Inside taXonomy, a next-gen closed claims solution for real-world evidence generation

Healthcare claims data are one of the most powerful sources of data in generating real-world evidence (RWE). These data help answer questions about therapeutic area incidence and prevalence, healthcare resource utilization and costs, treatment patterns and where care was rendered and by whom, etc.

Yet, even with modern data platforms, a familiar barrier remains: in a highly competitive environment shaped by shifting patient behavior and data fragmentation, execution of reliable and meaningful longitudinal studies demand assembling claims data that are robust, comprehensive and stable, not simply accessible.

This post is the first in a four-part series on HealthVerity taXonomy, the industry’s most comprehensive, consistent and curated closed claims dataset designed to power longitudinal study designs across health economics, patient outcomes, epidemiological, and safety and effectiveness initiatives. Today, we’ll discuss the taXonomy build and how it’s differentiated compared to legacy closed claims datasets.

In subsequent posts, we’ll focus on key topics and features of HealthVerity taXonomy, including the data model design, validation of its mortality data, its novel and proprietary cost data offering and responsible methods that can be used by researchers supporting study execution directly for, or on behalf of, life science organizations.

The challenge isn’t just data access, it’s reliable and consistent assembly of the data

In longitudinal RWE generation initiatives, the key consideration is often not how the data are applied, but whether the data are sufficiently reliable, consistent and stable to support defensible study design and interpretation. A few limitations that should be considered include:

• Fragmentation and fixed footprints: Many datasets are anchored to a fixed payer panel, or even a single payer. This inherently introduces the potential for biases, whether demographic or socioeconomic, geographical or by payer type, it can limit the generalizability of findings or analytic feasibility when study needs evolve.
  
  
• Market Dynamics: Payer membership, benefit design and provider networks change over time, which can alter the observable population even when study definitions remain constant. Datasets anchored to narrow or static contributors can be especially sensitive to these shifts, complicating interpretation of time trends and generalizability.
  
  
• Longitudinal completeness: Many RWE questions depend on establishing baseline periods, follow-up windows and continuous enrollment to ensure valid attribution and outcome measurement. Longitudinal completeness is a function of enrollment continuity, coverage breadth. Data sources constrained to a fixed or homogeneous set of payers may limit longitudinal visibility of patient journeys over time.

To summarize, the value of closed claims data is not defined by their availability, but by their methodological integrity and their ability to sustain rigorous, reproducible evidence generation.

The promise: closed claims that are configurable and transparent

HealthVerity taXonomy is a standardized, carefully assembled and de-duplicated closed claims dataset designed for analytics-ready use, but what differentiates it isn’t simply that it’s “clean” or “curated” (many modern datasets aim for that).

HealthVerity taXonomy payer type and enrollment distribution.

The key difference is that taXonomy is underpinned by the HealthVerity Marketplace model, which enables a dynamic approach to building a more reliable closed claims dataset. HealthVerity Marketplace includes the largest dataset of closed claims data in the industry, and we continue to add additional closed claims sources over time, expanding breadth and depth in ways that can help address coverage gaps and strengthen longitudinal analytics. The unique benefit of the marketplace model enables:

• Configurable supplier footprint over time: Teams can start with a targeted set of closed claims suppliers and have the option to augment by integrating additional closed claims assets without disrupting their broader approach. This directly addresses common limitations of fixed or narrow payer footprints. As the footprint expands, epidemiologic analyses become more statistically robust and generalizable, supporting detailed evaluation of both common and rare events. Greater scale and diversity can also improve power to detect small but meaningful associations between exposures and outcomes, reducing the risk of false negatives.
  
  
• Seamless expansion that strengthens longitudinal follow-up: Expanding the footprint can increase enrollment coverage and continuity, improving the ability to observe patients consistently across baseline and follow-up windows, an essential requirement for longitudinal cohort studies. By capturing a more complete view of enrollment over time, this approach helps reduce selection bias and supports a more accurate representation of patients’ healthcare activity, so outcomes and findings are comprehensive and less likely to be skewed by over- or under-representation of specific patient groups.

This “configurable foundation” matters because RWE research questions evolve, often quickly, and a static dataset can become a constraint.

Why we launched the taXonomy blog series for real-world evidence teams

We’re writing this series because we believe rigorous, transparent real-world evidence can improve decision-making across the life sciences industry, and higher quality data is the catalyst in achieving this. By pulling back the curtain on how taXonomy is designed and the value it possesses from a research perspective, we aim to assist research teams produce analyses that are faster to execute and stronger methodologically. In future posts, we’ll go deeper into the components that matter most to RWE teams:

• Post 2: Under the hood—data model design and why it’s intentionally client-friendly for cohort studies
  
  
• Post 3: Mortality in RWE—what’s included and how to use it responsibly
  
  
• Post 4: Cost in claims—how to think about allowed amounts vs standardized benchmarks and best practices for analysis