps.

Innovation in outcomes, starts and ends, as mentioned above; from a thorough description from patient's experience on how the disease impacts the patient, and the description of the therapeutic background (Step 1), thorough all the other activities up to the conclusion of a future research roadmap to validate the fitness of the instrument as a final summary of all the necessary processes required to progress in the therapeutic field (Step 7), as summarized in Table 1.

Whenever possible, activities needed to support a COA strategy should start before early-phase studies are even planned, ideally during the elaboration of the EIP, to allow enough time to collect the evidence needed to secure regulatory acceptance of new COAs. This evidence might require patient participation in interviews or focus groups, known as qualitative research, either to understand people with lived experienceof the disease or to confirm the instrument's content validity. In such cases, having patients participate in the cognitive debriefing of instruments/items intended for use in the trial ensures the correct understanding and adequacy of all items.

6. Impact to streamline the process for clinical outcomes decisions in neurosciences clinical trials

The framework produced by this Network ensures the collection of evidence needed to develop the COA dossiers for discussion and submission to regulatory agencies, and to document the validation process for later publication and acceptance. To increase the relevance of this initiative, the network considered it relevant to share the expert's opinions on the potential risks or pitfalls which may occur when an innovative COA strategy does not completely adhere to the standard process, which helped to identify persuasive topics useful to widespread its use by sponsors and researchers involved in trial study design. As a result, beyond the risk of failing to capture specific evidence of treatment benefit that could be relevant for future NDAs, additional concerns were suggested in a dedicated survey (Zaragoza Domingo et al., 2024b), summarized as follows:

• •Describing the disease impact model and therapeutic background (Step 1/7)Omission of this step entails the risk of “missing” relevant information about the disease and its impact on the patient and families.
• •Defining the scope of use for the COS/Outcomes (Step 2/7)If not done, there is the risk of distorting the final application of the study results in several ways, such as outcomes intended for clinical practice may not be appropriate for clinical trials.
• •Deciding on stakeholder involvement (Step 3/7)If not comprehensive enough, the absence of key stakeholders (clinicians, patients, carers, practitioners, policy makers, payers, etc.) can significantly impact the final generalizability approval/authorization, reimbursement, or even drug access to country markets.
• •Determining “What to Measure” in terms of domains or concepts of interest to be measured for efficacy and safety (Step 4/7)If this research is inaccurate, it can impact the specificity regarding the health domains ultimately measured. Therefore, relevant domains (meaningful aspects of health –MAH) and corresponding concepts of interest (COI) may be missed, including even the most important outcomes.
• •Determining “How to Measure” the outcomes, including validation principles (Step 5/7)If the outcome measurement instruments are not adequately selected, there is a risk of using instruments that do not capture the outcome domains, are insufficiently sensitive to detect clinical response, or fail to meet all technical or regulatory requirements.
• •Making final generic recommendations for optimal COS measurement (Step 6/7)This step aims to address the risk of repeating past errors. When lessons learned from previous lessons are not applied, there may be a lack of clarity about the rationale for the selection process, or, eventually, progress in the field can be delayed.
• •Drawing final conclusions concerning gap analysis and future research in the therapeutic field or condition (Step 7/7)If this step is skipped, there is a risk of producing insufficient evidence to support the use of endpoints, missing information necessary to advance innovation in outcomes and endpoints and failing to translate these benefits for patients.

For an infographic summary of the process and messages to take home for risks, see Supplementary Fig. S1. Moreover, all comments collected are included in Supplementary Table S1.

7. Current practices in drug development - survey results

During Network discussions, it was noted that, although regulations apply to all NDAs, a variety of practices for determining COA strategy exist. Overall, there is a global perception that the same standards are followed, but a closer look revealed nuances and differences in research practices across sponsors and researchers. With this in mind, the Network sought to achieve consensus on its acceptance, explore the level of endorsement, and learn more about the challenges to its implementation within the pharmaceutical and biotechnological industries.

