In this paper, we build upon the Clinical Evidence 2030 framework [1] to propose a structured and stakeholder-driven approach to evidence generation. In recent years, AstraZeneca (AZ) has been aligning its evidence generation initiatives with this methodology [2], aiming to refine evidence strategies that not only enhance regulatory and value and access pathways but also drive meaningful improvements in patient care. A critical success factor in implementing these initiatives is to have a strong contribution, from the methodological point of view, in their design, execution, and analysis. For instance, our team at AstraZeneca Spain defines and decides the statistical models and analytical frameworks to be used to guide data collection, interpretation, and the integration of RWE into the broader healthcare ecosystem. This methodological support ensures that the evidence generated is scientifically rigorous, relevant, and actionable across diverse therapeutic areas.

The variety of study types presented in this article (Fig. 1) allows the creation of specifically designed evidence plans aligned with the following two complementary pillars: (1) therapeutic area knowledge improvement and (2) strategic use of evidence for the company. These two interrelated components work together to drive evidence-based decisions that align with stakeholder needs, including HCPs, patients, and regulators. The therapeutic area knowledge improvement component focuses on better understanding disease progression, unmet needs, and treatment gaps within each therapeutic area. This foundational knowledge informs the strategic use of evidence, where the goal is to leverage the generated data to support personalized patient-centric decision making, market access, and the optimization of patient care pathways.

Fig. 1

Overview of types of evidence according to the two pillars of evidence generation. RWE real-world evidence, TA therapeutic area

Based on this experience, we present here a roadmap that aligns with the strategic transformation of MA (Fig. 2), incorporating examples of studies, innovative methodologies, and advanced data visualization techniques (Table S1 of the Electronic Supplementary Material). Importantly, the strategic evidence-generation plan is developed and refined using diverse internal and external inputs to ensure alignment with stakeholder decision needs. Internally, prioritization, study design, and methodological/statistical considerations are discussed within a cross-functional governance model involving evidence generation/MA and, as relevant for each use-case, regulatory affairs, value and access, commercial and other functions, to ensure that research questions, quantifiable objectives, and analytical approaches are fit for purpose and compliant.

Fig. 2

Proposed roadmap of evidence generation. HCP healthcare professional, P&R pricing and reimbursement, RWE real-world evidence

Externally, HCP perspectives are systematically incorporated through structured interactions (e.g., advisory boards and expert consultations) and are embedded in most projects. Patient perspectives are also incorporated by collecting their views in patient-centered research approaches, including ethnographic studies and the use of patient-reported outcome measures, which directly inform both evidence needs and study design. In addition, expectations from regulators and other relevant agencies are considered through applicable guidance and cross-functional input, helping ensure that evidence plans remain aligned with evolving stakeholder requirements.

2.1 Enhanced Therapeutic Area Knowledge

Innovations in clinical management are essential for improving outcomes and optimizing therapeutic approaches in complex diseases. Real-world studies that analyze disease characteristics, treatment patterns, adherence/persistence, and healthcare resource utilization provide actionable insights that can refine clinical practice and guideline development.

Our multi-faceted approach to evidence generation comprises a broad spectrum of study modalities to deepen therapeutic area knowledge and inform clinical management strategies. Depending on the sponsor type, studies can be categorized as externally sponsored research, whereby AZ provides financial support to study proposals meeting a high standard of scientific and methodological quality; and AZ-sponsored studies, which offer high strategic value and provide a flexible framework that supports diverse scientific objectives, accomplished in close collaboration with renowned external experts/institutions. These studies address data gaps in a range of areas, including epidemiology, disease management, effectiveness and tolerability of treatments, real-world treatment patterns, healthcare resource utilization, and predictive analytics. Furthermore, they can also evaluate changes occurring in the context of transforming care projects, designed to improve patient journeys and care processes. Finally, patient-centered studies offer an invaluable perspective of the patients’ experience of their own disease and its management.

2.1.1 RWE Generation About Diseases and Their Management

Different methodological approaches and study designs may be followed to investigate diseases and their management. In Spain, prospective observational studies involving medicines are not allowed when inclusion criteria require patients to be treated with a specific drug. This regulation aims to avoid a potential inducement to prescribe. Therefore, observational studies of this type must have a retrospective design, where the decision to treat a patient is clearly separated from the decision to include them in the study. As a result, database studies, chart reviews, registries, and hybrid designs are the most frequent designs used in the field of RWE generation.

Database studies represent a very convenient option in this area, as they have the advantage of increasing our understanding about a disease and its clinical management in a fast and versatile manner. In Spain, these projects draw on a variety of data sources, including proprietary and external databases managed by companies such as ATRYS [3], TELOMERA [4], and IQVIA [5], as well as established patient registries, retrospective chart reviews, and clinical trial data.

The following studies recently conducted by our company exemplify this approach and used the BIG-PAC® database, managed by ATRYS. The PATHWAYS-HF project analyzed a large cohort of patients with heart failure (HF) in Spain, providing valuable insights into the prevalence, clinical phenotypes, comorbidities, and treatment optimization in HF with reduced ejection fraction [6]. Results revealed substantial opportunities to improve guideline-directed therapy and reduce mortality and hospitalization rates.

The EXACOS-CV study is an international initiative assessing the risk of severe cardiovascular events after chronic obstructive pulmonary disease (COPD) exacerbations [7]. In the Spanish cohort, exacerbations were associated with a markedly increased short-term and longer term risk of severe cardiovascular events, underscoring the need for proactive prevention and management of these episodes [8]. Similarly, Alcázar-Navarrete et al. performed a real-world comparative analysis of single-inhaler triple therapy (SITT) versus multiple-inhaler triple therapy in COPD and showed that SITT was associated with better treatment persistence, fewer exacerbations, and lower healthcare resource use and costs [9]. These findings have contributed to the endorsement of SITT in international clinical guidance, such as the Canadian COPD recommendations [10], illustrating how RWE generated in Spain can inform best practice globally.

Finally, the CaReMe chronic kidney disease (CKD) study utilized data from 2.4 million patients across 11 countries, including Spain, to evaluate the prevalence of CKD, adverse outcomes, and associated costs [11]. It highlighted a global adult CKD prevalence of approximately 10%, with significant underdiagnosis and suboptimal treatment rates. This multi-national perspective underscores the broad public health impact of CKD and the need for early detection and intervention strategies.

Other national databases used in our studies include Telotron® [4] and IQVIA EMR [5]. Telotron® was used by Martínez-Montoro et al. in an analysis of the epidemiology and burden of CKD among people with type 2 diabetes mellitus (T2DM) managed by endocrinologists (ENDO-CKD study), which highlighted a high prevalence and clinical impact of CKD and the need to improve its screening and recording [12]. The IQVIA EMR database was used in the OCS-HeatMap study to reveal marked geographical disparities in oral corticosteroid use among patients with severe asthma in Spain, supporting the need for region-specific oral corticosteroid-sparing strategies and alternative treatments that reduce corticosteroid-related morbidity and healthcare costs [13].

Observational RWE studies based on chart reviews represent another type of design in this field, playing a pivotal role in understanding how medicines perform in routine clinical practice, complementing clinical trial data with insights into effectiveness, safety, and patient outcomes across diverse populations. These studies facilitate an evaluation of therapeutic benefits in real-life settings, capturing the heterogeneity of patient populations, adherence patterns, and long-term outcomes, thereby supporting informed decision making by clinicians, payers, and regulators.

The ORBE II study used a chart review design and exemplifies an innovative approach to RWE generation in severe eosinophilic asthma [14, 15], focusing on a study’s multidisciplinary collaboration (i.e., broad and diverse group of expert collaborators, both internal and external) and comprehensive data integration providing robust evidence supporting the effectiveness of benralizumab in real-world settings, helping to inform treatment strategies and health policy. The study also highlighted the value of biomarker stratification in predicting treatment response, demonstrating that blood eosinophil count is a robust predictor of clinical benefit [16].

Another example of this type of chart-review design is the ORESTES study, a large multicenter, retrospective, real-world study in Spain that described outcomes in patients with COPD initiating SITT combining budesonide, glycopyrronium, and formoterol [17]. In a cohort of 718 complex high-risk patients, ORESTES reported reductions in exacerbations, use of rescue and additional COPD medications, and healthcare resource utilization after treatment initiation, suggesting clinical benefit and high treatment persistence in routine practice. Importantly, it represents the first real-world study of SITT of this magnitude conducted in Spain, despite this being the third SITT to reach the Spanish market, illustrating how large national RWD studies can complement trial data and inform COPD management. Another example is the ongoing AZAHAR study, which uses electronic health record data to describe the characteristics and real-world outcomes of patients with systemic lupus erythematosus initiating anifrolumab in Spain [18]. Together, these studies illustrate how RWE on innovative management strategies can drive improvements in patient care, inform guideline updates, and shape MA initiatives aimed at optimizing treatment pathways and resource allocation.

There are also mixed models between database and chart review studies. An example is the COHERENT framework, which provides an innovative and adaptable visual model for tracking patient clinical outcomes, healthcare resource use, and related costs over time. In HF, it demonstrated differential outcomes and costs according to primary versus secondary HF diagnosis, facilitating resource-planning discussions [19]. When applied to COPD, the model identified high mortality and readmission rates alongside a substantial economic burden, emphasizing the need for targeted care programs to improve outcomes [20].

The generation of RWE on diseases and their management would not be complete without the investigation of healthcare processes and their transformation. An integral part of the advanced evidence generation model within MA departments are the transforming care initiatives aimed at optimizing clinical pathways and patient journeys. To assess the long-lasting impact of these activities, it is crucial to systematically collect RWD that provide evidence of changes in practice, efficiency, and patient care. AstraZeneca Spain has positioned itself as a strategic partner to the healthcare ecosystem, focusing beyond the availability of medicines. This approach is embodied in the CARABELA initiatives [21,22,23,24], where the analysis of care processes leads to the implementation of solutions addressing identified areas for improvement.

Correct monitoring of transforming care initiatives entails following the progress of those processes for which practical solutions have been implemented in each setting. In certain cases, the most appropriate method of evaluation is through formal observational studies that capture process-related and medical data from patients. A typical example is the CRIERFAC study, by Salar Ibáñez et al. [25], which evaluated the implementation of a CKD screening program in community pharmacies, to alleviate the burden in primary care centers and facilitate or accelerate diagnosis in patients at a high risk of developing the condition. An earlier CKD diagnosis can lead to improved quality of life and reduced morbidity by enabling a timely intervention. Although the RWE generated through this type of study is robust, such research can be lengthy and resource intensive. As an alternative, gathering insights or subjective perceptions from HCPs, through surveys or directly from MA in-field teams, can provide a clearer understanding of the real transformation in healthcare processes, often in a quicker and more cost-effective manner.

2.1.2 Health Economics and Outcomes Research Modeling

Health economics and outcomes research (HEOR) studies translate clinical evidence into quantified impacts on resources, costs, and long-term outcomes, addressing questions that cannot be answered by routine epidemiology or a database analysis alone. By linking RWD with decision-analytic or stochastic models, they estimate the clinical and economic consequences of alternative strategies, support value-based P&R, and guide prioritization of healthcare investments.

The PROMETHEUS model represents an example illustrating this added value for COPD management in Spain [26]. Drawing on patient-level characteristics and event rates from two clinical trials, as well as Spanish cost and epidemiology inputs, PROMETHEUS simulated two scenarios for 2024–33: maintaining current treatment patterns versus expanding SITT in line with guidelines. Despite acknowledging inevitable modeling assumptions and data limitations, the analysis projected that a broader SITT uptake would reduce exacerbations and all-cause mortality, yielding almost €1 billion in cumulative savings over 10 years. Such forward-looking estimates, impossible to obtain from a prospective study of comparable scope and horizon, provide compelling evidence for payers and clinicians when updating formularies, refining clinical pathways, or negotiating outcome-based agreements.

