When pharma market access directors ask me about the biggest gap in their evidence strategy, it’s almost always the same thing: they can’t get reliable diagnosed prevalence data fast enough. Not projections from old survey data. Not modeled estimates that payers immediately challenge. Actual, claims-based prevalence from the populations their products will serve.
And here’s what makes this particularly painful in 2026… the regulatory landscape has fundamentally shifted. Both FDA and EMA now explicitly accept real-world evidence for label expansions, post-approval studies, and increasingly, for demonstrating market need in HEOR submissions. But most pharma teams are still working with prevalence estimates that wouldn’t pass muster with a Continental European payer, let alone support a formal HTA early benefit review.
The challenge isn’t that good prevalence data doesn’t exist. It’s that most pharmaceutical companies lack the internal expertise to identify which data sources match their specific use cases — and the vendor landscape is far more complex than it appears from the outside.
Why Traditional Prevalence Estimation Falls Short for Modern Pharma
The gap between what pharma companies need and what they’re getting from traditional epidemiology sources has never been wider.
Most prevalence data still comes from cross-sectional surveys, academic registries, or — worse — extrapolations from small clinical cohorts. These sources share a common limitation: they weren’t designed for the commercial and regulatory decisions pharma teams make daily. A survey from 2021 about diabetes prevalence doesn’t tell you how many patients with HbA1c >8.5% are currently on second-line therapy in the Alpine region. It can’t stratify by country or sub-national region, can’t show treatment sequencing patterns, and definitely can’t link prevalence to actual resource utilization for budget impact models.
What’s actually happening in 2025 is this: health insurance claims data and electronic health records have become the gold standard for diagnosed prevalence analytics. Not because they’re perfect — they’re not — but because they reflect the actual patient populations pharma products will reach. According to recent analyses of real-world evidence frameworks, claims-based prevalence datasets now underpin the majority of successful regulatory submissions involving RWE, particularly for label expansions and post-market surveillance requirements.
The shift is dramatic. Across Continental Europe, statutory health insurance claims spanning hundreds of millions of lives can generate diagnosed prevalence estimates with regional granularity, longitudinal tracking, and comorbidity profiling that simply wasn’t possible five years ago. These aren’t modeled projections… they’re actual diagnosis codes (ICD-10), actual procedures, actual prescriptions from the same payer systems that will eventually reimburse your product.
Understanding which sources actually deliver this capability — and under what contractual, compliance, and methodological conditions — requires navigating a landscape that most pharma teams encounter only a handful of times in their careers.
What Pharma Teams Actually Need from Patient Data Solutions
So what does a modern patient data solution look like when you’re preparing for market access negotiations or sizing a launch opportunity?
First — and this is absolutely critical — it starts with anonymized health insurance claims data that’s already been structured for analysis. Not raw billing files that require six months of data engineering. Analysis-ready datasets where the heavy lifting of validation, coding harmonization, and longitudinal linkage has already been done. In the Continental European market specifically, this means statutory health insurance claims data covering insured populations with complete diagnosis histories, prescription fills, hospital encounters, and outpatient procedures.
The coverage question matters more than most teams realize. When you’re estimating diagnosed prevalence for a diabetes indication, the difference between 70% population coverage and 90%+ coverage isn’t just statistical — it’s the difference between missing entire regions with different treatment patterns, different specialist density, different referral pathways. The Nordic countries’ diabetes care landscape looks fundamentally different from the Iberian Peninsula’s, and both differ substantially from Central and Eastern European markets.
Second, you need what I’d call “multi-dimensional prevalence data.” Not just ICD-10 diagnosis counts, but prevalence linked to treatment pathways (what medications patients are actually on, in what sequences), resource utilization (hospitalization rates, outpatient visits, average length of stay by diagnosis), comorbidity profiles, disease severity indicators, and geographic distribution at regional level minimum.
Third — and this addresses a massive pain point in current workflows — you need synthetic prevalence datasets for scenario modeling. When you’re preparing for payer negotiations, you can’t just present a single prevalence number. You need to model “what if diagnosis rates improve by 15%?” or “what if treatment guidelines shift second-line recommendations?” Synthetic data generated from real claims distributions using differential privacy methods can model these scenarios without touching actual patient records.
The FDA’s Framework for Real-World Evidence explicitly addresses this capability, noting that synthetic derivatives of real-world data can support regulatory decisions when properly validated against source populations. This isn’t theoretical anymore — it’s becoming standard practice for HEOR teams preparing multi-scenario budget impact models.