With this purpose, a survey was distributed to experts in outcomes research involved in COA measurement decisions to solicit feedback on the proposed 7-Step process, the level of agreement and endorsement, and the expected challenges when implementing the full version of the standards. To ensure knowledge alignment, all participants were asked to read a summary of the 7-Step process before answering the survey (See Table 1). Open-ended questions were posed, including one question about the level of agreement with the method, level of endorsement and expected challenges during attempts of implementation of the method. An initial qualitative textual analysis of the open-ended questions is presented as a word cloud using the Voyant tools software (https://voyant-tools.org/).

A total of 26 experts participated in the survey, of whom 81% had experience defining COA measurement strategies in clinical trials, and importantly, 60% reported more than 10 years of experience. A total of 48% of participants represented pharma companies, 19% CRO and 19% from academia, and 12% equally distributed from health digital technology industry representatives, non-governmental organizations, healthcare providers and patient advocacy groups. All participants declared having read the pre-survey documents (see Table 1) and confirmed their acceptance. Noteworthy, a 30% considered worth including additional activities other than the ones listed, as for instance, considering the estimated evaluation time needed to conduct all the assessments as part of the strategy description and to specify the hierarchy of the COA measures, i,e. Primary/secondary objectives in the clinical trial design. In fact, the time to complete the evaluations has an impact on patient, caregiver, and research site staff burden, and thus negatively influences the quality of data derived from the COA. The hierarchy of the analysis has an impact on the statistical power by influencing sample size assumptions and the chances to find significant results, if there is a benefit.

Results on endorsement by the question “From your experience, to which extent is the pharmaceutical industry currently using the full 7-Step process?” a high variety of opinions were collected regarding how much the industry is currently adhering to such standards; answers were categorized into “not at all” used to “fully aligned” with them, therefore, showing substantial discrepancies as for current practices. A tag-cloud is presented in Fig. 2 with ideas collected and further categorized to offer descriptives. Overall, the answers pointed towards a substantial trend to customize the processes, and, more importantly, the low extent to which the full process is used in practice. In general, this result would support the need for efforts to harmonize drug development practices regarding outcomes strategy decisions.

Fig. 2.

Main survey results on the use of standard practices in outcomes strategy in study design, and on how to encourage their adoption (n = 26). Written responses to open-ended questions were categorized and plotted for frequency in a pie chart (A) or transformed into word clouds (B). The font size of the word cloud text reflects the frequency of term use in open-ended responses (high font size = higher use frequency).

Interesting suggestions to encourage the adoption of the 7-Step guidance were proposed, as summarized in Fig. 3 A. More specifically, these included promoting dissemination and engaging in educational programs and activities, framing the method as a playbook aligned with existing regulatory guidelines, and highlighting the advantages for the pharmaceutical industry to adopt the full methodology in terms of time and cost-effectiveness compared to other strategies (Fig. 3 B). Moreover, it was suggested that the method be piloted in real drug development programmes to determine whether it works and whether adaptations are needed. All answers provided by participants are included as Supplementary Tables S2–S3. All the suggestions received allowed us to describe a roadmap of actions for our Network to follow moving forward to have an impact in the field.

Fig. 3.

Main survey results on how to encourage the adoption of the proposed methodology (n = 26). Written responses to open-ended questions were categorized and plotted for frequency in a pie chart (A), or transformed into word clouds (B). The font size of the word cloud text reflects the frequency of term use in open-ended responses (high font size = higher use frequency).

8. The intriguing future of innovative endpoints for innovative therapies

This work represents the advocacy for the first published guidance on the standardization of clinical outcomes decisions in neuroscience for more effective and harmonized practices in neurologic, psychiatric, and related conditions to enhance the likelihood of success for novel treatments.

