The cost of care: Adding an economic lens to real-world outcomes research

Evelyn Liu: Our team has always focused on clinical outcomes research in terms of treatment patterns, overall survival, real world progression-free survival, time to next treatment, and other time-to-event analyses that show how patients do on their treatment regimens and in their care overall. Increasingly, our biopharmaceutical clients raise questions that clinical endpoints can’t directly answer on their own. What does HCRU look like for different patient cohorts across care settings such as inpatient stays, outpatient treatment, and physician office visits? And what is the economic burden of the disease in terms of cost of care?

Answering these questions unlocks a more holistic picture of the disease landscape, one that captures both the clinical and the economic burden of illness. By connecting the Tempus de-identified clinical and molecular dataset with healthcare claims data, we built an HCRU and cost analysis capability that sits directly on top of our existing outcomes research work.

Evelyn Liu: Raw claims data arrives with procedure codes, revenue codes, and bill type codes. Together they describe the medical service and the care setting, though in a messy, unstandardized format. Our healthcare claims subject matter experts and data science teams ingested this raw data and built a Tempus claims data model that translates it into standardized care-setting categories and costs. The model maps each code to the publicly available Medicare fee schedule and assigns costs based on that mapping.

We standardize to the national Medicare fee schedule to reduce the noise that comes from payer and regional variation. With standardized costs, two patients who received the same procedure or medical service become more comparable across geographies and insurance systems.

Evelyn Liu: What sets us apart most is our ability to pair claims with our existing multimodal dataset. With clinical and molecular data alongside claims, we can identify costs by clinical milestone, line of therapy, disease progression, and biomarker expression. That ties the clinical and molecular aspects of a patient’s care to the economic data.

A few other factors matter:

• Breadth: Our claims footprint spans many patient groups, so our view of the patient population extends beyond any one carrier’s covered lives or a single-payer data source.
• Care-setting granularity: We can break costs out into drug costs, inpatient stays, outpatient care, emergency room visits, and physician office visits. Because of our mapped data model, we can go deeper still into settings such as hospice and skilled nursing facilities.
• Attribution: In an oncology study, for example, diagnosis codes let us separate costs driven by cancer from costs driven by other comorbidities. That matters when you are trying to isolate the burden of a particular disease.

Evelyn Liu: We design and perform these analyses with our biopharmaceutical partners. For publications, the main audience is the wider medical community: providers, payers, and regulatory bodies.

These studies quantify clinical outcomes, biomarker patterns of interest, and the economic burden of disease together. A study might ask,

What are the patient demographics and clinical attributes? What are the treatment regimens and patterns in a given cancer? What does first-line therapy cost, and what does second-line therapy cost? What do costs look like between lines of therapy, and how do they break down across medical and pharmaceutical care settings?

On top of that, we can stratify by biomarker expression or other clinical attributes.

The results feed directly into partner decision-making. Clients tell us they use these cost analyses to update their internal budget impact models and as supporting evidence in payer and regulatory conversations.

Evelyn Liu: We recently completed a few cost analyses with clients that followed the cost of care for patients with a specific indication through first, second, and third lines of therapy. We combined Tempus’ curated treatment data with claims-derived costs and attributed costs to each line of therapy. We broke all-cause and cancer-specific costs down by care category, mapped each patient’s full treatment journey as costs accumulated, and identified the drivers of cost within each line of therapy.

Curated progression events in the Tempus data also let us look at costs tied directly to disease progression. We compared costs between patients who experienced progression and patients who did not, and quantified the increase. Overall, these analyses added an economic lens to the clinical research question that extends beyond treatment patterns and outcomes.

Evelyn Liu: Traditional clinical analyses characterize a patient’s clinical status, the treatment they receive, and how well a therapy works. The economic analysis completes that picture and shows what it costs to deliver a treatment regimen. Read together, they give a much more complete view of the disease and the treatment area. In practice, clients come away with sharper budget impact models and defensible evidence for payer and regulatory conversations, which supports market access and pricing strategy.

Evelyn Liu: Build the HCRU and cost analysis component into the study design from the beginning, so the desired outcomes and research questions carry through every analysis. As with traditional clinical outcomes research, well-defined research goals, research questions, and cohorts remain crucial to informing study-specific methods and design choices.

We’ve presented our cost analysis work at conferences and in publications, and it has drawn strong interest from partners and wider audiences. We look forward to continuing these analyses with our biopharmaceutical partners.