Using Integrated Clinical and Claims Data to Improve Oncology Care and Cost-Modeling

This article was originally published on Onco'Zine - The International Oncology Network.

For the fourth consecutive year, oncology will again be one of the top five causes of employer healthcare spending in 2026. While payers brace for these expenses, they recognize the value of recent advances in cell and gene therapies that have led to improved treatment outcomes for members.

In addition to improving clinical outcomes, payers must also consider the costs related to the growing number of people living with and beyond cancer. There are currently more than 18 million cancer survivors in the U.S., and that number is expected to exceed 26 million by 2040. These individuals will likely require extended care beyond the end of their initial therapy. Examples of extended care include managing chronic conditions, providing mental health support and coordinating ongoing care.

While new therapies offer the potential for transformational improvements for many of the people with few or no remaining treatment options, the high up-front cost of CAR-T and other cell-based therapies is daunting to health plans. Also, there is market uncertainty about how these therapies will function in the “real world.”

With a multitude of factors impacting decision-making for the treatment journey, access to accurate, complete data is critical to get a holistic view of how care is being provided to the patient. Unfortunately, data is typically fragmented and resides in silos. Clinical, claims, biomarker, and historical treatment data are generally found in separate databases, which not only slows down the processes but also obscures the full picture of a member’s care.

Fragmented data leaves clinical and operational leaders without the insights they need to act thoughtfully and decisively. What is really needed is a unified, clinically informed intelligence layer capable of interpreting what fragmented data actually means for care decisions, risk exposure, and downstream cost.

Using Both Clinical and Claims Data to Develop a Better Predictive Model for Oncology Risk Management‍

Many plans rely upon claims data for a system-wide view of the delivery of care and payment for services. From a positive perspective, claims data reflects the services provided to a patient and the payment made for those services, providing a reliable history of the care that was delivered and the costs that were incurred. Claims data is therefore most suitable for analyzing the utilization of treatments, the negotiated price paid for those treatments and the total cost of care.

However, relying solely upon claims data is not sufficient and may present some significant challenges. While a claims data report may say that a laboratory test was performed, it will not tell you what the result of that test was. In addition, claims data is structured around coding systems that limit the amount of detail that can be extracted for analysis.

Another issue is that claims data can be slow to report and may not be timely enough to assist with the acute decision-making process that involves caring for an oncology patient. Obtaining claims data and having it available from various payers, providers, and health plans can also be problematic, further delaying patient management.

Claims data is affected by the complexity of the billing and payment process. Variability in coding practices, inconsistency in documentation, and errors in coding and billing can all lead to inaccuracies in the results of claims data analyses. Historically, the use of claims data has helped reinforce current treatment trends rather than identify the optimal future treatment pathway.

Value of Clinical Data

Clinical data provides the level of detail missing from claims data. Clinical data shows the results of laboratory tests, biomarkers, imaging studies, treatment responses and physician notes indicating treatment intent.

Additional detail provides a more complete picture of a patient’s biological condition and how that condition changes over time. Clinical data allow clinicians and researchers to track disease progression, assess treatment response, and incorporate biomarkers into predictive models.

In addition, clinical data captures the variability of care delivery. There are many ways to deliver treatment. Each hospital, each physician and each care setting delivers treatment differently. Even the reference ranges for laboratory tests can impact treatment outcomes. Documenting and accounting for this variability is important for creating accurate and clinically relevant models of care.

Combine Claims + Clinical Data

Using both clinical and claims data gives decision makers a single source of truth. Claims data reflect treatment use and the costs of delivering treatments. Clinical data reflect the biological basis of the disease and the treatments provided to address that disease.

An analogy of this would be examining a patient’s tumor staging. A diagnostic study may provide structural information about a tumor, but without pathological and biomarker data, the biological basis of the disease is unknown.

Similarly, claims data without clinical context provides only part of the big picture. When these two data types are combined, a more comprehensive and actionable view of the patient’s care is developed. For example, when claims data showing a shift to second-line therapy is combined with biomarker data indicating a specific mutation profile, models can identify whether that escalation was clinically appropriate or driven by coding conventions.

When both clinical and claims data are used, care and cost models can identify emerging treatment patterns, identify risk trajectories and predict future care needs. This is especially useful when evaluating new treatments.

Developing Models of Emerging Oncology Treatments Using Real-World Pricing and Utilization Trends‍

Both hope and uncertainty exist when developing new oncology treatments. CAR-T therapy, an FDA- approved immunotherapy, provides outstanding remissions in certain forms of blood cancer for patients who have had no success with previous treatments.