Although the private pharmaceutical and biotech industries may assume the risk of not achieving the desired results and shift quickly to other pipeline assets when a treatment's benefits or risks are unclear, this should be regarded as a societal loss. This means that in some instances novel treatments can be prematurely abandoned, and it can be argued that, if properly evaluated, benefits might emerge. Conversely, any single advance in novel outcomes and endpoints intended for testing novel mechanisms of action or novel therapies, eventually represents a significant achievement that can benefit the pipeline of many industries and researchers working within the same therapeutic area. A more accurate look at the outcomes selected to prove efficacy and safety in early trials (i.e. ensuring their full validity, before pivotal trials are planned) is proposed in this paper as a way to increase the chances of success. At least, we advocate a potential way to increase the efficiency of clinical research by improving current practices for outcomes selection, provided that the novel therapies have a chance to benefit patients. In this context, it is worth mentioning that “absence of evidence is not evidence of absence”, in the sense that insufficient validated outcome measures, impact on poorly designed studies that can result in statistically or clinically irrelevant differences, supporting investor decision-making, and due to risk of failure, preventing progress in further development stages.

As already commented, common gaps on this line are: the use of instruments that were created to monitor drug treatments in one clinical context and then are used in a different one, impacting their sensitivity to find change and face validity, or those used to measure expected effects of a specific type of treatment mechanism of action, but not measuring dimensions that novel therapies might impact. However, conducting all required literature and COAs reviews and planning, and applying a consistent validation process that includes innovations in outcomes research, requires a start time ahead of ph2b and substantial resources.

In this regard, pre-competitive collaborations encouraged by the Innovative Health Initiative (IHI) of the European Commission (formerly known as Innovative Medicines Initiative –IMI-), or also offered from other programs or institutions, such as Critical Path Institute (C-Path), the National Institute of Health (NIH) Patient-Reported Outcomes Measurement Information System (PROMIS), or the International Consortium for Health Outcome Measurements (ICHOM), have a long trajectory and have already enhanced the field of developing novel outcomes in several indications, mainly PROs.

In reality, novel outcomes face significant challenges in gaining traction, highlighting the need for a bridge between sponsors and regulators to prevent regulatory rejections. Efforts to build a consistent COA plan aligned with regulatory early advice would ultimately save time for all stakeholders and accelerate drug development.

9. Regulatory rejections: insights from measure validity and study design issues

Examples of regulatory rejections can be found in review papers analysing the long-term success of a strategy. A review by Demuro et al. (2012) of labelling claims concluded that the primary reasons for regulatory denial included concerns about the validity of the selected measure, with the fitness of the measure accounting for more than 38% of denials. This was followed by issues related to study design, data quality, or interpretation of results, which were the second-largest identified category, accounting for approximately 27% of denials. As for validity concerns, the FDA specifically questioned the content validity and/or validity of measurement instruments in general, other issues related to the rationale for recall periods, and evidence of appropriateness of translations for use in multinational studies.

More recently, Gnanasakthly et al. (2021) conducted a review showing a clear increase in the inclusion of PRO instruments in the labelling claims, where the authors distinguished between applications for drugs and the application of new molecular entities. During the period 2016-2020, PRO-related labelling appeared in 60 out of the 228 drugs (26.3%), including statements based on PRO endpoints, and for new molecular entities, up to almost half, in 47 out of 90 (50%). As for central nervous system drugs, of the 34 recently approved drugs, 18 (52.9%) included a PRO in the labelling.

In spite of this high rates of PRO mentions in NDAs, a deep dive on this mention showed that the trend to mention PRO measures was often acompaigned by a lack of evidences supporting the appropriateness for the context of use of the instrument used; also, often it was identified an issue with the scoring algorithms used for the PRO analysis presenting a total score as the result of a multidimensional measure. This dilutes the results on the core experience of the disease with many less relevant concepts. Therefore, the PRO-related statements included in the drug labelling in these situations present a seemingly mixed message that may confuse or undermine the intended messaging regarding the benefit of the new treatment.