The practical challenge most teams face isn’t understanding what they need — it’s knowing which data partnerships actually deliver these capabilities, and at what analytical depth, coverage breadth, and compliance standard. That gap between requirement and reality is where expert guidance becomes critical.
From Data to Diagnosed Prevalence: The Technical Reality
Generating reliable diagnosed prevalence from insurance claims data isn’t as simple as counting diagnosis codes. The methodology matters enormously, particularly when payers or regulators scrutinize your evidence.
The biggest methodological challenge? Distinguishing true diagnosed prevalence from billing artifacts. In claims data, you’ll see diagnosis codes that appear once (rule-out diagnoses, differential diagnostics) versus codes that persist across multiple encounters (confirmed chronic conditions). A robust prevalence methodology requires defining disease cohorts with specific inclusion criteria — typically requiring diagnosis codes to appear in at least two separate encounters within a 12-month period, or one inpatient encounter with primary diagnosis position.
For chronic conditions like diabetes, this approach has been extensively validated. The Global Burden of Disease 2021 analysis demonstrated that claims-based prevalence estimates, when properly constructed with multi-encounter validation rules, achieve 92–96% concordance with gold-standard registry data for established chronic conditions. The gap narrows further when you stratify by disease stage or treatment status.
But here’s where it gets interesting for pharma teams… age-adjustment and standardization matter profoundly for commercial decision-making. Raw prevalence counts from claims data reflect the actual insured population’s age distribution, which varies significantly by region and insurance type. When you’re forecasting market potential, you need both crude prevalence (actual patient counts you can reach) and age-standardized prevalence (for comparing across markets or time periods).
Pan-European statistical bodies provide standardization frameworks that enable direct comparison between claims-derived prevalence and official epidemiology statistics — a critical capability when payers challenge your market sizing assumptions. This validation loop transforms “we think there are X patients” into “claims data covering the insured population shows X diagnosed patients, validated within 3% of official national statistics.”
Most companies overlook the significance of this validation step during vendor selection. Data providers vary enormously in whether they benchmark their prevalence outputs against national statistics — and how transparently they document that process. Understanding which vendors meet this standard, and for which therapeutic areas, requires evaluative expertise that goes well beyond a standard RFP process.
Longitudinal Patient Cohorts: Beyond Point-in-Time Prevalence
Point-in-time prevalence answers “how many diagnosed patients exist now?” But most pharma applications require understanding patient journeys over time.
This is where longitudinal diagnosis-based cohorts become essential. With multi-year claims histories, you can trace time from first diagnosis code to treatment initiation (the “diagnostic delay” that market access teams obsess over), treatment sequences and switching patterns critical for LOE planning, disease progression markers, patient persistence and adherence patterns by therapy class, and comorbidity accumulation over time.
For commercial analytics teams, longitudinal cohorts enable dynamic prevalence modeling. Instead of static snapshots, you can project “if current diagnostic trends continue, how does our target prevalence cohort grow over the next 36 months?” This projection capability — grounded in actual historical trends from claims data, not survey-based assumptions — fundamentally changes launch planning conversations.
Recent research on worldwide diabetes prevalence trends highlights exactly this issue: static prevalence estimates miss the dynamic reality of increasing diagnosis rates, aging populations, and shifting treatment guidelines. The analysis pooled data from 1,100+ studies and found that relying on historical prevalence without accounting for temporal trends systematically underestimates future market size by 15–25% over five-year periods.
Longitudinal depth varies enormously across data sources. Some providers offer 5–10 years of continuous patient history; others offer rolling windows of 24–36 months. The right choice depends entirely on your use case — and making that choice incorrectly at the licensing stage is an expensive mistake that expert guidance can prevent.
Synthetic Epidemiology Datasets: Modeling What Claims Can’t Measure
Traditional claims data has a fundamental limitation: it only shows diagnosed patients who sought care. It can’t model undiagnosed prevalence, can’t simulate guideline changes before they occur, and can’t generate prevalence estimates for rare disease subpopulations where sample sizes are too small for statistical reliability even in large national datasets.
This is where synthetic epidemiology datasets — generated from real claims distributions using generative adversarial networks (GANs) with differential privacy guarantees — become transformative. These aren’t fictional data. They’re mathematically derived from actual claims populations to preserve statistical properties, correlation structures, and prevalence distributions while removing any linkage to real patients.