But as previously noted, CAR-T costs between $400,000 and $500,000, and the true cost of care regularly exceeds $1 million when hospitalization, toxicity management and indirect institutional costs are included. This creates barriers for stakeholders who plan and manage financial risk for new treatments. To justify these expenditures, payers increasingly question the “real world” performance of these novel, innovative treatments.

While the results of clinical trials are an essential part of proving the safety and efficacy of therapies, they often fall short in fully replicating their actual performance. Many payers and regulators are requiring Real-World Evidence (RWE), the clinical information regarding the usage, safety and effectiveness of medical products derived from analysis of Real-World Data (RWD) collected outside of traditional, tightly controlled clinical trials. This complements Randomized Controlled Trials (RCTs) by evaluating long-term outcomes in diverse patient populations, helping regulators approve new indications and supporting payer utilization.

As new treatments move from clinical trials to plan-wide clinical use, RWD will become available to update the initial predictions, demonstrating the differences between the clinical trial population and the population of real-world patients. RWD data will also illustrate the differences in the rate of treatment adoption and the rate of treatment response. Health plans increasingly seek this level of evidence that accounts for variability in patient populations and clinical practices in order to accurately project the full downstream cost impacts.

Much of the uncertainty regarding these emerging therapies can be addressed with care and cost modeling, a technique that uses early clinical trial results along with RWD that enables stakeholders to simulate how new treatments may affect particular member populations. Models examine the following variables to determine how new treatments may affect the population:

• The effectiveness of the treatment
• Estimated cost of the treatment
• Whether the patient qualifies under the treatment’s eligibility criteria and existing treatment pathways.

By varying these parameters, stakeholders can then evaluate multiple potential outcomes. This could involve varying the percentage of patients who qualify for the treatment, the cost of the treatment and when the treatment is initiated. By using this type of care and cost modeling, stakeholders can forecast a variety of possible outcomes before recommending the widespread use of new treatments. This will allow them to systematically evaluate the clinical impact and financial risk of new treatments prior to implementation.

Stakeholders can continually refine their models as new data becomes available. This will ultimately allow them to transition from making generalized recommendations applicable to large populations to making targeted and personalized recommendations for specific subpopulations of patients.

Designing Targeted Oncology Treatment Pathways that Balance Treatment Outcomes and Financial Viability‍

Traditionally, oncology guidelines have established treatment eligibility using general criteria. For example, some oncology guidelines have generally recommended treatment based upon the lines of therapy that the patient has received. While guidelines provide a framework for treatment, they do not capture the variation in patient responses.

Research on actual use of treatments has indicated that there is significant variation in treatment response among specific groups of patients, regardless of the original diagnosis. These findings underscore the need for more detailed and nuanced methods for selecting treatments for patients and identifying which patients are more likely to experience positive results from particular therapies. The analysis process can include genetic and biomarker profiles, clinical characteristics of the patient, previous treatments received, and the patient’s previous response to treatment.

Armed with this data, clinicians gain a clearer picture of how each patient group will respond to a particular therapy. This level of information provides meaningful insights that inform treatment recommendations, leading to improved health outcomes and reduced waste.

A well-architected data integration strategy holds the key to forecasting the total cost of care over a longer term. There is consensus that delays in diagnosis and early treatment interventions have led to disease progression to later stages of cancer, which leads to increased costs.

By modeling the total cost of care over several years, stakeholders can determine whether intervening early will ultimately result in better health outcomes and lower overall expenditures. This modeling approach supports the development of value-based cancer care payment models that reward providers for providing the best long-term benefits to their patients, rather than focusing solely on containing costs in the near term.

Establishing a Biologically Precise and Financially Aligned Oncology System‍

With new technologies and vast amounts of data becoming available, oncology care is experiencing unprecedented growth and change. At the same time, the complexity of care and the financial constraints facing the healthcare system are growing exponentially.

To successfully address these challenges, it will take more than a collection of disparate data sources. Rather, we must integrate clinical data, claims data and real-world evidence to gain a comprehensive and actionable view of patient care. An integrated view of patient care will enable us to conduct predictive analyses, effectively evaluate new and innovative treatments, and design clinically relevant and cohort-specific treatment pathways.

Our ultimate goal is not merely to utilize better analytics tools. We want to make better decisions that will improve our patients’ health and ensure we can sustainably deliver quality care.

Contact us today to learn how Evidium's computational medical intelligence can help you understand the clinical drivers behind risk — and make decisions that are better informed, more precise, and grounded in what's actually happening for patients.