Beyond COPD, similar HEOR frameworks can be adapted to other therapeutic areas to explore the budget impact, cost effectiveness, and health-system sustainability. For example, the INSIDE-CKD study used a microsimulation model integrating national demographic and clinical data to project the future clinical and economic burden of CKD in Spain [27]. The study showed that both CKD prevalence and associated costs are expected to rise markedly over the coming years, with a substantial share of the economic burden driven by renal replacement therapy, underscoring the need for earlier detection and optimized management.

2.1.3 Predictive Analytics

An effective RWE-generation strategy must extend beyond the description of diseases and their management in real-world settings. It should include the development of robust predictive models capable of identifying patients at a higher risk of complications, disease progression, or non-response to treatments, as well as those most likely to benefit from interventions. These models help uncover factors associated with disease prognosis, particularly those that can be influenced by timely intervention.

We recently developed a predictive framework, Predict 2 Prevent (P2P), to explore the use of unstructured electronic health record data for predicting CKD development within 2 years in people with T2DM. The framework builds on the DIABETIC@ study, conducted with Savana Med, which used natural language processing and machine learning to analyze unstructured electronic health records from multiple Spanish hospitals and provided a large real-world cohort of individuals with diabetes [28]. From this cohort, Navarro-González et al. derived and validated a 2-year risk model for incident CKD in people with T2DM, achieving good discrimination and recall, and ultimately integrating the final logistic regression model into a web-based tool for educational use [29]. All analyses complied with strict data privacy and transparency standards. In a similar manner, additional models are being developed to predict HF in individuals with T2DM, with results in preparation at the time of writing.

Another example of predictive analytics is the COMFE Registry study, which investigated factors predicting myocardial recovery following hospitalization for de novo heart failure with reduced left ventricular ejection fraction [30]. Conducted across two referral centers in Spain, the study included 248 patients, 63% of whom showed a significant improvement in left ventricular ejection fraction within 3–4 months. The study identified non-ischemic etiologies, such as tachycardiomyopathy and valvular disease, as strong predictors of recovery, while ischemic cardiomyopathy was a negative predictor. N-terminal pro-B-type natriuretic peptide levels were consistently lower in patients who experienced myocardial improvement, making this biomarker valuable for monitoring prognosis. Additionally, the study observed a significant reduction in heart failure-related hospitalizations in the group that experienced improvement in left ventricular ejection fraction, highlighting the potential of predictive models to identify patients who are most likely to benefit from targeted therapeutic strategies aimed at improving myocardial function.

2.1.4 Patient-Centered Studies

Ethnographic studies offer invaluable qualitative insights into the lived experiences of patients, caregivers, and HCPs by exploring attitudes, behaviors, and contextual factors that quantitative data alone cannot capture. Through in-depth semi-structured interviews, these studies reveal nuanced aspects of patient journeys, including symptom perception, treatment challenges, and psychosocial impacts.

The LupusVoice study in Spain exemplifies this approach by highlighting significant gaps in disease awareness and the understanding of systemic lupus erythematosus among patients, caregivers, and healthcare providers [31]. The study uncovered diagnostic delays, emotional distress, and social isolation linked to limited disease recognition, emphasizing the critical need for increased education and support to improve patient quality of life.

Patient-reported outcome measures further complement ethnographic findings by quantifying symptom burden and health-related quality of life, enabling a comprehensive assessment of disease impact. Questionnaires such as RTT [32] and AIQR [33] have been validated and integrated into clinical practice. For example, AIRQ was included in the Spanish GEMA asthma guideline [34], supporting personalized disease management.

Moreover, publishing plain language summaries enhances the accessibility of complex research findings to patients and the broader public, fostering transparency and engagement in healthcare decisions. An example of this is the XALOC-1 study by Jackson et al. [35] on the use of benralizumab in real life, which has been recently published in three different languages, including Spanish. Together, these methodologies enrich MA understanding of patient needs and experiences, informing more patient-centric evidence generation and communication strategies.

2.2 Strategic Use of Evidence for the Company

The strategic use of evidence refers to the use of fit-for-purpose evidence to support patient-centered personalized decision making across the healthcare ecosystem, enabling the right innovative intervention for the right patient profile and care pathway at the right time. In practice, evidence generation is used to better characterize real-world patient phenotypes, unmet needs, care gaps, and outcomes, to make informed decisions. This strategic use of evidence is essential in ensuring that generated data drive improvements in patient care, market access, and healthcare ecosystem optimization. By applying evidence strategically, we empower stakeholders such as HCPs, payers, and system partners to make informed decisions that impact clinical practice and patient outcomes. This approach maximizes the availability and utility of RWD, aligning evidence generation with healthcare objectives and medical innovation. Presenting RWE in actionable formats enables informed clinical decisions, fostering continuous improvement in healthcare delivery.

2.2.1 Generation of the Value Story

The Value Story is a comprehensive data-driven narrative that communicates the clinical, economic, and humanistic value of a pharmaceutical product. It helps create confidence in regulators, payers, and prescribers to make informed decisions about the approval, financing, or prescription of pharmaceutical products, respectively, thereby strengthening their therapeutic positioning.

A core objective of a product’s Value Story is to differentiate treatments from competitors by demonstrating its added value not only in clinical outcomes but also in improving patient experiences and delivering sustainable benefits to healthcare systems. This process starts early with the identification of specific patient profiles or subpopulations (such as those at a higher risk of disease progression, non-responders, or those most likely to benefit from treatment). Such stratification enables more targeted interventions and supports evidence-based decisions to remove barriers to access following regulatory approval, thereby advancing personalized medicine strategies.

The Value Story is further strengthened by RWE that informs predictive market models (e.g., patient funnels) and sustains outcomes related to healthcare ecosystem sustainability, as exemplified by initiatives such as CARESA [36]. Incorporating HEOR and RWD into the Value Story ensures that both clinical outcomes and healthcare system benefits are effectively communicated to stakeholders. The CaReMe calculator described by Navarro-González et al. [37] represents a typical example of useful data to be integrated in a Value Story.

Ultimately, aligning RWE generation with value and access strategies ensures P&R readiness, facilitating dossier submissions and accelerating approvals, while securing broad and equitable patient access. The Value Story supports innovative P&R mechanisms going beyond traditional financial agreements (price per volume, discounts) that we call innovative value strategies. These include service-based products, such as offering virtual assistants or patient support programs upon product introduction, and outcome-based agreements, described previously, which tie reimbursement to the actual clinical and economic value delivered by the product in practice.

2.2.2 Actionable Data Visualization and Versatility

Actionable data and effective visualization are essential for the strategic use of evidence, enabling us to collect insights in real time, communicate the Value Story efficiently, and make informed decisions based on up-to-date data. These tools empower stakeholders to act on evidence more promptly, fostering an agile and responsive healthcare ecosystem. Medical affairs plays a pivotal role in delivering RWE in ways that are relevant and actionable, tailored to the unique requirements of clinicians, payers, and other decision makers. For example, interactive dashboards empower MA in-field teams to engage HCPs and payers in dynamic conversations, enabling exploration of patient-specific data and the application of real-time analytics.

To effectively meet the diverse and evolving needs of healthcare stakeholders, it is crucial to present evidence in formats that are timely, highly customized, and interactive. In order to address this challenge, AstraZeneca has recently developed a comprehensive healthcare data platform called ATLAS [2], customized for each different country (13 at the time of writing this article) and designed to enable data-driven planning and decision making for several chronic conditions. In Spain, it includes nationwide data on chronic kidney disease, asthma, COPD, heart failure, and hypertension.

The platform aggregates publicly available data relating to disease prevalence, cause-specific mortality, cause-specific hospitalizations, including associated costs and length of stay, and key sociodemographic variables. This information is sourced directly from the Spanish Ministry of Health and the National Institute of Statistics, compiled at the national, regional (19 autonomous communities), and provincial (52 provinces) levels. Public data are mapped from 2016 through to 2023, which is the most recently available data at the time of writing this article. Prevalence, mortality, and hospitalizations are analyzed using diagnosis code lists, with admission rates and crude mortality rates calculated at the provincial level. Additionally, the platform incorporates data on prescriptions, pathology, and care processes (e.g., laboratory parameters, referrals) at the national level, obtained from private vendors. All private data are analyzed using standardized definitions and code lists, with metrics mapped with a 1-year lag (currently up to 2024).

An additional distinctive strength of the ATLAS platform is its ability to systematically and dynamically integrate both public and private data sources: datasets that have traditionally been siloed and not analyzed together. Data integration within ATLAS follows a rigorous multi-stage process encompassing data selection, standardization, quality control, and traceability. Data are incorporated into ATLAS only after passing predefined inclusion criteria that assess relevance, coverage, update frequency, and data provider reliability. Once selected, datasets undergo automated and manual standardization routines that harmonize variable names, units, and taxonomies, ensuring that comparable indicators across domains are measured consistently.

Quality assurance rules are applied at several levels: automated validation checks detect anomalies in data structure and completeness; statistical consistency tests compare new data against historical series and external reference sources to identify potential deviations; and periodic manual audits verify the accuracy of automated procedures. All transformation steps (from ingestion to aggregation) are fully logged, maintaining end-to-end traceability that links every derived metric back to its original source.

This transparent lineage enables reproducibility, enhances interpretability of metrics, and reinforces confidence in the robustness of analyses performed within the ATLAS platform. Furthermore, the applied data engineering processes enable automation of key variable updates and validation, ensuring that the results displayed are reliable and reproducible. Ongoing external and internal user feedback is incorporated through iterative platform updates to maintain relevance and usability across stakeholder groups.

Geospatial mapping and advanced data visualization tools are used to create an interactive dashboard that is user friendly and intuitive to navigate. Data sources are refreshed regularly to ensure the platform reflects the most up-to-date information available. The dashboard’s heatmap feature provides a comprehensive perspective, allowing users to gain insights from the national level down to individual autonomous communities’ practices. This level of detail enables a thorough understanding of how diseases affect local healthcare services and helps to identify areas for quality improvement. The platform is designed to guide clinicians and healthcare decision makers by presenting actionable RWD in a digestible accessible format. By facilitating insight-driven service planning, as well as the design and implementation of clinical guidelines, the platform ultimately supports the delivery of effective patient care.

2.2.3 Innovation and New Technologies

Impactful outcomes of evidence generation cannot be achieved in isolation. Maximizing public-private collaboration is essential for building a robust evidence base that addresses medical needs and aligns with broader healthcare ecosystem objectives. AstraZeneca Spain has adopted an approach of driving innovation by actively participating in public-private consortia that pursue transformative strategies for managing chronic diseases across different healthcare system levels [38,39,40]. This collaborative approach allows for the implementation of cutting-edge methodologies such as federated data models, AI-driven insights, and personalized healthcare analytics. Through these efforts, we aim to contribute to the evolving landscape of evidence-based medicine, ensuring that MA plays a pivotal role in generating high-value insights that drive better patient outcomes and sustainable healthcare solutions.

The incorporation of innovative methodologies and collaborations with emerging technology companies and startups is transforming RWE generation within MA. These initiatives leverage advanced analytical tools, AI, and machine learning to extract deeper clinical insights from complex and voluminous RWD, enabling more precise and scalable understanding of disease and treatment landscapes. This represents a challenge difficult to address by a single institution alone. Therefore, we at AstraZeneca Spain decided to engage in partnerships with startups specialized in different innovative methodologies, such as MEDIWHALE [41] and IDOVEN [42], which exemplify this innovation ecosystem by providing validated tools that enhance data capture, patient monitoring, and predictive analytics. These collaborations facilitate the integration of novel data sources and the development of customized solutions that address unmet needs in clinical research and healthcare delivery.

These advancements underscore the critical role of AI-driven approaches in unlocking hidden data, refining disease phenotyping, and supporting personalized medicine. By embracing such cutting-edge methodologies, MA can lead the future of evidence generation delivering actionable insights that drive improved patient outcomes and healthcare sustainability.