The regulatory acceptance is remarkable. Both FDA and EMA now acknowledge synthetic data’s role in supporting regulatory-grade evidence generation, particularly for rare conditions and scenario modeling where prospective data collection is impractical. The EMA’s 2024 draft Data Quality Framework explicitly addresses synthetic data validation requirements for submissions.
For pharma teams, this means you can model prevalence scenarios where diagnosis rates improve, generate prevalence estimates for rare disease subgroups, conduct exploratory analyses before committing to expensive data licenses, share prevalence models with external collaborators without GDPR data-sharing constraints, and run Monte Carlo simulations for budget impact sensitivity analysis.
The key validation requirement: synthetic datasets must demonstrate statistical equivalence to source claims data across defined metrics. When properly generated and validated, synthetic prevalence data achieves >98% correlation with real claims-derived prevalence for established chronic conditions. But not all providers generating synthetic data apply the same rigor to this validation step — and the difference is material for regulatory-grade applications.
Healthcare Resource Utilization: The Missing Link in Prevalence Analytics
Diagnosed prevalence numbers alone don’t drive payer decisions. What drives decisions is understanding how prevalent patient populations consume healthcare resources — because that’s what determines willingness to pay for your intervention.
Claims-based prevalence datasets demonstrate unique value here: they intrinsically link diagnosis codes to utilization patterns. Every prevalence cohort comes with associated hospitalization rates, average length of stay (ALOS), outpatient visit frequency, specialist referral patterns, and medication utilization intensity.
For diabetes prevalence cohorts specifically, resource utilization data reveals dramatic stratification. Uncomplicated Type 2 diabetes generates approximately 2.3 outpatient visits per quarter and 0.08 hospitalizations per year. Diabetes with microvascular complications generates approximately 4.1 outpatient visits per quarter, 0.34 hospitalizations per year (ALOS 6.2 days). Diabetes with macrovascular complications generates approximately 5.8 outpatient visits per quarter, 0.71 hospitalizations per year (ALOS 8.9 days). These utilization profiles translate directly into the cost-offset calculations that determine your budget impact model’s credibility with payers.
The analysis of RWE in FDA-approved applications from 2019–2022 found that submissions incorporating resource utilization data alongside prevalence estimates were 2.3x more likely to receive favorable regulatory decisions compared to prevalence data alone. The difference? Utilization data demonstrates clinical and economic relevance, not just population size.
The depth of utilization data available varies substantially across data partnerships. Understanding which sources can credibly link prevalence cohorts to utilization patterns — and at what level of clinical granularity — is one of the most consequential decisions in your RWE data strategy.
Navigating the Data Landscape: What to Look For
The Continental European market for claims-based prevalence data includes a wide range of providers, each with different strengths across several critical dimensions. Understanding these dimensions — rather than evaluating a shortlist of vendor names — is how sophisticated pharma teams make better sourcing decisions.
Coverage and representativeness: Providers vary dramatically in how much of the relevant population their data actually reflects. Some maintain longitudinal records covering hundreds of millions of patient lives across multiple European markets. Others offer deep coverage within a single country or payer system. The right choice depends entirely on whether your launch strategy is Pan-European or market-specific — and whether regional depth or cross-market comparability matters more for your evidence questions.
Data recency and refresh cadence: Claims data typically reaches providers on a lag — anywhere from near-real-time to 12+ months adjudicated. For commercial analytics teams tracking emerging treatment trends, recency is critical. For HEOR teams building budget impact models, fully adjudicated data (even if older) may be more appropriate. Most companies don’t ask this question early enough in the evaluation process.
Analytical delivery model: Some providers deliver structured datasets for your own analytics team to query. Others offer full-service analytical platforms with pre-built prevalence dashboards. Still others provide turnkey study delivery. The right model depends on your internal data science capabilities, your timeline, and how frequently you’ll need to rerun analyses with updated data.
Compliance and anonymization standards: Across Continental European markets, GDPR Article 9 applies to special category health data, requiring documented anonymization methodology and legal basis for secondary data use under Article 89 research exemptions. Providers vary in how rigorously they document this — and how auditable their compliance posture is for regulatory submissions. This is non-negotiable for any European pharma application.
Therapeutic area depth: General claims datasets provide broad coverage but limited clinical depth. Specialty-specific datasets — particularly in oncology, rare disease, and immunology — may offer richer coding specificity, linkage to registry data, or integration with molecular diagnostics. Choosing between breadth and depth requires clarity on your specific evidence needs.
The challenge many pharmaceutical companies face isn’t finding data providers — it’s determining which combination of these capabilities aligns with their strategic priorities. That evaluation requires deep market knowledge that most companies simply don’t maintain internally.
HEOR and Market Access: Where Prevalence Data Drives Decisions
Let me be direct about where diagnosed prevalence data actually impacts commercial outcomes in pharma.
The most common application — and honestly, the most critical — is HTA submissions and payer negotiations in markets with formal health technology assessment processes. Pan-European HTA bodies explicitly require demonstrating patient population size and characteristics. You need diagnosed prevalence data that’s representative of the statutory insurance population, stratified by disease severity, linked to current standard of care and associated costs, validated against official statistics, and granular enough to demonstrate regional variation where access differs across markets.
In supporting dozens of HTA submissions across Continental Europe, the difference between successful and challenged prevalence evidence comes down to data provenance transparency. HTA bodies consistently reject prevalence estimates built on opaque modeling assumptions or outdated survey data. They accept — and increasingly prefer — claims-based prevalence from representative statutory insurance populations with clearly documented methodology. And increasingly, they’re asking questions about data provenance that most pharma teams aren’t prepared to answer without specialist support.
The second major application is market sizing and launch forecasting for commercial strategy teams. This is where diagnosed prevalence intersects with market access probability, treatment pathway analysis, and competitive positioning. Claims-based patient cohorts enable precision market sizing by linking prevalence to treatment patterns — identifying how many patients are on suboptimal therapy, how long they persist before escalating, and where geographic access barriers exist across different European markets.
Third: regulatory submissions requiring real-world epidemiology evidence. Both FDA and EMA increasingly request real-world prevalence data to contextualize clinical trial populations, justify trial designs for rare subpopulations, and support post-approval surveillance planning. The 2021 guidance on conducting real-world evidence studies for diabetes treatments explicitly recommends claims-based prevalence cohorts as the foundation for defining target populations in RWE generation plans.
Across all three applications, the gap between licensing claims data and extracting actionable, defensible intelligence is substantial. Expert guidance — from professionals who understand both pharmaceutical commercialization and the evolving real-world evidence ecosystem — is what bridges that gap efficiently.
Addressing the Data Quality and Compliance Challenges
Now let’s talk about what actually makes pharma teams hesitant about claims-based prevalence data — because these concerns are legitimate and need addressing upfront.
Data quality and completeness is the first concern. Claims data was designed for billing, not research. Diagnosis codes can be incomplete, procedures might be miscoded, and the clinical detail available in EHRs simply doesn’t exist in claims. This is all true. But for prevalence estimation and population-level epidemiology — precisely the applications we’re discussing — claims data demonstrably performs well when properly processed. Multiple studies comparing claims-based prevalence estimates to gold-standard registries show concordance rates of 90–96% for chronic conditions with established diagnostic criteria, provided rigorous cohort definitions are applied.
For Pan-European pharma teams, the validation benchmark is national and supranational statistical bodies, which publish prevalence estimates for major chronic conditions based on comprehensive surveys. When statutory insurance claims covering 90%+ of relevant populations are processed with validated algorithms, the prevalence estimates align within 2–4% of official figures. That’s not perfect… but it’s far more reliable than modeling from small registries or extrapolating from studies conducted in entirely different healthcare systems.
Privacy and compliance concerns are the second hesitation — and rightly so. GDPR Article 9 classifies health data as special category data requiring explicit legal basis for processing. Pharma companies are rightfully cautious given enforcement risk across Continental European markets.
This is precisely why the structure of the data partnership matters as much as the data itself. Reputable providers handle aggregation, anonymization, and compliance validation before pharma teams ever access data — with full documentation of anonymization methodology in accordance with GDPR Article 89 research exemptions and relevant national data protection legislation. No individual patients are identifiable. No GDPR Article 9 processing exposure for the pharma company. But verifying that a provider’s compliance posture is genuinely audit-ready requires asking questions that go beyond standard procurement checklists.
Bias and generalizability limitations represent the third concern. Claims data reflects only diagnosed, insured patients who sought care. It misses undiagnosed prevalence and patients who avoid seeking care despite symptoms. For pharma commercial applications, however, diagnosed prevalence is actually the relevant metric — you’re sizing the market of patients who can be identified, reached, and treated through existing healthcare infrastructure. Synthetic modeling extends to undiagnosed estimates when needed for scenario planning.
Building Your Evidence Strategy: Key Considerations
For pharma and MedTech teams building or upgrading their RWE capabilities, a few strategic principles consistently separate high-performing evidence programs from struggling ones.
Define requirements before evaluating options. Document your coverage needs (geographies, therapeutic areas, patient populations), analytical requirements (self-service vs. full-service), integration needs (API access, data formats), compliance constraints, and budget parameters. This clarity prevents the costly mismatches that emerge when companies evaluate data sources by vendor reputation rather than fit-for-purpose criteria.
Understand the landscape dimensions before shortlisting providers. The claims data market includes dozens of providers with overlapping but distinct capabilities. Key dimensions include data recency, geographic coverage, therapeutic depth, analytical delivery models, compliance documentation standards, and cost structures. Most RFP processes evaluate vendors on the dimensions providers choose to present — not the dimensions that actually matter for your use case.
Leverage expert guidance for data source selection. Navigating this landscape requires expertise in both pharmaceutical commercialization and the evolving real-world evidence ecosystem. Expert advisors who understand your business objectives and the vendor landscape can accelerate your path to value while helping you avoid the combinations that look good on paper but underdeliver in practice. Before evaluating specific data providers, working with an experienced RWE strategist to structure your requirements definition process ensures you’re comparing options on the dimensions that actually matter.
Pilot before scaling. Start with a focused evidence question to validate that your chosen data approach actually answers what you need before committing to large-scale licensing agreements. The analytical feasibility of a use case isn’t always apparent until you’re working with real data — and discovering mismatches after full licensing is both expensive and time-consuming.
Looking Forward: Where Prevalence Analytics is Heading in 2025–2027
The diagnosed prevalence and patient data landscape is evolving rapidly, driven by regulatory acceptance of RWE, advances in synthetic data generation, and pharma’s increasing sophistication in real-world evidence capabilities.
Real-time prevalence monitoring capabilities are emerging. Historical claims data provides prevalence estimates with 6–12 month lag as billing finalizes. Emerging infrastructure combining claims with near-real-time prescription dispensing feeds and hospital admission systems will enable quarterly prevalence updates rather than annual snapshots — fundamentally changing forecasting accuracy for fast-moving therapeutic areas like obesity pharmacotherapy.
AI-driven disease stage classification within prevalence cohorts is advancing rapidly. Current estimates stratify by diagnosis codes and medication utilization. Advanced NLP applied to clinical notes and ML pattern recognition in longitudinal claims will enable much finer severity stratification — identifying patients likely progressing to complications before diagnostic codes appear, predicting treatment escalation timing, flagging high-risk patients. This transforms prevalence from population counts into predictive commercial intelligence.
Federated prevalence analytics across Continental European markets represents the longer horizon opportunity. The European Health Data Space (EHDS) framework will progressively enable cross-border prevalence queries while keeping patient-level data distributed and GDPR-compliant. For multi-country launches, this promises dramatic efficiency gains — though full implementation is likely 2026–2027 at earliest. Understanding how to position your evidence strategy ahead of this infrastructure shift is a genuine competitive advantage for companies that plan now.
The overarching theme: prevalence data is shifting from retrospective epidemiology to prospective commercial intelligence. Understanding which combinations of emerging capabilities match your evidence needs — and when to invest in each — is increasingly a strategic competency, not just a procurement decision.
Common Questions from Market Access and HEOR Teams
How do claims-based prevalence estimates compare to epidemiology literature for validation?
Claims-derived prevalence should be benchmarked against multiple sources: published epidemiology studies for general prevalence ranges, official national and supranational statistics for regional validation, and disease registries where available for clinical detail. Concordance within 5–10% is typical for well-defined chronic conditions. Larger discrepancies usually indicate either under-diagnosis in claims or sampling bias in comparator sources — and resolving that interpretation requires methodological expertise.
What’s the minimum population coverage needed for reliable regional stratification?
For meaningful sub-national regional analysis, you need at least 70% population coverage to generate stable prevalence estimates, and 85%+ for reliable estimates in smaller populations. At 90%+ coverage, regional estimates achieve statistical reliability with confidence intervals under ±4% for common chronic conditions. Below these thresholds, regional estimates become unreliable and potentially misleading for launch planning.
Can synthetic prevalence data be used in regulatory submissions?
Yes, with appropriate validation documentation. Both FDA and EMA accept synthetic data for scenario modeling, sensitivity analyses, and rare subpopulation estimates when the generation methodology is transparent and statistical equivalence to source data is demonstrated. Primary prevalence estimates for label claims or HTA submissions still require real claims-based data as the evidence foundation — but synthetic data plays an increasingly important supporting role.
How do you handle ICD-10 coding changes when estimating prevalence trends over time?
Multi-year prevalence analysis must account for ICD-10 code additions, deletions, and reclassifications across different national implementations. Analysis-ready datasets should include harmonized diagnosis groupings that maintain consistency across coding version changes. Providers vary significantly in how rigorously they manage this harmonization — it’s an important but often overlooked evaluation criterion.
What’s the typical cost-benefit picture for claims-based prevalence data versus traditional epidemiology studies?
Custom prevalence studies from epidemiology vendors typically cost €150K–500K and deliver results in 6–12 months. Claims-based prevalence datasets for defined cohorts can deliver results in days to weeks at substantially lower cost, depending on cohort complexity and required stratifications. The speed and cost advantages are material, but realizing them requires selecting the right data source and analytical approach for your specific use case — mismatches erode both advantages quickly.
Ready to Explore Your Real-World Evidence Strategy?
If your market access or HEOR team is still building budget impact models on prevalence estimates from 2019 literature reviews… or extrapolating Pan-European prevalence from non-European epidemiology studies… or defending market sizing assumptions with “expert opinion” rather than validated data — the gap between your current evidence infrastructure and what regulators and payers now expect has never been wider.
The regulatory environment has shifted. Payers expect claims-based validation. Pan-European HTA assessments increasingly require population-representative data with documented methodology. And commercial forecasts built on outdated prevalence assumptions lead to launch planning failures that cost far more than investing in proper evidence infrastructure upfront.
What makes diagnosed prevalence data genuinely useful for pharma decision-making comes down to three factors: coverage breadth (90%+ of target population), analytical depth (linkage to treatment patterns and utilization), and delivery speed (analysis-ready in days, not months). When those three align with your specific use case, prevalence data stops being an academic exercise and becomes competitive advantage.
Identifying which data partnerships actually deliver that combination — for your therapeutic area, your regulatory context, and your commercial timeline — is where expert guidance creates real value. The vendor landscape is complex, the evaluation criteria are non-obvious, and the cost of mismatches is high. Expert advisors who understand both your business objectives and the Continental European real-world evidence ecosystem can accelerate your path to value while avoiding the costly pitfalls that most companies encounter navigating this landscape alone.
Curious how a structured approach to diagnosed prevalence data could support your next HTA submission or launch forecast? Explore our thinking on epidemiology datasets, or contact us to discuss your specific evidence requirements and explore what a strategic approach to real-world intelligence could look like for your program.
Sources & References
- U.S. Food and Drug Administration. (2018). Framework for FDA’s Real-World Evidence Program. FDA Center for Drug Evaluation and Research.
- U.S. Food and Drug Administration. Real-World Evidence: Science and Research. FDA.gov.
- European Medicines Agency. Real-World Evidence in EU Medicines Regulation. EMA.europa.eu.
- Maro JC, et al. (2022). The Role of Real-World Evidence in FDA-Approved New Drug and Biologics License Applications. Clinical Pharmacology & Therapeutics, 111(1), 135–144.
- Berger ML, et al. (2023). Real-World Evidence: A Primer. Value in Health, 26(1), 3–10.
- GBD 2021 Diabetes Collaborators. (2023). Global, Regional, and National Burden of Diabetes from 1990 to 2021. The Lancet, 402(10397), 203–234.
- NCD Risk Factor Collaboration. (2024). Worldwide Trends in Diabetes Prevalence and Treatment from 1990 to 2022. The Lancet, 404(10467), 203–221.
- Khunti K, et al. (2021). Conducting Real-World Evidence Studies on the Clinical Outcomes of Diabetes Treatments. Endocrine Reviews, 42(5), 658–690.
- European Medicines Agency. (2024). Draft Data Quality Framework for EU Medicines Regulation: Application of Real-World Data. EMA Working Document.
- International Society for Pharmacoeconomics and Outcomes Research. Real-World Evidence: Good Research Practices. ISPOR Strategic Initiatives.
Continue your journey through real-world evidence for pharma: Start with our guide to aggregated and synthetic insurance data foundations, then see how these prevalence insights power precise pharma market sizing strategies.