Additional information about the long run can also be obtained by examining marketed oncology drugs. Sams et al. (2023) published a review of FDA warnings and untitled letters for clinical outcome assessment claims between 2013 and 2021, identifying that 25% of the letters had clinical outcome assessment infringements (23% with PROs issues) on promotional materials that did not adhere to labelling claims (Sams et al., 2023). The authors also observed that, overall, the pharmaceutical industry's and researchers' increased awareness of FDA COA guidance requirements had a clear impact on the success of labelling strategies. In this way, the fact that regulatory agencies during recent years have set special early interaction opportunities, as processes for the pharmaceutical industry to discuss their approach to COA during the investigational new drug application, seeking advice and alignment to ensure a robust approach, has subsequently increased the likelihood of a successful labelling claim. Also, regulatory initiatives that include issuing guidance documents and increasing the support to sponsors by specific divisions are addressed to ensure good clinical and research practices. Recently, the EMA issued a reflection paper encouraging stakeholders and developers to engage at an early stage with regulators to discuss the inclusion of PED in medicine development programs and marketing authorization applications (European Medicines Agency (EMA)). Besides the encouraging increase in featured PRO-related statements in NDAs, this does not seem to reflect the increasing emphasis being placed on patient centricity in drug development. In future years, the pharmaceutical industry needs to be aware of how successfully this evolves and how all these issues are mitigated by early engagement of regulatory stakeholders in the selection of COA strategies.

Conclusions

The endorsement of standard methods in outcomes science, as the one produced by our Network commented on in this paper, constitutes an opportunity to ensure good communication with regulatory agencies when proposing drug development plans for NDAs either in the USA or Europe and to adhere to good research practices, already underscored by other organizations in recent decades.

The diversity of current practices documented by our survey also involves an important investment of resources and time ahead, but with clear advantages when starting in parallel with phase 1 studies consistently along early phases, which serves to collect sufficient evidence to inform study design in phase 2 trials and, most importantly, to pivotal clinical trials. Overall, our group agrees that the main risk for a negative or inconclusive trial result is to rely on legacy instruments that are not FFP and disregard the patient's experiences. Nowadays, these instruments would not pass quality scrutiny and might thus constitute an impediment to adequate sensitivity, thereby reducing the chances of acceptance and approval of novel therapeutics by clinicians and patients in the clinical context. In order to increase the chances of measuring what is intended to measure and relevant to the patient, the appraisal of the existing legacy instruments and the innovation on measurement strategies should go in parallel with the discovery of novel compounds with novel mechanisms of action, which in turn can impact additional disease dimensions. We propose covering all 7 steps to collect evidence needed to inform the drug development process, where each step can be completed at different levels of accuracy, but skipping any of them should be discouraged, as it is likely to cause issues and delays in market authorization. A number of actions are required to disseminate, educate, and, most importantly, highlight the advantages of its use in drug development for the pharmaceutical and biotechnological industries.

Financial disclosures

Silvia Zaragoza Domingo developed the EPICOG-SCH Brief Cognitive Battery in Schizophrenia and provided services as a consultant to the pharma industry (Lundbeck, Sanofi-Genzyme, Jazz Pharmaceuticals and Merck Healthcare and Alexion in different periods, 2018-26). Jan Kottner, Denise Bury-Maynard, and Daniella Tinoco has provided services as consultant to pharma industry (Sanofi-Genzyme, Novartis and Merck Healhcare). Dr. Pinkham has received consulting fees and travel reimbursements from Boehringer Ingelheim.

Disclaimer

For Florence Butlen-Ducuing was an employee of EMA during the conceptual design, drafting and review of the study. For Florence Butlen-Ducuing and Pavel Balabanov, the views expressed in this document are the personal views of the authors and may not be understood or quoted as being made for or reflecting the position of the European Medicines Agency (EMA) or any of its committees or working parties.

Funding

Publication of this paper has been possible with the collaboration and funding of the European College of Neuropsychopharmacology (ECNP), Netherlands.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:The authors declare the following interests. Silvia Zaragoza Domingo developed the EPICOG-SCH Brief Cognitive Battery in Schizophrenia and provided services as a consultant to the pharma industry (Lundbeck, Sanofi-Genzyme, Jazz Pharmaceuticals and Merck Healthcare in different periods, 2018-25). Jan Kottner, Denise Bury-Maynard, and Daniella Tinoco have no conflicts of interest. Amy Pinkham has received consulting fees and travel reimbursements from Boehringer Ingelheim. All authors declare that there are no additional conflicts of interest, including financial, personal, or other relationships with individuals or organizations within the past three years, that could inappropriately influence, or be perceived to influence, the work submitted.

Appendix A. Supplementary data

The following are the Supplementary data to this article: